#LyX 1.4.1 created this file. For more info see http://www.lyx.org/ \lyxformat 245 \begin_document \begin_header \textclass literate-article \begin_preamble % % This relaxes the noweb constraint that chunks are % never broken across pages. % % This is from the noweb FAQ % \def\nwendcode{\endtrivlist \endgroup} \let\nwdocspar=\smallbreak \end_preamble \language english \inputencoding default \fontscheme default \graphics default \paperfontsize default \spacing single \papersize default \use_geometry false \use_amsmath 0 \cite_engine basic \use_bibtopic false \paperorientation portrait \secnumdepth 3 \tocdepth 3 \paragraph_separation indent \defskip medskip \quotes_language english \papercolumns 1 \papersides 1 \paperpagestyle default \tracking_changes false \output_changes true \end_header \begin_body \begin_layout Title \noun on noweb2lyx \end_layout \begin_layout Author Kayvan A. Sylvan \end_layout \begin_layout Date May 6, 1999 \end_layout \begin_layout Abstract This document describes and implements a perl script for importing noweb files into LyX \end_layout \begin_layout Standard \begin_inset LatexCommand \tableofcontents{} \end_inset \end_layout \begin_layout Standard \newpage \end_layout \begin_layout Section Introduction \end_layout \begin_layout Standard Since version 1.0.1, LyX now supports Literate Programming using \noun on noweb \noun default . This addition to LyX made it very pleasant to write programs in the literate style (like this one). In addition to being able to write new literate programs, it would be quite useful if old \noun on noweb \noun default code could be imported into LyX in some fashion. That's where this program comes in. \end_layout \begin_layout Standard The purpose of \noun on noweb2lyx \noun default is to convert a \noun on noweb \noun default file to LyX. \end_layout \begin_layout Scrap <>= \newline #!@PERL@ \newline # \newline # Copyright (C) 1999 Kayvan A. Sylvan \newline # You are free to use and modify this code under the terms of \newline # the GNU General Public Licence version 2 or later. \newline # \newline # Written with assistance from: \newline # Edmar Wienskoski Jr. \newline # Amir Karger \newline # \newline # $Id: noweb2lyx.lyx,v 1.5 2005/07/18 09:42:27 jamatos Exp $ \newline # \newline # NOTE: This file was automatically generated from noweb2lyx.lyx using noweb. \newline # \newline <> \newline <> \newline <> \newline @ \end_layout \begin_layout Section The Noweb file defined \end_layout \begin_layout Standard A \noun on noweb \noun default file is a collection of documentation and code chunks. Documentation chunks simply start with an ``@'' and have no name: \end_layout \begin_layout LyX-Code @ Here is some documentation. \newline We can do arbitrary LaTeX code here. \newline [... blah blah blah ...] \end_layout \begin_layout Standard Code chunks look like this: \end_layout \begin_layout LyX-Code < \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset >= \newline {... code for the chunk goes here ...} \newline @ \end_layout \begin_layout Standard The ``@'' is a necessary delimiter to end the code chunk. The other form that the ``@'' line takes is as follows: \end_layout \begin_layout LyX-Code < \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset >= \newline {... code for the chunk ...} \newline @ %def identifier1 identifier2 \end_layout \begin_layout Standard In the latter form, we are declaring to \noun on noweb \noun default that this code chunk defines identifier1, identifier2, etc. \end_layout \begin_layout Standard When first tackling this problem, I spoke with members of the LyX team that knew about the literate programming extensions and reLyX (the LaTeX importing code). \end_layout \begin_layout Standard One of the first ideas was to extend the reLyX code to understand the \noun on noweb \noun default code chunks. This proved to be too hard and presents other problems \begin_inset Foot status collapsed \begin_layout Standard Not the least of these problems is the fact that << is a quote in French. \end_layout \end_inset . On the other hand, it turns out that reLyX contains a very useful literal quoting mechanism. If the input file contains the construct \end_layout \begin_layout LyX-Code \backslash begin{reLyXskip} \newline {... LaTeX stuff ...} \newline \backslash end{reLyXskip} \end_layout \begin_layout Standard then reLyX will copy the surrounded code to the output file verbatim. Given this, the first part of the translation is easy; we simply have to copy the code chunks into an intermediate file that surrounds them with \family typewriter \backslash begin{reLyXskip} \family default and \family typewriter \backslash end{reLyXskip} \family default . \end_layout \begin_layout Standard Once reLyX is done with the input file, the problem is reduced to changing the code chunks from LyX's LaTeX layout to the Scrap layout. \end_layout \begin_layout Standard There is one final constraint on \noun on noweb2lyx \noun default . We want to be able to run it as a simple pre-processor and post-processor from within reLyX. We can accomplish this by setting the flags \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[pre_only]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit and \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[post_only]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit before we reach the main conversion code. \end_layout \begin_layout Standard With all that preamble out of the way, we now have the basic high-level outline for our code: \end_layout \begin_layout Scrap <>= \newline if (!