def __detectmimetype(zipfd, odffile):
"""
detects the mime-type of an ODF file
@param zipfd an open zipfile.ZipFile instance
@param odffile this parameter is not used
@return a mime-type as a unicode string
"""
assert(isinstance(zipfd, zipfile.ZipFile))
assert(type(odffile)==type(u"") or 'rb' in repr(odffile) \
or 'BufferedReader' in repr(odffile) or 'BytesIO' in repr(odffile))
try:
mimetype = zipfd.read('mimetype').decode("utf-8")
return mimetype
except:
pass
# Fall-through to next mechanism
manifestpart = zipfd.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
for mentry,mvalue in manifest.items():
if mentry == "/":
assert(type(mvalue['media-type'])==type(u""))
return mvalue['media-type']
# Fall-through to last mechanism
return u'application/vnd.oasis.opendocument.text'
def __detectmimetype(zipfd, odffile):
"""
detects the mime-type of an ODF file
@param zipfd an open zipfile.ZipFile instance
@param odffile this parameter is not used
@return a mime-type as a unicode string
"""
assert (isinstance(zipfd, zipfile.ZipFile))
assert(type(odffile)==type(u"") or 'rb' in repr(odffile) \
or 'BufferedReader' in repr(odffile) or 'BytesIO' in repr(odffile))
try:
mimetype = zipfd.read('mimetype').decode("utf-8")
return mimetype
except:
pass
# Fall-through to next mechanism
manifestpart = zipfd.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
for mentry, mvalue in manifest.items():
if mentry == "/":
assert (type(mvalue['media-type']) == type(u""))
return mvalue['media-type']
# Fall-through to last mechanism
return u'application/vnd.oasis.opendocument.text'
def load(odffile):
from load import LoadParser
from xml.sax import make_parser, handler
z = zipfile.ZipFile(odffile)
mimetype = z.read('mimetype')
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
for xmlfile in ('settings.xml', 'meta.xml', 'content.xml', 'styles.xml'):
if not manifest.has_key(xmlfile):
continue
try:
xmlpart = z.read(xmlfile)
doc._parsing = xmlfile
parser = make_parser()
parser.setFeature(handler.feature_namespaces, 1)
parser.setContentHandler(LoadParser(doc))
parser.setErrorHandler(handler.ErrorHandler())
inpsrc = InputSource()
inpsrc.setByteStream(StringIO(xmlpart))
parser.parse(inpsrc)
del doc._parsing
except KeyError, v:
pass
def load(odffile):
""" Load an ODF file into memory
Returns a reference to the structure
"""
z = zipfile.ZipFile(odffile)
try:
mimetype = z.read('mimetype')
except KeyError: # Added by Kovid to handle malformed odt files
mimetype = 'application/vnd.oasis.opendocument.text'
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
__loadxmlparts(z, manifest, doc, '')
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
# Load subobjects into structure
elif mentry[:7] == "Object " and len(
mentry) < 11 and mentry[-1] == "/":
subdoc = OpenDocument(mvalue['media-type'], add_generator=False)
doc.addObject(subdoc, "/" + mentry[:-1])
__loadxmlparts(z, manifest, subdoc, mentry)
elif mentry[:7] == "Object ":
pass # Don't load subobjects as opaque objects
else:
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)
if mimetype[:39] == 'application/vnd.oasis.opendocument.text':
doc.text = b[0].firstChild
elif mimetype[:43] == 'application/vnd.oasis.opendocument.graphics':
doc.graphics = b[0].firstChild
elif mimetype[:47] == 'application/vnd.oasis.opendocument.presentation':
doc.presentation = b[0].firstChild
elif mimetype[:46] == 'application/vnd.oasis.opendocument.spreadsheet':
doc.spreadsheet = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.chart':
doc.chart = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.image':
doc.image = b[0].firstChild
elif mimetype[:42] == 'application/vnd.oasis.opendocument.formula':
doc.formula = b[0].firstChild
return doc
def load(odffile):
""" Load an ODF file into memory
Returns a reference to the structure
"""
from load import LoadParser
from xml.sax import make_parser, handler
z = zipfile.ZipFile(odffile)
mimetype = z.read("mimetype")
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read("META-INF/manifest.xml")
manifest = manifestlist(manifestpart)
