% This program by D. E. Knuth is not copyrighted and can be used freely.
% Version 0 was released in December, 1981.
% Version 1 was released in September, 1982, with version 0 of TeX.
% Slight changes were made in October, 1982, for version 0.6 of TeX.
% Version 1.1 changed "_" to "\_" if not within an identifier (November, 1982).
% Version 1.2 added @@= and @@\ and marked changed modules (December, 1982).
% Version 1.3 marked and indexed changed modules better (January, 1983).
% Version 1.4 added "history" (February, 1983).
% Version 1.5 conformed to TeX version 0.96 (March, 1983).
% Version 1.6 conformed to TeX version 0.98 (May, 1983).
% Version 1.7 introduced the new change file format (June, 1983).
% Version 2 was released in July, 1983, with version 0.999 of TeX.
% Version 2.1 corrected a bug in changed_module reckoning (August, 1983).
% Version 2.2 corrected it better (August, 1983).
% Version 2.3 starts the output with \input webmac (August, 1983).
% Version 2.4 fixed a bug in compress(#) (September, 1983).
% Version 2.5 cleared xrefswitch after module names (November, 1983).
% Version 2.6 fixed a bug in declaration of trans array (January, 1984).
% Version 2.7 fixed a bug in real constants (August, 1984).
% Version 2.8 fixed a bug in change_buffer movement (August, 1985).
% Version 2.9 increased max_refs and max_toks to 30000 each (January, 1987).
% Version 3, for Sewell's book, fixed long-line bug in input_ln (March, 1989).
% Version 3.1 fixed a bug for programs with only one module (April, 1989).
% Version 4 was major change to allow 8-bit input (September, 1989).
% Version 4.1, for Breitenlohner, avoids English-only output (March, 1990).
% Version 4.2 conforms to ANSI standard for-loop rules (September, 1990).
% Version 4.3 catches extra } in input (Breitenlohner, September, 1991).
% Version 4.4 corrects changed_module logic, %-overflow (January, 1992).
% Here is TeX material that gets inserted after \input webmac
\def\hang{\hangindent 3em\indent\ignorespaces}
\font\ninerm=cmr9
\let\mc=\ninerm % medium caps for names like SAIL
\def\PASCAL{Pascal}
\def\pb{$\.|\ldots\.|$} % Pascal brackets (|...|)
\def\v{\.{\char'174}} % vertical (|) in typewriter font
\def\dleft{[\![} \def\dright{]\!]} % double brackets
\mathchardef\RA="3221 % right arrow
\mathchardef\BA="3224 % double arrow
\def\({} % kludge for alphabetizing certain module names
\def\title{WEAVE}
\def\contentspagenumber{15} % should be odd
\def\topofcontents{\null\vfill
\titlefalse % include headline on the contents page
\def\rheader{\mainfont Appendix D\hfil \contentspagenumber}
\centerline{\titlefont The {\ttitlefont WEAVE} processor}
\vskip 15pt
\centerline{(Version 4.4)}
\vfill}
\pageno=\contentspagenumber \advance\pageno by 1
@* Introduction.
This program converts a \.{WEB} file to a \TeX\ file. It was written
by D. E. Knuth in October, 1981; a somewhat similar {\mc SAIL} program had
been developed in March, 1979, although the earlier program used a top-down
parsing method that is quite different from the present scheme.
The code uses a few features of the local \PASCAL\ compiler that may need
to be changed in other installations:
\yskip\item{1)} Case statements have a default.
\item{2)} Input-output routines may need to be adapted for use with a particular
character set and/or for printing messages on the user's terminal.
\yskip\noindent
These features are also present in the \PASCAL\ version of \TeX, where they
are used in a similar (but more complex) way. System-dependent portions
of \.{WEAVE} can be identified by looking at the entries for `system
dependencies' in the index below.
@!@^system dependencies@>
The ``banner line'' defined here should be changed whenever \.{WEAVE}
is modified.
@d banner=='This is WEAVE, Version 4.4'
@ The program begins with a fairly normal header, made up of pieces that
@^system dependencies@>
will mostly be filled in later. The \.{WEB} input comes from files |web_file|
and |change_file|, and the \TeX\ output goes to file |tex_file|.
If it is necessary to abort the job because of a fatal error, the program
calls the `|jump_out|' procedure, which goes to the label |end_of_WEAVE|.
@d end_of_WEAVE = 9999 {go here to wrap it up}
@p @t\4@>@<Compiler directives@>@/
program WEAVE(@!web_file,@!change_file,@!tex_file);
label end_of_WEAVE; {go here to finish}
const @<Constants in the outer block@>@/
type @<Types in the outer block@>@/
var @<Globals in the outer block@>@/
@<Error handling procedures@>@/
procedure initialize;
var @<Local variables for initialization@>@/
begin @<Set initial values@>@/
end;
@ Some of this code is optional for use when debugging only;
such material is enclosed between the delimiters |debug| and $|gubed|$.
Other parts, delimited by |stat| and $|tats|$, are optionally included
if statistics about \.{WEAVE}'s memory usage are desired.
@d debug==@{ {change this to `$\\{debug}\equiv\null$' when debugging}
@d gubed==@t@>@} {change this to `$\\{gubed}\equiv\null$' when debugging}
@f debug==begin
@f gubed==end
@#
@d stat==@{ {change this to `$\\{stat}\equiv\null$'
when gathering usage statistics}
@d tats==@t@>@} {change this to `$\\{tats}\equiv\null$'
when gathering usage statistics}
@f stat==begin
@f tats==end
@ The \PASCAL\ compiler used to develop this system has ``compiler
directives'' that can appear in comments whose first character is a dollar sign.
In production versions of \.{WEAVE} these directives tell the compiler that
@^system dependencies@>
it is safe to avoid range checks and to leave out the extra code it inserts
for the \PASCAL\ debugger's benefit, although interrupts will occur if
there is arithmetic overflow.
@<Compiler directives@>=
@{@&$C-,A+,D-@} {no range check, catch arithmetic overflow, no debug overhead}
@!debug @{@&$C+,D+@}@+ gubed {but turn everything on when debugging}
@ Labels are given symbolic names by the following definitions. We insert
the label `|exit|:' just before the `\ignorespaces|end|\unskip' of a
procedure in which we have used the `|return|' statement defined below;
the label `|restart|' is occasionally used at the very beginning of a
procedure; and the label `|reswitch|' is occasionally used just prior to
a \&{case} statement in which some cases change the conditions and we wish to
branch to the newly applicable case.
Loops that are set up with the \&{loop} construction defined below are
commonly exited by going to `|done|' or to `|found|' or to `|not_found|',
and they are sometimes repeated by going to `|continue|'.
@d exit=10 {go here to leave a procedure}
@d restart=20 {go here to start a procedure again}
@d reswitch=21 {go here to start a case statement again}
@d continue=22 {go here to resume a loop}
@d done=30 {go here to exit a loop}
@d found=31 {go here when you've found it}
@d not_found=32 {go here when you've found something else}
@ Here are some macros for common programming idioms.
@d incr(#) == #:=#+1 {increase a variable by unity}
@d decr(#) == #:=#-1 {decrease a variable by unity}
@d loop == @+ while true do@+ {repeat over and over until a |goto| happens}
@d do_nothing == {empty statement}
@d return == goto exit {terminate a procedure call}
@f return == nil
@f loop == xclause
@ We assume that |case| statements may include a default case that applies
if no matching label is found. Thus, we shall use constructions like
@^system dependencies@>
$$\vbox{\halign{#\hfil\cr
|case x of|\cr
1: $\langle\,$code for $x=1\,\rangle$;\cr
3: $\langle\,$code for $x=3\,\rangle$;\cr
|othercases| $\langle\,$code for |x<>1| and |x<>3|$\,\rangle$\cr
|endcases|\cr}}$$
since most \PASCAL\ compilers have plugged this hole in the language by
incorporating some sort of default mechanism. For example, the compiler
used to develop \.{WEB} and \TeX\ allows `|others|:' as a default label,
and other \PASCAL s allow syntaxes like `\ignorespaces|else|\unskip' or
`\&{otherwise}' or `\\{otherwise}:', etc. The definitions of |othercases|
and |endcases| should be changed to agree with local conventions.
(Of course, if no default mechanism is available, the |case| statements of
this program must be extended by listing all remaining cases.)
@d othercases == others: {default for cases not listed explicitly}
@d endcases == @+end {follows the default case in an extended |case| statement}
@f othercases == else
@f endcases == end
@ The following parameters are set big enough to handle \TeX, so they
should be sufficient for most applications of \.{WEAVE}.
@<Constants...@>=
@!max_bytes=45000; {|1/ww| times the number of bytes in identifiers,
index entries, and module names; must be less than 65536}
@!max_names=5000; {number of identifiers, index entries, and module names;
must be less than 10240}
@!max_modules=2000;{greater than the total number of modules}
@!hash_size=353; {should be prime}
@!buf_size=100; {maximum length of input line}
@!longest_name=400; {module names shouldn't be longer than this}
@!long_buf_size=500; {|buf_size+longest_name|}
@!line_length=80; {lines of \TeX\ output have at most this many characters,
should be less than 256}
@!max_refs=30000; {number of cross references; must be less than 65536}
@!max_toks=30000; {number of symbols in \PASCAL\ texts being parsed;
must be less than 65536}
@!max_texts=2000; {number of phrases in \PASCAL\ texts being parsed;
must be less than 10240}
@!max_scraps=1000; {number of tokens in \PASCAL\ texts being parsed}
@!stack_size=200; {number of simultaneous output levels}
@ A global variable called |history| will contain one of four values
at the end of every run: |spotless| means that no unusual messages were
printed; |harmless_message| means that a message of possible interest
was printed but no serious errors were detected; |error_message| means that
at least one error was found; |fatal_message| means that the program
terminated abnormally. The value of |history| does not influence the
behavior of the program; it is simply computed for the convenience
of systems that might want to use such information.
@d spotless=0 {|history| value for normal jobs}
@d harmless_message=1 {|history| value when non-serious info was printed}
@d error_message=2 {|history| value when an error was noted}
@d fatal_message=3 {|history| value when we had to stop prematurely}
@#
@d mark_harmless==@t@>@+if history=spotless then history:=harmless_message
@d mark_error==history:=error_message
@d mark_fatal==history:=fatal_message
@<Glob...@>=@!history:spotless..fatal_message; {how bad was this run?}
@ @<Set init...@>=history:=spotless;
@* The character set.
One of the main goals in the design of \.{WEB} has been to make it readily
portable between a wide variety of computers. Yet \.{WEB} by its very
nature must use a greater variety of characters than most computer
programs deal with, and character encoding is one of the areas in which
existing machines differ most widely from each other.
To resolve this problem, all input to \.{WEAVE} and \.{TANGLE} is
converted to an internal eight-bit code that is essentially standard
ASCII, the ``American Standard Code for Information Interchange.''
The conversion is done immediately when each character is read in.
Conversely, characters are converted from ASCII to the user's external
representation just before they are output. (The original ASCII code
was seven bits only; \.{WEB} now allows eight bits in an attempt to
keep up with modern times.)
Such an internal code is relevant to users of \.{WEB} only because it is
the code used for preprocessed constants like \.{"A"}. If you are writing
a program in \.{WEB} that makes use of such one-character constants, you
should convert your input to ASCII form, like \.{WEAVE} and \.{TANGLE} do.
Otherwise \.{WEB}'s internal coding scheme does not affect you.
@^ASCII code@>
Here is a table of the standard visible ASCII codes:
$$\def\:{\char\count255\global\advance\count255 by 1}
\count255='40
\vbox{
\hbox{\hbox to 40pt{\it\hfill0\/\hfill}%
\hbox to 40pt{\it\hfill1\/\hfill}%
\hbox to 40pt{\it\hfill2\/\hfill}%
\hbox to 40pt{\it\hfill3\/\hfill}%
\hbox to 40pt{\it\hfill4\/\hfill}%
\hbox to 40pt{\it\hfill5\/\hfill}%
\hbox to 40pt{\it\hfill6\/\hfill}%
\hbox to 40pt{\it\hfill7\/\hfill}}
\vskip 4pt
\hrule
\def\^{\vrule height 10.5pt depth 4.5pt}
\halign{\hbox to 0pt{\hskip -24pt\O{#0}\hfill}&\^
\hbox to 40pt{\tt\hfill#\hfill\^}&
&\hbox to 40pt{\tt\hfill#\hfill\^}\cr
04&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
05&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
06&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
07&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
10&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
11&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
12&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
13&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
14&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
15&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
16&\:&\:&\:&\:&\:&\:&\:&\:\cr\noalign{\hrule}
17&\:&\:&\:&\:&\:&\:&\:\cr}
\hrule width 280pt}$$
(Actually, of course, code @'040 is an invisible blank space.) Code @'136
was once an upward arrow (\.{\char'13}), and code @'137 was
once a left arrow (\.^^X), in olden times when the first draft
of ASCII code was prepared; but \.{WEB} works with today's standard
ASCII in which those codes represent circumflex and underline as shown.
@<Types...@>=
@!ASCII_code=0..255; {eight-bit numbers, a subrange of the integers}
@ The original \PASCAL\ compiler was designed in the late 60s, when six-bit
character sets were common, so it did not make provision for lowercase
letters. Nowadays, of course, we need to deal with both capital and small
letters in a convenient way, so \.{WEB} assumes that it is being used
with a \PASCAL\ whose character set contains at least the characters of
standard ASCII as listed above. Some \PASCAL\ compilers use the original
name |char| for the data type associated with the characters in text files,
while other \PASCAL s consider |char| to be a 64-element subrange of a larger
data type that has some other name.
In order to accommodate this difference, we shall use the name |text_char|
to stand for the data type of the characters in the input and output
files. We shall also assume that |text_char| consists of the elements
|chr(first_text_char)| through |chr(last_text_char)|, inclusive. The
following definitions should be adjusted if necessary.
@^system dependencies@>
@d text_char == char {the data type of characters in text files}
@d first_text_char=0 {ordinal number of the smallest element of |text_char|}
@d last_text_char=255 {ordinal number of the largest element of |text_char|}
@<Types...@>=
@!text_file=packed file of text_char;
@ The \.{WEAVE} and \.{TANGLE} processors convert between ASCII code and
the user's external character set by means of arrays |xord| and |xchr|
that are analogous to \PASCAL's |ord| and |chr| functions.
@<Globals...@>=
@!xord: array [text_char] of ASCII_code;
{specifies conversion of input characters}
@!xchr: array [ASCII_code] of text_char;
{specifies conversion of output characters}
@ If we assume that every system using \.{WEB} is able to read and write the
visible characters of standard ASCII (although not necessarily using the
ASCII codes to represent them), the following assignment statements initialize
most of the |xchr| array properly, without needing any system-dependent
changes. For example, the statement \.{xchr[@@\'101]:=\'A\'} that appears
in the present \.{WEB} file might be encoded in, say, {\mc EBCDIC} code
on the external medium on which it resides, but \.{TANGLE} will convert from
this external code to ASCII and back again. Therefore the assignment
statement \.{XCHR[65]:=\'A\'} will appear in the corresponding \PASCAL\ file,
and \PASCAL\ will compile this statement so that |xchr[65]| receives the
character \.A in the external (|char|) code. Note that it would be quite
incorrect to say \.{xchr[@@\'101]:="A"}, because |"A"| is a constant of
type |integer|, not |char|, and because we have $|"A"|=65$ regardless of
the external character set.
@<Set init...@>=
xchr[@'40]:=' ';
xchr[@'41]:='!';
xchr[@'42]:='"';
xchr[@'43]:='#';
xchr[@'44]:='$';
xchr[@'45]:='%';
xchr[@'46]:='&';
xchr[@'47]:='''';@/
xchr[@'50]:='(';
xchr[@'51]:=')';
xchr[@'52]:='*';
xchr[@'53]:='+';
xchr[@'54]:=',';
xchr[@'55]:='-';
xchr[@'56]:='.';
xchr[@'57]:='/';@/
xchr[@'60]:='0';
xchr[@'61]:='1';
xchr[@'62]:='2';
xchr[@'63]:='3';
xchr[@'64]:='4';
xchr[@'65]:='5';
xchr[@'66]:='6';
xchr[@'67]:='7';@/
xchr[@'70]:='8';
xchr[@'71]:='9';
xchr[@'72]:=':';
xchr[@'73]:=';';
xchr[@'74]:='<';
xchr[@'75]:='=';
xchr[@'76]:='>';
xchr[@'77]:='?';@/
xchr[@'100]:='@@';
xchr[@'101]:='A';
xchr[@'102]:='B';
xchr[@'103]:='C';
xchr[@'104]:='D';
xchr[@'105]:='E';
xchr[@'106]:='F';
xchr[@'107]:='G';@/
xchr[@'110]:='H';
xchr[@'111]:='I';
xchr[@'112]:='J';
xchr[@'113]:='K';
xchr[@'114]:='L';
xchr[@'115]:='M';
xchr[@'116]:='N';
xchr[@'117]:='O';@/
xchr[@'120]:='P';
xchr[@'121]:='Q';
xchr[@'122]:='R';
xchr[@'123]:='S';
xchr[@'124]:='T';
xchr[@'125]:='U';
xchr[@'126]:='V';
xchr[@'127]:='W';@/
xchr[@'130]:='X';
xchr[@'131]:='Y';
xchr[@'132]:='Z';
xchr[@'133]:='[';
xchr[@'134]:='\';
xchr[@'135]:=']';
xchr[@'136]:='^';
xchr[@'137]:='_';@/
xchr[@'140]:='`';
xchr[@'141]:='a';
xchr[@'142]:='b';
xchr[@'143]:='c';
xchr[@'144]:='d';
xchr[@'145]:='e';
xchr[@'146]:='f';
xchr[@'147]:='g';@/
xchr[@'150]:='h';
xchr[@'151]:='i';
xchr[@'152]:='j';
xchr[@'153]:='k';
xchr[@'154]:='l';
xchr[@'155]:='m';
xchr[@'156]:='n';
xchr[@'157]:='o';@/
xchr[@'160]:='p';
xchr[@'161]:='q';
xchr[@'162]:='r';
xchr[@'163]:='s';
xchr[@'164]:='t';
xchr[@'165]:='u';
xchr[@'166]:='v';
xchr[@'167]:='w';@/
xchr[@'170]:='x';
xchr[@'171]:='y';
xchr[@'172]:='z';
xchr[@'173]:='{';
xchr[@'174]:='|';
xchr[@'175]:='}';
xchr[@'176]:='~';@/
xchr[0]:=' '; xchr[@'177]:=' '; {these ASCII codes are not used}
@ Some of the ASCII codes below @'40 have been given symbolic names in
\.{WEAVE} and \.{TANGLE} because they are used with a special meaning.
@d and_sign=@'4 {equivalent to `\.{and}'}
@d not_sign=@'5 {equivalent to `\.{not}'}
@d set_element_sign=@'6 {equivalent to `\.{in}'}
@d tab_mark=@'11 {ASCII code used as tab-skip}
@d line_feed=@'12 {ASCII code thrown away at end of line}
@d form_feed=@'14 {ASCII code used at end of page}
@d carriage_return=@'15 {ASCII code used at end of line}
@d left_arrow=@'30 {equivalent to `\.{:=}'}
@d not_equal=@'32 {equivalent to `\.{<>}'}
@d less_or_equal=@'34 {equivalent to `\.{<=}'}
@d greater_or_equal=@'35 {equivalent to `\.{>=}'}
@d equivalence_sign=@'36 {equivalent to `\.{==}'}
@d or_sign=@'37 {equivalent to `\.{or}'}
@ When we initialize the |xord| array and the remaining parts of |xchr|,
it will be convenient to make use of an index variable, |i|.
@<Local variables for init...@>=
@!i:0..255;
@ Here now is the system-dependent part of the character set.
If \.{WEB} is being implemented on a garden-variety \PASCAL\ for which
only standard ASCII codes will appear in the input and output files, you
don't need to make any changes here. But if you have, for example, an extended
character set like the one in Appendix~C of {\sl The \TeX book}, the first
line of code in this module should be changed to
$$\hbox{|for i:=1 to @'37 do xchr[i]:=chr(i);|}$$
\.{WEB}'s character set is essentially identical to \TeX's, even with respect to
characters less than @'40.
@^system dependencies@>
Changes to the present module will make \.{WEB} more friendly on computers
that have an extended character set, so that one can type things like
\.^^Z\ instead of \.{<>}. If you have an extended set of characters that
are easily incorporated into text files, you can assign codes arbitrarily
here, giving an |xchr| equivalent to whatever characters the users of
\.{WEB} are allowed to have in their input files, provided that unsuitable
characters do not correspond to special codes like |carriage_return|
that are listed above.
(The present file \.{WEAVE.WEB} does not contain any of the non-ASCII
characters, because it is intended to be used with all implementations of
\.{WEB}. It was originally created on a Stanford system that has a
convenient extended character set, then ``sanitized'' by applying another
program that transliterated all of the non-standard characters into
standard equivalents.)
@<Set init...@>=
for i:=1 to @'37 do xchr[i]:=' ';
for i:=@'200 to @'377 do xchr[i]:=' ';
@ The following system-independent code makes the |xord| array contain a
suitable inverse to the information in |xchr|.
@<Set init...@>=
for i:=first_text_char to last_text_char do xord[chr(i)]:=" ";
for i:=1 to @'377 do xord[xchr[i]]:=i;
xord[' ']:=" ";
@* Input and output.
The input conventions of this program are intended to be very much like those
of \TeX\ (except, of course, that they are much simpler, because much less
needs to be done). Furthermore they are identical to those of \.{TANGLE}.
Therefore people who need to make modifications to all three systems
should be able to do so without too many headaches.
We use the standard \PASCAL\ input/output procedures in several places that
\TeX\ cannot, since \.{WEAVE} does not have to deal with files that are named
dynamically by the user, and since there is no input from the terminal.
@ Terminal output is done by writing on file |term_out|, which is assumed to
consist of characters of type |text_char|:
@^system dependencies@>
@d print(#)==write(term_out,#) {`|print|' means write on the terminal}
@d print_ln(#)==write_ln(term_out,#) {`|print|' and then start new line}
@d new_line==write_ln(term_out) {start new line}
@d print_nl(#)== {print information starting on a new line}
begin new_line; print(#);
end
@<Globals...@>=
@!term_out:text_file; {the terminal as an output file}
@ Different systems have different ways of specifying that the output on a
certain file will appear on the user's terminal. Here is one way to do this
on the \PASCAL\ system that was used in \.{TANGLE}'s initial development:
@^system dependencies@>
@<Set init...@>=
rewrite(term_out,'TTY:'); {send |term_out| output to the terminal}
@ The |update_terminal| procedure is called when we want
to make sure that everything we have output to the terminal so far has
actually left the computer's internal buffers and been sent.
@^system dependencies@>
@d update_terminal == break(term_out) {empty the terminal output buffer}
@ The main input comes from |web_file|; this input may be overridden
by changes in |change_file|. (If |change_file| is empty, there are no changes.)
@<Globals...@>=
@!web_file:text_file; {primary input}
@!change_file:text_file; {updates}
@ The following code opens the input files. Since these files were listed
in the program header, we assume that the \PASCAL\ runtime system has
already checked that suitable file names have been given; therefore no
additional error checking needs to be done. We will see below that
\.{WEAVE} reads through the entire input twice.
@^system dependencies@>
@p procedure open_input; {prepare to read |web_file| and |change_file|}
begin reset(web_file); reset(change_file);
end;
@ The main output goes to |tex_file|.
@<Globals...@>=
@!tex_file: text_file;
@ The following code opens |tex_file|.
Since this file was listed in the program header, we assume that the
\PASCAL\ runtime system has checked that a suitable external file name has
been given.
@^system dependencies@>
@<Set init...@>=
rewrite(tex_file);
@ Input goes into an array called |buffer|.
@<Globals...@>=@!buffer: array[0..long_buf_size] of ASCII_code;
@ The |input_ln| procedure brings the next line of input from the specified
file into the |buffer| array and returns the value |true|, unless the file has
already been entirely read, in which case it returns |false|. The conventions
of \TeX\ are followed; i.e., |ASCII_code| numbers representing the next line
of the file are input into |buffer[0]|, |buffer[1]|, \dots,
|buffer[limit-1]|; trailing blanks are ignored;
and the global variable |limit| is set to the length of the
@^system dependencies@>
line. The value of |limit| must be strictly less than |buf_size|.
We assume that none of the |ASCII_code| values
of |buffer[j]| for |0<=j<limit| is equal to 0, @'177, |line_feed|, |form_feed|,
or |carriage_return|. Since |buf_size| is strictly less than |long_buf_size|,
some of \.{WEAVE}'s routines use the fact that it is safe to refer to
|buffer[limit+2]| without overstepping the bounds of the array.
@p function input_ln(var f:text_file):boolean;
{inputs a line or returns |false|}
var final_limit:0..buf_size; {|limit| without trailing blanks}
begin limit:=0; final_limit:=0;
if eof(f) then input_ln:=false
else begin while not eoln(f) do
begin buffer[limit]:=xord[f^]; get(f);
incr(limit);
if buffer[limit-1]<>" " then final_limit:=limit;
if limit=buf_size then
begin while not eoln(f) do get(f);
decr(limit); {keep |buffer[buf_size]| empty}
if final_limit>limit then final_limit:=limit;
print_nl('! Input line too long'); loc:=0; error;
@.Input line too long@>
end;
end;
read_ln(f); limit:=final_limit; input_ln:=true;
end;
end;
@* Reporting errors to the user.
