ROB : fix offset correction in revised PDF

pubpub-zz · pubpub-zz · commit 9daeabd3e7fd · 2022-09-03T16:24:57.000+02:00
fixes #328 problem is observed in pdfs where the xref table in previous versions are not starting at 0 where _zero_xref was changing index which should not
diff --git a/PyPDF2/_reader.py b/PyPDF2/_reader.py
@@ -1347,15 +1347,19 @@ def read(self, stream: StreamType) -> None:
             for gen, xref_entry in self.xref.items():
                 if gen == 65535:
                     continue
-                for id in xref_entry:
+                xref_k = sorted(
+                    xref_entry.keys()
+                )  # must ensure ascendant to prevent damange
+                for id in xref_k:
                     stream.seek(xref_entry[id], 0)
                     try:
                         pid, _pgen = self.read_object_header(stream)
                     except ValueError:
                         break
                     if pid == id - self.xref_index:
-                        self._zero_xref(gen)
-                        break
+                        # fixing index item per item is required for revised PDF.
+                        self.xref[gen][pid] = self.xref[gen][id]
+                        del self.xref[gen][id]
                     # if not, then either it's just plain wrong, or the
                     # non-zero-index is actually correct
             stream.seek(loc, 0)  # return to where it was
@@ -1750,11 +1754,6 @@ def _read_xref_subsections(
                 elif self.strict:
                     raise PdfReadError(f"Unknown xref type: {xref_type}")
 
-    def _zero_xref(self, generation: int) -> None:
-        self.xref[generation] = {
-            k - self.xref_index: v for (k, v) in list(self.xref[generation].items())
-        }
-
     def _pairs(self, array: List[int]) -> Iterable[Tuple[int, int]]:
         i = 0
         while True:
diff --git a/tests/test_reader.py b/tests/test_reader.py
@@ -1136,3 +1136,11 @@ def test_reader(caplog):
     # ...and now no more required
     reader.pages[0].extract_text()
     assert caplog.text == ""
+
+
+def test_zeroing_xref():
+    # iss #328
+    url = "https://github.com/py-pdf/PyPDF2/files/9066120/UTA_OSHA_3115_Fall_Protection_Training_09162021_.pdf"
+    name = "UTA_OSHA.pdf"
+    reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+    len(reader.pages)
