attachment:docbook2wiki.py von DocBookDemystificationHowTo

   1 #!/usr/bin/python2.2
   2 
   3 import xml.parsers.expat
   4 import sys, string, re
   5 
   6 class xml2wiki:
   7     def __init__(self, infile=None, outfile=None):
   8         self.infile = infile
   9 	self.outfile = outfile
  10 	self.out = None
  11         self.p = xml.parsers.expat.ParserCreate()
  12         self.p.StartElementHandler = self.start_element
  13         self.p.EndElementHandler = self.end_element
  14         self.p.DefaultHandler = self.default_handler
  15         self.p.CharacterDataHandler = self.char_data
  16         self.level = 0 # sections
  17 	self.listlevel = 0 # lists
  18 	self.ignore = 0
  19 	self.ignorelist = ['articleinfo', 'indexterm', 'application']
  20         self.lastchars = ""
  21         self.vt = 0 # verbatim
  22 	
  23     def feed(self, data):
  24         self.p.Parse(data, 0)
  25 
  26     def close(self):
  27         self.p.Parse("", 1) # end of data
  28         del self.p # get rid of circular references
  29 
  30     # sections, lists, paragraphs
  31     def start_article(self, attrs): self.level+=1
  32     def end_article(self): self.level-=1
  33     def start_sect1(self, attrs): self.level+=1
  34     def end_sect1(self): self.level-=1
  35     def start_sect2(self, attrs): self.level+=1
  36     def end_sect2(self): self.level-=1
  37     def start_variablelist(self, attrs): self.listlevel+=1
  38     def end_variablelist(self): self.listlevel-=1
  39     def start_varlistentry(self, attrs): self.write(" "*self.listlevel+"* ")
  40     def end_varlistentry(self): pass
  41     def start_listitem(self, attrs): self.write(" "*self.listlevel+" ")
  42     def end_listitem(self): pass
  43     def start_title(self, attrs): self.write("\n"+"="*self.level+" ")
  44     def end_title(self): self.write(" "+"="*self.level+"\n")
  45     def start_abstract(self, attrs): pass
  46     def end_abstract(self): self.write("\n\n")
  47     def start_para(self, attrs): pass
  48     def end_para(self): self.write("\n\n")
  49 
  50     # emphasis
  51     def start_emphasis(self, attrs): self.write("''")
  52     def end_emphasis(self): self.write("''")
  53     def start_term(self, attrs): self.write("'''")
  54     def end_term(self): self.write("'''")
  55     def start_firstterm(self, attrs): self.write("'''")
  56     def end_firstterm(self): self.write("'''")
  57 
  58     # code
  59     def start_programlisting(self, attrs): self.vt+=1; self.write("\n{{{")
  60     def end_programlisting(self): self.vt-=1; self.write("}}}\n")
  61     def start_screen(self, attrs): self.vt+=1; self.write("\n{{{")
  62     def end_screen(self): self.vt-=1; self.write("}}}\n")
  63     def start_markup(self, attrs): self.write("{{{")
  64     def end_markup(self): self.write("}}}")
  65     def start_command(self, attrs): self.write("{{{")
  66     def end_command(self): self.write("}}}")
  67     def start_option(self, attrs): self.write("{{{")
  68     def end_option(self): self.write("}}}")
  69     def start_filename(self, attrs): self.write("{{{")
  70     def end_filename(self): self.write("}}}")
  71 
  72     # links
  73     def start_ulink(self, attrs): self.write("[%s " % attrs['url'])
  74     def end_ulink(self): self.write("]")
  75     
  76     def start_mediaobject(self, attrs): pass
  77     def end_mediaobject(self): pass
  78     def start_imageobject(self, attrs): pass
  79     def end_imageobject(self): pass
  80     def start_imagedata(self, attrs): self.write("\nattachment:%s\n" % attrs['fileref'])
  81     def end_imagedata(self): self.write("\n")
  82 
  83     # misc
  84     def start_citetitle(self, attrs): pass
  85     def end_citetitle(self): pass
  86     def start_citerefentry(self, attrs): pass
  87     def end_citerefentry(self): pass
  88     def start_refentrytitle(self, attrs): self.write("'''")
  89     def end_refentrytitle(self): self.write("'''")
  90     def start_manvolnum(self, attrs): self.write("(")
  91     def end_manvolnum(self): self.write(")")
  92 
  93     def start_element(self, name, attrs):
  94         if name in self.ignorelist: self.ignore+=1
  95         if not self.ignore:
  96 	    try:
  97                 return getattr(self, 'start_' + name)(attrs)
  98             except:
  99 	        self.write("<<<"+name+":")
 100 	    
 101     def end_element(self, name):
 102         if not self.ignore:
 103 	    try:
 104 	        return getattr(self, 'end_' + name)()
 105             except:
 106 	        self.write(":"+name+">>>")
 107         if name in self.ignorelist: self.ignore-=1 
 108 
 109     def char_data(self, data):
 110         if not self.ignore:
 111 	    if data.startswith("\n\n") and self.lastchars.endswith("\n\n"):
 112 	        data = data[2:]
 113 	    elif data.startswith("\n") and self.lastchars.endswith("\n"):
 114 	        data = data[1:]
 115 	    if not self.vt:
 116 	        data = ' '.join(data.strip().split())+" "
 117 	    if len(data):
 118 	        self.write(data)
 119 	        if len(data)>1:
 120 		    self.lastchars=data[-2]
 121                 else:
 122 		    self.lastchars=data[-1]
 123 
 124     def default_handler(self, data):
 125         pass # self.write('Default:', repr(data))
 126 
 127     def parse(self):
 128         if self.infile:
 129             f = open(self.infile)
 130 	else:
 131 	    f = sys.stdin
 132 	if self.outfile:
 133 	    self.out = open(self.outfile, "w")
 134 	else:
 135 	    self.out = sys.stdout
 136 	    
 137 	text = f.read()
 138 	f.close()
 139 	text = re.sub(r'>\s*<',r'><',text) # kill whitespace between tags
 140 	self.feed(text)
 141         self.close()
 142 
 143     def write(self, what):
 144         self.out.write(what)
 145 
 146 wiki = "/org/org.linuxwiki/data/text/"	
 147 o = xml2wiki("DocBook-Demystification-HOWTO.xml", wiki+"DocBookDemystificationHowTo")
 148 o.parse()