email date format flexibility

CHANGELOG.md

@@ -1,3 +1,13 @@
+## 0.18.13
+
+### Enhancements
+
+### Features
+
+### Fixes
+
+- **Parse a wider variety of date formats in email headers** The `partition_email` function is now more robust to non-standard date formats, including ISO-8601 dates with "Z" suffixes. This prevents `ValueError` exceptions when partitioning emails with these date formats. 
+
 ## 0.18.12
 
 ### Enhancements

example-docs/eml/test-invalid-date.eml

@@ -0,0 +1,6 @@
+Date: INVALID-DATE-FORMAT
+From: sender@example.com
+To: recipient@example.com
+Subject: Test invalid date format
+
+This is a test-email with an invalid date format.

example-docs/eml/test-iso-8601-date.eml

@@ -0,0 +1,6 @@
+Date: 2025-07-29T12:42:06.000Z
+From: sender@example.com
+To: recipient@example.com
+Subject: Test a Z-suffix date
+
+This is a test-email.

example-docs/eml/test-rfc2822-date.eml

@@ -0,0 +1,6 @@
+Date: Tue, 29 Jul 2025 12:42:06 +0000
+From: sender@example.com
+To: recipient@example.com
+Subject: Test a standard RFC-2822 date
+
+This is a test-email.

scripts/docker-smoke-test.sh

@@ -3,6 +3,7 @@
 # Start the containerized repository and run ingest tests
 
 # shellcheck disable=SC2317  # Shellcheck complains that trap functions are unreachable...
+# shellcheck disable=SC2329  # Functions are invoked indirectly
 
 set -eux -o pipefail
 

test_unstructured/partition/test_email.py

@@ -521,10 +521,29 @@ def it_uses_the_metadata_last_modified_arg_value_when_one_was_provided(self):
         ctx = EmailPartitioningContext(metadata_last_modified=metadata_last_modified)
         assert ctx.metadata_last_modified == metadata_last_modified
 
-    def and_it_uses_the_msg_Date_header_date_when_metadata_last_modified_was_not_provided(self):
+    def and_it_uses_the_msg_Date_header_date_when_metadata_last_modified_was_not_provided(
+        self,
+    ):
         ctx = EmailPartitioningContext(example_doc_path("eml/simple-rfc-822.eml"))
         assert ctx.metadata_last_modified == "2024-10-01T17:34:56+00:00"
 
+    @pytest.mark.parametrize(
+        ("date_format", "expected_date"),
+        [
+            ("test-iso-8601-date.eml", "2025-07-29T12:42:06+00:00"),
+            ("test-rfc2822-date.eml", "2025-07-29T12:42:06+00:00"),
+        ],
+    )
+    def and_it_correctly_parses_various_date_formats_like_the_ones_that_occur_in_the_wild(
+        self, date_format: str, expected_date: str
+    ):
+        ctx = EmailPartitioningContext(example_doc_path(f"eml/{date_format}"))
+        assert ctx.metadata_last_modified == expected_date
+
+    def and_it_returns_none_when_date_header_is_invalid(self):
+        ctx = EmailPartitioningContext(example_doc_path("eml/test-invalid-date.eml"))
+        assert ctx._sent_date is None
+
     def and_it_falls_back_to_filesystem_last_modified_when_no_Date_header_is_present(
         self, get_last_modified_date_: Mock
     ):

test_unstructured_ingest/src/azure.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/google-drive.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/kafka-local.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/local-single-file-basic-chunking.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/local-single-file-chunk-no-orig-elements.sh

@@ -6,6 +6,8 @@
 # option which otherwise has no other coverage.
 # ------------------------------------------------------------------------------------------------
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 # -- Test Parameters: These vary by test file, others are common computed values --

test_unstructured_ingest/src/local-single-file-with-encoding.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/local-single-file-with-pdf-infer-table-structure.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/local-single-file.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/pdf-fast-reprocess.sh

@@ -43,4 +43,21 @@ PYTHONPATH=${PYTHONPATH:-.} "$RUN_SCRIPT" \
   local \
   --output-dir "$OUTPUT_DIR"
 
+# Flatten outputs so paths match fixtures. New behavior for downloads in unstructured-ingest is to create a nested directory structure.
+mkdir -p "$OUTPUT_DIR/azure"
+find "$OUTPUT_DIR/azure" -type f -name '*.json' -path '*/unstructured_*/*' -print0 | while IFS= read -r -d '' f; do
+  mv "$f" "$OUTPUT_DIR/azure/$(basename "$f")"
+done
+find "$OUTPUT_DIR/azure" -type d -name 'unstructured_*' -exec rm -rf {} +
+
+# Normalize record_locator.path to drop unstructured_* in the download path
+python3 - "$OUTPUT_DIR/azure" <<'PY'
+import re, sys, pathlib
+root = pathlib.Path(sys.argv[1])
+for p in root.rglob('*.json'):
+    s = p.read_text()
+    s2 = re.sub(r'(/download/azure)/unstructured_[^/]+/', r'\1/', s)
+    if s2 != s:
+        p.write_text(s2)
+PY
 "$SCRIPT_DIR"/check-diff-expected-output.sh $OUTPUT_FOLDER_NAME

test_unstructured_ingest/src/s3-minio.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")
@@ -33,8 +35,7 @@ scripts/minio-test-helpers/create-and-check-minio.sh
 wait
 
 RUN_SCRIPT=${RUN_SCRIPT:-unstructured-ingest}
-AWS_SECRET_ACCESS_KEY=$secret_key AWS_ACCESS_KEY_ID=$access_key \
-  PYTHONPATH=${PYTHONPATH:-.} "$RUN_SCRIPT" \
+PYTHONPATH=${PYTHONPATH:-.} "$RUN_SCRIPT" \
   s3 \
   --num-processes "$max_processes" \
   --download-dir "$DOWNLOAD_DIR" \
@@ -45,6 +46,8 @@ AWS_SECRET_ACCESS_KEY=$secret_key AWS_ACCESS_KEY_ID=$access_key \
   --verbose \
   --remote-url s3://utic-dev-tech-fixtures/ \
   --endpoint-url http://localhost:9000 \
+  --key "$access_key" \
+  --secret "$secret_key" \
   --work-dir "$WORK_DIR" \
   local \
   --output-dir "$OUTPUT_DIR"

test_unstructured_ingest/src/s3.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

test_unstructured_ingest/src/sharepoint.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
 
+# shellcheck disable=SC2329  # Functions are invoked indirectly
+
 set -e
 
 SRC_PATH=$(dirname "$(realpath "$0")")

unstructured/__version__.py

@@ -1 +1 @@
-__version__ = "0.18.12"  # pragma: no cover
+__version__ = "0.18.13"  # pragma: no cover

unstructured/partition/email.py

@@ -14,6 +14,8 @@
 from email.message import EmailMessage, MIMEPart
 from typing import IO, Any, Final, Iterator, cast
 
+from dateutil import parser
+
 from unstructured.documents.elements import Element, ElementMetadata
 from unstructured.file_utils.model import FileType
 from unstructured.partition.common import UnsupportedFileFormatError
@@ -279,7 +281,11 @@ def _sent_date(self) -> str | None:
         date_str = self.msg.get("Date")
         if not date_str:
             return None
-        sent_date = email.utils.parsedate_to_datetime(date_str)
+        try:
+            sent_date = parser.parse(date_str)
+        except (parser.ParserError, TypeError, ValueError):
+            return None
+
         return sent_date.astimezone(dt.timezone.utc).isoformat(timespec="seconds")
 
     def _validate(self) -> EmailPartitioningContext: