From 82a55f1dada836797472da4485997fa9d73374b1 Mon Sep 17 00:00:00 2001
From: Chris Pomeroy <chris.pomeroy@hotmail.com>
Date: Sun, 21 Nov 2021 04:35:34 +0000
Subject: [PATCH] Added a function to normalize metadata

---
 aaxConvert.py |   91 ++++++++++++++++++++++++++-------------------
 1 files changed, 52 insertions(+), 39 deletions(-)

diff --git a/aaxConvert.py b/aaxConvert.py
index e5504d9..f83199f 100755
--- a/aaxConvert.py
+++ b/aaxConvert.py
@@ -3,20 +3,19 @@
 import os
 import glob
 import subprocess
-import shlex
 import json
+import re
 import requests
 import unicodedata
+import sys
 from querysubsonic import findalbumbyname
-from time import sleep
-
 
 # arguments
 # activation_key, file name, codec(default to mp3)
 
 parser = argparse.ArgumentParser()
-parser.add_argument("-s","--single", help="Use this option to create a single file. This is false by default", action="store_true")
-parser.add_argument("-d","--dpath", help="Use this to set the destination path. Otherwise I will use the current directory")
+parser.add_argument("-s", "--single", help="Use this option to create a single file. This is false by default", action="store_true")
+parser.add_argument("-d", "--dpath", help="Use this to set the destination path. Otherwise I will use the current directory")
 parser.add_argument("-v", "--verbose", help="Send output to stdout", action="store_true")
 parser.add_argument("filename", help="Filename to convert, or directory to look in")
 
@@ -44,38 +43,50 @@
 
 
 def getmetadata(aaxfile):
-# Returns the metadata from an aax file
-    ret = subprocess.run(["ffprobe", "-v", "info", "-hide_banner", "-show_format", "-show_chapters", "-print_format", "json", os.path.abspath(aaxfile)], capture_output=True)
+    # Returns the metadata from an aax file
+    ret = subprocess.run(["ffprobe", "-v", "info", "-hide_banner", "-show_format", "-show_chapters",
+                          "-print_format", "json", os.path.abspath(aaxfile)], capture_output=True)
+
     mdata = json.loads(ret.stdout)
     aret = ret.stderr.decode().split('\n')[0]
     mdata["checksum"] = aret.split()[-1]
     return mdata
 
+
 def getmetabitrate():
-#Return the bitrate of the media
+    # Return the bitrate of the media
     bit_rate = metadata['format']['bit_rate']
     return bit_rate[:2]
 
+
 def getmetacopyright():
-# Return normalized copyright data
-    copyright = unicodedata.normalize('NFKD', metadata['format']['tags']['copyright']).encode('ascii','ignore')
+    # Return normalized copyright data
+    copyright = unicodedata.normalize('NFKD', metadata['format']['tags']['copyright']).encode('ascii', 'ignore')
     return copyright.decode()
 
+
 def getmetadatatags(key):
-# get specific data
+    # get specific data
     tag = metadata['format']['tags'][key]
     return " ".join(tag.split())
 
 
+def normalize_data(data):
+    # Return a normalized title
+    data = data.replace(" ", "_")
+    pattern = re.compile('\W')
+    return re.sub(pattern, '', data)
+
+
 def reencode(aaxfile, outpath):
-# decrypt and reencode to mp3
+    # decrypt and reencode to mp3
     command = ("ffmpeg -loglevel error {} -activation_bytes {} -i {} -vn -codec:a libmp3lame -ab {}k -map_metadata -1 "
                "-metadata \"title={}\" -metadata 'artist={}' -metadata 'album_artist={}' -metadata \"album={}\" -metadata 'date={}' "
-               "-metadata track=1/1 -metadata 'genre={}' -metadata 'copyright={}' \"{}\" " ).format(stats, act_byte,aaxfile,getmetabitrate(),
-                                                                                           getmetadatatags('title'), getmetadatatags('artist'),
-                                                                                           getmetadatatags('album_artist'), getmetadatatags('album'),
-                                                                                           getmetadatatags('date'), getmetadatatags('genre'),
-                                                                                           getmetacopyright(),outpath)
+               "-metadata track=1/1 -metadata 'genre={}' -metadata 'copyright={}' \"{}\" ").format(stats, act_byte, aaxfile, getmetabitrate(),
+                                                                                                   getmetadatatags('title'), getmetadatatags('artist'),
+                                                                                                   getmetadatatags('album_artist'), getmetadatatags('album'),
+                                                                                                   getmetadatatags('date'), getmetadatatags('genre'),
+                                                                                                   getmetacopyright(), outpath)
     if args.verbose:
         print(command)
         process = subprocess.run(command, shell=True, capture_output=True)
@@ -89,28 +100,29 @@
         return rc
     else:
         process = subprocess.run(command, shell=True)
-    return 
+    return
 
