walmart-scraper / tool.py
bugrahans's picture
Upload 4 files
0cba968 verified
Raw History Blame Contribute Delete
17.3 kB
from __future__ import annotations
import json
import os
import re
import time
import urllib.parse
from concurrent.futures import ThreadPoolExecutor
from typing import Any, Dict, List, Optional, Tuple
import requests
from smolagents.tools import Tool
SCRAPE_DO_SIGNUP_URL = (
"https://dashboard.scrape.do/sign-up?utm_source=huggingface&utm_medium=walmart-scraper"
)
SEARCH_ENDPOINT = "https://api.scrape.do/plugin/walmart/search"
PRODUCT_ENDPOINT = "https://api.scrape.do/plugin/walmart/product"
# From the Scrape.do docs: 10 credits per successful call, failed calls are free.
CREDITS_PER_CALL = 10
# Secaucus, NJ. The store used in Scrape.do's Walmart scraping guide.
DEFAULT_LOCATION = "07094"
# Secaucus NJ, Deer Park IL, Dallas TX, Los Angeles CA.
DEFAULT_COMPARE_LOCATIONS = ["07094", "60074", "75201", "90012"]
MAX_COMPARE_LOCATIONS = 5
# ------------------------------------------------------------------ helpers
def parse_location(value: Optional[str]) -> Tuple[Optional[Dict[str, str]], Optional[str]]:
"""Read a 5-digit ZIP code or a Walmart store number.
Returns ({"zipcode": ...} or {"store": ...}, None) or (None, error message).
"""
text = (value or "").strip() or DEFAULT_LOCATION
cleaned = text.lower().replace("#", "").replace("no.", "").strip()
match = re.fullmatch(r"store\s*:?\s*(\d{1,6})", cleaned)
if match:
return {"store": match.group(1)}, None
if re.fullmatch(r"\d{5}", cleaned):
return {"zipcode": cleaned}, None
if re.fullmatch(r"\d{5}-\d{4}", cleaned):
return {"zipcode": cleaned[:5]}, None
if re.fullmatch(r"\d{1,4}", cleaned):
return {"store": cleaned}, None
return None, f"“{text}” isn't a 5-digit US ZIP code or a Walmart store number."
def split_locations(value: Optional[str]) -> List[str]:
parts = [p.strip() for p in re.split(r"[,;\n]+", value or "") if p.strip()]
seen: List[str] = []
for part in parts:
if part not in seen:
seen.append(part)
return seen
def parse_item_id(text: Optional[str]) -> Optional[str]:
"""Pull Walmart's numeric item id out of a product link, or accept a bare id."""
decoded = urllib.parse.unquote(text or "").strip()
match = re.search(r"/ip/(?:[^/?#]+/)*(\d{4,})", decoded)
if match:
return match.group(1)
match = re.search(r"[?&](?:itemId|item_id|usItemId|item)=(\d{4,})", decoded)
if match:
return match.group(1)
if re.fullmatch(r"\d{4,15}", decoded):
return decoded
return None
def _is_url(text: str) -> bool:
return text.lower().startswith(("http://", "https://", "www.", "walmart.com"))
def describe_store(store: Optional[Dict[str, Any]]) -> str:
if not isinstance(store, dict) or not store:
return ""
place = ", ".join(p for p in [store.get("city"), store.get("state")] if p)
label = f"Walmart #{store.get('store_id')}" if store.get("store_id") else "Walmart"
if place:
label += f", {place}"
if store.get("postal_code"):
label += f" {store['postal_code']}"
return label
# --------------------------------------------------------------------- tool
class ScrapeDoWalmartTool(Tool):
name = "walmart_scraper"
description = """
Scrape Walmart.com. Walmart prices and stock depend on the store, so every
call is tied to a US ZIP code or a Walmart store number. Three modes:
- 'search': products for a keyword at one store, with price, list price,
rating, review count, seller, stock status, pickup/delivery and item id.
- 'product': full details for one product: price, discount, stock, seller,
images, highlights, specifications, UPC/GTIN, model number, description.
- 'compare': the same product's price and stock at several stores at once.
For 'product' and 'compare', pass a Walmart product link, an item id, or a
product name. Powered by Scrape.do's Walmart API, which handles proxies,
bot protection and store selection.
"""
output_type = "string"
def __init__(self, api_token: Optional[str] = None) -> None:
self._api_token = api_token
self.inputs = {
"mode": {
"type": "string",
"description": "'search' for a keyword, 'product' for one product's details, 'compare' for one product's price across stores.",
},
"query": {
"type": "string",
"description": (
"For 'search': the keyword, e.g. 'whole milk'. For 'product' and 'compare': a Walmart "
"product link, the numeric item id, or a product name."
),
},
"location": {
"type": "string",
"description": (
"A 5-digit US ZIP code or a Walmart store number. For 'compare', a comma-separated list "
"of up to 5. Defaults to ZIP 07094, or four cities across the US for 'compare'."
