% This is DOCMAC.DOC, Release 1 of 12 Aug 1991. %--------------------------------------------------------- % This macro package is based on PROGDOC.DOC, which is % (c) 1988 by J.Schrod. Put into the public domain. % It is enhanced by incorporating some macros from WEBMAC.TEX to deal with % header lines and a table of contents and offers some additional typesetting % capabilities. The enhancements and changes are done by % Hans-Hermann Bode (HHBODE@DOSUNI1.BITNET). % % Macro package for the documentation of programs (and TeX macros) % english version % % Version of the basic PROGDOC.DOC: % first version (for ftp/Bitnet) (88-12-05) % % Release history of DOCMAC.DOC: % % Rel. Date Author Comment % 1 12 Aug 1991 H2B First change file to PROGDOC.DOC. % % documented with itself... \input docmac %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % local macros % \font\mc=cmr9 % for names like ASCII \def\title{DOCMAC} \def\topofcontents{\null\vfill \centerline{\titlefont\title} \bigskip \centerline{Joachim Schrod} \medskip \centerline{Hans-Hermann Bode} \vfill} \def\botofcontents{\vfill The macro package and the documentation was written initially by J.~Schrod under the name \.{progdoc}. H.-H.~Bode enhanced the package with some stuff from \.{webmac} and completed the documentation with material from the ``\.{WEB} User Manual''; this makes up the chapter on ``Formatting''---besides some minor changes in the remainder. \par} \def\WEB{{\tt WEB\/}} \def\LaTeX{{\rm L\kern-.36em\raise.3ex\hbox{\sc a}\kern-.15em\TeX}} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chap Introduction. \WEB{} systems allow the documentation of programs by supporting the separation in program fragments which can be collected and rearranged afterwards. This allows {\it top-down\/} programming as well as the {\it bottom-up\/} design of programs. Each program fragment can be documented, usually with \TeX{}. A disadvantadge is that \WEB{} actually exists only for a few programming languages (Pascal, C, Modula-2). Besides, building up \WEB{} systems for ``exotic'' programming languages like \TeX{} is very difficult. This macro package was built to allow good documentation for programs in languages for which \WEB{} doesn't exist. It separates a program text in sections that can be documented. All sections, collected sequentially, will result in the complete program. In every section begin and end of the program part are marked with |\beginprog| and |\endprog|, this program part will be formatted as it is input (``verbatim''). The separation of the sections is done in this version (which is based on Plain~\TeX{}) by the macros |\chap| and |\sect|. But this can be changed easily, e.g.\ for preparing the documentation with \LaTeX{}. Then the last section group of this macro package must be omitted and the rest must be realized as a style option. In the documentation part of a section text pieces, e.g.\ names of variables, can be inserted verbatim to demonstrate the connection to the program text. These parts are included in vertical bars (`{\tt\vbar}') while the vertical bar itself can be obtained by |\origvert|. This macro package does not offer the creation of an index because it is not known which syntactical tokens the language has that should be mentioned in an index. If you want a table of contents, just say |\contentspage| at the end of your document (you need not put a |\bye| after that, |\contentspage| does this for you). Some remarks about future enhancements which should be done urgently are set in slanted type. \sect We have to realize three parts: (1)~the formatting of rather small verbatim texts in a line, (2)~the formatting of larger parts of program and (3)~the document structuring elements for the separation of the sections. Before we start we declare some shorthands for category codes. By declaring the at sign~(`|@|') as well as the underscore~`(|_|)' as letters we can use them in our macros. {\tolerance=10000 (I agree with D.