ROB: Avoid negative seek values when reading partially broken files (#3157)

Closes #3151.

Author: stefan6419846

pypdf/_reader.py

@@ -1003,6 +1003,9 @@ def _get_xref_issues(stream: StreamType, startxref: int) -> int:
             0 means no issue, other values represent specific issues.
 
         """
+        if startxref == 0:
+            return 4
+
         stream.seek(startxref - 1, 0)  # -1 to check character before
         line = stream.read(1)
         if line == b"j":
@@ -1011,7 +1014,7 @@ def _get_xref_issues(stream: StreamType, startxref: int) -> int:
             return 1
         line = stream.read(4)
         if line != b"xref":
-            # not an xref so check if it is an XREF object
+            # not a xref so check if it is an XREF object
             line = b""
             while line in b"0123456789 \t":
                 line = stream.read(1)

tests/test_reader.py

@@ -1787,3 +1787,12 @@ def test_repair_root(caplog):
             'Searching object with "/Catalog" key',
         )
     )
+
+
+@pytest.mark.enable_socket
+def test_issue3151(caplog):
+    """Tests for #3151"""
+    url = "https://github.com/user-attachments/files/18941494/bible.pdf"
+    name = "issue3151.pdf"
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    assert len(reader.pages) == 742