test: cover standardization collection centroid

tools/3dtrees_metadata/.shed.yml

@@ -0,0 +1,13 @@
+name: 3dtrees_metadata
+owner: bgruening
+description: Extract administrative and ecoregion metadata for 3Dtrees collections.
+homepage_url: https://github.com/3dTrees-earth/3dtrees_metadata
+long_description: |
+  Extracts administrative and ecoregion context for a 3Dtrees collection summary
+  by computing the collection centroid from collection_summary.json and
+  intersecting it with selected GADM and WWF Terrestrial Ecoregions v2.0 vector
+  datasets.
+remote_repository_url: https://github.com/3dTrees-earth/galaxytools/tree/3dtrees_metadata_v1_0_0/tools/3dtrees_metadata
+type: unrestricted
+categories:
+- "Geo Science"

tools/3dtrees_metadata/3dtrees_metadata.xml

@@ -0,0 +1,106 @@
+<tool id="3dtrees_metadata" name="3Dtrees: Metadata" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="24.2">
+    <description>Extract administrative and ecoregion metadata for a point-cloud collection</description>
+    <macros>
+        <token name="@TOOL_VERSION@">1.0.0</token>
+        <token name="@VERSION_SUFFIX@">0</token>
+    </macros>
+    <requirements>
+        <container type="docker">ghcr.io/3dtrees-earth/3dtrees_metadata:@TOOL_VERSION@</container>
+    </requirements>
+    <command detect_errors="exit_code"><![CDATA[
+        ln -s '$collection_summary' collection_summary.json &&
+        python -u /src/run.py
+        --collection-summary collection_summary.json
+        --metadata-layers '$metadata_layers'
+        --output-file additional_metadata.json
+        2>&1
+    ]]></command>
+    <inputs>
+        <param name="collection_summary" type="data" format="json" label="Collection summary" help="collection_summary.json from the 3Dtrees: LAS/LAZ Standardization collection mode."/>
+        <param argument="--metadata-layers" type="select" multiple="true" display="checkboxes" min="1" label="Metadata layers to extract" help="Choose at least one metadata layer. Both are selected by default.">
+            <option value="gadm" selected="true">GADM administrative boundaries</option>
+            <option value="ecoregion" selected="true">WWF Terrestrial Ecoregions v2.0</option>
+        </param>
+    </inputs>
+    <outputs>
+        <data name="additional_metadata" format="json" label="additional_metadata" from_work_dir="additional_metadata.json"/>
+    </outputs>
+    <tests>
+        <test expect_num_outputs="1">
+            <param name="collection_summary" value="collection_summary.json" ftype="json"/>
+            <param name="metadata_layers" value="gadm,ecoregion"/>
+            <output name="additional_metadata" ftype="json">
+                <assert_contents>
+                    <has_text text='"matched": true'/>
+                    <has_text text='"selected_level": 1'/>
+                    <has_text text='"name": "Baden-Wuerttemberg"'/>
+                    <has_text text='"ecoregion_name": "Black Forest"'/>
+                    <has_text text='"biome_name": "Temperate Broadleaf &amp; Mixed Forests"'/>
+                    <has_text text='"is_forest_biome": true'/>
+                </assert_contents>
+            </output>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="collection_summary" value="collection_summary.json" ftype="json"/>
+            <param name="metadata_layers" value="gadm"/>
+            <output name="additional_metadata" ftype="json">
+                <assert_contents>
+                    <has_text text='"admin"'/>
+                    <has_text text='"selected_level": 1'/>
+                    <has_text text='"name": "Baden-Wuerttemberg"'/>
+                    <not_has_text text='"ecoregion"'/>
+                </assert_contents>
+            </output>
+        </test>
+        <test expect_num_outputs="1">
+            <param name="collection_summary" value="collection_summary.json" ftype="json"/>
+            <param name="metadata_layers" value="ecoregion"/>
+            <output name="additional_metadata" ftype="json">
+                <assert_contents>
+                    <has_text text='"ecoregion_name": "Black Forest"'/>
+                    <has_text text='"is_forest_biome": true'/>
+                    <not_has_text text='"admin"'/>
+                </assert_contents>
+            </output>
+        </test>
+    </tests>
+    <help format="markdown"><![CDATA[
+**What it does**
+
+This tool reads a 3Dtrees `collection_summary.json`, computes the WGS84 centroid
+from `collection.multipolygon_wkt`, and intersects that point with the selected
+reference layers. GADM provides administrative context; WWF Terrestrial
+Ecoregions v2.0 data provides ecoregion, realm, and biome context.
+
+-----
+
+**Input**
+
+- **Collection summary**: `collection_summary.json` produced by the 3Dtrees
+  standardization tool in collection mode.
+- **Metadata layers to extract**: one or more internal reference layers. The tool
+  reads the selected reference datasets from the Docker image.
+
+-----
+
+**Output**
+
+The tool writes `additional_metadata.json` with:
+
+- GADM source details and checked layers
+- Collection centroid longitude/latitude in EPSG:4326
+- Whether the centroid matched a GADM feature
+- Selected administrative level and hierarchy
+- Raw matched GADM fields for downstream metadata enrichment
+- Optional WWF ecoregion name/id, realm, biome code/name, and forest-biome flag
+    ]]></help>
+    <creator>
+        <person name="Kilian Gerberding" email="kilian.gerberding@geosense.uni-freiburg.de" url="https://orcid.org/0009-0002-5001-2571"/>
+        <organization name="3Dtrees-Team, University of Freiburg" url="https://github.com/3dTrees-earth"/>
+    </creator>
+    <citations>
+        <citation type="bibtex">
+            @misc{3dtrees_metadata, title = {3Dtrees: Metadata}, author = {3Dtrees Project}, year = {2026}}
+        </citation>
+    </citations>
+</tool>

