annotate tools/latex-to-docbook @ 651:cf006cabe489

Snapshot of conversion script

#!/usr/bin/python
#
# This is the most horrible of hacks. Pretend you're not looking.</para>

import cStringIO as StringIO
import re, sys

sections = {
    'chapter': 'chapter',
    'section': 'sect1',
    'subsection': 'sect2',
    'subsubsection': 'sect3',
    }

envs = {
    'codesample2': 'programlisting',
    'codesample4': 'programlisting',
    'enumerate': 'orderedlist',
    'figure': 'figure',
    'itemize': 'itemizedlist',
    'note': 'note',
    'quote': 'blockquote',
    }

def process(ifp, ofp):
    stack = []
    para = True
    inlist = False
    for line in ifp:
        if line.startswith('%%% Local Variables:'):
            break
        line = (line.rstrip()
                .replace(' ', ' ')
                .replace('&', '&amp;')
                .replace('&emdash;', '&emdash;')
                .replace('\_', '_')
                .replace('\{', '{')
                .replace('\}', '}')
                .replace('\$', '$')
                .replace('\%', '%')
                .replace('\#', '#')
                .replace('<', '&lt;')
                .replace('>', '&gt;')
                .replace('<quote>', '<quote>')
                .replace("</quote>", '</quote>')
                .replace('\\', '\\'))
        line = re.sub(r'\s*\\(?:centering|small)\b\s*', '', line)
        line = re.sub(r'\\(?:hgrc\\|hgrc)\b',
                      r'<filename role="special"> /.hgrc</filename>', line)
        line = re.sub(r'\\item\[(?P<key>[^]]+)\]', r'\item \g<key>:', line)
        line = re.sub(r'\\bug{(?P<id>\d+)}',
                      r'<ulink role="hg-bug" url="http://www.selenic.com/mercurial/bts/issue\g<id>">issue \g<id></ulink>', line)
        line = re.sub(r'\\cite{([^}]+)}', r'<citation>\1</citation>', line)
        line = re.sub(r'\\hggopt{(?P<opt>[^}]+)}',
                      r'<option role="hg-opt-global">\g<opt></option>', line)
        line = re.sub(r'\\hgxopt{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
                      r'<option role="hg-ext-\g<ext>-cmd-\g<cmd>-opt">\g<opt></option>', line)
        line = re.sub(r'\\hgxcmd{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}',
                      r'<command role="hg-ext-\g<ext>">\g<cmd></command>', line)
        line = re.sub(r'\\hgext{(?P<ext>[^}]+)}',
                      r'<literal role="hg-ext">\g<ext></literal>', line)
        line = re.sub(r'\\hgopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
                      r'<option role="hg-opt-\g<cmd>">\g<opt></option>',
                      line)
        line = re.sub(r'\\cmdopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
                      r'<option role="cmd-opt-\g<cmd>">\g<opt></option>',
                      line)
        line = re.sub(r'\\hgcmd{(?P<cmd>[^}]+)}',
                      r'<command role="hg-cmd">hg \g<cmd></command>', line)
        line = re.sub(r'\\caption{(?P<text>[^}]+?)}',
                      r'<caption>\g<text></caption>', line)
        line = re.sub(r'\\grafix{(?P<name>[^}]+)}',
                      r'<mediaobject><imageobject><imagedata fileref="\g<name>"/></imageobject><textobject><phrase>XXX add text</phrase></textobject></mediaobject>', line)
        line = re.sub(r'\\envar{(?P<name>[^}]+)}',
                      r'<envar>\g<name></envar>', line)
        line = re.sub(r'\\rcsection{(?P<sect>[^}]+)}',
                      r'<literal role="rc-\g<sect>">\g<sect></literal>', line)
        line = re.sub(r'\\rcitem{(?P<sect>[^}]+)}{(?P<name>[^}]+)}',
                      r'<envar role="rc-item-\g<sect>">\g<name></envar>', line)
        line = re.sub(r'\\dirname{(?P<dir>[^}]+?)}',
                      r'<filename class="directory">\g<dir></filename>', line)
        line = re.sub(r'\\filename{(?P<file>[^}]+?)}',
                      r'<filename>\g<file></filename>', line)
        line = re.sub(r'\\tildefile{(?P<file>[^}]+)}',
                      r'<filename role="home"> /\g<file></filename>', line)
        line = re.sub(r'\\sfilename{(?P<file>[^}]+)}',
                      r'<filename role="special">\g<file></filename>', line)
        line = re.sub(r'\\sdirname{(?P<dir>[^}]+)}',
                      r'<filename role="special" class="directory">\g<dir></filename>', line)
        line = re.sub(r'\\interaction{(?P<id>[^}]+)}',
                      r'<!-- &interaction.\g<id>; -->', line)
        line = re.sub(r'\\excode{(?P<id>[^}]+)}',
                      r'<!-- &example.\g<id>; -->', line)
        line = re.sub(r'\\pymod{(?P<mod>[^}]+)}',
