A structured JSON database of 5,000+ foods, dishes, and ingredients, with calories and macronutrients, in French and English.

Choose the language / Choisissez votre langue :

Live demo: https://chpalitom09-bot.github.io/Open-food-calories/

• Overview
• Installation (pip package)
• Repository structure
• Data format (schema v2)
• Data quality report
• Using the raw JSON
• Command-line tool (repository scripts)
• Local REST API
• Contributing
• License

• Coverage: 5,037 entries
• Structure: 100% schema-validated, bilingual (French / English)
• Accuracy: cross-referenced against CIQUAL and USDA reference values

The database is also published as a standalone Python package. This is the recommended way to use it: no need to clone the repository, the package fetches the dataset itself and caches it locally.

pip install open-food-calories

Two equivalent commands are installed: open-food-calories and the shorter alias ofc.

open-food-calories search pizza

On first use, the package downloads data/Open-food-calories.json from this repository and caches it in ~/.cache/open-food-calories/ for 24 hours. If the network is unavailable, it falls back to the last cached copy, or to a snapshot bundled inside the package.

open-food-calories search pizza
open-food-calories search poulet --lang fr --category meat
open-food-calories search "" --category fruits --sort kcal --desc --limit 20

open-food-calories get riz-blanc-cuit
open-food-calories get riz-blanc-cuit --json

Looks up a single entry by its exact id.

open-food-calories calc riz-blanc-cuit --grams 250
open-food-calories calc pizza-part-moyenne --units 2
open-food-calories calc poulet --lang fr --grams 150 --json

Computes calories and macronutrients for a given quantity of a food, identified by id or by a name search (the first match is used).

open-food-calories compare riz-blanc-cru riz-blanc-cuit
open-food-calories compare "blanc de poulet" "cuisse de poulet" --lang fr

Displays calories and macronutrients for two or more foods side by side, as a table (or as JSON with --json). Each argument can be an id or a name search.

open-food-calories random
open-food-calories random --category fruits
open-food-calories random --type liquid --lang en

Returns one random entry, optionally restricted by --category and --type.

open-food-calories categories

Lists every category with its entry count, sorted from the largest to the smallest.

open-food-calories stats

Prints the total entry count, macronutrient coverage, and the breakdown by confidence level.

open-food-calories cache info
open-food-calories cache clear

cache info shows where the local cache lives, its size, its age, and whether it is still considered fresh. cache clear deletes it, forcing a fresh download on the next command.

.
├── data/
│   ├── Open-food-calories.json          # current dataset (schema v2)
│   └── Open-food-calories-10_03_26.json # legacy snapshot (schema v1)
├── data-schema.json                     # JSON Schema for the v2 format
├── CLI/
│   ├── ofc_cli.py                       # command-line search tool
│   └── ofc_server.py                    # local REST API server
├── examples/
│   ├── app.py                           # Streamlit demo app
│   └── chek_data.py                     # duplicate-check script
├── index.html                           # static web app (GitHub Pages)
├── API.md                               # API reference
├── CONTRIBUTING.md                      # contribution guidelines
└── LICENSE

The dataset that is actively maintained and used by index.html and every tool in this repository is data/Open-food-calories.json. The -10_03_26 file is kept only as a historical snapshot of the older, flat schema and should not be used for new integrations.

Every entry in data/Open-food-calories.json follows the schema defined in data-schema.json.

{
  "id": "riz-blanc-cuit",
  "name": {
    "fr": "Riz blanc (cuit)",
    "en": "White rice (cooked)"
  },
  "category": "grains",
  "state": "cooked",
  "type": "solid",
  "kcal_per_100g": 130,
  "protein_g": 2.7,
  "fat_g": 0.3,
  "carbs_g": 28.0,
  "fiber_g": null,
  "sugars_g": null,
  "saturated_fat_g": null,
  "salt_g": null,
  "alcohol_g": null,
  "weight_per_unit": null,
  "emoji": "rice",
  "source": "CIQUAL",
  "confidence": "high"
}

This schema is a breaking change from the original flat format (name as a plain string, english_name, no id, no macronutrients). Any integration built against the old shape needs to be updated to read name.fr / name.en and, where relevant, the new macronutrient fields.

