Skip to content

v0.13.0

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 15 May 20:47
· 15 commits to main since this release
v0.13.0
54d3060

geoextent v0.13.0

The biggest release since the project went public: 79 commits, 23+ new content providers, a new plain-text source backed by spaCy NER, journal article landing-page support, and EPSG:4326-native coordinate order on by default.

Cite: 10.5281/zenodo.3925693 · PyPI: geoextent 0.13.0 · Docs: https://nuest.github.io/geoextent/

Highlights

🆕 Plain-text source via spaCy NER (#112)

A new handle_text handler runs spaCy NER over plain-text inputs, resolves place mentions through a gazetteer (Nominatim by default — no API key), and resolves named geological time periods through a bundled ICS GTS2020 chart (178 eons, eras, periods, epochs, ages). Calendar dates, decade and century envelopes, range expressions ("between 2010 and 2015"), and deep-time periods (Holocene, Mesozoic Era, Late Cretaceous) all flow into bbox / tbox. Pre-CE temporal extents use signed ISO 8601 year strings.

  • New optional install extra: pip install geoextent[nlp]
  • New from_text() API; new --text STRING / - (stdin) / file / directory inputs
  • New --annotate {auto,ansi,brackets,off} for in-terminal match highlighting
  • Standoff offsets contract: every mention carries char_start/char_end indices into an NFC-normalised source_text echo

See the text extraction how-to and the highlighting how-to.

🆕 Journal landing-page support: journals/ umbrella (#76)

A new journals/ content-provider package extracts spatial and temporal extent from journal article landing pages with the ojsGeo (Open Journal Systems) or janeway_geometadata plugins. Source-preference priority is richer-geometry-first: JSON-LD spatialCoverage → <link rel="alternate" type="application/geo+json"> → DC.SpatialCoverage (GeoJSON / WKT) → DC.box → ISO 19139 EX_GeographicBoundingBox → OJS administrativeUnits → ICBM / geo.position points.

The existing Pensoft provider was refactored into the same hierarchy (public API unchanged). Article DOIs are lifted from the HTML head — JSON-LD identifier, citation_doi, prism.doi, DC.Identifier — and fed into --external-metadata enrichment so a user can pass a journal article URL (not a DOI) and still get CrossRef / DataCite records.

See the journals examples page.

🆕 28 new content providers

Repositories: InvenioRDM (generalised from Zenodo, covers CaltechDATA, TU Wien, Frei-Data, GEO Knowledge Hub, TU Graz, Materials Cloud Archive, FDAT, DataPLANT ARChive, KTH, Prism, NYU Ultraviolet) · B2SHARE (EUDAT) · Mendeley Data · 4TU.ResearchData · RADAR · NSF Arctic Data Center · DEIMS-SDR (follows external DOIs) · HALO DB · GBIF · SEANOE · UKCEH · NFDI4Earth Knowledge Hub · DataONE (covers KNB, PISCO, EDI/LTER, NEON, BCO-DMO, ESS-DIVE …) · Wikidata (via SPARQL) · Dataverse instances: ioerDATA, heiDATA, Edmond · CSW-based: BAW, MDI-DE, GDI-DE · STAC (any compliant Collection) · Generic CKAN (data.gov.uk, GovData.de, data.gov.au, …) · GitHub, GitLab, Forgejo/Gitea, Software Heritage · GeoScienceWorld · journals/ umbrella (OJS, Janeway, Pensoft re-folded)

🆕 New file-format support

  • Cloud Optimized GeoTIFF (COG) over HTTP(S) via GDAL /vsicurl/ — header-only metadata extraction without downloading the file (#11)
  • Point clouds (LAS/LAZ) via laspy — header-only bbox extraction and temporal extent from creation date
  • Esri File Geodatabase (.gdb) via GDAL's OpenFileGDB driver
  • Zarr (.zarr) V2 and V3 via GDAL's Zarr driver (#9)

🆕 Other features

  • --metadata-first smart strategy: try metadata-only extraction, fall back to data download. Automatic metadata fallback after empty data downloads is on by default (opt out with --no-metadata-fallback).
  • --time-format for configurable temporal output: date-only (default), ISO 8601, or any strftime (#39)
  • -p / --parallel thread-based parallel file extraction within directories; workers=N API parameter (#34)
  • progress_callback structured progress for from_file() / from_directory() / from_remote() — three built-in callbacks (tqdm, logging, list-collecting) (#80)
  • --map, --preview, --map-dim for static map preview on OpenStreetMap tiles (terminal display via term-image; pip install geoextent[preview]) (#35)
  • --join to merge multiple --output exports into one file (GPKG / GeoJSON / CSV)
  • AppImage for portable Linux distribution — single-file executable bundling Python + GDAL + PROJ via conda-forge + appimagetool; built on every tag push (#40)
  • Interactive download-size confirmation (--max-download-size with download_size_soft_limit=True)
  • Temporal extent from raster files: NetCDF CF time dimensions, GeoTIFF TIFFTAG_DATETIME, ACDD time_coverage_start/end, band-level ACQUISITIONDATETIME (#22)
  • GeoCSV format: CoordX/CoordY columns, .prj sidecars, WKT geometry columns, EarthScope #-prefixed metadata (#52)

⚠️ Breaking changes

  • Default coordinate order is now EPSG:4326 native [lat, lon] for plain bounding boxes. Output bbox is [minlat, minlon, maxlat, maxlon] instead of [minlon, minlat, maxlon, maxlat]. GeoJSON output continues to use [lon, lat] per RFC 7946. Pass --legacy (or legacy=True in the API) to keep the previous [lon, lat] order for plain bboxes.
  • API renamed to snake_case (PEP 8): from_file(), from_directory(), from_remote(). The camelCase aliases (fromFile, fromDirectory, fromRemote) and from_repository() are removed. Internal handler modules and functions are also snake_case.
  • Drop support for bare numeric Zenodo record IDs (e.g. 820562); use the DOI (10.5281/zenodo.820562) or URL (https://zenodo.org/records/820562).
  • --placename default service changed from geonames (which requires GEONAMES_USERNAME) to nominatim (no API key). Use --placename-service geonames to opt back.

Selected bug fixes

  • Reject raster / vector files whose coordinates are projected but falsely declared WGS84
  • Validate bboxes against WGS84 ranges before merging
  • --convex-hull strips multi-hundred-KB admin polygons from place_names provenance after consuming them
  • --placename runs only on the final merged extent (no more per-file duplicate warnings)
  • Fix convex_hull_merge for degenerate single-point and two-point line-segment hulls
  • Text-file detection now combines extension, MIME, basename allowlist (README, LICENSE, CHANGELOG) and a content sniff — runs NER on extensionless project files while explicitly skipping source code (.py, .js, .sh, .yaml, …)
  • --geojsonio now reports the precise endpoint and cause when the geojson.io Gist fallback fails (commonly anonymous-gist 401)

Full diff: v0.12.0...v0.13.0
Full changelog: https://nuest.github.io/geoextent/changelog.html#id1

Install

pip install geoextent==0.13.0
# Optional extras
pip install 'geoextent[nlp]==0.13.0'         # spaCy NER for text inputs
pip install 'geoextent[preview]==0.13.0'     # static map / terminal preview

AppImage: see the Linux AppImage how-to.

Citing this release: Nüst, Daniel; Garzón, Sebastian; Drechsler, Lars and Qamaz, Yousef. (2026, May 15). geoextent (Version v0.13.0). Zenodo. https://doi.org/10.5281/zenodo.3925693

Assets 3
Loading