From 2a7384a10876c5ea81ef4bce3d3968b5b9074a61 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Tue, 8 Nov 2016 22:12:27 -0500
Subject: [PATCH] tag-mp3s.py: clean up song titles and album names

---
 tag-mp3s.py | 60 ++++++++++++++++++++++++++++++++++++++++-------------
 1 file changed, 46 insertions(+), 14 deletions(-)

diff --git a/tag-mp3s.py b/tag-mp3s.py
index 9dee0cb..6c631c8 100755
--- a/tag-mp3s.py
+++ b/tag-mp3s.py
@@ -6,6 +6,7 @@ import mutagen.id3
 import mutagen.mp3
 import os
 import sys
+import re
 
 min_mp3s_for_album = 4
 dry_run = False
@@ -22,16 +23,21 @@ def walk_tree(base_dir, visit_function):
 def process_dir(path):
     mp3s = glob.glob("%s/*.mp3" % path)
     jpgs = glob.glob("%s/*.jpg" % path)
+    artwork_path = choose_artwork(mp3s, jpgs)
+    for mp3 in mp3s:
+        process_mp3(mp3, artwork_path)
+
+def choose_artwork(mp3s, jpgs):
     if len(mp3s) < min_mp3s_for_album:
-        return
+        return None
     if len(jpgs) == 0:
         sys.stderr.write("No album artwork found in %s\n" % path)
-        return
+        return None
     if len(jpgs) > 1:
         sys.stderr.write("Multiple album artwork files found in %s\n" % path)
-        return
-    for mp3 in mp3s:
-        add_artwork(mp3, jpgs[0])
+        return None
+    return jpgs[0]
+
 
 def read_file(fname):
     f = open(fname, "rb")
@@ -39,17 +45,39 @@ def read_file(fname):
     f.close()
     return data
 
-def add_artwork(mp3_path, jpg_path):
+def clean_title(title):
+    new_title = re.sub(r'[\[\(]((explicit\s+)?album\s+version|explicit|clean)[\]\)]', '', title, flags = re.IGNORECASE)
+    new_title = re.sub(r'\s\s+', ' ', new_title)
+    return new_title.strip()
+
+def clean_tag(mp3, tagname):
+    if tagname in mp3:
+        tag = mp3[tagname]
+        if tag is not None:
+            orig_text = tag.text[0]
+            new_text = clean_title(orig_text)
+            if new_text != orig_text:
+                sys.stdout.write("Changing %s from \"%s\" to \"%s\"\n" % (tagname, orig_text, new_text))
+                tag.text = [new_text]
+                return True
+    return False
+
+def process_mp3(mp3_path, artwork_path):
     mp3 = mutagen.mp3.MP3(mp3_path)
     current_tags = mp3.tags
-    apic_tags = [tag for tag in current_tags if tag == "APIC" or tag.startswith("APIC:")]
-    if len(apic_tags) >= 1:
-        return
-    print("Adding APIC tag to %s..." % mp3_path)
-    jpg_data = read_file(jpg_path)
-    apic = mutagen.id3.APIC(0, "image/jpeg", 0, "", jpg_data)
-    mp3.tags.add(apic)
-    if not dry_run:
+    changed = False
+    changed = clean_tag(mp3, "TIT2") or changed
+    changed = clean_tag(mp3, "TALB") or changed
+    if artwork_path is not None:
+        apic_tags = [tag for tag in current_tags if tag == "APIC" or tag.startswith("APIC:")]
+        if len(apic_tags) < 1:
+            print("Adding APIC tag to %s..." % mp3_path)
+            jpg_data = read_file(artwork_path)
+            apic = mutagen.id3.APIC(0, "image/jpeg", 0, "", jpg_data)
+            mp3.tags.add(apic)
+            changed = True
+    if changed and not dry_run:
+        sys.stdout.write("Writing %s\n" % mp3_path)
         mp3.save()
 
 def main(argv):
@@ -59,7 +87,11 @@ def main(argv):
         if opt == '-d':
             base_dir = val
         elif opt == '-n':
+            global dry_run
             dry_run = True
+    if len(args) > 0:
+        sys.stderr.write("Usage: %s\n" % (sys.argv[0]))
+        return
     walk_tree(base_dir, process_dir)
 
 if __name__ == "__main__":