@natlibfi/melinda-record-matching 4.4.0-alpha.1 → 5.0.0-alpha.2

package/dist/match-detection/features/index.js.map

@@ -1 +1,7 @@
-{"version":3,"file":"index.js","names":["bib","_interopRequireWildcard","require","exports","_getRequireWildcardCache","e","WeakMap","r","t","__esModule","default","has","get","n","__proto__","a","Object","defineProperty","getOwnPropertyDescriptor","u","prototype","hasOwnProperty","call","i","set"],"sources":["../../../src/match-detection/features/index.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport * as bib from './bib';\nexport {bib};\n"],"mappings":";;;;;;AA4BA,IAAAA,GAAA,GAAAC,uBAAA,CAAAC,OAAA;AAA6BC,OAAA,CAAAH,GAAA,GAAAA,GAAA;AAAA,SAAAI,yBAAAC,CAAA,6BAAAC,OAAA,mBAAAC,CAAA,OAAAD,OAAA,IAAAE,CAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,CAAA,WAAAA,CAAA,GAAAG,CAAA,GAAAD,CAAA,KAAAF,CAAA;AAAA,SAAAJ,wBAAAI,CAAA,EAAAE,CAAA,SAAAA,CAAA,IAAAF,CAAA,IAAAA,CAAA,CAAAI,UAAA,SAAAJ,CAAA,eAAAA,CAAA,uBAAAA,CAAA,yBAAAA,CAAA,WAAAK,OAAA,EAAAL,CAAA,QAAAG,CAAA,GAAAJ,wBAAA,CAAAG,CAAA,OAAAC,CAAA,IAAAA,CAAA,CAAAG,GAAA,CAAAN,CAAA,UAAAG,CAAA,CAAAI,GAAA,CAAAP,CAAA,OAAAQ,CAAA,KAAAC,SAAA,UAAAC,CAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,CAAA,IAAAd,CAAA,oBAAAc,CAAA,IAAAH,MAAA,CAAAI,SAAA,CAAAC,cAAA,CAAAC,IAAA,CAAAjB,CAAA,EAAAc,CAAA,SAAAI,CAAA,GAAAR,CAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAb,CAAA,EAAAc,CAAA,UAAAI,CAAA,KAAAA,CAAA,CAAAX,GAAA,IAAAW,CAAA,CAAAC,GAAA,IAAAR,MAAA,CAAAC,cAAA,CAAAJ,CAAA,EAAAM,CAAA,EAAAI,CAAA,IAAAV,CAAA,CAAAM,CAAA,IAAAd,CAAA,CAAAc,CAAA,YAAAN,CAAA,CAAAH,OAAA,GAAAL,CAAA,EAAAG,CAAA,IAAAA,CAAA,CAAAgB,GAAA,CAAAnB,CAAA,EAAAQ,CAAA,GAAAA,CAAA"}
+{
+  "version": 3,
+  "sources": ["../../../src/match-detection/features/index.js"],
+  "sourcesContent": ["\nimport * as bib from './bib/index.js';\nexport {bib};\n"],
+  "mappings": "AACA,YAAY,SAAS;AACrB,SAAQ;",
+  "names": []
+}

package/dist/match-detection/index.js

@@ -1,187 +1,77 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.features = exports.default = void 0;
-var _debug = _interopRequireDefault(require("debug"));
-var features = _interopRequireWildcard(require("./features"));
-exports.features = features;
-function _getRequireWildcardCache(e) { if ("function" != typeof WeakMap) return null; var r = new WeakMap(), t = new WeakMap(); return (_getRequireWildcardCache = function (e) { return e ? t : r; })(e); }
-function _interopRequireWildcard(e, r) { if (!r && e && e.__esModule) return e; if (null === e || "object" != typeof e && "function" != typeof e) return { default: e }; var t = _getRequireWildcardCache(r); if (t && t.has(e)) return t.get(e); var n = { __proto__: null }, a = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var u in e) if ("default" !== u && Object.prototype.hasOwnProperty.call(e, u)) { var i = a ? Object.getOwnPropertyDescriptor(e, u) : null; i && (i.get || i.set) ? Object.defineProperty(n, u, i) : n[u] = e[u]; } return n.default = e, t && t.set(e, n), n; }
-function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
-/**
-*
-* @licstart  The following is the entire license notice for the JavaScript code in this file.
