fix pricing not being entered

.env.example

@@ -6,17 +6,24 @@ BRICKSET_API_KEY = "your_brickset_api_key"
 # Square environment - "sandbox" for testing, "production" for live store
 SQUARE_ENV = "sandbox"  # or "production"
 
-SQUARE_WEBHOOK_URL = "https://yourdomain.com/hook"
+# Public webhook URL that Square will POST events to
+SQUARE_WEBHOOK_URL = "https://yourdomain.com/webhook"
 
 # Sandbox credentials - test environment (fake data, safe for development)
 SANDBOX_SQUARE_APP_ID = "your_sandbox_app_id"
 SANDBOX_SQUARE_TOKEN = "your_sandbox_access_token"
-SANDBOX_SQUARE_WEBHOOK_INVENTORY_ID = "your_sandbox_webhook_signature_key"
+
+# Sandbox webhook subscription ID and signature key for verifying webhook requests
+SANDBOX_SQUARE_WEBHOOK_SUB_ID = "your_sandbox_webhook_subscription_id"
+SANDBOX_SQUARE_WEBHOOK_SIGNATURE_KEY = "your_sandbox_webhook_signature_key"
 
 # Production credentials - live store (real inventory and sales)
 PROD_SQUARE_APP_ID = "your_production_app_id"
 PROD_SQUARE_TOKEN = "your_production_access_token"
-PROD_SQUARE_WEBHOOK_INVENTORY_ID = "your_production_webhook_signature_key"
+
+# Production webhook subscription ID and signature key for verifying webhook requests
+PROD_SQUARE_WEBHOOK_SUB_ID = "your_production_webhook_subscription_id"
+PROD_SQUARE_WEBHOOK_SIGNATURE_KEY = "your_production_webhook_signature_key"
 
 # Category ID for LEGO items in Square catalog (find in Dashboard → Items → Categories)
 SQUARE_LEGO_CAT_ID = "your_lego_category_id"

.version

@@ -0,0 +1 @@
+0.3.0

config/gunicorn.conf.py

@@ -1,8 +1,18 @@
 import multiprocessing
 import os
+import git
 import src.banner as banner
 
 
+def get_git_commit():
+    """Get first 6 characters of current git commit hash"""
+    try:
+        repo = git.Repo(search_parent_directories=True)
+        return repo.head.commit.hexsha[:6]
+    except Exception:
+        return "unknown"
+
+
 # Server socket
 bind = "127.0.0.1:7433"  # nginx will proxy to this
 backlog = 2048
@@ -54,26 +64,13 @@ def on_starting(server):
     """Called just before the master process is initialized."""
     print("Studspider WEB starting...")
 
-    # Load .env file before workers start
-    import dotenv
+    # Verify .env file exists (workers will load it via get_sq_env.py)
     from pathlib import Path
 
     env_path = Path(__file__).parent.parent / ".env"
     print(f"[@DEBUG] Looking for .env at: {env_path.absolute()}")
     if env_path.exists():
-        dotenv.load_dotenv(env_path)
-        print(f"✓ Loaded environment from {env_path}")
-
-        # Verify key was loaded
-        import os
-
-        key = os.getenv("PROD_SQUARE_WEBHOOK_SIGNATURE_KEY") or os.getenv(
-            "SANDBOX_SQUARE_WEBHOOK_SIGNATURE_KEY"
-        )
-        if key:
-            print(f"✓ SQUARE_WEBHOOK_SIGNATURE_KEY loaded: {key[:10]}...")
-        else:
-            print(f"⚠ SQUARE_WEBHOOK_SIGNATURE_KEY not found in .env")
+        print(f"✓ .env file found (workers will load via get_sq_env.py)")
     else:
         print(f"⚠ No .env file found at {env_path.absolute()}")
 
