Update model_api.py

model_api.py

@@ -1,22 +1,162 @@
-import os
-from huggingface_hub import InferenceClient
-
-def query_model(prompt):
-    HF_TOKEN = os.getenv("HF_TOKEN")
-    client = InferenceClient(
-        model="Qwen/Qwen2.5-72B-Instruct",
-        token=HF_TOKEN
-    )
-    response = client.chat_completion(
-        messages=[
-            {
-                "role": "system",
-                "content": "You are a certified professional fitness trainer. Provide detailed, structured, day-by-day workout plans."
-            },
-            {"role": "user", "content": prompt}
-        ],
-        max_tokens=2500,
-        temperature=0.7
-    )
-    return response.choices[0].message.content
-    
+import os, time
+
+def query_model(prompt, max_tokens=1500):
+    """Single Groq API call."""
+    GROQ_API_KEY = os.getenv("GROQ_API_KEY", "").strip()
+
+    if not GROQ_API_KEY:
+        raise ValueError(
+            "GROQ_API_KEY is not set.\n\n"
+            "Steps to fix:\n"
+            "1. Go to https://console.groq.com → sign up free\n"
+            "2. Click 'API Keys' → 'Create API Key'\n"
+            "3. Copy the key (starts with gsk_...)\n"
+            "4. In your HuggingFace Space → Settings → Variables and Secrets\n"
+            "   → Add secret: Name = GROQ_API_KEY, Value = gsk_xxxxxxxxxxxx\n"
+            "5. Restart the Space"
+        )
+
+    try:
+        from groq import Groq
+        client = Groq(api_key=GROQ_API_KEY)
+
+        response = client.chat.completions.create(
+            model="llama-3.3-70b-versatile",   # fastest + best quality on Groq
+            messages=[
+                {
+                    "role": "system",
+                    "content": (
+                        "You are a certified professional fitness trainer. "
+                        "Provide detailed, structured, day-by-day workout plans exactly as instructed."
+                    )
+                },
+                {"role": "user", "content": prompt}
+            ],
+            max_tokens=max_tokens,
+            temperature=0.7,
+        )
+        return response.choices[0].message.content
+
+    except Exception as e:
+        err = str(e)
+
+        if "401" in err or "invalid_api_key" in err.lower() or "authentication" in err.lower():
+            raise ValueError(
+                "❌ Groq API key is invalid or expired.\n\n"
+                "Fix: Go to https://console.groq.com/keys → create a new key → "
+                "update GROQ_API_KEY in your HuggingFace Space Secrets."
+            ) from None
+
+        if "429" in err or "rate_limit" in err.lower() or "rate limit" in err.lower():
+            raise ValueError(
+                "⏳ Groq rate limit hit. Wait 60 seconds and try again.\n\n"
+                "Free tier: 6,000 tokens/min. Upgrade at console.groq.com for higher limits."
+            ) from None
+
+        if "503" in err or "504" in err or "timeout" in err.lower():
+            raise ValueError(
+                "⚡ Groq server busy. Please try again in a few seconds."
+            ) from None
+
+        raise
+
+
+def query_model_chunked(name, gender, height, weight, goal, fitness_level,
+                         equipment, days_per_week=5, months=1,
+                         progress_callback=None):
+    """
+    Generate a long plan by splitting into chunks of 3 days each.
+    Groq is fast enough (~2-3s per chunk) so even 30-day plans finish in ~30s.
+    progress_callback(chunk_num, total_chunks, days_done, total_days) — optional.
+    Returns: (full_plan_text, bmi, bmi_cat)
+    """
+    from prompt_builder import calculate_bmi, bmi_category, bmi_advice
+
+    bmi      = calculate_bmi(weight, height)
+    bmi_cat  = bmi_category(bmi)
+    advice   = bmi_advice(bmi_cat)
+
+    total_days = min(days_per_week * 4 * months, 30)
+    eq_list    = ", ".join(equipment) if equipment else "Bodyweight only"
+
+    intensity_map = {
+        "Beginner":     "2–3 sets, moderate weight, 90s rest. Prioritise form.",
+        "Intermediate": "3–4 sets, progressive overload, 60–75s rest.",
+        "Advanced":     "4–5 sets, heavy compounds, 45–60s rest, supersets.",
+    }
+    intensity = intensity_map.get(fitness_level, "3 sets, 60s rest.")
+
+    CHUNK_SIZE   = 3       # 3 days per call — well within Groq token limits
+    chunks       = []
+    start_day    = 1
+    total_chunks = max(1, (total_days + CHUNK_SIZE - 1) // CHUNK_SIZE)
+
+    while start_day <= total_days:
+        end_day   = min(start_day + CHUNK_SIZE - 1, total_days)
+        chunk_num = len(chunks) + 1
+
+        if progress_callback:
+            progress_callback(chunk_num, total_chunks, start_day - 1, total_days)
+
+        context_note = ""
+        if chunks:
+            context_note = (
+                f"(Continue from Day {start_day}. Do NOT repeat Days 1–{start_day-1}. "
+                f"Vary muscle groups from previous days.)"
+            )
+
+        prompt = f"""Fitness trainer. Generate ONLY Days {start_day}-{end_day} of a {total_days}-day plan.
+Client: {name}, {gender}, {height}cm/{weight}kg, BMI {bmi:.1f} ({bmi_cat}), Goal: {goal}, Level: {fitness_level}, Equipment: {eq_list}.
+{context_note}
+
+For EACH day use EXACTLY this format:
+## Day N - [Muscle Group]
+**Warm-Up** - Exercise: 2x10, Exercise: 2x10
+**Main Workout**
+- ExerciseName — 3x12 reps (rest 60s)
+- ExerciseName — 3x12 reps (rest 60s)
+- ExerciseName — 3x12 reps (rest 60s)
+- ExerciseName — 3x12 reps (rest 60s)
+- ExerciseName — 3x12 reps (rest 60s)
+**Cool-Down** - Stretch1, Stretch2
+
+Rules: vary muscle groups each day, {fitness_level} appropriate, use {eq_list} only, intensity: {intensity}.
+{"End with 1 motivational sentence for "+name+"." if end_day == total_days else "No motivational text — more days follow."}
+Output ONLY Days {start_day}-{end_day}. No preamble. No extra text.
+"""
+
+        # Retry up to 3 times on transient errors
+        for attempt in range(3):
+            try:
+                chunk_text = query_model(prompt, max_tokens=1500)
+                break
+            except ValueError as e:
+                # Rate limit — wait and retry
+                err = str(e)
+                if "rate limit" in err.lower() and attempt < 2:
+                    wait = 65  # wait just over 1 minute for rate limit window to reset
+                    if progress_callback:
+                        progress_callback(chunk_num, total_chunks, start_day - 1, total_days,
+                                          status=f"Rate limit — waiting {wait}s…")
+                    time.sleep(wait)
+                    continue
+                raise
+            except Exception as e:
+                err = str(e)
+                if attempt < 2:
+                    time.sleep(5 * (attempt + 1))
+                    continue
+                raise
+
+        chunks.append(chunk_text.strip())
+        start_day = end_day + 1
+
+        # Small pause between chunks to respect rate limits on free tier
+        if start_day <= total_days:
+            time.sleep(1)
+
+    if progress_callback:
+        progress_callback(total_chunks, total_chunks, total_days, total_days)
+
+    full_plan = "\n\n".join(chunks)
+    return full_plan, bmi, bmi_cat