bos@552: #!/usr/bin/python
bos@552: #
bos@552: # This is the most horrible of hacks. Pretend you're not looking.</para>
bos@552: 
bos@552: import cStringIO as StringIO
bos@552: import re, sys
bos@552: 
bos@552: sections = {
bos@552:     'chapter': 'chapter',
bos@552:     'section': 'sect1',
bos@552:     'subsection': 'sect2',
bos@552:     'subsubsection': 'sect3',
bos@552:     }
bos@552: 
bos@552: envs = {
bos@552:     'codesample2': 'programlisting',
bos@552:     'codesample4': 'programlisting',
bos@552:     'enumerate': 'orderedlist',
bos@556:     'figure': 'informalfigure',
bos@552:     'itemize': 'itemizedlist',
bos@552:     'note': 'note',
bos@552:     'quote': 'blockquote',
bos@552:     }
bos@552: 
bos@552: def process(ifp, ofp):
bos@556:     print >> ofp, '<!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : -->\n'
bos@552:     stack = []
bos@552:     para = True
bos@556:     inlist = 0
bos@552:     for line in ifp:
bos@552:         if line.startswith('%%% Local Variables:'):
bos@552:             break
bos@552:         line = (line.rstrip()
bos@556:                 .replace('~', ' ')
bos@552:                 .replace('&', '&amp;')
bos@559:                 .replace('---', '&emdash;')
bos@552:                 .replace('\_', '_')
bos@552:                 .replace('\{', '{')
bos@552:                 .replace('\}', '}')
bos@552:                 .replace('\$', '$')
bos@552:                 .replace('\%', '%')
bos@552:                 .replace('\#', '#')
bos@552:                 .replace('<', '&lt;')
bos@552:                 .replace('>', '&gt;')
bos@556:                 .replace('``', '<quote>')
bos@556:                 .replace("''", '</quote>')
bos@552:                 .replace('\\', '\\'))
bos@552:         line = re.sub(r'\s*\\(?:centering|small)\b\s*', '', line)
bos@552:         line = re.sub(r'\\(?:hgrc\\|hgrc)\b',
bos@552:                       r'<filename role="special"> /.hgrc</filename>', line)
bos@552:         line = re.sub(r'\\item\[(?P<key>[^]]+)\]', r'\item \g<key>:', line)
bos@552:         line = re.sub(r'\\bug{(?P<id>\d+)}',
bos@552:                       r'<ulink role="hg-bug" url="http://www.selenic.com/mercurial/bts/issue\g<id>">issue \g<id></ulink>', line)
bos@552:         line = re.sub(r'\\cite{([^}]+)}', r'<citation>\1</citation>', line)
bos@552:         line = re.sub(r'\\hggopt{(?P<opt>[^}]+)}',
bos@552:                       r'<option role="hg-opt-global">\g<opt></option>', line)
bos@552:         line = re.sub(r'\\hgxopt{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
bos@552:                       r'<option role="hg-ext-\g<ext>-cmd-\g<cmd>-opt">\g<opt></option>', line)
bos@552:         line = re.sub(r'\\hgxcmd{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}',
bos@552:                       r'<command role="hg-ext-\g<ext>">\g<cmd></command>', line)
bos@552:         line = re.sub(r'\\hgext{(?P<ext>[^}]+)}',
bos@552:                       r'<literal role="hg-ext">\g<ext></literal>', line)
bos@552:         line = re.sub(r'\\hgopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
bos@552:                       r'<option role="hg-opt-\g<cmd>">\g<opt></option>',
bos@552:                       line)
bos@552:         line = re.sub(r'\\cmdopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
bos@552:                       r'<option role="cmd-opt-\g<cmd>">\g<opt></option>',
bos@552:                       line)
bos@552:         line = re.sub(r'\\hgcmd{(?P<cmd>[^}]+)}',
bos@552:                       r'<command role="hg-cmd">hg \g<cmd></command>', line)
bos@552:         line = re.sub(r'\\caption{(?P<text>[^}]+?)}',
bos@556:                       r'<caption><para>\g<text></para></caption>', line)
