Fix ckan export

[karlbrink]

No description provided.

[github-actions]

Code Review by Gemini

Here are the bugs and issues identified in the pull request along with suggested fixes.

---

1. Swapped X/Y Coordinates in Bounding Box Calculation

File: http/php/mod_linkedDataProxy.php
Lines: 308–313, 325–328

In calculateBboxFromGeojsonGeometry(), $minLat / $maxLat represent Y coordinates (latitude) and $minLon / $maxLon represent X coordinates (longitude). When assigning properties to the $bbox object (and when comparing geometry collection parts), minx was assigned $minLat and miny was assigned $minLon. This swaps X and Y coordinates.

// http/php/mod_linkedDataProxy.php

// Lines 308-313
			if ($bboxPart !== false) {
				if ($minLon > $bboxPart->minx) { $minLon = $bboxPart->minx; }
				if ($minLat > $bboxPart->miny) { $minLat = $bboxPart->miny; }
				if ($maxLon < $bboxPart->maxx) { $maxLon = $bboxPart->maxx; }
				if ($maxLat < $bboxPart->maxy) { $maxLat = $bboxPart->maxy; }
			}

// Lines 325-328
	$bbox = new stdClass();
	$bbox->minx = $minLon;
	$bbox->miny = $minLat;
	$bbox->maxx = $maxLon;
	$bbox->maxy = $maxLat;

---

2. Relative Path Used in file_get_contents

File: http/plugins/mb_downloadFeedServer.php
Line: 11

file_exists() uses dirname(__FILE__) for an absolute path, but file_get_contents() uses a relative path ../../conf/excludeFromAtomFeedClient.json. If the current working directory differs from the script directory, file_get_contents() will fail.

// http/plugins/mb_downloadFeedServer.php line 11
	if (file_exists ( dirname ( __FILE__ ) . "/../../conf/excludeFromAtomFeedClient.json" )) {
	    $configObject = json_decode ( file_get_contents ( dirname ( __FILE__ ) . "/../../conf/excludeFromAtomFeedClient.json" ) );
	}

---

3. Typo in Standard Name

File: http/php/mod_exportMapbenderMetadata2Ckan.php
Line: 1089

There is a typo in the ISO standard description string (ISO19168-1:20202). The standard version year contains an extra 2.

// http/php/mod_exportMapbenderMetadata2Ckan.php line 1089
"description" =>   "Objektart: " . $value1->resourceName. " - ISO19168-1:2020 API",

---

4. Docblock Parameter Mismatch

File: http/classes/class_metadata.php
Lines: 2360–2367

The PHPDoc docblock for computeIsHVD() documents $inspireCats as a second parameter (@param array $inspireCats), but the function signature only accepts $customCats. Update the docblock to match the method signature.

// http/classes/class_metadata.php lines 2360-2367
	/**
	 * Computes whether a resource is a High-Value Dataset (HVD)
	 * based on the hvd_cats.json configuration.
	 * Checks if any of the resource's custom category IDs
	 * match those configured as HVD categories.
	 *
	 * @param array $customCats Array of custom category IDs (strings)
	 * @return bool true if the resource qualifies as HVD
	 */
	private function computeIsHVD($customCats) {

[github-actions]

Code Review by Gemini

Here is a code review highlighting issues and suggested improvements found in the pull request:

---

1. http/delete.me

Issue: Temporary / debug file included in the pull request.

• Line 1:

  this is my testfile for ai checks
  

• Recommendation: Remove http/delete.me before merging.

---

2. http/php/mod_exportMapbenderMetadata2Ckan.php

Issue: Duplicate case "wfsrequest": label inside the switch ($value1->type) block. In PHP, duplicate switch cases result in dead code because execution will always jump to the first matching case, rendering the second wfsrequest block unreachable.

• Lines to change (http/php/mod_exportMapbenderMetadata2Ckan.php):

  case "directwfs":
      $featuretypeAccessResource_2 = array("name" => "WFS-Schnittstelle",
          "description" =>   "Objektart: " . $value1->resourceName. " - WFS",
          "format" => "WFS",
          "url" => str_replace($mapbenderWebserviceUrl, $mapbenderBaseUrl, $value1->accessClient),
          "original_url" => str_replace("http://127.0.0.1", $baseUrlPortal, str_replace("http://localhost", $baseUrlPortal, $value1->originalGetCapabilitiesUrl)),
          "id" => $gpDataset->uuid . "_wfs_interface_" . str_replace(":", "__colon__", $value1->resourceName) . "_" . $value1->serviceId,
          "license_id" => $inspireAtomFeedsLicenseId,
          "license_source_note" => $value1->licenseSourceNote
      );
      $resourceArray[] = $featuretypeAccessResource_2;
      break;

---

3. http/classes/class_metadata.php

Issue 1: In computeIsHVD(), casting non-numeric or empty string values like "" using (int)$catId evaluates to 0. If 0 exists in $this->hvdCustomCats, this yields false positives.

