catalog-api

Expose a product recommandation API, based on Google Vision API

List package used

• ORM : sequelize pros : faster to develop cons : could be slower with a large dataset
• DB : Postgres
• Web Framework : express
• Color Proximity Calcul : colour-proximity
• image-downloader

How to use it

Step 0

Clone this repo and npm install

Step 1 - loading csv data

prerequisite :

• Node installed
• CSV file with comma delimiter
• Postgres DB available, you can install it quickly with Docker docker run --name catalog-postgres -e POSTGRES_PASSWORD=kT6xMdTqahVCvYAN -e POSTGRES_USER=catalog-user -e POSTGRES_DB=catalog-db -p 5432:5432 -d postgres
• Configure your database information on site-config.js

Launch this command line to load your data :
node src/cmd/step1-loading.js path/to/file.csv

Exemple :
node src/cmd/step1-loading.js data/products_eb_test_technique.csv

Step 2 - Identify color for each products

prerequisite :

• A key authentification configured on your env : see more

Launch this command line to populate database with Metadata from Google Vision API :
node src/cmd/step2-identifyColors.js

TIPS : a Backup with Data from step 1 and 2 are available on data/backup-data.sql

Step 3 - API ENDPOINTS

Launch this command line :
node index.js

Go to : http://localhost:3000/api/recommandation/[productid]
Example : http://localhost:3000/api/recommandation/L1212-00-001

Also available : http://localhost:3000/api/products in order to list every products

What's next

• Used Score of the dominant color to exclude some erroneous classification
• Improve API robustness (configure CORS, authentication...)
• Manage Google Vision API error without downloaded image
• Add some Unit Test (nevertheless, some manual and basic functionnal test are possible with catalog-front)