@@ -84,11 +81,21 @@ def on_reload(server):
 
 
 def when_ready(server):
-    print(banner.generate("studspider v0.1"))
+    commit_hash = get_git_commit()
+    banner_text = banner.generate("studspider v0.3.0")
+
+    # Add commit hash to the last line of banner
+    banner_lines = banner_text.rstrip("\n").split("\n")
+    if banner_lines:
+        banner_lines[-1] += f"  - {commit_hash}"
+
+    print("\n".join(banner_lines))
     print("Built for PLAY-WELL TEKNOLOGIES by contact@studspider.sh\n\n")
-    print(f"STUDSPIDER HOOK v0.1 RUNNING ON {bind}")
-    print(f"WORKERS: {workers}")
-    print(f"WORKING CLASS: {worker_class}\n")
+    print("PRODUCTION MODE\n" if os.getenv("SQUARE_ENV") == "production" else "SANDBOX MODE\n")
+    print(f"HOST:\t{bind}")
+    print(f"WORKERS:\t{workers}")
+    print(f"WORKER TYPE:\t{worker_class}")
+    print(f"{'=' * 60}\n")
 
 
 def worker_int(worker):

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "studpider"
-version = "0.1.0"
+version = "0.3.0"
 description = ""
 authors = [{ name = "CF", email = "ph0sgene@pm.me" }]
 requires-python = ">=3.13,<3.14"
@@ -21,6 +21,8 @@ dependencies = [
     "fastapi>=0.128.0",
     "uvicorn>=0.40.0",
     "gunicorn>=23.0.0",
+    "async-lru>=2.0.4",
+    "gitpython>=3.1.0",
 ]
 
 [build-system]

setup.py

@@ -2,7 +2,7 @@ from setuptools import setup
 
 setup(
     name="studspider",
-    version="0.1",
+    version="0.3.0",
     description="Bricket + Square integration application",
     author="CF",
     author_email="chris-forst@pm.me",

src/abricksetapi.py

@@ -11,7 +11,9 @@ import math
 import dotenv
 from typing import Dict, Any, Optional
 from datetime import datetime, timedelta
+from async_lru import alru_cache
 from . import throttler
+from . import get_sq_env
 
 Throttler = throttler.GlobalThrottler
 
@@ -21,11 +23,12 @@ dotenv.load_dotenv()
 class AsyncBricksetAPI:
     """Asynchronous client for the Brickset API"""
 
-    def __init__(self, debug: bool = False, logger=None):
+    def __init__(self, debug: bool = None, logger=None):
         self.username = os.getenv("BRICKSET_USERNAME")
         self.password = os.getenv("BRICKSET_PASS")
         self.api_key = os.getenv("BRICKSET_API_KEY")
-        self.debug = debug
+        # Use DEBUG_MODE from env if debug not explicitly set
+        self.debug = debug if debug is not None else get_sq_env.DEBUG_MODE
         self.headers = {"UserAgent": "studspider/0.1"}
         self.throttler = Throttler(rate_limit=500, period=60)
         self.user_hash = None
@@ -155,9 +158,10 @@ class AsyncBricksetAPI:
                     self.logger.info("Successfully authenticated with Brickset API")
                 return result
 
+    @alru_cache(maxsize=1000, ttl=604800)  # Cache 1000 sets, 7 days TTL
     async def get_set(self, set_number: int) -> Dict[str, Any]:
         """
-        Fetch LEGO set information asynchronously
+        Fetch LEGO set information asynchronously (cached for 7 days)
 
         Args:
             set_number (str): LEGO set number

src/asquare.py

@@ -124,8 +124,9 @@ async def get_item_by_variation_id(variation_id: str) -> dict:
 
         cursor = None
         page_count = 0
+        max_pages = 50  # Safety limit to prevent infinite loops
 
-        while True:
+        while page_count < max_pages:
             page_count += 1
             print(f"[@DEBUG] Searching page {page_count}...")
 
