Mercurial > hg-public > dreambox_tools / file comparison
comparison: dupecheck.py
dupecheck.py
- changeset 37
- 5be334b71b08
- parent 36
- a1ad6f4728be
equal
deleted
inserted
replaced
| 36:a1ad6f4728be | 37:5be334b71b08 |
|---|---|
| 1 #!/usr/bin/env python | 1 #!/usr/bin/env python |
| 2 # -*- coding: utf-8 -*- | 2 # -*- coding: utf-8 -*- |
| 3 """ | 3 """ |
| 4 Toolkit / executable to scan for duplicate filenames in movie database | 4 Toolkit / executable to scan for duplicate filenames in movie database |
| 5 | 5 More functions: |
| 6 2017-2019 by mdd | 6 * sanitize filenames |
| 7 * statistics | |
| 8 | |
| 9 2017-2020 by mdd | |
| 7 """ | 10 """ |
| 8 | 11 |
| 9 #pylint: disable=line-too-long | 12 #pylint: disable=line-too-long |
| 10 #pylint: disable=invalid-name | 13 #pylint: disable=invalid-name |
| 11 | 14 |
| 69 | 72 |
| 70 # remove parentheses with contents in title | 73 # remove parentheses with contents in title |
| 71 title = RE_PARENTHESES.sub("", title) | 74 title = RE_PARENTHESES.sub("", title) |
| 72 | 75 |
| 73 self.filelist.append([title, filename, root, ext]) | 76 self.filelist.append([title, filename, root, ext]) |
| 74 elif ext in ['.mkv', '.avi', '.mpg', '.mpeg', '.mp4']: | 77 elif ext in ['.mkv', '.avi', '.mpg', '.mpeg', '.mp4', '.ogg', '.mp3', '.iso']: |
| 75 title = filename[:-4].lower() | 78 title = filename[:-4].lower() |
| 76 title = RE_PARENTHESES.sub("", title) | 79 title = RE_PARENTHESES.sub("", title) |
| 77 self.filelist.append([title, filename, root, ext]) | 80 self.filelist.append([title, filename, root, ext]) |
| 78 elif ext in extra: | 81 elif ext in extra: |
| 79 title = filename[:-4].lower() | 82 title = filename[:-4].lower() |
| 166 """ | 169 """ |
| 167 Search for defect filenames and remove illegal characters | 170 Search for defect filenames and remove illegal characters |
| 168 """ | 171 """ |
| 169 import re | 172 import re |
| 170 for item in self.filelist: | 173 for item in self.filelist: |
| 171 if not item[3] in ['.mkv', '.txt']: | 174 if not item[3] in ['.mkv', '.txt', '.nfo']: |
| 172 continue | 175 continue |
| 173 # any non-alphanumeric characters in filename? | 176 # any non-alphanumeric characters in filename? |
| 174 cleanfn = replace_all(item[1], { | 177 cleanfn = replace_all(item[1], { |
| 175 #'ä':'ae', 'Ä':'Ae', | 178 #'ä':'ae', 'Ä':'Ae', |
| 176 #'ö':'oe', 'Ö':'Oe', | 179 #'ö':'oe', 'Ö':'Oe', |
| 177 #'ü':'ue', 'Ü':'Ue', | 180 #'ü':'ue', 'Ü':'Ue', |
| 178 'ß':'ss', | 181 'ß': 'ss', |
| 182 ':': ' -', | |
| 179 }) | 183 }) |
| 180 cleanfn = re.sub(r'[^A-Za-z0-9\.\_\-\(\)\&öäüÖÄÜ\' ]', '-', cleanfn) | 184 cleanfn = re.sub(r'[^A-Za-z0-9\.\_\-\(\)\[\]\{\}\&öäüÖÄÜ\' ]', '-', cleanfn) |
| 185 | |
| 186 # if cleanfn.endswith(".nfo"): | |
| 187 # # force .txt ending of nfo files | |
| 188 # # TODO: later there should nfo files for kodi | |
| 189 # cleanfn = cleanfn[:-4] + ".txt" | |
| 190 | |
| 191 checklist = re.findall(r'\([^\(\)]+\)', cleanfn) | |
| 192 for nonyear in checklist: | |
| 193 if re.match(r'\(\d{4}\)', nonyear): | |
| 194 continue | |
| 195 cleanfn = replace_all(cleanfn, { | |
| 196 nonyear: replace_all(nonyear, {'(':'[', ')':']'}) | |
| 197 }) | |
| 198 #print ("NONYEAR: ", nonyear) | |
| 199 | |
| 200 checklist = re.findall(r'\[\d{4}[^\]]+\]', cleanfn) | |
| 201 for year in checklist: | |
| 202 cleanfn = replace_all(cleanfn, { | |
| 203 year: replace_all(year, { | |
| 204 year[:5]: '(' + year[1:5] + ') [' | |
| 205 }) | |
| 206 }) | |
| 207 # print ("YEAR: ", year) | |
| 208 | |
| 209 | |
| 181 if item[1] == cleanfn: | 210 if item[1] == cleanfn: |
| 182 continue | 211 continue |
| 183 print (item[1]) | 212 print (item[1], " -> ", cleanfn) |
| 184 os.rename( | 213 os.rename( |
| 185 os.path.join(item[2], item[1]), | 214 os.path.join(item[2], item[1]), |
| 186 os.path.join(item[2], cleanfn) | 215 os.path.join(item[2], cleanfn) |
| 187 ) | 216 ) |
| 188 | 217 |
| 309 print("Loading ignore filename prefixes file for dupe checking...") | 338 print("Loading ignore filename prefixes file for dupe checking...") |
| 310 dupe.ignore_fileprefix = [line.rstrip('\n').rstrip('\r') for line in open("dupecheck-ignore.txt", "rb")] | 339 dupe.ignore_fileprefix = [line.rstrip('\n').rstrip('\r') for line in open("dupecheck-ignore.txt", "rb")] |
| 311 | 340 |
| 312 if args.fixnames: | 341 if args.fixnames: |
| 313 for srcstr in args.basedir: | 342 for srcstr in args.basedir: |
| 314 dupe.scandir(srcstr, ['.txt']) | 343 dupe.scandir(srcstr, ['.txt', '.nfo']) |
| 315 if len(dupe.filelist) > 0: | 344 if len(dupe.filelist) > 0: |
| 316 print ("Checking %i file names..." % len(dupe.filelist)) | 345 print ("Checking %i file names..." % len(dupe.filelist)) |
| 317 dupe.fixnames() | 346 dupe.fixnames() |
| 318 dupe.filelist = [] | 347 dupe.filelist = [] |
| 319 sys.exit(0) | 348 sys.exit(0) |
