54 lines
1.6 KiB
Python
54 lines
1.6 KiB
Python
|
import functools
|
||
|
|
from typing import Any, Dict, Iterable, List, Optional, Tuple
|
||
|
|
|
||
|
|
from pdfminer import settings
|
||
|
|
from pdfminer.pdfparser import PDFSyntaxError
|
||
|
|
from pdfminer.pdftypes import list_value, int_value, dict_value
|
||
|
|
from pdfminer.utils import choplist
|
||
|
|
|
||
|
|
|
||
|
|
class NumberTree:
|
||
|
|
"""A PDF number tree.
|
||
|
|
|
||
|
|
See Section 3.8.6 of the PDF Reference.
|
||
|
|
"""
|
||
|
|
def __init__(self, obj: Any):
|
||
|
|
self._obj = dict_value(obj)
|
||
|
|
self.nums: Optional[Iterable[Any]] = None
|
||
|
|
self.kids: Optional[Iterable[Any]] = None
|
||
|
|
self.limits: Optional[Iterable[Any]] = None
|
||
|
|
|
||
|
|
if 'Nums' in self._obj:
|
||
|
|
self.nums = list_value(self._obj['Nums'])
|
||
|
|
if 'Kids' in self._obj:
|
||
|
|
self.kids = list_value(self._obj['Kids'])
|
||
|
|
if 'Limits' in self._obj:
|
||
|
|
self.limits = list_value(self._obj['Limits'])
|
||
|
|
|
||
|
|
def _parse(self) -> List[Tuple[int, Any]]:
|
||
|
|
l = []
|
||
|
|
if self.nums: # Leaf node
|
||
|
|
for k, v in choplist(2, self.nums):
|
||
|
|
l.append((int_value(k), v))
|
||
|
|
|
||
|
|
if self.kids: # Root or intermediate node
|
||
|
|
for child_ref in self.kids:
|
||
|
|
l += NumberTree(child_ref)._parse()
|
||
|
|
|
||
|
|
return l
|
||
|
|
|
||
|
|
values: List[Tuple[int, Any]] # workaround decorators unsupported by mypy
|
||
|
|
|
||
|
|
@property # type: ignore [no-redef,misc]
|
||
|
|
@functools.lru_cache
|
||
|
|
def values(self) -> List[Tuple[int, Any]]:
|
||
|
|
values = self._parse()
|
||
|
|
|
||
|
|
if settings.STRICT:
|
||
|
|
if not all(a[0] <= b[0] for a, b in zip(values, values[1:])):
|
||
|
|
raise PDFSyntaxError('Number tree elements are out of order')
|
||
|
|
else:
|
||
|
|
values.sort(key=lambda t: t[0])
|
||
|
|
|
||
|
|
return values
|