date_utils: make date specificity sorting safer

Author: chapmanjacobd

library/createdb/av.py

@@ -191,20 +191,25 @@ def munge_av_tags(args, m) -> dict:
                 full_scan_if_corrupt=args.full_scan_if_corrupt,
                 threads=1,
             )
+        except (TimeoutError, subprocess.TimeoutExpired):
+            log.error(f"FFProbe timed out. {path}")
+            m["error"] = "FFProbe timed out"
         except Exception:
             print(path)
             raise
-
-        if media_check.corruption_threshold_exceeded(
-            args.delete_corrupt, corruption, duration
-        ) and not file_utils.is_file_open(path):
-            threshold_str = (
-                strings.percent(args.delete_corrupt) if 0 < args.delete_corrupt < 1 else (args.delete_corrupt + "s")
-            )
-            log.warning("Deleting %s corruption %.1f%% exceeded threshold %s", path, corruption * 100, threshold_str)
-            file_utils.trash(args, path, detach=False)
-            m["time_deleted"] = consts.APPLICATION_START
-            m["error"] = "Media check failed"
+        else:
+            if media_check.corruption_threshold_exceeded(
+                args.delete_corrupt, corruption, duration
+            ) and not file_utils.is_file_open(path):
+                threshold_str = (
+                    strings.percent(args.delete_corrupt) if 0 < args.delete_corrupt < 1 else (args.delete_corrupt + "s")
+                )
+                log.warning(
+                    "Deleting %s corruption %.1f%% exceeded threshold %s", path, corruption * 100, threshold_str
+                )
+                file_utils.trash(args, path, detach=False)
+                m["time_deleted"] = consts.APPLICATION_START
+                m["error"] = "Media check failed"
 
     tags = format_.pop("tags", None)
     if tags:

library/createdb/tube_backend.py

@@ -1,4 +1,4 @@
-import json, re, sys
+import json, re, subprocess, sys
 from copy import deepcopy
 from pathlib import Path
 from subprocess import CalledProcessError
@@ -594,7 +594,7 @@ def blocklist_check(info, *pargs, incomplete):
                     threads=args.same_file_threads,
                 )
                 info["corruption"] = int(corruption * 100)
-            except (RuntimeError, CalledProcessError):
+            except (RuntimeError, CalledProcessError, TimeoutError, subprocess.TimeoutExpired):
                 info["corruption"] = 50
             if info["corruption"] > 7:
                 media_check_failed = True

library/mediafiles/media_check.py

@@ -213,7 +213,7 @@ def media_check() -> None:
                 corruption = future.result()
                 print(strings.percent(corruption), shlex.quote(path), sep="\t")
             except Exception as excinfo:
-                print(f"Error hashing {path}: {excinfo}")
+                print(f"Error checking {path}: {excinfo}")
                 if args.verbose >= consts.LOG_DEBUG:
                     raise
             else:

library/utils/date_utils.py

@@ -3,7 +3,7 @@
 
 import dateutil.parser
 
-from library.utils import iterables, nums
+from library.utils import consts, iterables, nums
 
 
 def is_tz_aware(dt: datetime.datetime) -> bool:
@@ -37,15 +37,26 @@ def super_parser(date_str, fallback=False):
     return None
 
 
+def date_specificity_sort_key(d):
+    month_flag = bool(getattr(d, "month", None))
+    day_flag = bool(getattr(d, "day", None))
+
+    try:
+        ts = d.timestamp()
+        ts_key = -ts
+    except Exception:  # fallback for invalid year/timestamp
+        ts_key = consts.APPLICATION_START
+
+    return (month_flag, day_flag, ts_key)
+
+
 def specific_date(*dates):
     valid_dates = [super_parser(s) for s in dates if s]
     past_dates = [d for d in valid_dates if d and d < maybe_tz_now(d)]
     if not past_dates:
         return None
 
-    earliest_specific_date = sorted(
-        past_dates, key=lambda d: (bool(d.month), bool(d.day), -d.timestamp()), reverse=True
-    )[0]
+    earliest_specific_date = sorted(past_dates, key=date_specificity_sort_key, reverse=True)[0]
     return nums.to_timestamp(earliest_specific_date)
 
 

library/utils/nums.py

@@ -1,5 +1,7 @@
 import math, re, statistics
 
+from library.utils import consts
+
 
 def percent(value, total):
     if total == 0:
@@ -26,7 +28,10 @@ def percentage_difference(value1, value2):
 
 
 def to_timestamp(dt_object):
-    return int(dt_object.timestamp())
+    try:
+        return int(dt_object.timestamp())
+    except ValueError:
+        return consts.APPLICATION_START
 
 
 def safe_int(s) -> int | None: