1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/contrib/latex-to-docbook	Tue Aug 18 17:31:11 2009 +0200
     1.3 @@ -0,0 +1,198 @@
     1.4 +#!/usr/bin/python
     1.5 +#
     1.6 +# This is the most horrible of hacks. Pretend you're not looking.</para>
     1.7 +
     1.8 +import cStringIO as StringIO
     1.9 +import re, sys
    1.10 +
    1.11 +sections = {
    1.12 +    'chapter': 'chapter',
    1.13 +    'section': 'sect1',
    1.14 +    'subsection': 'sect2',
    1.15 +    'subsubsection': 'sect3',
    1.16 +    }
    1.17 +
    1.18 +envs = {
    1.19 +    'codesample2': 'programlisting',
    1.20 +    'codesample4': 'programlisting',
    1.21 +    'enumerate': 'orderedlist',
    1.22 +    'figure': 'informalfigure',
    1.23 +    'itemize': 'itemizedlist',
    1.24 +    'note': 'note',
    1.25 +    'quote': 'blockquote',
    1.26 +    }
    1.27 +
    1.28 +def process(ifp, ofp):
    1.29 +    print >> ofp, '<!-- vim: set filetype=docbkxml shiftwidth=2 autoindent expandtab tw=77 : -->\n'
    1.30 +    stack = []
    1.31 +    para = True
    1.32 +    inlist = 0
    1.33 +    for line in ifp:
    1.34 +        if line.startswith('%%% Local Variables:'):
    1.35 +            break
    1.36 +        line = (line.rstrip()
    1.37 +                .replace('~', ' ')
    1.38 +                .replace('&', '&amp;')
    1.39 +                .replace('---', '&emdash;')
    1.40 +                .replace('\_', '_')
    1.41 +                .replace('\{', '{')
    1.42 +                .replace('\}', '}')
    1.43 +                .replace('\$', '$')
    1.44 +                .replace('\%', '%')
    1.45 +                .replace('\#', '#')
    1.46 +                .replace('<', '&lt;')
    1.47 +                .replace('>', '&gt;')
    1.48 +                .replace('``', '<quote>')
    1.49 +                .replace("''", '</quote>')
    1.50 +                .replace('\\', '\\'))
    1.51 +        line = re.sub(r'\s*\\(?:centering|small)\b\s*', '', line)
    1.52 +        line = re.sub(r'\\(?:hgrc\\|hgrc)\b',
    1.53 +                      r'<filename role="special"> /.hgrc</filename>', line)
    1.54 +        line = re.sub(r'\\item\[(?P<key>[^]]+)\]', r'\item \g<key>:', line)
    1.55 +        line = re.sub(r'\\bug{(?P<id>\d+)}',
    1.56 +                      r'<ulink role="hg-bug" url="http://www.selenic.com/mercurial/bts/issue\g<id>">issue \g<id></ulink>', line)
    1.57 +        line = re.sub(r'\\cite{([^}]+)}', r'<citation>\1</citation>', line)
    1.58 +        line = re.sub(r'\\hggopt{(?P<opt>[^}]+)}',
    1.59 +                      r'<option role="hg-opt-global">\g<opt></option>', line)
    1.60 +        line = re.sub(r'\\hgxopt{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
    1.61 +                      r'<option role="hg-ext-\g<ext>-cmd-\g<cmd>-opt">\g<opt></option>', line)
    1.62 +        line = re.sub(r'\\hgxcmd{(?P<ext>[^}]+)}{(?P<cmd>[^}]+)}',
    1.63 +                      r'<command role="hg-ext-\g<ext>">\g<cmd></command>', line)
    1.64 +        line = re.sub(r'\\hgext{(?P<ext>[^}]+)}',
    1.65 +                      r'<literal role="hg-ext">\g<ext></literal>', line)
    1.66 +        line = re.sub(r'\\hgopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
    1.67 +                      r'<option role="hg-opt-\g<cmd>">\g<opt></option>',
    1.68 +                      line)
    1.69 +        line = re.sub(r'\\cmdopt{(?P<cmd>[^}]+)}{(?P<opt>[^}]+)}',
    1.70 +                      r'<option role="cmd-opt-\g<cmd>">\g<opt></option>',
    1.71 +                      line)
    1.72 +        line = re.sub(r'\\hgcmd{(?P<cmd>[^}]+)}',
    1.73 +                      r'<command role="hg-cmd">hg \g<cmd></command>', line)
    1.74 +        line = re.sub(r'\\caption{(?P<text>[^}]+?)}',