~Knuth that |\identifier_several_words_long| is much better readable than |\IdentifierSeveralWordsLong| and both are better than |\pr@@@s|.) With the at sign we can use the ``private'' Plain macros and with the underscore we can make our own macros more readable. But as we have to restore these category codes at the end of this macro file we store their former values in the control sequences |\atcode| and |\uscode|. This method is better than the usage of a group because not all macros have to be defined global this way. \par} \beginprog \chardef\escape=0 \chardef\letter=11 \chardef\other=12 %\chardef\active=13 % is defined in Plain already \chardef\atcode=\catcode`\@ \chardef\uscode=\catcode`\_ \catcode`\@=\letter \catcode`\_=\letter \endprog \chap Local Verbatim Formatting. The main point of every verbatim formatting is the switching of the character codes of all characters that have a special \TeX{} meaning. This can be done with the control sequence |\dospecials| that applies the control sequence |\do| to all special characters. Additionally, every line is regarded as a paragraph without indentation. Between two paragraphs, i.e.\ between two lines, no extra space is set. Finally all blanks and tabular characters shall be obeyed and the inter word space after sentence terminators shall not be enlarged. The activation of the tabular characters with |\obeytabs| is equivalent to |\obeyspaces| in {\tt plain.tex\/}. Verbatim texts are set in monospace, we use (like in \WEB{}) the character set with the extended {\mc ASCII}. For \LaTeX{} we would have to add here |\ttex| to all size modifications like |\xpt|, |\xiipt|, etc. \beginprog \font\tentex=cmtex10 % typewriter extended ASCII 10pt \let\ttex=\tentex % only for PLAIN with base size 10pt \def\setup_verbatim{% \def\do##1{\catcode`##1\other}\dospecials \parskip\z@skip \parindent\z@ \obeylines \obeyspaces \obeytabs \frenchspacing \ttex } \let\tab=\space \begingroup \catcode`\^^I=\active% % Attention: no tabs! \gdef\obeytabs{\catcode`\^^I=\active\def^^I{\tab}} \global\let^^I=\tab% % if an active tab appears in a \write \endgroup \endprog \sect After having saved the old meaning of `{\tt\vbar}' in |\origvert| and after declaring |\vbar| as a synonym for the character that has the code of a vertical bar in the actual font, the vertical bar can be made active. Then we call |\setup_verbatim|. But the newline characters shall not be processed, they shall be regarded like blank space. This can be reached by defining |\par| as |\space|. The next vertical bar in the input closes the group which becomes an (unbreakable) |\hbox| then. The old meanings of the special characters and of the vertical bar are restored and \TeX{} is in normal (horizontal) mode again. \beginprog \let\origvert=| \chardef\vbar=`\| \catcode`\|=\active \def|{% \leavevmode \hbox\bgroup \let\par\space \setup_verbatim \let|\egroup } \endprog \chap Program Fragments in Verbatim. We need macros to format the program fragments without any linebreaking. Such a text area shall start with the macro |\beginprog| and end with |\endprog|. The macro |\endprog| must stand at the very beginning of a line and must be followed by white space (blank, tab or newline character). After |\beginprog| as well as after |\endprog| the rest of the line is ignored. Two demands must be regarded: There should be no length restrictions for the processed text, and the tabular characters should be expanded so that this macro works on PC's and on VAXes, too. \sect The implementation method is quite simple: We read the next line, test, wether the end is reached (by comparing with |\endprog|) and otherwise set the character actually read. Every character is inspected and tabular characters are expanded. Wether a line is set or wether the end of the processed area is reached is indicated by the switch |\if@print|. At the beginning of the |\beginprog| macro most settings are done with |\setup_verbatim| (the vertical bar must be handled separately) and the rest of the line is