The \.{WEAVE} processor operates in three phases: first it inputs the source
file and stores cross-reference data, then it inputs the source once again and
produces the \TeX\ output file, and finally it sorts and outputs the index.
The global variables |phase_one| and |phase_three| tell which Phase we are in.
@<Globals...@>=
@!phase_one: boolean; {|true| in Phase I, |false| in Phases II and III}
@!phase_three: boolean; {|true| in Phase III, |false| in Phases I and II}
@ If an error is detected while we are debugging,
we usually want to look at the contents of memory.
A special procedure will be declared later for this purpose.
@<Error handling...@>=
@!debug@+ procedure debug_help; forward;@+gubed
@ The command `|err_print('! Error message')|' will report a syntax error to
the user, by printing the error message at the beginning of a new line and
then giving an indication of where the error was spotted in the source file.
Note that no period follows the error message, since the error routine
will automatically supply a period.
The actual error indications are provided by a procedure called |error|.
However, error messages are not actually reported during phase one,
since errors detected on the first pass will be detected again
during the second.
@d err_print(#)==
begin if not phase_one then
begin new_line; print(#); error;
end;
end
@<Error handling...@>=
procedure error; {prints `\..' and location of error message}
var@!k,@!l: 0..long_buf_size; {indices into |buffer|}
begin @<Print error location based on input buffer@>;
update_terminal; mark_error;
@!debug debug_skipped:=debug_cycle;debug_help;@+gubed
end;
@ The error locations can be indicated by using the global variables
|loc|, |line|, and |changing|, which tell respectively the first
unlooked-at position in |buffer|, the current line number, and whether or not
the current line is from |change_file| or |web_file|.
This routine should be modified on systems whose standard text editor
has special line-numbering conventions.
@^system dependencies@>
@<Print error location based on input buffer@>=
begin if changing then print('. (change file ')@+else print('. (');
print_ln('l.', line:1, ')');
if loc>=limit then l:=limit else l:=loc;
for k:=1 to l do
if buffer[k-1]=tab_mark then print(' ')
else print(xchr[buffer[k-1]]); {print the characters already read}
new_line;
for k:=1 to l do print(' '); {space out the next line}
for k:=l+1 to limit do print(xchr[buffer[k-1]]); {print the part not yet read}
if buffer[limit]="|" then print(xchr["|"]);
{end of \PASCAL\ text in module names}
print(' '); {this space separates the message from future asterisks}
end
@ The |jump_out| procedure just cuts across all active procedure levels
and jumps out of the program. This is the only non-local \&{goto} statement
in \.{WEAVE}. It is used when no recovery from a particular error has
been provided.
Some \PASCAL\ compilers do not implement non-local |goto| statements.
@^system dependencies@>
In such cases the code that appears at label |end_of_WEAVE| should be
copied into the |jump_out| procedure, followed by a call to a system procedure
that terminates the program.
@d fatal_error(#)==begin new_line; print(#); error; mark_fatal; jump_out;
end
@<Error handling...@>=
procedure jump_out;
begin goto end_of_WEAVE;
end;
@ Sometimes the program's behavior is far different from what it should be,
and \.{WEAVE} prints an error message that is really for the \.{WEAVE}
maintenance person, not the user. In such cases the program says
|confusion('indication of where we are')|.
@d confusion(#)==fatal_error('! This can''t happen (',#,')')
@.This can't happen@>
@ An overflow stop occurs if \.{WEAVE}'s tables aren't large enough.
@d overflow(#)==fatal_error('! Sorry, ',#,' capacity exceeded')
@.Sorry, x capacity exceeded@>
@* Data structures.
During the first phase of its processing, \.{WEAVE} puts identifier names,
index entries, and module names into the large |byte_mem| array, which is
packed with eight-bit integers. Allocation is sequential, since names are
never deleted.
An auxiliary array |byte_start| is used as a directory for |byte_mem|,
and the |link|, |ilk|, and |xref| arrays give further information about names.
These auxiliary arrays consist of sixteen-bit items.
@<Types...@>=
@!eight_bits=0..255; {unsigned one-byte quantity}
@!sixteen_bits=0..65535; {unsigned two-byte quantity}
@ \.{WEAVE} has been designed to avoid the need for indices that are more
than sixteen bits wide, so that it can be used on most computers. But
there are programs that need more than 65536 bytes; \TeX\ is one of these.
To get around this problem, a slight complication has been added to the
data structures: |byte_mem| is a two-dimensional array, whose first index
is either 0 or 1. (For generality, the first index is actually allowed to
run between 0 and |ww-1|, where |ww| is defined to be 2; the program will
work for any positive value of |ww|, and it can be simplified in obvious
ways if |ww=1|.)
@d ww=2 {we multiply the byte capacity by approximately this amount}
@<Globals...@>=
@!byte_mem: packed array [0..ww-1,0..max_bytes] of ASCII_code;
{characters of names}
@!byte_start: array [0..max_names] of sixteen_bits; {directory into |byte_mem|}
@!link: array [0..max_names] of sixteen_bits; {hash table or tree links}
@!ilk: array [0..max_names] of sixteen_bits; {type codes or tree links}
@!xref: array [0..max_names] of sixteen_bits; {heads of cross-reference lists}
@ The names of identifiers are found by computing a hash address |h| and
then looking at strings of bytes signified by |hash[h]|, |link[hash[h]]|,
|link[link[hash[h]]]|, \dots, until either finding the desired name
or encountering a zero.
A `|name_pointer|' variable, which signifies a name, is an index into
|byte_start|. The actual sequence of characters in the name pointed to by
|p| appears in positions |byte_start[p]| to |byte_start[p+ww]-1|, inclusive,
in the segment of |byte_mem| whose first index is |p mod ww|. Thus, when
|ww=2| the even-numbered name bytes appear in |byte_mem[0,@t$*$@>]|
and the odd-numbered ones appear in |byte_mem[1,@t$*$@>]|.
The pointer 0 is used for undefined module names; we don't
want to use it for the names of identifiers, since 0 stands for a null
pointer in a linked list.
We usually have |byte_start[name_ptr+w]=byte_ptr[(name_ptr+w) mod ww]|
for |0<=w<ww|, since these are the starting positions for the next |ww|
names to be stored in |byte_mem|.
@d length(#)==byte_start[#+ww]-byte_start[#] {the length of a name}
@<Types...@>=
@!name_pointer=0..max_names; {identifies a name}
@ @<Global...@>=
@!name_ptr:name_pointer; {first unused position in |byte_start|}
@!byte_ptr:array [0..ww-1] of 0..max_bytes;
{first unused position in |byte_mem|}
@ @<Local variables for init...@>=
@!wi: 0..ww-1; {to initialize the |byte_mem| indices}
@ @<Set init...@>=
for wi:=0 to ww-1 do
begin byte_start[wi]:=0; byte_ptr[wi]:=0;
end;
byte_start[ww]:=0; {this makes name 0 of length zero}
name_ptr:=1;
@ Several types of identifiers are distinguished by their |ilk|:
\yskip\hang |normal| identifiers are part of the \PASCAL\ program and
will appear in italic type.
\yskip\hang |roman| identifiers are index entries that appear after
\.{@@\^} in the \.{WEB} file.
\yskip\hang |wildcard| identifiers are index entries that appear after
\.{@@:} in the \.{WEB} file.
\yskip\hang |typewriter| identifiers are index entries that appear after
\.{@@.} in the \.{WEB} file.
\yskip\hang |array_like|, |begin_like|, \dots, |var_like|
identifiers are \PASCAL\ reserved words whose |ilk| explains how they are
to be treated when \PASCAL\ code is being formatted.
\yskip\hang Finally, if |c| is an ASCII code, an |ilk| equal to
|char_like+c| denotes a reserved word that will be converted to character
|c|.
@d normal=0 {ordinary identifiers have |normal| ilk}
@d roman=1 {normal index entries have |roman| ilk}
@d wildcard=2 {user-formatted index entries have |wildcard| ilk}
@d typewriter=3 {`typewriter type' entries have |typewriter| ilk}
@d reserved(#)==(ilk[#]>typewriter) {tells if a name is a reserved word}
@d array_like=4 {\&{array}, \&{file}, \&{set}}
@d begin_like=5 {\&{begin}}
@d case_like=6 {\&{case}}
@d const_like=7 {\&{const}, \&{label}, \&{type}}
@d div_like=8 {\&{div}, \&{mod}}
@d do_like=9 {\&{do}, \&{of}, \&{then}}
@d else_like=10 {\&{else}}
@d end_like=11 {\&{end}}
@d for_like=12 {\&{for}, \&{while}, \&{with}}
@d goto_like=13 {\&{goto}, \&{packed}}
@d if_like=14 {\&{if}}
@d in_like=15 {\&{in}}
@d nil_like=16 {\&{nil}}
@d proc_like=17 {\&{function}, \&{procedure}, \&{program}}
@d record_like=18 {\&{record}}
@d repeat_like=19 {\&{repeat}}
@d to_like=20 {\&{downto}, \&{to}}
@d until_like=21 {\&{until}}
@d var_like=22 {\&{var}}
@d loop_like=23 {\&{loop}, \&{xclause}}
@d char_like=24 {\&{and}, \&{or}, \&{not}, \&{in}}
@ The names of modules are stored in |byte_mem| together
with the identifier names, but a hash table is not used for them because
\.{WEAVE} needs to be able to recognize a module name when given a prefix of
that name. A conventional binary seach tree is used to retrieve module names,
with fields called |llink| and |rlink| in place of |link| and |ilk|. The
root of this tree is |rlink[0]|.
@d llink==link {left link in binary search tree for module names}
@d rlink==ilk {right link in binary search tree for module names}
@d root==rlink[0] {the root of the binary search tree for module names}
@<Set init...@>=
root:=0; {the binary search tree starts out with nothing in it}
@ Here is a little procedure that prints the text of a given name on the
user's terminal.
@p procedure print_id(@!p:name_pointer); {print identifier or module name}
var k:0..max_bytes; {index into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
begin if p>=name_ptr then print('IMPOSSIBLE')
else begin w:=p mod ww;
for k:=byte_start[p] to byte_start[p+ww]-1 do
print(xchr[byte_mem[w,k]]);
end;
end;
@ We keep track of the current module number in
|module_count|, which is the total number of modules that have started.
Modules which have been altered by a change file entry
have their |changed_module| flag turned on during the first phase.
@<Globals...@>=
@!module_count:0..max_modules; {the current module number}
@!changed_module: packed array [0..max_modules] of boolean; {is it changed?}
@!change_exists: boolean; {has any module changed?}
@ The other large memory area in \.{WEAVE} keeps the cross-reference data.
All uses of the name |p| are recorded in a linked list beginning at
|xref[p]|, which points into the |xmem| array. Entries in |xmem| consist
of two sixteen-bit items per word, called the |num| and |xlink| fields.
If |x| is an index into |xmem|, reached from name |p|, the value of |num(x)|
is either a module number where |p| is used, or it is |def_flag| plus a
module number where |p| is defined; and |xlink(x)| points to the next such
cross reference for |p|, if any. This list of cross references is in
decreasing order by module number. The current number of cross references
is |xref_ptr|.
The global variable |xref_switch| is set either to |def_flag| or to zero,
depending on whether the next cross reference to an identifier is to be
underlined or not in the index. This switch is set to |def_flag| when
\.{@@!} or \.{@@d} or \.{@@f} is scanned, and it is cleared to zero when
the next identifier or index entry cross reference has been made. Similarly,
the global variable |mod_xref_switch| is either |def_flag| or zero, depending
on whether a module name is being defined or used.
@d num(#)==xmem[#].num_field
@d xlink(#)==xmem[#].xlink_field
@d def_flag=10240 {must be strictly larger than |max_modules|}
@ @<Types...@>=
@!xref_number=0..max_refs;
@ @<Globals...@>=
@!xmem:array[xref_number] of packed record@t@>@/
@!num_field: sixteen_bits; {module number plus zero or |def_flag|}
@!xlink_field: sixteen_bits; {pointer to the previous cross reference}
end;
@!xref_ptr:xref_number; {the largest occupied position in |xmem|}
@!xref_switch,@!mod_xref_switch:0..def_flag; {either zero or |def_flag|}
@ @<Set init...@>=xref_ptr:=0; xref_switch:=0; mod_xref_switch:=0; num(0):=0;
xref[0]:=0; {cross references to undefined modules}
@ A new cross reference for an identifier is formed by calling |new_xref|,
which discards duplicate entries and ignores non-underlined references
to one-letter identifiers or \PASCAL's reserved words.
@d append_xref(#)==if xref_ptr=max_refs then overflow('cross reference')
else begin incr(xref_ptr); num(xref_ptr):=#;
end
@p procedure new_xref(@!p:name_pointer);
label exit;
var q:xref_number; {pointer to previous cross reference}
@!m,@!n: sixteen_bits; {new and previous cross-reference value}
begin if (reserved(p)or(byte_start[p]+1=byte_start[p+ww]))and
(xref_switch=0) then return;
m:=module_count+xref_switch; xref_switch:=0; q:=xref[p];
if q>0 then
begin n:=num(q);
if (n=m)or(n=m+def_flag) then return
else if m=n+def_flag then
begin num(q):=m; return;
end;
end;
append_xref(m); xlink(xref_ptr):=q; xref[p]:=xref_ptr;
exit: end;
@ The cross reference lists for module names are slightly different. Suppose
that a module name is defined in modules $m_1$, \dots, $m_k$ and used in
modules $n_1$, \dots, $n_l$. Then its list will contain $m_1+|def_flag|$,
$m_k+|def_flag|$, \dots, $m_2+|def_flag|$, $n_l$, \dots, $n_1$, in
this order. After Phase II, however, the order will be
$m_1+|def_flag|$, \dots, $m_k+|def_flag|$, $n_1$, \dots, $n_l$.
@p procedure new_mod_xref(@!p:name_pointer);
var q,@!r:xref_number; {pointers to previous cross references}
begin q:=xref[p]; r:=0;
if q>0 then
begin if mod_xref_switch=0 then while num(q)>=def_flag do
begin r:=q; q:=xlink(q);
end
else if num(q)>=def_flag then
begin r:=q; q:=xlink(q);
end;
end;
append_xref(module_count+mod_xref_switch); xlink(xref_ptr):=q;
mod_xref_switch:=0;
if r=0 then xref[p]:=xref_ptr
else xlink(r):=xref_ptr;
end;
@ A third large area of memory is used for sixteen-bit `tokens', which appear
in short lists similar to the strings of characters in |byte_mem|. Token lists
are used to contain the result of \PASCAL\ code translated into \TeX\ form;
further details about them will be explained later. A |text_pointer| variable
is an index into |tok_start|.
@<Types...@>=
@!text_pointer=0..max_texts; {identifies a token list}
@ The first position of |tok_mem|
that is unoccupied by replacement text is called |tok_ptr|, and the first
unused location of |tok_start| is called |text_ptr|.
Thus, we usually have |tok_start[text_ptr]=tok_ptr|.
@<Glob...@>=
@t\hskip1em@>@!tok_mem: packed array [0..max_toks] of sixteen_bits; {tokens}
@t\hskip1em@>@!tok_start: array [text_pointer] of sixteen_bits;
{directory into |tok_mem|}
@t\hskip1em@>@!text_ptr:text_pointer; {first unused position in |tok_start|}
@t\hskip1em@>@!tok_ptr:0..max_toks; {first unused position in |tok_mem|}
stat@!max_tok_ptr,@!max_txt_ptr:0..max_toks; {largest values occurring}
tats
@ @<Set init...@>=
tok_ptr:=1; text_ptr:=1; tok_start[0]:=1; tok_start[1]:=1;
stat max_tok_ptr:=1; max_txt_ptr:=1;@+tats
@* Searching for identifiers.
The hash table described above is updated by the |id_lookup| procedure,
which finds a given identifier and returns a pointer to its index in
|byte_start|. The identifier is supposed to match character by character
and it is also supposed to have a given |ilk| code; the same name may be
present more than once if it is supposed to appear in the index with
different typesetting conventions.
If the identifier was not already present, it is inserted into the table.
Because of the way \.{WEAVE}'s scanning mechanism works, it is most convenient
to let |id_lookup| search for an identifier that is present in the |buffer|
array. Two other global variables specify its position in the buffer: the
first character is |buffer[id_first]|, and the last is |buffer[id_loc-1]|.
@<Glob...@>=
@!id_first:0..long_buf_size; {where the current identifier begins in the buffer}
@!id_loc:0..long_buf_size; {just after the current identifier in the buffer}
@#
@!hash:array [0..hash_size] of sixteen_bits; {heads of hash lists}
@ Initially all the hash lists are empty.
@<Local variables for init...@>=
@!h:0..hash_size; {index into hash-head array}
@ @<Set init...@>=
for h:=0 to hash_size-1 do hash[h]:=0;
@ Here now is the main procedure for finding identifiers (and index
entries). The parameter |t| is set to the desired |ilk| code. The
identifier must either have |ilk=t|, or we must have
|t=normal| and the identifier must be a reserved word.
@p function id_lookup(@!t:eight_bits):name_pointer; {finds current identifier}
label found;
var i:0..long_buf_size; {index into |buffer|}
@!h:0..hash_size; {hash code}
@!k:0..max_bytes; {index into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
@!l:0..long_buf_size; {length of the given identifier}
@!p:name_pointer; {where the identifier is being sought}
begin l:=id_loc-id_first; {compute the length}
@<Compute the hash code |h|@>;
@<Compute the name location |p|@>;
if p=name_ptr then @<Enter a new name into the table at position |p|@>;
id_lookup:=p;
end;
@ A simple hash code is used: If the sequence of
ASCII codes is $c_1c_2\ldots c_m$, its hash value will be
$$(2^{n-1}c_1+2^{n-2}c_2+\cdots+c_n)\,\bmod\,|hash_size|.$$
@<Compute the hash...@>=
h:=buffer[id_first]; i:=id_first+1;
while i<id_loc do
begin h:=(h+h+buffer[i]) mod hash_size; incr(i);
end
@ If the identifier is new, it will be placed in position |p=name_ptr|,
otherwise |p| will point to its existing location.
@<Compute the name location...@>=
p:=hash[h];
while p<>0 do
begin if (length(p)=l)and((ilk[p]=t)or((t=normal)and reserved(p))) then
@<Compare name |p| with current identifier,
|goto found| if equal@>;
p:=link[p];
end;
p:=name_ptr; {the current identifier is new}
link[p]:=hash[h]; hash[h]:=p; {insert |p| at beginning of hash list}
found:
@ @<Compare name |p|...@>=
begin i:=id_first; k:=byte_start[p]; w:=p mod ww;
while (i<id_loc)and(buffer[i]=byte_mem[w,k]) do
begin incr(i); incr(k);
end;
if i=id_loc then goto found; {all characters agree}
end
@ When we begin the following segment of the program, |p=name_ptr|.
@<Enter a new name...@>=
begin w:=name_ptr mod ww;
if byte_ptr[w]+l>max_bytes then overflow('byte memory');
if name_ptr+ww>max_names then overflow('name');
i:=id_first; k:=byte_ptr[w]; {get ready to move the identifier into |byte_mem|}
while i<id_loc do
begin byte_mem[w,k]:=buffer[i]; incr(k); incr(i);
end;
byte_ptr[w]:=k; byte_start[name_ptr+ww]:=k; incr(name_ptr);
ilk[p]:=t; xref[p]:=0;
end
@* Initializing the table of reserved words.
We have to get \PASCAL's reserved words into the hash table, and the
simplest way to do this is to insert them every time \.{WEAVE} is run.
A few macros permit us to do the initialization with a compact program.
@d sid9(#)==buffer[9]:=#;cur_name:=id_lookup
@d sid8(#)==buffer[8]:=#;sid9
@d sid7(#)==buffer[7]:=#;sid8
@d sid6(#)==buffer[6]:=#;sid7
@d sid5(#)==buffer[5]:=#;sid6
@d sid4(#)==buffer[4]:=#;sid5
@d sid3(#)==buffer[3]:=#;sid4
@d sid2(#)==buffer[2]:=#;sid3
@d sid1(#)==buffer[1]:=#;sid2
@d id2==id_first:=8; sid8
@d id3==id_first:=7; sid7
@d id4==id_first:=6; sid6
@d id5==id_first:=5; sid5
@d id6==id_first:=4; sid4
@d id7==id_first:=3; sid3
@d id8==id_first:=2; sid2
@d id9==id_first:=1; sid1
@<Globals...@>=
@!cur_name:name_pointer; {points to the identifier just inserted}
@ The intended use of the macros above might not be immediately obvious,
but the riddle is answered by the following:
@<Store all the reserved words@>=
id_loc:=10;@/
id3("a")("n")("d")(char_like+and_sign);@/
id5("a")("r")("r")("a")("y")(array_like);@/
id5("b")("e")("g")("i")("n")(begin_like);@/
id4("c")("a")("s")("e")(case_like);@/
id5("c")("o")("n")("s")("t")(const_like);@/
id3("d")("i")("v")(div_like);@/
id2("d")("o")(do_like);@/
id6("d")("o")("w")("n")("t")("o")(to_like);@/
id4("e")("l")("s")("e")(else_like);@/
id3("e")("n")("d")(end_like);@/
id4("f")("i")("l")("e")(array_like);@/
id3("f")("o")("r")(for_like);@/
id8("f")("u")("n")("c")("t")("i")("o")("n")(proc_like);@/
id4("g")("o")("t")("o")(goto_like);@/
id2("i")("f")(if_like);@/
id2("i")("n")(char_like+set_element_sign);@/
id5("l")("a")("b")("e")("l")(const_like);@/
id3("m")("o")("d")(div_like);@/
id3("n")("i")("l")(nil_like);@/
id3("n")("o")("t")(char_like+not_sign);@/
id2("o")("f")(do_like);@/
id2("o")("r")(char_like+or_sign);@/
id6("p")("a")("c")("k")("e")("d")(goto_like);@/
id9("p")("r")("o")("c")("e")("d")("u")("r")("e")(proc_like);@/
id7("p")("r")("o")("g")("r")("a")("m")(proc_like);@/
id6("r")("e")("c")("o")("r")("d")(record_like);@/
id6("r")("e")("p")("e")("a")("t")(repeat_like);@/
id3("s")("e")("t")(array_like);@/
id4("t")("h")("e")("n")(do_like);@/
id2("t")("o")(to_like);@/
id4("t")("y")("p")("e")(const_like);@/
id5("u")("n")("t")("i")("l")(until_like);@/
id3("v")("a")("r")(var_like);@/
id5("w")("h")("i")("l")("e")(for_like);@/
id4("w")("i")("t")("h")(for_like);@/
id7("x")("c")("l")("a")("u")("s")("e")(loop_like);@/
@* Searching for module names.
The |mod_lookup| procedure finds the module name |mod_text[1..l]| in the
search tree, after inserting it if necessary, and returns a pointer to
where it was found.
@<Glob...@>=
@!mod_text:array [0..longest_name] of ASCII_code; {name being sought for}
@ According to the rules of \.{WEB}, no module name
should be a proper prefix of another, so a ``clean'' comparison should
occur between any two names. The result of |mod_lookup| is 0 if this
prefix condition is violated. An error message is printed when such violations
are detected during phase two of \.{WEAVE}.
@d less=0 {the first name is lexicographically less than the second}
@d equal=1 {the first name is equal to the second}
@d greater=2 {the first name is lexicographically greater than the second}
@d prefix=3 {the first name is a proper prefix of the second}
@d extension=4 {the first name is a proper extension of the second}
@p function mod_lookup(@!l:sixteen_bits):name_pointer; {finds module name}
label found;
var c:less..extension; {comparison between two names}
@!j:0..longest_name; {index into |mod_text|}
@!k:0..max_bytes; {index into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
@!p:name_pointer; {current node of the search tree}
@!q:name_pointer; {father of node |p|}
begin c:=greater; q:=0; p:=root;
while p<>0 do
begin @<Set variable |c| to the result of comparing the given name
to name |p|@>;
q:=p;
if c=less then p:=llink[q]
else if c=greater then p:=rlink[q]
else goto found;
end;
@<Enter a new module name into the tree@>;
found: if c<>equal then
begin err_print('! Incompatible section names'); p:=0;
@.Incompatible section names@>
end;
mod_lookup:=p;
end;
@ @<Enter a new module name...@>=
w:=name_ptr mod ww; k:=byte_ptr[w];
if k+l>max_bytes then overflow('byte memory');
if name_ptr>max_names-ww then overflow('name');
p:=name_ptr;
if c=less then llink[q]:=p else rlink[q]:=p;
llink[p]:=0; rlink[p]:=0; xref[p]:=0; c:=equal;
for j:=1 to l do byte_mem[w,k+j-1]:=mod_text[j];
byte_ptr[w]:=k+l; byte_start[name_ptr+ww]:=k+l; incr(name_ptr);
@ @<Set variable |c|...@>=
begin k:=byte_start[p]; w:=p mod ww; c:=equal; j:=1;
while (k<byte_start[p+ww]) and (j<=l) and (mod_text[j]=byte_mem[w,k]) do
begin incr(k); incr(j);
end;
if k=byte_start[p+ww] then
if j>l then c:=equal
else c:=extension
else if j>l then c:=prefix
else if mod_text[j]<byte_mem[w,k] then c:=less
else c:=greater;
end
@ The |prefix_lookup| procedure is supposed to find exactly one module
name that has |mod_text[1..l]| as a prefix. Actually the algorithm
silently accepts also the situation that some module name is a prefix of
|mod_text[1..l]|, because the user who painstakingly typed in more than
necessary probably doesn't want to be told about the wasted effort.
