bellingcat/auto-archiver
1
0
Fork 0
mirror of https://github.com/bellingcat/auto-archiver.git synced 2026-06-08 03:18:28 +03:00
Code Issues Packages Projects Releases Wiki Activity

Update generic_extractor.py for general/ youtube extraction.

Browse Source
This commit is contained in:
erinhmclark
2025-03-18 21:10:06 +00:00
parent 7d972ee9b8
commit a577228465
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/auto_archiver/modules/generic_extractor/generic_extractor.py
View File

@@ -203,7 +203,7 @@ class GenericExtractor(Extractor):
if not result.get("url"):
result.set_url(url)
if "description" in video_data and not result.get_content():
if "description" in video_data and not result.get("content"):
result.set_content(video_data["description"])
# extract comments if enabled
if self.comments:
@@ -220,10 +220,13 @@ class GenericExtractor(Extractor):
)
# then add the common metadata
if timestamp := video_data.pop("timestamp", None) and not result.get("timestamp"):
timestamp = video_data.pop("timestamp", None)
if timestamp and not result.get("timestamp"):
timestamp = datetime.datetime.fromtimestamp(timestamp, tz=datetime.timezone.utc).isoformat()
result.set_timestamp(timestamp)
if upload_date := video_data.pop("upload_date", None) and not result.get("upload_date"):
upload_date = video_data.pop("upload_date", None)
if upload_date and not result.get("upload_date"):
upload_date = get_datetime_from_str(upload_date, "%Y%m%d").replace(tzinfo=datetime.timezone.utc)
result.set("upload_date", upload_date)