Journal Landing Page Examples

This page demonstrates how to extract geospatial extent from journal article landing pages that advertise spatial metadata in their HTML head. Geoextent ships with a journals/ umbrella content provider that detects three publishing platforms out of the box and pulls a normalised bbox / tbox out of whatever encoding the publisher chose.

Overview

The journals/ umbrella covers:

  • OJS (Open Journal Systems) with the ojsGeo plugin — fingerprinted via the <meta name="generator" content="Open Journal Systems …"> tag every OJS theme emits.

  • Janeway with the janeway_geometadata plugin — fingerprinted via the <link rel="alternate" type="application/geo+json"> link the plugin always renders, plus the /article/id/{N}/ URL pattern.

  • Pensoft journals (e.g. Biodiversity Data Journal, ZooKeys) — DOI-prefix fast-path on 10.3897/ plus a JSON-LD contentLocation walker. See Examples for the existing Pensoft example.

For OJS and Janeway you can pass either:

  • the article landing-page URL (http://journal-host/article/view/{N} or /article/id/{N}/), or

  • a DOI that resolves to the article (any prefix — the OJS/Janeway providers do not need a per-publisher allowlist).

A journal that advertises the platform but does not have the geo plugin installed is still recognised; geoextent then returns no spatial extent but will still surface the publication date and, when --ext-metadata is set, look up the article DOI on CrossRef / DataCite.

Source-Preference Priority

When a page emits the same article in multiple encodings, the provider picks the one most likely to carry the richest geometry, not the one easiest to parse. Priority order (first match wins):

  1. JSON-LD <script type="application/ld+json"> with spatialCoverage or contentLocation — may carry Point, Polygon, MultiPolygon, GeometryCollection, Feature, or FeatureCollection.

  2. <link rel="alternate" type="application/geo+json"> — fetched inline from the plugin’s canonical export endpoint.

  3. DC.SpatialCoverage with scheme="GeoJSON" (handles both the OJS FeatureCollection wrapper and the Janeway single-Feature wrapper).

  4. DC.SpatialCoverage with scheme="WKT".

  5. Inlined ISO 19139 EX_GeographicBoundingBox snippets.

  6. DCMI Box (DC.box) bounding-box fields.

  7. OJS administrativeUnits[].bbox fallback (when the plugin recorded an admin unit but no real geometry).

  8. ICBM / geo.position centroids — last resort, single-point bbox.

The temporal extent chain is analogous, but more conservative: JSON-LD temporalCoverage → DC.temporal / DC.PeriodOfTime ISO 8601 intervals → embedded GeoJSON temporal_periods. Publication-date meta tags (``DC.Date.issued``, ``citation_publication_date``, ``citation_date``) are deliberately NOT used as a fallback — a journal article’s publication date is metadata about the article, not about what the article studied. If no research-period source is present, tbox stays None.

Basic Extraction

OJS article with a real polygon and research period

Article 40 in the TIB OJS demo journal carries a polygon over Brandenburg and a research period of 2008–2018 advertised via <meta name="DC.temporal" scheme="ISO8601" content="2008-01-01/2018-12-31"/> — note that this is distinct from the publication date in DC.Date.issued (2025-07-14). The temporal resolver prefers the research period because it appears earlier in the priority list:

python -m geoextent -b -t \\
  https://service.tib.eu/komet/ojs330/index.php/gmdj/article/view/40

Output:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Polygon",
        "coordinates": [[
          [11.2657725, 51.359064],
          [14.7658159, 51.359064],
          [14.7658159, 53.5590907],
          [11.2657725, 53.5590907],
          [11.2657725, 51.359064]
        ]]
      },
      "properties": {
        "tbox": ["2008-01-01", "2018-12-31"]
      }
    }
  ],
  "geoextent_extraction": {
    "inputs": ["https://service.tib.eu/komet/ojs330/index.php/gmdj/article/view/40"],
    "format": "remote",
    "crs": "4326",
    "extent_type": "bounding_box"
  }
}

OJS article with point coordinates and no research period

Article 32 on the same demo journal references a study site in Sulawesi as a single point and has no DC.temporal / DC.PeriodOfTime / JSON-LD temporalCoverage — only a DC.Date.issued publication date, which is intentionally not used as a fallback. tbox is therefore absent from the output. The point itself comes from the GeoJSON-encoded DC.SpatialCoverage (the OJS plugin’s first-class encoding):

python -m geoextent -b -t \\
  https://service.tib.eu/komet/ojs330/index.php/gmdj/article/view/32

Output (geometry shortened):

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Point",
        "coordinates": [121.34369641542438, -3.9118318544582156]
      },
      "properties": {}
    }
  ],
  "geoextent_extraction": {"extent_type": "point", "format": "remote", "crs": "4326"}
}