Recall that error messages are not printed during phase one. It is
possible that the |prefix_lookup| procedure will fail on the first pass,
because there is no match, yet the second pass might detect no error if a
matching module name has occurred after the offending prefix. In such a
case the cross-reference information will be incorrect and \.{WEAVE} will
report no error. However, such a mistake will be detected by the
\.{TANGLE} processor.
@p function prefix_lookup(@!l:sixteen_bits):name_pointer; {finds name extension}
var c:less..extension; {comparison between two names}
@!count:0..max_names; {the number of hits}
@!j:0..longest_name; {index into |mod_text|}
@!k:0..max_bytes; {index into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
@!p:name_pointer; {current node of the search tree}
@!q:name_pointer; {another place to resume the search after one branch is done}
@!r:name_pointer; {extension found}
begin q:=0; p:=root; count:=0; r:=0; {begin search at root of tree}
while p<>0 do
begin @<Set variable |c| to the result of comparing...@>;
if c=less then p:=llink[p]
else if c=greater then p:=rlink[p]
else begin r:=p; incr(count); q:=rlink[p]; p:=llink[p];
end;
if p=0 then
begin p:=q; q:=0;
end;
end;
if count<>1 then
if count=0 then err_print('! Name does not match')
@.Name does not match@>
else err_print('! Ambiguous prefix');
@.Ambiguous prefix@>
prefix_lookup:=r; {the result will be 0 if there was no match}
end;
@* Lexical scanning.
Let us now consider the subroutines that read the \.{WEB} source file
and break it into meaningful units. There are four such procedures:
One simply skips to the next `\.{@@\ }' or `\.{@@*}' that begins a
module; another passes over the \TeX\ text at the beginning of a
module; the third passes over the \TeX\ text in a \PASCAL\ comment;
and the last, which is the most interesting, gets the next token of
a \PASCAL\ text.
@ But first we need to consider the low-level routine |get_line|
that takes care of merging |change_file| into |web_file|. The |get_line|
procedure also updates the line numbers for error messages.
@<Globals...@>=
@!ii:integer; {general purpose |for| loop variable in the outer block}
@!line:integer; {the number of the current line in the current file}
@!other_line:integer; {the number of the current line in the input file that
is not currently being read}
@!temp_line:integer; {used when interchanging |line| with |other_line|}
@!limit:0..long_buf_size; {the last character position occupied in the buffer}
@!loc:0..long_buf_size; {the next character position to be read from the buffer}
@!input_has_ended: boolean; {if |true|, there is no more input}
@!changing: boolean; {if |true|, the current line is from |change_file|}
@!change_pending: boolean; {if |true|, the current change is not yet
recorded in |changed_module[module_count]|}
@ As we change |changing| from |true| to |false| and back again, we must
remember to swap the values of |line| and |other_line| so that the |err_print|
routine will be sure to report the correct line number.
@d change_changing==
changing := not changing;
temp_line:=other_line; other_line:=line; line:=temp_line
{|line @t$\null\BA\null$@> other_line|}
@ When |changing| is |false|, the next line of |change_file| is kept in
|change_buffer[0..change_limit]|, for purposes of comparison with the next
line of |web_file|. After the change file has been completely input, we
set |change_limit:=0|, so that no further matches will be made.
@<Globals...@>=
@!change_buffer:array[0..buf_size] of ASCII_code;
@!change_limit:0..buf_size; {the last position occupied in |change_buffer|}
@ Here's a simple function that checks if the two buffers are different.
@p function lines_dont_match:boolean;
label exit;
var k:0..buf_size; {index into the buffers}
begin lines_dont_match:=true;
if change_limit<>limit then return;
if limit>0 then
for k:=0 to limit-1 do if change_buffer[k]<>buffer[k] then return;
lines_dont_match:=false;
exit: end;
@ Procedure |prime_the_change_buffer| sets |change_buffer| in preparation
for the next matching operation. Since blank lines in the change file are
not used for matching, we have |(change_limit=0)and not changing| if and
only if the change file is exhausted. This procedure is called only
when |changing| is true; hence error messages will be reported correctly.
@p procedure prime_the_change_buffer;
label continue, done, exit;
var k:0..buf_size; {index into the buffers}
begin change_limit:=0; {this value will be used if the change file ends}
@<Skip over comment lines in the change file; |return| if end of file@>;
@<Skip to the next nonblank line; |return| if end of file@>;
@<Move |buffer| and |limit| to |change_buffer| and |change_limit|@>;
exit: end;
@ While looking for a line that begins with \.{@@x} in the change file,
we allow lines that begin with \.{@@}, as long as they don't begin with
\.{@@y} or \.{@@z} (which would probably indicate that the change file is
fouled up).
@<Skip over comment lines in the change file...@>=
loop@+ begin incr(line);
if not input_ln(change_file) then return;
if limit<2 then goto continue;
if buffer[0]<>"@@" then goto continue;
if (buffer[1]>="X")and(buffer[1]<="Z") then
buffer[1]:=buffer[1]+"z"-"Z"; {lowercasify}
if buffer[1]="x" then goto done;
if (buffer[1]="y")or(buffer[1]="z") then
begin loc:=2; err_print('! Where is the matching @@x?');
@.Where is the match...@>
end;
continue: end;
done:
@ Here we are looking at lines following the \.{@@x}.
@<Skip to the next nonblank line...@>=
repeat incr(line);
if not input_ln(change_file) then
begin err_print('! Change file ended after @@x');
@.Change file ended...@>
return;
end;
until limit>0;
@ @<Move |buffer| and |limit| to |change_buffer| and |change_limit|@>=
begin change_limit:=limit;
if limit>0 then for k:=0 to limit-1 do change_buffer[k]:=buffer[k];
end
@ The following procedure is used to see if the next change entry should
go into effect; it is called only when |changing| is false.
The idea is to test whether or not the current
contents of |buffer| matches the current contents of |change_buffer|.
If not, there's nothing more to do; but if so, a change is called for:
All of the text down to the \.{@@y} is supposed to match. An error
message is issued if any discrepancy is found. Then the procedure
prepares to read the next line from |change_file|.
When a match is found, the current module is marked as changed unless
the first line after the \.{@@x} and after the \.{@@y} both start with
either |'@@*'| or |'@@ '| (possibly preceded by whitespace).
@d if_module_start_then_make_change_pending(#)==
loc:=0; buffer[limit]:="!";
while (buffer[loc]=" ")or(buffer[loc]=tab_mark) do incr(loc);
buffer[limit]:=" ";
if buffer[loc]="@@" then
if (buffer[loc+1]="*") or
(buffer[loc+1]=" ") or (buffer[loc+1]=tab_mark) then
change_pending:=#
@p procedure check_change; {switches to |change_file| if the buffers match}
label exit;
var n:integer; {the number of discrepancies found}
@!k:0..buf_size; {index into the buffers}
begin if lines_dont_match then return;
change_pending:=false;
if not changed_module[module_count] then
begin if_module_start_then_make_change_pending(true);
if not change_pending then changed_module[module_count]:=true;
end;
n:=0;
loop@+ begin change_changing; {now it's |true|}
incr(line);
if not input_ln(change_file) then
begin err_print('! Change file ended before @@y');
@.Change file ended...@>
change_limit:=0; change_changing; {|false| again}
return;
end;
@<If the current line starts with \.{@@y},
report any discrepancies and |return|@>;
@<Move |buffer| and |limit|...@>;
change_changing; {now it's |false|}
incr(line);
if not input_ln(web_file) then
begin err_print('! WEB file ended during a change');
@.WEB file ended...@>
input_has_ended:=true; return;
end;
if lines_dont_match then incr(n);
end;
exit: end;
@ @<If the current line starts with \.{@@y}...@>=
if limit>1 then if buffer[0]="@@" then
begin if (buffer[1]>="X")and(buffer[1]<="Z") then
buffer[1]:=buffer[1]+"z"-"Z"; {lowercasify}
if (buffer[1]="x")or(buffer[1]="z") then
begin loc:=2; err_print('! Where is the matching @@y?');
@.Where is the match...@>
end
else if buffer[1]="y" then
begin if n>0 then
begin loc:=2; err_print('! Hmm... ',n:1,
' of the preceding lines failed to match');
@.Hmm... n of the preceding...@>
end;
return;
end;
end
@ The |reset_input| procedure, which gets \.{WEAVE} ready to read the
user's \.{WEB} input, is used at the beginning of phases one and two.
@p procedure reset_input;
begin open_input; line:=0; other_line:=0;@/
changing:=true; prime_the_change_buffer; change_changing;@/
limit:=0; loc:=1; buffer[0]:=" "; input_has_ended:=false;
end;
@ The |get_line| procedure is called when |loc>limit|; it puts the next
line of merged input into the buffer and updates the other variables
appropriately. A space is placed at the right end of the line.
@p procedure get_line; {inputs the next line}
label restart;
begin restart:if changing then
@<Read from |change_file| and maybe turn off |changing|@>;
if not changing then
begin @<Read from |web_file| and maybe turn on |changing|@>;
if changing then goto restart;
end;
loc:=0; buffer[limit]:=" ";
end;
@ @<Read from |web_file|...@>=
begin incr(line);
if not input_ln(web_file) then input_has_ended:=true
else if limit=change_limit then
if buffer[0]=change_buffer[0] then
if change_limit>0 then check_change;
end
@ @<Read from |change_file|...@>=
begin incr(line);
if not input_ln(change_file) then
begin err_print('! Change file ended without @@z');
@.Change file ended...@>
buffer[0]:="@@"; buffer[1]:="z"; limit:=2;
end;
if limit>0 then {check if the change has ended}
begin if change_pending then
begin if_module_start_then_make_change_pending(false);
if change_pending then
begin changed_module[module_count]:=true; change_pending:=false;
end;
end;
buffer[limit]:=" ";
if buffer[0]="@@" then
begin if (buffer[1]>="X")and(buffer[1]<="Z") then
buffer[1]:=buffer[1]+"z"-"Z"; {lowercasify}
if (buffer[1]="x")or(buffer[1]="y") then
begin loc:=2; err_print('! Where is the matching @@z?');
@.Where is the match...@>
end
else if buffer[1]="z" then
begin prime_the_change_buffer; change_changing;
end;
end;
end;
end
@ At the end of the program, we will tell the user if the change file
had a line that didn't match any relevant line in |web_file|.
@<Check that all changes have been read@>=
if change_limit<>0 then {|changing| is false}
begin for ii:=0 to change_limit do buffer[ii]:=change_buffer[ii];
limit:=change_limit; changing:=true; line:=other_line; loc:=change_limit;
err_print('! Change file entry did not match');
@.Change file entry did not match@>
end
@ Control codes in \.{WEB}, which begin with `\.{@@}', are converted
into a numeric code designed to simplify \.{WEAVE}'s logic; for example,
larger numbers are given to the control codes that denote more significant
milestones, and the code of |new_module| should be the largest of
all. Some of these numeric control codes take the place of ASCII
control codes that will not otherwise appear in the output of the
scanning routines.
@^ASCII code@>
@d ignore=0 {control code of no interest to \.{WEAVE}}
@d verbatim=@'2 {extended ASCII alpha will not appear}
@d force_line=@'3 {extended ASCII beta will not appear}
@d begin_comment=@'11 {ASCII tab mark will not appear}
@d end_comment=@'12 {ASCII line feed will not appear}
@d octal=@'14 {ASCII form feed will not appear}
@d hex=@'15 {ASCII carriage return will not appear}
@d double_dot=@'40 {ASCII space will not appear except in strings}
@d no_underline=@'175 {this code will be intercepted without confusion}
@d underline=@'176 {this code will be intercepted without confusion}
@d param=@'177 {ASCII delete will not appear}
@d xref_roman=@'203 {control code for `\.{@@\^}'}
@d xref_wildcard=@'204 {control code for `\.{@@:}'}
@d xref_typewriter=@'205 {control code for `\.{@@.}'}
@d TeX_string=@'206 {control code for `\.{@@t}'}
@d check_sum=@'207 {control code for `\.{@@\$}'}
@d join=@'210 {control code for `\.{@@\&}'}
@d thin_space=@'211 {control code for `\.{@@,}'}
@d math_break=@'212 {control code for `\.{@@\char'174}'}
@d line_break=@'213 {control code for `\.{@@/}'}
@d big_line_break=@'214 {control code for `\.{@@\#}'}
@d no_line_break=@'215 {control code for `\.{@@+}'}
@d pseudo_semi=@'216 {control code for `\.{@@;}'}
@d format=@'217 {control code for `\.{@@f}'}
@d definition=@'220 {control code for `\.{@@d}'}
@d begin_Pascal=@'221 {control code for `\.{@@p}'}
@d module_name=@'222 {control code for `\.{@@<}'}
@d new_module=@'223 {control code for `\.{@@\ }' and `\.{@@*}'}
@ Control codes are converted from ASCII to \.{WEAVE}'s internal
representation by the |control_code| routine.
@p function control_code(@!c:ASCII_code):eight_bits; {convert |c|
after \.{@@}}
begin case c of
"@@": control_code:="@@"; {`quoted' at sign}
"'": control_code:=octal; {precedes octal constant}
"""": control_code:=hex; {precedes hexadecimal constant}
"$": control_code:=check_sum; {precedes check sum constant}
" ",tab_mark,"*": control_code:=new_module; {beginning of a new module}
"=": control_code:=verbatim;
"\": control_code:=force_line;
"D","d": control_code:=definition; {macro definition}
"F","f": control_code:=format; {format definition}
"{": control_code:=begin_comment; {begin-comment delimiter}
"}": control_code:=end_comment; {end-comment delimiter}
"P","p": control_code:=begin_Pascal; {\PASCAL\ text in unnamed module}
"&": control_code:=join; {concatenate two tokens}
"<": control_code:=module_name; {beginning of a module name}
">": begin err_print('! Extra @@>'); control_code:=ignore;
@.Extra \AT!>@>
end; {end of module name should not be discovered in this way}
"T","t": control_code:=TeX_string; {\TeX\ box within \PASCAL}
"!": control_code:=underline; {set definition flag}
"?": control_code:=no_underline; {reset definition flag}
"^": control_code:=xref_roman; {index entry to be typeset normally}
":": control_code:=xref_wildcard; {index entry to be in user format}
".": control_code:=xref_typewriter; {index entry to be in typewriter type}
",": control_code:=thin_space; {puts extra space in \PASCAL\ format}
"|": control_code:=math_break; {allows a break in a formula}
"/": control_code:=line_break; {forces end-of-line in \PASCAL\ format}
"#": control_code:=big_line_break; {forces end-of-line and some space besides}
"+": control_code:=no_line_break; {cancels end-of-line down to single space}
";": control_code:=pseudo_semi; {acts like a semicolon, but is invisible}
@t\4@>@<Special control codes allowed only when debugging@>@;
othercases begin err_print('! Unknown control code'); control_code:=ignore;
@.Unknown control code@>
end
endcases;
end;
@ If \.{WEAVE} is compiled with debugging commands, one can write
\.{@@2}, \.{@@1}, and \.{@@0} to turn tracing fully on, partly on,
and off, respectively.
@.\AT!2@>
@.\AT!1@>
@<Special control codes...@>=
@!debug@t@>@/
"0","1","2": begin tracing:=c-"0"; control_code:=ignore;
end;
gubed
@ The |skip_limbo| routine is used on the first pass to skip through
portions of the input that are not in any modules, i.e., that precede
the first module. After this procedure has been called, the value of
|input_has_ended| will tell whether or not a new module has
actually been found.
@p procedure skip_limbo; {skip to next module}
label exit;
var c:ASCII_code; {character following \.{@@}}
begin loop if loc>limit then
begin get_line;
if input_has_ended then return;
end
else begin buffer[limit+1]:="@@";
while buffer[loc]<>"@@" do incr(loc);
if loc<=limit then
begin loc:=loc+2; c:=buffer[loc-1];
if (c=" ")or(c=tab_mark)or(c="*") then return;
end;
end;
exit: end;
@ The |skip_TeX| routine is used on the first pass to skip through
the \TeX\ code at the beginning of a module. It returns the next
control code or `\v' found in the input. A |new_module| is
assumed to exist at the very end of the file.
@p function skip_TeX: eight_bits; {skip past pure \TeX\ code}
label done;
var c:eight_bits; {control code found}
begin loop begin if loc>limit then
begin get_line;
if input_has_ended then
begin c:=new_module; goto done;
end;
end;
buffer[limit+1]:="@@";
repeat c:=buffer[loc]; incr(loc);
if c="|" then goto done;
until c="@@";
if loc<=limit then
begin c:=control_code(buffer[loc]); incr(loc); goto done;
end;
end;
done:skip_TeX:=c;
end;
@ The |skip_comment| routine is used on the first pass to skip
through \TeX\ code in \PASCAL\ comments. The |bal| parameter
tells how many left braces are assumed to have been scanned when
this routine is called, and the procedure returns a corresponding
value of |bal| at the point that scanning has stopped. Scanning
stops either at a `\v' that introduces \PASCAL\ text,
in which case the returned value is positive, or it stops at the
end of the comment, in which case the returned value is zero.
The scanning also stops in anomalous situations when the comment
doesn't end or when it contains an illegal use of \.{@@}.
One should call |skip_comment(1)| when beginning to scan a comment.
@p function skip_comment(@!bal:eight_bits):eight_bits; {skips \TeX\
code in comments}
label done;
var c:ASCII_code; {the current character}
begin loop begin if loc>limit then
begin get_line;
if input_has_ended then
begin bal:=0; goto done;
end; {an error message will occur in phase two}
end;
c:=buffer[loc]; incr(loc);
if c="|" then goto done;
@<Do special things when |c="@@", "\", "{", "}"|; |goto done| at end@>;
end;
done: skip_comment:=bal;
end;
@ @<Do special things when |c="@@"...@>=
if c="@@" then
begin c:=buffer[loc];
if (c<>" ")and(c<>tab_mark)and(c<>"*") then incr(loc)
else begin decr(loc); bal:=0; goto done;
end {an error message will occur in phase two}
end
else if (c="\")and(buffer[loc]<>"@@") then incr(loc)
else if c="{" then incr(bal)
else if c="}" then
begin decr(bal);
if bal=0 then goto done;
end
@* Inputting the next token.
As stated above, \.{WEAVE}'s most interesting lexical scanning routine is the
|get_next| function that inputs the next token of \PASCAL\ input. However,
|get_next| is not especially complicated.
The result of |get_next| is either an ASCII code for some special character,
or it is a special code representing a pair of characters (e.g., `\.{:=}'
or `\.{..}'), or it is the numeric value computed by the |control_code|
procedure, or it is one of the following special codes:
\yskip\hang |exponent|: The `\.E' in a real constant.
\yskip\hang |identifier|: In this case the global variables |id_first|
and |id_loc| will have been set to the appropriate values needed by the
|id_lookup| routine.
\yskip\hang |string|: In this case the global variables |id_first| and
|id_loc| will have been set to the beginning and ending-plus-one locations
in the buffer. The string ends with the first reappearance of its initial
delimiter; thus, for example, $$\.{\'This isn\'\'t a single string\'}$$
will be treated as two consecutive strings, the first being \.{\'This
isn\'}.
\yskip\noindent Furthermore, some of the control codes cause
|get_next| to take additional actions:
\yskip\hang |xref_roman|, |xref_wildcard|,
|xref_typewriter|, |TeX_string|: The values of
|id_first| and |id_loc| will be set so that the string in question appears
in |buffer[id_first..(id_loc-1)]|.
\yskip\hang |module_name|: In this case the global variable |cur_module| will
point to the |byte_start| entry for the module name that has just been scanned.
\yskip\noindent If |get_next| sees `\.{@@!}' or `\.{@@?}',
it sets |xref_switch| to |def_flag| or zero and goes on to the next token.
A global variable called |scanning_hex| is set |true| during the time that
the letters \.A through \.F should be treated as if they were digits.
@d exponent=@'200 {\.E or \.e following a digit}
@d string=@'201 {\PASCAL\ string or \.{WEB} precomputed string}
@d identifier=@'202 {\PASCAL\ identifier or reserved word}
@<Globals...@>=
@!cur_module: name_pointer; {name of module just scanned}
@!scanning_hex: boolean; {are we scanning a hexadecimal constant?}
@ @<Set init...@>=
scanning_hex:=false;
@ As one might expect, |get_next| consists mostly of a big switch
that branches to the various special cases that can arise.
@d up_to(#)==#-24,#-23,#-22,#-21,#-20,#-19,#-18,#-17,#-16,#-15,#-14,
#-13,#-12,#-11,#-10,#-9,#-8,#-7,#-6,#-5,#-4,#-3,#-2,#-1,#
@p function get_next:eight_bits; {produces the next input token}
label restart,done,found;
var c:eight_bits; {the current character}
@!d:eight_bits; {the next character}
@!j,@!k:0..longest_name; {indices into |mod_text|}
begin restart: if loc>limit then
begin get_line;
if input_has_ended then
begin c:=new_module; goto found;
end;
end;
c:=buffer[loc]; incr(loc);
if scanning_hex then @<Go to |found| if |c| is a hexadecimal digit,
otherwise set |scanning_hex:=false|@>;
case c of
"A",up_to("Z"),"a",up_to("z"): @<Get an identifier@>;
"'","""": @<Get a string@>;
"@@": @<Get control code and possible module name@>;
@t\4@>@<Compress two-symbol combinations like `\.{:=}'@>@;
" ",tab_mark: goto restart; {ignore spaces and tabs}
"}": begin err_print('! Extra }'); goto restart;
@.Extra \}@>
end;
othercases if c>=128 then goto restart {ignore nonstandard characters}
else do_nothing
endcases;
found:@!debug if trouble_shooting then debug_help;@;@+gubed@/
get_next:=c;
end;
@ @<Go to |found| if |c| is a hexadecimal digit...@>=
if ((c>="0")and(c<="9"))or((c>="A")and(c<="F")) then goto found
else scanning_hex:=false
@ Note that the following code substitutes \.{@@\{} and \.{@@\}} for the
respective combinations `\.{(*}' and `\.{*)}'. Explicit braces should be used
for \TeX\ comments in \PASCAL\ text.
@d compress(#)==begin if loc<=limit then begin c:=#; incr(loc); end; end
@<Compress two-symbol...@>=
".": if buffer[loc]="." then compress(double_dot)
else if buffer[loc]=")" then compress("]");
":": if buffer[loc]="=" then compress(left_arrow);
"=": if buffer[loc]="=" then compress(equivalence_sign);
">": if buffer[loc]="=" then compress(greater_or_equal);
"<": if buffer[loc]="=" then compress(less_or_equal)
else if buffer[loc]=">" then compress(not_equal);
"(": if buffer[loc]="*" then compress(begin_comment)
else if buffer[loc]="." then compress("[");
"*": if buffer[loc]=")" then compress(end_comment);
@ @<Get an identifier@>=
begin if ((c="E")or(c="e"))and(loc>1) then
if (buffer[loc-2]<="9")and(buffer[loc-2]>="0") then c:=exponent;
if c<>exponent then
begin decr(loc); id_first:=loc;
repeat incr(loc); d:=buffer[loc];
until ((d<"0")or((d>"9")and(d<"A"))or((d>"Z")and(d<"a"))or(d>"z"))and(d<>"_");
c:=identifier; id_loc:=loc;
end;
end
@ A string that starts and ends with single or double quote marks is
scanned by the following piece of the program.
@<Get a string@>=
begin id_first:=loc-1;
repeat d:=buffer[loc]; incr(loc);
if loc>limit then
begin err_print('! String constant didn''t end');
@.String constant didn't end@>
loc:=limit; d:=c;
end;
until d=c;
id_loc:=loc; c:=string;
end
@ After an \.{@@} sign has been scanned, the next character tells us
whether there is more work to do.
@<Get control code and possible module name@>=
begin c:=control_code(buffer[loc]); incr(loc);
if c=underline then
begin xref_switch:=def_flag; goto restart;
end
else if c=no_underline then
begin xref_switch:=0; goto restart;
end
else if (c<=TeX_string)and(c>=xref_roman) then
@<Scan to the next \.{@@>}@>
else if c=hex then scanning_hex:=true
else if c=module_name then
@<Scan the module name and make |cur_module| point to it@>
else if c=verbatim then @<Scan a verbatim string@>;
end
@ The occurrence of a module name sets |xref_switch| to zero,
because the module name might (for example) follow \&{var}.
