Update app.py

app.py

@@ -2,44 +2,37 @@ import gradio as gr
 from transformers import pipeline
 import json
 import re
-import os
 
-# Load the phrase mapping from the JSON file
+# Load the phrase mapping from the JSON file (English to Swahili)
 with open('phrase_mappings.json', 'r') as f:
     phrase_mapping = json.load(f)
 
-# Define the UnifiedTranslator class
+# Define the UnifiedTranslator class for English-to-Swahili translation
 class UnifiedTranslator:
-    def __init__(self, model_path, phrase_mapping):
-        """
-        model_path: Local folder path to the model (e.g., 'translation_model/')
-        phrase_mapping: dict of custom phrases
-        """
-        if not os.path.exists(model_path):
-            raise FileNotFoundError(f"Model folder not found: {model_path}")
-
-        # Load translation pipeline from local folder
-        self.model = pipeline(
-            "translation",
-            model=model_path,
-            tokenizer=model_path
-        )
+    def __init__(self, model_name, phrase_mapping):
+        self.model = pipeline("translation", model=model_name)
         self.phrase_mapping = phrase_mapping
 
     def translate(self, text):
         # Normalize text to lowercase and strip extra spaces
         text_lower = text.lower().strip()
         
-        # Check custom phrase mappings first
+        # Debugging output
+        print(f"Input text: {text_lower}")
+
+        # Check if the text matches any pattern in the phrase_mapping (English to Swahili)
         for pattern, translation in self.phrase_mapping.items():
             try:
-                # Convert placeholders to regex
+                # Use regex to match the pattern with placeholders
                 pattern_regex = re.compile(
-                    re.escape(pattern).replace(r"\{name\}", r"([\w'-]+)"),
+                    re.escape(pattern).replace(r"\{name\}", r"([\w'-]+)").strip(), 
                     re.IGNORECASE
                 )
+                print(f"Checking pattern: {pattern}")
                 match = pattern_regex.fullmatch(text_lower)
                 if match:
+                    print(f"Match found: {match.group(0)}")
+                    # Replace the placeholder with the actual value if needed
                     if '{name}' in pattern:
                         return translation.format(name=match.group(1))
                     else:
@@ -47,16 +40,17 @@ class UnifiedTranslator:
             except re.error as e:
                 print(f"Regex error with pattern {pattern}: {e}")
         
-        # Fallback to model translation
+        # Fallback to model translation if no pattern matches
         try:
+            print(f"Fallback to model translation for text: {text}")
             translation = self.model(text)[0]
             return translation['translation_text']
         except Exception as e:
             print(f"Model translation error: {e}")
             return "Translation error occurred"
 
-# Initialize the translator with the local model folder
-translator = UnifiedTranslator("translation_model", phrase_mapping)
+# Initialize the UnifiedTranslator with your model and custom phrases
+translator = UnifiedTranslator("Bildad/English-Swahili_Translation", phrase_mapping)
 
 # Define the Gradio interface
 def translate_text(text):
@@ -71,4 +65,4 @@ iface = gr.Interface(
 )
 
 # Launch the interface
-iface.launch(share=True)
+iface.launch(share=True)