#!/usr/bin/env python
"""
Naive DocBook to asciidoc translatiion.
"""

import re, sys

header412 = r'''<\?xml version="1.0" encoding="ISO-8859-1"\?>
<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
\s*"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd">
'''

refentry = """<refentry id='[^>]*'>
<refmeta>
<refentrytitle>[^>]*</refentrytitle>
<manvolnum>([^>]*)</manvolnum>
<refmiscinfo class='source'>[^>]*</refmiscinfo>
</refmeta>
<refnamediv id='name'>
<refname>([^>]*)</refname>
<refpurpose>([^>]*)</refpurpose>
</refnamediv>
"""

adocentry = r"""= \2(\1) =
:doctype: manpage

== NAME ==
\2 - \3
"""

transforms = (
    (header412, ""),
    (refentry, adocentry), 
    ("<para>", ""),
    ("</para>", ""),
    ("<emphasis>", "_"),
    ("</emphasis>", "_"),
    ("<bold>", "*"),
    ("</bold>", "*"),
    ("<quote>", "\""),
    ("</quote>", "\""),
    ("<firstterm>", ""),
    ("</firstterm>", ""),
    ("<application>", ""),
    ("</application>", ""),
    ("<email>", "<"),
    ("</email>", ">"),
    ("<command>", "+"),
    ("</command>", "+"),
    ("&mdash;", "--"),	# Maybe not the right thing?
    # Section markup
    ("<sect1><title>([^>]*)</title>", r'== \1 =='),
    ("<sect1 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n== \2 =='),
    ("</sect1>", ""),
    ("<sect2><title>([^>]*)</title>", r'=== \1 ==='),
    ("<sect2 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n=== \2 ==='),
    ("</sect2>", ""),
    ("<sect3><title>([^>]*)</title>", r'=== \1 ==='),
    ("<sect3 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n==== \2 ===='),
    ("</sect3>", ""),
    ("<refsect1><title>([^>]*)</title>", r'== \1 =='),
    ("<refsect1 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n== \2 =='),
    ("</refsect1>", ""),
    ("<refsect2><title>([^>]*)</title>", r'=== \1 ==='),
    ("<refsect2 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n=== \2 ==='),
    ("</refsect2>", ""),
    ("<refsect3><title>([^>]*)</title>", r'=== \1 ==='),
    ("<refsect3 id='([^']*)'><title>([^>]*)</title>", r'[[\1]]\n==== \2 ===='),
    ("</refsect3>", ""),
    ("</article>\n", "// end"),
    ("</refentry>\n", "// end"),
    # Miscellanea 
    ("&lt;", "<"),
    ("&gt;", ">"),
    ("&apos;", "'"),
    ("<programlisting remap='.nf'>", "---"),
    ("</programlisting> <!-- .fi -->", "---"),
    ("<programlisting>", "---"),
    ("</programlisting>", "---"),
    ("<citerefentry><refentrytitle>([^>]*)</refentrytitle><manvolnum>([^>]*)</manvolnum></citerefentry>", r"\1(\2)"),
    ("<literallayout[^>]*>", "---"), 
    ("</literallayout>", "---"), 
    ("<variablelist>\n", ""), 
    ("</variablelist>\n", ""), 
    ("<varlistentry>\n", ""), 
    ("</varlistentry>\n", ""), 
    ("</varlistentry>\n", ""), 
    ("<term>([^>]*)</term>", r"\1::"),
    ("<arg choice='opt'>([^>]*)</arg>", r"[\1]"),
    )

xml = re.compile("<.*>|&.*;")

def dedoc(inpath, outpath):
    contents = open(inpath).read()
    for (source, target) in compiled_transforms:
        contents = source.sub(target, contents)
    open(outpath, "w").write(contents)
    #for (n, text) in enumerate(open(outpath).readlines()):
    #    if xml.searc(text):
    #        sys.stderr.write("%s:%d: %s" % (outpath, n+1, text))

if __name__ == "__main__":
    compiled_transforms = [(re.compile(f), t) for (f, t) in transforms]
    if len(sys.argv) == 1:
        dedoc("/dev/stdin", "/dev/stdout")
    else:
        for fn in sys.argv[1:]:
            dedoc(fn, fn)