Added resync function to line text with correct timestamps.

parse-subs.py

@@ -26,14 +26,35 @@ def parse_vtt(file_path):
 def remove_duplicates(subtitles):
     no_dupes = list()
     prev = subtitles[0]
-    # print(type(prev), ":", prev)
     for caption in subtitles:
         if prev["text"] != str() and caption["text"] != prev["text"]:
-            # print(prev)
             no_dupes.append(prev)
         prev = caption
     return no_dupes
 
+def resync(subtitles):
+    # create placeholder timestamp for the first
+    init = "00:00:00.000 --> 00:00:5.000 align:start position:0%"
+    new_subs = list()
+    new_subs.append({
+                "time": init,
+                "text": subtitles[0]["text"]
+            })
+    for i in range(len(subtitles)):
+        if i == len(subtitles) -1:
+            break
+        # we don't need the first text
+        # but we still need the fuck timestamp
+        curr  = subtitles[i] 
+        next_ = subtitles[i+1]
+        new_time = curr["time"]
+        new_text = next_["text"]
+        new_subs.append({
+                "time": new_time,
+                "text": new_text
+            })
+    return new_subs
+
 def export_to_vtt(subtitles, output_file_path):
     with open(output_file_path, 'w', encoding='utf-8') as file:
         # Write the WebVTT header
@@ -55,9 +76,9 @@ if __name__ == "__main__":
     print(vtt_file_path)
 
     export_to_vtt(
-        remove_duplicates(
+        resync(remove_duplicates(
             parse_vtt(vtt_file_path)
-        ), "output.vtt"
+        )), "output.vtt"
     )