IGN:Update odfpy and make copying in rtf2xml a little nicer

src/calibre/ebooks/rtf2xml/copy.py

@@ -15,7 +15,8 @@
 #                                                                       #
 #                                                                       #
 #########################################################################
-import sys, os
+import sys, os, shutil
+
 class Copy:
     """Copy each changed file to a directory for debugging purposes"""
     __dir = ""
@@ -64,25 +65,7 @@ class Copy:
         of cp. Otherwise, use a safe python method.
         """
         write_file = os.path.join(Copy.__dir,new_file)
-        platform = sys.platform
+        shutil.copyfile(file, write_file)
-        if platform[:5] == 'linux':
+    
-            command = 'cp %(file)s %(write_file)s' % vars()
-            os.system(command)
-        else:
-            read_obj = open(file,'r')
-            write_obj = open(write_file, 'w')
-            line = "dummy"
-            while line:
-                    line = read_obj.read(1000)
-                    write_obj.write(line )
-            read_obj.close()
-            write_obj.close()
     def rename(self, source, dest):
-        read_obj = open(source, 'r')
+        shutil.copyfile(source, dest)
-        write_obj = open(dest, 'w')
-        line = 1
-        while line:
-            line = read_obj.readline()
-            write_obj.write(line)
-        read_obj.close()
-        write_obj.close()

src/odf/grammar.py

@@ -624,7 +624,7 @@ allowed_children = {
 		(TEXTNS,u'object-index'),
 		(TEXTNS,u'p'),
 		(TEXTNS,u'section'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'user-index'),
@@ -1073,7 +1073,7 @@ allowed_children = {
 		(TEXTNS,u'page-sequence'),
 		(TEXTNS,u'section'),
 		(TEXTNS,u'sequence-decls'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'tracked-changes'),
@@ -1529,7 +1529,7 @@ allowed_children = {
 		(TEXTNS,u'object-index'),
 		(TEXTNS,u'p'),
 		(TEXTNS,u'section'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'user-index'),
@@ -1759,7 +1759,7 @@ allowed_children = {
 	(TABLENS,u'table') : (
 		(OFFICENS,u'dde-source'),
 		(OFFICENS,u'forms'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TABLENS,u'scenario'),
 		(TABLENS,u'shapes'),
 		(TABLENS,u'table-column'),
@@ -1807,7 +1807,7 @@ allowed_children = {
 		(TEXTNS,u'object-index'),
 		(TEXTNS,u'p'),
 		(TEXTNS,u'section'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'user-index'),
@@ -1828,7 +1828,7 @@ allowed_children = {
 	),
 	(TABLENS,u'table-header-rows') : (
 		(TABLENS,u'table-row'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 	),
 	(TABLENS,u'table-row') : (
 		(TABLENS,u'covered-table-cell'),
@@ -1842,7 +1842,7 @@ allowed_children = {
 	),
 	(TABLENS,u'table-rows') : (
 		(TABLENS,u'table-row'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 	),
 	(TABLENS,u'table-source') : (
 	),
@@ -2060,6 +2060,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'chapter') : (
 	),
+	(TEXTNS,u'character-count') : (
+	),
 	(TEXTNS,u'conditional-text') : (
 	),
 	(TEXTNS,u'creation-date') : (
@@ -2126,7 +2128,7 @@ allowed_children = {
 		(TEXTNS,u'object-index'),
 		(TEXTNS,u'p'),
 		(TEXTNS,u'section'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'user-index'),
@@ -2254,7 +2256,7 @@ allowed_children = {
 		(TEXTNS,u'sequence'),
 		(TEXTNS,u'sequence-ref'),
 		(TEXTNS,u'sheet-name'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'span'),
 		(TEXTNS,u'subject'),
 		(TEXTNS,u'tab'),
@@ -2294,6 +2296,8 @@ allowed_children = {
 		(TEXTNS,u'illustration-index-entry-template'),
 		(TEXTNS,u'index-title-template'),
 	),
+	(TEXTNS,u'image-count') : (
+	),
 # allowed_children
 	(TEXTNS,u'index-body') : (
 		(DR3DNS,u'scene'),
@@ -2328,7 +2332,7 @@ allowed_children = {
 		(TEXTNS,u'object-index'),
 		(TEXTNS,u'p'),
 		(TEXTNS,u'section'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'table-index'),
 		(TEXTNS,u'table-of-content'),
 		(TEXTNS,u'user-index'),
@@ -2423,7 +2427,7 @@ allowed_children = {
 		(TEXTNS,u'list'),
 		(TEXTNS,u'number'),
 		(TEXTNS,u'p'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 	),
 	(TEXTNS,u'list-level-style-bullet') : (
 		(STYLENS,u'list-level-properties'),
@@ -2635,7 +2639,7 @@ allowed_children = {
 		(TEXTNS,u'sequence'),
 		(TEXTNS,u'sequence-ref'),
 		(TEXTNS,u'sheet-name'),
-                (TEXTNS,u'soft-page-break'),
+		(TEXTNS,u'soft-page-break'),
 		(TEXTNS,u'span'),
 		(TEXTNS,u'subject'),
 		(TEXTNS,u'tab'),
@@ -2661,6 +2665,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'page') : (
 	),
+	(TEXTNS,u'page-count') : (
+	),
 	(TEXTNS,u'page-continuation') : (
 	),
 	(TEXTNS,u'page-number') : (
@@ -2672,6 +2678,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'page-variable-set') : (
 	),
+	(TEXTNS,u'paragraph-count') : (
+	),
 	(TEXTNS,u'placeholder') : (
 	),
 	(TEXTNS,u'print-date') : (
@@ -2686,6 +2694,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'reference-mark-start') : (
 	),
+	(TEXTNS,u'reference-ref') : (
+	),
 	(TEXTNS,u'ruby') : (
 		(TEXTNS,u'ruby-base'),
 		(TEXTNS,u'ruby-text'),
@@ -3035,6 +3045,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'tab') : (
 	),
+	(TEXTNS,u'table-count') : (
+	),
 	(TEXTNS,u'table-formula') : (
 	),
 	(TEXTNS,u'table-index') : (
@@ -3132,6 +3144,8 @@ allowed_children = {
 	),
 	(TEXTNS,u'variable-set') : (
 	),
+	(TEXTNS,u'word-count') : (
+	),
 }
 