ignored. As everything is done within a group, the end of the verbatim text can be processed by simply closing this group. For the user it looks as if |\endprog| terminates the processing, but it just serves for the identification of the end, the true processing is done with the internal macro |\end_verbatim|. \beginprog \newif\if@print \def\beginprog{% \endgraf \bigbreak \begingroup \setup_verbatim \catcode`\|\other \@printtrue \ignore_rest_line } \let\end_verbatim=\endgroup % internal command ! \endprog \sect The first line is ignored, all the other lines are identified with |\set_next_line| and processed with |\do_set|. This separation in identification and processing allows that the line end character is active in the definition only for a short time. When a line is to be formatted, we first check with |\check_print| wether it contains |\endprog|, otherwise it is printed with |\print_char|. The printing must be done for every character individually because we want to check for tabular characters; the exact algorithm is described below. Here we just have to note that |\print_char| is used with two parameters of which the second one is finished with the token |\end_line|. The first parameter is the first character of the line, the second parameter is the rest of the line. If the line is empty, the argument of |\do_set| is empty, too; so the activation of |\print_char| must be finished with two |\end_line|. Then the first |\end_line| is the first argument for |\print_char| and the second argument is empty. But if the line did contain something, the second |\end_line| is evaluated, for this case it is defined as |\relax|. At last we call |\set_next_line| again to format the next line. If the end is reached, i.e.\ if a line with |\endprog| was found, |\set_next_line| is redefined as |\relax|. This can be done because the original meaning is restored while closing the group with |\end_verbatim|. \beginprog \begingroup \obeylines% % ^^M is active! ==> every line must end with % \gdef\ignore_rest_line#1^^M{\set_next_line}% \gdef\set_next_line#1^^M{\do_set{#1}}% \endgroup \def\do_set#1{% \endgraf \check_print{#1}% \if@print \indent \print_char#1\end_line\end_line \else \let\set_next_line\end_verbatim \fi \set_next_line } \let\end_line=\relax \endprog \sect Before we look at the problem of formatting a line, we declare |\check_print| that checks the end of the verbatim mode. We have to do two things: we must split everything in front of the first blank or tabular character and compare for identity with |\endprog|. The splitting is easy because the line which is our first argument contains blanks and tabulators as active characters. First we call |\cut_at_tab| that demands a tabular character as separator for its two pramenters so that everything in the line in front of the first tabulator is part of the first parameter. If there is no tabular character in the line, we append one so that the second parameter is empty. The same trick is used to separate the part in front of the first blank character from the resulting first part. The check is done with |\do_check|. We use a separate macro here so that we can indent it (in the following definition blanks are active!) \beginprog \begingroup \obeyspaces\obeytabs \gdef\check_print#1{\cut_at_tab#1^^I\end_line} \gdef\cut_at_tab#1^^I#2\end_line{\check_first_part#1 \end_line}% blank ! \gdef\check_first_part#1 #2\end_line{\do_check{#1}} \endgroup \endprog \sect |\do_check| compares the line with a sample line that is available in |\@endverbatim|. During the definition of |\@endverbatim| it must be cared for that the escape character `|\|' is a printable character: A comparison with |\ifx| demands identical category codes. As a temporary escape character we use the vertical bar. \beginprog \def\do_check#1{% \def\@line{#1}% \ifx \@line\@endverbatim \@printfalse \fi } {\catcode`\|=\escape \catcode`\\=\other % | is temporary escape char |gdef|@endverbatim{\endprog} % sample line } % here \endgroup can't