@<Scan the module name...@>=
begin @<Put module name into |mod_text[1..k]|@>;
if k>3 then
begin if (mod_text[k]=".")and(mod_text[k-1]=".")and(mod_text[k-2]=".") then
cur_module:=prefix_lookup(k-3)
else cur_module:=mod_lookup(k);
end
else cur_module:=mod_lookup(k);
xref_switch:=0;
end
@ Module names are placed into the |mod_text| array with consecutive spaces,
tabs, and carriage-returns replaced by single spaces. There will be no
spaces at the beginning or the end. (We set |mod_text[0]:=" "| to facilitate
this, since the |mod_lookup| routine uses |mod_text[1]| as the first
character of the name.)
@<Set init...@>=mod_text[0]:=" ";
@ @<Put module name...@>=
k:=0;
loop@+ begin if loc>limit then
begin get_line;
if input_has_ended then
begin err_print('! Input ended in section name');
@.Input ended in section name@>
loc:=1; goto done;
end;
end;
d:=buffer[loc];
@<If end of name, |goto done|@>;
incr(loc); if k<longest_name-1 then incr(k);
if (d=" ")or(d=tab_mark) then
begin d:=" "; if mod_text[k-1]=" " then decr(k);
end;
mod_text[k]:=d;
end;
done: @<Check for overlong name@>;
if (mod_text[k]=" ")and(k>0) then decr(k)
@ @<If end of name,...@>=
if d="@@" then
begin d:=buffer[loc+1];
if d=">" then
begin loc:=loc+2; goto done;
end;
if (d=" ")or(d=tab_mark)or(d="*") then
begin err_print('! Section name didn''t end'); goto done;
@.Section name didn't end@>
end;
incr(k); mod_text[k]:="@@"; incr(loc); {now |d=buffer[loc]| again}
end
@ @<Check for overlong name@>=
if k>=longest_name-2 then
begin print_nl('! Section name too long: ');
@.Section name too long@>
for j:=1 to 25 do print(xchr[mod_text[j]]);
print('...'); mark_harmless;
end
@ @<Scan to the next...@>=
begin id_first:=loc; buffer[limit+1]:="@@";
while buffer[loc]<>"@@" do incr(loc);
id_loc:=loc;
if loc>limit then
begin err_print('! Control text didn''t end'); loc:=limit;
@.Control text didn't end@>
end
else begin loc:=loc+2;
if buffer[loc-1]<>">" then
err_print('! Control codes are forbidden in control text');
@.Control codes are forbidden...@>
end;
end
@ A verbatim \PASCAL\ string will be treated like ordinary strings, but
with no surrounding delimiters. At the present point in the program we
have |buffer[loc-1]=verbatim|; we must set |id_first| to the beginning
of the string itself, and |id_loc| to its ending-plus-one location in the
buffer. We also set |loc| to the position just after the ending delimiter.
@<Scan a verbatim string@>=
begin id_first:=loc; incr(loc);
buffer[limit+1]:="@@"; buffer[limit+2]:=">";
while (buffer[loc]<>"@@")or(buffer[loc+1]<>">") do incr(loc);
if loc>=limit then err_print('! Verbatim string didn''t end');
@.Verbatim string didn't end@>
id_loc:=loc; loc:=loc+2;
end
@* Phase one processing.
We now have accumulated enough subroutines to make it possible to carry out
\.{WEAVE}'s first pass over the source file. If everything works right,
both phase one and phase two of \.{WEAVE} will assign the same numbers to
modules, and these numbers will agree with what \.{TANGLE} does.
The global variable |next_control| often contains the most recent output of
|get_next|; in interesting cases, this will be the control code that
ended a module or part of a module.
@<Glob...@>=@!next_control:eight_bits; {control code waiting to be acting upon}
@ The overall processing strategy in phase one has the following
straightforward outline.
@<Phase I: Read all the user's text and store the cross references@>=
phase_one:=true; phase_three:=false;
reset_input;
module_count:=0; skip_limbo; change_exists:=false;
while not input_has_ended do
@<Store cross reference data for the current module@>;
changed_module[module_count]:=change_exists;
{the index changes if anything does}
phase_one:=false; {prepare for second phase}
@<Print error messages about unused or undefined module names@>;
@ @<Store cross reference data...@>=
begin incr(module_count);
if module_count=max_modules then overflow('section number');
changed_module[module_count]:=changing;
{it will become |true| if any line changes}
if buffer[loc-1]="*" then
begin print('*',module_count:1);
update_terminal; {print a progress report}
end;
@<Store cross references in the \TeX\ part of a module@>;
@<Store cross references in the \(definition part of a module@>;
@<Store cross references in the \PASCAL\ part of a module@>;
if changed_module[module_count] then change_exists:=true;
end
@ The |Pascal_xref| subroutine stores references to identifiers in
\PASCAL\ text material beginning with the current value of |next_control|
and continuing until |next_control| is `\.\{' or `\v', or until the next
``milestone'' is passed (i.e., |next_control>=format|). If
|next_control>=format| when |Pascal_xref| is called, nothing will happen;
but if |next_control="|"| upon entry, the procedure assumes that this is
the `\v' preceding \PASCAL\ text that is to be processed.
The program uses the fact that our internal code numbers satisfy
the relations |xref_roman=identifier+roman| and |xref_wildcard=identifier
+wildcard| and |xref_typewriter=identifier+
typewriter| and |normal=0|. An implied `\.{@@!}' is inserted after
\&{function}, \&{procedure}, \&{program}, and \&{var}.
@p procedure Pascal_xref; {makes cross references for \PASCAL\ identifiers}
label exit;
var p:name_pointer; {a referenced name}
begin while next_control<format do
begin if (next_control>=identifier)and
(next_control<=xref_typewriter) then
begin p:=id_lookup(next_control-identifier); new_xref(p);
if (ilk[p]=proc_like)or(ilk[p]=var_like) then
xref_switch:=def_flag; {implied `\.{@@!}'}
end;
next_control:=get_next;
if (next_control="|")or(next_control="{") then return;
end;
exit:end;
@ The |outer_xref| subroutine is like |Pascal_xref| but it begins
with |next_control<>"|"| and ends with |next_control>=format|. Thus, it
handles \PASCAL\ text with embedded comments.
@p procedure outer_xref; {extension of |Pascal_xref|}
var bal:eight_bits; {brace level in comment}
begin while next_control<format do
if next_control<>"{" then Pascal_xref
else begin bal:=skip_comment(1); next_control:="|";
while bal>0 do
begin Pascal_xref;
if next_control="|" then bal:=skip_comment(bal)
else bal:=0; {an error will be reported in phase two}
end;
end;
end;
@ In the \TeX\ part of a module, cross reference entries are made only for
the identifiers in \PASCAL\ texts enclosed in \pb, or for control texts
enclosed in \.{@@\^}$\,\ldots\,$\.{@@>} or \.{@@.}$\,\ldots\,$\.{@@>}
or \.{@@:}$\,\ldots\,$\.{@@>}.
@<Store cross references in the \T...@>=
repeat next_control:=skip_TeX;
case next_control of
underline: xref_switch:=def_flag;
no_underline: xref_switch:=0;
"|": Pascal_xref;
xref_roman, xref_wildcard, xref_typewriter, module_name:
begin loc:=loc-2; next_control:=get_next; {scan to \.{@@>}}
if next_control<>module_name then
new_xref(id_lookup(next_control-identifier));
end;
othercases do_nothing
endcases;
until next_control>=format
@ During the definition and \PASCAL\ parts of a module, cross references
are made for all identifiers except reserved words; however, the
identifiers in a format definition are referenced even if they are
reserved. The \TeX\ code in comments is, of course, ignored, except for
\PASCAL\ portions enclosed in \pb; the text of a module name is skipped
entirely, even if it contains \pb\ constructions.
The variables |lhs| and |rhs| point to the respective identifiers involved
in a format definition.
@<Global...@>=
@!lhs,@!rhs:name_pointer; {indices into |byte_start| for format identifiers}
@ When we get to the following code we have |next_control>=format|.
@<Store cross references in the \(d...@>=
while next_control<=definition do {|format| or |definition|}
begin xref_switch:=def_flag; {implied \.{@@!}}
if next_control=definition then next_control:=get_next
else @<Process a format definition@>;
outer_xref;
end
@ Error messages for improper format definitions will be issued in phase
two. Our job in phase one is to define the |ilk| of a properly formatted
identifier, and to fool the |new_xref| routine into thinking that the
identifier on the right-hand side of the format definition is not a
reserved word.
@<Process a form...@>=
begin next_control:=get_next;
if next_control=identifier then
begin lhs:=id_lookup(normal); ilk[lhs]:=normal; new_xref(lhs);
next_control:=get_next;
if next_control=equivalence_sign then
begin next_control:=get_next;
if next_control=identifier then
begin rhs:=id_lookup(normal);
ilk[lhs]:=ilk[rhs]; ilk[rhs]:=normal; new_xref(rhs);
ilk[rhs]:=ilk[lhs]; next_control:=get_next;
end;
end;
end;
end
@ Finally, when the \TeX\ and definition parts have been treated, we have
|next_control>=begin_Pascal|.
@<Store cross references in the \P...@>=
if next_control<=module_name then {|begin_Pascal| or |module_name|}
begin if next_control=begin_Pascal then mod_xref_switch:=0
else mod_xref_switch:=def_flag;
repeat if next_control=module_name then new_mod_xref(cur_module);
next_control:=get_next; outer_xref;
until next_control>module_name;
end
@ After phase one has looked at everything, we want to check that each
module name was both defined and used.
The variable |cur_xref| will point to cross references for the
current module name of interest.
@<Glob...@>=@!cur_xref:xref_number; {temporary cross reference pointer}
@ The following recursive procedure
walks through the tree of module names and prints out anomalies.
@^recursion@>
@p procedure mod_check(@!p:name_pointer); {print anomalies in subtree |p|}
begin if p>0 then
begin mod_check(llink[p]);@/
cur_xref:=xref[p];
if num(cur_xref)<def_flag then
begin print_nl('! Never defined: <'); print_id(p);
@.Never defined: <section name>@>
print('>'); mark_harmless;
end;
while num(cur_xref)>=def_flag do cur_xref:=xlink(cur_xref);
if cur_xref=0 then
begin print_nl('! Never used: <'); print_id(p); print('>');
@.Never used: <section name>@>
mark_harmless;
end;
mod_check(rlink[p]);
end;
end;
@ @<Print error messages about un...@>=@+mod_check(root)
@* Low-level output routines.
The \TeX\ output is supposed to appear in lines at most |line_length|
characters long, so we place it into an output buffer. During the output
process, |out_line| will hold the current line number of the line about to
be output.
@<Glo...@>=
@!out_buf:array[0..line_length] of ASCII_code; {assembled characters}
@!out_ptr:0..line_length; {number of characters in |out_buf|}
@!out_line: integer; {coordinates of next line to be output}
@ The |flush_buffer| routine empties the buffer up to a given breakpoint,
and moves any remaining characters to the beginning of the next line.
If the |per_cent| parameter is |true|, a |"%"| is appended to the line
that is being output; in this case the breakpoint |b| should be strictly
less than |line_length|. If the |per_cent| parameter is |false|,
trailing blanks are suppressed.
The characters emptied from the buffer form a new line of output;
if the |carryover| parameter is true, a |"%"| in that line will be
carried over to the next line (so that \TeX\ will ignore the completion
of commented-out text).
@p procedure flush_buffer(@!b:eight_bits;@!per_cent,@!carryover:boolean);
{outputs |out_buf[1..b]|, where |b<=out_ptr|}
label done,found;
var j,@!k:0..line_length;
begin j:=b;
if not per_cent then {remove trailing blanks}
loop@+ begin if j=0 then goto done;
if out_buf[j]<>" " then goto done;
decr(j);
end;
done: for k:=1 to j do write(tex_file,xchr[out_buf[k]]);
if per_cent then write(tex_file,xchr["%"]);
write_ln(tex_file); incr(out_line);
if carryover then
for k:=1 to j do
if out_buf[k]="%" then
if (k=1)or(out_buf[k-1]<>"\") then {comment mode should be preserved}
begin out_buf[b]:="%"; decr(b); goto found;
end;
found: if (b<out_ptr) then
for k:=b+1 to out_ptr do out_buf[k-b]:=out_buf[k];
out_ptr:=out_ptr-b;
end;
@ When we are copying \TeX\ source material, we retain line breaks
that occur in the input, except that an empty line is not
output when the \TeX\ source line was nonempty. For example, a line
of the \TeX\ file that contains only an index cross-reference entry
will not be copied. The |finish_line| routine is called just before
|get_line| inputs a new line, and just after a line break token has
been emitted during the output of translated \PASCAL\ text.
@p procedure finish_line; {do this at the end of a line}
label exit;
var k:0..buf_size; {index into |buffer|}
begin if out_ptr>0 then flush_buffer(out_ptr,false,false)
else begin for k:=0 to limit do
if (buffer[k]<>" ")and(buffer[k]<>tab_mark) then return;
flush_buffer(0,false,false);
end;
exit:end;
@ In particular, the |finish_line| procedure is called near the very
beginning of phase two. We initialize the output variables in a slightly
tricky way so that the first line of the output file will be
`\.{\\input webmac}'.
@.\\input webmac@>
@.webmac@>
@<Set init...@>=
out_ptr:=1; out_line:=1; out_buf[1]:="c"; write(tex_file,'\input webma');
@ When we wish to append the character |c| to the output buffer, we write
`$|out|(c)$'; this will cause the buffer to be emptied if it was already
full. Similarly, `$|out2|(c_1)(c_2)$' appends a pair of characters.
A line break will occur at a space or after a single-nonletter
\TeX\ control sequence.
@d oot(#)==@;@/
if out_ptr=line_length then break_out;
incr(out_ptr); out_buf[out_ptr]:=#;
@d oot1(#)==oot(#)@+end
@d oot2(#)==oot(#)@,oot1
@d oot3(#)==oot(#)@,oot2
@d oot4(#)==oot(#)@,oot3
@d oot5(#)==oot(#)@,oot4
@d out==@+begin oot1
@d out2==@+begin oot2
@d out3==@+begin oot3
@d out4==@+begin oot4
@d out5==@+begin oot5
@ The |break_out| routine is called just before the output buffer is about
to overflow. To make this routine a little faster, we initialize position
0 of the output buffer to `\.\\'; this character isn't really output.
@<Set init...@>=
out_buf[0]:="\";
@ A long line is broken at a blank space or just before a backslash that isn't
preceded by another backslash. In the latter case, a |"%"| is output at
the break.
@p procedure break_out; {finds a way to break the output line}
label exit;
var k:0..line_length; {index into |out_buf|}
@!d:ASCII_code; {character from the buffer}
begin k:=out_ptr;
loop@+ begin if k=0 then
@<Print warning message, break the line, |return|@>;
d:=out_buf[k];
if d=" " then
begin flush_buffer(k,false,true); return;
end;
if (d="\")and(out_buf[k-1]<>"\") then {in this case |k>1|}
begin flush_buffer(k-1,true,true); return;
end;
decr(k);
end;
exit:end;
@ We get to this module only in unusual cases that the entire output line
consists of a string of backslashes followed by a string of nonblank
non-backslashes. In such cases it is almost always safe to break the
line by putting a |"%"| just before the last character.
@<Print warning message...@>=
begin print_nl('! Line had to be broken (output l.',out_line:1);
@.Line had to be broken@>
print_ln('):');
for k:=1 to out_ptr-1 do print(xchr[out_buf[k]]);
new_line; mark_harmless;
flush_buffer(out_ptr-1,true,true); return;
end
@ Here is a procedure that outputs a module number in decimal notation.
@<Glob...@>=@!dig:array[0..4] of 0..9; {digits to output}
@ The number to be converted by |out_mod| is known to be less than
|def_flag|, so it cannot have more than five decimal digits. If
the module is changed, we output `\.{\\*}' just after the number.
@p procedure out_mod(@!m:integer); {output a module number}
var k:0..5; {index into |dig|}
@!a:integer; {accumulator}
begin k:=0; a:=m;
repeat dig[k]:=a mod 10; a:=a div 10; incr(k);
until a=0;
repeat decr(k); out(dig[k]+"0");
until k=0;
if changed_module[m] then out2("\")("*");
@.\\*@>
end;
@ The |out_name| subroutine is used to output an identifier or index
entry, enclosing it in braces.
@p procedure out_name(@!p:name_pointer); {outputs a name}
var k:0..max_bytes; {index into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
begin out("{"); w:=p mod ww;
for k:=byte_start[p] to byte_start[p+ww]-1 do
begin if byte_mem[w,k]="_" then out("\");
@.\\_@>
out(byte_mem[w,k]);
end;
out("}");
end;
@* Routines that copy \TeX\ material.
During phase two, we use the subroutines |copy_limbo|, |copy_TeX|, and
|copy_comment| in place of the analogous |skip_limbo|, |skip_TeX|, and
|skip_comment| that were used in phase one.
The |copy_limbo| routine, for example, takes \TeX\ material that is not
part of any module and transcribes it almost verbatim to the output file.
No `\.{@@}' signs should occur in such material except in `\.{@@@@}'
pairs; such pairs are replaced by singletons.
@p procedure copy_limbo; {copy \TeX\ code until the next module begins}
label exit;
var c:ASCII_code; {character following \.{@@} sign}
begin loop if loc>limit then
begin finish_line; get_line;
if input_has_ended then return;
end
else begin buffer[limit+1]:="@@";
@<Copy up to control code, |return| if finished@>;
end;
exit:end;
@ @<Copy up to control...@>=
while buffer[loc]<>"@@" do
begin out(buffer[loc]); incr(loc);
end;
if loc<=limit then
begin loc:=loc+2; c:=buffer[loc-1];
if (c=" ")or(c=tab_mark)or(c="*") then return;
if (c<>"z")and(c<>"Z") then
begin out("@@");
if c<>"@@" then err_print('! Double @@ required outside of sections');
@.Double \AT! required...@>
end;
end
@ The |copy_TeX| routine processes the \TeX\ code at the beginning of a
module; for example, the words you are now reading were copied in this
way. It returns the next control code or `\v' found in the input.
@p function copy_TeX:eight_bits; {copy pure \TeX\ material}
label done;
var c:eight_bits; {control code found}
begin loop begin if loc>limit then
begin finish_line; get_line;
if input_has_ended then
begin c:=new_module; goto done;
end;
end;
buffer[limit+1]:="@@";
@<Copy up to `\v' or control code, |goto done| if finished@>;
end;
done:copy_TeX:=c;
end;
@ We don't copy spaces or tab marks into the beginning of a line. This
makes the test for empty lines in |finish_line| work.
@<Copy up to `\v'...@>=
repeat c:=buffer[loc]; incr(loc);
if c="|" then goto done;
if c<>"@@" then
begin out(c);
if (out_ptr=1)and((c=" ")or(c=tab_mark)) then decr(out_ptr);
end;
until c="@@";
if loc<=limit then
begin c:=control_code(buffer[loc]); incr(loc);
goto done;
end
@ The |copy_comment| uses and returns a brace-balance value, following the
conventions of |skip_comment| above. Instead of copying the \TeX\ material
into the output buffer, this procedure copies it into the token memory.
The abbreviation |app_tok(t)| is used to append token |t| to the current
token list, and it also makes sure that it is possible to append at least
one further token without overflow.
@d app_tok(#)==begin if tok_ptr+2>max_toks then overflow('token');
tok_mem[tok_ptr]:=#; incr(tok_ptr);
end
@p function copy_comment(@!bal:eight_bits):eight_bits; {copies \TeX\ code in
comments}
label done;
var c:ASCII_code; {current character being copied}
begin loop begin if loc>limit then
begin get_line;
if input_has_ended then
begin err_print('! Input ended in mid-comment');
@.Input ended in mid-comment@>
loc:=1; @<Clear |bal| and |goto done|@>;
end;
end;
c:=buffer[loc]; incr(loc);
if c="|" then goto done;
app_tok(c);
@<Copy special things when |c="@@", "\", "{", "}"|;
|goto done| at end@>;
end;
done: copy_comment:=bal;
end;
@ @<Copy special things when |c="@@"...@>=
if c="@@" then
begin incr(loc);
if buffer[loc-1]<>"@@" then
begin err_print('! Illegal use of @@ in comment');
@.Illegal use of \AT!...@>
loc:=loc-2; decr(tok_ptr); @<Clear |bal|...@>;
end;
end
else if (c="\")and(buffer[loc]<>"@@") then
begin app_tok(buffer[loc]); incr(loc);
end
else if c="{" then incr(bal)
else if c="}" then
begin decr(bal);
if bal=0 then goto done;
end
@ When the comment has terminated abruptly due to an error, we output
enough right braces to keep \TeX\ happy.
@<Clear |bal|...@>=
app_tok(" "); {this is done in case the previous character was `\.\\'}
repeat app_tok("}"); decr(bal);
until bal=0;
goto done;
@* Parsing.
The most intricate part of \.{WEAVE} is its mechanism for converting
\PASCAL-like code into \TeX\ code, and we might as well plunge into this
aspect of the program now. A ``bottom up'' approach is used to parse the
\PASCAL-like material, since \.{WEAVE} must deal with fragmentary
constructions whose overall ``part of speech'' is not known.
At the lowest level, the input is represented as a sequence of entities
that we shall call {\it scraps}, where each scrap of information consists
of two parts, its {\it category} and its {\it translation}. The category
is essentially a syntactic class, and the translation is a token list that
represents \TeX\ code. Rules of syntax and semantics tell us how to
combine adjacent scraps into larger ones, and if we are lucky an entire
\PASCAL\ text that starts out as hundreds of small scraps will join
together into one gigantic scrap whose translation is the desired \TeX\
code. If we are unlucky, we will be left with several scraps that don't
combine; their translations will simply be output, one by one.
The combination rules are given as context-sensitive productions that are
applied from left to right. Suppose that we are currently working on the
sequence of scraps $s_1\,s_2\ldots s_n$. We try first to find the longest
production that applies to an initial substring $s_1\,s_2\ldots\,$; but if
no such productions exist, we find to find the longest production
applicable to the next substring $s_2\,s_3\ldots\,$; and if that fails, we
try to match $s_3\,s_4\ldots\,$, etc.
A production applies if the category codes have a given pattern. For
example, one of the productions is
$$|open|\;|math|\;|semi|\;\RA\;|open|\;|math|$$
and it means that three consecutive scraps whose respective categories are
|open|, |math|, and |semi| are con\-verted to two scraps whose categories
are |open| and |math|. This production also has an associated rule that
tells how to combine the translation parts:
$$\eqalign{O_2&=O_1\cr
M_2&=M_1\,S\,\.{\\,}\,\hbox{|opt|\thinspace\tt5}\cr}$$
This means that the |open| scrap has not changed, while the new |math| scrap
has a translation $M_2$ composed of the translation $M_1$ of the original
|math| scrap followed by the translation |S| of the |semi| scrap followed
by `\.{\\,}' followed by `|opt|' followed by `\.5'. (In the \TeX\ file,
this will specify an additional thin space after the semicolon, followed
by an optional line break with penalty 50.) Translation rules use subscripts
to distinguish between translations of scraps whose categories have the
same initial letter; these subscripts are assigned from left to right.
$\.{WEAVE}$ also has the production rule
$$|semi|\;\RA\;|terminator|$$
(meaning that a semicolon can terminate a \PASCAL\ statement). Since
productions are applied from left to right, this rule will be activated
only if the |semi| is not preceded by scraps that match other productions;
in particular, a |semi| that is preceded by `|open| |math|' will have
disappeared because of the production above, and such semicolons do not
act as statement terminators. This incidentally is how \.{WEAVE} is able
to treat semicolons in two distinctly different ways, the first of which
is intended for semicolons in the parameter list of a procedure
declaration.
The translation rule corresponding to $|semi|\;\RA\;|terminator|$ is
$$T=S$$
but we shall not mention translation rules in the common case that the
translation of the new scrap on the right-hand side is simply the
concatenation of the disappearing scraps on the left-hand side.
@ Here is a list of the category codes that scraps can have.
@d simp=1 {the translation can be used both in horizontal mode
and in math mode of \TeX}
@d math=2 {the translation should be used only in \TeX\ math mode}
@d intro=3 {a statement is expected to follow this, after a space and
an optional break}
@d open=4 {denotes an incomplete parenthesized quantity to be used in
math mode}
@d beginning=5 {denotes an incomplete compound statement to be used in
horizontal mode}
@d close=6 {ends a parenthesis or compound statement}
@d alpha=7 {denotes the beginning of a clause}
@d omega=8 {denotes the ending of a clause and possible comment following}
@d semi=9 {denotes a semicolon and possible comment following it}
@d terminator=10 {something that ends a statement or declaration}
@d stmt=11 {denotes a statement or declaration including its terminator}
@d cond=12 {precedes an \&{if} clause that might have a matching \&{else}}
@d clause=13 {precedes a statement after which indentation ends}
@d colon=14 {denotes a colon}
@d exp=15 {stands for the E in a floating point constant}
@d proc=16 {denotes a procedure or program or function heading}
@d case_head=17 {denotes a case statement or record heading}
@d record_head=18 {denotes a record heading without indentation}
@d var_head=19 {denotes a variable declaration heading}
@d elsie=20 {\&{else}}
@d casey=21 {\&{case}}
@d mod_scrap=22 {denotes a module name}
@p @!debug procedure print_cat(@!c:eight_bits);
{symbolic printout of a category}
begin case c of
simp: print('simp');
math: print('math');
intro: print('intro');
open: print('open');
beginning: print('beginning');
close: print('close');
alpha: print('alpha');
omega: print('omega');
semi: print('semi');
terminator: print('terminator');
stmt: print('stmt');
cond: print('cond');
clause: print('clause');
colon: print('colon');
exp: print('exp');
proc: print('proc');
case_head: print('casehead');
record_head: print('recordhead');
var_head: print('varhead');
elsie: print('elsie');
casey: print('casey');
mod_scrap: print('module');
othercases print('UNKNOWN')
endcases;
end;
gubed
@ The token lists for translated \TeX\ output contain some special control
symbols as well as ordinary characters. These control symbols are
interpreted by \.{WEAVE} before they are written to the output file.