for xmlfile in ("settings.xml", "meta.xml", "content.xml", "styles.xml"):
if not manifest.has_key(xmlfile):
continue
try:
xmlpart = z.read(xmlfile)
doc._parsing = xmlfile
parser = make_parser()
parser.setFeature(handler.feature_namespaces, 1)
parser.setContentHandler(LoadParser(doc))
parser.setErrorHandler(handler.ErrorHandler())
inpsrc = InputSource()
inpsrc.setByteStream(StringIO(xmlpart))
parser.parse(inpsrc)
del doc._parsing
except KeyError, v:
pass
def load(odffile):
"""
Load an ODF file into memory
@param odffile unicode string: name of a file, or as an alternative,
an open readable stream
@return a reference to the structure (an OpenDocument instance)
"""
assert(type(odffile) == type(u"") or 'rb' in repr(odffile)
or 'BufferedReader' in repr(odffile) or 'BytesIO' in repr(odffile))
z = zipfile.ZipFile(odffile)
mimetype = __detectmimetype(z, odffile)
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
__loadxmlparts(z, manifest, doc, u'')
for mentry, mvalue in manifest.items():
if mentry[:9] == u"Pictures/" and len(mentry) > 9:
doc.addPicture(
mvalue['full-path'], mvalue['media-type'], z.read(mentry))
elif mentry == u"Thumbnails/thumbnail.png":
doc.addThumbnail(z.read(mentry))
elif mentry in (u'settings.xml', u'meta.xml', u'content.xml', u'styles.xml'):
pass
# Load subobjects into structure
elif mentry[:7] == u"Object " and len(mentry) < 11 and mentry[-1] == u"/":
subdoc = OpenDocument(mvalue['media-type'], add_generator=False)
doc.addObject(subdoc, u"/" + mentry[:-1])
__loadxmlparts(z, manifest, subdoc, mentry)
elif mentry[:7] == u"Object ":
pass # Don't load subobjects as opaque objects
else:
if mvalue['full-path'][-1] == u'/':
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)
if mimetype[:39] == u'application/vnd.oasis.opendocument.text':
doc.text = b[0].firstChild
elif mimetype[:43] == u'application/vnd.oasis.opendocument.graphics':
doc.graphics = b[0].firstChild
elif mimetype[:47] == u'application/vnd.oasis.opendocument.presentation':
doc.presentation = b[0].firstChild
elif mimetype[:46] == u'application/vnd.oasis.opendocument.spreadsheet':
doc.spreadsheet = b[0].firstChild
elif mimetype[:40] == u'application/vnd.oasis.opendocument.chart':
doc.chart = b[0].firstChild
elif mimetype[:40] == u'application/vnd.oasis.opendocument.image':
doc.image = b[0].firstChild
elif mimetype[:42] == u'application/vnd.oasis.opendocument.formula':
doc.formula = b[0].firstChild
return doc
def load(odffile):
from load import LoadParser
from xml.sax import make_parser, handler
z = zipfile.ZipFile(odffile)
mimetype = z.read('mimetype')
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
for xmlfile in ('settings.xml', 'meta.xml', 'content.xml', 'styles.xml'):
if not manifest.has_key(xmlfile):
continue
try:
xmlpart = z.read(xmlfile)
doc._parsing = xmlfile
parser = make_parser()
parser.setFeature(handler.feature_namespaces, 1)
parser.setContentHandler(LoadParser(doc))
parser.setErrorHandler(handler.ErrorHandler())
inpsrc = InputSource()
inpsrc.setByteStream(StringIO(xmlpart))
parser.parse(inpsrc)
del doc._parsing
except KeyError, v: pass
# FIXME: Add subobjects correctly 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:
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)
if mimetype[:39] == 'application/vnd.oasis.opendocument.text':
doc.text = b[0].firstChild
elif mimetype[:43] == 'application/vnd.oasis.opendocument.graphics':
doc.graphics = b[0].firstChild
elif mimetype[:47] == 'application/vnd.oasis.opendocument.presentation':
doc.presentation = b[0].firstChild
elif mimetype[:46] == 'application/vnd.oasis.opendocument.spreadsheet':
doc.spreadsheet = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.chart':
doc.chart = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.image':
doc.image = b[0].firstChild
elif mimetype[:42] == 'application/vnd.oasis.opendocument.formula':
doc.formula = b[0].firstChild
return doc
def load(odffile):
"""
Load an ODF file into memory
@param odffile unicode string: name of a file, or as an alternative,
an open readable stream
@return a reference to the structure (an OpenDocument instance)