be used \endprog \sect Now we can set a line: we start with the first character, followed by the rest of the line. Each character is counted in |\char_count|. At the beginning of a line |\char_count| is~1, this is reset at the end of the line. \goodbreak \beginprog \newcount\char_count \char_count\@ne \def\print_char#1#2\end_line{% \print_first_char{#1}% \print_rest_of_line{#2}% } \endprog \sect For each character that is set |\char_count| is incremented. If a character is a tabulator, we set with |\print_tab| the fitting amount of blank characters, otherwise the character itself. We must compare the character that is stored in |\@char| with a macro of which the ``first-level'' expansion is an active tabulator. For this case we declare |\@tab|. \beginprog {\obeytabs\gdef\@tab{^^I}} \def\print_first_char#1{% \def\@char{#1}% \advance \char_count\@ne \ifx \@char\@tab \print_tab \else \@char \fi } \endprog \sect If we want to fill the line with blank spaces up to the next column with a number that can be divided by~8, we must be able to compute the column number modulo~8, but \TeX{} has no modulo operator. So we define the macro |\mod_viii| that computes its argument modulo~8 and returns the result in the counter |\count_mod_viii|. For the computation we need the temporary counter |\count@|. \beginprog \newcount\count_mod_viii \def\mod_viii#1{% \count@ #1\relax \count_mod_viii\count@ \divide \count@ 8\relax \multiply \count@ 8\relax \advance \count_mod_viii -\count@ } \endprog \sect Now we can declare |\print_tab|. We must remember that |\char_count| was incremented already, if we set only one blank character the counter keeps untouched. \beginprog \def\print_tab{% \loop \space \mod_viii\char_count \ifnum \count_mod_viii>\z@ \advance \char_count\@ne \repeat } \endprog \sect If the rest of the line is empty, we are ready. |\char_count| is reset to~1 for the next line. Inside the |\else| part of |\ifx| |\print_char| should not be used directly because this costs too much storage of \TeX{}. Instead we set a control sequence |\next| that is processed afterwards, depending on the result of the comparison. If there is still something to set, we use |\print_char| again, otherwise a syntactically similar macro that expands to |\relax|. \beginprog \def\print_rest_of_line#1{% \def\@line{#1}% \ifx \@line\empty \char_count\@ne \def\next##1\end_line{\relax}% \else \let\next\print_char \fi \next#1\end_line } \endprog \chap Document Structuring. The layout of the document shall be like in \WEB{}. This can be done easily in Plain. If \LaTeX{} shall be used for the document preparation, this part of the macro must be eliminated. It will also be better to include the program parts \LaTeX-like as an environment between |\begin{prog}| and |\end{prog}| and to define control sequences |\makebaractive| and |\makebarother| for the switch of the special meaning of the vertical bar (`{\tt \origvert\/}') (for the {\tt tabular\/} environment!) All sections are enumerated, the number of the next section is stored in the counter |\sectno|. We differenciate between main sections which start a group of sections and between normal sections within a group. \beginprog \newcount\sectno \sectno=\@ne \endprog \sect The main sections are started with the macro |\chap| that has one parameter, the title of the section group. This parameter must be terminated by a dot. We start a new page, typeset the title in bold face and separate it from the section text with a |\medskip|. This text, the documentation part of the section, is formatted without paragraph indentation. The deletion of the paragraph indentation is done with |\everypar|. We do not care if |\everypar| is used for something else and delete it after usage. The section number is marked to appear {\it via\/} |\topmark| in the headline, the title goes as a running head into the headline too. The section number is displayed on the terminal as a progress report, and title, section and page number are recorded in the contents file. \begingroup \sl The dealing with |\everypar| should not be so cruel. \endgroup \beginprog \def\chap#1.