\yskip\hang |break_space| denotes an optional line break or an en space;
\yskip\hang |force| denotes a line break;
\yskip\hang |big_force| denotes a line break with additional vertical space;
\yskip\hang |opt| denotes an optional line break (with the continuation
line indented two ems with respect to the normal starting position)---this
code is followed by an integer |n|, and the break will occur with penalty
$10n$;
\yskip\hang |backup| denotes a backspace of one em;
\yskip\hang |cancel| obliterates any |break_space| or |force| or |big_force|
tokens that immediately precede or follow it and also cancels any
|backup| tokens that follow it;
\yskip\hang |indent| causes future lines to be indented one more em;
\yskip\hang |outdent| causes future lines to be indented one less em.
\yskip\noindent All of these tokens are removed from the \TeX\ output that
comes from \PASCAL\ text between \pb\ signs; |break_space| and |force| and
|big_force| become single spaces in this mode. The translation of other
\PASCAL\ texts results in \TeX\ control sequences \.{\\1}, \.{\\2},
\.{\\3}, \.{\\4}, \.{\\5}, \.{\\6}, \.{\\7} corresponding respectively to
|indent|, |outdent|, |opt|, |backup|, |break_space|, |force|, and
|big_force|. However, a sequence of consecutive `\.\ ', |break_space|,
|force|, and/or |big_force| tokens is first replaced by a single token
(the maximum of the given ones).
The tokens |math_rel|, |math_bin|, |math_op| will be translated into
\.{\\mathrel\{}, \.{\\mathbin\{}, and \.{\\mathop\{}, respectively.
Other control sequences in the \TeX\ output will be `\.{\\\\\{}$\,\ldots\,$\.\}'
surrounding identifiers, `\.{\\\&\{}$\,\ldots\,$\.\}' surrounding
reserved words, `\.{\\.\{}$\,\ldots\,$\.\}' surrounding strings,
`\.{\\C\{}$\,\ldots\,$\.\}$\,$|force|' surrounding comments, and
`\.{\\X$n$:}$\,\ldots\,$\.{\\X}' surrounding module names, where
|n| is the module number.
@d math_bin=@'203
@d math_rel=@'204
@d math_op=@'205
@d big_cancel=@'206 {like |cancel|, also overrides spaces}
@d cancel=@'207 {overrides |backup|, |break_space|, |force|, |big_force|}
@d indent=cancel+1 {one more tab (\.{\\1})}
@d outdent=cancel+2 {one less tab (\.{\\2})}
@d opt=cancel+3 {optional break in mid-statement (\.{\\3})}
@d backup=cancel+4 {stick out one unit to the left (\.{\\4})}
@d break_space=cancel+5 {optional break between statements (\.{\\5})}
@d force=cancel+6 {forced break between statements (\.{\\6})}
@d big_force=cancel+7 {forced break with additional space (\.{\\7})}
@d end_translation=big_force+1 {special sentinel token at end of list}
@ The raw input is converted into scraps according to the following table,
which gives category codes followed by the translations. Sometimes a single
item of input produces more than one scrap.
\def\stars {\.{**}}%
(The symbol `\stars' stands for `\.{\\\&\{{\rm identifier}\}}',
i.e., the identifier itself treated as a reserved word. In a few cases the
category is given as `|@!comment|'; this is not an actual category code, it
means that the translation will be treated as a comment, as explained
below.)
\yskip\halign{\quad#\hfil&\quad#\hfil\cr
\.{<>}&|math:|\.{\\I}\cr
\.{<=}&|math:|\.{\\L}\cr
\.{>=}&|math:|\.{\\G}\cr
\.{:=}&|math:|\.{\\K}\cr
\.{==}&|math:|\.{\\S}\cr
\.{(*}&|math:|\.{\\B}\cr
\.{*)}&|math:|\.{\\T}\cr
\.{(.}&|open:|\.[\cr
\.{.)}&|close:|\.]\cr
\."$\,$string$\,$\."&|simp:|\.{\\.\{"{\rm$\,$modified string$\,$}"\}}\cr
\.\'$\,$string$\,$\.\'&|simp:|\.{\\.\{\\\'{\rm$\,$modified
string$\,$}\\\'\}}\cr
\.{@@=}$\,$string$\,$\.{@@>}&|simp:|\.{\\=\{{\rm$\,$modified string$\,$}\}}\cr
\#&|math:|\.{\\\#}\cr
\.\$&|math:|\.{\\\$}\cr
\.\_&|math:|\.{\\\_}\cr
\.\%&|math:|\.{\\\%}\cr
\.\^&|math:|\.{\\\^}\cr
\.(&|open:|\.(\cr
\.)&|close:|\.)\cr
\.[&|open:|\.[\cr
\.]&|close:|\.]\cr
\.*&|math:|\.{\\ast}\cr
\.,&|math:|\.,|@,opt@,|\.9\cr
\.{..}&|math:|\.{\\to}\cr
\..&|simp:|\..\cr
\.:&|colon:|\.:\cr
\.;&|semi:|\.;\cr
identifier&|simp:|\.{\\\\\{{\rm$\,$identifier$\,$}\}}\cr
\.E in constant&|exp:|\.{\\E\{}\cr
digit $d$&|simp:|$d$\cr
other character $c$&|math:|$c$\cr
\.{and}&|math:|\.{\\W}\cr
\.{array}&|alpha:|\stars\cr
\.{begin}&|beginning:|$|force|\,\stars\,|cancel|$\qquad|intro:|\cr
\.{case}&|casey:|\qquad|alpha:|$|force|\,\stars$\cr
\.{const}&|intro:|$|force|\,|backup|\,\stars$\cr
\.{div}&|math:|$|math_bin|\,\stars\,\.\}$\cr
\.{do}&|omega:|\stars\cr
\.{downto}&|math:|$|math_rel|\,\stars\,\.\}$\cr
\.{else}&|terminator:|\qquad|elsie:|$|force|\,|backup|\,\stars$\cr
\.{end}&|terminator:|\qquad|close:|$|force|\,\stars$\cr
\.{file}&|alpha:|\stars\cr
\.{for}&|alpha:|$|force|\,\stars$\cr
\.{function}&|proc:|$|force|\,|backup|\,\stars\,|cancel|$\qquad
|intro:|$|indent|\,\.{\\\ }$\cr
\.{goto}&|intro:|\stars\cr
\.{if}&|cond:|\qquad|alpha:|$|force|\,\stars$\cr
\.{in}&|math:|\.{\\in}\cr
\.{label}&|intro:|$|force|\,|backup|\,\stars$\cr
\.{mod}&|math:|$|math_bin|\,\stars\,\.\}$\cr
\.{nil}&|simp:|\stars\cr
\.{not}&|math:|\.{\\R}\cr
\.{of}&|omega:|\stars\cr
\.{or}&|math:|\.{\\V}\cr
\.{packed}&|intro:|\stars\cr
\.{procedure}&|proc:|$|force|\,|backup|\,\stars\,|cancel|$\qquad
|intro:|$|indent|\,\.{\\\ }$\cr
\.{program}&|proc:|$|force|\,|backup|\,\stars\,|cancel|$\qquad
|intro:|$|indent|\,\.{\\\ }$\cr
\.{record}&|record_head:|\stars\qquad|intro:|\cr
\.{repeat}&|beginning:|$|force|\,|indent|\,\stars\,|cancel|$\qquad|intro:|\cr
\.{set}&|alpha:|\stars\cr
\.{then}&|omega:|\stars\cr
\.{to}&|math:|$|math_rel|\,\stars\,\.\}$\cr
\.{type}&|intro:|$|force|\,|backup|\,\stars$\cr
\.{until}&|terminator:|\qquad|close:|$|force|\,|backup|\,\stars$\qquad
|clause:|\cr
\.{var}&|var_head:|$|force|\,|backup|\,\stars\,|cancel|$\qquad|intro:|\cr
\.{while}&|alpha:|$|force|\,\stars$\cr
\.{with}&|alpha:|$|force|\,\stars$\cr
\.{xclause}&|alpha:|$|force|\,\.{\\\~}$\qquad|omega:|\stars\cr
\.{@@\'}$\,$const&|simp:|\.{\\O\{}\hbox{const}\.\}\cr
\.{@@"}$\,$const&|simp:|\.{\\H\{}\hbox{const}\.\}\cr
\.{@@\$}&|simp:|\.{\\)}\cr
\.{@@\\}&|simp:|\.{\\]}\cr
\.{@@,}&|math:|\.{\\,}\cr
\.{@@t}$\,$stuff$\,$\.{@@>}&|simp:|\.{\\hbox\{{\rm$\,$stuff$\,$}\}}\cr
\.{@@<}$\,$module$\,$\.{@@>}&|mod_scrap:|\.{\\X$n$:{\rm$\,$module$\,$}\\X}\cr
\.{@@\#}&|comment:||big_force|\cr
\.{@@/}&|comment:||force|\cr
\.{@@\char'174}&|simp:|$|opt|\,\.0$\cr
\.{@@+}&|comment:|$|big_cancel|\,\.{\\\ }\,|big_cancel|$\cr
\.{@@;}&|semi:|\cr
\.{@@\&}&|math:|\.{\\J}\cr
\.{@@\{}&|math:|\.{\\B}\cr
\.{@@\}}&|math:|\.{\\T}\cr}
\yskip\noindent When a string is output, certain characters are preceded by
`\.\\' signs so that they will print properly.
A comment in the input will be combined with the preceding
|omega| or |semi| scrap, or with the following |terminator| scrap, if
possible; otherwise it will be inserted as a separate |terminator| scrap.
An additional ``comment'' is effectively appended at the end of the
\PASCAL\ text, just before translation begins; this consists of a |cancel|
token in the case of \PASCAL\ text in \pb, otherwise it consists of a
|force| token.
From this table it is evident that \.{WEAVE} will parse a lot of non-\PASCAL\
programs. For example, the reserved words `\.{for}' and `\.{array}' are
treated in an identical way by \.{WEAVE} from a syntactic standpoint,
and semantically they are equivalent except that a forced line break occurs
just before `\&{for}'; \PASCAL\ programmers may well be surprised at this
similarity. The idea is to keep \.{WEAVE}'s rules as simple as possible,
consistent with doing a reasonable job on syntactically correct \PASCAL\
programs. The production rules below have been formulated in the same
spirit of ``almost anything goes.''
@ Here is a table of all the productions. The reader can best get a feel for
@^productions, table of@>
how they work by trying them out by hand on small examples; no amount of
explanation will be as effective as watching the rules in action.
Parsing can also be watched by debugging with `\.{@@2}'.
\def\[#1]{\quad$\dleft#1\dright$}
\def\sp{\.{\ }}
\yskip
\halign to\the\hsize{\hfil\it# &
#\hfil\hskip-200pt\tabskip 0pt plus 100pt&
#\hfil\tabskip0pt\cr
&Production categories\[\hbox{translations}]&Remarks\cr
\noalign{\yskip}
1&|alpha@,math@,colon| $\RA$ |alpha@,math|&e.g., |case v:boolean of|\cr
2&|alpha@,math@,omega| $\RA$ |clause|\[C=A\,\sp\,\.\$\,M\,\.\$\,\sp\,|indent|\,
O]&e.g., |while x>0 do|\cr
3&|alpha@,omega| $\RA$ |clause|\[C=A\,\sp\,|indent|\,O]&e.g., |file of|\cr
4&|alpha@,simp| $\RA$ |alpha@,math|&convert to math mode\cr
5&|beginning@,close@,(terminator@t or @>stmt)| $\RA$ |stmt|&compound statement
ends\cr
6&|beginning@,stmt| $\RA$ |beginning|\[B_2=B_1\,|break_space|\,S]&compound
statement grows\cr
7&|case_head@,casey@,clause| $\RA$ |case_head|\[C_4=C_1\,|outdent|\,C_2\,C_3]&
variant records\cr
8&|case_head@,close@,terminator| $\RA$ |stmt|\[S=C_1\,|cancel|\,|outdent|\,
C_2\,T]&end of case statement\cr
9&|case_head@,stmt| $\RA$ |case_head|\[C_2=C_1\,|force|\,S]&case statement
grows\cr
10&|casey@,clause| $\RA$ |case_head|&beginning of case statement\cr
11&|clause@,stmt| $\RA$ |stmt|\[S_2=C\,|break_space|\,S_1\,|cancel|\,|outdent|\,
|force|]&end of controlled statement\cr
12&|cond@,clause@,stmt@,elsie| $\RA$ |clause|\[C_3=C_1\,C_2\,|break_space|\,S\,
E\,\sp\,|cancel|]&complete conditional\cr
13&|cond@,clause@,stmt| $\RA$ |stmt|\cr
&\qquad\[S_2=C_1\,C_2\,|break_space|\,S_1\,
|cancel|\,|outdent|\,|force|]&incomplete conditional\cr
14&|elsie| $\RA$ |intro|&unmatched else\cr
15&|exp@,math@,simp|* $\RA$ |math|\[M_2=E\,M_1\,S\,\.\}]&signed exponent\cr
16&|exp@,simp|* $\RA$ |math|\[M=E\,S\,\.\}]&unsigned exponent\cr
17&|intro@,stmt| $\RA$ |stmt|\[S_2=I\,\sp\,|opt|\,\.7\,|cancel|\,S_1]&labeled
statement, etc.\cr
18&|math@,close| $\RA$ |stmt@,close|\[S=\.\$\,M\,\.\$]&end of field list\cr
19&|math@,colon| $\RA$ |intro|\[I=|force|\,|backup|\,\.\$\,M\,\.\$\,C]&compound
label\cr
20&|math@,math| $\RA$ |math|&simple concatenation\cr
21&|math@,simp| $\RA$ |math|&simple concatenation\cr
22&|math@,stmt| $\RA$ |stmt|\cr
&\qquad\[S_2=\.\$\,M\,\.\$\,|indent|\,|break_space|\,
S_1\,|cancel|\,|outdent|\,|force|]¯o or type definition\cr
23&|math@,terminator| $\RA$ |stmt|\[S=\.\$\,M\,\.\$\,T]&statement involving
math\cr
24&|mod_scrap@,(terminator@t or @>semi)| $\RA$ |stmt|\[S=M\,T\,|force|]&module
like a statement\cr
25&|mod_scrap| $\RA$ |simp|&module unlike a statement\cr
26&|open@,case_head@,close| $\RA$ |math|\[M=O\,\.\$\,|cancel|\,C_1\,
|cancel|\,|outdent|\,\.\$\,C_2]&case in field list\cr
27&|open@,close| $\RA$ |math|\[M=O\,\.\\\,\.,\,C]&empty set |[]|\cr
28&|open@,math@,case_head@,close| $\RA$ |math|\cr
&\qquad\[M_2=O\,M_1\,\.\$\,|cancel|\,
C_1\,|cancel|\,|outdent|\,\.\$\,C_2]&case in field list\cr
29&|open@,math@,close| $\RA$ |math|&parenthesized group\cr
30&|open@,math@,colon| $\RA$ |open@,math|&colon in parentheses\cr
31&|open@,math@,proc@,intro| $\RA$ |open@,math|\[M_2=M_1\,|math_op|\,|cancel|\,
P\,\.\}]&|procedure| in parentheses\cr
32&|open@,math@,semi| $\RA$ |open@,math|\[M_2=M_1\,S\,\.\\\,\.,\,|opt|\,
\.5]&semicolon in parentheses\cr
33&|open@,math@,var_head@,intro| $\RA$ |open@,math|\[M_2=M_1\,|math_op|\,
|cancel|\,V\,\.\}]&|var| in parentheses\cr
34&|open@,proc@,intro| $\RA$ |open@,math|\[M=|math_op|\,|cancel|\,
P\,\.\}]&|procedure| in parentheses\cr
35&|open@,simp| $\RA$ |open@,math|&convert to math mode\cr
36&|open@,stmt@,close| $\RA$ |math|\[M=O\,\.\$\,|cancel|\,S\,|cancel|\,
\.\$\,C]&field list\cr
37&|open@,var_head@,intro| $\RA$ |open@,math|\[M=|math_op|\,|cancel|\,V\,
\.\}]&|var| in parentheses\cr
38&|proc@,beginning@,close@,terminator| $\RA$ |stmt|\[S=P\,|cancel|\,
|outdent|\,B\,C\,T]&end of procedure declaration\cr
39&|proc@,stmt| $\RA$ |proc|\[P_2=P_1\,|break_space|\,S]&procedure declaration
grows\cr
40&|record_head@,intro@,casey| $\RA$ |casey|\[C_2=R\,I\,\sp\,|cancel|\,C_1]&
\&{record case} $\ldots$\cr
41&|record_head| $\RA$ |case_head|\[C=|indent|\,R\,|cancel|]&other \&{record}
structures\cr
42&|semi| $\RA$ |terminator|&semicolon after statement\cr
43&|simp@,close| $\RA$ |stmt@,close|&end of field list\cr
44&|simp@,colon| $\RA$ |intro|\[I=|force|\,|backup|\,S\,C]&simple label\cr
45&|simp@,math| $\RA$ |math|&simple concatenation\cr
46&|simp@,mod_scrap| $\RA$ |mod_scrap|&in emergencies\cr
47&|simp@,simp| $\RA$ |simp|&simple concatenation\cr
48&|simp@,terminator| $\RA$ |stmt|&simple statement\cr
49&|stmt@,stmt| $\RA$ |stmt|\[S_3=S_1\,|break_space|\,S_2]&adjacent
statements\cr
50&|terminator| $\RA$ |stmt|&empty statement\cr
51&|var_head@,beginning| $\RA$ |stmt@,beginning|&end of variable
declarations\cr
52&|var_head@,math@,colon| $\RA$ |var_head@,intro|\[I=\.\$\,M\,\.\$\,C]&
variable declaration\cr
53&|var_head@,simp@,colon| $\RA$ |var_head@,intro|&variable declaration\cr
54&|var_head@,stmt| $\RA$ |var_head|\[V_2=V_1\,|break_space|\,S]&variable
declarations grow\cr}
\yskip\noindent
Translations are not specified here when they are simple concatenations
of the scraps that change. For example, the full translation of
`|open@,math@,colon| $\RA$ |open@,math|' is $O_2=O_1$, $M_2=M_1C$.
The notation `|simp|*', in the |exp|-related productions above,
stands for a |simp| scrap that isn't followed by another |simp|.
@* Implementing the productions.
When \PASCAL\ text is to be processed with the grammar above, we put its
initial scraps $s_1\ldots s_n$ into two arrays |cat[1..n]| and |trans[1..n]|.
The value of |cat[k]| is simply a category code from the list above; the
value of |trans[k]| is a text pointer, i.e., an index into |tok_start|.
Our production rules have the nice property that the right-hand side is never
longer than the left-hand side. Therefore it is convenient to use sequential
allocation for the current sequence of scraps. Five pointers are used to
manage the parsing:
\yskip\hang |pp| (the parsing pointer) is such that we are trying to match
the category codes |cat[pp]@,cat[pp+1]|$\,\ldots\,$ to the left-hand sides
of productions.
\yskip\hang |scrap_base|, |lo_ptr|, |hi_ptr|, and |scrap_ptr| are such that
the current sequence of scraps appears in positions |scrap_base| through
|lo_ptr| and |hi_ptr| through |scrap_ptr|, inclusive, in the |cat| and
|trans| arrays. Scraps located between |scrap_base| and |lo_ptr| have
been examined, while those in positions |>=hi_ptr| have not yet been
looked at by the parsing process.
\yskip\noindent Initially |scrap_ptr| is set to the position of the final
scrap to be parsed, and it doesn't change its value. The parsing process
makes sure that |lo_ptr>=pp+3|, since productions have as many as four terms,
by moving scraps from |hi_ptr| to |lo_ptr|. If there are
fewer than |pp+3| scraps left, the positions up to |pp+3| are filled with
blanks that will not match in any productions. Parsing stops when
|pp=lo_ptr+1| and |hi_ptr=scrap_ptr+1|.
The |trans| array elements are declared to be of type |0..10239| instead
of type |text_pointer|, because the final sorting phase of \.{WEAVE}
uses this array to contain elements of type |name_pointer|. Both
of these types are subranges of |0..10239|.
@<Glo...@>=
@!cat:array[0..max_scraps] of eight_bits; {category codes of scraps}
@!trans:array[0..max_scraps] of 0..10239; {translation texts of scraps}
@!pp:0..max_scraps; {current position for reducing productions}
@!scrap_base:0..max_scraps; {beginning of the current scrap sequence}
@!scrap_ptr:0..max_scraps; {ending of the current scrap sequence}
@!lo_ptr:0..max_scraps; {last scrap that has been examined}
@!hi_ptr:0..max_scraps; {first scrap that has not been examined}
stat@!max_scr_ptr:0..max_scraps; {largest value assumed by |scrap_ptr|}
tats
@ @<Set init...@>=
scrap_base:=1; scrap_ptr:=0;
stat max_scr_ptr:=0; @+tats
@ Token lists in |@!tok_mem| are composed of the following kinds of
items for \TeX\ output.
\yskip\item{$\bullet$}ASCII codes and special codes like |force| and
|math_rel| represent themselves;
\item{$\bullet$}|id_flag+p| represents \.{\\\\\{{\rm identifier $p$}\}};
\item{$\bullet$}|res_flag+p| represents \.{\\\&\{{\rm identifier $p$}\}};
\item{$\bullet$}|mod_flag+p| represents module name |p|;
\item{$\bullet$}|tok_flag+p| represents token list number |p|;
\item{$\bullet$}|inner_tok_flag+p| represents token list number |p|, to be
translated without line-break controls.
@d id_flag=10240 {signifies an identifier}
@d res_flag=id_flag+id_flag {signifies a reserved word}
@d mod_flag=res_flag+id_flag {signifies a module name}
@d tok_flag==mod_flag+id_flag {signifies a token list}
@d inner_tok_flag==tok_flag+id_flag {signifies a token list in `\pb'}
@#
@d lbrace==xchr["{"] {this avoids possible \PASCAL\ compiler confusion}
@d rbrace==xchr["}"] {because these braces might occur within comments}
@p @!debug procedure print_text(@!p:text_pointer); {prints a token list}
var j:0..max_toks; {index into |tok_mem|}
@!r:0..id_flag-1; {remainder of token after the flag has been stripped off}
begin if p>=text_ptr then print('BAD')
else for j:=tok_start[p] to tok_start[p+1]-1 do
begin r:=tok_mem[j] mod id_flag;
case tok_mem[j] div id_flag of
1: begin print('\\',lbrace); print_id(r); print(rbrace);
end; {|id_flag|}
2: begin print('\&',lbrace); print_id(r); print(rbrace);
end; {|res_flag|}
3: begin print('<'); print_id(r); print('>');
end; {|mod_flag|}
4: print('[[',r:1,']]'); {|tok_flag|}
5: print('|[[',r:1,']]|'); {|inner_tok_flag|}
othercases @<Print token |r| in symbolic form@>
endcases;
end;
end;
gubed
@ @<Print token |r|...@>=
case r of
math_bin: print('\mathbin',lbrace);
math_rel: print('\mathrel',lbrace);
math_op: print('\mathop',lbrace);
big_cancel: print('[ccancel]');
cancel: print('[cancel]');
indent: print('[indent]');
outdent: print('[outdent]');
backup: print('[backup]');
opt: print('[opt]');
break_space: print('[break]');
force: print('[force]');
big_force: print('[fforce]');
end_translation: print('[quit]');
othercases print(xchr[r])
endcases
@ The production rules listed above are embedded directly into the \.{WEAVE}
program, since it is easier to do this than to write an interpretive system
that would handle production systems in general. Several macros are defined
here so that the program for each production is fairly short.
All of our productions conform to the general notion that some |k|
consecutive scraps starting at some position |j| are to be replaced by a
single scrap of some category |c| whose translation is composed from the
translations of the disappearing scraps. After this production has been
applied, the production pointer |pp| should change by an amount |d|. Such
a production can be represented by the quadruple $(j,k,c,d)$. For example,
the production `|simp@,math| $\RA$ |math|' would be represented by
`$(|pp|,2,|math|,-1)$'; in this case the pointer $pp$ should decrease by 1
after the production has been applied, because some productions with
|math| in their second positions might now match, but no productions have
|math| in the third or fourth position of their left-hand sides. Note that
the value of |d| is determined by the whole collection of productions, not
by an individual one. Consider the further example
`|var_head@,math@,colon| $\RA$ |var_head@,intro|', which is represented by
`$(|pp|+1,2,|intro|,+1)$'; the $+1$ here is deduced by looking at the
grammar and seeing that no matches could possibly occur at positions |<=pp|
after this production has been applied. The determination of |d| has been
done by hand in each case, based on the full set of productions but not on
the grammar of \PASCAL\ or on the rules for constructing the initial
scraps.
