Readers for national statistics offices everywhere

Changes since 2.19.0.

#New

• KNBS Data Portal (Kenya) and any other .Stat Data Browser hub: every table, complete, with area names and their hierarchy. Tables that list the country, counties and sub-counties together are joined one level at a time (the 2019 census county table: all 47 counties).
• Statistics office websites: for offices without an API, OrionGIS reads the CSV and Excel (xlsx and xls) tables they publish. The walk follows robots.txt, stays on the office's own site, reads publication pages first and has page and depth limits. PDFs are not indexed. Sites behind bot protection are reported, never bypassed.
• UK ONS (dataset API, latest version of every dataset), Statistics Canada (every current table), IBGE SIDRA (Brazil, by state), and Japan e-Stat and India data.gov.in, which need a free API key set in Orion Core. Keys are never stored in records or download links.
• ABS (Australia) and Eurostat use the existing SDMX reader.
• Tables of many countries (Eurostat-style) are joined to country outlines. Statistical codes such as EL (Greece) and UK are recognised.
• Columns that mix levels (for example councils, counties and regions in one list) are joined when they match nearly every area of one boundary level. The other names are listed, not guessed.
• Excel tables with title rows: the header row is found automatically.
• NADA microdata catalogues: a reader for study metadata as link-only records (no microdata copied). Its sources stay off until a decision on how link-only records are published.

#Fixed

• Statistics office servers with incomplete TLS chains (Uganda, Benin, Kenya's main site and others) are read with verification on. The missing intermediate certificates ship with Orion Core, each one checked against the standard root store.
• Other official spellings of boundary names are recognised (KNBS "Tharaka Nithi" = "Tharaka").