"""
z = zipfile.ZipFile(odffile)
mimetype = __detectmimetype(z, odffile)
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
__loadxmlparts(z, manifest, doc, u'')
for mentry,mvalue in manifest.items():
if mentry[:9] == u"Pictures/" and len(mentry) > 9:
doc.addPicture(mvalue['full-path'], mvalue['media-type'], z.read(mentry))
elif mentry == u"Thumbnails/thumbnail.png":
doc.addThumbnail(z.read(mentry))
elif mentry in (u'settings.xml', u'meta.xml', u'content.xml', u'styles.xml'):
pass
# Load subobjects into structure
elif mentry[:7] == u"Object " and len(mentry) < 11 and mentry[-1] == u"/":
subdoc = OpenDocument(mvalue['media-type'], add_generator=False)
doc.addObject(subdoc, u"/" + mentry[:-1])
__loadxmlparts(z, manifest, subdoc, mentry)
elif mentry[:7] == u"Object ":
pass # Don't load subobjects as opaque objects
else:
if mvalue['full-path'][-1] == u'/':
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)
if mimetype[:39] == u'application/vnd.oasis.opendocument.text':
doc.text = b[0].firstChild
elif mimetype[:43] == u'application/vnd.oasis.opendocument.graphics':
doc.graphics = b[0].firstChild
elif mimetype[:47] == u'application/vnd.oasis.opendocument.presentation':
doc.presentation = b[0].firstChild
elif mimetype[:46] == u'application/vnd.oasis.opendocument.spreadsheet':
doc.spreadsheet = b[0].firstChild
elif mimetype[:40] == u'application/vnd.oasis.opendocument.chart':
doc.chart = b[0].firstChild
elif mimetype[:40] == u'application/vnd.oasis.opendocument.image':
doc.image = b[0].firstChild
elif mimetype[:42] == u'application/vnd.oasis.opendocument.formula':
doc.formula = b[0].firstChild
return doc
def load(odffile):
""" Load an ODF file into memory
Returns a reference to the structure
"""
z = zipfile.ZipFile(odffile)
try:
mimetype = z.read('mimetype')
except KeyError: # Added by Kovid to handle malformed odt files
mimetype = 'application/vnd.oasis.opendocument.text'
doc = OpenDocument(mimetype, add_generator=False)
# Look in the manifest file to see if which of the four files there are
manifestpart = z.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
__loadxmlparts(z, manifest, doc, '')
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
# Load subobjects into structure
elif mentry[:7] == "Object " and len(mentry) < 11 and mentry[-1] == "/":
subdoc = OpenDocument(mvalue['media-type'], add_generator=False)
doc.addObject(subdoc, "/" + mentry[:-1])
__loadxmlparts(z, manifest, subdoc, mentry)
elif mentry[:7] == "Object ":
pass # Don't load subobjects as opaque objects
else:
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)
if mimetype[:39] == 'application/vnd.oasis.opendocument.text':
doc.text = b[0].firstChild
elif mimetype[:43] == 'application/vnd.oasis.opendocument.graphics':
doc.graphics = b[0].firstChild
elif mimetype[:47] == 'application/vnd.oasis.opendocument.presentation':
doc.presentation = b[0].firstChild
elif mimetype[:46] == 'application/vnd.oasis.opendocument.spreadsheet':
doc.spreadsheet = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.chart':
doc.chart = b[0].firstChild
elif mimetype[:40] == 'application/vnd.oasis.opendocument.image':
doc.image = b[0].firstChild
elif mimetype[:42] == 'application/vnd.oasis.opendocument.formula':
doc.formula = b[0].firstChild
return doc
def __detectmimetype(zipfd, odffile):
try:
mimetype = zipfd.read("mimetype")
return mimetype
except:
pass
# Fall-through to next mechanism
manifestpart = zipfd.read("META-INF/manifest.xml")
manifest = manifestlist(manifestpart)
for mentry, mvalue in manifest.items():
if mentry == "/":
return mvalue["media-type"]
# Fall-through to last mechanism
return "application/vnd.oasis.opendocument.text"
def __detectmimetype(zipfd, odffile):
try:
mimetype = zipfd.read('mimetype')
return mimetype
except:
pass
# Fall-through to next mechanism
manifestpart = zipfd.read('META-INF/manifest.xml')
manifest = manifestlist(manifestpart)
for mentry,mvalue in manifest.items():
if mentry == "/":
return mvalue['media-type']
# Fall-through to last mechanism
return 'application/vnd.oasis.opendocument.text'