),
"nullable": True,
},
}
super().__init__()
# ------------------------------------------------------------ plumbing
def _resolve_token(self) -> Optional[str]:
if self._api_token:
return self._api_token.strip()
env_token = os.getenv("SCRAPE_DO_API_TOKEN")
return env_token.strip() if env_token else None
@staticmethod
def _missing_token() -> Dict[str, Any]:
return {
"status": "error",
"errorMessage": "Missing Scrape.do API token.",
"hint": f"Get a free token at {SCRAPE_DO_SIGNUP_URL}",
}
@staticmethod
def _as_error(data: Dict[str, Any]) -> Dict[str, Any]:
result = {"status": "error", "errorMessage": data.get("_error", "Unknown error")}
if data.get("_auth"):
result["hint"] = f"Check your token or create one at {SCRAPE_DO_SIGNUP_URL}"
return result
def _get(self, url: str, params: Dict[str, Any], retries: int = 1) -> Dict[str, Any]:
"""GET a Scrape.do endpoint. Retries once on 5xx, which Scrape.do doesn't charge for."""
token = str(params.get("token") or "")
def redact(text: str) -> str:
return text.replace(token, "***") if token else text
full_url = url + "?" + urllib.parse.urlencode(params, quote_via=urllib.parse.quote_plus)
last: Dict[str, Any] = {"_error": "Request failed."}
for attempt in range(retries + 1):
try:
response = requests.get(full_url, timeout=120)
except requests.exceptions.RequestException as exc:
last = {"_error": redact(f"Request failed: {exc}")}
if attempt < retries:
time.sleep(1.5)
continue
return last
if response.status_code == 401:
return {"_error": "Scrape.do rejected the token (HTTP 401).", "_auth": True}
try:
data = response.json()
except ValueError:
data = None
if response.ok and isinstance(data, dict):
return data
message = ""
if isinstance(data, dict):
parts = [str(data.get(key)).strip().rstrip(".") for key in ("error", "message") if data.get(key)]
message = ": ".join(parts)
if not message:
message = (response.text or "").strip()[:200] or "no details"
if response.status_code == 502:
text = (
"Walmart didn't return a valid page for this store. Failed calls aren't charged, "
f"so try again or use another ZIP code. ({message})"
)
else:
text = f"Scrape.do returned HTTP {response.status_code}: {message}"
last = {"_error": redact(text), "_status": response.status_code}
if response.status_code in (500, 502, 503, 504) and attempt < retries:
time.sleep(1.5)
continue
return last
return last
def _resolve_item(
self, token: str, ref: str, location: Dict[str, str]
) -> Dict[str, Any]:
"""Item id from a link, an id, or (one extra search) a product name."""
text = (ref or "").strip()
if not text:
return {"error": "Paste a Walmart product link, an item id, or a product name."}
item_id = parse_item_id(text)
if item_id:
return {"item_id": item_id, "credits": 0, "note": None}
if _is_url(text):
if "walmart." not in text.lower():
return {"error": "That isn't a Walmart link. Paste a walmart.com product link."}
return {
"error": "Couldn't find a product id in that link. Open the product page on Walmart "
"(the address contains /ip/) and copy that link, or use the Search tab."
}
data = self._get(SEARCH_ENDPOINT, dict({"token": token, "q": text}, **location))
if "_error" in data:
return {"error": data["_error"], "auth": data.get("_auth")}
results = [r for r in data.get("results") or [] if r.get("item_id")]
if not results:
return {"error": f"Walmart found no product for “{text}”."}
best = next((r for r in results if not r.get("sponsored")), results[0])
note = (
f"Searched Walmart for “{text}” and used the first match: {best.get('name')} "
f"({CREDITS_PER_CALL} credits)."