• Lines to change (http/classes/class_metadata.php):

  private function computeIsHVD($customCats) {
      if (!empty($this->hvdCustomCats) && is_array($customCats)) {
          foreach ($customCats as $catId) {
              if (is_numeric($catId) && in_array((int)$catId, $this->hvdCustomCats, true)) {
                  return true;
              }
          }
      }
      return false;
  }

Issue 2: The DocBlock comment for computeIsHVD() documents $inspireCats as a second parameter, but the method signature accepts only one argument ($customCats). Update the PHPDoc comment to match the function signature.

---

4. http/plugins/mb_downloadFeedServer.php

Issue: Potential file_get_contents() failure due to relative path usage. While file_exists() uses dirname(__FILE__) . "/../../conf/excludeFromAtomFeedClient.json", file_get_contents() uses ../../conf/excludeFromAtomFeedClient.json". If the script is executed from a different working directory, file_get_contents() will fail.

• Lines to change (http/plugins/mb_downloadFeedServer.php):

  if (file_exists ( dirname ( __FILE__ ) . "/../../conf/excludeFromAtomFeedClient.json" )) {
      $configObject = json_decode ( file_get_contents ( dirname ( __FILE__ ) . "/../../conf/excludeFromAtomFeedClient.json" ) );
  }

[github-actions]

ERROR: # Code Review by Gemini
An unexpected error occurred in Client call: 503 UNAVAILABLE. {'error': {'code': 503, 'message': 'This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.', 'status': 'UNAVAILABLE'}}

[github-actions]

Code Review by Gemini

Here is the review of the changes in this pull request along with identified bugs and suggested improvements:

---

1. Leftover Test File

http/delete.me

A test file http/delete.me was added in this commit. Test/temporary trigger files should be removed before merging into production.

Action Required: Remove http/delete.me.

---

2. Inverted Bounding Box Coordinates and Coordinate System Boundary Handling

http/php/mod_linkedDataProxy.php

In calculateBboxFromGeojsonGeometry:

function calculateBboxFromGeojsonGeometry($geometry) {
	if (!is_object($geometry) || !isset($geometry->type)) {
		return false;
	}
	$minLat = 90;
	$minLon = 180;
	$maxLat = -90;
	$maxLon = -180;
...
	$bbox = new stdClass();
	$bbox->minx = $minLat;
	$bbox->miny = $minLon;
	$bbox->maxx = $maxLat;
	$bbox->maxy = $maxLon;
	return $bbox;
}

Issues:

1. Inverted Coordinates: minx and maxx are assigned $minLat/$maxLat (Y coordinates/latitude), while miny and maxy are assigned $minLon/$maxLon (X coordinates/longitude). This swaps the X and Y axes in generated bounding box outputs.
2. Hardcoded Degree Range: Initializing bounds to 90/-90 and 180/-180 fails for non-geographic projected coordinate systems (e.g. UTM / EPSG:25832 meter-based coordinates).

Suggested Change:
In http/php/mod_linkedDataProxy.php, update the initialization, assignment, and properties:

function calculateBboxFromGeojsonGeometry($geometry) {
	if (!is_object($geometry) || !isset($geometry->type)) {
		return false;
	}
	$minLat = INF;
	$minLon = INF;
	$maxLat = -INF;
	$maxLon = -INF;

	if ($geometry->type === "GeometryCollection" && isset($geometry->geometries) && is_array($geometry->geometries)) {
		foreach ($geometry->geometries as $geometryPart) {
			$bboxPart = calculateBboxFromGeojsonGeometry($geometryPart);
			if ($bboxPart !== false) {
				if ($minLon > $bboxPart->minx) { $minLon = $bboxPart->minx; }
				if ($minLat > $bboxPart->miny) { $minLat = $bboxPart->miny; }
				if ($maxLon < $bboxPart->maxx) { $maxLon = $bboxPart->maxx; }
				if ($maxLat < $bboxPart->maxy) { $maxLat = $bboxPart->maxy; }
			}
		}
	}
	else if (isset($geometry->coordinates)) {
		$coords = json_decode(json_encode($geometry->coordinates), true);
		updateBboxAccumulatorFromCoordinates($coords, $minLat, $minLon, $maxLat, $maxLon);
	}

	if ($minLat === INF || $minLon === INF || $maxLat === -INF || $maxLon === -INF) {
		return false;
	}

	$bbox = new stdClass();
	$bbox->minx = $minLon;
	$bbox->miny = $minLat;
	$bbox->maxx = $maxLon;
	$bbox->maxy = $maxLat;
	return $bbox;
}

---

3. Partial String Matching in Subdomain/Path for Whitelist Checks

http/plugins/mb_downloadFeedServer.php

function isWhitelistedDownloadFeedUrl($serviceFeedUrl) {
	global $urlsWhitelist;
	if ($urlsWhitelist === false || !is_array($urlsWhitelist) || count($urlsWhitelist) === 0) {
		return false;
	}
	$parsedUrl = parse_url($serviceFeedUrl);
	$host = isset($parsedUrl['host']) ? strtolower($parsedUrl['host']) : '';
	foreach ($urlsWhitelist as $whitelistEntry) {
		$whitelistEntry = strtolower(trim($whitelistEntry));
		if ($whitelistEntry === '') {
			continue;
		}
		if ($host !== '' && $host === $whitelistEntry) {
			return true;
		}
		if (strpos(strtolower($serviceFeedUrl), $whitelistEntry) !== false) {
			return true;
		}
	}
	return false;
}

Issue:
The strpos(strtolower($serviceFeedUrl), $whitelistEntry) !== false check matches any occurrences of the whitelist string anywhere inside the URL (such as path or query parameters). This can cause arbitrary non-whitelisted domains to bypass blacklist validation.

