From e42deb6d0086a57e88b65e6c6d7b087d807e78bb Mon Sep 17 00:00:00 2001
From: Antoine Lambert <anlambert@softwareheritage.org>
Date: Fri, 8 Nov 2024 12:46:40 +0100
Subject: [PATCH] bibtex: Robustify code extracting year and month from date
Previous implementation could lead to errors when an invalid date
is present in source codemeta.
---
swh/indexer/bibtex.py | 13 +++++++++----
swh/indexer/tests/test_bibtex.py | 23 +++++++++++++++++++++++
2 files changed, 32 insertions(+), 4 deletions(-)
diff --git a/swh/indexer/bibtex.py b/swh/indexer/bibtex.py
index 446ce4d..891cd11 100644
--- a/swh/indexer/bibtex.py
+++ b/swh/indexer/bibtex.py
@@ -11,6 +11,7 @@ import sys
from typing import Any, Dict, List, Optional
import uuid
+import iso8601
from pybtex.database import Entry, Person
from pybtex.database.output.bibtex import Writer
from pybtex.plugin import register_plugin
@@ -130,10 +131,14 @@ def codemeta_to_bibtex(
fields["date"] = date
break
if "date" in fields:
- (fields["year"], month_number, _) = fields["date"].split("-")
- fields["month"] = (
- f"{MACRO_PREFIX}:{calendar.month_abbr[int(month_number)].lower()}"
- )
+ try:
+ parsed_date = iso8601.parse_date(fields["date"])
+ fields["year"] = str(parsed_date.year)
+ fields["month"] = (
+ f"{MACRO_PREFIX}:{calendar.month_abbr[parsed_date.month].lower()}"
+ )
+ except iso8601.ParseError:
+ pass
# identifier, doi, hal_id
entry_key = None
diff --git a/swh/indexer/tests/test_bibtex.py b/swh/indexer/tests/test_bibtex.py
index fa6fbf1..a820861 100644
--- a/swh/indexer/tests/test_bibtex.py
+++ b/swh/indexer/tests/test_bibtex.py
@@ -276,6 +276,29 @@ def test_affiliation():
)
+def test_invalid_date():
+ assert codemeta_to_bibtex(
+ {
+ "@context": "https://doi.org/10.5063/schema/codemeta-2.0",
+ "author": {"name": "Jane Doe"},
+ "name": "Example Software",
+ "url": "http://example.org/",
+ "datePublished": "TBD",
+ "license": "https://spdx.org/licenses/Apache-2.0",
+ }
+ ) == textwrap.dedent(
+ """\
+ @software{REPLACEME,
+ author = "Doe, Jane",
+ license = "Apache-2.0",
+ date = "TBD",
+ title = "Example Software",
+ url = "http://example.org/"
+ }
+ """
+ )
+
+
def test_cff_empty():
assert cff_to_bibtex("") == textwrap.dedent(
"""\
--
GitLab