 
 def getchaptercount():
-# Get the number of chapters
+    # Get the number of chapters
     ccount = metadata['chapters']
     return len(ccount)
 
 
 def getchaptermetadata(cid, key):
-# get the Chapter metadata
+    # get the Chapter metadata
     for i in metadata['chapters']:
         if i['id'] == cid:
             return i[key]
 
-def movetochapters(path, outpath, chapter, title, start,end):
-# Creating individual chapters
+
+def movetochapters(path, outpath, chapter, title, start, end):
+    # Creating individual chapters
 
     outfile = "{}/Ch-{}_{}.mp3".format(outpath, chapter, title.replace(' ', '_'))
     command = "ffmpeg -loglevel error {} -i \"{}\" -ss {} -to {} -codec:a copy -metadata 'track={}' \"{}\"".format(stats, path,
-                                                                                                               start, end,
-                                                                                                               chapter, outfile)
+                                                                                                                   start, end,
+                                                                                                                   chapter, outfile)
     if args.verbose:
         print(command)
         process = subprocess.run(command, shell=True, capture_output=True)
@@ -127,17 +139,18 @@
     return
 
 
-def getcoverart(path,outpath):
-# Pull the coverart from the file
+def getcoverart(path, outpath):
+    # Pull the coverart from the file
     command = "ffmpeg -loglevel error -activation_bytes {} -i \"{}\" -an -codec:v copy \"{}/cover.jpg\"".format(act_byte,
-                                                                                                               path, outpath)
+                                                                                                                path, outpath)
     if args.verbose:
         print(command)
-    process = process = subprocess.run(command, shell=True)
+        subprocess.run(command, shell=True)
     return
 
+
 def getcorrectkey():
-#request the key for the checksum
+    # request the key for the checksum
     try:
         r = requests.post('http://prod-faas-gw.darkurthe.net:8080/function/checkkey', metadata['checksum'], verify=False, timeout=None)
         return r.text.strip()
@@ -145,20 +158,21 @@
         raise err
         return None
 
+
 for rfile in glob.glob(args.filename):
     if rfile.find("aax") != -1 and os.path.isfile(rfile):
         metadata = getmetadata(rfile)
         album = getmetadatatags('album')
-        #See if we got it already
-        if (findalbumbyname(album) == False):
-            artist = getmetadatatags('artist')
-            title = getmetadatatags('title')
+        # See if we got it already
+        if not findalbumbyname(album):
+            artist = normalize_data(getmetadatatags('artist'))
+            title = normalize_data(getmetadatatags('title'))
             act_byte = getcorrectkey()
-            if act_byte == None:
+            if act_byte is None:
                 sys.exit("Can't continue with this file {rfile}")
             else:
-                ddir = "%s/%s/%s" % (path, artist.replace(' ', '_'), title.replace(' ', '_'))
-                single_file_path = "/processing/%s.mp3" % (title.replace(' ','_'))
+                ddir = "%s/%s/%s" % (path, artist, title)
+                single_file_path = "/processing/%s.mp3" % (title)
                 if not os.path.exists(ddir):
                     os.makedirs(ddir)
                 print(ddir)
@@ -166,15 +180,14 @@
                 if mode == 'chapter':
                     chapter = 0
                     numchapters = getchaptercount()
-                    while (numchapters > 0 ):
+                    while (numchapters > 0):
                         cstart = getchaptermetadata(chapter, 'start_time')
                         cend = getchaptermetadata(chapter, 'end_time')
                         chapter += 1
                         numchapters -= 1
                         schap = str(chapter).zfill(2)
-                        movetochapters(single_file_path, ddir, schap, title, cstart,cend)
+                        movetochapters(single_file_path, ddir, schap, title, cstart, cend)
                     os.remove(single_file_path)
                 getcoverart(rfile, ddir)
         else:
             print('We have that book already')
-            
\ No newline at end of file

--
Gitblit v1.10.0