@@ -151,6 +152,7 @@ async def get_item_by_variation_id(variation_id: str) -> dict:
                     raise ValueError(f"No items found in LEGO category")
                 else:
                     # No more results
+                    print(f"[@DEBUG] No more results on page {page_count}")
                     break
 
             # Search through items on this page to find which has this variation
@@ -188,8 +190,15 @@ async def get_item_by_variation_id(variation_id: str) -> dict:
             cursor = result_json.get("cursor")
             if not cursor:
                 # No more pages
+                print(f"[@DEBUG] No more pages after page {page_count}")
                 break
 
+        # If we get here, variation wasn't found
+        if page_count >= max_pages:
+            raise ValueError(
+                f"Variation {variation_id} not found after searching {max_pages} pages (possible infinite loop prevented)"
+            )
+
         raise ValueError(
             f"Variation {variation_id} not found in any LEGO category item (searched {page_count} pages)"
         )
@@ -294,7 +303,6 @@ async def update_itm(item_id: str, lego_set: dict, item_data: dict = None):
         pw_name = lego_set["pw_name"]
         set_adjustedprice = lego_set["set_adjustredprice"]
 
-        # Convert price to cents for Square API
         price_in_cents = int(set_adjustedprice * 100)
 
         # Retry logic for version conflicts (race conditions with concurrent updates)
@@ -377,11 +385,19 @@ async def update_itm(item_id: str, lego_set: dict, item_data: dict = None):
                         "item_variation_data", {}
                     ).copy()
 
-                    # Update only specific fields
-                    updated_var_data["price_money"] = {
-                        "amount": price_in_cents,
-                        "currency": "USD",
-                    }
+                    # Check if this variation has VARIABLE_PRICING
+                    current_pricing_type = updated_var_data.get("pricing_type")
+                    
+                    # Only set price if not using variable pricing
+                    if current_pricing_type != "VARIABLE_PRICING":
+                        updated_var_data["price_money"] = {
+                            "amount": price_in_cents,
+                            "currency": "USD",
+                        }
+                        print(f"[@DEBUG] Setting price: ${set_adjustedprice} for variation {variation_id}")
+                    else:
+                        print(f"[@WARNING] Skipping price update for {variation_id} - has VARIABLE_PRICING enabled")
+
                     updated_var_data["sku"] = set_number  # LEGO set number
                     updated_var_data["upc"] = (
                         set_upc if set_upc else None
@@ -396,7 +412,11 @@ async def update_itm(item_id: str, lego_set: dict, item_data: dict = None):
                     updated_variations.append(updated_var)
                     variation_found = True
                 else:
-                    # Keep other variations as-is
+                    # Keeprice_money": {
+                            "amount": price_in_cents,
+                            "currency": "USD",
+                        },
+                        "p other variations as-is
                     updated_variations.append(existing_var)
 
             # If variation wasn't found (shouldn't happen), create new one
@@ -408,10 +428,6 @@ async def update_itm(item_id: str, lego_set: dict, item_data: dict = None):
                         "item_id": item_id,
                         "name": "Regular",
                         "pricing_type": "FIXED_PRICING",
-                        "price_money": {
-                            "amount": price_in_cents,
-                            "currency": "USD",
-                        },
                         "sku": set_number,
                         "upc": set_upc if set_upc else None,
                     },

src/get_sq_env.py

@@ -1,14 +1,21 @@
 import os
 import dotenv
+from pathlib import Path
 from square.environment import SquareEnvironment
 
-dotenv.load_dotenv()
+# Load .env from project root (parent of src/)
+env_path = Path(__file__).parent.parent / ".env"
+dotenv.load_dotenv(env_path)
 
 LEGO_CAT_ID = os.getenv("SQUARE_LEGO_CAT_ID")
 
 SQUARE_WEBHOOK_URL = os.getenv("SQUARE_WEBHOOK_URL")
 
-if os.getenv("SQUARE_ENV") == "production":
+# Check environment mode
+IS_PRODUCTION = os.getenv("SQUARE_ENV") == "production"
+DEBUG_MODE = not IS_PRODUCTION  # Debug logging enabled in sandbox/dev
+
+if IS_PRODUCTION:
     SQUARE_APP_ID = os.getenv("PROD_SQUARE_APP_ID")
     SQUARE_TOKEN = os.getenv("PROD_SQUARE_TOKEN")
     SQUARE_WEBHOOK_SIGNATURE_KEY = os.getenv("PROD_SQUARE_WEBHOOK_SIGNATURE_KEY")

src/glue.py

@@ -96,7 +96,8 @@ async def enrich_single_item(item_id: str, item_data: dict = None, logger=None)
             print(f"[@DEBUG] Extracted set number: {set_number}")
 
