Romania Open Data Scraper - data.gov.ro API

dataset-row

One dataset with its full catalogue record: title, Romanian description, publishing institution, author and maintainer with emails, licence, categories, tags, user rating, portal URL, creation and update dates with a computed staleness flag, and every file nested with its format, size, checksum, dates, portal download URL and whether it is queryable through the DataStore.

resource-row

One downloadable file or service flattened to its own row: direct URL, portal download URL, format as published and as cleaned, MIME type, byte size, checksum, position and dates, plus whether it is loaded into the DataStore, carrying the identity, publisher and licence of the dataset it belongs to.

contact-row

One named author or maintainer with their email address and the institution they publish for, deduplicated across the run, linked to the dataset the contact was recorded on.

organization-row

One of the 180 publishing institutions on data.gov.ro: slug, official Romanian title, description, logo URL, dataset count, follower count, creation date and portal URL.

category-row

One of the 14 portal categories with its Romanian title, description, dataset count and portal URL.

tag-row

One tag out of the 1,761 in use, with the number of datasets carrying it and the portal URL that lists them. Respects the search filters, so it doubles as a tag frequency report for any slice of the catalogue.

license-row

One of the licences in use on the portal with its title, URL, open-data conformance flags and the number of datasets published under it.

showcase-row

One published showcase: the applications and analyses built on data.gov.ro data, with title, description, author, image and tags.

datastore-table-row

One of the 5,753 SQL tables the portal exposes through its DataStore, with its resource id, PostgreSQL object id, alias and the ready-to-call query URL.

archive-status

The portal archiver verdict for every file on one dataset: broken flag, status text, failure reason, failure count, first failure, last success, redirect target, content hash, ETag, archived byte size and the cached copy URL. Charged once per dataset, only when the portal actually holds archiver data.

openness-score

Tim Berners-Lee Five Stars openness grade for every file on one dataset, with the portal's reason text and the format it actually detected inside the file, which often differs from the declared one. Charged once per dataset, only when the portal has scored it.

view-stats

Total and recent page-view counts for one dataset, read from the portal tracking table. Costs one extra request per dataset and is charged only when the portal returns a tracking record.

activity-history

The last 25 recorded changes to one dataset with their timestamps and change types, plus the first and last change dates and a revision count. Costs one extra request per dataset and is charged only when the portal returns activity.

organization-profile

The publishing institution behind one row: description, logo, dataset count, follower count, member count and creation date. Cached per organization for the whole run, so a 500-dataset run from one ministry is charged once, not 500 times.

datastore-schema

Column names, PostgreSQL column types and the exact row count for the queryable tables on one dataset, up to 10 tables. Costs one extra request per table and is charged only when a table actually exists.

datastore-preview

Real data rows read out of the portal SQL tables on one dataset, up to 5 tables and up to 100 rows each, with the true total row count. Costs one extra request per table and is charged only when rows came back.

resource-views

The preview widgets the portal offers for the files on one dataset (table, map, graph, text, PDF), with their titles and view types, up to 10 files. Costs one extra request per file and is charged only when a widget exists.

link-check

A live HEAD request against every file URL on one dataset, up to 10 files, returning the HTTP status, content type, content length and a live broken count. This is the current truth, unlike the archiver status which is the portal cached verdict. Charged once per dataset, only when at least one file was reachable enough to answer.