ci: add metadata blocks properties file check

As creating the properties files for metadata blocks is a
tedious manual process, this script ensures in CI everything
is present. It adds to checking for duplicates.

Author: poikilotherm

.github/workflows/check_property_files.yml

@@ -3,6 +3,7 @@ on:
     pull_request:
         #paths:
         #    - "**/*.properties"
+        #    - "scripts/api/data/metadatablocks/*"
 jobs:
     duplicate_keys:
         name: Duplicate Keys
@@ -32,3 +33,56 @@ jobs:
                   if [ "$FAIL" -eq 1 ]; then
                       exit 1
                   fi
+
+    metadata_blocks_properties:
+        name: Metadata Blocks Properties
+        runs-on: ubuntu-latest
+        steps:
+            - uses: actions/checkout@v3
+            - name: Run metadata block properties verification script
+              shell: bash
+              run: |
+                  for MDB in $(find scripts/api/data/metadatablocks -name '*.tsv'); do
+                      BLOCK_NAME=$(sed -n "2p" "$MDB" | cut -f2)
+                      BLOCK_DISPLAYNAME=$(sed -n "2p" "$MDB" | cut -f4)
+                      PROPERTIES_FILE="src/main/java/propertyFiles/$BLOCK_NAME.properties"
+                  
+                      # Check correct file exists
+                      if [ ! -r "$PROPERTIES_FILE" ]; then
+                          echo "::error::Missing properties file for metadata block '$BLOCK_NAME', expected at '$PROPERTIES_FILE'"
+                          continue
+                      fi
+                  
+                      # Check metadata block properties exist and are equal to TSV source
+                      if ! grep -a -q -e "^metadatablock.name=$BLOCK_NAME$" "$PROPERTIES_FILE"; then
+                          echo "::error::Missing 'metadatablock.name=$BLOCK_NAME' or different from TSV source"
+                      fi
+                      if ! grep -a -q -e "^metadatablock.displayName=$BLOCK_DISPLAYNAME$" "$PROPERTIES_FILE"; then
+                          echo "::error::Missing 'metadatablock.displayName=$BLOCK_DISPLAYNAME' or different from TSV source"
+                      fi
+                      if ! grep -a -q -e "^metadatablock.displayFacet=" "$PROPERTIES_FILE"; then
+                          echo "::error::Missing 'metadatablock.displayFacet=...'"
+                      fi
+                  
+                      # Check dataset fields
+                      for FIELD in $(grep -a -A1000 "^#datasetField" "$MDB" | tail -n+2 | grep -a -B1000 "^#controlledVocabulary" | head -n-1 | cut -f2); do
+                          for ENTRY in title description watermark; do
+                              if ! grep -a -q -e "^datasetfieldtype.$FIELD.$ENTRY=" "$PROPERTIES_FILE"; then
+                                  echo "::error::Missing key 'datasetfieldtype.$FIELD.$ENTRY=...'"
+                              fi  
+                          done
+                      done
+                  
+                      # Check CV entries
+                      grep -a -A1000 "^#controlledVocabulary" "$MDB" | tail -n+2 | 
+                      {
+                          while read LINE; do
+                              FIELD_NAME=$(echo "$LINE" | cut -f1)
+                              # TODO: needs to replace UTF-8 chars with nearest ascii here!
+                              FIELD_VALUE=$(echo "$LINE" | cut -f2 | tr '[:upper:]' '[:lower:]' | tr " " "_")
+                              if ! grep -q -a -e "^controlledvocabulary.$FIELD_NAME.$FIELD_VALUE=" "$PROPERTIES_FILE"; then
+                                  echo "::error::Missing key 'controlledvocabulary.$FIELD_NAME.$FIELD_VALUE=...'"
+                              fi 
+                          done
+                      };
+                  done