[2026-09-27] eagle: personal/plans/music-artist-albumartist-repair.md personal/plans/tagwriter-container.md personal/tech/id3-tag-repair.md personal/tech/music-tags-repair-truenas.md personal/tech/scripts/find_aa_hijack.py personal/tech/scripts/find_va_misuse.py

personal/tech/scripts/find_aa_hijack.py

@@ -1,26 +1,29 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
-"""Детектор класса «albumartist размножен от чужого файла в корне папки».
+"""Detector A1 — albumartist hijacked from a random ROOT file of the folder.
 
-Сессия 6, 2026-09-27. Нашёл 7 испорченных папок из 11 подозреваемых
-(The Prodigy->DJ Feel, Банд Эрос->VIP77, Дискотека Авария, Жуки,
-Потап & Каменских, Король и Шут, Ленинград).
+Rule (Alex): a track's albumartist = its OWN artist, unless the album is a genuine
+compilation. A value taken from one stray file in the root must NOT be propagated.
 
-ПРИЗНАК ПОРЧИ:
-  для папки, где new_albumartist ОДИН на всю папку:
-    он не совпадает с top-артистом папки
-    И не совпадает с именем папки
-    И совпадает с МИНОРИТАРНЫМ артистом папки
-  -> ПОДОЗРЕВАЕМЫЙ
+Finds folders where the folder-level albumartist equals the `artist` of a MINORITY
+file, but is neither the top artist nor the folder name.
 
-ЗАПУСК:  python3 find_aa_hijack.py
-Не меняет файлы. Только печатает.
+Benign hits to review manually (NOT corruption):
+  - eponymous compilation        (ДДТ/2008 L'Echoppe — files are a duet credit)
+  - letter sub-folder of a set   (Сборка Песни для детей/L)
+  - category root                (z Другое, Enigma, Ленинград)
+  - any folder the user explicitly decided (Re Sfaction, S Is Here)
+
+Usage:
+    python3 find_aa_hijack.py [путь_к_APPLY-PLAN.csv]
 """
+import sys
+
 import csv
 import collections
 import os
 
-P = '/Users/admin/Downloads/music-tags/APPLY-PLAN.csv'
+P = sys.argv[1] if len(sys.argv) > 1 else '/Users/admin/Downloads/music-tags/APPLY-PLAN.csv'
 rows = list(csv.DictReader(open(P, encoding='utf-8-sig', newline='')))
 
 
@@ -36,13 +39,14 @@ suspects = []
 for folder, rs in sorted(by_folder.items()):
     aa_vals = collections.Counter(s(r['new_albumartist']) for r in rs if s(r['new_albumartist']))
     if len(aa_vals) != 1:
-        continue  # смешанные папки — отдельный случай
+        continue  # mixed folder — separate case
     aa = next(iter(aa_vals))
     art = collections.Counter(s(r['new_artist']) for r in rs if s(r['new_artist']))
     if not art:
         continue
     top_artist, top_n = art.most_common(1)[0]
     folder_name = os.path.basename(folder)
+    # albumartist matches a MINORITY artist, not the top one, not the folder name
     if aa != top_artist and aa != folder_name and aa in art:
         minority = art[aa]
         if minority < top_n: