encoding, sync subtitles.py

oxdbarchive/cron.py

@@ -12,6 +12,10 @@ def extractNew():
   for f in ArchiveFile.select(ArchiveFile.q.extracted == False):
     f.extractAll()
 
+def cleanScrapeitCache():
+  #remove files older than 30 days from scrapeit cache
+  os.system("""find /var/cache/scrapeit/ -type f -ctime +30 -exec rm '{}' \;""")
+
 
 def cleanClipMovieCache():
   cache = os.path.abspath('oxdb/cache/mini/')

oxdbarchive/subtitles.py

@@ -12,14 +12,14 @@ import chardet
 
 img_extension = 'jpg'
 
-def srt2txt(srt, encoding = "latin-1"):
+def srt2txt(srt, encoding = "utf-8"):
   subtitles = srt2dict(srt, encoding)
   txt = ''
   for k in sorted([int(k) for k in subtitles.keys()]):
     txt += "%s\n\n" % subtitles["%s" % k]['text']
   return txt.strip()
   
-def srt2dict(srt, encoding = "latin-1"):
+def srt2dict(srt, encoding = "utf-8"):
   '''convert srt string into a dict in the form
      dict(num = dict(start, stop, text))
   '''
@@ -39,7 +39,7 @@ def srt2dict(srt, encoding = "latin-1"):
           }
   return subdict
 
-def dict2srt(subtitles, encoding = "latin-1"):
+def dict2srt(subtitles, encoding = "utf-8"):
   '''convert dict in the form dict(num = dict(start, stop, text))
      into an srt file
   '''