-*
-* Melinda record matching modules for Javascript
-*
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
-*
-* This file is part of melinda-record-matching-js
-*
-* melinda-record-matching-js program is free software: you can redistribute it and/or modify
-* it under the terms of the GNU Lesser General Public License as
-* published by the Free Software Foundation, either version 3 of the
-* License, or (at your option) any later version.
-*
-* melinda-record-matching-js is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-* GNU Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Affero General Public License
-* along with this program.  If not, see <http://www.gnu.org/licenses/>.
-*
-* @licend  The above is the entire license notice
-* for the JavaScript code in this file.
-*
-*/
-var _default = ({
-  strategy,
-  treshold = 0.9
-}, returnStrategy = false, localSettings = {}) => ({
-  recordA,
-  recordB,
-  recordAExternal = {},
-  recordBExternal = {}
-}) => {
+import createDebugLogger from "debug";
+import * as features from "./features/index.js";
+export { features };
+export default ({ strategy, treshold = 0.9 }, returnStrategy = false, localSettings = {}) => ({ recordA, recordB, recordAExternal = {}, recordBExternal = {} }) => {
   const minProbabilityQuantifier = 0.5;
-  const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection');
-  const debugData = debug.extend('data');
+  const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection");
+  const debugData = debug.extend("data");
   debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
-  debugData(`Records: A: ${recordA}\nB: ${recordB}`);
+  debugData(`Records: A: ${recordA}
+B: ${recordB}`);
   debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);
-  // We could add here labels for records if we didn't get external labels
-
-  const featuresA = extractFeatures({
-    record: recordA,
-    recordExternal: recordAExternal,
-    localSettings
-  });
+  const featuresA = extractFeatures({ record: recordA, recordExternal: recordAExternal, localSettings });
   debug(`We got an array of records: ${Array.isArray(recordB)}`);
   const recordsB = Array.isArray(recordB) ? recordB : [recordB];
   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
+  const detectionResults = recordsB.map((record, index) => actualDetection({ featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index }));
   return Array.isArray(recordB) ? detectionResults : detectionResults[0];
-  function actualDetection({
-    featuresA,
-    record,
-    recordExternal,
-    index,
-    localSettings
-  }) {
-    const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';
-    const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';
+  function actualDetection({ featuresA: featuresA2, record, recordExternal, index, localSettings: localSettings2 }) {
+    const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : "a";
+    const labelB = recordExternal && recordExternal.label ? recordExternal.label : "b";
     debug(`Actual detection for record ${index + 1} ${labelB}`);
-    const featuresB = extractFeatures({
-      record,
-      recordExternal,
-      localSettings
-    });
-    debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);
+    const featuresB = extractFeatures({ record, recordExternal, localSettings: localSettings2 });
+    debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA2)}`);
     debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);
-    const featurePairs = generateFeaturePairs(featuresA, featuresB);
+    const featurePairs = generateFeaturePairs(featuresA2, featuresB);
     const similarityVector = generateSimilarityVector(featurePairs);
-    if (similarityVector.some(v => v >= minProbabilityQuantifier)) {
+    if (similarityVector.some((v) => v >= minProbabilityQuantifier)) {
       const probability = calculateProbability(similarityVector);
       debug(`probability: ${probability} (Treshold: ${treshold})`);
-      return returnResult({
-        match: probability >= treshold,
-        probability
-      });
+      return returnResult({ match: probability >= treshold, probability });