 struct_elements = (  # Unused?
@@ -3182,6 +3196,7 @@ allows_text = (
 	(PRESENTATIONNS,u'footer-decl'),
 	(PRESENTATIONNS,u'header-decl'),
 	(SVGNS,u'desc'),
+	(SVGNS,u'title'),
 	(TEXTNS,u'a'),
 	(TEXTNS,u'author-initials'),
 	(TEXTNS,u'author-name'),
@@ -3403,7 +3418,6 @@ required_attributes = {
 	(DRAWNS,u'glue-point'): (
 		(SVGNS,u'y'),
 		(SVGNS,u'x'),
-		(DRAWNS,u'align'),
 		(DRAWNS,u'id'),
 	),
 	(DRAWNS,u'gradient'): (
@@ -4297,35 +4311,47 @@ allowed_attributes = {
 		(ANIMNS,u'value'),
 	),
 	(ANIMNS,u'seq'):(
-		(PRESENTATIONNS,u'node-type'),
-		(SMILNS,u'decelerate'),
-		(SMILNS,u'begin'),
-		(SMILNS,u'end'),
-		(PRESENTATIONNS,u'group-id'),
-		(SMILNS,u'accelerate'),
-		(SMILNS,u'repeatDur'),
-		(SMILNS,u'endsync'),
-		(SMILNS,u'restartDefault'),
-		(PRESENTATIONNS,u'preset-class'),
-		(SMILNS,u'fillDefault'),
-		(PRESENTATIONNS,u'preset-id'),
-		(SMILNS,u'autoReverse'),
-		(PRESENTATIONNS,u'preset-sub-type'),
-		(SMILNS,u'repeatCount'),
-		(SMILNS,u'dur'),
-		(SMILNS,u'fill'),
 		(ANIMNS,u'id'),
-		(SMILNS,u'restart'),
+		(PRESENTATIONNS,u'group-id'),
 		(PRESENTATIONNS,u'master-element'),
+		(PRESENTATIONNS,u'node-type'),
+		(PRESENTATIONNS,u'preset-class'),
+		(PRESENTATIONNS,u'preset-id'),
+		(PRESENTATIONNS,u'preset-sub-type'),
+		(SMILNS,u'accelerate'),
+		(SMILNS,u'autoReverse'),
+		(SMILNS,u'begin'),
+		(SMILNS,u'decelerate'),
+		(SMILNS,u'dur'),
+		(SMILNS,u'end'),
+		(SMILNS,u'endsync'),
+		(SMILNS,u'fill'),
+		(SMILNS,u'fillDefault'),
+		(SMILNS,u'repeatCount'),
+		(SMILNS,u'repeatDur'),
+		(SMILNS,u'restart'),
+		(SMILNS,u'restartDefault'),
 	),
 	(ANIMNS,u'set'):(
+		(ANIMNS,u'sub-item'),
+		(SMILNS,u'accelerate'),
+		(SMILNS,u'accumulate'),
+		(SMILNS,u'autoReverse'),
 		(SMILNS,u'additive'),
 		(SMILNS,u'attributeName'),
-		(SMILNS,u'to'),
+		(SMILNS,u'begin'),
-		(ANIMNS,u'sub-item'),
+		(SMILNS,u'decelerate'),
-		(SMILNS,u'targetElement'),
+		(SMILNS,u'dur'),
-		(SMILNS,u'accumulate'),
+		(SMILNS,u'end'),
 		(SMILNS,u'fill'),
+		(SMILNS,u'fillDefault'),
+		(SMILNS,u'repeatCount'),
+		(SMILNS,u'repeatDur'),
+		(SMILNS,u'restart'),
+		(SMILNS,u'restartDefault'),
+		(SMILNS,u'targetElement'),
+		(SMILNS,u'to'),
+
 	),
 	(ANIMNS,u'transitionFilter'):(
 		(SMILNS,u'direction'),
@@ -5789,6 +5815,8 @@ allowed_attributes = {
