recent_video_capture_exists

Author: Barbara Miller

brozzler/video_data.py

@@ -18,8 +18,9 @@
 
 import os
 from dataclasses import dataclass
-from typing import Any, List, Optional
+from typing import Any, Bool, List, Optional
 
+import datetime
 import structlog
 import urlcanon
 
@@ -83,7 +84,24 @@ def _execute_pg_query(self, query_tuple, fetchall=False) -> Optional[Any]:
             logger.warn("postgres query failed: %s", e)
         return None
 
-    def get_recent_video_capture(self, site=None, containing_page_url=None) -> List:
+    def _timestamp4datetime(timestamp):
+        """split `timestamp` into a tuple of 6 integers.
+
+        :param timestamp: full-length timestamp
+        """
+        timestamp = timestamp[:14]
+        return (
+            int(timestamp[:-10]),
+            int(timestamp[-10:-8]),
+            int(timestamp[-8:-6]),
+            int(timestamp[-6:-4]),
+            int(timestamp[-4:-2]),
+            int(timestamp[-2:]),
+        )
+
+    def recent_video_capture_exists(
+        self, site=None, containing_page_url=None, recent=30
+    ) -> Bool:
         # using ait_account_id as postgres partition id
         partition_id = (
             site["metadata"]["ait_account_id"]
@@ -93,7 +111,7 @@ def get_recent_video_capture(self, site=None, containing_page_url=None) -> List:
         seed_id = (
             site["metadata"]["ait_seed_id"] if site["metadata"]["ait_seed_id"] else None
         )
-        result = None
+        result = False
 
         if partition_id and seed_id and containing_page_url:
             # check for postgres query for most recent record
@@ -105,7 +123,20 @@ def get_recent_video_capture(self, site=None, containing_page_url=None) -> List:
                 result_tuple = self._execute_pg_query(pg_query)
                 if result_tuple:
                     result = result_tuple[0]
-                    logger.info("found most recent video capture record: %s", result)
+                    logger.info("found most recent capture timestamp: %s", result)
+                    capture_timestamp = datetime.datetime(
+                        *self._timestamp4datetime(result)
+                    )
+                    time_diff = (
+                        datetime.datetime.now(datetime.timezone.utc)()
+                        - capture_timestamp
+                    )
+                    if time_diff < datetime.timedelta(recent):
+                        logger.info(
+                            "recent video capture from %s exists",
+                            containing_page_url,
+                        )
+                        result = True
 
             except Exception as e:
                 logger.warn("postgres query failed: %s", e)

brozzler/worker.py

@@ -279,21 +279,6 @@ def thumb_jpeg(self, full_jpeg):
         img.save(out, "jpeg", quality=95)
         return out.getbuffer()
 
-    def _timestamp4datetime(timestamp):
-        """split `timestamp` into a tuple of 6 integers.
-
-        :param timestamp: full-length timestamp
-        """
-        timestamp = timestamp[:14]
-        return (
-            int(timestamp[:-10]),
-            int(timestamp[-10:-8]),
-            int(timestamp[-8:-6]),
-            int(timestamp[-6:-4]),
-            int(timestamp[-4:-2]),
-            int(timestamp[-2:]),
-        )
-
     def should_ytdlp(self, logger, site, page, page_status):
         # called only after we've passed needs_browsing() check
 
@@ -316,25 +301,15 @@ def should_ytdlp(self, logger, site, page, page_status):
         logger.info("checking for recent previous captures of %s", ytdlp_url)
         if "youtube.com/watch" in ytdlp_url:
             try:
-                previous_capture = self._video_data.get_recent_video_capture(
-                    site, ytdlp_url
+                recent_capture_exists = self._video_data.recent_video_capture_exists(
+                    site, ytdlp_url, recent=90
                 )
-                if previous_capture:
-                    capture_timestamp = datetime.datetime(
-                        *self._timestamp4datetime(previous_capture)
-                    )
-                    logger.info("capture_timestamp: %s", capture_timestamp)
-                    time_diff = (
-                        datetime.datetime.now(datetime.timezone.utc)()
-                        - capture_timestamp
+                if recent_capture_exists:
+                    logger.info(
+                        "recent previous capture of %s found, skipping ytdlp",
+                        ytdlp_url,
                     )
-                    # TODO: make variable for timedelta
-                    if time_diff < datetime.timedelta(days=90):
-                        logger.info(
-                            "skipping ytdlp for %s since there's a recent capture",
-                            ytdlp_url,
-                        )
-                        return False
+                    return False
             except Exception as e:
                 logger.warning(
                     "exception querying for previous capture for %s: %s",