updated error handling in utils

utils.py

@@ -35,10 +35,11 @@ claude_client = Anthropic(api_key=os.getenv("ANTHROPIC_API_KEY"))
 # --------------------------------------------------------------------------
 # OpenAI
 # --------------------------------------------------------------------------
-
-def call_openai_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
+def call_openai_with_retries(prompt: str, model: str = None, max_retries: int = 3, backoff: float = 1.0) -> str:
+    """Call OpenAI model with retry logic."""
     model = model or os.getenv("OPENAI_MODEL", "gpt-4o-mini")
     last_err = None
+
     for attempt in range(1, max_retries + 1):
         try:
             resp = openai.ChatCompletion.create(
@@ -47,25 +48,26 @@ def call_openai_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
                 temperature=0.2,
                 max_tokens=1200,
             )
-            if "choices" in resp and len(resp.choices) > 0:
-                text = resp.choices[0].message.get("content", "").strip()
-                if text:
-                    return text
+            if resp.choices and resp.choices[0].message.get("content"):
+                return resp.choices[0].message["content"].strip()
             raise RuntimeError("Empty response from OpenAI")
+
         except Exception as e:
             last_err = e
+            logging.warning(f"OpenAI call failed (attempt {attempt}): {e}")
             if attempt < max_retries:
                 time.sleep(backoff * attempt)
             else:
-                raise
+                raise RuntimeError(f"OpenAI failed after {max_retries} attempts: {last_err}")
 
 # --------------------------------------------------------------------------
 # Claude
 # --------------------------------------------------------------------------
-
-def call_claude_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
+def call_claude_with_retries(prompt: str, model: str = None, max_retries: int = 3, backoff: float = 1.0) -> str:
+    """Call Anthropic Claude model with retry logic."""
     model = model or os.getenv("CLAUDE_MODEL", "claude-3-sonnet-20240229")
     last_err = None
+
     for attempt in range(1, max_retries + 1):
         try:
             resp = claude_client.messages.create(
@@ -74,26 +76,27 @@ def call_claude_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
                 temperature=0.2,
                 messages=[{"role": "user", "content": prompt}],
             )
-            if resp.content and len(resp.content) > 0 and resp.content[0].type == "text":
+            if resp.content and resp.content[0].type == "text":
                 text = resp.content[0].text.strip()
                 if text:
                     return text
             raise RuntimeError("Empty response from Claude")
+
         except Exception as e:
             last_err = e
+            logging.warning(f"Claude call failed (attempt {attempt}): {e}")
             if attempt < max_retries:
                 time.sleep(backoff * attempt)
             else:
-                raise
+                raise RuntimeError(f"Claude failed after {max_retries} attempts: {last_err}")
 
 # --------------------------------------------------------------------------
 # Groq
 # --------------------------------------------------------------------------
-
-def call_groq_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
+def call_groq_with_retries(prompt: str, model: str = None, max_retries: int = 3, backoff: float = 1.0) -> str:
+    """Call Groq model with retry logic."""
     api_key = os.getenv("GROQ_API_KEY")
     model = model or os.getenv("GROQ_MODEL", "llama3-8b-8192")
-
     if not api_key:
         raise RuntimeError("Missing GROQ_API_KEY")
 
@@ -106,25 +109,31 @@ def call_groq_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
         "model": model,
         "messages": [{"role": "user", "content": prompt}],
         "temperature": 0.2,
-        "max_tokens": 1000
+        "max_tokens": 1000,
     }
 
     last_err = None
     for attempt in range(1, max_retries + 1):
         try:
-            resp = requests.post("https://api.groq.com/openai/v1/chat/completions", headers=headers, json=payload)
+            resp = requests.post(
+                "https://api.groq.com/openai/v1/chat/completions",
+                headers=headers,
+                json=payload,
+                timeout=30,
+            )
+
             if resp.status_code == 200:
                 data = resp.json()
-                if data.get("choices"):
+                if data.get("choices") and data["choices"][0]["message"].get("content"):
                     return data["choices"][0]["message"]["content"].strip()
                 raise RuntimeError("Empty response from Groq")
 
-            # Capture 400s with context
-            print("Groq API error response:", resp.text)
+            logging.error(f"Groq API error ({resp.status_code}): {resp.text}")
             resp.raise_for_status()
 
-        except Exception as e:
+        except (RequestException, Exception) as e:
             last_err = e
+            logging.warning(f"Groq call failed (attempt {attempt}): {e}")
             if attempt < max_retries:
                 time.sleep(backoff * attempt)
             else:
@@ -134,8 +143,10 @@ def call_groq_with_retries(prompt, model=None, max_retries=3, backoff=1.0):
 # Generic handler
 # --------------------------------------------------------------------------
 
-def call_llm(prompt):
+def call_llm(prompt: str) -> str:
+    """Route LLM calls based on PROVIDER environment variable."""
     provider = os.getenv("PROVIDER", "openai").lower()
+
     if provider == "claude":
         return call_claude_with_retries(prompt)
     elif provider == "openai":
@@ -143,4 +154,4 @@ def call_llm(prompt):
     elif provider == "groq":
         return call_groq_with_retries(prompt)
     else:
-        raise RuntimeError(f"Unsupported PROVIDER: {provider}")
+        raise RuntimeError(f"Unsupported PROVIDER: {provider}")