$post_only) { \newline <> \newline } \newline if ((!$pre_only) && (!$post_only)) { \newline <> \newline } \newline if (!$pre_only) { \newline <> \newline } \newline <> \newline @ \end_layout \begin_layout Section Making a file that reLyX can process \end_layout \begin_layout Standard In this section, we present the code that performs the task of creating the intermediate file that reLyX can process, using the algorithm that we just outlined. This algorithm is outlined in the code that follows: \end_layout \begin_layout Scrap <>= \newline <> \newline inputline: while() \newline { \newline if (/^ \backslash s* \backslash < \backslash <.* \backslash > \backslash >=/) { # Beginning of a noweb scrap \newline <> \newline } elsif (/^@ \backslash s+(.*)/) { # Beginning of a documentation chunk \newline print OUTPUT $1; # We do not need the ``@'' part \newline } elsif (/ \backslash [ \backslash [.+ \backslash ] \backslash ]/) { # noweb quoted code \newline <> \newline } else { \newline print OUTPUT; # Just let the line pass through \newline } \newline } \newline <> \newline @ \end_layout \begin_layout Standard In the code above, we do some pre-processing of the noweb ``[[...]]'' construct. This avoids some problems with reLyX confusing lists composed of ``[[...]]'' constructs. \end_layout \begin_layout Scrap <>= \newline s/ \backslash [ \backslash [.+? \backslash ]{2,}/{$&}/g; \newline print OUTPUT; \newline @ \end_layout \begin_layout Standard While reading in the \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[INPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit file, once we have identified a \noun on noweb \noun default code chunk, we transform it into a form that is usable by reLyX. \end_layout \begin_layout Scrap <>= \newline <> \newline <> \newline <> \newline @ \end_layout \begin_layout Subsection File input and output for the pre-processing step \end_layout \begin_layout Standard In \noun on noweb2lyx \noun default , we will use \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[INPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit and \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[OUTPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit to read and write files. In the code fragment above, we need to read from the input file and write to a file that will be later transformed by reLyX. If we are being called only to pre-process the input file, then there is no need to create a temporary file. \end_layout \begin_layout Scrap <>= \newline if ($pre_only) { \newline &setup_files($input_file, $output_file); \newline } else { \newline $relyx_file = "temp$$"; \newline &setup_files($input_file, $relyx_file); \newline } \newline @ \end_layout \begin_layout Standard This code uses a small perl subroutine, \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[setup_files]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit , which we define below: \end_layout \begin_layout Scrap <>= \newline sub setup_files { \newline my($in, $out) = @_; \newline open(INPUT, "<$in") || die "Can not read $in: $! \backslash n"; \newline open(OUTPUT, ">$out") || die "Can not write $out: $! \backslash n"; \newline } \newline @ %def setup_files \end_layout \begin_layout Subsection Reading in the \noun on noweb \noun default scrap \end_layout \begin_layout Standard After we see the beginning of the scrap, we need to read in and save the rest of the scrap for output. \end_layout \begin_layout Scrap <>= \newline $savedScrap = $_; \newline $endLine = ""; \newline @ \end_layout \begin_layout Scrap <>= \newline scrapline: while () { \newline last scrapline if /^@ \backslash s+/; \newline $savedScrap .= $_; \newline }; \newline switch: { \newline if (/^@ \backslash s+$/) {$savedScrap .= $_; last switch; } \newline if (/^@ \backslash s+%def.*$/) {$savedScrap .= $_; last switch; } \newline if (/^@ \backslash s+(.*)$/) {$savedScrap .