Convert blocks of space to newline

Optional because I don't know how general it is, but it's common for Chinese subtitles to use multiple spaces instead of newlines to separate speakers.

GUI/SettingsDialog.py

@@ -30,6 +30,7 @@ class SettingsDialog(QDialog):
             'max_batch_size': int,
             'scene_threshold': float,
             'batch_threshold': float,
+            'whitespaces_to_newline': bool,
             'max_context_summaries': int,
             'max_characters': int,
             'max_newlines': int,

PySubtitleGPT/Options.py

@@ -44,6 +44,7 @@ def env_bool(key, default=False):
     'max_context_summaries': int(os.getenv('MAX_CONTEXT_SUMMARIES', 10)),
     'max_characters': int(os.getenv('MAX_CHARACTERS', 120)),
     'max_newlines': int(os.getenv('MAX_NEWLINES', 3)),
+    'whitespaces_to_newline' : env_bool('WHITESPACES_TO_NEWLINE', False),
     'max_lines': int(os.getenv('MAX_LINES')) if os.getenv('MAX_LINES') else None, 
     'rate_limit': float(os.getenv('RATE_LIMIT')) if os.getenv('RATE_LIMIT') else None,
     'max_threads': int(os.getenv('MAX_THREADS', 4)),

PySubtitleGPT/SubtitleBatch.py

@@ -1,4 +1,5 @@
 from datetime import timedelta
+import re
 from PySubtitleGPT.SubtitleValidator import SubtitleValidator
 from PySubtitleGPT.SubtitleError import SubtitleError, TranslationError
 from PySubtitleGPT.Helpers import PerformSubstitutions
@@ -157,6 +158,14 @@ def PerformOutputSubstitutions(self, substitutions):
                     item.text = replacements.get(item.text) or item.text
 
             return replacements
+
+    def ConvertWhitespaceBlocksToNewlines(self):
+        """
+        Convert blocks of 3 or more spaces to a newline, unless there are newlines already
+        """
+        for item in self.originals:
+            if '\n' not in item.text:
+                item.text = re.sub(r' {3,}', '\n', item.text)
 
     def MergeLines(self, original_lines : list[int], translated_lines : list[int]):
         first_line = next((line for line in self.originals if line.number == original_lines[0]), None)

PySubtitleGPT/SubtitleTranslator.py

@@ -179,6 +179,10 @@ def TranslateBatches(self, batches : list[SubtitleBatch], context : dict, remain
             # Apply any substitutions to the input
             replacements = batch.PerformInputSubstitutions(substitutions)
 
+            # Split single lines with blocks of whitespace
+            if options.get('whitespaces_to_newline'):
+                batch.ConvertWhitespaceBlocksToNewlines()
+
             # Filter out empty lines
             originals = [line for line in batch.originals if line.text.strip()]