Italy Open Data Scraper - dati.gov.it CKAN Catalog
Pricing
from $3.56 / 1,000 catalogue page scanneds
Scrape all 65,960 datasets from Italy's national open data portal dati.gov.it: full DCAT-AP_IT metadata, distributions, 399 publishers, licences and live link checks.
Free-text queries run against the whole catalogue (title, description, tags). Italian works best: bilancio, turismo, mobilita. Leave empty to sweep every dataset.
Specific dati.gov.it dataset pages, for example https://www.dati.gov.it/dataset/zone-turistiche-citta-metropolitana-di-firenze. When you supply these, only these datasets are scraped and the filters below are ignored.
CKAN dataset names or UUIDs, one per line. Same effect as Dataset URLs, without the URL.
Free users: limited to 10 items (preview). Paid users: up to 1,000,000.
EU data themes as dati.gov.it groups them. Counts on 2026-08-27: governo 31,664 | societa 9,833 | economia 6,609 | ambiente 5,860 | cultura 4,356 | regioni 3,824 | trasporti 2,162 | scienza 1,576 | salute 1,162 | agricoltura 795 | giustizia 763 | energia 297 | internazionali 56.
Publishing body slugs, one per line, for example regione-toscana, comune-di-milano, inps, ministero-dell-economia-e-delle-finanze. 447 are registered and 399 have at least one dataset; tick 'Organizations directory' below to export the full list with dataset counts.
Exact holder names, one per line, for example Comune di Milano or Istituto Nazionale Previdenza Sociale. The holder is the body legally responsible for the data and is often narrower than the publishing organization. 1,763 exist.
Exact DCAT publisher names, one per line. Populated on about 95 percent of datasets and often names a department rather than the whole body.
The regional or municipal portal a dataset was harvested from, exact title, for example Open Data Veneto or MEF OpenBDAP. 328 feed the national catalogue.
Keep only datasets that publish at least one file in these formats. Each choice expands to every spelling the catalogue really uses, because the index is case sensitive (GML matches 1,521 datasets and gml another 92).
Each choice expands to every spelling the portal stores for that licence. CC BY 4.0 alone is recorded four different ways and covers 61,448 datasets.
DCAT accrual periodicity. The portal stores 44 different spellings for about 20 real frequencies (NOT_PLANNED and NOTPLANNED are the same thing and are counted separately upstream); each choice here matches all of them.
ISO 639-3 code recorded on the dataset. Each choice matches every spelling the portal stores, including the multi-language ones: Italian 48,398 datasets, German 274, English 138, Spanish 5.
Keep only datasets that ship at least this many distributions. 9,413 datasets have 5 or more.
Only datasets whose catalogue record changed on or after this date (YYYY-MM-DD). 60,030 changed during 2026.
Only datasets whose catalogue record changed on or before this date (YYYY-MM-DD).
Only datasets first indexed on dati.gov.it on or after this date (YYYY-MM-DD).
Only datasets first indexed on dati.gov.it on or before this date (YYYY-MM-DD).
The publisher's own issue date, not the harvest date (YYYY-MM-DD). Populated on 92 percent of datasets.
Upper bound for the publisher's own issue date (YYYY-MM-DD).
Keep only the 6,208 datasets that carry a spatial extent, which is what fills the boundingBox column.
Keep only the 598 datasets loaded straight into the national portal instead of harvested from a regional catalogue.
Keep only the 4,026 datasets the portal itself flags as open. Note this flag is under-populated upstream: 61,448 datasets carry a CC BY 4.0 licence but only 4,026 have the flag set, so the Licences filter is usually the better tool.
Raw CKAN fq clause ANDed with everything above, for example res_format:PARQUET AND -organization:regione-toscana. A field name that does not exist returns zero rows rather than an error.
Only these four orders plus relevance actually work upstream; any other value is silently ignored by the portal and falls back to relevance.
Emit one extra row per downloadable file, with its own URL, format, size, checksum and licence. Datasets average 2.4 files each.
Emit one row for each of the 447 registered publishing bodies with its IPA code, certified email, website, Italian region and dataset count.
Emit the 13 EU data themes with their Italian names and dataset counts.
Emit the keyword tags used by the datasets that match your filters, with counts.
Emit every licence string in the matched result set with its normalised code and dataset count.
Emit all 92 format spellings with their canonical name, machine-readable flag and dataset count.
Emit the 1,763 bodies legally responsible for the data (titolare del dato), with dataset counts.
Emit the 1,452 DCAT publisher names in the matched result set, with dataset counts.
Emit the 328 regional and municipal portals that feed dati.gov.it, with how many datasets each one contributes.
Request each distribution and record the HTTP status, redirect target, content type and size. dati.gov.it harvests 328 catalogues and never revalidates the links; in a 193-file sample 14 percent were unreachable.
Download the first 64 KB of each distribution and report what the bytes really are, the CSV delimiter and the column headers. The declared format is often wrong: the first file checked in research was declared CSV and turned out to be a ZIP.
Add the publishing body's IPA code, contact email, telephone, website and region to every dataset row. Each organization is fetched once and reused.