     }
     debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);
-    return returnResult({
-      match: false,
-      probability: 0.0
-    });
+    return returnResult({ match: false, probability: 0 });
   }
-  function extractFeatures({
-    record,
-    recordExternal
-  }) {
-    return strategy.reduce((acc, {
-      name,
-      extract
-    }) => acc.concat({
-      name,
-      value: extract({
-        record,
-        recordExternal
-      })
-    }), []);
+  function extractFeatures({ record, recordExternal }) {
+    return strategy.reduce((acc, { name, extract }) => acc.concat({ name, value: extract({ record, recordExternal }) }), []);
   }
   function returnResult(result) {
     if (returnStrategy) {
       debug(`Returning detection strategy with the result`);
-      const resultWithStrategy = {
-        match: result.match,
-        probability: result.probability,
-        strategy: formatStrategy(strategy),
-        treshold
-      };
+      const resultWithStrategy = { match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold };
       debugData(`${JSON.stringify(resultWithStrategy)}`);
       return resultWithStrategy;
     }
     return result;
   }
-  function formatStrategy(strategy) {
-    const strategyNames = strategy.map(element => element.name);
+  function formatStrategy(strategy2) {
+    const strategyNames = strategy2.map((element) => element.name);
     return strategyNames || [];
   }
   function calculateProbability(similarityVector) {
-    const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);
-    return probability > 1.0 ? 1.0 : probability;
+    const probability = similarityVector.reduce((acc, v) => acc + v, 0);
+    return probability > 1 ? 1 : probability;
   }
   function generateSimilarityVector(featurePairs) {
-    const compared = featurePairs.map(({
-      name,
-      a,
-      b
-    }) => {
-      const {
-        compare
-      } = strategy.find(({
-        name: featureName
-      }) => name === featureName);
+    const compared = featurePairs.map(({ name, a, b }) => {
+      const { compare } = strategy.find(({ name: featureName }) => name === featureName);
       const points = compare(a, b);
-      return {
-        name,
-        points
-      };
+      return { name, points };
     });
     debugData(`Points: ${JSON.stringify(compared)}`);
-    return compared.map(({
-      points
-    }) => points);
+    return compared.map(({ points }) => points);
   }
-  function generateFeaturePairs(featuresA, featuresB) {
+  function generateFeaturePairs(featuresA2, featuresB) {
     const pairs = generatePairs();
     const missingFeatures = findMissing();
     debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);
     return pairs;
     function generatePairs() {
-      return featuresA.reduce((acc, {
-        name,
-        value
-      }, index) => acc.concat({
+      return featuresA2.reduce((acc, { name, value }, index) => acc.concat({
         name,
         a: value,
         b: featuresB[index].value
-      }), []).filter(({
-        a,
-        b
-      }) => {
+      }), []).filter(({ a, b }) => {
         if (a.length === 0 || b.length === 0) {
           return false;
         }
@@ -189,13 +79,8 @@ var _default = ({
       });
     }
     function findMissing() {
-      return featuresA.map(({
-        name
-      }) => name).filter(v => pairs.some(({
-        name
-      }) => name === v) === false);
+      return featuresA2.map(({ name }) => name).filter((v) => pairs.some(({ name }) => name === v) === false);
     }
   }
 };
-exports.default = _default;
-//# sourceMappingURL=index.js.map
+//# sourceMappingURL=index.js.map

package/dist/match-detection/index.js.map

@@ -1 +1,7 @@