Total entries: 5,037

Macronutrient coverage:

• Inherited from the original file: 1,010
• Added from CIQUAL/USDA: 121
• Still missing (null): 3,906

By confidence level:

• unknown: 3,906
• medium: 1,010
• high: 121

By category:

Known issues:

• Near-duplicate French names: 776 entries share a name or a close variant (for example multiple cheddar, mozzarella, or oil entries with numeric suffixes). These are distinct products, not errors, but deduplication tooling should account for this.
• Atwater-inconsistent entries: 119 entries where the declared calorie value diverges noticeably from the value calculated from protein/fat/carb grams (for example carotte: declared 41, calculated 31).
• Alcohol not accounted for: 69 alcoholic beverages where the calculated calories do not include the contribution of alcohol_g, producing a large gap between declared and calculated values (for example vodka: declared 231, calculated 0).
• Out-of-range value: 1 entry (epice-mastic-larme-de-mastic) at exactly 1000 kcal/100g, at the edge of the schema's allowed range.

These are tracked for cleanup; contributions that correct any of the above are welcome, see Contributing.

const response = await fetch('./data/Open-food-calories.json');
const foods = await response.json();

const item = foods.find(f => f.name.en === 'White rice (cooked)');

if (item) {
  console.log(`${item.name.fr} has ${item.kcal_per_100g} kcal per 100g.`);
}

const weight = 250;
const totalKcal = (item.kcal_per_100g * weight) / 100;
console.log(`Total: ${totalKcal} kcal for ${weight}g`);

import json

with open('data/Open-food-calories.json', 'r', encoding='utf-8') as f:
    data = json.load(f)

search_term = 'oeuf'
results = [food for food in data if search_term.lower() in food['name']['fr'].lower()]

for r in results:
    if r['weight_per_unit']:
        kcal_unit = (r['kcal_per_100g'] * r['weight_per_unit']) / 100
        print(f"{r['name']['fr']}: {kcal_unit} kcal per unit")
    else:
        print(f"{r['name']['fr']}: {r['kcal_per_100g']} kcal per 100g")

weight_per_unit is null when the food is normally measured by weight (rice, flour), and set to the average weight in grams when the food is normally counted by piece (an egg, an apple). An apple with kcal_per_100g: 52 and weight_per_unit: 150 amounts to 78 kcal per apple.

For working directly against a local checkout of this repository (for example while contributing data), CLI/ofc_cli.py is a dependency-free, standalone version of the search command that does not require installing the pip package. It resolves data/Open-food-calories.json relative to its own location, so it works whether it is run from the repository root or from inside CLI/.

python3 CLI/ofc_cli.py search pizza
python3 CLI/ofc_cli.py search poulet --lang fr --category meat
python3 CLI/ofc_cli.py get riz-blanc-cuit
python3 CLI/ofc_cli.py get riz-blanc-cuit --json
python3 CLI/ofc_cli.py categories
python3 CLI/ofc_cli.py stats

Options for search:

CLI/ofc_server.py starts a local HTTP server exposing the dataset, using only the Python standard library, no installation required.

python3 CLI/ofc_server.py --port 8000

Examples:

curl "http://localhost:8000/search?q=pizza&category=prepared"
curl "http://localhost:8000/food/riz-blanc-cuit"
curl "http://localhost:8000/categories"

Responses are JSON, served with Access-Control-Allow-Origin: *, so the server can be queried directly from a browser-based prototype during development.

The goal is to grow this dataset well beyond 5,000 entries while keeping it clean. Contributions of new entries, corrections to the issues listed in Data quality report, and improvements to the tooling are all welcome.

All data changes are made in data/Open-food-calories.json, following the schema described above and validated by data-schema.json. Before opening a pull request:

• Check for an existing entry with the same or a very similar name before adding a new one.
• Provide both name.fr and name.en.
• Set state and type accurately; use liquid for drinks and oils, unit for foods normally counted by piece, solid otherwise.
• Set weight_per_unit when the food is normally counted by piece, leave it null otherwise.
• Set source and confidence honestly; use estimated and low/medium rather than guessing a high-confidence value.
• Validate the JSON syntax before committing.
• Run python3 CLI/ofc_cli.py stats against the updated file to sanity-check totals and confidence levels before opening the pull request.

See CONTRIBUTING.md for the full process.

This project is licensed under the MIT License, see LICENSE for details.

Démo en direct : https://chpalitom09-bot.github.io/Open-food-calories/

• Aperçu
• Installation (paquet pip)
• Structure du dépôt
• Format des données (schéma v2)
• Rapport sur la qualité des données
• Utilisation du JSON brut
• Outil en ligne de commande (scripts du dépôt)
• API REST locale
• Contribution
• Licence

Ce projet est sous la licence MIT, voir LICENSE pour des détails.