We also attach a serial number to each production, so that additional
information is available when debugging. For example, the program below
contains the statement `|reduce(pp+1,2,intro,+1)(52)|' when it implements
the production just mentioned.
Before calling |reduce|, the program should have appended the tokens of
the new translation to the |tok_mem| array. We commonly want to append
copies of several existing translations, and macros are defined to
simplify these common cases. For example, |app2(pp)| will append the
translations of two consecutive scraps, |trans[pp]| and |trans[pp+1]|, to
the current token list. If the entire new translation is formed in this
way, we write `$|squash|(j,k,c,d)$' instead of `$|reduce|(j,k,c,d)$'. For
example, `|squash(pp,2,math,-1)|' is an abbreviation for `|app2(pp);
reduce(pp,2,math,-1)|'.
The code below is an exact translation of the production rules into
\PASCAL, using such macros, and the reader should have no difficulty
understanding the format by comparing the code with the symbolic
productions as they were listed earlier.
{\sl Caution:\/} The macros |app|, |app1|, |app2|, and |app3| are
sequences of statements that are not enclosed with |begin| and $|end|$,
because such delimiters would make the \PASCAL\ program much longer. This
means that it is necessary to write |begin| and |end| explicitly when such
a macro is used as a single statement. Several mysterious bugs in the
original programming of \.{WEAVE} were caused by a failure to remember
this fact. Next time the author will know better.
@d production(#)==@!debug prod(#) gubed; goto found
@d reduce(#)==red(#); production
@d production_end(#)==@!debug prod(#) gubed; goto found;
end
@d squash(#)==begin sq(#); production_end
@d app(#)==tok_mem[tok_ptr]:=#; incr(tok_ptr) {this is like |app_tok|,
but it doesn't test for overflow}
@d app1(#)==tok_mem[tok_ptr]:=tok_flag+trans[#]; incr(tok_ptr)
@d app2(#)==app1(#);app1(#+1)
@d app3(#)==app2(#);app1(#+2)
@ Let us consider the big case statement for productions now, before looking
at its context. We want to design the program so that this case statement
works, so we might as well not keep ourselves in suspense about exactly what
code needs to be provided with a proper environment.
The code here is more complicated than it need be, since some popular
\PASCAL\ compilers are unable to deal with procedures that contain a lot
of program text. The |translate| procedure, which incorporates the |case|
statement here, would become too long for those compilers if we did
not do something to split the cases into parts. Therefore
a separate procedure called |five_cases| has been introduced.
@^split procedures@>
This auxiliary procedure contains approximately half of the program text
that |translate| would otherwise have had. There's also a procedure
called |alpha_cases|, which turned out to be necessary because the best
two-way split wasn't good enough. The procedure could be split further
in an analogous manner, but the present scheme works on all compilers
known to the author.
@<Match a production at |pp|, or increase |pp| if there is no match@>=
if cat[pp]<=alpha then
if cat[pp]<alpha then five_cases@+else alpha_cases
else begin case cat[pp] of
case_head: @<Cases for |case_head|@>;
casey: @<Cases for |casey|@>;
clause: @<Cases for |clause|@>;
cond: @<Cases for |cond|@>;
elsie: @<Cases for |elsie|@>;
exp: @<Cases for |exp|@>;
mod_scrap: @<Cases for |mod_scrap|@>;
proc: @<Cases for |proc|@>;
record_head: @<Cases for |record_head|@>;
semi: @<Cases for |semi|@>;
stmt: @<Cases for |stmt|@>;
terminator: @<Cases for |terminator|@>;
var_head: @<Cases for |var_head|@>;
othercases do_nothing
endcases;@/
incr(pp); {if no match was found, we move to the right}
found: end
@ Here are the procedures that need to be present for the reason just
explained.
@<Declaration of subprocedures for |translate|@>=
procedure five_cases; {handles almost half of the syntax}
label found;
begin case cat[pp] of
beginning: @<Cases for |beginning|@>;
intro: @<Cases for |intro|@>;
math: @<Cases for |math|@>;
open: @<Cases for |open|@>;
simp: @<Cases for |simp|@>;
othercases do_nothing
endcases;@/
incr(pp); {if no match was found, we move to the right}
found: end;
@#
procedure alpha_cases;
label found;
begin @<Cases for |alpha|@>;
incr(pp); {if no match was found, we move to the right}
found: end;
@ Now comes the code that tries to match each production starting
with a particular type of scrap. Whenever a match is discovered,
the |squash| or |reduce| macro will cause the appropriate action
to be performed, followed by |goto found|.
@<Cases for |alpha|@>=
if cat[pp+1]=math then
begin if cat[pp+2]=colon then squash(pp+1,2,math,0)(1)
else if cat[pp+2]=omega then
begin app1(pp); app(" "); app("$"); app1(pp+1);
app("$"); app(" "); app(indent); app1(pp+2);
reduce(pp,3,clause,-2)(2);
end;
end
else if cat[pp+1]=omega then
begin app1(pp); app(" "); app(indent); app1(pp+1);
reduce(pp,2,clause,-2)(3);
end
else if cat[pp+1]=simp then squash(pp+1,1,math,0)(4)
@ @<Cases for |beginning|@>=
if cat[pp+1]=close then
begin if (cat[pp+2]=terminator)or(cat[pp+2]=stmt) then
squash(pp,3,stmt,-2)(5);
end
else if cat[pp+1]=stmt then
begin app1(pp); app(break_space); app1(pp+1);
reduce(pp,2,beginning,-1)(6);
end
@ @<Cases for |case_head|@>=
if cat[pp+1]=casey then
begin if cat[pp+2]=clause then
begin app1(pp); app(outdent); app2(pp+1);
reduce(pp,3,case_head,0)(7);
end;
end
else if cat[pp+1]=close then
begin if cat[pp+2]=terminator then
begin app1(pp); app(cancel); app(outdent); app2(pp+1);
reduce(pp,3,stmt,-2)(8);
end;
end
else if cat[pp+1]=stmt then
begin app1(pp); app(force); app1(pp+1);
reduce(pp,2,case_head,0)(9);
end
@ @<Cases for |casey|@>=
if cat[pp+1]=clause then squash(pp,2,case_head,0)(10)
@ @<Cases for |clause|@>=
if cat[pp+1]=stmt then
begin app1(pp); app(break_space); app1(pp+1);
app(cancel); app(outdent);
app(force); reduce(pp,2,stmt,-2)(11);
end
@ @<Cases for |cond|@>=
if (cat[pp+1]=clause)and(cat[pp+2]=stmt) then
if cat[pp+3]=elsie then
begin app2(pp); app(break_space); app2(pp+2); app(" ");
app(cancel); reduce(pp,4,clause,-2)(12);
end
else begin app2(pp); app(break_space); app1(pp+2); app(cancel);
app(outdent); app(force); reduce(pp,3,stmt,-2)(13);
end
@ @<Cases for |elsie|@>=
squash(pp,1,intro,-3)(14)
@ @<Cases for |exp|@>=
if cat[pp+1]=math then
begin if cat[pp+2]=simp then if cat[pp+3]<>simp then
begin app3(pp); app("}"); reduce(pp,3,math,-1)(15);
end;
end
else if cat[pp+1]=simp then if cat[pp+2]<>simp then
begin app2(pp); app("}"); reduce(pp,2,math,-1)(16);
end
@ @<Cases for |intro|@>=
if cat[pp+1]=stmt then
begin app1(pp); app(" "); app(opt); app("7");
app(cancel); app1(pp+1); reduce(pp,2,stmt,-2)(17);
end
@ @<Cases for |math|@>=
if cat[pp+1]=close then
begin app("$"); app1(pp); app("$"); reduce(pp,1,stmt,-2)(18);
end
else if cat[pp+1]=colon then
begin app(force); app(backup); app("$"); app1(pp);
app("$"); app1(pp+1); reduce(pp,2,intro,-3)(19);
end
else if cat[pp+1]=math then squash(pp,2,math,-1)(20)
else if cat[pp+1]=simp then squash(pp,2,math,-1)(21)
else if cat[pp+1]=stmt then
begin app("$"); app1(pp); app("$"); app(indent);
app(break_space); app1(pp+1); app(cancel); app(outdent);
app(force); reduce(pp,2,stmt,-2)(22);
end
else if cat[pp+1]=terminator then
begin app("$"); app1(pp); app("$"); app1(pp+1);
reduce(pp,2,stmt,-2)(23);
end
@ @<Cases for |mod_scrap|@>=
if (cat[pp+1]=terminator)or(cat[pp+1]=semi) then
begin app2(pp); app(force); reduce(pp,2,stmt,-2)(24);
end
else squash(pp,1,simp,-2)(25)
@ @<Cases for |open|@>=
if (cat[pp+1]=case_head)and(cat[pp+2]=close) then
begin app1(pp); app("$"); app(cancel); app1(pp+1); app(cancel);
app(outdent); app("$"); app1(pp+2); reduce(pp,3,math,-1)(26);
end
else if cat[pp+1]=close then
begin app1(pp); app("\"); app(","); app1(pp+1);
@.\\,@>
reduce(pp,2,math,-1)(27);
end
else if cat[pp+1]=math then @<Cases for |open@,math|@>
else if cat[pp+1]=proc then
begin if cat[pp+2]=intro then
begin app(math_op); app(cancel); app1(pp+1); app("}");
reduce(pp+1,2,math,0)(34);
end;
end
else if cat[pp+1]=simp then squash(pp+1,1,math,0)(35)
else if (cat[pp+1]=stmt)and(cat[pp+2]=close) then
begin app1(pp); app("$"); app(cancel); app1(pp+1); app(cancel);
app("$"); app1(pp+2); reduce(pp,3,math,-1)(36);
end
else if cat[pp+1]=var_head then
begin if cat[pp+2]=intro then
begin app(math_op); app(cancel); app1(pp+1); app("}");
reduce(pp+1,2,math,0)(37);
end;
end
@ @<Cases for |open@,math|@>=
begin if (cat[pp+2]=case_head)and(cat[pp+3]=close) then
begin app2(pp); app("$"); app(cancel); app1(pp+2); app(cancel);
app(outdent); app("$"); app1(pp+3); reduce(pp,4,math,-1)(28);
end
else if cat[pp+2]=close then squash(pp,3,math,-1)(29)
else if cat[pp+2]=colon then squash(pp+1,2,math,0)(30)
else if cat[pp+2]=proc then
begin if cat[pp+3]=intro then
begin app1(pp+1); app(math_op); app(cancel);
app1(pp+2); app("}"); reduce(pp+1,3,math,0)(31);
end;
end
else if cat[pp+2]=semi then
begin app2(pp+1); app("\"); app(","); app(opt); app("5");
@.\\,@>
reduce(pp+1,2,math,0)(32);
end
else if cat[pp+2]=var_head then
begin if cat[pp+3]=intro then
begin app1(pp+1); app(math_op); app(cancel);
app1(pp+2); app("}"); reduce(pp+1,3,math,0)(31);
end;
end;
end
@ @<Cases for |proc|@>=
if cat[pp+1]=beginning then
begin if (cat[pp+2]=close)and(cat[pp+3]=terminator) then
begin app1(pp); app(cancel); app(outdent); app3(pp+1);
reduce(pp,4,stmt,-2)(38);
end;
end
else if cat[pp+1]=stmt then
begin app1(pp); app(break_space); app1(pp+1);
reduce(pp,2,proc,-2)(39);
end
@ @<Cases for |record_head|@>=
if (cat[pp+1]=intro)and(cat[pp+2]=casey) then
begin app2(pp); app(" "); app(cancel); app1(pp+2);
reduce(pp,3,casey,-2)(40);
end
else begin app(indent); app1(pp); app(cancel);
reduce(pp,1,case_head,0)(41);
end
@ @<Cases for |semi|@>=
squash(pp,1,terminator,-3)(42)
@ @<Cases for |simp|@>=
if cat[pp+1]=close then squash(pp,1,stmt,-2)(43)
else if cat[pp+1]=colon then
begin app(force); app(backup); app2(pp); reduce(pp,2,intro,-3)(44);
end
else if cat[pp+1]=math then squash(pp,2,math,-1)(45)
else if cat[pp+1]=mod_scrap then squash(pp,2,mod_scrap,0)(46)
else if cat[pp+1]=simp then squash(pp,2,simp,-2)(47)
else if cat[pp+1]=terminator then squash(pp,2,stmt,-2)(48)
@ @<Cases for |stmt|@>=
if cat[pp+1]=stmt then
begin app1(pp); app(break_space); app1(pp+1);
reduce(pp,2,stmt,-2)(49);
end
@ @<Cases for |terminator|@>=
squash(pp,1,stmt,-2)(50)
@ @<Cases for |var_head|@>=
if cat[pp+1]=beginning then squash(pp,1,stmt,-2)(51)
else if cat[pp+1]=math then
begin if cat[pp+2]=colon then
begin app("$"); app1(pp+1); app("$"); app1(pp+2);
reduce(pp+1,2,intro,+1)(52);
end;
end
else if cat[pp+1]=simp then
begin if cat[pp+2]=colon then squash(pp+1,2,intro,+1)(53);
end
else if cat[pp+1]=stmt then
begin app1(pp); app(break_space); app1(pp+1);
reduce(pp,2,var_head,-2)(54);
end
@ The `|freeze_text|' macro is used to give official status to a token list.
Before saying |freeze_text|, items are appended to the current token list,
and we know that the eventual number of this token list will be the current
value of |text_ptr|. But no list of that number really exists as yet,
because no ending point for the current list has been
stored in the |tok_start| array. After saying |freeze_text|, the
old current token list becomes legitimate, and its number is the current
value of |text_ptr-1| since |text_ptr| has been increased. The new
current token list is empty and ready to be appended to.
Note that |freeze_text| does not check to see that |text_ptr| hasn't gotten
too large, since it is assumed that this test was done beforehand.
@d freeze_text==incr(text_ptr); tok_start[text_ptr]:=tok_ptr
@ The `|reduce|' macro used in our code for productions actually calls on
a procedure named `|red|', which makes the appropriate changes to the
scrap list.
@p procedure red(@!j:sixteen_bits; @!k:eight_bits; @!c:eight_bits;
@!d:integer);
var i:0..max_scraps; {index into scrap memory}
begin cat[j]:=c; trans[j]:=text_ptr; freeze_text;
if k>1 then
begin for i:=j+k to lo_ptr do
begin cat[i-k+1]:=cat[i]; trans[i-k+1]:=trans[i];
end;
lo_ptr:=lo_ptr-k+1;
end;
@<Change |pp| to $\max(|scrap_base|,|pp+d|)$@>;
end;
@ @<Change |pp| to $\max(|scrap_base|,|pp+d|)$@>=
if pp+d>=scrap_base then pp:=pp+d
else pp:=scrap_base
@ Similarly, the `|squash|' macro invokes a procedure called `|sq|'. This
procedure takes advantage of the simplification that occurs when |k=1|.
@p procedure sq(@!j:sixteen_bits; @!k:eight_bits; @!c:eight_bits;
@!d:integer);
var i:0..max_scraps; {index into scrap memory}
begin if k=1 then
begin cat[j]:=c; @<Change |pp|...@>;
end
else begin for i:=j to j+k-1 do
begin app1(i);
end;
red(j,k,c,d);
end;
end;
@ Here now is the code that applies productions as long as possible. It
requires two local labels (|found| and |done|), as well as a local
variable (|i|).
@<Reduce the scraps using the productions until no more rules apply@>=
loop@+begin @<Make sure the entries |cat[pp..(pp+3)]| are defined@>;
if (tok_ptr+8>max_toks)or(text_ptr+4>max_texts) then
begin stat if tok_ptr>max_tok_ptr then max_tok_ptr:=tok_ptr;
if text_ptr>max_txt_ptr then max_txt_ptr:=text_ptr;
tats@;@/
overflow('token/text');
end;
if pp>lo_ptr then goto done;
@<Match a production...@>;
end;
done:
@ If we get to the end of the scrap list, category codes equal to zero are
stored, since zero does not match anything in a production.
@<Make sure the entries...@>=
if lo_ptr<pp+3 then
begin repeat if hi_ptr<=scrap_ptr then
begin incr(lo_ptr);@/
cat[lo_ptr]:=cat[hi_ptr]; trans[lo_ptr]:=trans[hi_ptr];@/
incr(hi_ptr);
end;
until (hi_ptr>scrap_ptr)or(lo_ptr=pp+3);
for i:=lo_ptr+1 to pp+3 do cat[i]:=0;
end
@ If \.{WEAVE} is being run in debugging mode, the production numbers and
current stack categories will be printed out when |tracing| is set to 2;
a sequence of two or more irreducible scraps will be printed out when
|tracing| is set to 1.
@.\AT!2@>
@.\AT!1@>
@<Glo...@>=
@!debug@!tracing:0..2; {can be used to show parsing details}
gubed
@ The |prod| procedure is called in debugging mode just after |reduce| or
|squash|; its parameter is the number of the production that has just
been applied.
@p @!debug procedure prod(@!n:eight_bits); {shows current categories}
var k:1..max_scraps; {index into |cat|}
begin if tracing=2 then
begin print_nl(n:1,':');
for k:=scrap_base to lo_ptr do
begin if k=pp then print('*') @+ else print(' ');
print_cat(cat[k]);
end;
if hi_ptr<=scrap_ptr then print('...'); {indicate that more is coming}
end;
end;
gubed
@ The |translate| function assumes that scraps have been stored in
positions |scrap_base| through |scrap_ptr| of |cat| and |trans|. It
appends a |terminator| scrap and begins to apply productions as much as
possible. The result is a token list containing the translation of
the given sequence of scraps.
After calling |translate|, we will have |text_ptr+3<=max_texts| and
|tok_ptr+6<=max_toks|, so it will be possible to create up to three token
lists with up to six tokens without checking for overflow. Before calling
|translate|, we should have |text_ptr<max_texts| and |scrap_ptr<max_scraps|,
since |translate| might add a new text and a new scrap before it checks
for overflow.
@p @<Declaration of subprocedures for |translate|@>@;
function translate:text_pointer; {converts a sequence of scraps}
label done,found;
var i: 1..max_scraps; {index into |cat|}
@!j:0..max_scraps; {runs through final scraps}
@!k:0..long_buf_size; {index into |buffer|}
begin pp:=scrap_base; lo_ptr:=pp-1; hi_ptr:=pp;
@<If tracing, print an indication of where we are@>;
@<Reduce the scraps...@>;
if (lo_ptr=scrap_base)and(cat[lo_ptr]<>math) then translate:=trans[lo_ptr]
else @<Combine the irreducible scraps that remain@>;
end;
@ If the initial sequence of scraps does not reduce to a single scrap,
we concatenate the translations of all remaining scraps, separated by
blank spaces, with dollar signs surrounding the translations of |math|
scraps.
@<Combine the irreducible...@>=
begin @<If semi-tracing, show the irreducible scraps@>;
for j:=scrap_base to lo_ptr do
begin if j<>scrap_base then
begin app(" ");
end;
if cat[j]=math then
begin app("$");
end;
app1(j);
if cat[j]=math then
begin app("$");
end;
if tok_ptr+6>max_toks then overflow('token');
end;
freeze_text; translate:=text_ptr-1;
end
@ @<If semi-tracing, show the irreducible scraps@>=
@!debug if (lo_ptr>scrap_base)and(tracing=1) then
begin print_nl('Irreducible scrap sequence in section ',module_count:1);
print_ln(':'); mark_harmless;
for j:=scrap_base to lo_ptr do
begin print(' '); print_cat(cat[j]);
end;
end;
gubed
@ @<If tracing,...@>=
@!debug if tracing=2 then
begin print_nl('Tracing after l.',line:1,':'); mark_harmless;
if loc>50 then
begin print('...');
for k:=loc-50 to loc do print(xchr[buffer[k-1]]);
end
else for k:=1 to loc do print(xchr[buffer[k-1]]);
end
gubed
@* Initializing the scraps.
If we are going to use the powerful production mechanism just developed, we
must get the scraps set up in the first place, given a \PASCAL\ text. A table
of the initial scraps corresponding to \PASCAL\ tokens appeared above in the
section on parsing; our goal now is to implement that table. We shall do this
by implementing a subroutine called |Pascal_parse| that is analogous to the
|Pascal_xref| routine used during phase one.
Like |Pascal_xref|, the |Pascal_parse| procedure starts with the current
value of |next_control| and it uses the operation |next_control:=get_next|
repeatedly to read \PASCAL\ text until encountering the next `\v' or
`\.\{', or until |next_control>=format|. The scraps corresponding to what
it reads are appended into the |cat| and |trans| arrays, and |scrap_ptr|
is advanced.
Like |prod|, this procedure has to split into pieces so that each
part is short enough to be handled by \PASCAL\ compilers that discriminate
against long subroutines. This time there are two split-off routines,
called |easy_cases| and |sub_cases|.
@^split procedures@>
After studying |Pascal_parse|, we will look at the sub-procedures
|app_comment|, |app_octal|, and |app_hex| that are used in some of its
branches.
@p @<Declaration of the |app_comment| procedure@>@;
@<Declaration of the |app_octal| and |app_hex| procedures@>@;
@<Declaration of the |easy_cases| procedure@>@;
@<Declaration of the |sub_cases| procedure@>@;
procedure Pascal_parse; {creates scraps from \PASCAL\ tokens}
label reswitch, exit;
var j:0..long_buf_size; {index into |buffer|}
@!p:name_pointer; {identifier designator}
begin while next_control<format do
begin @<Append the scrap appropriate to |next_control|@>;
next_control:=get_next;
if (next_control="|")or(next_control="{") then return;
end;
exit:end;
@ The macros defined here are helpful abbreviations for the operations
needed when generating the scraps. A scrap of category |c| whose
translation has three tokens $t_1$, $t_2$, $t_3$ is generated by
|sc3|$(t_1)(t_2)(t_3)(c)$, etc.
@d s0(#)==incr(scrap_ptr); cat[scrap_ptr]:=#; trans[scrap_ptr]:=text_ptr;
freeze_text;
end
@d s1(#)==app(#);s0
@d s2(#)==app(#);s1
@d s3(#)==app(#);s2
@d s4(#)==app(#);s3
@d sc4==@+begin s4
@d sc3==@+begin s3
@d sc2==@+begin s2
@d sc1==@+begin s1
@d sc0(#)==begin incr(scrap_ptr); cat[scrap_ptr]:=#; trans[scrap_ptr]:=0;
end
@d comment_scrap(#)==begin app(#); app_comment;
end
@ @<Append the scr...@>=
@<Make sure that there is room for at least four more scraps, six more
tokens, and four more texts@>;
reswitch: case next_control of
string,verbatim: @<Append a \(string scrap@>;
identifier: @<Append an identifier scrap@>;
TeX_string: @<Append a \TeX\ string scrap@>;
othercases easy_cases
endcases
@ The |easy_cases| each result in straightforward scraps.
@<Declaration of the |easy_cases| procedure@>=
procedure easy_cases; {a subprocedure of |Pascal_parse|}
begin case next_control of
set_element_sign: sc3("\")("i")("n")(math);
@.\\in@>
double_dot: sc3("\")("t")("o")(math);
@.\\to@>
"#","$","%","^","_": sc2("\")(next_control)(math);
@.\\\#@>
@.\\\$@>
@.\\\%@>
@.\\\^@>
ignore,"|",xref_roman,xref_wildcard,xref_typewriter: do_nothing;
"(","[": sc1(next_control)(open);
")","]": sc1(next_control)(close);
"*": sc4("\")("a")("s")("t")(math);
@.\\ast@>
",": sc3(",")(opt)("9")(math);
".","0","1","2","3","4","5","6","7","8","9": sc1(next_control)(simp);
";": sc1(";")(semi);
":": sc1(":")(colon);
@t\4@> @<Cases involving nonstandard ASCII characters@>@;
exponent: sc3("\")("E")("{")(exp);
@.\\E@>
begin_comment: sc2("\")("B")(math);
@.\\B@>
end_comment: sc2("\")("T")(math);
@.\\T@>
octal: app_octal;
hex: app_hex;
check_sum: sc2("\")(")")(simp);
@.\\)@>
force_line: sc2("\")("]")(simp);
@.\\]@>
thin_space: sc2("\")(",")(math);
@.\\,@>
math_break: sc2(opt)("0")(simp);
line_break: comment_scrap(force);
big_line_break: comment_scrap(big_force);
no_line_break: begin app(big_cancel); app("\"); app(" ");
@.\\\ @>
comment_scrap(big_cancel);
end;
pseudo_semi: sc0(semi);
join: sc2("\")("J")(math);
@.\\J@>
othercases sc1(next_control)(math)
endcases;
end;
@ @<Make sure that there is room for at least four...@>=
if (scrap_ptr+4>max_scraps)or(tok_ptr+6>max_toks)or(text_ptr+4>max_texts) then
begin stat if scrap_ptr>max_scr_ptr then max_scr_ptr:=scrap_ptr;
if tok_ptr>max_tok_ptr then max_tok_ptr:=tok_ptr;
if text_ptr>max_txt_ptr then max_txt_ptr:=text_ptr;
tats@;@/
overflow('scrap/token/text');
end
@ Some nonstandard ASCII characters may have entered \.{WEAVE} by means of
standard ones. They are converted to \TeX\ control sequences so that it is
possible to keep \.{WEAVE} from stepping beyond standard ASCII.