Edge case: platform detected, no spatial metadata

Real-world OJS journals that have not installed the geo plugin yet are still recognised as OJS via the generator meta tag, so the provider claims the request even though it has nothing spatial to return. Article 170 in JOSIS is a real peer-reviewed paper without spatial metadata on its landing page:

python -m geoextent -b -t \\
  https://josis.org/index.php/josis/article/view/170

Output:

{
  "format": "remote",
  "_extracted_doi": "10.5311/JOSIS.2021.23.170"
}

The provider was unable to compute a bounding box (no spatial extent warning is logged) and no temporal extent is reported either — only the article DOI is lifted from the head, which sets up the next example.

External Metadata Enrichment via Extracted DOI

This is what makes the journal providers genuinely useful in citation pipelines: pass the URL of any journal article and combine --ext-metadata with the platform fingerprint to get CrossRef / DataCite metadata back, even though the user never typed the DOI:

python -m geoextent -b -t --ext-metadata \\
  https://josis.org/index.php/josis/article/view/170

The provider:

  1. Resolves the URL → fetches the landing page.

  2. Detects the OJS generator meta tag → claims the request.

  3. Lifts the article DOI out of the HTML head — in order: JSON-LD identifier blocks, citation_doi meta, prism.doi, then DC.Identifier if it looks like a DOI.

  4. Substitutes that extracted DOI for the URL when calling external_metadata.get_external_metadata so CrossRef returns a hit.

Output (single-resource shape):

{
  "format": "remote",
  "_extracted_doi": "10.5311/JOSIS.2021.23.170",
  "external_metadata": [
    {
      "source": "CrossRef",
      "doi": "10.5311/JOSIS.2021.23.170",
      "title": "Identifying regional variation in place visit behavior during a global pandemic",
      "authors": ["Grant McKenzie", "Kevin Mwenda"],
      "publisher": "Journal of Spatial Information Science",
      "publication_year": 2021,
      "url": "https://doi.org/10.5311/josis.2021.23.170",
      "license": "https://creativecommons.org/licenses/by/3.0/"
    }
  ]
}

Without --ext-metadata the external_metadata array is omitted; with the flag it is populated whenever the provider could lift a DOI from the head, regardless of whether the original input was a URL or already a DOI.

Python API

Same flow from Python:

import geoextent.lib.extent as geoextent

result = geoextent.from_remote(
    "https://josis.org/index.php/josis/article/view/170",
    bbox=True,
    tbox=True,
    ext_metadata=True,
)

print(result["_extracted_doi"])
# "10.5311/JOSIS.2021.23.170"

print(result["external_metadata"][0]["title"])
# "Identifying regional variation in place visit behavior during a global pandemic"

print(result.get("tbox"))
# None — JOSIS article 170 has no research-period meta tag, and the
# publication date is intentionally not used as a tbox fallback.

Janeway example

The Janeway plugin emits more meta tags than ojsGeo — DC.SpatialCoverage in both WKT and GeoJSON, DC.box, ISO 19139, schema.org JSON-LD with spatialCoverage Place, plus a separate application/geo+json alternate link. Because Janeway has no public demo URL, this example uses the local Delta Quadrant Journal dev instance (article IDs drift as the dev DB is reset):

python -m geoextent -b -t \\
  http://localhost:8000/dqj/article/id/251/

Output (Laos polygon, geometry abbreviated):

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Polygon",
        "coordinates": [[
          [100.1, 13.9], [107.7, 13.9],
          [107.7, 22.5], [100.1, 22.5],
          [100.1, 13.9]
        ]]
      },
      "properties": {}
    }
  ],
  "geoextent_extraction": {"extent_type": "bounding_box", "format": "remote", "crs": "4326"}
}

(The Janeway demo article 251 emits an empty temporal_periods list and no DC.temporal, so no research-period source is available; the publication date is excluded as a fallback per the rule above.)

Notes and Limitations

  • The DOI extracted from the HTML head is exposed in the per-resource dictionary as _extracted_doi. With --ext-metadata it is used as the CrossRef / DataCite lookup key whenever the user-supplied identifier is a URL, not a DOI.

  • The alternate application/geo+json link is fetched inline as one additional GET; failures fall through to the next priority rule silently (debug-logged). Set --debug to see the fetch attempt.

  • Coordinate order at the public API boundary is EPSG:4326 native [lat, lon]. GeoJSON output remains [lon, lat] per RFC 7946. Use --legacy to switch the bbox output to [lon, lat].

  • See Advanced Features for a list of every meta tag and JSON-LD shape the journals/ provider recognises, and #76 for the implementation history.