    1.75 +                      r'<caption><para>\g<text></para></caption>', line)
    1.76 +        line = re.sub(r'\\grafix{(?P<name>[^}]+)}',
    1.77 +                      r'<mediaobject><imageobject><imagedata fileref="\g<name>"/></imageobject><textobject><phrase>XXX add text</phrase></textobject></mediaobject>', line)
    1.78 +        line = re.sub(r'\\envar{(?P<name>[^}]+)}',
    1.79 +                      r'<envar>\g<name></envar>', line)
    1.80 +        line = re.sub(r'\\rcsection{(?P<sect>[^}]+)}',
    1.81 +                      r'<literal role="rc-\g<sect>">\g<sect></literal>', line)
    1.82 +        line = re.sub(r'\\rcitem{(?P<sect>[^}]+)}{(?P<name>[^}]+)}',
    1.83 +                      r'<envar role="rc-item-\g<sect>">\g<name></envar>', line)
    1.84 +        line = re.sub(r'\\dirname{(?P<dir>[^}]+?)}',
    1.85 +                      r'<filename class="directory">\g<dir></filename>', line)
    1.86 +        line = re.sub(r'\\filename{(?P<file>[^}]+?)}',
    1.87 +                      r'<filename>\g<file></filename>', line)
    1.88 +        line = re.sub(r'\\tildefile{(?P<file>[^}]+)}',
    1.89 +                      r'<filename role="home">~/\g<file></filename>', line)
    1.90 +        line = re.sub(r'\\sfilename{(?P<file>[^}]+)}',
    1.91 +                      r'<filename role="special">\g<file></filename>', line)
    1.92 +        line = re.sub(r'\\sdirname{(?P<dir>[^}]+)}',
    1.93 +                      r'<filename role="special" class="directory">\g<dir></filename>', line)
    1.94 +        line = re.sub(r'\\interaction{(?P<id>[^}]+)}',
    1.95 +                      r'<!-- &interaction.\g<id>; -->', line)
    1.96 +        line = re.sub(r'\\excode{(?P<id>[^}]+)}',
    1.97 +                      r'<!-- &example.\g<id>; -->', line)
    1.98 +        line = re.sub(r'\\pymod{(?P<mod>[^}]+)}',
    1.99 +                      r'<literal role="py-mod">\g<mod></literal>', line)
   1.100 +        line = re.sub(r'\\pymodclass{(?P<mod>[^}]+)}{(?P<class>[^}]+)}',
   1.101 +                      r'<literal role="py-mod-\g<mod>">\g<class></literal>', line)
   1.102 +        line = re.sub(r'\\url{(?P<url>[^}]+)}',
   1.103 +                      r'<ulink url="\g<url>">\g<url></ulink>', line)
   1.104 +        line = re.sub(r'\\href{(?P<url>[^}]+)}{(?P<text>[^}]+)}',
   1.105 +                      r'<ulink url="\g<url>">\g<text></ulink>', line)
   1.106 +        line = re.sub(r'\\command{(?P<cmd>[^}]+)}',
   1.107 +                      r'<command>\g<cmd></command>', line)
   1.108 +        line = re.sub(r'\\option{(?P<opt>[^}]+)}',
   1.109 +                      r'<option>\g<opt></option>', line)
   1.110 +        line = re.sub(r'\\ref{(?P<id>[^}]+)}', r'<xref linkend="\g<id>"/>', line)
   1.111 +        line = re.sub(r'\\emph{(?P<txt>[^}]+)}',
   1.112 +                      r'<emphasis>\g<txt></emphasis>', line)
   1.113 +        line = re.sub(r'\\texttt{(?P<txt>[^}]+)}',
   1.114 +                      r'<literal>\g<txt></literal>', line)
   1.115 +        line = re.sub(r'\\textbf{(?P<txt>[^}]+)}',
   1.116 +                      r'<emphasis role="bold">\g<txt></emphasis>', line)
   1.117 +        line = re.sub(r'\\hook{(?P<name>[^}]+)}',
   1.118 +                      r'<literal role="hook">\g<name></literal>', line)
   1.119 +        line = re.sub(r'\\tplfilter{(?P<name>[^}]+)}',
   1.120 +                      r'<literal role="template-filter">\g<name></literal>', line)
   1.121 +        line = re.sub(r'\\tplkword{(?P<name>[^}]+)}',
   1.122 +                      r'<literal role="template-keyword">\g<name></literal>', line)
   1.123 +        line = re.sub(r'\\tplkwfilt{(?P<tpl>[^}]+)}{(?P<name>[^}]+)}',
   1.124 +                      r'<literal role="template-kw-filt-\g<tpl>">\g<name></literal>', line)
   1.125 +        line = re.sub(r'\\[vV]erb(.)(?P<txt>[^\1]+?)\1',