{% \mark{\number\sectno} \endgraf \vfill\eject \def\rhead{\uppercase{\ignorespaces#1}} % define running headline \message{*\number\sectno} % progress report \let\Z=\let % now you can \send the control sequence \Z \edef\next{\write\cont{\Z{#1}{\number\sectno}{\the\pageno}}}\next % to contents file \noindent {\bf \number\sectno.\quad #1.}% \advance \sectno\@ne \endgraf \medskip \nobreak \everypar{% \setbox0\lastbox \global\everypar{}% }% } \endprog \sect Normal sections are started with |\sect|. This macro has no parameter. Between two paragraphs we set 2~pica space (if possible) and prefer the page breaking. Between the section number and the text one quad of space is set. \begingroup \sl The page break manipulation could be more sophisticated. \endgroup \beginprog \def\sect{% \mark{\number\sectno} \endgraf \vskip 2pc plus 1pc minus 6dd \goodbreak \noindent {\bf \number\sectno.}\advance \sectno\@ne \quad \ignorespaces } \endprog \chap Formatting. The macros in this chapter are extracted from \.{WEBMAC.TEX}, while most of the documentation is taken from Appendix~G of the ``\.{WEB} User Manual''. The macros in \.{docmac} make it possible to produce a variety of formats, and the purpose of this appendix is to explain some of the possibilities. Displayed examples, that are included in the documentation, come out centered to distinguish them from program text, which is flushleft. \def\lpile{\def\cr{\hfill\endline}\matrix} % I [D E Knuth] only use \lpile by itself \sect We begin the formatting part with some declarations that should make look the output a little better for documenting programs. \beginprog \parskip 0pt % no stretch between paragraphs \parindent 1em % for paragraphs and for the first line of Pascal text \endprog \sect Three fonts have been declared in addition to the standard fonts of \.{PLAIN} format: You can say `\.{\{\\sc stuff\}}' to get {\sc stuff} in small caps; and you can select the largish fonts \.{\\titlefont} and \.{\\ttitlefont} in the title of your document, where \.{\\ttitlefont} is a typewriter style of type. \beginprog \font\eightrm=cmr8 \font\sc=cmcsc10 % this IS a caps-and-small-caps font! \let\mainfont=\tenrm \font\titlefont=cmr7 scaled\magstep4 % title on the contents page \font\ttitlefont=cmtt10 scaled\magstep2 % typewriter type in title %\font\tentex=cmtex10 % TeX extended character set (used in strings) % already defined above \fontdimen7\tentex=0pt % no extra space after punctuation \endprog \sect You can say `\.{\\\\\{identifier\}}' to set \\{identifier} in italics. To get typewriter-like type, as when referring to `\.{WEB}', you can use the `\.{\\.}' macro (e.g., `\.{\\.\{WEB\}}'). In the argument to this macro you should insert an additional backslash before the symbols listed below, i.e., before backslashes and dollar signs and the like. A `\.{\\\ }' here will result in the visible space symbol; to get an invisible space following a control sequence you can say `\.{\{\ \}}'. \beginprog \def\\#1{\hbox{\it#1\/\kern.05em}} % italic type for identifiers \def\|#1{\hbox{$#1$}} % one-letter identifiers look a bit better this way \def\{\hbox{\bf#1\/}} % boldface type for reserved words \def\.#1{\hbox{\tentex % typewriter type for strings \let\\=\BS % backslash in a string \let\'=\RQ % right quote in a string \let\`=\LQ % left quote in a string \let\{=\LB % left brace in a string \let\}=\RB % right brace in a string \let\~=\TL % tilde in a string \let\ =\SP % space in a string \let\_=\UL % underline in a string \let\&=\AM % ampersand in a string #1}} \def\#{\hbox{\tt\char`\#}} % parameter sign \def\${\hbox{\tt\char`\$}} % dollar sign \def\%{\hbox{\tt\char`\%}} % percent sign \def\^{\ifmmode\mathchar"222 \else\char`^ \fi} % pointer or hat % circumflex accents can be obtained from \^^D instead of \^ \def\AT!