steadfetch/exceltur-indicadores-extractor

Extract published Exceltur tourism indicators from its public XLSX workbooks with Apify pay-per-event pricing and schema-validated output.

Data source: Exceltur

What it does

Exceltur publishes its tourism studies as XLSX workbooks through the public WordPress media API. The Actor lists those public files, downloads the selected workbooks and saves one validated dataset item per indicator series:

• the published indicator name, unit, column label and source note;
• every numeric (period, value) observation found in the workbook;
• the source file (mediaId, URL, publication timestamp and title);
• publisher attribution (source.publisher = "Exceltur", source URL and retrieval timestamp).

It reads both layouts used by Exceltur:

• Time series workbooks such as PIB-Anual-3T26.xlsx and PIB-Trimestral-3T26.xlsx (years, or year plus quarter, with one PIB Turistico column);
• Matrix workbooks such as Datos-IMPACTUR-jul26.xlsx (indicators as rows, autonomous communities and Spain as published columns).

Hidden rows are ignored. Non-numeric cells are never emitted and nothing is inferred: if the workbook does not publish a value, no observation is created.

Input

Empty input reads every XLSX workbook published in the last 90 days, up to 25 indicator series.

{
  "mediaTypes": ["xlsx"],
  "search": "pib",
  "sinceDate": "2026-07-23",
  "maxItems": 25
}

• mediaTypes: only ["xlsx"] is supported; any other value is rejected.
• search: optional WordPress media title filter (for example "pib"); empty means no filter.
• sinceDate: only workbooks published on or after this date (YYYY-MM-DD); empty means 90 days before the run.
• maxItems: 1 to 100 indicator series saved to the dataset (default 25; values above 100 are rejected).

Example output

Real output for PIB-Anual-3T26.xlsx (media id 51525) on 2026-10-21:

{
  "indicator": "ISTE. PIB Tur\u00edstico en constantes. Indicador Sint\u00e9tico del Turismo en Espa\u00f1a",
  "unit": "Tasa de variaci\u00f3n interanual (%)",
  "columnLabel": "PIB Tur\u00edstico",
  "sourceNote": "Fuente: Exceltur",
  "observations": [
    { "period": "2023", "value": 3.2 },
    { "period": "2024", "value": 3.1 },
    { "period": "2025", "value": 2.1 },
    { "period": "2026 (P)", "value": 2.7 }
  ],
  "file": {
    "mediaId": 51525,
    "url": "https://www.exceltur.org/wp-content/uploads/2026/10/PIB-Anual-3T26.xlsx",
    "publishedAt": "2026-10-06T16:27:22",
    "title": "PIB-Anual-3T26"
  },
  "source": {
    "publisher": "Exceltur",
    "license": "No content-use clause located on the publisher site; documented source review in docs/legal-gate.md",
    "url": "https://www.exceltur.org/wp-content/uploads/2026/10/PIB-Anual-3T26.xlsx",
    "retrievedAt": "2026-10-21T06:30:00.000Z"
  },
  "url": "https://www.exceltur.org/wp-content/uploads/2026/10/PIB-Anual-3T26.xlsx",
  "scrapedAt": "2026-10-21T06:30:00.000Z"
}

Pricing

Event	When it fires	Price
apify-actor-start	once per run (synthetic, automatic)	0.00005 USD
apify-default-dataset-item	one validated indicator series saved to the dataset (synthetic, automatic)	0.004 USD per series = 4.00 USD per 1,000 series

One result = one indicator series (a dataset item), with all of its observations inside. Observations are not counted or billed separately. Only series with at least one numeric observation are pushed to the dataset. A workbook without parseable rows delivers nothing and bills nothing. Charging is automatic through the synthetic dataset event: there is no custom per-item or per-observation event, so an item is never billed twice. At 4.00 USD per 1,000 series the reference series of the example (4 observations) works out at 1.00 USD per 1,000 observations; the mean in the local maxItems=25 run was 5.5 observations per series, which is 0.73 USD per 1,000 observations at this price.

Cost per 1,000 results (measured locally)

Two local measurements on 2026-10-21, with no Apify credit spent:

Measurement	Value
Fixtures: 3 workbooks, 49,131 bytes	4 indicator series, 35 numeric observations
Parse + extract + forward fill	1.98 ms per three-workbook batch (1,000 iterations)
Live apify run with {"search":"pib","maxItems":1}	6.3 s wall, 2 HTTP requests, 9,936-byte workbook
Delivered by the live run	1 series with the 4 reference observations
Memory ceiling	256 MB (Actor minimum); 96 MB process RSS in the parse benchmark

The live run includes local tsx startup and the mandatory 500 ms between the two HTTP requests. At 256 MB and Apify's 0.20 USD/GB-hour reference, 6.3 s is about 0.0000875 USD per run/series: 0.09 USD per 1,000 series and, with four observations per reference series, 0.022 USD per 1,000 observations. The second figure is below the 0.05 USD estimate in the specification. The production image runs dist/main.js without the local tsx startup, so this is a conservative upper bound. At 4.00 USD per 1,000 series the measured compute cost (0.09 USD) leaves a gross margin of about 97.7% before payment fees. The first paid platform run will replace this local measurement with the billed figure.

Legal and polite access

The public review is documented in docs/legal-gate.md:

• robots.txt disallows only /wp-admin/; /wp-json/ and /wp-content/uploads/ are allowed;
• the published sitemap and WordPress page search contain no content-use or anti-scraping clause;
• only aggregate tourism indicators are read: no personal data, no login, no paywall, no CAPTCHA;
• every item carries the Exceltur attribution and the exact source workbook;
• the Actor waits at least 500 ms between requests and aborts on 401/403/429/451 or a WAF page instead of retrying or evading the block.

FAQ

Does it need an API key, a login or a proxy? No. The WordPress media API and the XLSX uploads are public. The Actor uses plain HTTP, no browser and no proxy.

Why is one worksheet sometimes split into several items? Each numeric indicator column (time-series layout) or indicator row (matrix layout) becomes one item. That keeps every series attributable to its own published label and avoids mixing units.

Why do some items have null for unit or sourceNote? Those workbooks do not print that information. The Actor reports null; it never invents a unit or an attribution line that Exceltur did not publish.

What happens if a workbook has no numeric rows? The Actor logs a diagnostic and pushes no item for that workbook, so the run bills nothing for it.

Does it include PDF reports or website HTML? No. Version 0.1 covers published XLSX workbooks only. PDF reports and HTML pages are outside the declared scope.

Where does the data come from? Exceltur, the Spanish tourism business association, through its public WordPress media library. Every item cites the source workbook URL and the publisher.

Maintenance

Tested daily: a scheduled run validates the public media listing and the dataset schema every day. Failures become issues in the Steadfetch Actors repository. npm test, npm run publish:check and apify-cli validate-schema run in CI on every push.