Difference between revisions of "Product Opener"
(new project to open source the code for Open Food Facts) |
(No difference)
|
Revision as of 10:40, 4 September 2014
Introduction
ProductOpener is the software used by Open Food Facts to create and maintain its collaborative database of food products in open data.
Open source
ProductOpener will be released under an open source licence so that it can be reused by other projets, for instance to open the data for other types of products.
The ProductOpener original code includes codes related to the original developer's other projects. Efforts are under way to clean, decouple and document the ProductOpener code so that it can be released as open source.
Components
ProductOpener Database
Product data and history
The data for each product is stored in a structured object (several levels of hashes and arrays). The data is stored on disk in individual files in Perl's binary format for objects (Storable). The history of each change to the product data is also saved in the same format. There is one binary file for each version.
Product data index
The current version of each product is stored in a MongoDB database. The database includes indexes and is used to search and display products. MongoDB allow storing structured objects as-is, so there is an exact correspondence between Perl's internal format, MongoDB's format and the JSON export of the API.
Product images
Images are saved on disk for each product. Original images uploaded through the web server or mobile applications are kept. Thumbnails are also generated for the final product image.
Aggregated data
Some of the data is periodically aggregated and saved in Perl binary files (e.g. average nutrition facts for products of each category).
Exported data
The data for all products is periodically exported in CSV and RDF formats.
ProductOpener Web Server
The web interface for searching, displaying, adding and editing products. (e.g. http://world.openfoodfacts.org for Open Food Facts)
The server is written in Perl. In production we use a light weight Apache web server for static files (e.g. images) that does reverse proxying to a modperl Apache server that dynamically generates the HTML pages.
The form to add and edit product uses Javascript and jquery.
In the backend, other libraries and software are used, such as Tesseract for optical character recognition for ingredients.
ProductOpener API
ProductOpener provides a JSON API to search products and read their data.
There is also a limited API to upload product images and edit some of the product data that is used by the mobile applications.
The API functionality is currently provided by the ProductOpener Web Server, but it may change in the future (e.g. at some point we could implement a full API for accessing (reading + writing) the ProductOpener database, and have the Web Server call the API instead of going directly to the database).
ProductOpener client applications
Apache Cordova mobile app for Android, iOS and Windows Phone
The current Open Food Facts mobile apps for Android, iOS and Windows Phone are developed with Apache Cordova (previously known as Phonegap) with the same HTML + Javascript (including jquery and jquerymobile) code base.
The Android app uses Moodstocks image recognizer and barcode scanner plugin. The iOS and Windows Phone apps use the BarcodeScanner plugin.
The code for the Open Food Facts phonegap app is already in open source:
- https://github.com/openfoodfacts/openfoodfacts-android/tree/master/phonegap/Open%20Food%20Facts (with the BarcodeScanner plugin)
- https://github.com/openfoodfacts/openfoodfacts-moodstocks (with the Moodstocks plugin)
Installation
(In progress.)