BUG: Cope with different ISO date length (#1999)

Ensure compatibility with all optional date field formats as specified in the PDF  1.7 specification.

closes #1972
closes #1996

Author: pubpub-zz

pypdf/_reader.py

@@ -59,6 +59,7 @@
     deprecation_no_replacement,
     deprecation_with_replacement,
     logger_warning,
+    parse_iso8824_date,
     read_non_whitespace,
     read_previous_line,
     read_until_whitespace,
@@ -240,12 +241,7 @@ def producer_raw(self) -> Optional[str]:
     @property
     def creation_date(self) -> Optional[datetime]:
         """Read-only property accessing the document's creation date."""
-        text = self._get_text(DI.CREATION_DATE)
-        if text is None:
-            return None
-        return datetime.strptime(
-            text.replace("Z", "+").replace("'", ""), "D:%Y%m%d%H%M%S%z"
-        )
+        return parse_iso8824_date(self._get_text(DI.CREATION_DATE))
 
     @property
     def creation_date_raw(self) -> Optional[str]:
@@ -264,12 +260,7 @@ def modification_date(self) -> Optional[datetime]:
 
         The date and time the document was most recently modified.
         """
-        text = self._get_text(DI.MOD_DATE)
-        if text is None:
-            return None
-        return datetime.strptime(
-            text.replace("Z", "+").replace("'", ""), "D:%Y%m%d%H%M%S%z"
-        )
+        return parse_iso8824_date(self._get_text(DI.MOD_DATE))
 
     @property
     def modification_date_raw(self) -> Optional[str]:

pypdf/_utils.py

@@ -33,6 +33,7 @@
 import logging
 import warnings
 from dataclasses import dataclass
+from datetime import datetime, timezone
 from io import DEFAULT_BUFFER_SIZE, BytesIO
 from os import SEEK_CUR
 from typing import (
@@ -76,6 +77,38 @@
 DEPR_MSG_HAPPENED = "{} is deprecated and was removed in pypdf {}. Use {} instead."
 
 
+def parse_iso8824_date(text: Optional[str]) -> Optional[datetime]:
+    orgtext = text
+    if text is None:
+        return None
+    if text[0].isdigit():
+        text = "D:" + text
+    if text.endswith(("Z", "z")):
+        text += "0000"
+    text = text.replace("z", "+").replace("Z", "+").replace("'", "")
+    i = max(text.find("+"), text.find("-"))
+    if i > 0 and i != len(text) - 5:
+        text += "00"
+    for f in (
+        "D:%Y",
+        "D:%Y%m",
+        "D:%Y%m%d",
+        "D:%Y%m%d%H",
+        "D:%Y%m%d%H%M",
+        "D:%Y%m%d%H%M%S",
+        "D:%Y%m%d%H%M%S%z",
+    ):
+        try:
+            d = datetime.strptime(text, f)  # noqa: DTZ007
+        except ValueError:
+            continue
+        else:
+            if text[-5:] == "+0000":
+                d = d.replace(tzinfo=timezone.utc)
+            return d
+    raise ValueError(f"Can not convert date: {orgtext}")
+
+
 def _get_max_pdf_version_header(header1: bytes, header2: bytes) -> bytes:
     versions = (
         b"%PDF-1.3",

tests/test_utils.py

@@ -15,6 +15,7 @@
     deprecation_no_replacement,
     mark_location,
     matrix_multiply,
+    parse_iso8824_date,
     read_block_backwards,
     read_previous_line,
     read_until_regex,
@@ -337,3 +338,33 @@ def test_file_class():
     f = File(name="image.png", data=b"")
     assert str(f) == "File(name=image.png, data: 0 Byte)"
     assert repr(f) == "File(name=image.png, data: 0 Byte, hash: 0)"
+
+
+@pytest.mark.parametrize(
+    ("text", "expected"),
+    [
+        ("D:20210318000756", "2021-03-18T00:07:56"),
+        ("20210318000756", "2021-03-18T00:07:56"),
+        ("D:2021", "2021-01-01T00:00:00"),
+        ("D:202103", "2021-03-01T00:00:00"),
+        ("D:20210304", "2021-03-04T00:00:00"),
+        ("D:2021030402", "2021-03-04T02:00:00"),
+        ("D:20210408054711", "2021-04-08T05:47:11"),
+        ("D:20210408054711Z", "2021-04-08T05:47:11+00:00"),
+        ("D:20210408054711Z00", "2021-04-08T05:47:11+00:00"),
+        ("D:20210408054711Z0000", "2021-04-08T05:47:11+00:00"),
+        ("D:20210408075331+02'00'", "2021-04-08T07:53:31+02:00"),
+        ("D:20210408075331-03'00'", "2021-04-08T07:53:31-03:00"),
+    ],
+)
+def test_parse_datetime(text, expected):
+    date = parse_iso8824_date(text)
+    date_str = (date.isoformat() + date.strftime("%z"))[: len(expected)]
+    assert date_str == expected
+
+
+def test_parse_datetime_err():
+    with pytest.raises(ValueError) as ex:
+        parse_iso8824_date("D:20210408T054711Z")
+    assert ex.value.args[0] == "Can not convert date: D:20210408T054711Z"
+    assert parse_iso8824_date("D:20210408054711").tzinfo is None