{@} % at sign for control text \chardef\AM=`\& % ampersand character in a string \chardef\BS=`\\ % backslash in a string \chardef\LB=`\{ % left brace in a string \def\LQ{{\tt\char'22}} % left quote in a string \chardef\RB=`\} % right brace in a string \def\RQ{{\tt\char'23}} % right quote in a string \def\SP{{\tt\char`\ }} % (visible) space in a string \chardef\TL=`\~ % tilde in a string \chardef\UL=`\_ % underline character in a string \endprog \sect The three control sequences \.{\\pagewidth}, \.{\\pageheight}, and \.{\\fullpageheight} can be redefined in the limbo section at the beginning of your source file, to change the dimensions of each page. The standard settings $$\lpile{\.{\\pagewidth=6.5in}\cr \.{\\pageheight=8.7in}\cr \.{\\fullpageheight=9in}\cr}$$ were used to prepare the present report; \.{\\fullpageheight} is \.{\\pageheight} plus room for the additional heading and page numbers at the top of each page. If you change any of these quantities, you should call the macro \.{\\setpage} immediately after making the change. The \.{\\pageshift} macro defines an amount by which right-hand pages (i.e., odd-numbered pages) are shifted right with respect to left-hand (even-numbered) ones. By adjusting this amount you may be able to get two-sided output in which the page numbers line up on opposite sides of each sheet. \beginprog \newdimen\pagewidth \pagewidth=6.5in % the width of each page \newdimen\pageheight \pageheight=8.7in % the height of each page \newdimen\fullpageheight \fullpageheight=9in % page height including headlines \newdimen\pageshift \pageshift=0in % shift righthand pages wrt lefthand ones \def\magnify#1{\mag=#1\pagewidth=6.5truein\pageheight=8.7truein \fullpageheight=9truein\setpage} \def\setpage{\hsize\pagewidth\vsize\pageheight} % use after changing page size \endprog \sect The \.{\\title} macro will appear at the top of each page in small caps. For example, this document was produced after saying `\.{\\def\\title\{DOCMAC\}}'. \beginprog \def\lheader{\mainfont\the\pageno\eightrm\qquad\rhead\hfill\title\qquad \tensy x\mainfont\topmark} % top line on left-hand pages \def\rheader{\tensy x\mainfont\topmark\eightrm\qquad\title\hfill\rhead \qquad\mainfont\the\pageno} % top line on right-hand pages \def\page{\box255 } \def\normaloutput#1#2#3{\ifodd\pageno\hoffset=\pageshift\fi \shipout\vbox{ \vbox to\fullpageheight{ \iftitle\global\titlefalse \else\hbox to\pagewidth{\vbox to10pt{}\ifodd\pageno #3\else#2\fi}\fi \vfill#1}} % parameter #1 is the page itself \global\advance\pageno by1} \def\rhead{\.{DOCMAC} OUTPUT} % this running head is reset by starred sections \def\title{} % an optional title can be set by the user \endprog \sect The first page usually is assigned page number 1. To start on page 16, with contents on page 15, say this: `\.{\\def\\contentspagenumber\{15\}} \.{\\pageno=\\contentspagenumber} \.{\\advance\\pageno by 1}'. \beginprog \def\contentspagenumber{0} % default page number for table of contents \endprog \sect The macro \.{\\iftitle} will suppress the header line if it is defined by `\.{\\titletrue}'. The normal value is \.{\\titlefalse} except for the table of contents; thus, the contents page is usually unnumbered. Two macros are provided to give flexibility to the table of contents: \.{\\topofcontents} is invoked just before the contents info is read, and \.{\\botofcontents} is invoked just after. For example, you can define: $$\lpile{\.{\\def\\topofcontents\{\\null\\vfill}\cr \.{ { }\\titlefalse \% include headline on the contents page}\cr \.{ { }\\def\\rheader\{\\mainfont Appendix D\\hfil 15\}}\cr \.{ { }\\centerline\{\\titlefont The \{\\ttitlefont WEAVE\}{ }processor\}}\cr \.{ { }\\vskip 15pt \\centerline\{(Version 4)\}{ }\\vfill\}}\cr}$$ Redefining \.{\\rheader}, which is the headline for right-hand pages, suffices in this case to put the desired information at the top of the contents page. \beginprog \newif\iftitle \def\topofcontents{\centerline{\titlefont\title} \vfill} % this material will start the table of contents page \def\botofcontents{\vfill} % this material will end the table of contents page \endprog \sect Data for the table of contents is written to a file that is read when |\contentspage| is called; there's one line of data for every starred module. For example, when this document was being created, a file \.