def read_manifest(self):
if '/manifest' not in self.entries:
raise LitError('Lit file does not have a valid manifest')
raw = self.get_file('/manifest')
self.manifest = {}
self.paths = {self.opf_path: None}
while raw:
slen, raw = ord(raw[0:1]), raw[1:]
if slen == 0:
break
root, raw = raw[:slen].decode('utf8'), raw[slen:]
if not raw:
raise LitError('Truncated manifest')
for state in ['spine', 'not spine', 'css', 'images']:
num_files, raw = int32(raw), raw[4:]
if num_files == 0:
continue
for i in range(num_files):
if len(raw) < 5:
raise LitError('Truncated manifest')
offset, raw = u32(raw), raw[4:]
internal, raw = consume_sized_utf8_string(raw)
original, raw = consume_sized_utf8_string(raw)
# The path should be stored unquoted, but not always
original = urlunquote(original)
# Is this last one UTF-8 or ASCIIZ?
mime_type, raw = consume_sized_utf8_string(raw, zpad=True)
self.manifest[internal] = ManifestItem(
original, internal, mime_type, offset, root, state)
mlist = list(itervalues(self.manifest))
# Remove any common path elements
if len(mlist) > 1:
shared = mlist[0].path
for item in mlist[1:]:
path = item.path
while shared and not path.startswith(shared):
try:
shared = shared[:shared.rindex("/", 0, -2) + 1]
except ValueError:
shared = None
if not shared:
break
if shared:
slen = len(shared)
for item in mlist:
item.path = item.path[slen:]
# Fix any straggling absolute paths
for item in mlist:
if item.path[0] == '/':
item.path = os.path.basename(item.path)
self.paths[item.path] = item
def read_manifest(self):
if '/manifest' not in self.entries:
raise LitError('Lit file does not have a valid manifest')
raw = self.get_file('/manifest')
self.manifest = {}
self.paths = {self.opf_path: None}
while raw:
slen, raw = ord(raw[0:1]), raw[1:]
if slen == 0:
break
root, raw = raw[:slen].decode('utf8'), raw[slen:]
if not raw:
raise LitError('Truncated manifest')
for state in ['spine', 'not spine', 'css', 'images']:
num_files, raw = int32(raw), raw[4:]
if num_files == 0:
continue
for i in range(num_files):
if len(raw) < 5:
raise LitError('Truncated manifest')
offset, raw = u32(raw), raw[4:]
internal, raw = consume_sized_utf8_string(raw)
original, raw = consume_sized_utf8_string(raw)
# The path should be stored unquoted, but not always
original = urlunquote(original)
# Is this last one UTF-8 or ASCIIZ?
mime_type, raw = consume_sized_utf8_string(raw, zpad=True)
self.manifest[internal] = ManifestItem(
original, internal, mime_type, offset, root, state)
mlist = list(itervalues(self.manifest))
# Remove any common path elements
if len(mlist) > 1:
shared = mlist[0].path
for item in mlist[1:]:
path = item.path
while shared and not path.startswith(shared):
try:
shared = shared[:shared.rindex("/", 0, -2) + 1]
except ValueError:
shared = None
if not shared:
break
if shared:
slen = len(shared)
for item in mlist:
item.path = item.path[slen:]
# Fix any straggling absolute paths
for item in mlist:
if item.path[0] == '/':
item.path = os.path.basename(item.path)
self.paths[item.path] = item
def read(self, name):
entry = self._litfile.paths[urlunquote(name)] if name else None
if entry is None:
content = OPF_DECL + self._read_meta()
elif 'spine' in entry.state:
internal = '/'.join(('/data', entry.internal, 'content'))
raw = self._litfile.get_file(internal)
manifest = self._litfile.manifest
atoms = self._litfile.get_atoms(entry)
unbin = UnBinary(raw, name, manifest, HTML_MAP, atoms)
content = HTML_DECL + unbin.unicode_representation
tags = ('personname', 'place', 'city', 'country-region')
pat = r'(?i)</{0,1}st1:(%s)>' % ('|'.join(tags))
content = re.sub(pat, '', content)
content = re.sub(r'<(/{0,1})form>', r'<\1div>', content)
else:
internal = '/'.join(('/data', entry.internal))
content = self._litfile.get_file(internal)
return content
def read(self, name):
entry = self._litfile.paths[urlunquote(name)] if name else None
if entry is None:
content = OPF_DECL + self._read_meta()
elif 'spine' in entry.state:
internal = '/'.join(('/data', entry.internal, 'content'))
raw = self._litfile.get_file(internal)
manifest = self._litfile.manifest
atoms = self._litfile.get_atoms(entry)
unbin = UnBinary(raw, name, manifest, HTML_MAP, atoms)
content = HTML_DECL + unbin.unicode_representation
tags = ('personname', 'place', 'city', 'country-region')
pat = r'(?i)</{0,1}st1:(%s)>'%('|'.join(tags))
content = re.sub(pat, '', content)
content = re.sub(r'<(/{0,1})form>', r'<\1div>', content)
else:
internal = '/'.join(('/data', entry.internal))
content = self._litfile.get_file(internal)
return content
def exists(self, name):
return urlunquote(name) in self._litfile.paths
def exists(self, name):
return urlunquote(name) in self._litfile.paths
