fix dups

hms-dbmi · Jul 18, 2024 · 69dc831 · 69dc831
1 parent d3d93c2
commit 69dc831
Show file tree

Hide file tree

Showing 4 changed files with 655 additions and 200 deletions.
diff --git a/out.json b/out.json
@@ -0,0 +1,169 @@
+[
+  {
+    "name": "study_ids_dataset_ids",
+    "display": "Study IDs/Dataset IDs",
+    "description": "",
+    "facets": [
+      {
+        "name": "1",
+        "display": "GIC",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "2",
+        "display": "National Health and Nutrition Examination Survey",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "3",
+        "display": "1000 Genomes Project",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs002715",
+        "display": "NSRR CFS",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs000284",
+        "display": "CFS",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs000007",
+        "display": "FHS",
+        "description": null,
+        "count": 3,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs000007",
+        "display": "FHS",
+        "description": null,
+        "count": 1,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs002385",
+        "display": "HCT_for_SCD",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs003463",
+        "display": "RECOVER_Adult",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs003543",
+        "display": "NSRR_HSHC",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs002808",
+        "display": "nuMoM2b",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs003566",
+        "display": "SPRINT",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      },
+      {
+        "name": "phs001963",
+        "display": "DEMENTIA-SEQ",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "study_ids_dataset_ids",
+        "meta": null
+      }
+    ]
+  },
+  {
+    "name": "nsrr_harmonized",
+    "display": "Common Data Element Collection",
+    "description": "",
+    "facets": [
+      {
+        "name": "LOINC",
+        "display": "LOINC",
+        "description": null,
+        "count": 1,
+        "children": null,
+        "category": "nsrr_harmonized",
+        "meta": null
+      },
+      {
+        "name": "PhenX",
+        "display": "PhenX",
+        "description": null,
+        "count": 1,
+        "children": null,
+        "category": "nsrr_harmonized",
+        "meta": null
+      },
+      {
+        "name": "gad_7",
+        "display": "Generalized Anxiety Disorder Assessment (GAD-7)",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "nsrr_harmonized",
+        "meta": null
+      },
+      {
+        "name": "taps_tool",
+        "display": "NIDA CTN Common Data Elements = TAPS Tool",
+        "description": null,
+        "count": 0,
+        "children": null,
+        "category": "nsrr_harmonized",
+        "meta": null
+      }
+    ]
+  }
+]
diff --git a/src/main/java/edu/harvard/dbmi/avillach/dictionary/facet/FacetQueryGenerator.java b/src/main/java/edu/harvard/dbmi/avillach/dictionary/facet/FacetQueryGenerator.java
@@ -184,24 +184,27 @@ private String createMultiCategorySQLNoSearch(Map<String, List<Facet>> facets, M
            and INTERSECT them. This creates the concepts for this category
          */
         String selectedFacetsQuery = facets.keySet().stream().map(category -> {
-                String allConceptsForCategory = categoryKeys.values().stream()
-                    .filter(key -> !categoryKeys.get(category).equals(key))
-                    .map(key -> "SELECT * FROM facet_category_" + key + "_concepts")
-                    .collect(Collectors.joining(" UNION "));
-                params.addValue("", "");
-                return """
-                    (
-                        SELECT
-                            facet.facet_id, count(*) as facet_count
-                        FROM
-                            facet
-                            JOIN facet__concept_node fcn ON fcn.facet_id = facet.facet_id
-                        WHERE
-                            fcn.concept_node_id IN (%s)
-                        GROUP BY
-                            facet.facet_id
-                    )
-                    """.formatted(allConceptsForCategory);
+            params.addValue("facet_category_" + categoryKeys.get(category), category);
+            String allConceptsForCategory = categoryKeys.values().stream()
+                .filter(key -> !categoryKeys.get(category).equals(key))
+                .map(key -> "SELECT * FROM facet_category_" + key + "_concepts")
+                .collect(Collectors.joining(" UNION "));
+            params.addValue("", "");
+            return """
+                (
+                    SELECT
+                        facet.facet_id, count(*) as facet_count
+                    FROM
+                        facet
+                        JOIN facet__concept_node fcn ON fcn.facet_id = facet.facet_id
+                        JOIN facet_category fc on fc.facet_category_id = facet.facet_category_id
+                    WHERE
+                        fcn.concept_node_id IN (%s)
+                        AND fc.name = :facet_category_%s
+                    GROUP BY
+                        facet.facet_id
+                )
+                """.formatted(allConceptsForCategory, categoryKeys.get(category));
             })
             .collect(Collectors.joining("\n\tUNION\n"));
 

diff --git a/src/test/java/edu/harvard/dbmi/avillach/dictionary/facet/FacetQueryGeneratorTest.java b/src/test/java/edu/harvard/dbmi/avillach/dictionary/facet/FacetQueryGeneratorTest.java
@@ -20,7 +20,7 @@ class FacetQueryGeneratorTest {
     void shouldaaa() {
         Filter jim = new Filter(List.of(
             new Facet("phs000007", "", "", null, null, "study_ids_dataset_ids", null),
-            new Facet("PhenX", "", "", null, null, "nsrr_harmonized", null)
+            new Facet("LOINC", "", "", null, null, "nsrr_harmonized", null)
         ), "");
         String query = generator.createFacetSQLAndPopulateParams(jim, new MapSqlParameterSource());
         System.out.println(query);