Support microsecond timestamp resolution

The dateTimeStamp XML standard type supports fractional
second notation: https://www.w3.org/TR/xmlschema11-2/#dateTimeStamp

Before this patch, the tool crashed when fractional timestamps were given
in e.g. CreationInfo.

Current implementation is limited to microsecond resolution.
More fine-grained timestamps (nanosecond), will be silently
truncated to microsecond resolution.

Signed-off-by: Zalan Blenessy <zalan.blenessy@volvocars.com>

Author: zbleness

src/spdx_tools/spdx/datetime_conversions.py

@@ -1,15 +1,22 @@
 # SPDX-FileCopyrightText: 2022 spdx contributors
 #
 # SPDX-License-Identifier: Apache-2.0
+import re
 from datetime import datetime, timezone
 
 
 def datetime_from_str(date_str: str) -> datetime:
     if not isinstance(date_str, str):
         raise TypeError(f"Could not convert str to datetime, invalid type: {type(date_str).__name__}")
 
-    date = datetime.strptime(date_str, "%Y-%m-%dT%H:%M:%SZ")  # raises ValueError if format does not match
-    return date
+    if "." not in date_str:
+        return datetime.strptime(date_str, "%Y-%m-%dT%H:%M:%SZ")  # raises ValueError if format does not match
+
+    # Based on the https://www.w3.org/TR/xmlschema11-2/#dateTimeStamp
+    # The secondFrag allows fractional second notation as well.
+    # normalize to micro seconds so that we can use %f with strptime
+    date_str = re.sub(r"\.(\d{1,6})\d*Z$", r".\1Z", date_str)
+    return datetime.strptime(date_str, "%Y-%m-%dT%H:%M:%S.%fZ")  # raises ValueError if format does not match
 
 
 def datetime_to_iso_string(date: datetime) -> str:

tests/spdx/test_datetime_conversions.py

@@ -43,10 +43,22 @@ def test_local_datetime_to_iso_string():
 
 def test_datetime_from_str():
     date_str = "2010-03-04T05:45:11Z"
+    assert datetime_from_str(date_str) == datetime(2010, 3, 4, 5, 45, 11)
 
-    date = datetime_from_str(date_str)
+    date_str = "2010-03-04T05:45:11.0Z"
+    assert datetime_from_str(date_str) == datetime(2010, 3, 4, 5, 45, 11)
 
-    assert date == datetime(2010, 3, 4, 5, 45, 11)
+    # implicit notation
+    date_str = "2010-03-04T05:45:11.1Z"
+    assert datetime_from_str(date_str) == datetime(2010, 3, 4, 5, 45, 11, 100000)
+
+    # explicity notation
+    date_str = "2010-03-04T05:45:11.123456Z"
+    assert datetime_from_str(date_str) == datetime(2010, 3, 4, 5, 45, 11, 123456)
+
+    # truncation of nano seconds
+    date_str = "2010-03-04T05:45:11.1234567Z"
+    assert datetime_from_str(date_str) == datetime(2010, 3, 4, 5, 45, 11, 123456)
 
 
 @pytest.mark.parametrize(

tests/spdx3/writer/tag_value/test_write_document.py

@@ -29,9 +29,7 @@ def test_render_creation_info():
     output_str = io.StringIO()
     write_spdx_document(spdx_document, text_output=output_str)
 
-    assert (
-        output_str.getvalue()
-        == """\
+    assert output_str.getvalue() == """\
 ## SPDX Document
 SPDXID: SPDXRef-FOO
 name: BAR
@@ -42,4 +40,3 @@ def test_render_creation_info():
   data license: CC0-1.0
 elements: 
 """  # noqa: W291 # elements: are printed with a space
-    )