                      r'<literal role="py-mod">\g<mod></literal>', line)
        line = re.sub(r'\\pymodclass{(?P<mod>[^}]+)}{(?P<class>[^}]+)}',
                      r'<literal url="py-mod-\g<mod>">\g<class></ulink>', line)
        line = re.sub(r'\\url{(?P<url>[^}]+)}',
                      r'<ulink url="\g<url>">\g<url></ulink>', line)
        line = re.sub(r'\\href{(?P<url>[^}]+)}{(?P<text>[^}]+)}',
                      r'<ulink url="\g<url>">\g<text></ulink>', line)
        line = re.sub(r'\\command{(?P<cmd>[^}]+)}',
                      r'<command>\g<cmd></command>', line)
        line = re.sub(r'\\option{(?P<opt>[^}]+)}',
                      r'<option>\g<opt></option>', line)
        line = re.sub(r'\\ref{(?P<id>[^}]+)}', r'<xref id="\g<id>"/>', line)
        line = re.sub(r'\\emph{(?P<txt>[^}]+)}',
                      r'<emphasis>\g<txt></emphasis>', line)
        line = re.sub(r'\\texttt{(?P<txt>[^}]+)}',
                      r'<literal>\g<txt></literal>', line)
        line = re.sub(r'\\textbf{(?P<txt>[^}]+)}',
                      r'<emphasis role="bold">\g<txt></emphasis>', line)
        line = re.sub(r'\\hook{(?P<name>[^}]+)}',
                      r'<literal role="hook">\g<name></literal>', line)
        line = re.sub(r'\\tplfilter{(?P<name>[^}]+)}',
                      r'<literal role="template-filter">\g<name></literal>', line)
        line = re.sub(r'\\tplkword{(?P<name>[^}]+)}',
                      r'<literal role="template-keyword">\g<name></literal>', line)
        line = re.sub(r'\\tplkwfilt{(?P<tpl>[^}]+)}{(?P<name>[^}]+)}',
                      r'<literal role="template-kw-filt-\g<tpl>">\g<name></literal>', line)
        line = re.sub(r'\\[vV]erb(.)(?P<txt>[^\1]+?)\1',
                      r'<literal>\g<txt></literal>', line)
        line = re.sub(r'\\package{(?P<name>[^}]+)}',
                      r'<literal role="package">\g<name></literal>', line)
        line = re.sub(r'\\hgcmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
                      r'<command role="hg-cmd">hg \g<cmd> \g<args></command>',
                      line)
        line = re.sub(r'\\cmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
                      r'<command>\g<cmd> \g<args></command>',
                      line)
        m = re.match(r'\\(chapter|section|subsection|subsubsection){(.*)}', line)
        if m:
            kind, content = m.groups()
            sec = sections[kind]
            while stack and stack[-1] >= sec:
                close = stack.pop()
                print >> ofp, '</%s>' % close
            stack.append(sec)
            print >> ofp, '<%s>\n<title>%s</title>' % (sec, content)
        else:
            m = re.match(r'\s*\\(begin|end){(?P<sect>[^}]+)}', line)
            if m:
                if not para:
                    print >> ofp, '</para>'
                    if inlist:
                        ofp.write('</listitem>')
                    para = True
                state, env = m.groups()
                env = envs[env]
                if state == 'begin':
                    ofp.write('<')
                    if env == 'itemizedlist':
                        inlist = True
                else:
                    ofp.write('</')
                    if env == 'itemizedlist':
                        inlist = False
                print >> ofp, env + '>'
            else:
                if line.startswith('\\item '):
                    para = True
                    line = line[6:]
                if line and para:
                    if inlist:
                        ofp.write('<listitem>')
                    ofp.write('<para>')
                    para = False
                if not line and not para:
                    print >> ofp, '</para>'
                    if inlist:
                        ofp.write('</listitem>')
                    para = True
                print >> ofp, line
    while stack:
        print >> ofp, '</%s>' % stack.pop()
    ofp.write('\n'.join(['<!--',
                         'local variables: ',
                         'sgml-parent-document: ("00book.xml" "book" "chapter")',
                         'end:',
                         '-->']))


if __name__ == '__main__':
    for name in sys.argv[1:]:
        if not name.endswith('.tex'):
            continue
        newname = name[:-3] + 'xml'
        ofp = StringIO.StringIO()
        process(open(name), ofp)
        s = ofp.getvalue()
        s = re.sub('\n+</para>', '</para>', s, re.M)
        open(newname, 'w').write(s)