def _read_xref(self,
context: TransformerContext,
initial_offset: Optional[int] = None) -> None:
"""
This function attempts to read the XREF table, first as plaintext, then as a stream
:param context: the TransformerContext (containing the io source)
:type context: TransformerContext
:param initial_offset: the initial byte offset at which to read (set to None to allow this method to find the XREF)
:type initial_offset: int
"""
doc = context.root_object
src = context.source
tok = context.tokenizer
most_recent_xref = None
exceptions_to_rethrow = []
# attempt to read plaintext XREF
try:
most_recent_xref = PlainTextXREF()
most_recent_xref.parent = doc
most_recent_xref.read(src, tok, initial_offset)
if "XRef" in doc:
doc["XRef"] = doc["XRef"].merge(most_recent_xref)
else:
doc["XRef"] = most_recent_xref
except Exception as ex0:
most_recent_xref = None
exceptions_to_rethrow.append(ex0)
# attempt to read stream XREF
if most_recent_xref is None:
try:
most_recent_xref = StreamXREF()
most_recent_xref.parent = doc
most_recent_xref.read(src, tok, initial_offset)
if "XRef" in doc:
doc["XRef"] = doc["XRef"].merge(most_recent_xref)
else:
doc["XRef"] = most_recent_xref
except Exception as ex0:
raise ex0
most_recent_xref = None
exceptions_to_rethrow.append(ex0)
# unable to read XREF
# re-throw exceptions
if most_recent_xref is None:
for e in exceptions_to_rethrow:
raise e
# handle Prev, Previous
prev = None
if "Prev" in most_recent_xref["Trailer"]:
prev = int(most_recent_xref["Trailer"]["Prev"])
if "Previous" in most_recent_xref["Trailer"]:
prev = int(most_recent_xref["Trailer"]["Previous"])
if prev is not None:
self._read_xref(context, initial_offset=prev)
def insert_page(
self,
page: Page,
index: int = -1) -> "Document": # type: ignore [name-defined]
# build XRef
if "XRef" not in self:
self["XRef"] = PlainTextXREF()
# build Trailer
if "Trailer" not in self["XRef"]:
self["XRef"]["Trailer"] = Dictionary()
self["XRef"][Name("Size")] = Decimal(0)
# build Root
if "Root" not in self["XRef"]["Trailer"]:
self["XRef"]["Trailer"][Name("Root")] = Dictionary()
# build Pages
if "Pages" not in self["XRef"]["Trailer"]["Root"]:
self["XRef"]["Trailer"][Name("Root")][Name("Pages")] = Dictionary()
self["XRef"]["Trailer"][Name("Root")][Name("Pages")][Name(
"Count")] = Decimal(0)
self["XRef"]["Trailer"][Name("Root")][Name("Pages")][Name(
"Kids")] = List()
# update /Kids
kids = self["XRef"]["Trailer"]["Root"]["Pages"]["Kids"]
assert kids is not None
assert isinstance(kids, List)
kids.insert(index, page)
# update /Count
prev_count = self["XRef"]["Trailer"]["Root"]["Pages"]["Count"]
self["XRef"]["Trailer"]["Root"]["Pages"]["Count"] = Decimal(
prev_count + 1)
# return
return self
def insert_page(
self,
page: Page,
index: typing.Optional[int] = None
) -> "Document": # type: ignore [name-defined]
"""
This method appends a page (from another Document) to this Document at a given index
"""
# build XRef
if "XRef" not in self:
self[Name("XRef")] = PlainTextXREF()
self[Name("XRef")].set_parent(self)
# build Trailer
if "Trailer" not in self["XRef"]:
self["XRef"][Name("Trailer")] = Dictionary()
self["XRef"][Name("Size")] = Decimal(0)
self["XRef"]["Trailer"].set_parent(self["XRef"])
# build Root
if "Root" not in self["XRef"]["Trailer"]:
self["XRef"]["Trailer"][Name("Root")] = Dictionary()
self["XRef"]["Trailer"]["Root"].set_parent(self["XRef"]["Trailer"])
# build Pages
if "Pages" not in self["XRef"]["Trailer"]["Root"]:
self["XRef"]["Trailer"][Name("Root")][Name("Pages")] = Dictionary()
self["XRef"]["Trailer"][Name("Root")][Name("Pages")][Name(
"Count")] = Decimal(0)
self["XRef"]["Trailer"][Name("Root")][Name("Pages")][Name(
"Kids")] = List()
self["XRef"]["Trailer"][Name("Root")][Name("Pages")][Name(
"Type")] = Name("Pages")
self["XRef"]["Trailer"]["Root"]["Pages"].set_parent(
self["XRef"]["Trailer"]["Root"])
self["XRef"]["Trailer"]["Root"]["Pages"]["Kids"].set_parent(
self["XRef"]["Trailer"]["Root"]["Pages"])
# update /Kids
kids = self["XRef"]["Trailer"]["Root"]["Pages"]["Kids"]
assert kids is not None
assert isinstance(kids, List)
if index is None:
index = len(kids)
kids.insert(index, page)
# update /Count
prev_count = self["XRef"]["Trailer"]["Root"]["Pages"]["Count"]
self["XRef"]["Trailer"]["Root"]["Pages"][Name("Count")] = Decimal(
prev_count + 1)
# set /Parent
page[Name("Parent")] = self["XRef"]["Trailer"]["Root"]["Pages"]
page.set_parent(kids) # type: ignore [attr-defined]
# return
return self