= "@ \backslash n"; $endLine = "$1 \backslash n"; } \newline } \newline @ \end_layout \begin_layout Subsection Printing out the scrap \end_layout \begin_layout Standard The final piece of the first pass of the conversion is done by this code. \end_layout \begin_layout Scrap <>= \newline print OUTPUT " \backslash \backslash begin{reLyXskip} \backslash n"; \newline print OUTPUT $savedScrap; \newline print OUTPUT " \backslash \backslash end{reLyXskip} \backslash n \backslash n"; \newline print OUTPUT "$endLine"; \newline @ \end_layout \begin_layout Standard Finally, we need to close the \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[INPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit and \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[OUTPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit files. \end_layout \begin_layout Scrap <>= \newline close(INPUT); \newline close(OUTPUT); \newline @ \end_layout \begin_layout Section Running reLyX \end_layout \begin_layout Standard In this section, we describe and implement the code that runs reLyX on the intermediate file \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[relyx_file]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit . \end_layout \begin_layout Subsection Selecting the document class \end_layout \begin_layout Standard In order to run reLyX, we need to know the article class of the input document (to choose the corresponding literate document layout). For this, we need to parse the intermediate file. \end_layout \begin_layout Scrap <>= \newline <> \newline <> \newline @ \end_layout \begin_layout Standard In the code below, you'll see a strange regular expression to search for the document class. The reason for this kludge is that without it, we can't run \noun on noweb2lyx \noun default on the \emph on noweb2lyx.nw \emph default file that is generated by LyX \begin_inset Foot status collapsed \begin_layout Standard reLyX searches for \backslash \backslash doc \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit ument \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard {} \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit class and gets confused, so we have to obfuscate it slightly. \end_layout \end_inset . With the regular expression as it is, we can actually run \noun on noweb2lyx \noun default on itself and a produce a quite reasonable LyX file. \end_layout \begin_layout Scrap <>= \newline open(INPUT, "<$relyx_file") || \newline die "Can not read $relyx_file: $! \backslash n"; \newline $class = "article"; # default if none found \newline parse: while() { \newline if (/ \backslash \backslash docu[m]entclass{(.*)}/) { \newline $class = $1; \newline last parse; \newline } \newline } \newline close(INPUT); \newline @ \end_layout \begin_layout Subsection Running reLyX with the corresponding literate document layout \end_layout \begin_layout Standard Now that we know what the document class ought to be, we do: \end_layout \begin_layout Scrap <>= \newline $doc_class = "literate-" . $class; \newline die "reLyX returned non-zero: $! \backslash n" \newline if (system("reLyX -c $doc_class $relyx_file")); \newline @ \end_layout \begin_layout Standard reLyX performs the main bulk of the translation work. Note that if the ``literate- \emph on class \emph default '' document layout is not found, then reLyX will fail with an error. In that case, you may need to modify your \noun on noweb \noun default input file to a supported document type. \end_layout \begin_layout Section Fixing the reLyX output \end_layout \begin_layout Standard We need to perform some post-processing of what reLyX produces in order to have the best output for our literate document. The outline of the post-processing steps are: \end_layout \begin_layout Scrap <>= \newline <> \newline line: while() \newline { \newline <> \newline <> \newline print OUTPUT; # default \newline } \newline <> \newline @ \end_layout \begin_layout Standard Note that in the perl code that is contained in the \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[while()]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit loop above, the perl construct \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[next line]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit is sufficient to restart the loop. We can use this construct to do some relatively complex parsing of the reLyX generated file. \end_layout \begin_layout Subsection File input and output for the post-processing \end_layout \begin_layout Standard Setting up the \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[INPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit and \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard [[OUTPUT]] \end_layout \end_inset \family default \series default \shape default \size default \emph default \bar default \noun default \color inherit is taken care of by this code: \end_layout \begin_layout Scrap <>= \newline if ($post_only) { \newline &setup_files("$input_file", "$output_file"); \newline } else { \newline &setup_files("$relyx_file.lyx", "$output_file"); \newline } \newline @ \end_layout \begin_layout Subsection Making sure the code chunks are in the Scrap layout \end_layout \begin_layout Standard Now, as we outlined before, the final step is transforming the code-chunks which have been put into a LaTeX layout by LyX into the scrap layout. \end_layout \begin_layout Scrap <>= \newline if (/ \backslash \backslash latex latex/) { # Beginning of some latex code \newline if (($line = ) =~ /^ \backslash s*<> \newline } else { \newline # print the \backslash latex latex line + next line \newline print OUTPUT "$_$line"; \newline } \newline next line; \newline } \newline @ \end_layout \begin_layout Standard When we are sure that we are in a code chunk, we must read in the rest of the code chunk and output a scrap layout for it: \end_layout \begin_layout Scrap <>= \newline $savedScrap = " \backslash \backslash layout Scrap \backslash n \backslash n$line"; \newline codeline: while () { \newline $savedScrap .