 		(MANIFESTNS,'salt'),
 		(MANIFESTNS,'iteration-count'),
 	),
+	(MANIFESTNS,u'manifest'):(
+	),
 # allowed_attributes
 	(METANS,u'auto-reload'):(
 		(METANS,u'delay'),
@@ -6096,7 +6124,7 @@ allowed_attributes = {
 		(CHARTNS,u'gap-width'),
 		(CHARTNS,u'interpolation'),
 		(CHARTNS,u'interval-major'),
-		(CHARTNS,u'interval-minor'),
+		(CHARTNS,u'interval-minor-divisor'),
 		(CHARTNS,u'japanese-candle-stick'),
 		(CHARTNS,u'label-arrangement'),
 		(CHARTNS,u'lines'),
@@ -6117,6 +6145,7 @@ allowed_attributes = {
 		(CHARTNS,u'spline-resolution'),
 		(CHARTNS,u'stacked'),
 		(CHARTNS,u'symbol-height'),
+		(CHARTNS,u'symbol-name'),
 		(CHARTNS,u'symbol-type'),
 		(CHARTNS,u'symbol-width'),
 		(CHARTNS,u'text-overlap'),
@@ -6236,11 +6265,9 @@ allowed_attributes = {
 	),
 	(STYLENS,u'footer'):(
 		(STYLENS,u'display'),
-		(STYLENS,u'dynamic-spacing'),
 	),
 	(STYLENS,u'footer-left'):(
 		(STYLENS,u'display'),
-		(STYLENS,u'dynamic-spacing'),
 	),
 	(STYLENS,u'footer-style'):(
 	),
@@ -6437,7 +6464,6 @@ allowed_attributes = {
 	),
 	(STYLENS,u'header'):(
 		(STYLENS,u'display'),
-		(STYLENS,u'dynamic-spacing'),
 	),
 	(STYLENS,u'header-footer-properties'): (
 		(FONS,u'background-color'),
@@ -6468,7 +6494,6 @@ allowed_attributes = {
 	),
 	(STYLENS,u'header-left'):(
 		(STYLENS,u'display'),
-		(STYLENS,u'dynamic-spacing'),
 	),
 	(STYLENS,u'header-style'):(
 	),
@@ -6480,6 +6505,7 @@ allowed_attributes = {
 		(STYLENS,u'font-name'),
 		(STYLENS,u'vertical-pos'),
 		(STYLENS,u'vertical-rel'),
+		(SVGNS,u'y'),
 		(TEXTNS,u'min-label-distance'),
 		(TEXTNS,u'min-label-width'),
 		(TEXTNS,u'space-before'),
@@ -6534,12 +6560,10 @@ allowed_attributes = {
 		(STYLENS,u'layout-grid-print'),
 		(STYLENS,u'layout-grid-ruby-below'),
 		(STYLENS,u'layout-grid-ruby-height'),
-		(STYLENS,u'name'),
 		(STYLENS,u'num-format'),
 		(STYLENS,u'num-letter-sync'),
 		(STYLENS,u'num-prefix'),
 		(STYLENS,u'num-suffix'),
-		(STYLENS,u'page-usage'),
 		(STYLENS,u'paper-tray-name'),
 		(STYLENS,u'print'),
 		(STYLENS,u'print-orientation'),
@@ -7156,7 +7180,6 @@ allowed_attributes = {
 		(TABLENS,u'cell-range-address'),
 	),
 	(TABLENS,u'null-date'):(
-		(TABLENS,u'date-value-type'),
 		(TABLENS,u'value-type'),
 	),
 	(TABLENS,u'odd-columns'):(

