overlapping subtitles

subtitles.py

@@ -113,15 +113,64 @@ def render_subtitles(item_json, output_json, output_srt, lang):
                 elif slang == lang:
                     subs[sub_id]['value'].append(value)
         position += clip['duration']
+
+    if isinstance(lang, list):
+        #fixme = [sub for sub in subs.values() if [s for s in list(subs.values())]]
+        fixme = [sub for sub in subs.values() if set(sub['languages']) != set(lang) and sub['value']]
+        if fixme:
+            remove = []
+            for key, sub in list(subs.items()):
+                intersections = []
+                for s in list(subs.values()):
+                    intersections += overlaps(sub, s)
+                if intersections:
+                    points = list(sorted(set([sub['in'], sub['out']] + intersections)))
+                    #print(points, sub['value'])
+                    sub_in = points[0]
+                    for sub_out in points[1:]:
+                        sub_id = '%0.3f-%0.3f' % (sub_in, sub_out)
+                        if sub_id not in subs:
+                            subs[sub_id] = {
+                                'in': sub_in,
+                                'out': sub_out,
+                                'value': [],
+                                'ids': [],
+                                'languages': []
+                            }
+                        if set(subs[sub_id]['languages']) != set(lang):
+                            if not subs[sub_id]['languages']:
+                                subs[sub_id]['value'] += sub['value']
+                                subs[sub_id]['languages'] += sub['languages']
+                                subs[sub_id]['ids'] += sub['ids']
+                            elif subs[sub_id]['languages'] == [lang[0]] \
+                                    and sub['languages'][0] not in subs[sub_id]['languages']:
+                                subs[sub_id]['value'].append(sub['value'][0])
+                                subs[sub_id]['languages'].append(sub['languages'][0])
+                                subs[sub_id]['ids'] += sub['ids']
+                            elif subs[sub_id]['languages'] == [lang[1]] \
+                                    and sub['languages'][0] not in subs[sub_id]['languages']:
+                                subs[sub_id]['value'].insert(0, sub['value'][0])
+                                subs[sub_id]['languages'].insert(0, sub['languages'][0])
+                                subs[sub_id]['ids'] += sub['ids']
+                            #else:
+                            #    print('WTF', sub['languages'], subs[sub_id]['languages'])
+
+                        sub_in = sub_out
+                    remove.append(key)
+            #for key, sub in list(subs.items()):
+            #    if len(sub['languages']) == 1:
+            #        del subs[key]
+            for key in remove:
+                if len(subs[key]['languages']) == 1:
+                    del subs[key]
+            for key, sub in list(subs.items()):
+                if abs(sub['out'] - sub['in']) <= 0.040001:
+                    del subs[key]
     subs = sorted(subs.values(), key=lambda c: (c['in'], c['out']))
     for sub in subs:
         sub['value'] = '\n'.join(sub['value'])
-        subtitles.append(sub)
+        if sub['value'].strip():
-
+            subtitles.append(sub)
-    if isinstance(lang, list):
-        fixme = [sub for sub in subs if set(sub['languages']) != set(lang)]
-        if fixme:
-            print('split/merge overlaps', output_srt, output_json)
 
     if output_srt:
         with open(output_srt, 'wb') as fd:
@@ -130,6 +179,20 @@ def render_subtitles(item_json, output_json, output_srt, lang):
         json.dump(subtitles, fd, indent=4, ensure_ascii=False, sort_keys=True)
 
 
+def overlaps(src, other):
+    src_in = float('%0.3f' % src['in'])
+    src_out = float('%0.3f' % src['out'])
+    other_in = float('%0.3f' % other['in'])
+    other_out = float('%0.3f' % other['out'])
+    points = []
+    if src_in != other_in or src_out != other_out:
+        # src inside
+        if other_in >= src_in and other_in < src_out:
+            points += [other['in']]
+        if other_out > src_in and other_out <= src_out:
+            points += [other['out']]
+    return points
+
 if __name__ == '__main__':
     if os.path.exists('subtitles.json'):
         items = json.load(open('subtitles.json'))