-{"version":3,"file":"index.js","names":["_debug","_interopRequireDefault","require","features","_interopRequireWildcard","exports","_getRequireWildcardCache","e","WeakMap","r","t","__esModule","default","has","get","n","__proto__","a","Object","defineProperty","getOwnPropertyDescriptor","u","prototype","hasOwnProperty","call","i","set","obj","_default","strategy","treshold","returnStrategy","localSettings","recordA","recordB","recordAExternal","recordBExternal","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","record","recordExternal","Array","isArray","recordsB","recordsBExternal","detectionResults","map","index","actualDetection","labelA","label","labelB","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateProbability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport * as features from './features';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false, localSettings = {}) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {\n  const minProbabilityQuantifier = 0.5;\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n  const debugData = debug.extend('data');\n\n  debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n  debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n  debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n  // We could add here labels for records if we didn't get external labels\n\n  const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal, localSettings});\n\n  debug(`We got an array of records: ${Array.isArray(recordB)}`);\n  const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n  const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n  const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n  // if we got array of records, we return an array of result\n  // if we got a singular record, we return a singular result\n  return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n  function actualDetection({featuresA, record, recordExternal, index, localSettings}) {\n    const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n    const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n    debug(`Actual detection for record ${index + 1} ${labelB}`);\n    const featuresB = extractFeatures({record, recordExternal, localSettings});\n\n    debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n    debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n    const featurePairs = generateFeaturePairs(featuresA, featuresB);\n    const similarityVector = generateSimilarityVector(featurePairs);\n\n    if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n      const probability = calculateProbability(similarityVector);\n      debug(`probability: ${probability} (Treshold: ${treshold})`);\n      return returnResult({match: probability >= treshold, probability});\n    }\n\n    debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n    return returnResult({match: false, probability: 0.0});\n  }\n\n  function extractFeatures({record, recordExternal}) {\n    return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n  }\n\n  function returnResult(result) {\n    if (returnStrategy) {\n      debug(`Returning detection strategy with the result`);\n      const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n      debugData(`${JSON.stringify(resultWithStrategy)}`);\n      return resultWithStrategy;\n    }\n    return result;\n  }\n\n  function formatStrategy(strategy) {\n    const strategyNames = strategy.map(element => element.name);\n    return strategyNames || [];\n  }\n\n  function calculateProbability(similarityVector) {\n    const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n    return probability > 1.0 ? 1.0 : probability;\n  }\n\n  function generateSimilarityVector(featurePairs) {\n    const compared = featurePairs.map(({name, a, b}) => {\n      const {compare} = strategy.find(({name: featureName}) => name === featureName);\n      const points = compare(a, b);\n      return {name, points};\n    });\n\n    debugData(`Points: ${JSON.stringify(compared)}`);\n    return compared.map(({points}) => points);\n  }\n\n  function generateFeaturePairs(featuresA, featuresB) {\n    const pairs = generatePairs();\n    const missingFeatures = findMissing();\n\n    debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n    return pairs;\n\n    function generatePairs() {\n      return featuresA\n        .reduce((acc, {name, value}, index) => acc.concat({\n          name,\n          a: value,\n          b: featuresB[index].value\n        }), [])\n        .filter(({a, b}) => {\n          if (a.length === 