src/odf/manifest.py

@@ -39,15 +39,3 @@ def Algorithm(**args):
 def KeyDerivation(**args):
     return Element(qname = (MANIFESTNS,'key-derivation'), **args)
 
-
-
-if __name__ == "__main__":
-    import cStringIO
-    xml=cStringIO.StringIO()
-    m = Manifest()
-    f = FileEntry(mediatype="text/xml", fullpath="content.xml")
-    m.addElement(f)
-
-    m.toXml(0,xml)
-    print xml.getvalue()
-

src/odf/namespaces.py

@@ -17,7 +17,7 @@
 #
 # Contributor(s):
 #
-TOOLSVERSION = u"ODFPY/0.8.1dev"
+TOOLSVERSION = u"ODFPY/0.8.2dev"
 
 ANIMNS         = u"urn:oasis:names:tc:opendocument:xmlns:animation:1.0"
 DBNS           = u"urn:oasis:names:tc:opendocument:xmlns:database:1.0"
@@ -49,7 +49,7 @@ TABLENS        = u"urn:oasis:names:tc:opendocument:xmlns:table:1.0"
 TEXTNS         = u"urn:oasis:names:tc:opendocument:xmlns:text:1.0"
 XFORMSNS       = u"http://www.w3.org/2002/xforms"
 XLINKNS        = u"http://www.w3.org/1999/xlink"
-XMLNS          = "http://www.w3.org/XML/1998/namespace"
+XMLNS          = u"http://www.w3.org/XML/1998/namespace"
 
 
 nsdict = {

src/odf/opendocument.py

@@ -64,6 +64,12 @@ odmimetypes = {
  'application/vnd.oasis.opendocument.text-web':              '.oth',
 }
 
+class OpaqueObject:
+    def __init__(self, filename, mediatype, content=None):
+       self.mediatype = mediatype
+       self.filename = filename
+       self.content = content
+
 class OpenDocument:
     """ A class to hold the content of an OpenDocument document
         Use the xml method to write the XML
@@ -76,6 +82,7 @@ class OpenDocument:
     def __init__(self, mimetype, add_generator=True):
         self.mimetype = mimetype
         self.childobjects = []
+        self._extra = []
         self.folder = "" # Always empty for toplevel documents
         self.topnode = Document(mimetype=self.mimetype)
         self.topnode.ownerDocument = self
@@ -303,12 +310,15 @@ class OpenDocument:
         else:
             self.thumbnail = filecontent
 
