[#19] Extract thumbnail from ISO documents

ckanext/spatial/harvesters/base.py

@@ -259,6 +259,17 @@ class SpatialHarvester(HarvesterBase):
 
         extras['access_constraints'] = iso_values.get('limitations-on-public-access', '')
 
+        # Grpahic preview
+        browse_graphic = iso_values.get('browse-graphic')
+        if browse_graphic:
+            browse_graphic = browse_graphic[0]
+            extras['graphic-preview-file'] = browse_graphic.get('file')
+            if browse_graphic.get('description'):
+                extras['graphic-preview-description'] = browse_graphic.get('description')
+            if browse_graphic.get('type'):
+                extras['graphic-preview-type'] = browse_graphic.get('type')
+
+
         for key in ['temporal-extent-begin', 'temporal-extent-end']:
             if len(iso_values[key]) > 0:
                 extras[key] = iso_values[key][0]

ckanext/spatial/model/harvested_metadata.py

@@ -344,6 +344,33 @@ class ISOBoundingBox(ISOElement):
         ),
     ]
 
+class ISOBrowseGraphic(ISOElement):
+
+    elements = [
+        ISOElement(
+            name="file",
+            search_paths=[
+                "gmd:fileName/gco:CharacterString/text()",
+            ],
+            multiplicity="1",
+        ),
+        ISOElement(
+            name="description",
+            search_paths=[
+                "gmd:fileDescription/gco:CharacterString/text()",
+            ],
+            multiplicity="0..1",
+        ),
+        ISOElement(
+            name="type",
+            search_paths=[
+                "gmd:fileType/gco:CharacterString/text()",
+            ],
+            multiplicity="0..1",
+        ),
+    ]
+
+
 class ISODocument(MappedXmlDocument):
 
     # Attribute specifications from "XPaths for GEMINI" by Peter Parslow.
@@ -651,7 +678,16 @@ class ISODocument(MappedXmlDocument):
                 "gmd:dataQualityInfo/gmd:DQ_DataQuality/gmd:lineage/gmd:LI_Lineage/gmd:statement/gco:CharacterString/text()",
             ],
             multiplicity="0..1",
-        )
+        ),
+        ISOBrowseGraphic(
+            name="browse-graphic",
+            search_paths=[
+                "gmd:identificationInfo/gmd:MD_DataIdentification/gmd:graphicOverview/gmd:MD_BrowseGraphic",
+                "gmd:identificationInfo/srv:SV_ServiceIdentification/gmd:graphicOverview/gmd:MD_BrowseGraphic",
+            ],
+            multiplicity="*",
+        ),
+
     ]
 
     def infer_values(self, values):