Suggested Change:
In http/plugins/mb_downloadFeedServer.php, replace the matching loop:

function isWhitelistedDownloadFeedUrl($serviceFeedUrl) {
	global $urlsWhitelist;
	if ($urlsWhitelist === false || !is_array($urlsWhitelist) || count($urlsWhitelist) === 0) {
		return false;
	}
	$parsedUrl = parse_url($serviceFeedUrl);
	$host = isset($parsedUrl['host']) ? strtolower($parsedUrl['host']) : '';
	if ($host === '') {
		return false;
	}
	foreach ($urlsWhitelist as $whitelistEntry) {
		$whitelistEntry = strtolower(trim($whitelistEntry));
		if ($whitelistEntry === '') {
			continue;
		}
		if ($host === $whitelistEntry || sprintf('.%s', $host) === strstr($host, sprintf('.%s', $whitelistEntry))) {
			return true;
		}
	}
	return false;
}

---

4. Whitespace Parsing in GML Coordinate Delimiters

http/classes/class_gml_3_factory.php

$coordArray = explode(' ', trim($coordsDom->nodeValue));

Issue:
GML coordinate blocks frequently contain multi-space separations, newlines, or tabs. Using explode(' ', ...) creates empty array elements which leads to invalid coordinate indices during iteration.

Suggested Change:
In http/classes/class_gml_3_factory.php, update coordinate parsing in parseCurve (and existing methods using explode(' ', ...)):

Lines 198, 226, 290, etc.:

- $coordArray = explode(' ', trim($coordsDom->nodeValue));
+ $coordArray = preg_split('/\s+/', trim($coordsDom->nodeValue));

---

5. Docblock Signature Mismatch

http/classes/class_metadata.php

	/**
	 * Computes whether a resource is a High-Value Dataset (HVD)
	 * based on the hvd_cats.json configuration.
	 * Checks if any of the resource's custom or inspire category IDs
	 * match those configured as HVD categories.
	 *
	 * @param array $customCats Array of custom category IDs (strings)
	 * @param array $inspireCats Array of inspire category IDs (strings)
	 * @return bool true if the resource qualifies as HVD
	 */
	private function computeIsHVD($customCats) {

Issue:
The docblock specifies $inspireCats as a second parameter, but computeIsHVD accepts only $customCats.

Suggested Change:
In http/classes/class_metadata.php, update the PHPDoc block:

	/**
	 * Computes whether a resource is a High-Value Dataset (HVD)
	 * based on the hvd_cats.json configuration.
	 *
	 * @param array $customCats Array of custom category IDs (strings)
	 * @return bool true if the resource qualifies as HVD
	 */
	private function computeIsHVD($customCats) {

[Copilot]

Pull request overview

This PR improves CKAN export output and hardens/extends several Mapbender HTTP endpoints involved in WFS/linked-data access and INSPIRE download feed handling.

Changes:

• Add whitelist support to the download feed server’s URL filtering and expand the default blacklist distribution file.
• Improve mod_linkedDataProxy.php robustness: WFS version fallback chain, better bbox handling for more GeoJSON geometry types, and STARTINDEX compatibility.
• Extend CKAN metadata export handling for additional AtomFeed-linked resource types; add metadata search facets and HVD logic updates.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
http/plugins/mb_downloadFeedServer.php	Adds whitelist logic to bypass blacklist for trusted feed URLs (needs tightening + path fix).
http/php/mod_linkedDataProxy.php	Adds WFS paging/version fallbacks and generalized GeoJSON bbox computation.
http/php/mod_exportMapbenderMetadata2Ckan.php	Exports additional AtomFeed-linked resource types into CKAN distributions.
http/delete.me	New test file added (should be removed).
http/classes/class_wfs.php	Adjusts WFS KVP parameter casing / typeName parameter handling and paging request building.
http/classes/class_metadata.php	Adds adminTypes facet/filtering, improves HVD filtering logic, and enriches metadata JSON fields.
http/classes/class_gml_feature.php	Outputs null for missing geometry in GeoJSON.
http/classes/class_gml_3_factory.php	Improves GML3 geometry parsing (Curve + namespace-agnostic XPath), geometry detection, and SRS handling.
conf/excludeFromAtomFeedClient.json-dist	Expands blacklist patterns and adds a whitelist example section.
.github/workflows/gemini-review.yml	Changes workflow trigger to run only when the gemini-review label is added.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.