-    def addObject(self, document):
+    def addObject(self, document, objectname=None):
         """ Add an object. The object must be an OpenDocument class
             The return value will be the folder in the zipfile the object is stored in
         """
         self.childobjects.append(document)
-        document.folder = "%s/Object %d" % (self.folder, len(self.childobjects))
+        if objectname is None:
+            document.folder = "%s/Object %d" % (self.folder, len(self.childobjects))
+        else:
+            document.folder = objectname
         return ".%s" % document.folder
 
     def _savePictures(self, object, folder):
@@ -348,7 +358,7 @@ class OpenDocument:
         else:
             if addsuffix:
                 outputfile = outputfile + odmimetypes.get(self.mimetype,'.xxx')
-            outputfp = zipfile.ZipFile(outputfile,"w")
+            outputfp = zipfile.ZipFile(outputfile, "w")
         self._zipwrite(outputfp)
         outputfp.close()
 
@@ -382,6 +392,14 @@ class OpenDocument:
             zi.external_attr = UNIXPERMS
             self._z.writestr(zi, self.thumbnail)
 
+        # Write any extra files
+        for op in self._extra:
+            self.manifest.addElement(manifest.FileEntry(fullpath=op.filename, mediatype=op.mediatype))
+            zi = zipfile.ZipInfo(op.filename.encode('utf-8'), self._now)
+            zi.compress_type = zipfile.ZIP_DEFLATED
+            zi.external_attr = UNIXPERMS
+            if op.content is not None:
+                self._z.writestr(zi, op.content)
         # Write manifest
         zi = zipfile.ZipInfo("META-INF/manifest.xml", self._now)
         zi.compress_type = zipfile.ZIP_DEFLATED
@@ -528,15 +546,20 @@ def load(odffile):
             parser.parse(inpsrc)
             del doc._parsing
         except KeyError, v: pass
-    # Add the thumbnail here
+    # FIXME: Add subobjects correctly here
-    # Add the images here
     for mentry,mvalue in manifest.items():
         if mentry[:9] == "Pictures/" and len(mentry) > 9:
             doc.addPicture(mvalue['full-path'], mvalue['media-type'], z.read(mentry))
         elif mentry == "Thumbnails/thumbnail.png":
             doc.addThumbnail(z.read(mentry))
+        elif mentry in ('settings.xml', 'meta.xml', 'content.xml', 'styles.xml'):
+            pass
         else:
-            pass  # Add the SUN junk here to the struct somewhere
+            if mvalue['full-path'][-1] == '/':
+                doc._extra.append(OpaqueObject(mvalue['full-path'], mvalue['media-type'], None))
+            else:
+                doc._extra.append(OpaqueObject(mvalue['full-path'], mvalue['media-type'], z.read(mentry)))
+            # Add the SUN junk here to the struct somewhere
             # It is cached data, so it can be out-of-date
     z.close()
     b = doc.getElementsByType(Body)

src/odf/userfield.py

@@ -231,6 +231,19 @@ class ODFContentParser(xml.sax.saxutils.XMLGenerator):
         self._callback_func = callback_func
         xml.sax.saxutils.XMLGenerator.__init__(self, out, encoding)
 
+    def _qname(self, name):
+        """Builds a qualified name from a (ns_url, localname) pair"""
+        if name[0]:
+            if name[0] == u'http://www.w3.org/XML/1998/namespace':
+                return u'xml' + ":" + name[1]
+            # The name is in a non-empty namespace
+            prefix = self._current_context[name[0]]
+            if prefix:
+                # If it is not the default namespace, prepend the prefix
+                return prefix + ":" + name[1]
+        # Return the unqualified name
+        return name[1]
+
     def startElementNS(self, name, qname, attrs):
         if name == (TEXTNS, u'user-field-decl'):
             field_name = attrs.get((TEXTNS, u'name'))