thetvdb_rename: python3 and more robust name matching

Signed-off-by: Olivier Mehani <shtrom@ssji.net>

1 files changed, 35 insertions, 54 deletions

diff --git a/thetvdb_rename.py b/thetvdb_rename.py
index f525391..ff9db9e 100755
--- a/thetvdb_rename.py
+++ b/thetvdb_rename.py
@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 
 # pip install tvnamer fuzzywuzzy
 
@@ -6,6 +6,7 @@ import os, sys
 import logging
 import tvdb_api # use tvnamer
 import re
+import html
 # Need python-levenshtein for faster sequence matching
 from fuzzywuzzy import fuzz
 
@@ -14,6 +15,7 @@ ext="mkv"
 TVNAMER_THETVDB_APIKEY='fb51f9b848ffac9750bada89ecba0225' # https://github.com/dbr/tvnamer/blob/24a9642f261611499a026f95b4fa5496de553e45/tvnamer/main.py#L55
 
 debuglevel=logging.INFO
+debuglevel=logging.DEBUG
 logging.basicConfig(level=debuglevel)
 
 inparen=re.compile('.*\(([^\)]\+)\)')
@@ -40,62 +42,41 @@ except OSError:
     pass
 
 count=0
+eps = {e['episodeName']: e for e in s.values()}
 for f in os.listdir("."):
-    e = None
-    title = None
+    title = f.split("- ")[-1].rstrip(".%s" % ext)
     if f != outdir and f.split(".")[-1] == ext:
-        e = []
-        logging.debug("considering '%s'..." % f)
-        title = f.split("- ")[-1].rstrip(".%s" % ext)
-        logging.debug("considering '%s'..." % title)
+        em = {fuzz.ratio(title, t): t for t in eps.keys()}
+        logging.debug("found %s", em)
+
+        maxt = em[max(em.keys())]
+        maxe = eps[maxt]
+        clean_title = maxe['episodeName']
+
+        logging.debug("found %s", clean_title)
+
+        episodeNumber = maxe['dvdEpisodeNumber']
+        if not episodeNumber:
+            episodeNumber = maxe['airedEpisodeNumber']
+
+        d = "%s/%s - S%02dE%02d - %s.%s" % (outdir,
+                maxe.season.show['seriesName'],
+                season,
+                episodeNumber,
+                clean_title,
+                ext
+                )
+        logging.info("%s -> S%02dE%02d " % (
+                f,
+                season,
+                episodeNumber,
+                ))
+
+        count += 1
         try:
-            e.extend(s.search(title))
-        except tvdb_exceptions.tvdb_episodenotfound:
-            logging.debug("%s not matched" % title)
-        title_noep = title.lower().replace("episode ","")
-        logging.debug("considering '%s'..." % title_noep)
-        try:
-            e.extend(s.search(title_noep))
-        except tvdb_exceptions.tvdb_episodenotfound:
-            logging.debug("%s not matched" % title_noep)
-
-        for ss in reversed(re.split("[-()]", title)):
-            if len(ss.strip()) > 0:
-                logging.debug("considering '%s'..." % ss)
-                try:
-                    e.extend(s.search(ss))
-                except tvdb_exceptions.tvdb_episodenotfound:
-                    logging.debug("%s not matched" % ss)
-                    continue
-
-        logging.debug(e)
-
-        em = {fuzz.ratio(title, em['episodename']): em for em in e}
-        if len(em) < 1:
-            logging.warning("couldn't match %s" % (f))
-        else:
-            e = em[max(em.keys())]
-
-            logging.debug("found %s", e['episodename'])
-
-            d = "%s/%s - S%02dE%02d - %s.%s" % (outdir,
-                    e.season.show['seriesname'],
-                    season,
-                    int(e['episodenumber']),
-                    title,
-                    ext
-                    )
-            logging.info("%s -> S%02dE%02d " % (
-                    f,
-                    season,
-                    int(e['episodenumber'])
-                    ))
-
-            count += 1
-            try:
-                os.link(f, d)
-            except OSError as e:
-                logging.warning("%s: %s" % (d, e.strerror))
+            os.link(f, d)
+        except OSError as e:
+            logging.warning("%s: %s" % (d, e.strerror))
 
 if count != len(s):
     logging.warning("processed %d files but expected %d" % (count, len(s)))