   1.126 +                      r'<literal>\g<txt></literal>', line)
   1.127 +        line = re.sub(r'\\package{(?P<name>[^}]+)}',
   1.128 +                      r'<literal role="package">\g<name></literal>', line)
   1.129 +        line = re.sub(r'\\hgcmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
   1.130 +                      r'<command role="hg-cmd">hg \g<cmd> \g<args></command>',
   1.131 +                      line)
   1.132 +        line = re.sub(r'\\cmdargs{(?P<cmd>[^}]+)}{(?P<args>[^}]+)}',
   1.133 +                      r'<command>\g<cmd> \g<args></command>',
   1.134 +                      line)
   1.135 +        m = re.match(r'\\(chapter|section|subsection|subsubsection){(.*)}', line)
   1.136 +        if m:
   1.137 +            kind, content = m.groups()
   1.138 +            sec = sections[kind]
   1.139 +            while stack and stack[-1] >= sec:
   1.140 +                close = stack.pop()
   1.141 +                print >> ofp, '</%s>' % close
   1.142 +            stack.append(sec)
   1.143 +            print >> ofp, '<%s>\n<title>%s</title>' % (sec, content)
   1.144 +        else:
   1.145 +            m = re.match(r'\s*\\(begin|end){(?P<sect>[^}]+)}', line)
   1.146 +            if m:
   1.147 +                if not para:
   1.148 +                    print >> ofp, '</para>'
   1.149 +                    if inlist:
   1.150 +                        ofp.write('</listitem>')
   1.151 +                    para = True
   1.152 +                state, env = m.groups()
   1.153 +                env = envs[env]
   1.154 +                if state == 'begin':
   1.155 +                    ofp.write('<')
   1.156 +                    if env in ('itemizedlist', 'orderedlist'):
   1.157 +                        inlist = 1
   1.158 +                else:
   1.159 +                    ofp.write('</')
   1.160 +                    if env in ('itemizedlist', 'orderedlist'):
   1.161 +                        inlist = 0
   1.162 +                print >> ofp, env + '>'
   1.163 +            else:
   1.164 +                if line.startswith('\\item '):
   1.165 +                    if inlist > 1:
   1.166 +                        print >> ofp, '</para>'
   1.167 +                        print >> ofp, '</listitem>'
   1.168 +                    else:
   1.169 +                        inlist = 2
   1.170 +                    para = True
   1.171 +                    line = line[6:]
   1.172 +                if line and para:
   1.173 +                    if inlist:
   1.174 +                        ofp.write('<listitem>')
   1.175 +                    ofp.write('<para>')
   1.176 +                    para = False
   1.177 +                if not line and not para:
   1.178 +                    print >> ofp, '</para>'
   1.179 +                    if inlist:
   1.180 +                        ofp.write('</listitem>')
   1.181 +                    para = True
   1.182 +                print >> ofp, line
   1.183 +    while stack:
   1.184 +        print >> ofp, '</%s>' % stack.pop()
   1.185 +    ofp.write('\n'.join(['\n<!--',
   1.186 +                         'local variables: ',
   1.187 +                         'sgml-parent-document: ("00book.xml" "book" "chapter")',
   1.188 +                         'end:',
   1.189 +                         '-->']))
   1.190 +
   1.191 +
   1.192 +if __name__ == '__main__':
   1.193 +    for name in sys.argv[1:]:
   1.194 +        if not name.endswith('.tex'):
   1.195 +            continue
   1.196 +        newname = name[:-3] + 'xml'
   1.197 +        ofp = StringIO.StringIO()
   1.198 +        process(open(name), ofp)
   1.199 +        s = ofp.getvalue()
   1.200 +        s = re.sub('\n+</para>', '</para>', s, re.M)
   1.201 +        open(newname, 'w').write(s)