Fix fetching of empty categories (#125)

Closes #124
OpenRefine · Oct 1, 2024 · b6a1120 · b6a1120
1 parent 8b202dd
commit b6a1120
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 22 deletions.
diff --git a/src/main/java/org/openrefine/extensions/commons/importer/CommonsImporter.java b/src/main/java/org/openrefine/extensions/commons/importer/CommonsImporter.java
@@ -10,6 +10,7 @@
 import com.fasterxml.jackson.databind.node.ObjectNode;
 import com.google.common.collect.Iterators;
 import com.google.refine.ProjectMetadata;
+import com.google.refine.importers.ImporterUtilities;
 import com.google.refine.importers.TabularImportingParserBase;
 import com.google.refine.importing.ImportingJob;
 import com.google.refine.model.Column;
@@ -76,6 +77,30 @@ static public void parse(
 
         // initializes progress reporting with the name of the first category
         setProgress(job, categoriesWithDepth.get(0).categoryName, 0);
+
+        // pre-allocate columns
+        List<String> columnNames = new ArrayList<>();
+        Column col = ImporterUtilities.getOrAllocateColumn(project, columnNames, 0, false);
+        StandardReconConfig cfg = new StandardReconConfig(
+                service,
+                "https://commons.wikimedia.org/entity/",
+                "http://www.wikidata.org/prop/direct/",
+                "",
+                "entity",
+                true,
+                1,
+                new ArrayList<ColumnDetail>());
+        col.setReconStats(ReconStats.create(project, 0));
+        col.setReconConfig(cfg);
+        col.setName("File");
+        if (mIdsColumn) {
+            ImporterUtilities.getOrAllocateColumn(project, columnNames, 1, false).setName("M-ids");
+            if (categoriesColumn) {
+                ImporterUtilities.getOrAllocateColumn(project, columnNames, 2, false).setName("Categories");
+            }
+        } else if (categoriesColumn) {
+            ImporterUtilities.getOrAllocateColumn(project, columnNames, 1, false).setName("Categories");
+        }
 
         for(CategoryWithDepth categoryWithDepth: categoriesWithDepth) {
             fetchedFiles = Iterators.concat(fetchedFiles,
@@ -96,28 +121,6 @@ static public void parse(
                 exceptions
         );
 
-        Column col = project.columnModel.columns.get(0);
-        StandardReconConfig cfg = new StandardReconConfig(
-                service,
-                "https://commons.wikimedia.org/entity/",
-                "http://www.wikidata.org/prop/direct/",
-                "",
-                "entity",
-                true,
-                1,
-                new ArrayList<ColumnDetail>());
-        col.setReconStats(ReconStats.create(project, 0));
-        col.setReconConfig(cfg);
-        col.setName("File");
-        if (mIdsColumn) {
-            project.columnModel.columns.get(1).setName("M-ids");
-            if (categoriesColumn) {
-                project.columnModel.columns.get(2).setName("Categories");
-            }
-        } else if (categoriesColumn) {
-            project.columnModel.columns.get(1).setName("Categories");
-        }
-
         setProgress(job, categoriesWithDepth.get(categoriesWithDepth.size()-1).categoryName, 100);
     }
 

diff --git a/src/test/java/org/openrefine/extensions/commons/importer/CommonsImporterTest.java b/src/test/java/org/openrefine/extensions/commons/importer/CommonsImporterTest.java
@@ -70,4 +70,41 @@ public void testParse() throws Exception {
 
         }
     }
+
+    /**
+     * Test column names upon project creation as well as reconciled cells
+     */
+    @Test
+    public void testParseEmptyCategory() throws Exception {
+
+        try (MockWebServer server = new MockWebServer()) {
+            server.start();
+            HttpUrl url = server.url("/w/api.php");
+            String jsonResponse = "{\"batchcomplete\":\"\",\"query\":{\"categorymembers\":[]}}";
+            server.enqueue(new MockResponse().setBody(jsonResponse));
+            servlet = new RefineServlet();
+            ImportingManager.initialize(servlet);
+            ProjectManager.singleton = Mockito.mock(ProjectManager.class);
+            project = new Project();
+            metadata = new ProjectMetadata();
+            metadata.setName("Commons Import Test Project");
+            job = Mockito.mock(ImportingJob.class);
+            ObjectNode options = ParsingUtilities.evaluateJsonStringToObjectNode(
+                    "{\"categoryJsonValue\":[{\"category\":\"Category:Costa Rica\",\"depth\":\"0\"}],\"skipDataLines\":0,"
+                    + "\"limit\":-1,\"disableAutoPreview\":false,\"categoriesColumn\":true,\"mIdsColumn\":true}");
+            List<Exception> exceptions = new ArrayList<Exception>();
+            CommonsImporter importer = new CommonsImporter();
+
+            importer.setApiUrl(url.toString());
+            CommonsImporter.parse(project, metadata, job, 0, options, exceptions);
+            project.update();
+
+            Assert.assertEquals(project.rows.size(), 0);
+            Assert.assertEquals(project.columnModel.columns.get(0).getName(), "File");
+            Assert.assertEquals(project.columnModel.columns.get(1).getName(), "M-ids");
+            Assert.assertEquals(project.columnModel.columns.get(2).getName(), "Categories");
+            server.close();
+
+        }
+    }
 }