@<Cases involving nonstandard...@>=
not_equal: sc2("\")("I")(math);
@.\\I@>
less_or_equal: sc2("\")("L")(math);
@.\\L@>
greater_or_equal: sc2("\")("G")(math);
@.\\G@>
equivalence_sign: sc2("\")("S")(math);
@.\\S@>
and_sign: sc2("\")("W")(math);
@.\\W@>
or_sign: sc2("\")("V")(math);
@.\\V@>
not_sign: sc2("\")("R")(math);
@.\\R@>
left_arrow: sc2("\")("K")(math);
@.\\K@>
@ The following code must use |app_tok| instead of |app| in order to
protect against overflow. Note that |tok_ptr+1<=max_toks| after |app_tok|
has been used, so another |app| is legitimate before testing again.
Many of the special characters in a string must be prefixed by `\.\\' so that
\TeX\ will print them properly.
@^special string characters@>
@<Append a \(string scrap@>=
begin app("\");
if next_control=verbatim then
begin app("=");
@.\\=@>
end
else begin app(".");
@.\\.@>
end;
app("{"); j:=id_first;
while j<id_loc do
begin case buffer[j] of
" ","\","#","%","$","^","'","`","{","}","~","&","_":
begin app("\");
end;
@.\\\ @>
@.\\\\@>
@.\\\#@>
@.\\\%@>
@.\\\$@>
@.\\\^@>
@.\\\'@>
@.\\\`@>
@.\\\{@>
@.\\\}@>
@.\\\~@>
@.\\\&@>
@.\\_@>
"@@": if buffer[j+1]="@@" then incr(j)
else err_print('! Double @@ should be used in strings');
@.Double \AT! should be used...@>
othercases do_nothing
endcases;@/
app_tok(buffer[j]); incr(j);
end;
sc1("}")(simp);
end
@ @<Append a \TeX\ string scrap@>=
begin app("\"); app("h"); app("b"); app("o"); app("x");
app("{");
for j:=id_first to id_loc-1 do app_tok(buffer[j]);
sc1("}")(simp);
end
@ @<Append an identifier scrap@>=
begin p:=id_lookup(normal);
case ilk[p] of
normal,array_like,const_like,div_like,
do_like,for_like,goto_like,nil_like,to_like: sub_cases(p);
@t\4@>@<Cases that generate more than one scrap@>@;
othercases begin next_control:=ilk[p]-char_like; goto reswitch;
end {\&{and}, \&{in}, \&{not}, \&{or}}
endcases;
end
@ The |sub_cases| also result in straightforward scraps.
@<Declaration of the |sub_cases| procedure@>=
procedure sub_cases(@!p:name_pointer); {a subprocedure of |Pascal_parse|}
begin case ilk[p] of
normal: sc1(id_flag+p)(simp); {not a reserved word}
array_like: sc1(res_flag+p)(alpha); {\&{array}, \&{file}, \&{set}}
const_like: sc3(force)(backup)(res_flag+p)(intro);
{\&{const}, \&{label}, \&{type}}
div_like: sc3(math_bin)(res_flag+p)("}")(math); {\&{div}, \&{mod}}
do_like: sc1(res_flag+p)(omega); {\&{do}, \&{of}, \&{then}}
for_like: sc2(force)(res_flag+p)(alpha); {\&{for}, \&{while}, \&{with}}
goto_like: sc1(res_flag+p)(intro); {\&{goto}, \&{packed}}
nil_like: sc1(res_flag+p)(simp); {\&{nil}}
to_like: sc3(math_rel)(res_flag+p)("}")(math); {\&{downto}, \&{to}}
end;
end;
@ @<Cases that generate more than one scrap@>=
begin_like: begin sc3(force)(res_flag+p)(cancel)(beginning); sc0(intro);
end; {\&{begin}}
case_like: begin sc0(casey); sc2(force)(res_flag+p)(alpha);
end; {\&{case}}
else_like: begin @<Append |terminator| if not already present@>;
sc3(force)(backup)(res_flag+p)(elsie);
end; {\&{else}}
end_like: begin @<Append |term...@>;
sc2(force)(res_flag+p)(close);
end; {\&{end}}
if_like: begin sc0(cond); sc2(force)(res_flag+p)(alpha);
end; {\&{if}}
loop_like: begin sc3(force)("\")("~")(alpha);
@.\\\~@>
sc1(res_flag+p)(omega);
end; {\&{xclause}}
proc_like: begin sc4(force)(backup)(res_flag+p)(cancel)(proc);
sc3(indent)("\")(" ")(intro);
@.\\\ @>
end; {\&{function}, \&{procedure}, \&{program}}
record_like: begin sc1(res_flag+p)(record_head); sc0(intro);
end; {\&{record}}
repeat_like: begin sc4(force)(indent)(res_flag+p)(cancel)(beginning);
sc0(intro);
end; {\&{repeat}}
until_like: begin @<Append |term...@>;
sc3(force)(backup)(res_flag+p)(close); sc0(clause);
end; {\&{until}}
var_like: begin sc4(force)(backup)(res_flag+p)(cancel)(var_head); sc0(intro);
end; {\&{var}}
@ If a comment or semicolon appears before the reserved words \&{end},
\&{else}, or \&{until}, the |semi| or |terminator| scrap that is already
present overrides the |terminator| scrap belonging to this reserved word.
@<Append |termin...@>=
if (scrap_ptr<scrap_base)or((cat[scrap_ptr]<>terminator)and
(cat[scrap_ptr]<>semi)) then sc0(terminator)
@ A comment is incorporated into the previous scrap if that scrap is of type
|omega| or |semi| or |terminator|. (These three categories have consecutive
category codes.) Otherwise the comment is entered as a separate scrap
of type |terminator|, and it will combine with a |terminator| scrap that
immediately follows~it.
The |app_comment| procedure takes care of placing a comment at the end of the
current scrap list. When |app_comment| is called, we assume that the current
token list is the translation of the comment involved.
@<Declaration of the |app_comment|...@>=
procedure app_comment; {append a comment to the scrap list}
begin freeze_text;
if (scrap_ptr<scrap_base)or(cat[scrap_ptr]<omega)or
(cat[scrap_ptr]>terminator) then sc0(terminator)
else begin app1(scrap_ptr); {|cat[scrap_ptr]| is
|omega| or |semi| or |terminator|}
end;
app(text_ptr-1+tok_flag); trans[scrap_ptr]:=text_ptr; freeze_text;
end;
@ We are now finished with |Pascal_parse|, except for two relatively
trivial subprocedures that convert constants into tokens.
@<Declaration of the |app_octal| and...@>=
procedure app_octal;
begin app("\"); app("O"); app("{");
@.\\O@>
while (buffer[loc]>="0")and(buffer[loc]<="7") do
begin app_tok(buffer[loc]); incr(loc);
end;
sc1("}")(simp);
end;
@#
procedure app_hex;
begin app("\"); app("H"); app("{");
@.\\H@>
while ((buffer[loc]>="0")and(buffer[loc]<="9"))or@|
((buffer[loc]>="A")and(buffer[loc]<="F")) do
begin app_tok(buffer[loc]); incr(loc);
end;
sc1("}")(simp);
end;
@ When the `\v' that introduces \PASCAL\ text is sensed, a call on
|Pascal_translate| will return a pointer to the \TeX\ translation of
that text. If scraps exist in the |cat| and |trans| arrays, they are
unaffected by this translation process.
@p function Pascal_translate: text_pointer;
var p:text_pointer; {points to the translation}
@!save_base:0..max_scraps; {holds original value of |scrap_base|}
begin save_base:=scrap_base; scrap_base:=scrap_ptr+1;
Pascal_parse; {get the scraps together}
if next_control<>"|" then err_print('! Missing "|" after Pascal text');
@.Missing "|"...@>
app_tok(cancel); app_comment; {place a |cancel| token as a final ``comment''}
p:=translate; {make the translation}
stat if scrap_ptr>max_scr_ptr then max_scr_ptr:=scrap_ptr;@;@+tats@;@/
scrap_ptr:=scrap_base-1; scrap_base:=save_base; {scrap the scraps}
Pascal_translate:=p;
end;
@ The |outer_parse| routine is to |Pascal_parse| as |outer_xref|
is to |Pascal_xref|: It constructs a sequence of scraps for \PASCAL\ text
until |next_control>=format|. Thus, it takes care of embedded comments.
@p procedure outer_parse; {makes scraps from \PASCAL\ tokens and comments}
var bal:eight_bits; {brace level in comment}
@!p,@!q:text_pointer; {partial comments}
begin while next_control<format do
if next_control<>"{" then Pascal_parse
else begin @<Make sure that there is room for at least seven more
tokens, three more texts, and one more scrap@>;
app("\"); app("C"); app("{");
@.\\C@>
bal:=copy_comment(1); next_control:="|";
while bal>0 do
begin p:=text_ptr; freeze_text; q:=Pascal_translate;
{at this point we have |tok_ptr+6<=max_toks|}
app(tok_flag+p); app(inner_tok_flag+q);
if next_control="|" then bal:=copy_comment(bal)
else bal:=0; {an error has been reported}
end;
app(force); app_comment; {the full comment becomes a scrap}
end;
end;
@ @<Make sure that there is room for at least seven more...@>=
if (tok_ptr+7>max_toks)or(text_ptr+3>max_texts)or(scrap_ptr>=max_scraps) then
begin stat if scrap_ptr>max_scr_ptr then max_scr_ptr:=scrap_ptr;
if tok_ptr>max_tok_ptr then max_tok_ptr:=tok_ptr;
if text_ptr>max_txt_ptr then max_txt_ptr:=text_ptr;
tats@;@/
overflow('token/text/scrap');
end
@* Output of tokens.
So far our programs have only built up multi-layered token lists in
\.{WEAVE}'s internal memory; we have to figure out how to get them into
the desired final form. The job of converting token lists to characters in
the \TeX\ output file is not difficult, although it is an implicitly
recursive process. Four main considerations had to be kept in mind when
this part of \.{WEAVE} was designed. (a) There are two modes of output:
|outer| mode, which translates tokens like |force| into line-breaking
control sequences, and |inner| mode, which ignores them except that blank
spaces take the place of line breaks. (b) The |cancel| instruction applies
to adjacent token or tokens that are output, and this cuts across levels
of recursion since `|cancel|' occurs at the beginning or end of a token
list on one level. (c) The \TeX\ output file will be semi-readable if line
breaks are inserted after the result of tokens like |break_space| and
|force|. (d) The final line break should be suppressed, and there should
be no |force| token output immediately after `\.{\\Y\\P}'.
@ The output process uses a stack to keep track of what is going on at
different ``levels'' as the token lists are being written out. Entries on
this stack have three parts:
\yskip\hang |end_field| is the |tok_mem| location where the token list of a
particular level will end;
\yskip\hang |tok_field| is the |tok_mem| location from which the next token
on a particular level will be read;
\yskip\hang |mode_field| is the current mode, either |inner| or |outer|.
\yskip\noindent The current values of these quantities are referred to
quite frequently, so they are stored in a separate place instead of in the
|stack| array. We call the current values |cur_end|, |cur_tok|, and
|cur_mode|.
The global variable |stack_ptr| tells how many levels of output are
currently in progress. The end of output occurs when an |end_translation|
token is found, so the stack is never empty except when we first begin the
output process.
@d inner=0 {value of |mode| for \PASCAL\ texts within \TeX\ texts}
@d outer=1 {value of |mode| for \PASCAL\ texts in modules}
@<Types...@>=
@!mode=inner..outer;@/
@!output_state=record@!end_field:sixteen_bits; {ending location of token list}
@!tok_field:sixteen_bits; {present location within token list}
@!mode_field:mode; {interpretation of control tokens}
end;
@ @d cur_end==cur_state.end_field {current ending location in |tok_mem|}
@d cur_tok==cur_state.tok_field {location of next output token in |tok_mem|}
@d cur_mode==cur_state.mode_field {current mode of interpretation}
@d init_stack==stack_ptr:=0;cur_mode:=outer {do this to initialize the stack}
@<Glob...@>=
@!cur_state:output_state; {|cur_end|, |cur_tok|, |cur_mode|}
@!stack:array[1..stack_size] of output_state; {info for non-current levels}
@!stack_ptr:0..stack_size; {first unused location in the output state stack}
stat@!max_stack_ptr:0..stack_size; {largest value assumed by |stack_ptr|}
tats
@ @<Set init...@>=stat max_stack_ptr:=0;@+tats
@ To insert token-list |p| into the output, the |push_level| subroutine
is called; it saves the old level of output and gets a new one going.
The value of |cur_mode| is not changed.
@p procedure push_level(@!p:text_pointer); {suspends the current level}
begin if stack_ptr=stack_size then overflow('stack')
else begin if stack_ptr>0 then
stack[stack_ptr]:=cur_state; {save |cur_end|$\,\ldots\,$|cur_mode|}
incr(stack_ptr);
stat if stack_ptr>max_stack_ptr then
max_stack_ptr:=stack_ptr;@;@+tats@;@/
cur_tok:=tok_start[p]; cur_end:=tok_start[p+1];
end;
end;
@ Conversely, the |pop_level| routine restores the conditions that were in
force when the current level was begun. This subroutine will never be
called when |stack_ptr=1|. It is so simple, we declare it as a macro:
@d pop_level==begin decr(stack_ptr); cur_state:=stack[stack_ptr];
end {do this when |cur_tok| reaches |cur_end|}
@ The |get_output| function returns the next byte of output that is not a
reference to a token list. It returns the values |identifier| or |res_word|
or |mod_name| if the next token is to be an identifier (typeset in
italics), a reserved word (typeset in boldface) or a module name (typeset
by a complex routine that might generate additional levels of output).
In these cases |cur_name| points to the identifier or module name in
question.
@d res_word=@'201 {returned by |get_output| for reserved words}
@d mod_name=@'200 {returned by |get_output| for module names}
@p function get_output:eight_bits; {returns the next token of output}
label restart;
var a:sixteen_bits; {current item read from |tok_mem|}
begin restart: while cur_tok=cur_end do pop_level;
a:=tok_mem[cur_tok]; incr(cur_tok);
if a>=@'400 then
begin cur_name:=a mod id_flag;
case a div id_flag of
2: a:=res_word; {|a=res_flag+cur_name|}
3: a:=mod_name; {|a=mod_flag+cur_name|}
4: begin push_level(cur_name); goto restart;
end; {|a=tok_flag+cur_name|}
5: begin push_level(cur_name); cur_mode:=inner; goto restart;
end; {|a=inner_tok_flag+cur_name|}
othercases a:=identifier {|a=id_flag+cur_name|}
endcases;
end;
@!debug if trouble_shooting then debug_help; @+ gubed@/
get_output:=a;
end;
@ The real work associated with token output is done by |make_output|.
This procedure appends an |end_translation| token to the current token list,
and then it repeatedly calls |get_output| and feeds characters to the output
buffer until reaching the |end_translation| sentinel. It is possible for
|make_output| to
be called recursively, since a module name may include embedded \PASCAL\
text; however, the depth of recursion never exceeds one level, since
module names cannot be inside of module names.
A procedure called |output_Pascal| does the scanning, translation, and
output of \PASCAL\ text within `\pb' brackets, and this procedure uses
|make_output| to output the current token list. Thus, the recursive call
of |make_output| actually occurs when |make_output| calls |output_Pascal|
while outputting the name of a module.
@^recursion@>
@p procedure make_output; forward; @t\2@>@#
procedure output_Pascal; {outputs the current token list}
var save_tok_ptr,@!save_text_ptr,@!save_next_control:sixteen_bits;
{values to be restored}
p:text_pointer; {translation of the \PASCAL\ text}
begin save_tok_ptr:=tok_ptr; save_text_ptr:=text_ptr;
save_next_control:=next_control; next_control:="|"; p:=Pascal_translate;
app(p+inner_tok_flag);
make_output; {output the list}
stat if text_ptr>max_txt_ptr then max_txt_ptr:=text_ptr;
if tok_ptr>max_tok_ptr then max_tok_ptr:=tok_ptr;@;@+tats@;@/
text_ptr:=save_text_ptr; tok_ptr:=save_tok_ptr; {forget the tokens}
next_control:=save_next_control; {restore |next_control| to original state}
end;
@ Here is \.{WEAVE}'s major output handler.
@p procedure make_output; {outputs the equivalents of tokens}
label reswitch,exit,found;
var a:eight_bits; {current output byte}
@!b:eight_bits; {next output byte}
@!k,@!k_limit:0..max_bytes; {indices into |byte_mem|}
@!w:0..ww-1; {row of |byte_mem|}
@!j:0..long_buf_size; {index into |buffer|}
@!string_delimiter:ASCII_code; {first and last character of
string being copied}
@!save_loc,@!save_limit:0..long_buf_size; {|loc| and |limit| to be restored}
@!cur_mod_name:name_pointer; {name of module being output}
@!save_mode:mode; {value of |cur_mode| before a sequence of breaks}
begin app(end_translation); {append a sentinel}
freeze_text; push_level(text_ptr-1);
loop@+ begin a:=get_output;
reswitch: case a of
end_translation: return;
identifier,res_word:@<Output an identifier@>;
mod_name:@<Output a module name@>;
math_bin,math_op,math_rel:@<Output a \.{\\math} operator@>;
cancel: begin repeat a:=get_output;
until (a<backup)or(a>big_force);
goto reswitch;
end;
big_cancel: begin repeat a:=get_output;
until ((a<backup)and(a<>" "))or(a>big_force);
goto reswitch;
end;
indent,outdent,opt,backup,break_space,force,big_force:@<Output a
\(control, look ahead in case of line breaks,
possibly |goto reswitch|@>;
othercases out(a) {otherwise |a| is an ASCII character}
endcases;
end;
exit:end;
@ An identifier of length one does not have to be enclosed in braces, and it
looks slightly better if set in a math-italic font instead of a (slightly
narrower) text-italic font. Thus we output `\.{\\\char'174a}' but
`\.{\\\\\{aa\}}'.
@<Output an identifier@>=
begin out("\");
if a=identifier then
if length(cur_name)=1 then out("|")
@.\\|@>
else out("\")
@.\\\\@>
else out("&"); {|a=res_word|}
@.\\\&@>
if length(cur_name)=1 then out(byte_mem[cur_name mod ww,byte_start[cur_name]])
else out_name(cur_name);
end
@ @<Output a \....@>=
begin out5("\")("m")("a")("t")("h");
if a=math_bin then out3("b")("i")("n")
else if a=math_rel then out3("r")("e")("l")
else out2("o")("p");
out("{");
end
@ The current mode does not affect the behavior of \.{WEAVE}'s output routine
except when we are outputting control tokens.
@<Output a \(control...@>=
if a<break_space then
begin if cur_mode=outer then
begin out2("\")(a-cancel+"0");
@.\\1@>
@.\\2@>
@.\\3@>
@.\\4@>
@.\\5@>
@.\\6@>
@.\\7@>
if a=opt then out(get_output) {|opt| is followed by a digit}
end
else if a=opt then b:=get_output {ignore digit following |opt|}
end
else @<Look ahead for strongest line break, |goto reswitch|@>
@ If several of the tokens |break_space|, |force|, |big_force| occur in a
row, possibly mixed with blank spaces (which are ignored),
the largest one is used. A line break also occurs in the output file,
except at the very end of the translation. The very first line break
is suppressed (i.e., a line break that follows `\.{\\Y\\P}').
@<Look ahead for st...@>=
begin b:=a; save_mode:=cur_mode;
loop@+ begin a:=get_output;
if (a=cancel)or(a=big_cancel) then goto reswitch;
{|cancel| overrides everything}
if ((a<>" ")and(a<break_space))or(a>big_force) then
begin if save_mode=outer then
begin if out_ptr>3 then
if (out_buf[out_ptr]="P")and
(out_buf[out_ptr-1]="\")and
@.\\P@>
@.\\Y@>
(out_buf[out_ptr-2]="Y")and
(out_buf[out_ptr-3]="\") then
goto reswitch;
@.\\1@>
@.\\2@>
@.\\3@>
@.\\4@>
@.\\5@>
@.\\6@>
@.\\7@>
out2("\")(b-cancel+"0");
if a<>end_translation then finish_line;
end
else if (a<>end_translation)and(cur_mode=inner) then out(" ");
goto reswitch;
end;
if a>b then b:=a; {if |a=" "| we have |a<b|}
end;
end
@ The remaining part of |make_output| is somewhat more complicated. When we
output a module name, we may need to enter the parsing and translation
routines, since the name may contain \PASCAL\ code embedded in
\pb\ constructions. This \PASCAL\ code is placed at the end of the active
input buffer and the translation process uses the end of the active
|tok_mem| area.
@<Output a module name@>=
begin out2("\")("X");
@.\\X@>
cur_xref:=xref[cur_name];
if num(cur_xref)>=def_flag then
begin out_mod(num(cur_xref)-def_flag);
if phase_three then
begin cur_xref:=xlink(cur_xref);
while num(cur_xref)>=def_flag do
begin out2(",")(" ");
out_mod(num(cur_xref)-def_flag);
cur_xref:=xlink(cur_xref);
end;
end;
end
else out("0"); {output the module number, or zero if it was undefined}
out(":"); @<Output the text of the module name@>;
out2("\")("X");
end
@ @<Output the text...@>=
k:=byte_start[cur_name]; w:=cur_name mod ww; k_limit:=byte_start[cur_name+ww];
cur_mod_name:=cur_name;
while k<k_limit do
begin b:=byte_mem[w,k]; incr(k);
if b="@@" then @<Skip next character, give error if not `\.{@@}'@>;
if b<>"|" then out(b)
else begin @<Copy the \PASCAL\ text into |buffer[(limit+1)..j]|@>;
save_loc:=loc; save_limit:=limit; loc:=limit+2; limit:=j+1;
buffer[limit]:="|"; output_Pascal;
loc:=save_loc; limit:=save_limit;
end;
end
@ @<Skip next char...@>=
begin if byte_mem[w,k]<>"@@" then
begin print_nl('! Illegal control code in section name:');
@.Illegal control code...@>
print_nl('<'); print_id(cur_mod_name); print('> '); mark_error;
end;
incr(k);
end
@ The \PASCAL\ text enclosed in \pb\ should not contain `\v' characters,
except within strings. We put a `\v' at the front of the buffer, so that an
error message that displays the whole buffer will look a little bit sensible.
The variable |string_delimiter| is zero outside of strings, otherwise it
equals the delimiter that began the string being copied.
@<Copy the \PASCAL\ text into...@>=
j:=limit+1; buffer[j]:="|"; string_delimiter:=0;
loop@+ begin if k>=k_limit then
begin print_nl('! Pascal text in section name didn''t end:');
@.Pascal text...didn't end@>
print_nl('<'); print_id(cur_mod_name); print('> '); mark_error;
goto found;
end;
b:=byte_mem[w,k]; incr(k);
if b="@@" then @<Copy a control code into the buffer@>
else begin if (b="""")or(b="'") then
if string_delimiter=0 then string_delimiter:=b
else if string_delimiter=b then string_delimiter:=0;
if (b<>"|")or(string_delimiter<>0) then
begin if j>long_buf_size-3 then overflow('buffer');
incr(j); buffer[j]:=b;
end
else goto found;
end;
end;
found:
@ @<Copy a control code into the buffer@>=
begin if j>long_buf_size-4 then overflow('buffer');
buffer[j+1]:="@@"; buffer[j+2]:=byte_mem[w,k]; j:=j+2; incr(k);
end
@* Phase two processing.
We have assembled enough pieces of the puzzle in order to be ready to specify
the processing in \.{WEAVE}'s main pass over the source file. Phase two
is analogous to phase one, except that more work is involved because we must
actually output the \TeX\ material instead of merely looking at the
\.{WEB} specifications.
@<Phase II: Read all the text again and translate it to \TeX\ form@>=
reset_input; print_nl('Writing the output file...');
module_count:=0;
copy_limbo;
finish_line; flush_buffer(0,false,false); {insert a blank line, it looks nice}
while not input_has_ended do @<Translate the \(current module@>
@ The output file will contain the control sequence \.{\\Y} between non-null
sections of a module, e.g., between the \TeX\ and definition parts if both
are nonempty. This puts a little white space between the parts when they are
printed. However, we don't want \.{\\Y} to occur between two definitions
within a single module. The variables |out_line| or |out_ptr| will
change if a section is non-null, so the following macros `|save_position|'
and `|emit_space_if_needed|' are able to handle the situation:
@d save_position==save_line:=out_line; save_place:=out_ptr
@d emit_space_if_needed==if (save_line<>out_line)or(save_place<>out_ptr) then
out2("\")("Y")
@.\\Y@>
@<Glo...@>=
@!save_line:integer; {former value of |out_line|}
@!save_place:sixteen_bits; {former value of |out_ptr|}
@ @<Translate the \(current module@>=
begin incr(module_count);@/
@<Output the code for the beginning of a new module@>;
save_position;@/
@<Translate the \TeX\ part of the current module@>;
@<Translate the \(definition part of the current module@>;
@<Translate the \PASCAL\ part of the current module@>;
@<Show cross references to this module@>;
@<Output the code for the end of a module@>;
end
@ Modules beginning with the \.{WEB} control sequence `\.{@@\ }' start in the
output with the \TeX\ control sequence `\.{\\M}', followed by the module
number. Similarly, `\.{@@*}' modules lead to the control sequence `\.{\\N}'.