tools/3dtrees_metadata/test-data/collection_summary.json

@@ -0,0 +1,8 @@
+{
+  "collection": {
+    "multipolygon_wkt": "MULTIPOLYGON(((7.84 47.98, 7.86 47.98, 7.86 48.00, 7.84 48.00, 7.84 47.98)))"
+  },
+  "n_tiles": 2,
+  "homogeneous_crs": true,
+  "homogeneous_attribute_names": true
+}

tools/3dtrees_metadata/test-data/gadm_fixture.geojson

@@ -0,0 +1,35 @@
+{
+  "type": "FeatureCollection",
+  "name": "gadm_fixture",
+  "crs": {
+    "type": "name",
+    "properties": {
+      "name": "urn:ogc:def:crs:OGC:1.3:CRS84"
+    }
+  },
+  "features": [
+    {
+      "type": "Feature",
+      "properties": {
+        "GID_0": "DEU",
+        "COUNTRY": "Germany",
+        "GID_1": "DEU.1_1",
+        "NAME_1": "Baden-Wuerttemberg",
+        "TYPE_1": "Land",
+        "ENGTYPE_1": "State"
+      },
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [
+          [
+            [7.70, 47.80],
+            [8.00, 47.80],
+            [8.00, 48.20],
+            [7.70, 48.20],
+            [7.70, 47.80]
+          ]
+        ]
+      }
+    }
+  ]
+}

tools/3dtrees_metadata/test-data/wwf_ecoregions_fixture.geojson

@@ -0,0 +1,34 @@
+{
+  "type": "FeatureCollection",
+  "name": "wwf_ecoregions_fixture",
+  "crs": {
+    "type": "name",
+    "properties": {
+      "name": "urn:ogc:def:crs:OGC:1.3:CRS84"
+    }
+  },
+  "features": [
+    {
+      "type": "Feature",
+      "properties": {
+        "ECO_NAME": "Black Forest",
+        "ECO_ID": "PA0414",
+        "REALM": "PA",
+        "BIOME": 4,
+        "BIOME_DESC": "Temperate Broadleaf & Mixed Forests"
+      },
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [
+          [
+            [7.70, 47.80],
+            [8.00, 47.80],
+            [8.00, 48.20],
+            [7.70, 48.20],
+            [7.70, 47.80]
+          ]
+        ]
+      }
+    }
+  ]
+}

