Update app.py

app.py

@@ -37,6 +37,7 @@ import difflib
 import time
 import shutil
 import requests
+import re
 
 import json
 import markdown
@@ -370,6 +371,18 @@ def save_to_google_storage(bucket_name, file_path, destination_blob_name, expira
     )
     print(f"✅ File uploaded to Google Cloud Storage. Signed URL: {signed_url}")
     return signed_url
+
+
+# Function to check if content is Simplified Chinese
+def is_simplified(text):
+    simplified_range = re.compile('[\u4e00-\u9fff]')  # Han characters in general
+    simplified_characters = [char for char in text if simplified_range.match(char)]
+    return len(simplified_characters) > len(text) * 0.5  # Threshold of 50% to be considered simplified
+
+# Function to choose the appropriate font for the content
+def choose_font_for_content(content):
+    return 'NotoSansSC' if is_simplified(content) else 'NotoSansTC'
+
 # Function to generate and save a document using ReportLab
 def generate_document(task_description, md_content, user_name='jayw', bucket_name='curify'):
     logger.debug("Starting to generate document")
@@ -385,29 +398,39 @@ def generate_document(task_description, md_content, user_name='jayw', bucket_nam
     local_filename = f"{truncated_hash}.pdf"  # Use the truncated hash as the local file name
     c = canvas.Canvas(local_filename, pagesize=letter)
 
-    # Load the .ttc font using ReportLab's TTFont and set it as the font for the canvas
-    font_path = 'NotoSansCJK-Regular.ttc'  # Path to your .ttc file
+    # Paths to the TTF fonts for Simplified and Traditional Chinese
+    sc_font_path = 'NotoSansSC-Regular.ttf'  # Path to Simplified Chinese font
+    tc_font_path = 'NotoSansTC-Regular.ttf'  # Path to Traditional Chinese font
+
     try:
-        # Register the font from the .ttc file
-        cttc_font = TTFont('NotoSansCJK', font_path)
-        c.ttfonts.addRegisteredFont('NotoSansCJK', cttc_font)
-        c.setFont('NotoSansCJK', 12)
+        # Register the Simplified Chinese font
+        sc_font = TTFont('NotoSansSC', sc_font_path)
+        c.ttfonts.addRegisteredFont('NotoSansSC', sc_font)
+
+        # Register the Traditional Chinese font
+        tc_font = TTFont('NotoSansTC', tc_font_path)
+        c.ttfonts.addRegisteredFont('NotoSansTC', tc_font)
+        
+        # Set default font (Simplified Chinese or Traditional Chinese depending on content)
+        c.setFont('NotoSansSC', 12)
     except Exception as e:
-        logger.error(f"Error loading font from {font_path}: {e}")
-        raise RuntimeError("Failed to load the unified font. Ensure the font file is accessible.")
+        logger.error(f"Error loading font files: {e}")
+        raise RuntimeError("Failed to load one or more fonts. Ensure the font files are accessible.")
 
     # Set initial Y position for drawing text
     y_position = 750  # Starting position for text
 
     # Process dictionary and render content
     for key, value in md_content.items():
-        # Add key as a header
-        c.setFont('NotoSansCJK', 14)
+        # Choose the font based on the key (header)
+        c.setFont(choose_font_for_content(key), 14)
         c.drawString(100, y_position, f"# {key}")
         y_position -= 20
 
+        # Choose the font for the value
+        c.setFont(choose_font_for_content(str(value)), 12)
+
         # Add value
-        c.setFont('NotoSansCJK', 12)  # Regular font for content
         if isinstance(value, list):  # Handle lists
             for item in value:
                 c.drawString(100, y_position, f"- {item}")
@@ -419,7 +442,7 @@ def generate_document(task_description, md_content, user_name='jayw', bucket_nam
         # Check if the page needs to be broken (if Y position is too low)
         if y_position < 100:
             c.showPage()  # Create a new page
-            c.setFont('NotoSansCJK', 12)  # Reset font
+            c.setFont('NotoSansSC', 12)  # Reset font
             y_position = 750  # Reset the Y position for the new page
 
     # Save the PDF
@@ -432,7 +455,8 @@ def generate_document(task_description, md_content, user_name='jayw', bucket_nam
     public_url = save_to_google_storage(bucket_name, local_filename, destination_blob_name)
     logger.debug("Finished generating document")
     return public_url
-    
+
+
 # In[10]: