# Czech food products

**Search Czech food products and inspect candidate recipe ingredient matches.** The catalogue combines Rohlík product facts with inferred food attributes and ingredient candidates from international TheMealDB recipes. The published table contains 13,503 unique products and 790 observed recipes. Toprecepty is registered for the next build but is not included in the published result. Three product…

Czech food products is an open, machine-readable dataset published on Mostly Right as immutable Parquet versions, with a documented column schema, a JSON query API and a direct download. It is free to read and open to AI agents.

## Facts

- Slug: `czech-food-products`
- Canonical page: https://mostlyright.md/datasets/czech-food-products
- Publisher: mostlyrightmd (https://mostlyright.md/publishers/mostlyrightmd)
- Published: 2026-10-11T23:16:53.028Z
- Tables: 1
- Formats: Apache Parquet (download), JSON (query and sample)
- Cost: free to read; querying and downloading need a free account
- Metadata API: `GET https://mostlyright.md/api/v2/public/datasets/czech-food-products`
- Topics: czech republic, groceries, barcodes, product catalogue, ingredients

## How to use it

1. Read a table's schema — no account needed.
2. Fetch its published sample rows — no account needed.
3. Sign in at https://mostlyright.md/login, connect this dataset to a workspace on its page, and create an `mr_use_` key.
4. Query bounded JSON, or download the complete Parquet snapshot and read it locally.

Agents can do all of this over MCP at `https://mostlyright.md/api/mcp` — searching, reading schemas and sampling rows work with no credential at all.

## Table: rohlik_catalogue_pages

- Table id: `a50f498d-1692-43ee-be62-f807b57e69a5`
- Slug: `rohlik-catalogue-pages`
- Immutable version: `4dc0fcf5-e3f9-5683-a579-117405c0515d`
- Schema: `GET https://mostlyright.md/api/v2/public/tables/a50f498d-1692-43ee-be62-f807b57e69a5/schema`
- Sample rows, no key: `GET https://mostlyright.md/api/v2/public/tables/a50f498d-1692-43ee-be62-f807b57e69a5/preview`
- Bounded JSON query: `POST https://mostlyright.md/api/v2/public/tables/a50f498d-1692-43ee-be62-f807b57e69a5/query` with `x-api-key: mr_use_…`
- Full Parquet snapshot: `GET https://mostlyright.md/api/v2/public/tables/a50f498d-1692-43ee-be62-f807b57e69a5/current` with `x-api-key: mr_use_…` (application/vnd.apache.parquet)

### Columns (32)

| Column | Type | Description |
| --- | --- | --- |
| `product_id` | string | Stable Rohlik URL numeric identifier, not a GTIN. |
| `product_name` | string | Original target product heading. |
| `standardized_name_cs` | string | Controlled ingredient identity, cut and form; unknown preserves original title. Does not replace original commercial name. |
| `ingredient_key` | string | Clef controlled product identity; uncertain classifications stay unknown. |
| `food_scope` | string | food, beverage, non_food or unknown; unknown rows require review. |
| `cut` | string | Explicit meat cut classification, else unknown/not_applicable. |
| `form` | string | Physical form classification. |
| `preparation_state` | string | Preparation state classification; missing facts are unknown. |
| `storage_state` | string | Storage state classification from title only. |
| `classification_confidence` | number | Lowest confidence among identity and food scope decisions; model probability, not calibrated accuracy. |
| `classification_json` | string | All raw Clef decisions/probabilities for audit. |
| `classification_model` | string | Model actually served by platform. |
| `taxonomy_version` | string | Controlled ingredient vocabulary version. |
| `review_status` | string | Unknown or low-confidence labels remain flagged for review. |
| `pack_quantity_text` | string | Target heading-adjacent quantity text; extraction is conservative and may be missing. |
| `description_text` | string | Target product description section from Popis to nutrition/ingredients/footer boundary; layout-sensitive extraction. |
| `ingredients_text` | string | Target ingredients section when recognized; missing means unknown. |
| `allergens_text` | string | Original allergen and traces section when recognized; absence is never inferred. |
| `nutrition_text` | string | Original nutrition section including original basis/units; not normalized numeric facts. |
| `page_text` | string | Original full captured page including unrelated recommendations, retained for extraction audit only. |
| `source_url` | string | Exact captured Rohlik URL. |
| `fetched_at` | timestamp_utc | UTC acquisition timestamp. |
| `snapshot_date` | date | UTC acquisition date. |
| `classification_input_hash` | string | Hash of title used in classifier request. |
| `search_text` | string | Standardized identity and original title for retrieval; excludes recommendations. |
| `image_url` | string | Observed primary product image URL; gallery first image with og:image fallback. |
| `image_urls` | string | JSON array of observed product gallery URLs in page order; primary metadata fallback when gallery is absent. Null if no image is observed. More than 20 gallery… |
| `recipe_matches` | string | Ordered inferred recipe ingredient candidates; original ingredient/measure, constraints and evidence retained. No quantity conversion, serving normalization, a… |
| `matched_recipe_count` | integer | Distinct acquired recipes having an ingredient candidate for this product. |
| `matched_ingredient_count` | integer | Recipe ingredient positions having this product as a candidate; many products may match a position. |
| `recipe_match_coverage` | string | Observed recipe/ingredient counts; does not attest acquisition completeness. Includes, classified positions, uncertain positions and matched/unmatched counts;… |
| `recipe_matching_integrity_verified` | integer | 1 only when every observed nonempty recipe ingredient position has exactly one pinned classifier answer and product classifier models match the same pin. |

## Citing this dataset

Cite the dataset by name and canonical URL (https://mostlyright.md/datasets/czech-food-products), the table, and the immutable `version_id` the rows came from. A query response carries `table.version_id` and `table.content_digest`; a download carries a strong ETag and a `Content-Digest` header.