         # Fetch Brickset data (with throttling)
-        brickset_client = abricksetapi.AsyncBricksetAPI(debug=True, logger=logger)
+        # debug flag will be auto-set based on SQUARE_ENV
+        brickset_client = abricksetapi.AsyncBricksetAPI(logger=logger)
         lego_set = await brickset_client.throttled_get_set(set_number)
 
         if logger:
@@ -104,24 +105,10 @@ async def enrich_single_item(item_id: str, item_data: dict = None, logger=None)
         else:
             print(f"[@DEBUG] Retrieved Brickset data for {lego_set['pw_name']}")
 
-        # Check if manual price review needed
-        if lego_set.get("requires_manual_price"):
-            if logger:
-                logger.info(f"Item {item_id} requires manual price review (≥$110)")
-            else:
-                print(f"[@DEBUG] Item {item_id} requires manual price review")
+        # Note: Price updates disabled - enriching without price
+        # REVIEW table logic skipped
 
-            await db.upsert(
-                "REVIEW",
-                {
-                    "sq_object": item_id,
-                    "price": lego_set["set_adjustredprice"],
-                    "date_modified": int(time()),
-                },
-                "sq_object",
-            )
-
-        # Update Square item with enriched data
+        # Update Square item with enriched data (without price)
         success = await asquare.update_itm(item_id, lego_set, item_data=item)
 
         if success:
@@ -179,18 +166,37 @@ async def enrich_bulk(logger=None) -> None:
         else:
             print(f"[@DEBUG] Found {len(items)} items in LEGO category")
 
-        # Enrich each item
+        # Enrich each item with concurrent batches
         successful = 0
         failed = 0
+        batch_size = 20  # Process 20 items concurrently
 
-        for item in items:
-            item_id = item["id"]
-            result = await enrich_single_item(item_id, logger=logger)
+        for i in range(0, len(items), batch_size):
+            batch = items[i : i + batch_size]
+            batch_num = (i // batch_size) + 1
+            total_batches = (len(items) + batch_size - 1) // batch_size
 
-            if result:
-                successful += 1
+            if logger:
+                logger.info(
+                    f"Processing batch {batch_num}/{total_batches} ({len(batch)} items)"
+                )
             else:
-                failed += 1
+                print(
+                    f"[@DEBUG] Processing batch {batch_num}/{total_batches} ({len(batch)} items)"
+                )
+
+            # Process batch concurrently
+            tasks = [enrich_single_item(item["id"], logger=logger) for item in batch]
+            results = await asyncio.gather(*tasks, return_exceptions=True)
+
+            # Count results
+            for result in results:
+                if isinstance(result, Exception):
+                    failed += 1
+                elif result:
+                    successful += 1
+                else:
+                    failed += 1
 
         if logger:
             logger.info(

src/stats.py

@@ -4,15 +4,24 @@ Queries Square catalog to collect metrics about LEGO inventory quality.
 """
 
 import json
+import git
 from typing import Dict, Any
 from square import AsyncSquare
-import get_sq_env
+from . import get_sq_env
+from . import abricksetapi
+
+
+def get_git_commit() -> str:
+    """Get first 6 characters of current git commit hash"""
+    try:
+        repo = git.Repo(search_parent_directories=True)
+        return repo.head.commit.hexsha[:6]
+    except Exception:
+        return "unknown"
+
 
 # Initialize Square client
-client = AsyncSquare(
-    token=get_sq_env.SQUARE_TOKEN,
-    environment=get_sq_env.environment
-)
+client = AsyncSquare(token=get_sq_env.SQUARE_TOKEN, environment=get_sq_env.environment)
 
 LEGO_CAT_ID = get_sq_env.LEGO_CAT_ID
 
@@ -104,12 +113,62 @@ async def get_daily_stats() -> Dict[str, Any]:
     }
 
 
-async def get_total_items() -> int:
+async def get_brickset_cache_stats() -> Dict[str, Any]:
     """
-    Count total items in LEGO category.
+    Get Brickset API cache statistics.
 