If this is a changed module, we put \.{*} just before the module number.
@<Output the code for the beginning...@>=
out("\");
if buffer[loc-1]<>"*" then out("M")
@.\\M@>
else begin out("N"); print('*',module_count:1);
@.\\N@>
update_terminal; {print a progress report}
end;
out_mod(module_count); out2(".")(" ")
@ In the \TeX\ part of a module, we simply copy the source text, except that
index entries are not copied and \PASCAL\ text within \pb\ is translated.
@<Translate the \T...@>=
repeat next_control:=copy_TeX;
case next_control of
"|": begin init_stack; output_Pascal;
end;
"@@": out("@@");
octal: @<Translate an octal constant appearing in \TeX\ text@>;
hex: @<Translate a hexadecimal constant appearing in \TeX\ text@>;
TeX_string,xref_roman,xref_wildcard,xref_typewriter,module_name:
begin loc:=loc-2; next_control:=get_next; {skip to \.{@@>}}
if next_control=TeX_string then
err_print('! TeX string should be in Pascal text only');
@.TeX string should be...@>
end;
begin_comment,end_comment,check_sum,thin_space,math_break,line_break,
big_line_break,no_line_break,join,pseudo_semi:
err_print('! You can''t do that in TeX text');
@.You can't do that...@>
othercases do_nothing
endcases;
until next_control>=format
@ @<Translate an octal constant appearing in \TeX\ text@>=
begin out3("\")("O")("{");
@.\\O@>
while (buffer[loc]>="0")and(buffer[loc]<="7") do
begin out(buffer[loc]); incr(loc);
end; {since |buffer[limit]=" "|, this loop will end}
out("}");
end
@ @<Translate a hexadecimal constant appearing in \TeX\ text@>=
begin out3("\")("H")("{");
@.\\H@>
while ((buffer[loc]>="0")and(buffer[loc]<="9"))or@|
((buffer[loc]>="A")and(buffer[loc]<="F")) do
begin out(buffer[loc]); incr(loc);
end;
out("}");
end
@ When we get to the following code we have |next_control>=format|, and
the token memory is in its initial empty state.
@<Translate the \(d...@>=
if next_control<=definition then {definition part non-empty}
begin emit_space_if_needed; save_position;
end;
while next_control<=definition do {|format| or |definition|}
begin init_stack;
if next_control=definition then @<Start a macro definition@>
else @<Start a format definition@>;
outer_parse; finish_Pascal;
end
@ The |finish_Pascal| procedure outputs the translation of the current
scraps, preceded by the control sequence `\.{\\P}' and followed by the
control sequence `\.{\\par}'. It also restores the token and scrap
memories to their initial empty state.
A |force| token is appended to the current scraps before translation
takes place, so that the translation will normally end with \.{\\6} or
\.{\\7} (the \TeX\ macros for |force| and |big_force|). This \.{\\6} or
\.{\\7} is replaced by the concluding \.{\\par} or by \.{\\Y\\par}.
@p procedure finish_Pascal; {finishes a definition or a \PASCAL\ part}
var p:text_pointer; {translation of the scraps}
begin out2("\")("P"); app_tok(force); app_comment; p:=translate;
@.\\P@>
app(p+tok_flag); make_output; {output the list}
if out_ptr>1 then
if out_buf[out_ptr-1]="\" then
@.\\6@>
@.\\7@>
@.\\Y@>
if out_buf[out_ptr]="6" then out_ptr:=out_ptr-2
else if out_buf[out_ptr]="7" then out_buf[out_ptr]:="Y";
out4("\")("p")("a")("r"); finish_line;
stat if text_ptr>max_txt_ptr then max_txt_ptr:=text_ptr;
if tok_ptr>max_tok_ptr then max_tok_ptr:=tok_ptr;
if scrap_ptr>max_scr_ptr then max_scr_ptr:=scrap_ptr;
tats@;@/
tok_ptr:=1; text_ptr:=1; scrap_ptr:=0; {forget the tokens and the scraps}
end;
@ @<Start a macro...@>=
begin sc2("\")("D")(intro); {this will produce `\&{define }'}
@.\\D@>
next_control:=get_next;
if next_control<>identifier then err_print('! Improper macro definition')
@.Improper macro definition@>
else sc1(id_flag+id_lookup(normal))(math);
next_control:=get_next;
end
@ @<Start a format...@>=
begin sc2("\")("F")(intro); {this will produce `\&{format }'}
@.\\F@>
next_control:=get_next;
if next_control=identifier then
begin sc1(id_flag+id_lookup(normal))(math);
next_control:=get_next;
if next_control=equivalence_sign then
begin sc2("\")("S")(math); {output an equivalence sign}
@.\\S@>
next_control:=get_next;
if next_control=identifier then
begin sc1(id_flag+id_lookup(normal))(math);
sc0(semi); {insert an invisible semicolon}
next_control:=get_next;
end;
end;
end;
if scrap_ptr<>5 then err_print('! Improper format definition');
@.Improper format definition@>
end
@ Finally, when the \TeX\ and definition parts have been treated, we have
|next_control>=begin_Pascal|. We will make the global variable |this_module|
point to the current module name, if it has a name.
@<Glob...@>=@!this_module:name_pointer; {the current module name, or zero}
@ @<Translate the \P...@>=
this_module:=0;
if next_control<=module_name then
begin emit_space_if_needed; init_stack;
if next_control=begin_Pascal then next_control:=get_next
else begin this_module:=cur_module;
@<Check that |=| or |==| follows this module name, and
emit the scraps to start the module definition@>;
end;
while next_control<=module_name do
begin outer_parse;
@<Emit the scrap for a module name if present@>;
end;
finish_Pascal;
end
@ @<Check that |=|...@>=
repeat next_control:=get_next;
until next_control<>"+"; {allow optional `\.{+=}'}
if (next_control<>"=")and(next_control<>equivalence_sign) then
err_print('! You need an = sign after the section name')
@.You need an = sign...@>
else next_control:=get_next;
if out_ptr>1 then
if (out_buf[out_ptr]="Y")and(out_buf[out_ptr-1]="\") then
@.\\Y@>
begin app(backup); {the module name will be flush left}
end;
sc1(mod_flag+this_module)(mod_scrap);
cur_xref:=xref[this_module];
if num(cur_xref)<>module_count+def_flag then
begin sc3(math_rel)("+")("}")(math);
{module name is multiply defined}
this_module:=0; {so we won't give cross-reference info here}
end;
sc2("\")("S")(math); {output an equivalence sign}
@.\\S@>
sc1(force)(semi); {this forces a line break unless `\.{@@+}' follows}
@ @<Emit the scrap...@>=
if next_control<module_name then
begin err_print('! You can''t do that in Pascal text');
@.You can't do that...@>
next_control:=get_next;
end
else if next_control=module_name then
begin sc1(mod_flag+cur_module)(mod_scrap); next_control:=get_next;
end
@ Cross references relating to a named module are given after the module ends.
@<Show cross...@>=
if this_module>0 then
begin @<Rearrange the list pointed to by |cur_xref|@>;
footnote(def_flag); footnote(0);
end
@ To rearrange the order of the linked list of cross references, we need
four more variables that point to cross reference entries. We'll end up
with a list pointed to by |cur_xref|.
@<Glob...@>=
@!next_xref,@!this_xref,@!first_xref,@!mid_xref:xref_number;
{pointer variables for rearranging a list}
@ We want to rearrange the cross reference list so that all the entries with
|def_flag| come first, in ascending order; then come all the other
entries, in ascending order. There may be no entries in either one or both
of these categories.
@<Rearrange the list...@>=
first_xref:=xref[this_module];
this_xref:=xlink(first_xref); {bypass current module number}
if num(this_xref)>def_flag then
begin mid_xref:=this_xref; cur_xref:=0; {this value doesn't matter}
repeat next_xref:=xlink(this_xref); xlink(this_xref):=cur_xref;
cur_xref:=this_xref; this_xref:=next_xref;
until num(this_xref)<=def_flag;
xlink(first_xref):=cur_xref;
end
else mid_xref:=0; {first list null}
cur_xref:=0;
while this_xref<>0 do
begin next_xref:=xlink(this_xref); xlink(this_xref):=cur_xref;
cur_xref:=this_xref; this_xref:=next_xref;
end;
if mid_xref>0 then xlink(mid_xref):=cur_xref
else xlink(first_xref):=cur_xref;
cur_xref:=xlink(first_xref)
@ The |footnote| procedure gives cross reference information about
multiply defined module names (if the |flag| parameter is |def_flag|), or about
the uses of a module name (if the |flag| parameter is zero). It assumes that
|cur_xref| points to the first cross-reference entry of interest, and it
leaves |cur_xref| pointing to the first element not printed. Typical outputs:
`\.{\\A101.}'; `\.{\\Us370\\ET1009.}'; `\.{\\As8, 27\\*, 51\\ETs64.}'.
@p procedure footnote(@!flag:sixteen_bits); {outputs module cross-references}
label done,exit;
var q:xref_number; {cross-reference pointer variable}
begin if num(cur_xref)<=flag then return;
finish_line; out("\");
@.\\A@>
@.\\U@>
if flag=0 then out("U")@+else out("A");
@<Output all the module numbers on the reference list |cur_xref|@>;
out(".");
exit:end;
@ The following code distinguishes three cases, according as the number
of cross references is one, two, or more than two. Variable |q| points
to the first cross reference, and the last link is a zero.
@<Output all the module numbers...@>=
q:=cur_xref; if num(xlink(q))>flag then out("s"); {plural}
@.\\As@>
@.\\Us@>
loop@+ begin out_mod(num(cur_xref)-flag);
cur_xref:=xlink(cur_xref); {point to the next cross reference to output}
if num(cur_xref)<=flag then goto done;
if num(xlink(cur_xref))>flag then out2(",")(" ") {not the last}
else begin out3("\")("E")("T"); {the last}
@.\\ET@>
if cur_xref<>xlink(q) then out("s"); {the last of more than two}
@.\\ETs@>
end;
end;
done:
@ @<Output the code for the end of a module@>=
out3("\")("f")("i"); finish_line;
flush_buffer(0,false,false); {insert a blank line, it looks nice}
@.\\fi@>
@* Phase three processing.
We are nearly finished! \.{WEAVE}'s only remaining task is to write out the
index, after sorting the identifiers and index entries.
@<Phase III: Output the cross-reference index@>=
phase_three:=true; print_nl('Writing the index...');
if change_exists then
begin finish_line; @<Tell about changed modules@>;
end;
finish_line; out4("\")("i")("n")("x"); finish_line;
@.\\inx@>
@<Do the first pass of sorting@>;
@<Sort and output the index@>;
out4("\")("f")("i")("n"); finish_line;
@.\\fin@>
@<Output all the module names@>;
out4("\")("c")("o")("n"); finish_line;
@.\\con@>
print('Done.');
@ Just before the index comes a list of all the changed modules, including
the index module itself.
@<Glob...@>=
@!k_module:0..max_modules; {runs through the modules}
@ @<Tell about changed modules@>=
begin {remember that the index is already marked as changed}
k_module:=1;
out4("\")("c")("h")(" ");
while k_module<module_count do
begin if changed_module[k_module] then
begin out_mod(k_module); out2(",")(" ");
end;
incr(k_module);
end;
out_mod(k_module);
out(".");
end
@ A left-to-right radix sorting method is used, since this makes it easy to
adjust the collating sequence and since the running time will be at worst
proportional to the total length of all entries in the index. We put the
identifiers into 230 different lists based on their first characters.
(Uppercase letters are put into the same list as the corresponding lowercase
letters, since we want to have `$t<\\{TeX}<\&{to}$'.) The
list for character |c| begins at location |bucket[c]| and continues through
the |blink| array.
@<Glob...@>=
@!bucket:array[ASCII_code] of name_pointer;
@!next_name: name_pointer; {successor of |cur_name| when sorting}
@!c:ASCII_code; {index into |bucket|}
@!h:0..hash_size; {index into |hash|}
@!blink:array[0..max_names] of sixteen_bits; {links in the buckets}
@ To begin the sorting, we go through all the hash lists and put each entry
having a nonempty cross-reference list into the proper bucket.
@<Do the first pass...@>=
for c:=0 to 255 do bucket[c]:=0;
for h:=0 to hash_size-1 do
begin next_name:=hash[h];
while next_name<>0 do
begin cur_name:=next_name; next_name:=link[cur_name];
if xref[cur_name]<>0 then
begin c:=byte_mem[cur_name mod ww,byte_start[cur_name]];
if (c<="Z")and(c>="A") then c:=c+@'40;
blink[cur_name]:=bucket[c]; bucket[c]:=cur_name;
end;
end;
end
@ During the sorting phase we shall use the |cat| and |trans| arrays from
\.{WEAVE}'s parsing algorithm and rename them |depth| and |head|. They now
represent a stack of identifier lists for all the index entries that have
not yet been output. The variable |sort_ptr| tells how many such lists are
present; the lists are output in reverse order (first |sort_ptr|, then
|sort_ptr-1|, etc.). The |j|th list starts at |head[j]|, and if the first
|k| characters of all entries on this list are known to be equal we have
|depth[j]=k|.
@d depth==cat {reclaims memory that is no longer needed for parsing}
@d head==trans {ditto}
@d sort_ptr==scrap_ptr {ditto}
@d max_sorts==max_scraps {ditto}
@<Globals...@>=
@!cur_depth:eight_bits; {depth of current buckets}
@!cur_byte:0..max_bytes; {index into |byte_mem|}
@!cur_bank:0..ww-1; {row of |byte_mem|}
@!cur_val:sixteen_bits; {current cross reference number}
stat@!max_sort_ptr:0..max_sorts;@+tats {largest value of |sort_ptr|}
@ @<Set init...@>=stat max_sort_ptr:=0;@+tats
@ The desired alphabetic order is specified by the |collate| array; namely,
|collate[0]<collate[1]<@t$\cdots$@><collate[229]|.
@<Glob...@>=@!collate:array[0..229] of ASCII_code; {collation order}
@ @<Local variables for init...@>=
@!c:ASCII_code; {used to initialize |collate|}
@ We use the order $\hbox{null}<\.\ <\hbox{other characters}<\.\_<
\.A=\.a<\cdots<\.Z=\.z<\.0<\cdots<\.9.$
@<Set init...@>=
collate[0]:=0; collate[1]:=" ";
for c:=1 to " "-1 do collate[c+1]:=c;
for c:=" "+1 to "0"-1 do collate[c]:=c;
for c:="9"+1 to "A"-1 do collate[c-10]:=c;
for c:="Z"+1 to "_"-1 do collate[c-36]:=c;
collate["_"-36]:="_"+1;
for c:="z"+1 to 255 do collate[c-63]:=c;
collate[193]:="_";
for c:="a" to "z" do collate[c-"a"+194]:=c;
for c:="0" to "9" do collate[c-"0"+220]:=c;
@ Procedure |unbucket| goes through the buckets and adds nonempty lists
to the stack, using the collating sequence specified in the |collate| array.
The parameter to |unbucket| tells the current depth in the buckets.
Any two sequences that agree in their first 255 character positions are
regarded as identical.
@d infinity=255 {$\infty$ (approximately)}
@p procedure unbucket(@!d:eight_bits); {empties buckets having depth |d|}
var c:ASCII_code; {index into |bucket|}
begin for c:=229 downto 0 do if bucket[collate[c]]>0 then
begin if sort_ptr>max_sorts then overflow('sorting');
incr(sort_ptr);
stat if sort_ptr>max_sort_ptr then max_sort_ptr:=sort_ptr;@;@+tats@;@/
if c=0 then depth[sort_ptr]:=infinity else depth[sort_ptr]:=d;
head[sort_ptr]:=bucket[collate[c]]; bucket[collate[c]]:=0;
end;
end;
@ @<Sort and output...@>=
sort_ptr:=0; unbucket(1);
while sort_ptr>0 do
begin cur_depth:=cat[sort_ptr];
if (blink[head[sort_ptr]]=0)or(cur_depth=infinity) then
@<Output index entries for the list at |sort_ptr|@>
else @<Split the list at |sort_ptr| into further lists@>;
end
@ @<Split the list...@>=
begin next_name:=head[sort_ptr];
repeat cur_name:=next_name; next_name:=blink[cur_name];
cur_byte:=byte_start[cur_name]+cur_depth; cur_bank:=cur_name mod ww;
if cur_byte=byte_start[cur_name+ww] then c:=0 {we hit the end of the name}
else begin c:=byte_mem[cur_bank,cur_byte];
if (c<="Z")and(c>="A") then c:=c+@'40;
end;
blink[cur_name]:=bucket[c]; bucket[c]:=cur_name;
until next_name=0;
decr(sort_ptr); unbucket(cur_depth+1);
end
@ @<Output index...@>=
begin cur_name:=head[sort_ptr];
@!debug if trouble_shooting then debug_help;@;@+gubed@/
repeat out2("\")(":");
@.\\:@>
@<Output the name at |cur_name|@>;
@<Output the cross-references at |cur_name|@>;
cur_name:=blink[cur_name];
until cur_name=0;
decr(sort_ptr);
end
@ @<Output the name...@>=
case ilk[cur_name] of
normal: if length(cur_name)=1 then out2("\")("|")@+else out2("\")("\");
@.\\|@>
@.\\\\@>
roman: do_nothing;
wildcard: out2("\")("9");
@.\\9@>
typewriter: out2("\")(".");
@.\\.@>
othercases out2("\")("&")
@.\\\&@>
endcases;@/
out_name(cur_name)
@ Section numbers that are to be underlined are enclosed in
`\.{\\[}$\,\ldots\,$\.]'.
@<Output the cross-references...@>=
@<Invert the cross-reference list at |cur_name|, making |cur_xref| the head@>;
repeat out2(",")(" "); cur_val:=num(cur_xref);
if cur_val<def_flag then out_mod(cur_val)
else begin out2("\")("["); out_mod(cur_val-def_flag); out("]");
@.\\[@>
end;
cur_xref:=xlink(cur_xref);
until cur_xref=0;
out("."); finish_line
@ List inversion is best thought of as popping elements off one stack and
pushing them onto another. In this case |cur_xref| will be the head of
the stack that we push things onto.
@<Invert the cross-reference list at |cur_name|, making |cur_xref| the head@>=
this_xref:=xref[cur_name]; cur_xref:=0;
repeat next_xref:=xlink(this_xref); xlink(this_xref):=cur_xref;
cur_xref:=this_xref; this_xref:=next_xref;
until this_xref=0
@ The following recursive procedure walks through the tree of module names and
prints them.
@^recursion@>
@p procedure mod_print(p:name_pointer); {print all module names in subtree |p|}
begin if p>0 then
begin mod_print(llink[p]);@/
out2("\")(":");@/
@.\\:@>
tok_ptr:=1; text_ptr:=1; scrap_ptr:=0; init_stack;
app(p+mod_flag); make_output;
footnote(0); {|cur_xref| was set by |make_output|}
finish_line;@/
mod_print(rlink[p]);
end;
end;
@ @<Output all the module names@>=@+mod_print(root)
@* Debugging.
The \PASCAL\ debugger with which \.{WEAVE} was developed allows breakpoints
to be set, and variables can be read and changed, but procedures cannot be
executed. Therefore a `|debug_help|' procedure has been inserted in the main
loops of each phase of the program; when |ddt| and |dd| are set to appropriate
values, symbolic printouts of various tables will appear.
The idea is to set a breakpoint inside the |debug_help| routine, at the
place of `\ignorespaces|breakpoint:|\unskip' below. Then when
|debug_help| is to be activated, set |trouble_shooting| equal to |true|.
The |debug_help| routine will prompt you for values of |ddt| and |dd|,
discontinuing this when |ddt<=0|; thus you type $2n+1$ integers, ending
with zero or a negative number. Then control either passes to the
breakpoint, allowing you to look at and/or change variables (if you typed
zero), or to exit the routine (if you typed a negative value).
Another global variable, |debug_cycle|, can be used to skip silently
past calls on |debug_help|. If you set |debug_cycle>1|, the program stops
only every |debug_cycle| times |debug_help| is called; however,
any error stop will set |debug_cycle| to zero.
@<Globals...@>=
@!debug@!trouble_shooting:boolean; {is |debug_help| wanted?}
@!ddt:integer; {operation code for the |debug_help| routine}
@!dd:integer; {operand in procedures performed by |debug_help|}
@!debug_cycle:integer; {threshold for |debug_help| stopping}
@!debug_skipped:integer; {we have skipped this many |debug_help| calls}
@!term_in:text_file; {the user's terminal as an input file}
gubed
@ The debugging routine needs to read from the user's terminal.
@^system dependencies@>
@<Set init...@>=
@!debug trouble_shooting:=true; debug_cycle:=1; debug_skipped:=0; tracing:=0;@/
trouble_shooting:=false; debug_cycle:=99999; {use these when it almost works}
reset(term_in,'TTY:','/I'); {open |term_in| as the terminal, don't do a |get|}
gubed
@ @d breakpoint=888 {place where a breakpoint is desirable}
@^system dependencies@>
@p @!debug procedure debug_help; {routine to display various things}
label breakpoint,exit;
var k:integer; {index into various arrays}
begin incr(debug_skipped);
if debug_skipped<debug_cycle then return;
debug_skipped:=0;
loop@+ begin write(term_out,'#'); update_terminal; {prompt}
read(term_in,ddt); {read a list of integers}
if ddt<0 then return
else if ddt=0 then
begin goto breakpoint;@\ {go to every label at least once}
breakpoint: ddt:=0;@\
end
else begin read(term_in,dd);
case ddt of
1: print_id(dd);
2: print_text(dd);
3: for k:=1 to dd do print(xchr[buffer[k]]);
4: for k:=1 to dd do print(xchr[mod_text[k]]);
5: for k:=1 to out_ptr do print(xchr[out_buf[k]]);
6: for k:=1 to dd do
begin print_cat(cat[k]); print(' ');
end;
othercases print('?')
endcases;
end;
end;
exit:end;
gubed
@* The main program.
Let's put it all together now: \.{WEAVE} starts and ends here.
@^system dependencies@>
The main procedure has been split into three sub-procedures in order to
keep certain \PASCAL\ compilers from overflowing their capacity.
@^split procedures@>
@p procedure Phase_I;
begin @<Phase I:...@>;
end;
@#
procedure Phase_II;
begin @<Phase II:...@>;
end;
@#
begin initialize; {beginning of the main program}
print_ln(banner); {print a ``banner line''}
@<Store all the reserved words@>;
Phase_I; Phase_II;@/
@<Phase III:...@>;
@<Check that all changes have been read@>;
end_of_WEAVE:
stat @<Print statistics about memory usage@>;@+tats@;@/
@t\4\4@>{here files should be closed if the operating system requires it}
@<Print the job |history|@>;
end.
@ @<Print statistics about memory usage@>=
print_nl('Memory usage statistics: ',
name_ptr:1,' names, ', xref_ptr:1,' cross references, ',
byte_ptr[0]:1);
for cur_bank:=1 to ww-1 do print('+',byte_ptr[cur_bank]:1);
print(' bytes;');
print_nl('parsing required ',max_scr_ptr:1,' scraps, ',max_txt_ptr:1,
' texts, ',max_tok_ptr:1,' tokens, ', max_stack_ptr:1,' levels;');
print_nl('sorting required ',max_sort_ptr:1, ' levels.')
@ Some implementations may wish to pass the |history| value to the
operating system so that it can be used to govern whether or not other
programs are started. Here we simply report the history to the user.
@^system dependencies@>
@<Print the job |history|@>=
case history of
spotless: print_nl('(No errors were found.)');
harmless_message: print_nl('(Did you see the warning message above?)');
error_message: print_nl('(Pardon me, but I think I spotted something wrong.)');
fatal_message: print_nl('(That was a fatal error, my friend.)');
end {there are no other cases}
@* System-dependent changes.
This module should be replaced, if necessary, by changes to the program
that are necessary to make \.{WEAVE} work at a particular installation.
It is usually best to design your change file so that all changes to
previous modules preserve the module numbering; then everybody's version
will be consistent with the printed program. More extensive changes,
which introduce new modules, can be inserted here; then only the index
itself will get a new module number.
@^system dependencies@>
@* Index.
If you have read and understood the code for Phase III above, you know what
is in this index and how it got here. All modules in which an identifier is
used are listed with that identifier, except that reserved words are
indexed only when they appear in format definitions, and the appearances
of identifiers in module names are not indexed. Underlined entries
correspond to where the identifier was declared. Error messages, control
sequences put into the output, and a few
other things like ``recursion'' are indexed here too.
|