= $_; \newline last codeline if /^@ \backslash s+/; \newline }; \newline print OUTPUT $savedScrap; \newline <> \newline @ \end_layout \begin_layout Standard Okay, now we just need to eat the rest of the latex layout. There should only be a few different types of lines for us to match: \end_layout \begin_layout Scrap <>= \newline slurp: while () { \newline last slurp if / \backslash \backslash latex /; \newline next slurp if / \backslash \backslash newline/; \newline next slurp if /^ \backslash s*$/; \newline warn "confused by line: $_"; \newline } \newline @ \end_layout \begin_layout Subsection Taking care of the \noun on noweb \noun default \emph on [[quoted code]] \emph default construct \end_layout \begin_layout Standard \noun on noweb \noun default allows the user to use a special code quoting mechanism in documentation chunks. Fixing this ``[[quoted-code]]'' \noun on noweb \noun default syntax means putting the ``[[quoted-code]]'' in a LaTeX layout in the LyX file. Otherwise, LyX will backslash-quote the brackets, creating ugly output. The quoted-code is transformed by \noun on noweb \noun default when it generates the final LaTeX code. \end_layout \begin_layout Scrap <>= \newline if (/ \backslash [ \backslash [.+ \backslash ] \backslash ]/) { # special code for [[var]] \newline s/ \backslash [ \backslash [.+? \backslash ]{2,}/ \backslash n \backslash \backslash latex latex \backslash n$& \backslash n \backslash \backslash latex default \backslash n/g; \newline print OUTPUT; \newline next line; \newline } \newline @ \end_layout \begin_layout Section Cleaning up intermediate files \end_layout \begin_layout Standard The cleanup code is very simple: \end_layout \begin_layout Scrap <>= \newline system("rm -f $relyx_file*") unless ($post_only || $pre_only); \newline @ \end_layout \begin_layout Section User supplied arguments \end_layout \begin_layout Standard The \noun on noweb2lyx \noun default script understands two arguments, input-file and output-file. It is also set up to be used internally by reLyX to pre-process or postprocess files in the import pipeline. \end_layout \begin_layout Scrap <>= \newline &usage() if ($#ARGV < 1); # zero or one argument \newline if ($ARGV[0] eq "-pre") { \newline &usage unless ($#ARGV == 2); \newline $input_file = $ARGV[1]; $output_file = $ARGV[2]; $pre_only = 1; \newline } elsif ($ARGV[0] eq "-post") { \newline &usage unless ($#ARGV == 2); \newline $input_file = $ARGV[1]; $output_file = $ARGV[2]; $post_only = 1; \newline } else { \newline &usage unless ($#ARGV == 1); \newline $input_file = $ARGV[0]; $output_file = $ARGV[1]; \newline $pre_only = 0; $post_only = 0; \newline } \newline @ %def input_file output_file pre_only post_only \end_layout \begin_layout Scrap <>= \newline sub usage() { \newline print "Usage: noweb2lyx [-pre | -post] input-file output-file \newline \newline If -pre is specified, only pre-processes the input-file for reLyX. \newline Similarly, in the case of -post, post-processes reLyX output. \newline In case of bugs, Email Kayvan Sylvan . \backslash n"; \newline exit; \newline } \newline @ %def usage \end_layout \begin_layout Section Generating the \noun on noweb2lyx \noun default script \end_layout \begin_layout Standard The noweb2lyx script can be tangled from LyX if you set \family typewriter \backslash build_command \family default to call a generic script that always extracts a scrap named \family typewriter build-script \family default and executes it. Here is an example of such a script: \end_layout \begin_layout LyX-Code #!/bin/sh \newline notangle -Rbuild-script $1 | sh \end_layout \begin_layout Scrap <>= \newline PREFIX=/usr \newline notangle -Rnoweb2lyx.in noweb2lyx.nw > noweb2lyx.in \newline sed -e "s=@PERL@=$PREFIX/bin/perl=" noweb2lyx.in > noweb2lyx \newline chmod +x noweb2lyx \newline @ \end_layout \begin_layout Standard \newpage \end_layout \begin_layout Section* Macros \end_layout \begin_layout Standard \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard \backslash nowebchunks \end_layout \end_inset \end_layout \begin_layout Section* Identifiers \end_layout \begin_layout Standard \family roman \series medium \shape up \size normal \emph off \bar no \noun off \color none \begin_inset ERT status collapsed \begin_layout Standard \backslash nowebindex \end_layout \end_inset \end_layout \end_body \end_document