Right to Left Direction in Word to Persian

app.py

@@ -7,6 +7,8 @@ from fpdf import FPDF  # For PDF output
 from pptx import Presentation  # For PowerPoint output
 import subprocess  # To use ffmpeg for embedding subtitles
 import shlex  # For better command-line argument handling
+from docx.oxml.ns import qn
+from docx.oxml import OxmlElement
 
 # Load the Whisper model
 model = whisper.load_model("tiny")  # Smaller model for faster transcription
@@ -81,16 +83,35 @@ def embed_hardsub_in_video(video_file, srt_file, output_video):
     except Exception as e:
         raise RuntimeError(f"Error running ffmpeg: {e}")
 
-def write_word(transcription, output_file, tokenizer=None, translation_model=None):
-    """Creates a Word document from the transcription without timestamps."""
+from docx.oxml.ns import qn
+from docx.oxml import OxmlElement
+
+def write_word(transcription, output_file, tokenizer=None, translation_model=None, target_language=None):
+    """Creates a Word document from the transcription with support for RTL when translating to Persian."""
     doc = Document()
+    
+    # Check if the target language is Persian for RTL text direction
+    rtl = target_language == "fa"
+    
     for i, segment in enumerate(transcription['segments']):
         text = segment['text']
         
         if translation_model:
             text = translate_text(text, tokenizer, translation_model)
+
+        # Add a paragraph with the text
+        para = doc.add_paragraph(f"{i + 1}. {text.strip()}")
         
-        doc.add_paragraph(f"{i + 1}. {text.strip()}")  # No timestamps
+        # If RTL is required, modify the paragraph's properties
+        if rtl:
+            # Set the paragraph direction to RTL
+            para_format = para.paragraph_format
+            para_format.right_to_left = True
+            
+            # Set RTL for the text itself
+            run = para.runs[0]
+            run._element.rPr.append(OxmlElement('w:bidi'))
+
     doc.save(output_file)
 
 def write_pdf(transcription, output_file, tokenizer=None, translation_model=None):