metadata script

metadata remove genre: script to remove keywords ie audiobook from genre
rank from list: change pattern and separate author and title, rank is
str now

quiethost/metadata_rank_from_list.py

@@ -70,9 +70,11 @@ def tag_file(path: Path, rank: int, year: str, dry_run: bool):
     else:
         tag = "grouping"
         existing = ""
-    pattern = r"Top 100 SciFi Books #\d+"
+    #pattern = r"Top 100 SciFi Books #\d+(?:\.\d+)?"
-#    pattern = r"Enderverse"
+    pattern = r"Enderverse Publication #\d+(?:\.\d+)?"
-    replace = r"Top 100 SciFi Books"
+    #replace = r"Top 100 SciFi Books"
+    replace = r"Enderverse Publication"
+
     if existing:
         if re.search(pattern, existing):
             new_group = re.sub(
@@ -142,9 +144,9 @@ def main():
         parts = line.split(" - ")
 
         if len(parts) == 4:
-            rank, author, title, year = parts
+            rank, title, author, year = parts
         elif len(parts) == 3:
-            rank, author, title = parts
+            rank, title, author = parts
             year = None
         else:
             raise ValueError(f"Invalid line: {line}")
@@ -186,7 +188,7 @@ def main():
 
         found.add(rank)
 
-        print(f"{GREEN}FOUND{RESET}   #{int(rank):3} {title} in {book_dir}")
+        print(f"{GREEN}FOUND{RESET}   #{rank:3} {title} in {book_dir}")
 
         if args.find_missing:
             continue
@@ -195,7 +197,7 @@ def main():
             if file.suffix.lower() not in AUDIO_EXTENSIONS:
                 continue
 
-            ret=tag_file(file, int(rank), year, args.dry_run)
+            ret=tag_file(file, rank, year, args.dry_run)
             if ret == "skip":
                 break
         #print("next")

quiethost/metadata_remove_genre.sh

@@ -0,0 +1,103 @@
+#!/usr/bin/env bash
+
+LIBRARY="$1"
+MODE="${2:-}"
+REPORT="${3:-/home/lucas/audiobook_genre_cleanup.txt}"
+
+if [[ -z "$LIBRARY" ]]; then
+    echo "Usage: $0 <library_path> [--delete] [report_file]"
+    exit 1
+fi
+
+mapfile -t FILES < <(
+    find "$LIBRARY" -type f \( \
+        -iname "*.mp3" -o \
+        -iname "*.m4b" -o \
+        -iname "*.m4a" \
+    \)
+)
+
+TOTAL=${#FILES[@]}
+PROCESSED=0
+MATCHES=0
+MODIFIED=0
+
+{
+    echo "Audiobook Genre Cleanup Report"
+    echo "Generated: $(date)"
+    echo "Library: $LIBRARY"
+    echo "Total files: $TOTAL"
+    echo
+} > "$REPORT"
+
+for file in "${FILES[@]}"; do
+    ((PROCESSED++))
+
+    if (( PROCESSED % 100 == 0 )); then
+        echo "Progress: $PROCESSED/$TOTAL files processed"
+    fi
+    genre=$(kid3-cli -c "get genre" "$file" 2>/dev/null)
+
+    [[ -z "$genre" ]] && continue
+    cleaned=$(printf '%s\n' "$genre" | sed -E '
+        s/[Aa]udio[[:space:]]*[Bb]ooks?//g
+    
+        # Normalize genre names
+        s/\b(Sci[ -]?Fi|Science Fiction|SF)\b/SciFi/Ig
+        s/\bnonfiction\b/NonFiction/Ig
+    
+        # Normalize separators (&, :, / -> ;)
+        s/[[:space:]]*[,&:\/][[:space:]]*/;/g
+    
+        # Normalize commas
+        s/[[:space:]]*;[[:space:]]*/; /g
+    
+        # Trim leading/trailing punctuation and whitespace
+        s/^[[:space:],;:-]+//
+        s/[[:space:],;-]+$//
+    ')
+#           s/.*//;
+    
+    if [[ "$genre" != "$cleaned" ]]; then
+        ((MATCHES++))
+
+        status="MATCH"
+
+        if [[ "$MODE" == "--delete" ]]; then
+            printf "D "
+            kid3-cli \
+                -c "select-tag 1" \
+                -c "set genre \"$cleaned\"" \
+                -c "select-tag 2" \
+                -c "set genre \"$cleaned\"" \
+                -c save \
+                "$file" >/dev/null
+
+            ((MODIFIED++))
+            status="UPDATED"
+        fi
+
+        printf '[%d/%d] [%s] %s | OLD="%s" | NEW="%s"\n' \
+            "$PROCESSED" "$TOTAL" "$status" "$file" "$genre" "$cleaned"
+
+        printf '[%d/%d] [%s] %s | OLD="%s" | NEW="%s"\n' \
+            "$PROCESSED" "$TOTAL" "$status" "$file" "$genre" "$cleaned" >> "$REPORT"
+    fi
+
+done
+
+{
+    echo
+    echo "========== SUMMARY =========="
+    echo "Files scanned : $TOTAL"
+    echo "Matches found : $MATCHES"
+    echo "Files updated : $MODIFIED"
+} >> "$REPORT"
+
+echo
+echo "========== SUMMARY =========="
+echo "Files scanned : $TOTAL"
+echo "Matches found : $MATCHES"
+echo "Files updated : $MODIFIED"
+echo "Report written to: $REPORT"
+