tools/3dtrees_smart_tile/smart_tile.xml

@@ -1,10 +1,13 @@
 <tool id="3dtrees_smart_tile" name="3DTrees: SmartTile" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="24.2">
     <description>Create COPC tiles, subsample outputs, filter border overlap, and remap prediction dimensions</description>
     <macros>
-        <token name="@TOOL_VERSION@">2.0.0</token>
+        <token name="@TOOL_VERSION@">2.0.1</token>
         <token name="@VERSION_SUFFIX@">0</token>
         <xml name="produce_merged_file_param">
-            <param argument="--produce-merged-file" type="boolean" truevalue="True" falsevalue="False" checked="true" label="Also write one merged output (LAZ)" help="If enabled (default), produce a single merged LAZ in addition to the per-file outputs."/>
+            <param name="produce_merged_file" type="select" label="Also write one merged output (LAZ)" help="If enabled (default), produce a single merged LAZ in addition to the per-file outputs.">
+                <option value="False">No</option>
+                <option value="True" selected="true">Yes</option>
+            </param>
         </xml>
         <xml name="merge_enrichment_conditional">
             <conditional name="merge_enrichment_mode">
@@ -59,6 +62,7 @@
                 --resolution-1 '$operation.resolution_1'
                 --resolution-2 '$operation.resolution_2'
                 --output-copc-res1 $operation.output_copc_res1
+                --output-copc-res2 $operation.output_copc_res2
                 --dimension-reduction $operation.dimension_reduction
                 --chunk-size '$operation.chunk_size'
                 --num-spatial-chunks \${GALAXY_SLOTS:-4}
@@ -229,6 +233,7 @@
                 <param argument="--resolution-1" type="float" min="0.001" max="1.0" value="0.01" label="Subsample resolution 1 (m)" help="First subsampling resolution, in meters."/>
                 <param argument="--resolution-2" type="float" min="0.001" max="1.0" value="0.1" label="Subsample resolution 2 (m)" help="Second subsampling resolution, in meters."/>
                 <param argument="--output-copc-res1" type="boolean" truevalue="True" falsevalue="False" checked="true" label="Write resolution 1 as COPC (.copc.laz)" help="If enabled (default), the first subsampled output collection is written as COPC LAZ (`*.copc.laz`). If disabled, it is written as standard LAZ (`*.laz`). This does not affect the internal COPC-normalized source cache."/>
+                <param argument="--output-copc-res2" type="boolean" truevalue="True" falsevalue="False" checked="false" label="Write resolution 2 as COPC (.copc.laz)" help="If enabled, the second subsampled output collection is written as COPC LAZ (`*.copc.laz`). If disabled (default), it is written as standard LAZ (`*.laz`). This matches the new SmartTile res2 COPC output option."/>
                 <param argument="--dimension-reduction" type="boolean" truevalue="True" falsevalue="False" checked="true" label="Reduce dimensions in subsampled outputs" help="If enabled (default), subsampled outputs are written with a minimal LAS schema (smaller files). The initial source-to-COPC normalization step still preserves all input dimensions."/>
                 <param argument="--chunk-size" type="integer" min="100000" max="200000000" value="20000000" label="Chunk size (points)" help="Performance knob. Controls chunked reads used during COPC normalization and later spatial slicing steps."/>
             </when>
@@ -363,7 +368,7 @@
             <output_collection name="output_subsampled_res2" type="list" count="1"/>
             <output name="output_tiling_preview">
                 <assert_contents>
-                    <has_image_center_of_mass center_of_mass="1813, 1768" eps="100"/>
+                    <has_image_center_of_mass center_of_mass="1974, 1785" eps="100"/>
                 </assert_contents>
             </output>
             <output name="output_tile_bounds_json">
@@ -449,6 +454,7 @@
 - It then creates overlapping tiles and writes two subsampled output collections.
 - `Reduce dimensions in subsampled outputs` only affects the subsampled outputs, not the internal COPC cache.
 - `Write resolution 1 as COPC (.copc.laz)` only affects the first subsampled output collection (default: enabled).
+- `Write resolution 2 as COPC (.copc.laz)` only affects the second subsampled output collection (default: disabled).
 
 **Filter**
 

tools/3dtrees_standardization/standard.xml

@@ -1,7 +1,7 @@
 <tool id="3dtrees_standardization" name="3DTrees: LAS/LAZ Standardization" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="24.2">
     <description>Standardize LAS/LAZ files or validate collections for consistency</description>
     <macros>
-        <token name="@TOOL_VERSION@">1.1.0</token>
+        <token name="@TOOL_VERSION@">1.2.0</token>
         <token name="@VERSION_SUFFIX@">0</token>
     </macros>
     <requirements>
@@ -27,6 +27,12 @@
             --dataset-path ./input_files/
             --output-dir ./
             --method collection
+            #if $mode.approx_lon is not None:
+                --approx-lon '$mode.approx_lon'
+            #end if
+            #if $mode.approx_lat is not None:
+                --approx-lat '$mode.approx_lat'
+            #end if
         #end if
     ]]></command>
 
