troubleshooting costco header extraction

pm/scrape-giant.org

@@ -125,6 +125,14 @@ request-context: appId=cid-v1:75750625-0c81-4f08-9f5d-ce4f73198e54
 X-Firefox-Spdy: h2
 
 * costco requests
+- localstorage idToken has the auth token, but needs "Bearer " prepended
+- localstorage clientID has the COSTCO_X_WCS_CLIENTID
+- I don't see the client_identifier uuid anywhere.
+
+we will pull from .env first (may have to hardcode)
+then overwrite with session data (token)
+hopefully this doesnt change.
+
 ** warehouse
 *** POST
 https://ecom-api.costco.com/ebusiness/order/v1/orders/graphql

retailer_sessions.py

@@ -1,4 +1,6 @@
+import os
 from dataclasses import dataclass
+from dotenv import load_dotenv
 
 from browser_session import (
     find_json_storage_value,
@@ -32,29 +34,60 @@ def load_giant_session(browser="firefox", profile_dir=None):
     )
     return RetailerSession(cookies=context.cookies, headers={})
 
-
 def load_costco_session(browser="firefox", profile_dir=None):
+    load_dotenv()
+
+    headers = {
+        "costco-x-authorization": os.getenv("COSTCO_X_AUTHORIZATION", "").strip(),
+        "costco-x-wcs-clientId": os.getenv("COSTCO_WCS_CLIENT_ID", "").strip(),
+        "client-identifier": os.getenv("COSTCO_CLIENT_IDENTIFIER", "").strip(),
+    }
+
     context = load_browser_context(
         browser=browser,
         domain_name=".costco.com",
-        storage_origins=COSTCO_STORAGE_ORIGINS,
+        storage_origins=["costco.com"],
         profile_dir=profile_dir,
     )
-    headers = extract_costco_headers(context.storage_entries)
-    missing = [
-        header_name for header_name, value in headers.items() if not value
-    ]
-    if missing:
-        available_keys = ", ".join(
-            list_storage_keys(context.storage_entries, COSTCO_STORAGE_ORIGINS)
-        )
-        raise ValueError(
-            "missing Costco browser session headers: "
-            f"{', '.join(missing)}. "
-            f"Available Costco storage keys: {available_keys or '(none)'}"
+
+    storage = {entry.key: entry.value for entry in context.storage_entries}
+
+    id_token = storage.get("idToken", "").strip()
+    client_id = storage.get("clientID", "").strip()
+
+    if id_token:
+        headers["costco-x-authorization"] = (
+            id_token if id_token.startswith("Bearer ") else f"Bearer {id_token}"
         )
+    if client_id:
+        headers["costco-x-wcs-clientId"] = client_id
+
+    headers = {k: v for k, v in headers.items() if v}
+
     return RetailerSession(cookies=context.cookies, headers=headers)
 
+#def load_costco_session(browser="firefox", profile_dir=None):
+#    context = load_browser_context(
+#        browser=browser,
+#        domain_name=".costco.com",
+#        storage_origins=COSTCO_STORAGE_ORIGINS,
+#        profile_dir=profile_dir,
+#    )
+#    headers = extract_costco_headers(context.storage_entries)
+#    missing = [
+#        header_name for header_name, value in headers.items() if not value
+#    ]
+#    if missing:
+#        available_keys = ", ".join(
+#            list_storage_keys(context.storage_entries, COSTCO_STORAGE_ORIGINS)
+#        )
+#        raise ValueError(
+#            "missing Costco browser session headers: "
+#            f"{', '.join(missing)}. "
+#            f"Available Costco storage keys: {available_keys or '(none)'}"
+#        )
+#    return RetailerSession(cookies=context.cookies, headers=headers)
+
 
 def extract_costco_headers(storage_entries):
     headers = {}

scrape_costco.py

@@ -1,3 +1,4 @@
+import os
 import csv
 import json
 import time
@@ -5,7 +6,7 @@ import re
 from calendar import monthrange
 from datetime import datetime, timedelta
 from pathlib import Path
-
+from dotenv import load_dotenv
 import click
 from curl_cffi import requests
 
@@ -23,7 +24,7 @@ query receiptsWithCounts($startDate: String!, $endDate: String!, $documentType:
     gasAndCarWash
     receipts {
       warehouseName
-      receiptType
+      receiptType 
       documentType
       transactionDateTime
       transactionBarcode
@@ -225,11 +226,11 @@ def build_headers(auth_headers):
     headers.update(auth_headers)
     return headers
 
-def build_session(profile_dir=None):
-    retailer_session = load_costco_session(profile_dir=profile_dir)
+def build_session(retailer_session):
     session = requests.Session()
     session.cookies.update(retailer_session.cookies)
-    session.headers.update(build_headers(retailer_session.headers))
+    session.headers.update(build_headers())
+    session.headers.update(retailer_session.headers)
     return session
 
 
@@ -593,23 +594,28 @@ def main(
 ):
     outdir = Path(outdir)
     raw_dir = outdir / "raw"
-    try:
-        session = build_session(profile_dir=firefox_profile_dir)
-    except Exception as exc:
-        if firefox_profile_dir:
-            raise click.ClickException(
-                f"failed to load Costco browser session: {exc}"
-            ) from exc
-        prompted_profile = click.prompt(
-            "Firefox profile dir",
-            type=click.Path(exists=True, file_okay=False, path_type=Path),
+    if firefox_profile_dir is None:
+        firefox_profile_dir = next(
+            (Path(os.getenv("APPDATA")) / "Mozilla" / "Firefox" / "Profiles").iterdir()
         )
-        try:
-            session = build_session(profile_dir=prompted_profile)
-        except Exception as prompt_exc:
-            raise click.ClickException(
-                f"failed to load Costco browser session: {prompt_exc}"
-            ) from prompt_exc
+    try:
+        retailer_session = load_costco_session(
+            browser="firefox",
+            profile_dir=firefox_profile_dir,
+        )
+        click.echo(
+            "session bootstrap: "
+            f"cookies={bool(retailer_session.cookies)}, "
+            f"authorization={'costco-x-authorization' in retailer_session.headers}, "
+            f"client_id={'costco-x-wcs-clientId' in retailer_session.headers}, "
+            f"client_identifier={'client-identifier' in retailer_session.headers}"
+        )
+        session = build_session(retailer_session)
+    except Exception as exc:
+        raise click.ClickException(
+            f"failed to load Costco browser session: {exc}"
+        ) from exc
+
     start_date, end_date = resolve_date_range(months_back)
 
     summary_payload, request_metadata = fetch_summary_windows(
@@ -645,4 +651,5 @@ def main(
 
 if __name__ == "__main__":
     main()
+