@natlibfi/melinda-record-matching 5.0.5 → 5.1.0

package/src/match-detection/features/auth/index.test.js

@@ -0,0 +1,47 @@
+
+import assert from 'node:assert';
+import {describe} from 'node:test';
+import createDebugLogger from 'debug';
+import generateTests from '@natlibfi/fixugen';
+import {READERS} from '@natlibfi/fixura';
+import {MarcRecord} from '@natlibfi/marc-record';
+import * as features from './index.js';
+
+
+const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/auth:test');
+const debugData = debug.extend('data');
+
+
+describe('match-detection/features/auth/', () => {
+  generateTests({
+    path: [import.meta.dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'auth'],
+    useMetadataFile: true,
+    fixura: {
+      reader: READERS.JSON
+    },
+
+    callback: ({feature, options, type, ...expectations}) => {
+      debug(`Testing: ${feature} ${type}`);
+
+      if (type === 'extract') {
+        const {expectedFeatures, inputRecord} = expectations;
+        const record = new MarcRecord(inputRecord, {subfieldValues: false});
+        debugData(`Record: ${record}`);
+        const {extract} = features[feature](options);
+
+        assert.deepStrictEqual(extract({record}), expectedFeatures);
+        return;
+      }
+
+      if (type === 'compare') {
+        const {featuresA, featuresB, expectedPoints} = expectations;
+        const {compare} = features[feature](options);
+
+        assert.equal(Math.round(compare(featuresA, featuresB) *100)/100, expectedPoints);
+        return;
+      }
+
+      throw new Error(`Invalid type ${type}`);
+    }
+  });
+});

package/src/match-detection/features/auth/yso.js

@@ -0,0 +1,53 @@
+
+import createDebugLogger from 'debug';
+import {recordGetAuthURX} from '../../../candidate-search/query-list/auth.js';
+
+const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:auth:yso');
+const debugData = debug.extend('data');
+
+function getSubjectHeadingThesaurus(record) {
+  // Quick'n'dirty: we could/should check 008/11 first. Now we just assume it is 'z'
+  const [f040] = record.get('040');
+  if (f040) {
+    // debug(`FOUND 040: ${fieldToString(f040)}`);
+    const sf = f040.subfields?.find(sf => sf.code === 'f');
+    if (sf) {
+      return sf.value;
+    }
+  }
+  return null;
+}
+
+export default () => ({
+  name: 'yso',
+  extract: ({record/*, recordExternal*/}) => {
+    const identifiers = recordGetAuthURX(record);
+
+    const thesaurus = getSubjectHeadingThesaurus(record);
+
+    debug(`EXTRACT ${thesaurus}: ${identifiers.join(', ')}`);
+    return {identifiers, thesaurus};
+  },
+
+  compare: (aa, bb) => {
+    const aIdentifiers = aa.identifiers;
+    const aThesaurus = aa.thesaurus;
+    const bIdentifiers = bb.identifiers;
+    const bThesaurus =  bb.thesaurus;
+
+    // Require identical existing thesauri
+    if (!aThesaurus || aThesaurus !== bThesaurus) {
+      return -1.0;
+    }
+    debugData(`Shared thesaurus: ${aThesaurus}`)
+
+    const sharedIdentifier = aIdentifiers.find(id => bIdentifiers.includes(id));
+    if (!sharedIdentifier) {
+      return -1.0;
+    }
+    debugData(`Shared identifier: ${sharedIdentifier} => MATCH`);
+
+    return 1.0;
+  }
+});
+

package/src/match-detection/features/bib/index.test.js

@@ -20,12 +20,7 @@ describe('match-detection/features/bib/', () => {
       reader: READERS.JSON
     },
 
-    callback: ({enabled = true, feature, options, type, ...expectations}) => {
-
-      if (!enabled) {
-        return;
-      }
-
+    callback: ({feature, options, type, ...expectations}) => {
       debug(`Testing: ${feature} ${type}`);
 
       if (type === 'extract') {

package/src/match-detection/features/bib/isbn.js

@@ -16,7 +16,6 @@ export default () => ({
 
     return record.get('020').filter(f => f.subfields?.some(sf => ['a', 'z'].includes(sf.code) && sf.value));
   },
-  // eslint-disable-next-line max-statements
   compare: (aa, bb) => {
     debugData(`Comparing ISBN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);
     if (aa.length === 0 || bb.length === 0) {

package/src/match-detection/features/bib/issn.js

@@ -37,7 +37,6 @@ export default () => ({
       return /^[0-9][0-9][0-9][0-9]-[0-9][0-9][0-9][0-9X]$/u.test(val);
     }
   },
-  // eslint-disable-next-line max-statements
   compare: (aa, bb) => {
     debugData(`Comparing ISSN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);
     if (aa.length === 0 || bb.length === 0) {

package/src/match-detection/features/bib/language.js

@@ -1,4 +1,3 @@
-
 import createDebugLogger from 'debug';
 
 import {MarcRecord} from '@natlibfi/marc-record';
@@ -37,7 +36,6 @@ export default () => ({
     // Should we return all the fields, or just the relevant fields?
     return [clonedRecord.fields, label];
   },
-  // eslint-disable-next-line max-statements
   compare: (aa, bb) => {
     const [a, aLabel] = aa;
     const [b, bLabel] = bb;

package/src/match-detection/features/bib/media-type.js

@@ -1,6 +1,3 @@
-/* eslint-disable max-statements */
-
-
 import createDebugLogger from 'debug';
 
 const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:media-type');

package/src/match-detection/features/bib/publication-time-allow-cons-years-multi.js

@@ -147,7 +147,6 @@ export default () => ({
     }
 
   },
-  // eslint-disable-next-line max-statements
   compare: (a, b) => {
     const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');
     debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);

package/src/match-detection/features/bib/standard-identifier-factory.js

@@ -1,5 +1,3 @@
-/* eslint-disable max-statements */
-
 import createDebugLogger from 'debug';
 import {extractSubfieldsFromField, uniqueSubfields} from '../../../matching-utils.js';
 

package/src/match-detection/features/index.js

@@ -1,3 +1,5 @@
 
+import * as auth from './auth/index.js';
 import * as bib from './bib/index.js';
-export {bib};
+
+export {auth, bib};

package/src/match-detection/index.test.js

@@ -20,13 +20,7 @@ describe('match-detection', () => {
     fixura: {
       reader: READERS.JSON
     },
-    callback: ({getFixture, options, expectedResults, array, enabled = true}) => {
-
-      if (!enabled) {
-        debug(`*** DISABLED TEST! ***`);
-        return;
-      }
-
+    callback: ({getFixture, options, expectedResults, array}) => {
       const detect = createDetectionInterface(formatOptions());
       const recordA = new MarcRecord(getFixture('recordA.json'), {subfieldValues: false});
       debugData(inspect(recordA));