updated readme with Review steps

This commit is contained in:
ben
2026-03-17 09:14:14 -04:00
parent 91bfd3597e
commit 7f8c3ed8eb
3 changed files with 147 additions and 191 deletions

273
README.md
View File

@@ -1,227 +1,118 @@
# scrape-giant
Small grocery-history pipeline for Giant and Costco receipt data.
Small CLI pipeline for pulling purchase history from Giant and Costco, enriching line items, and building a reviewable cross-retailer purchase dataset.
This repo is still a manual, stepwise pipeline. There is no single orchestrator
script yet. Each stage is run directly, and later stages depend on files
produced by earlier stages.
There is no one-shot runner yet. Today, you run the scripts step by step from the terminal.
## What The Project Does
## What It Does
The current flow is:
- `scrape_giant.py`: download Giant orders and items
- `enrich_giant.py`: normalize Giant line items
- `scrape_costco.py`: download Costco orders and items
- `enrich_costco.py`: normalize Costco line items
- `build_purchases.py`: combine retailer outputs into one purchase table
- `review_products.py`: review unresolved product matches in the terminal
1. acquire raw Giant receipt/history data
2. enrich Giant line items into a shared enriched-item schema
3. acquire raw Costco receipt data
4. enrich Costco line items into the same shared enriched-item schema
5. build observed-product, review, and canonical-product layers
6. validate that Giant and Costco can flow through the same downstream model
## Requirements
Raw retailer JSON remains the source of truth.
- Python 3.10+
- Firefox installed with active Giant and Costco sessions
## Current Scripts
- `scrape_giant.py`
Fetch Giant in-store history and order detail payloads from an active Firefox
session.
- `scrape_costco.py`
Fetch Costco receipt summary/detail payloads from an active Firefox session.
Costco currently prefers `.env` header values first, then falls back to exact
Firefox local-storage values for session auth.
- `enrich_giant.py`
Parse Giant raw order JSON into `giant_output/items_enriched.csv`.
- `enrich_costco.py`
Parse Costco raw receipt JSON into `costco_output/items_enriched.csv`.
- `build_observed_products.py`
Build retailer-facing observed products from enriched rows.
- `build_review_queue.py`
Build a manual review queue for low-confidence or unresolved observed
products.
- `build_canonical_layer.py`
Build shared canonical products and observed-to-canonical links.
- `validate_cross_retailer_flow.py`
Write a proof/check output showing that Giant and Costco can meet in the same
downstream model.
## Manual Pipeline
Run these from the repo root with the venv active, or call them through
`./venv/bin/python`.
### 1. Acquire Giant raw data
## Install
```bash
./venv/bin/python scrape_giant.py
python -m venv venv
./venv/scripts/activate
pip install -r requirements.txt
```
Inputs:
- active Firefox session for `giantfood.com`
- `GIANT_USER_ID` and `GIANT_LOYALTY_NUMBER` from `.env`, shell env, or prompt
## Optional `.env`
Outputs:
- `giant_output/raw/history.json`
- `giant_output/raw/<order_id>.json`
Current version works best with `.env` in the project root. The scraper will prompt for these values if they are not found in the current browser session.
- `scrape_giant` prompts if `GIANT_USER_ID` or `GIANT_LOYALTY_NUMBER` is missing.
- `scrape_costco` tries `.env` first, then Firefox local storage for session-backed values; `COSTCO_CLIENT_IDENTIFIER` should still be set explicitly.
```env
GIANT_USER_ID=...
GIANT_LOYALTY_NUMBER=...
# Costco can use these if present, but it can also pull session values from Firefox.
COSTCO_X_AUTHORIZATION=...
COSTCO_X_WCS_CLIENTID=...
COSTCO_CLIENT_IDENTIFIER=...
```
## Run Order
Run the pipeline in this order:
```bash
python scrape_giant.py
python enrich_giant.py
python scrape_costco.py
python enrich_costco.py
python build_purchases.py
python review_products.py
python build_purchases.py
```
Why run `build_purchases.py` twice:
- first pass builds the current combined dataset and review queue inputs
- `review_products.py` writes durable review decisions
- second pass reapplies those decisions into the purchase output
If you only want to refresh the queue without reviewing interactively:
```bash
python review_products.py --refresh-only
```
## Key Outputs
Giant:
- `giant_output/orders.csv`
- `giant_output/items.csv`
### 2. Enrich Giant data
```bash
./venv/bin/python enrich_giant.py
```
Input:
- `giant_output/raw/*.json`
Output:
- `giant_output/items_enriched.csv`
