Update llama_api.py

llama_api.py

@@ -1,92 +1,125 @@
-import os
-import requests
+# import ollama
+
+# # Synchronous ask (kept for caching or non-stream calls)
+# def ask_ollama(prompt: str, model_name: str = "llama3"):
+#     response = ollama.chat(
+#         model=model_name,
+#         messages=[
+#             {"role": "system", "content": "You are a helpful assistant for college queries."},
+#             {"role": "user", "content": prompt}
+#         ]
+#     )
+#     return response.get("message", {}).get("content", "")
+
+# # Streaming generator: yields incremental text chunks
+# def ask_ollama_stream(prompt: str, model_name: str = "llama3"):
+#     stream = ollama.chat(
+#         model=model_name,
+#         messages=[
+#             {"role": "system", "content": "You are a helpful assistant for college queries."},
+#             {"role": "user", "content": prompt}
+#         ],
+#         stream=True
+#     )
+#     buffer = ""
+#     for chunk in stream:
+#         # chunk may contain partial content; combine
+#         text = chunk.get("message", {}).get("content", "")
+#         if text:
+#             # yield incremental text (could be full or partial)
+#             yield text
+
 from dotenv import load_dotenv
+import os
+from openai import OpenAI
 
+# Load environment variables
 load_dotenv()
 
-OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY")
+OPENROUTER_KEY = os.getenv("OPENROUTER_API_KEY")
+
+if not OPENROUTER_KEY:
+    raise ValueError(" Missing OPENROUTER_API_KEY in .env")
 
-# You can change this to any valid OpenRouter model ID
-# Avoid the ":free" suffix unless you're sure it's valid.
-# Example options (depending on your account/plan):
-#   "meta-llama/llama-3.1-8b-instruct"
-#   "meta-llama/llama-3.1-70b-instruct"
-MODEL_NAME = os.getenv(
-    "OPENROUTER_MODEL",
-    "meta-llama/llama-3.3-70b-instruct:free",  # safer default
+print("Loaded key prefix:", OPENROUTER_KEY[:15])
+
+client = OpenAI(
+    base_url="https://openrouter.ai/api/v1",
+    api_key=OPENROUTER_KEY,
 )
 
 
-def ask_ollama(prompt: str) -> str:
+
+def ask_ollama(prompt: str, model_name: str = "meta-llama/llama-3.3-70b-instruct:free"):
     """
-    Call OpenRouter chat completion API and return the assistant's message text.
-    If there's an error or no choices, return a readable error message and log details.
+    Sends a prompt to OpenRouter (Llama-3.3-70B-Instruct) and returns the response text.
+    Handles missing fields, errors, and empty responses gracefully.
     """
-    if not OPENROUTER_API_KEY:
-        return (
-            "⚠️ The language model is not configured. "
-            "OPENROUTER_API_KEY is missing on the server."
+    try:
+        print(f"🚀 Sending request to OpenRouter model: {model_name}")
+        completion = client.chat.completions.create(
+            extra_headers={
+                "HTTP-Referer": "https://ifheindia.org",
+                "X-Title": "IFHE Chatbot",
+            },
+            model=model_name,
+            messages=[
+                {"role": "system", "content": "You are a helpful academic assistant for IFHE University. Quote only factual content from context."},
+                {"role": "user", "content": prompt},
+            ],
         )
 
-    url = "https://openrouter.ai/api/v1/chat/completions"
-    headers = {
-        "Authorization": f"Bearer {OPENROUTER_API_KEY}",
-        "Content-Type": "application/json",
-        # Optional but recommended
-        "HTTP-Referer": "https://icfai-chatbot.local",
-        "X-Title": "ICFAI Digital MBA Assistant",
-    }
-
-    payload = {
-        "model": MODEL_NAME,
-        "messages": [
-            {
-                "role": "user",
-                "content": prompt,
-            }
-        ],
-        "temperature": 0.2,
-    }
-
-    print(f"🚀 Sending request to OpenRouter model: {MODEL_NAME}")
-    try:
-        resp = requests.post(url, json=payload, headers=headers, timeout=60)
+        
+        if not hasattr(completion, "choices") or not completion.choices:
+            print(" No choices returned from OpenRouter.")
+            return " No valid response received from the model."
+
+        message = getattr(completion.choices[0].message, "content", None)
+        if not message or not message.strip():
+            print(" Empty message content in completion.")
+            return " The model did not return any text."
+
+        print(" Model raw response:", message[:250])
+        return message.strip()
+
     except Exception as e:
-        print(f"❌ Network error calling OpenRouter: {e}")
-        return f"⚠️ Error contacting the language model: {e}"
+        print(" OpenRouter / Llama API Error:", e)
+        return f" Error communicating with the model: {e}"
+
 
-    print(f"🔹 OpenRouter status: {resp.status_code}")
-    # Always print the raw text once if something looks off
+
+def ask_ollama_stream(prompt: str, model_name: str = "meta-llama/llama-3.3-70b-instruct:free"):
+    """
+    Streams response token-by-token for real-time output.
+    Includes detailed logging for debugging.
+    """
     try:
-        data = resp.json()
-    except Exception as e:
-        print(f"❌ Failed to parse JSON from OpenRouter: {e}")
-        print("Raw response text:", resp.text[:500])
-        return "⚠️ Received an invalid response from the language model."
-
-    # If OpenRouter returned an error object
-    if "error" in data:
-        print("⚠️ OpenRouter error payload:", data["error"])
-        msg = data["error"].get("message") or str(data["error"])
-        return f"⚠️ LLM error from OpenRouter: {msg}"
-
-    choices = data.get("choices")
-    if not choices:
-        # <== THIS is your current situation
-        print("⚠️ No choices returned from OpenRouter. Full payload:")
-        print(data)
-        return (
-            "⚠️ I couldn't generate a response from the language model. "
-            "Please try again in a moment."
+        print(f" Connecting to OpenRouter model: {model_name}")
+        stream = client.chat.completions.create(
+            extra_headers={
+                "HTTP-Referer": "https://ifheindia.org",
+                "X-Title": "IFHE Chatbot",
+            },
+            model=model_name,
+            messages=[
+                {"role": "system", "content": "You are a helpful academic assistant for IFHE University."},
+                {"role": "user", "content": prompt},
+            ],
+            stream=True,
         )
 
-    try:
-        content = choices[0]["message"]["content"]
+        for chunk in stream:
+            # Log structure of each chunk
+            print(f" Chunk received: {chunk}")
+            if hasattr(chunk.choices[0].delta, "content"):
+                text = chunk.choices[0].delta.content
+                if text:
+                    print(f" Token: {text!r}")
+                    yield text
+
+        print(" Streaming complete.")
+
     except Exception as e:
-        print(f"⚠️ Unexpected response structure: {e}")
-        print("Full payload:", data)
-        return (
-            "⚠️ I received an unexpected response structure from the language model."
-        )
+        print(" Streaming error (inside llama_api):", e)
+        yield f" Error while streaming: {str(e)}"
 
-    return content.strip()