     Returns:
-        int: Total number of items
+        dict: {
+            "cache_enabled": bool,
+            "cache_hits": int,
+            "cache_misses": int,
+            "cache_size": int,
+            "cache_maxsize": int,
+            "hit_rate": float (0-100%)
+        }
+    """
+    # Create a temporary client instance to access cache info
+    temp_client = abricksetapi.AsyncBricksetAPI()
+
+    # Check if get_set has cache_info (it should with @alru_cache)
+    if hasattr(temp_client.get_set, "cache_info"):
+        info = await temp_client.get_set.cache_info()
+        hits = info.hits
+        misses = info.misses
+        total = hits + misses
+        hit_rate = (hits / total * 100) if total > 0 else 0.0
+
+        return {
+            "cache_enabled": True,
+            "cache_hits": hits,
+            "cache_misses": misses,
+            "cache_size": info.currsize,
+            "cache_maxsize": info.maxsize,
+            "hit_rate": round(hit_rate, 2),
+        }
+    else:
+        return {
+            "cache_enabled": False,
+            "cache_hits": 0,
+            "cache_misses": 0,
+            "cache_size": 0,
+            "cache_maxsize": 0,
+            "hit_rate": 0.0,
+        }
+
+
+async def get_daily_stats() -> Dict[str, Any]:
+    """
+    Gather all metrics for daily_stats table in a single pass.
+
+    Returns:
+        dict: {
+            "total_items": int,
+            "enriched_items": int,
+            "missing_upcs": int,
+            "missing_images": int,
+            "total_catalog_value": float
+        }
     """
     result = await client.catalog.search(
         object_types=["ITEM"],
@@ -293,6 +352,8 @@ if __name__ == "__main__":
         print(f"Missing UPCs:         {stats['missing_upcs']}")
         print(f"Missing Images:       {stats['missing_images']}")
         print(f"Total Catalog Value:  ${stats['total_catalog_value']:,.2f}")
-        print(f"Completion Rate:      {(stats['enriched_items'] / stats['total_items'] * 100):.1f}%")
+        print(
+            f"Completion Rate:      {(stats['enriched_items'] / stats['total_items'] * 100):.1f}%"
+        )
 
     asyncio.run(main())

studspider.py

@@ -15,18 +15,6 @@ SQUARE_WEBHOOK_SIGNATURE_KEY = sq_env.SQUARE_WEBHOOK_SIGNATURE_KEY
 SQUARE_WEBHOOK_URL = sq_env.SQUARE_WEBHOOK_URL
 SQUARE_LEGO_CAT_ID = sq_env.LEGO_CAT_ID
 
-
-# Debug: Print configuration (redact key)
-if SQUARE_WEBHOOK_SIGNATURE_KEY:
-    print(f"[CONFIG] Signature key loaded: {SQUARE_WEBHOOK_SIGNATURE_KEY[:10]}...")
-else:
-    print("[CONFIG] WARNING: SQUARE_WEBHOOK_SIGNATURE_KEY not set!")
-
-if SQUARE_WEBHOOK_URL:
-    print(f"[CONFIG] Webhook URL: {SQUARE_WEBHOOK_URL}")
-else:
-    print("[CONFIG] WARNING: SQUARE_WEBHOOK_URL not set!")
-
 # In-memory store for processed event IDs (idempotency)
 processed_event_ids = set()
 MAX_STORED_EVENT_IDS = 10000
@@ -65,10 +53,11 @@ async def get_stats():
     """
     Get daily statistics for LEGO inventory.
 
-    Returns metrics about enrichment progress, missing data, and catalog value.
+    Returns metrics about enrichment progress, missing data, catalog value, and cache stats.
     """
     try:
         daily_stats = await stats.get_daily_stats()
+        cache_stats = await stats.get_brickset_cache_stats()
 
         # Calculate completion percentage
         if daily_stats["total_items"] > 0:
@@ -80,8 +69,11 @@ async def get_stats():
 
         return {
             "status": "success",
+            "version": "0.3.0",
+            "git_commit": stats.get_git_commit(),
             "stats": daily_stats,
             "completion_rate": round(completion_rate, 1),
+            "brickset_cache": cache_stats,
         }
     except Exception as e:
         return JSONResponse(