@@ -49,6 +55,12 @@
             <when value="collection">
                 <param name="input_collection" type="data" format="laz" multiple="true" label="Point Cloud Collection" 
                        help="Multiple LAZ/LAS files to validate for consistency (CRS homogeneity, attribute consistency, overlaps)"/>
+                <param argument="--approx-lon" type="float" optional="true" min="-180" max="180"
+                       label="Approximate longitude"
+                       help="Optional WGS84 longitude that overrides the calculated collection centroid when paired with approximate latitude."/>
+                <param argument="--approx-lat" type="float" optional="true" min="-90" max="90"
+                       label="Approximate latitude"
+                       help="Optional WGS84 latitude that overrides the calculated collection centroid when paired with approximate longitude."/>
             </when>
         </conditional>
     </inputs>
@@ -90,16 +102,55 @@
                 </assert_contents>
             </output>
         </test>
+        <test expect_num_outputs="1">
+            <conditional name="mode">
+                <param name="task" value="collection"/>
+                <param name="input_collection" value="mikro_no_crs.laz,mikro2_no_crs.laz" ftype="laz"/>
+                <param name="approx_lon" value="7.85"/>
+                <param name="approx_lat" value="47.99"/>
+            </conditional>
+            <output name="collection_summary" ftype="json">
+                <assert_contents>
+                    <has_json_property_with_value property="n_tiles" value="2"/>
+                    <has_json_property_with_value property="homogeneous_crs" value="true"/>
+                    <has_json_property_with_value property="homogeneous_attribute_names" value="true"/>
+                    <has_json_property_with_value property="longitude" value="7.85"/>
+                    <has_json_property_with_value property="latitude" value="47.99"/>
+                    <has_json_property_with_text property="crs" text="EPSG:4326"/>
+                    <has_json_property_with_text property="source" text="approximate_input"/>
+                    <has_json_property_with_value property="available" value="true"/>
+                    <has_json_property_with_text property="multipolygon_crs" text="LOCAL_OR_UNKNOWN"/>
+                    <has_json_property_with_text property="multipolygon_source" text="source"/>
+                    <has_json_property_with_value property="multipolygon_available" value="true"/>
+                    <has_json_property_with_value property="multipolygon_wgs84_available" value="false"/>
+                    <has_json_property_with_text property="concave_hull_crs" text="LOCAL_OR_UNKNOWN"/>
+                    <has_json_property_with_text property="concave_hull_source" text="source"/>
+                </assert_contents>
+            </output>
+        </test>
         <test expect_num_outputs="1">
             <conditional name="mode">
                 <param name="task" value="collection"/>
                 <param name="input_collection" value="mikro.laz,mikro2.laz" ftype="laz"/>
+                <param name="approx_lon" value="7.86"/>
+                <param name="approx_lat" value="48.01"/>
             </conditional>
             <output name="collection_summary" ftype="json">
                 <assert_contents>
                     <has_json_property_with_value property="n_tiles" value="2"/>
                     <has_json_property_with_value property="homogeneous_crs" value="true"/>
                     <has_json_property_with_value property="homogeneous_attribute_names" value="true"/>
+                    <has_json_property_with_value property="longitude" value="7.86"/>
+                    <has_json_property_with_value property="latitude" value="48.01"/>
+                    <has_json_property_with_text property="crs" text="EPSG:4326"/>
+                    <has_json_property_with_text property="source" text="approximate_input"/>
+                    <has_json_property_with_value property="available" value="true"/>
+                    <has_json_property_with_text property="multipolygon_crs" text="EPSG:4326"/>
+                    <has_json_property_with_text property="multipolygon_source" text="wgs84_transform"/>
+                    <has_json_property_with_value property="multipolygon_available" value="true"/>
+                    <has_json_property_with_value property="multipolygon_wgs84_available" value="true"/>
+                    <has_json_property_with_text property="concave_hull_crs" text="EPSG:4326"/>
+                    <has_json_property_with_text property="concave_hull_source" text="wgs84_transform"/>
                 </assert_contents>
             </output>
         </test>
@@ -134,6 +185,7 @@ Validates consistency of multiple LAS/LAZ files:
 - **Tile Geometry**: Detects overlaps and gaps between tiles
 - **Global Statistics**: Computes aggregated statistics across all files
 - **Removeable Attributes**: Identifies attributes that are all-NA or constant-zero
+- **Collection Centroid**: Computes a WGS84 centroid from valid collection geometry, or uses optional approximate longitude/latitude as a trusted override when provided
 
 **Outputs (collection mode):**
 - ``collection_summary.json`` containing collection-level checks and flags, per-file metadata (CRS, attributes, point counts, convex hulls), and suggested removeable attributes for standardization