bos@552:         line = re.sub(r'\\grafix{(?P<name>[^}]+)}',
bos@552:                       r'<mediaobject><imageobject><imagedata fileref="\g<name>"/></imageobject><textobject><phrase>XXX add text</phrase></textobject></mediaobject>', line)
bos@552:         line = re.sub(r'\\envar{(?P<name>[^}]+)}',
bos@552:                       r'<envar>\g<name></envar>', line)
bos@552:         line = re.sub(r'\\rcsection{(?P<sect>[^}]+)}',
bos@552:                       r'<literal role="rc-\g<sect>">\g<sect></literal>', line)
bos@552:         line = re.sub(r'\\rcitem{(?P<sect>[^}]+)}{(?P<name>[^}]+)}',
bos@552:                       r'<envar role="rc-item-\g<sect>">\g<name></envar>', line)
bos@552:         line = re.sub(r'\\dirname{(?P<dir>[^}]+?)}',
bos@552:                       r'<filename class="directory">\g<dir></filename>', line)
bos@552:         line = re.sub(r'\\filename{(?P<file>[^}]+?)}',
bos@552:                       r'<filename>\g<file></filename>', line)
bos@552:         line = re.sub(r'\\tildefile{(?P<file>[^}]+)}',
bos@559:                       r'<filename role="home">~/\g<file></filename>', line)
bos@552:         line = re.sub(r'\\sfilename{(?P<file>[^}]+)}',
bos@552:                       r'<filename role="special">\g<file></filename>', line)
bos@552:         line = re.sub(r'\\sdirname{(?P<dir>[^}]+)}',
bos@552:                       r'<filename role="special" class="directory">\g<dir></filename>', line)
bos@552:         line = re.sub(r'\\interaction{(?P<id>[^}]+)}',
bos@552:                       r'<!-- &interaction.\g<id>; -->', line)
bos@552:         line = re.sub(r'\\excode{(?P<id>[^}]+)}',
bos@552:                       r'<!-- &example.\g<id>; -->', line)
bos@552:         line = re.sub(r'\\pymod{(?P<mod>[^}]+)}',
bos@552:                       r'<literal role="py-mod">\g<mod></literal>', line)
bos@552:         line = re.sub(r'\\pymodclass{(?P<mod>[^}]+)}{(?P<class>[^}]+)}',
bos@559:                       r'<literal role="py-mod-\g<mod>">\g<class></literal>', line)
bos@552:         line = re.sub(r'\\url{(?P<url>[^}]+)}',
bos@552:                       r'<ulink url="\g<url>">\g<url></ulink>', line)
bos@552:         line = re.sub(r'\\href{(?P<url>[^}]+)}{(?P<text>[^}]+)}',
bos@552:                       r'<ulink url="\g<url>">\g<text></ulink>', line)
bos@552:         line = re.sub(r'\\command{(?P<cmd>[^}]+)}',
bos@552:                       r'<command>\g<cmd></command>', line)
bos@552:         line = re.sub(r'\\option{(?P<opt>[^}]+)}',
bos@552:                       r'<option>\g<opt></option>', line)
bos@556:         line = re.sub(r'\\ref{(?P<id>[^}]+)}', r'<xref linkend="\g<id>"/>', line)
bos@552:         line = re.sub(r'\\emph{(?P<txt>[^}]+)}',
bos@552:                       r'<emphasis>\g<txt></emphasis>', line)
bos@552:         line = re.sub(r'\\texttt{(?P<txt>[^}]+)}',
bos@552:                       r'<literal>\g<txt></literal>', line)
bos@552:         line = re.sub(r'\\textbf{(?P<txt>[^}]+)}',
bos@552:                       r'<emphasis role="bold">\g<txt></emphasis>', line)
bos@552:         line = re.sub(r'\\hook{(?P<name>[^}]+)}',
bos@552:                       r'<literal role="hook">\g<name></literal>', line)
bos@552:         line = re.sub(r'\\tplfilter{(?P<name>[^}]+)}',
bos@552:                       r'<literal role="template-filter">\g<name></literal>', line)
bos@552:         line = re.sub(r'\\tplkword{(?P<name>[^}]+)}',
bos@552:                       r'<literal role="template-keyword">\g<name></literal>', line)
bos@552:         line = re.sub(r'\\tplkwfilt{(?P<tpl>[^}]+)}{(?P<name>[^}]+)}',
bos@552:                       r'<literal role="template-kw-filt-\g<tpl>">\g<name></literal>', line)
bos@552:         line = re.sub(r'\\[vV]erb(.)(?P<txt>[^\1]+?)\1',