### 3. Acquire Costco raw data
```bash
./venv/bin/python scrape_costco.py
```
Optional useful flags:
```bash
./venv/bin/python scrape_costco.py --months-back 36
./venv/bin/python scrape_costco.py --firefox-profile-dir "C:\\Users\\you\\AppData\\Roaming\\Mozilla\\Firefox\\Profiles\\xxxx.default-release"
```
Inputs:
- active Firefox session for `costco.com`
- optional `.env` values:
- `COSTCO_X_AUTHORIZATION`
- `COSTCO_X_WCS_CLIENTID`
- `COSTCO_CLIENT_IDENTIFIER`
- if `COSTCO_X_AUTHORIZATION` is absent, the script falls back to exact Firefox
local-storage values:
- `idToken` -> sent as `Bearer <idToken>`
- `clientID` -> used as `costco-x-wcs-clientId` when env is blank
Outputs:
- `costco_output/raw/summary.json`
- `costco_output/raw/summary_requests.json`
- `costco_output/raw/<receipt_id>-<timestamp>.json`
Costco:
- `costco_output/orders.csv`
- `costco_output/items.csv`
### 4. Enrich Costco data
```bash
./venv/bin/python enrich_costco.py
```
Input:
- `costco_output/raw/*.json`
Output:
- `costco_output/items_enriched.csv`
### 5. Build shared downstream layers
Combined:
- `combined_output/purchases.csv`
- `combined_output/review_queue.csv`
- `combined_output/review_resolutions.csv`
- `combined_output/canonical_catalog.csv`
- `combined_output/product_links.csv`
- `combined_output/comparison_examples.csv`
```bash
./venv/bin/python build_observed_products.py
./venv/bin/python build_review_queue.py
./venv/bin/python build_canonical_layer.py
```
## Review Workflow
These scripts consume the enriched item files and generate the downstream
product-model outputs.
`review_products.py` is the manual cleanup step for unresolved or weakly unified items.
Current outputs on disk:
In the terminal, you can:
- link an item to an existing canonical product
- create a new canonical product
- exclude an item
- skip it for later
- retailer-facing:
- `giant_output/products_observed.csv`
- `giant_output/review_queue.csv`
- `giant_output/products_canonical.csv`
- `giant_output/product_links.csv`
- cross-retailer proof/check output:
- `combined_output/products_observed.csv`
- `combined_output/products_canonical.csv`
- `combined_output/product_links.csv`
- `combined_output/proof_examples.csv`
### 6. Validate cross-retailer flow
```bash
./venv/bin/python validate_cross_retailer_flow.py
```
This is a proof/check step, not the main acquisition path.
## Inputs And Outputs By Directory
### `giant_output/`
Inputs to this layer:
- Firefox session data for Giant
- Giant raw JSON payloads
Generated files:
- `raw/history.json`
- `raw/<order_id>.json`
- `orders.csv`
- `items.csv`
- `items_enriched.csv`
- `products_observed.csv`
- `review_queue.csv`
- `products_canonical.csv`
- `product_links.csv`
### `costco_output/`
Inputs to this layer:
- Firefox session data for Costco
- Costco raw GraphQL receipt payloads
Generated files:
- `raw/summary.json`
- `raw/summary_requests.json`
- `raw/<receipt_id>-<timestamp>.json`
- `orders.csv`
- `items.csv`
- `items_enriched.csv`
### `combined_output/`
Generated by cross-retailer proof/build scripts:
- `products_observed.csv`
- `products_canonical.csv`
- `product_links.csv`
- `proof_examples.csv`
Those decisions are saved and reused on later runs.
## Notes
- The pipeline is intentionally simple and currently manual.
- Scraping is retailer-specific and fragile; downstream modeling is shared only
after enrichment.
- `summary_requests.json` is diagnostic metadata from Costco summary enumeration
and is not a receipt payload.
- `enrich_costco.py` skips that file and only parses receipt payloads.
- The repo may contain archived or sample output files under `archive/`; they
are not part of the active scrape path.
- This project is designed around fragile retailer scraping flows, so the code favors explicit retailer-specific steps over heavy abstraction.
- `scrape_giant.py` and `scrape_costco.py` are meant to work as standalone acquisition scripts.
- `validate_cross_retailer_flow.py` is a proof/check script, not a required production step.
## Verification
Run the full test suite with:
## Test
```bash
./venv/bin/python -m unittest discover -s tests
```
Useful one-off checks:
```bash
./venv/bin/python scrape_giant.py --help
./venv/bin/python scrape_costco.py --help
./venv/bin/python enrich_giant.py
./venv/bin/python enrich_costco.py
```
## Project Docs
- `pm/tasks.org`
- `pm/data-model.org`
- `pm/scrape-giant.org`
- `pm/tasks.org`: task tracking
- `pm/data-model.org`: current data model notes
- `pm/review-workflow.org`: review and resolution workflow