{DOCMAC.TOC} containing $$\lpile{\.{\\Z \{{ }Introduction\}\{1\}\{1\}}\cr \.{\\Z \{{ }Local Verbatim Formatting\}\{3\}\{2\}}\cr}$$ and similar lines was created. The \.{\\topofcontents} macro could redefine \.{\\Z} so that the information appears in another format. \beginprog \def\contentsfile{\jobname.toc} % file that gets table of contents info \def\readcontents{\expandafter\input \contentsfile} \newwrite\cont \output{\setbox0=\page % the first page is garbage \openout\cont=\contentsfile \global\output{\normaloutput\page\lheader\rheader}} \setpage \vbox to \vsize{} % the first \topmark won't be null \def\con{% \write\cont{} % ensure that the contents file isn't empty \closeout\cont % the contents information has been fully gathered \par\vfill\eject % finish the preceding page \rightskip 0pt \hyphenpenalty 50 \tolerance 200 \setpage \output{\normaloutput\page\lheader\rheader} \titletrue % prepare to output the table of contents \pageno=\contentspagenumber \def\rhead{TABLE OF CONTENTS} \message{Table of contents:} \topofcontents \line{\hfil Section\hbox to3em{\hss Page}} \def\Z##1##2##3{\line{\ignorespaces##1 \leaders\hbox to .5em{.\hfil}\hfil\ ##2\hbox to3em{\hss##3}}} \readcontents\relax % read the contents info \botofcontents \end} % print the contents page(s) and terminate \let\contentspage=\con \endprog \sect Sometimes it is necessary or desirable to divide the output of \.{docmac} into subfiles that can be processed separately. For example, the listing of \TeX\ runs to more than 500 pages, and that is enough to exceed the capacity of many printing devices and/or their software. When an extremely large job isn't cut into smaller pieces, the entire process might be spoiled by a single error of some sort, making it necessary to start everything over. Here's a safe way to break a docmac'ed file into three parts: Say the pieces are $\alpha$, $\beta$, and $\gamma$, where each piece begins with a starred module. All macros should be defined in the opening limbo section of $\alpha$, and copies of this \TeX\ code should be placed at the beginning of $\beta$ and of $\gamma$. In order to process the parts separately, we need to take care of two things: The starting page numbers of $\beta$ and $\gamma$ need to be set up properly, and the table of contents data from all three runs needs to be accumulated. The \.{docmac} macros include two control sequences \.{\\contentsfile} and \.{\\readcontents} that facilitate the necessary processing. We include `\.{\\def\\contentsfile\{CONT1\}}' in the limbo section of~$\alpha$, we include `\.{\\def\\contentsfile\{CONT2\}}' in the limbo section of~$\beta$, and we include `\.{\\def\\contentsfile\{CONT3\}}' in the limbo section of~$\gamma$; this causes \TeX\ to write the contents data for $\alpha$,~$\beta$, and~$\gamma$ into \.{CONT1.TEX}, \.{CONT2.TEX}, and \.{CONT3.TEX}. Now in~$\gamma$ we say $$\.{\\def\\readcontents\{\\input CONT1 \\input CONT2 \\input CONT3\}};$$ this brings in the data from all three pieces, in the proper order. However, we still need to solve the page-numbering problem. One way to do it is to include the following in the limbo material for $\beta$: $$\lpile{\.{\\message\{Please type the last page number of part 1: \}}\cr \.{\\read -1 to \\temp \\pageno=\\temp \\advance\\pageno by 1}\cr}$$ Then you simply provide the necessary data when \TeX\ requests it; a similar construction is used at the beginning of $\gamma$. This method can, of course, be used to divide a docmac'ed file into any number of pieces. \sect The counter \.{\\sectno} is set to the number of the module being typeset. \sect We are finished and just have to restore the category codes. \beginprog \catcode`\@=\atcode \catcode`\_=\uscode \endinput \endprog \contentspage