)
return {"item_id": str(best["item_id"]), "credits": CREDITS_PER_CALL, "note": note}
# --------------------------------------------------------------- search
def search(self, query: str, location: Optional[str] = None) -> Dict[str, Any]:
token = self._resolve_token()
if not token:
return self._missing_token()
if not query or not query.strip():
return {"status": "error", "errorMessage": "Type what you're looking for, e.g. 'whole milk'."}
loc, error = parse_location(location)
if error:
return {"status": "error", "errorMessage": error}
data = self._get(SEARCH_ENDPOINT, dict({"token": token, "q": query.strip()}, **loc))
if "_error" in data:
return self._as_error(data)
results = [r for r in data.get("results") or [] if isinstance(r, dict)]
return {
"status": "success",
"query": query.strip(),
"store": data.get("store") or {},
"total_results": data.get("total_results"),
"total_pages": data.get("total_pages"),
"results": results,
"credits_used": CREDITS_PER_CALL,
}
# -------------------------------------------------------------- product
def product(self, ref: str, location: Optional[str] = None) -> Dict[str, Any]:
token = self._resolve_token()
if not token:
return self._missing_token()
loc, error = parse_location(location)
if error:
return {"status": "error", "errorMessage": error}
resolved = self._resolve_item(token, ref, loc)
if "error" in resolved:
result = {"status": "error", "errorMessage": resolved["error"]}
if resolved.get("auth"):
result["hint"] = f"Check your token or create one at {SCRAPE_DO_SIGNUP_URL}"
return result
data = self._get(PRODUCT_ENDPOINT, dict({"token": token, "item": resolved["item_id"]}, **loc))
if "_error" in data:
return self._as_error(data)
product = data.get("product")
if not isinstance(product, dict) or not product:
return {"status": "error", "errorMessage": "Walmart returned no details for this product."}
result: Dict[str, Any] = {
"status": "success",
"store": data.get("store") or {},
"product": product,
"credits_used": resolved["credits"] + CREDITS_PER_CALL,
}
if resolved.get("note"):
result["note"] = resolved["note"]
return result
# -------------------------------------------------------------- compare
def compare(self, ref: str, locations: Optional[str] = None) -> Dict[str, Any]:
token = self._resolve_token()
if not token:
return self._missing_token()
raw = split_locations(locations) or list(DEFAULT_COMPARE_LOCATIONS)
warnings: List[str] = []
if len(raw) > MAX_COMPARE_LOCATIONS:
warnings.append(f"Only the first {MAX_COMPARE_LOCATIONS} locations were checked.")
raw = raw[:MAX_COMPARE_LOCATIONS]
parsed: List[Tuple[str, Dict[str, str]]] = []
for item in raw:
loc, error = parse_location(item)
if error:
warnings.append(error)
else:
parsed.append((item, loc))
if not parsed:
return {"status": "error", "errorMessage": "Enter at least one 5-digit ZIP code or store number."}
resolved = self._resolve_item(token, ref, parsed[0][1])
if "error" in resolved:
result = {"status": "error", "errorMessage": resolved["error"]}
if resolved.get("auth"):
result["hint"] = f"Check your token or create one at {SCRAPE_DO_SIGNUP_URL}"
return result
item_id = resolved["item_id"]
def fetch(entry: Tuple[str, Dict[str, str]]) -> Dict[str, Any]:
label, loc = entry
data = self._get(PRODUCT_ENDPOINT, dict({"token": token, "item": item_id}, **loc))
return {"label": label, "data": data}
# Three at a time keeps free-plan accounts inside their concurrency limit.
with ThreadPoolExecutor(max_workers=3) as pool:
fetched = list(pool.map(fetch, parsed))
credits = resolved["credits"]
rows: List[Dict[str, Any]] = []
product_info: Dict[str, Any] = {}
auth_failed = False
for entry in fetched:
data = entry["data"]
if "_error" in data:
auth_failed = auth_failed or bool(data.get("_auth"))
rows.append({"location": entry["label"], "error": data["_error"]})
continue
credits += CREDITS_PER_CALL
product = data.get("product") or {}
store = data.get("store") or {}
if product and not product_info:
product_info = {
key: product.get(key)
for key in ("item_id", "name", "brand", "url", "rating", "reviews_count", "category")
}
images = product.get("images") or []
product_info["image"] = images[0] if images else None
rows.append(
{
"location": entry["label"],
"store_id": store.get("store_id"),
"city": store.get("city"),
"state": store.get("state"),
"postal_code": store.get("postal_code"),
"price": product.get("price"),
"list_price": product.get("list_price"),
"currency": product.get("currency"),
"availability": product.get("availability"),
"fulfillment": product.get("fulfillment") or [],
"seller": product.get("seller"),
}
)
if all("error" in row for row in rows):
result = {"status": "error", "errorMessage": rows[0]["error"]}
if auth_failed:
result["hint"] = f"Check your token or create one at {SCRAPE_DO_SIGNUP_URL}"
return result
priced = [r for r in rows if isinstance(r.get("price"), (int, float))]
result: Dict[str, Any] = {
"status": "success",
"item_id": item_id,
"product": product_info,
"stores": rows,
"credits_used": credits,
}
if priced:
cheapest = min(priced, key=lambda r: r["price"])
dearest = max(priced, key=lambda r: r["price"])
result["cheapest"] = cheapest
result["most_expensive"] = dearest
result["price_gap"] = round(dearest["price"] - cheapest["price"], 2)
if resolved.get("note"):
result["note"] = resolved["note"]
if warnings:
result["warnings"] = warnings
return result
# -------------------------------------------------------------- forward
def forward(self, mode: str, query: str, location: Optional[str] = None) -> str:
mode_key = (mode or "").strip().lower()
if mode_key in ("search", "find"):
result = self.search(query, location)
elif mode_key in ("product", "details", "item"):
result = self.product(query, location)
elif mode_key in ("compare", "prices", "price_compare"):
result = self.compare(query, location)
else:
result = {"status": "error", "errorMessage": "mode must be 'search', 'product' or 'compare'."}
return json.dumps(result, ensure_ascii=False, indent=2)