About Food DB
What is Food DB?
Food DB is a sourced database for stable food identity and reusable facts such as weight, nutrition, seasonality, storage, substitutions, and shopping units.
What problem does it solve?
Everyday cooking requires decisions across names, forms, and quantities: how many grams are in one onion, whether firm and silken tofu should be grouped together, or whether dried and rehydrated hijiki share the same nutrition values.
Food DB links wording found in recipes to stable Food and Food Variant identities, then connects those identities to conversions, nutrition, shopping grouping, seasonality, storage, and substitutions.
What Food DB contains
Canonical food identity plus parts, product forms, and MEXT states such as raw, boiled, or dried when they require different values.
Conversions from pieces, tablespoons, cloves, and other units into grams and shopping-list identities.
Primarily MEXT-derived nutrient values per 100g, attached to the appropriate Food Variant.
Facts with their region, conditions, scope, and evidence kept explicit.
Aliases, scripts, product forms, and recipe wording mapped to canonical Food identities.
Sources, confidence, review status, claims, and conflicts are separated so uncertain values do not look final.
Food and Food Variant
Food is the stable identity people select in everyday cooking. Food Variant represents a type, part, or state that needs a distinct nutrient or conversion reference.
A spinach Food can have separate MEXT-linked raw, boiled, and oil-sauteed variants. These are source food identities, not arbitrary states in a cooking-process graph.
How to read nutrition values
Nutrition values are per 100g for the selected Food Variant. When a reviewed conversion exists, the site can also show values per piece, spoon, pack, or other practical unit.
These are reference values for standard foods, not chemical analyses of a finished dish. Actual values vary with products and cooking conditions.
Canonical data and public projection
The canonical source is Neon Postgres under food_db.*. This website is a static public projection containing only data that is suitable for publication.
- Candidate, under-review, or weakly sourced records may be hidden or labeled.
- Missing values are not silently replaced with zero or unsupported estimates.
- Conflicting source values remain separate claims until reviewed.
- Generated JSON and SQLite files are publication, evaluation, or operations surfaces—not canonical truth.
How data is updated
- Collect: record official sources, primary information, and recurring gaps found in real recipes
- Propose: create candidate identities, conversions, nutrients, storage rules, or other facts
- Review: check evidence, duplicates, applicability, and conflicts
- Promote: write reviewed facts to canonical tables
- Publish: project only eligible fields to the public website
Technical architecture
Canonical Foods, Variants, nutrients, conversions, sources, and claims.
Builds static pages from the publication projection.
Hosts and serves the public Food DB website.