more fixes

loculus-project · Jan 17, 2025 · a899740 · a899740
1 parent d87c684
commit a899740
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 9 deletions.
diff --git a/website/src/components/Submission/FileUpload/ColumnMapping.spec.ts b/website/src/components/Submission/FileUpload/ColumnMapping.spec.ts
@@ -19,19 +19,20 @@ describe('ColumnMapping', () => {
     });
 
     it('should create a mapping from columns with sensible column mapping', () => {
-        const sourceColumns = ['state', 'geoLocAdmin2'];
+        const sourceColumns = ['state', 'geoLocAdmin', 'geoLocAdmin2'];
         const inputFields = [
             { name: 'date' },
-            { name: 'geoLocAdmin1', displayName: 'Collection subdivision level 1' },
             { name: 'geoLocAdmin2', displayName: 'Collection subdivision level 2' },
+            { name: 'geoLocAdmin1', displayName: 'Collection subdivision level 1' },
         ];
 
         const mapping = ColumnMapping.fromColumns(sourceColumns, inputFields);
         const entries = mapping.entries();
 
         expect(entries).toEqual([
-            ['state', null],
             ['geoLocAdmin2', 'geoLocAdmin2'],
+            ['state', null],
+            ['geoLocAdmin', 'geoLocAdmin1'],
         ]);
     });
 
@@ -71,16 +72,16 @@ describe('ColumnMapping', () => {
         const mapping = ColumnMapping.fromColumns(sourceColumns, inputFields);
         let entries = mapping.entries();
         expect(entries).toEqual([
-            ['loc', null],
             ['date', 'date'],
+            ['loc', null],
         ]);
 
         const updatedMapping = mapping.updateWith('loc', 'date');
 
         entries = updatedMapping.entries();
         expect(entries).toEqual([
-            ['loc', 'date'],
             ['date', null],
+            ['loc', 'date'],
         ]);
     });
 
@@ -97,6 +98,6 @@ describe('ColumnMapping', () => {
         const remappedFile = await updatedMapping.applyTo(new RawFile(tsvFile));
         const remappedContent = await remappedFile.text();
 
-        expect(remappedContent).toBe('location\tdate\n' + '"U\nS\nA"\t2023-01-01\n' + 'Canada\t2023-01-02\n');
+        expect(remappedContent).toBe('date\tlocation\n' + '2023-01-01\t"U\nS\nA"\n' + '2023-01-02\tCanada\n');
     });
 });
diff --git a/website/src/components/Submission/FileUpload/ColumnMapping.ts b/website/src/components/Submission/FileUpload/ColumnMapping.ts
@@ -9,6 +9,7 @@ export class ColumnMapping {
     private constructor(private readonly map: ReadonlyMap<string, string | null>) {}
 
     private static getBestMatchingTargetColumn(sourceColumn: string, inputFields: InputField[]): string | null {
+        if (inputFields.length === 0) return null;
         const [bestMatch, score] = inputFields
             .map((field): [string, number] => {
                 const score = Math.max(
@@ -25,24 +26,37 @@ export class ColumnMapping {
     public static fromColumns(sourceColumns: string[], inputFields: InputField[]) {
         const mapping = new Map();
         let availableFields = inputFields;
+        let remainingSourceColumns = sourceColumns;
+        // assign exact matches first
         sourceColumns.forEach((sourceColumn) => {
+            const foundField = availableFields.find(
+                (inputField) => inputField.name === sourceColumn || inputField.displayName === sourceColumn,
+            );
+            if (foundField) {
+                mapping.set(sourceColumn, foundField.name);
+                availableFields = availableFields.filter((f) => f.name !== sourceColumn);
+                remainingSourceColumns = remainingSourceColumns.filter((f) => f !== sourceColumn);
+            }
+        });
+        // do best effort matching second
+        remainingSourceColumns.forEach((sourceColumn) => {
             const bestMatch = this.getBestMatchingTargetColumn(sourceColumn, availableFields);
             mapping.set(sourceColumn, bestMatch);
             availableFields = availableFields.filter((field) => field.name !== bestMatch);
         });
         return new ColumnMapping(mapping);
     }
 
-    /* Update the mapping with new source and target columns, trying to keep as much of the 
-       mapping intact as possible. */
+    /* Update the mapping with new source and target columns, keeping previously mapped values. */
     public update(newSourceColumns: string[], newInputFields: InputField[]): ColumnMapping {
+        // keep entries that existed before
         const newMapping = new Map(
             newSourceColumns.map((newSourceCol) => {
                 const prevTargetCol = this.map.get(newSourceCol);
                 if (prevTargetCol && newInputFields.map((f) => f.name).includes(prevTargetCol)) {
                     return [newSourceCol, prevTargetCol];
                 } else {
-                    return [newSourceCol, ColumnMapping.getBestMatchingTargetColumn(newSourceCol, newInputFields)];
+                    return [newSourceCol, null];
                 }
             }),
         );