Git-Mirrors/brozzler
1
0
Fork 0
mirror of https://github.com/internetarchive/brozzler.git synced 2025-08-08 22:42:32 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

keep it simple for now

Browse source
This commit is contained in:
Barbara Miller 2025-06-30 14:38:21 -07:00
parent 046db4b6cc
commit 7d58a9ae3b
1 changed files with 4 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

23
brozzler/ydl.py
View file

@ -72,13 +72,14 @@ class VideoDataClient:
return cur.fetchone()
if fetchall:
return cur.fetchall()
return None
def get_pg_video_captures(self, site=None, source=None) -> List[str]:
account_id = site.account_id if site.account_id else None
seed = site.metadata.ait_seed_id if site.metadata.ait_seed_id else None
# TODO: generalize, maybe make variable?
containing_page_timestamp_pattern = "2025%"
containing_page_timestamp_pattern = "2025%" # for future pre-dup additions
if source == "youtube":
containing_page_url_pattern = "http://youtube.com/watch%" # yes, video data canonicalization uses "http"
@ -93,34 +94,18 @@ class VideoDataClient:
containing_page_url_pattern,
),
)
elif account_id and seed:
pg_query = (
"SELECT distinct(containing_page_url) from video where account_id = %s and seed = %s and containing_page_timestamp like %s",
(
account_id,
seed,
containing_page_timestamp_pattern,
),
)
elif seed and source:
elif seed and source: # account_id should usually be present
pg_query = (
"SELECT distinct(containing_page_url) from video where seed = %s and containing_page_url like %s",
(seed, containing_page_url_pattern),
)
elif seed:
pg_query = (
"SELECT distinct(containing_page_url) from video where seed = %s and containing_page_timestamp like %s",
(
seed,
containing_page_timestamp_pattern,
),
)
try:
results = self._execute_query(
pg_query, row_factory=psycopg.rows.scalar_row, fetchall=True
)
except Exception as e:
logger.warn("postgres query failed: %s", e)
results = []
return results