Update app.py

app.py

@@ -1,9 +1,11 @@
 import gradio as gr
 import pandas as pd
-import tempfile
 import os
+import json
+import tempfile
 from openai import OpenAI
 
+# Initialize OpenAI client
 client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
 
 # -------- Speech to Text --------
@@ -15,73 +17,118 @@ def transcribe_audio(file_path):
         )
     return transcript.text
 
-# -------- Extract CRM Fields --------
+
+# -------- Extract CRM Fields (SAFE JSON) --------
 def extract_fields(text):
     prompt = f"""
     Extract the following fields from the conversation:
+
     Name, Phone, Product, Budget, Location, Intent.
-    
+
+    Return ONLY valid JSON like:
+    {{
+      "Name": "",
+      "Phone": "",
+      "Product": "",
+      "Budget": "",
+      "Location": "",
+      "Intent": ""
+    }}
+
     Conversation:
     {text}
-    
-    Return in JSON format.
     """
 
     response = client.chat.completions.create(
         model="gpt-4o-mini",
-        messages=[{"role": "user", "content": prompt}]
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0
     )
 
+    content = response.choices[0].message.content
+
     try:
-        data = eval(response.choices[0].message.content)
+        data = json.loads(content)
     except:
-        data = {}
+        data = {
+            "Name": "",
+            "Phone": "",
+            "Product": "",
+            "Budget": "",
+            "Location": "",
+            "Intent": ""
+        }
 
     return data
 
+
 # -------- Main Processing --------
 def process_audio(audio_file):
     if audio_file is None:
-        return "No audio provided", None, None
+        return "No audio provided", pd.DataFrame(), None
+
+    try:
+        file_path = audio_file  # Gradio gives filepath directly
+
+        # Step 1: Transcription
+        text = transcribe_audio(file_path)
 
-    # Save temp file
-    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
-    temp_file.write(audio_file.read())
-    temp_file.close()
+        # Step 2: Extraction
+        data = extract_fields(text)
 
-    # Step 1: Transcribe
-    text = transcribe_audio(temp_file.name)
+        # Step 3: Convert to DataFrame
+        df = pd.DataFrame([data])
 
-    # Step 2: Extract fields
-    data = extract_fields(text)
+        # Step 4: Save Excel
+        excel_path = os.path.join(tempfile.gettempdir(), "crm_output.xlsx")
+        df.to_excel(excel_path, index=False)
 
-    # Convert to DataFrame
-    df = pd.DataFrame([data])
+        return text, df, excel_path
 
-    # Save Excel
-    excel_path = os.path.join(tempfile.gettempdir(), "crm_output.xlsx")
-    df.to_excel(excel_path, index=False)
+    except Exception as e:
+        return f"Error: {str(e)}", pd.DataFrame(), None
 
-    return text, df, excel_path
 
 # -------- UI --------
 with gr.Blocks() as app:
     gr.Markdown("# 🎙️ AI Voice to CRM Auto Filler")
 
     with gr.Tabs():
+        # 🎤 Record
         with gr.Tab("🎤 Record Inquiry"):
-            mic_input = gr.Audio(source="microphone", type="file")
+            mic_input = gr.Audio(
+                sources=["microphone"],
+                type="filepath",
+                label="Record Audio"
+            )
             btn1 = gr.Button("Process Recording")
 
+        # 📁 Upload
         with gr.Tab("📁 Upload Voice"):
-            file_input = gr.File(file_types=["audio"])
+            file_input = gr.Audio(
+                sources=["upload"],
+                type="filepath",
+                label="Upload Audio File"
+            )
             btn2 = gr.Button("Process File")
 
+    # Outputs
     transcript_output = gr.Textbox(label="Transcription")
     table_output = gr.Dataframe(label="Extracted CRM Data")
     download_btn = gr.File(label="Download Excel")
 
-    btn1.click(process_audio, inputs=mic_input, outputs=[transcript_output, table_output, download_btn])
-    btn2.click(process_audio, inputs=file_input, outputs=[transcript_output, table_output, download_btn])
+    # Actions
+    btn1.click(
+        fn=process_audio,
+        inputs=mic_input,
+        outputs=[transcript_output, table_output, download_btn]
+    )
+
+    btn2.click(
+        fn=process_audio,
+        inputs=file_input,
+        outputs=[transcript_output, table_output, download_btn]
+    )
 
+# Launch
 app.launch()