ox/file.py

@@ -31,11 +31,11 @@ EXTENSIONS = {
     'video': [
         '3gp',
         'avi', 'divx', 'dv', 'flv', 'm2t', 'm2ts', 'm2v', 'm4v', 'mkv', 'mov', 'mp4',
-        'mpeg', 'mpg', 'mts', 'ogm', 'ogv', 'vob', 'webm', 'wmv', 'asf',
+        'mpeg', 'mpg', 'mts', 'ogm', 'ogv', 'rm', 'rmvb', 'vob', 'webm', 'wmv', 'asf',
         'mod', 'tod',  # http://en.wikipedia.org/wiki/MOD_and_TOD
         'mxf', 'ts',
         'dat',  # VOD files
-        'rm', 'rmvb',  # Real Media
+        'rm',  # Real Media
     ],
 }
 

ox/web/imdb.py

@@ -158,6 +158,15 @@ class Imdb(SiteParser):
             'type': 'list'
         },
         'country': zebra_list('Country', more=['<a.*?>(.*?)</a>']),
+        'creator': {
+            'page': '',
+            're': [
+                '<div class="credit_summary_item">.*?<h4.*?>Creator.?:</h4>(.*?)</div>',
+                '<a href="/name/.*?>(.*?)</a>',
+                lambda ll: strip_tags(ll)
+            ],
+            'type': 'list'
+        },
         'director': reference_section('directors'),
         'editor': reference_section('editors'),
         'composer': reference_section('composers'),
@@ -320,8 +329,6 @@ class Imdb(SiteParser):
     }
 
     def read_url(self, url, timeout):
-        if self.debug:
-            print(url)
         if url not in self._cache:
             self._cache[url] = read_url(url, timeout=timeout, unicode=True)
         return self._cache[url]
@@ -444,6 +451,49 @@ class Imdb(SiteParser):
         if 'episodeTitle' in self:
             self['episodeTitle'] = re.sub('Episode \#\d+\.\d+', '', self['episodeTitle'])
 
+        if 'series' in self:
+            series = Imdb(self['series'], timeout=timeout)
+            self['seriesTitle'] = series['title']
+            if 'episodeTitle' in self:
+                self['seriesTitle'] = series['title']
+                if 'season' in self and 'episode' in self:
+                    self['title'] = "%s (S%02dE%02d) %s" % (
+                        self['seriesTitle'], self['season'], self['episode'], self['episodeTitle'])
+                else:
+                    self['title'] = "%s (S01) %s" % (self['seriesTitle'], self['episodeTitle'])
+                    self['season'] = 1
+                self['title'] = self['title'].strip()
+            if 'director' in self:
+                self['episodeDirector'] = self['director']
+
+            if 'creator' not in series and 'director' in series:
+                series['creator'] = series['director']
+                if len(series['creator']) > 10:
+                    series['creator'] = series['director'][:1]
+
+            for key in ['creator', 'country']:
+                if key in series:
+                    self[key] = series[key]
+
+            if 'year' in series:
+                self['seriesYear'] = series['year']
+                if 'year' not in self:
+                    self['year'] = series['year']
+
+            if 'year' in self:
+                self['episodeYear'] = self['year']
+            if 'creator' in self:
+                self['seriesDirector'] = self['creator']
+            if 'originalTitle' in self:
+                del self['originalTitle']
+        else:
+            for key in ('seriesTitle', 'episodeTitle', 'season', 'episode'):
+                if key in self:
+                    del self[key]
+        if 'creator' in self:
+            if 'director' in self:
+                self['episodeDirector'] = self['director']
+            self['director'] = self['creator']
 
         #make lists unique but keep order
         for key in ('director', 'language'):
@@ -531,57 +581,6 @@ class Imdb(SiteParser):
                     series_credit = [c for c in self['credits'] if c.get('deparment') == deparment]
                     if series_credit:
                         self[key] = [c['name'] for c in series_credit]
-        creator = []
-        for c in self.get('credits', []):
-            if '(created by)' in c['roles'] and c['name'] not in creator:
-                creator.append(c['name'])
-        if creator:
-            self['creator'] = creator
-
-        if 'series' in self:
-            series = Imdb(self['series'], timeout=timeout)
-            self['seriesTitle'] = series['title']
-            if 'episodeTitle' in self:
-                self['seriesTitle'] = series['title']
-                if 'season' in self and 'episode' in self:
-                    self['title'] = "%s (S%02dE%02d) %s" % (
-                        self['seriesTitle'], self['season'], self['episode'], self['episodeTitle'])
-                else:
-                    self['title'] = "%s (S01) %s" % (self['seriesTitle'], self['episodeTitle'])
-                    self['season'] = 1
-                self['title'] = self['title'].strip()
-            if 'director' in self:
-                self['episodeDirector'] = self['director']
-
-            if 'creator' not in series and 'director' in series:
-                series['creator'] = series['director']
-                if len(series['creator']) > 10:
-                    series['creator'] = series['director'][:1]
-
-            for key in ['creator', 'country']:
-                if key in series:
-                    self[key] = series[key]
-
-            if 'year' in series:
-                self['seriesYear'] = series['year']
-                if 'year' not in self:
-                    self['year'] = series['year']
-
-            if 'year' in self:
-                self['episodeYear'] = self['year']
-            if 'creator' in self:
-                self['seriesDirector'] = self['creator']
-            if 'originalTitle' in self:
-                del self['originalTitle']
-        else:
-            for key in ('seriesTitle', 'episodeTitle', 'season', 'episode'):
-                if key in self:
-                    del self[key]
-        if 'creator' in self:
-            if 'director' in self:
-                self['episodeDirector'] = self['director']
-            self['director'] = self['creator']
-
 
 class ImdbCombined(Imdb):
     def __init__(self, id, timeout=-1):