bos@552:                       r'<literal>\g<txt></literal>', line)
bos@552:         line = re.sub(r'\\package{(?P<name>[^}]+)}',
bos@552:                       r'<literal role="package">\g<name></literal>', line)
bos@552:         line = re.sub(r'\\hgcmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
bos@552:                       r'<command role="hg-cmd">hg \g<cmd> \g<args></command>',
bos@552:                       line)
bos@552:         line = re.sub(r'\\cmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
bos@552:                       r'<command>\g<cmd> \g<args></command>',
bos@552:                       line)
bos@552:         m = re.match(r'\\(chapter|section|subsection|subsubsection){(.*)}', line)
bos@552:         if m:
bos@552:             kind, content = m.groups()
bos@552:             sec = sections[kind]
bos@552:             while stack and stack[-1] >= sec:
bos@552:                 close = stack.pop()
bos@552:                 print >> ofp, '</%s>' % close
bos@552:             stack.append(sec)
bos@552:             print >> ofp, '<%s>\n<title>%s</title>' % (sec, content)
bos@552:         else:
bos@552:             m = re.match(r'\s*\\(begin|end){(?P<sect>[^}]+)}', line)
bos@552:             if m:
bos@552:                 if not para:
bos@552:                     print >> ofp, '</para>'
bos@552:                     if inlist:
bos@552:                         ofp.write('</listitem>')
bos@552:                     para = True
bos@552:                 state, env = m.groups()
bos@552:                 env = envs[env]
bos@552:                 if state == 'begin':
bos@552:                     ofp.write('<')
bos@556:                     if env in ('itemizedlist', 'orderedlist'):
bos@556:                         inlist = 1
bos@552:                 else:
bos@552:                     ofp.write('</')
bos@559:                     if env in ('itemizedlist', 'orderedlist'):
bos@556:                         inlist = 0
bos@552:                 print >> ofp, env + '>'
bos@552:             else:
bos@552:                 if line.startswith('\\item '):
bos@556:                     if inlist > 1:
bos@556:                         print >> ofp, '</para>'
bos@556:                         print >> ofp, '</listitem>'
bos@556:                     else:
bos@556:                         inlist = 2
bos@552:                     para = True
bos@552:                     line = line[6:]
bos@552:                 if line and para:
bos@552:                     if inlist:
bos@552:                         ofp.write('<listitem>')
bos@552:                     ofp.write('<para>')
bos@552:                     para = False
bos@552:                 if not line and not para:
bos@552:                     print >> ofp, '</para>'
bos@552:                     if inlist:
bos@552:                         ofp.write('</listitem>')
bos@552:                     para = True
bos@552:                 print >> ofp, line
bos@552:     while stack:
bos@552:         print >> ofp, '</%s>' % stack.pop()
bos@556:     ofp.write('\n'.join(['\n<!--',
bos@552:                          'local variables: ',
bos@552:                          'sgml-parent-document: ("00book.xml" "book" "chapter")',
bos@552:                          'end:',
bos@552:                          '-->']))
bos@552: 
bos@552: 
bos@552: if __name__ == '__main__':
bos@552:     for name in sys.argv[1:]:
bos@552:         if not name.endswith('.tex'):
bos@552:             continue
bos@552:         newname = name[:-3] + 'xml'
bos@552:         ofp = StringIO.StringIO()
bos@552:         process(open(name), ofp)
bos@552:         s = ofp.getvalue()
bos@552:         s = re.sub('\n+</para>', '</para>', s, re.M)
bos@552:         open(newname, 'w').write(s)