@natlibfi/melinda-record-matching 2.2.1 → 3.0.0-alpha.1

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

@@ -33,22 +33,26 @@ import {testStringOrNumber} from '../../../matching-utils';
 
 export default ({nameTreshold = 10} = {}) => ({
   name: 'Authors',
-  extract: record => record.get(/^(?<def>100|700)$/u)
-    .map(({subfields}) => {
-      return subfields
-        .filter(({code, value}) => code && value)
-        .filter(({code}) => ['a', '0'].includes(code))
-        .map(toObj)
-        .reduce((acc, v) => ({...acc, ...v}), {});
-
-      function toObj({code, value}) {
-        if (code === 'a') {
-          return {name: testStringOrNumber(value) ? String(value).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''};
+  extract: ({record}) => {
+    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
+    const authors = record.get(/^(?<def>100|700)$/u)
+      .map(({subfields}) => {
+        return subfields
+          .filter(({code, value}) => code && value)
+          .filter(({code}) => ['a', '0'].includes(code))
+          .map(toObj)
+          .reduce((acc, v) => ({...acc, ...v}), {});
+
+        function toObj({code, value}) {
+          if (code === 'a') {
+            return {name: testStringOrNumber(value) ? String(value).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''};
+          }
+
+          return {id: value};
         }
-
-        return {id: value};
-      }
-    }),
+      });
+    return authors;
+  },
   compare: (a, b) => {
     const maxAuthors = a.length > b.length ? a.length : b.length;
     const matchingIds = findMatchingIds();

package/src/match-detection/features/bib/bibliographic-level.js

@@ -28,6 +28,6 @@
 
 export default () => ({
   name: 'Bibliographic level',
-  extract: r => r.leader[7] ? [r.leader[7]] : [],
+  extract: ({record}) => record.leader[7] ? [record.leader[7]] : [],
   compare: (a, b) => a[0] === b[0] ? 0.1 : -0.2
 });

package/src/match-detection/features/bib/host-component.js

@@ -28,6 +28,6 @@
 
 export default () => ({
   name: 'Host/Component record',
-  extract: r => r.get(/^773$/u).length > 0 ? ['component'] : ['host'],
+  extract: ({record}) => record.get(/^773$/u).length > 0 ? ['component'] : ['host'],
   compare: (a, b) => a[0] === b[0] ? 0.0 : -1.0
 });

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

@@ -52,7 +52,7 @@ describe('match-detection/features/bib/', () => {
         const record = new MarcRecord(inputRecord, {subfieldValues: false});
         const {extract} = features[feature](options);
 
-        expect(extract(record)).to.eql(expectedFeatures);
+        expect(extract({record})).to.eql(expectedFeatures);
         return;
       }
 

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

@@ -34,10 +34,12 @@ const debugData = debug.extend('data');
 
 export default () => ({
   name: 'Language',
-  extract: record => {
+  extract: ({record, recordExternal}) => {
+    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
+
     const value008 = get008Value();
     const values041 = get041Values();
-    debugData(`008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
+    debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
 
     if (value008 && values041.length > 0) {
       debugData(`There's both 008 and 041, searching for value in both`);

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

@@ -43,7 +43,7 @@ export default () => {
 
   return {extract, compare};
 
-  function extract(record) {
+  function extract({record, recordExternal}) {
 
     const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');
     const [f001] = record.get('001').map(field => field.value);
@@ -54,7 +54,7 @@ export default () => {
       f001 === undefined &&
       f035MelindaIds.length < 1) {
 
-      debug(`No Melinda-IDs found`);
+      debug(`${recordExternal.label} No Melinda-IDs found`);
       return {};
     }
 

package/src/match-detection/features/bib/publication-time.js

@@ -33,7 +33,7 @@ import {testStringOrNumber} from '../../../matching-utils';
 
 export default () => ({
   name: 'Publication time',
-  extract: record => {
+  extract: ({record}) => {
     const value = record.get(/^008$/u)?.[0]?.value || undefined;
     return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
   },

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

@@ -30,6 +30,6 @@
 
 export default () => ({
   name: 'Record type',
-  extract: r => r.leader[6] ? [r.leader[6]] : [],
+  extract: ({record}) => record.leader[6] ? [record.leader[6]] : [],
   compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5
 });

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

@@ -35,7 +35,7 @@ const debugData = debug.extend('data');
 export default ({pattern, subfieldCodes}) => {
   return {extract, compare};
 
-  function extract(record) {
+  function extract({record}) {
     const [field] = record.get(pattern);
 
     if (field) {

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

@@ -32,7 +32,7 @@ import {testStringOrNumber} from '../../../matching-utils';
 
 export default ({treshold = 10} = {}) => ({
   name: 'Title',
-  extract: record => {
+  extract: ({record}) => {
     const title = getTitle();
 
     if (testStringOrNumber(title)) {

package/src/match-detection/index.js

@@ -32,30 +32,38 @@ import * as features from './features';
 
 export {features};
 
-export default ({strategy, treshold = 0.9}, returnStrategy = false) => (recordA, recordB) => {
+export default ({strategy, treshold = 0.9}, returnStrategy = false) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {
   const minProbabilityQuantifier = 0.5;
 
   const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');
   const debugData = debug.extend('data');
 
-  debugData(`Strategy: ${JSON.stringify(strategy)}`);
-  debugData(`Treshold: ${JSON.stringify(treshold)}`);
-  debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
+  debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
+  debugData(`Records: A: ${recordA}\nB: ${recordB}`);
+  debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);
 
-  const featuresA = extractFeatures(recordA);
+  const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal});
 
   debug(`We got an array of records: ${Array.isArray(recordB)}`);
   const recordsB = Array.isArray(recordB) ? recordB : [recordB];
-  const detectionResults = recordsB.map(record => actualDetection(featuresA, record));
+  const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];
 
+  const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));
+
+  // if we got array of records, we return an array of result
+  // if we got a singular record, we return a singular result
   return Array.isArray(recordB) ? detectionResults : detectionResults[0];
 
-  function actualDetection(featuresA, record) {
+  function actualDetection({featuresA, record, recordExternal, index}) {
+    const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';
+    const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';
+
 
-    const featuresB = extractFeatures(record);
+    debug(`Actual detection for record ${index + 1} ${labelB}`);
+    const featuresB = extractFeatures({record, recordExternal});
 
-    debugData(`Features (a): ${JSON.stringify(featuresA)}`);
-    debugData(`Features (b): ${JSON.stringify(featuresB)}`);
+    debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);
+    debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);
 
     const featurePairs = generateFeaturePairs(featuresA, featuresB);
     const similarityVector = generateSimilarityVector(featurePairs);
@@ -70,8 +78,8 @@ export default ({strategy, treshold = 0.9}, returnStrategy = false) => (recordA,
     return returnResult({match: false, probability: 0.0});
   }
 
-  function extractFeatures(record) {
-    return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract(record)}), []);
+  function extractFeatures({record, recordExternal}) {
+    return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);
   }
 
   function returnResult(result) {

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

@@ -63,7 +63,7 @@ describe('match-detection', () => {
         : new MarcRecord(getFixture('recordB.json'), {subfieldValues: false});
       debugData(inspect(recordB));
 
-      const results = detect(recordA, recordB);
+      const results = detect({recordA, recordB});
       debugData(`${JSON.stringify(results)}`);
 
       expect(results).to.eql(expectedResults);