html2odt

   1 #!/usr/bin/python
   2 """Convert html files to ODF.
   3
   4 html2odt source.html destination.odt
   5 """
   6
   7 from __future__ import with_statement
   8 import sys, os, subprocess, time
   9
  10 import uno
  11 from com.sun.star.beans import PropertyValue
  12 from com.sun.star.connection import NoConnectException
  13
  14 def _inspect_obj(o):
  15     print >> sys.stderr, 'inspecting %r' % o
  16     for a in  dir(o):
  17         try:
  18             print >> sys.stderr, "%25s %s" % (a, getattr(o, a))
  19         except Exception, e:
  20             print >> sys.stderr, "%s DOES NOT WORK! (%s)" % (a, e)
  21
  22 def file_url(path):
  23     if path.startswith('file:///'):
  24         return path
  25     return "file://" + os.path.abspath(path)
  26
  27 class Oo(object):
  28     def __init__(self):
  29         """Start up an open office and connect to it."""
  30         accept_string = "socket,host=localhost,port=2002;urp;StarOffice.ComponentContext"
  31
  32         self.ooffice = subprocess.Popen(["ooffice", "-nologo", "-nodefault",
  33                                          "-norestore", "-nofirststartwizard",
  34                                          "-headless", "-invisible", "-nolockcheck",
  35                                          "-accept=%s" % accept_string])
  36
  37         for i in range(10):
  38             time.sleep(0.5)
  39             try:
  40                 local = uno.getComponentContext()
  41                 self.resolver = local.ServiceManager.createInstanceWithContext("com.sun.star.bridge.UnoUrlResolver", local)
  42                 self.context = self.resolver.resolve("uno:" + accept_string)
  43                 self.desktop = self.unobject("com.sun.star.frame.Desktop", self.context)
  44                 return
  45
  46             except NoConnectException:
  47                 print >> sys.stderr, '.',
  48
  49     def unobject(self, klass, context=None):
  50         """get an instance of the class named by <klass>.  It will
  51         probably be a string that looks like
  52         'com.sun.something.SomeThing'."""
  53         if context is None:
  54             return self.context.ServiceManager.createInstance(klass)
  55         return self.context.ServiceManager.createInstanceWithContext(klass, context)
  56
  57     def load(self, src):
  58         """Attempt to load as TextDocument format, but fall back to
  59         WebDocument if that doesn't work.  (WebDocument is called
  60         writer/web in the gui and is less exportable."""
  61         # import property values:
  62         # http://api.openoffice.org/docs/common/ref/com/sun/star/document/MediaDescriptor.html
  63         try:
  64             return self.desktop.loadComponentFromURL(src, "_blank", 0,
  65                                                      (PropertyValue("Hidden" , 0 , True, 0),
  66                                                       PropertyValue("FilterName" , 0 , 'HTML (StarWriter)', 0),
  67                                                       ))
  68         except Exception, e:
  69             print >> sys.stderr, e
  70         #fall back on default WebDocument format
  71         return self.desktop.loadComponentFromURL(src, "_blank", 0,
  72                                                  (PropertyValue("Hidden" , 0 , True, 0),
  73                                                   ))
  74
  75     def embed_graphics(self, doc):
  76         """Reset each graphic object to an embedded copy of itself."""
  77         gp = self.unobject("com.sun.star.graphic.GraphicProvider")
  78         for i in range(doc.GraphicObjects.Count):
  79             g = doc.GraphicObjects.getByIndex(i)
  80             props = (PropertyValue("URL", 0, g.GraphicURL, 0),)
  81             g.setPropertyValue("Graphic", gp.queryGraphic(props))
  82
  83     def convert(self, src, dest):
  84         """Use the connected open office instance to convert the file
  85         named by <src> into odf and save it as <dest>.
  86
  87         The main trick here is forcing the images to be stored inline."""
  88         src = file_url(src)
  89         dest = file_url(dest)
  90         print >> sys.stderr, src
  91         print >> sys.stderr, dest
  92
  93         doc = self.load(src)
  94         self.embed_graphics(doc)
  95         doc.storeToURL(dest, (PropertyValue("FilterName", 0, 'writer8', 0),
  96                               PropertyValue("Overwrite", 0, True, 0 )))
  97         doc.dispose()
  98
  99     def __enter__(self):
 100         return self
 101
 102     def __exit__(self, exc_type, exc_value, traceback):
 103         self.desktop.dispose()
 104         self.context.dispose()
 105         for x in range(10):
 106             os.kill(self.ooffice.pid, 15)
 107             time.sleep(0.25)
 108             if self.ooffice.poll():
 109                 break
 110             print >> sys.stderr, '*',
 111         else:
 112             os.kill(self.ooffice.pid, 9)
 113
 114
 115 def set_env(workdir):
 116     workdir = os.path.abspath(workdir)
 117     os.environ['HOME'] = workdir
 118     os.chdir(workdir)
 119     print >> sys.stderr, os.environ
 120
 121 if __name__ == '__main__':
 122     workdir, src, dest = sys.argv[1:4]
 123     set_env(workdir)
 124
 125     with Oo() as oo:
 126         oo.convert(src, dest)
 127