0 || b.length === 0) {\n            return false;\n          }\n\n          return true;\n        });\n    }\n\n    function findMissing() {\n      return featuresA\n        .map(({name}) => name)\n        .filter(v => pairs.some(({name}) => name === v) === false);\n    }\n  }\n\n};\n"],"mappings":";;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,QAAA,GAAAC,uBAAA,CAAAF,OAAA;AAAuCG,OAAA,CAAAF,QAAA,GAAAA,QAAA;AAAA,SAAAG,yBAAAC,CAAA,6BAAAC,OAAA,mBAAAC,CAAA,OAAAD,OAAA,IAAAE,CAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,CAAA,WAAAA,CAAA,GAAAG,CAAA,GAAAD,CAAA,KAAAF,CAAA;AAAA,SAAAH,wBAAAG,CAAA,EAAAE,CAAA,SAAAA,CAAA,IAAAF,CAAA,IAAAA,CAAA,CAAAI,UAAA,SAAAJ,CAAA,eAAAA,CAAA,uBAAAA,CAAA,yBAAAA,CAAA,WAAAK,OAAA,EAAAL,CAAA,QAAAG,CAAA,GAAAJ,wBAAA,CAAAG,CAAA,OAAAC,CAAA,IAAAA,CAAA,CAAAG,GAAA,CAAAN,CAAA,UAAAG,CAAA,CAAAI,GAAA,CAAAP,CAAA,OAAAQ,CAAA,KAAAC,SAAA,UAAAC,CAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,CAAA,IAAAd,CAAA,oBAAAc,CAAA,IAAAH,MAAA,CAAAI,SAAA,CAAAC,cAAA,CAAAC,IAAA,CAAAjB,CAAA,EAAAc,CAAA,SAAAI,CAAA,GAAAR,CAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAb,CAAA,EAAAc,CAAA,UAAAI,CAAA,KAAAA,CAAA,CAAAX,GAAA,IAAAW,CAAA,CAAAC,GAAA,IAAAR,MAAA,CAAAC,cAAA,CAAAJ,CAAA,EAAAM,CAAA,EAAAI,CAAA,IAAAV,CAAA,CAAAM,CAAA,IAAAd,CAAA,CAAAc,CAAA,YAAAN,CAAA,CAAAH,OAAA,GAAAL,CAAA,EAAAG,CAAA,IAAAA,CAAA,CAAAgB,GAAA,CAAAnB,CAAA,EAAAQ,CAAA,GAAAA,CAAA;AAAA,SAAAd,uBAAA0B,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAhB,UAAA,GAAAgB,GAAA,KAAAf,OAAA,EAAAe,GAAA;AA7BvC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,IAAAC,QAAA,GAiCeA,CAAC;EAACC,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,EAAEC,aAAa,GAAG,CAAC,CAAC,KAAK,CAAC;EAACC,OAAO;EAAEC,OAAO;EAAEC,eAAe,GAAG,CAAC,CAAC;EAAEC,eAAe,GAAG,CAAC;AAAC,CAAC,KAAK;EAC7J,MAAMC,wBAAwB,GAAG,GAAG;EAEpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,mDAAmD,CAAC;EACpF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtCD,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACd,QAAQ,CAAE,eAAca,IAAI,CAACC,SAAS,CAACb,QAAQ,CAAE,qBAAoBY,IAAI,CAACC,SAAS,CAACZ,cAAc,CAAE,EAAC,CAAC;EAC5IS,SAAS,CAAE,eAAcP,OAAQ,QAAOC,OAAQ,EAAC,CAAC;EAClDI,KAAK,CAAE,iBAAgBI,IAAI,CAACC,SAAS,CAACR,eAAe,CAAE,QAAOO,IAAI,CAACC,SAAS,CAACP,eAAe,CAAE,EAAC,CAAC;EAChG;;EAEA,MAAMQ,SAAS,GAAGC,eAAe,CAAC;IAACC,MAAM,EAAEb,OAAO;IAAEc,cAAc,EAAEZ,eAAe;IAAEH;EAAa,CAAC,CAAC;EAEpGM,KAAK,CAAE,+BAA8BU,KAAK,CAACC,OAAO,CAACf,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMgB,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMiB,gBAAgB,GAAGH,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGE,eAAe,GAAG,CAACA,eAAe,CAAC;EAErF,MAAMgB,gBAAgB,GAAGF,QAAQ,CAACG,GAAG,CAAC,CAACP,MAAM,EAAEQ,KAAK,KAAKC,eAAe,CAAC;IAACX,SAAS;IAAET,eAAe;IAAEW,MAAM;IAAEC,cAAc,EAAEI,gBAAgB,CAACG,KAAK,CAAC;IAAEA;EAAK,CAAC,CAAC,CAAC;;EAE/J;EACA;EACA,OAAON,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGkB,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAeA,CAAC;IAACX,SAAS;IAAEE,MAAM;IAAEC,cAAc;IAAEO,KAAK;IAAEtB;EAAa,CAAC,EAAE;IAClF,MAAMwB,MAAM,GAAGrB,eAAe,IAAIA,eAAe,CAACsB,KAAK,GAAGtB,eAAe,CAACsB,KAAK,GAAG,GAAG;IACrF,MAAMC,MAAM,GAAGX,cAAc,IAAIA,cAAc,CAACU,KAAK,GAAGV,cAAc,CAACU,KAAK,GAAG,GAAG;IAGlFnB,KAAK,CAAE,+BAA8BgB,KAAK,GAAG,CAAE,IAAGI,MAAO,EAAC,CAAC;IAC3D,MAAMC,SAAS,GAAGd,eAAe,CAAC;MAACC,MAAM;MAAEC,cAAc;MAAEf;IAAa,CAAC,CAAC;IAE1EQ,SAAS,CAAE,gBAAegB,MAAO,MAAKd,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IAClEJ,SAAS,CAAE,gBAAekB,MAAO,MAAKhB,IAAI,CAACC,SAAS,CAACgB,SAAS,CAAE,EAAC,CAAC;IAElE,MAAMC,YAAY,GAAGC,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAI5B,wBAAwB,CAAC,EAAE;MAC7D,MAAM6B,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DxB,KAAK,CAAE,gBAAe4B,WAAY,eAAcpC,QAAS,GAAE,CAAC;MAC5D,OAAOsC,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAIpC,QAAQ;QAAEoC;MAAW,CAAC,CAAC;IACpE;IAEA1B,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAO+B,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASrB,eAAeA,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,EAAE;IACjD,OAAOlB,QAAQ,CAACyC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAC;QAAC3B,MAAM;QAAEC;MAAc,CAAC;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EACpH;EAEA,SAASqB,YAAYA,CAACQ,MAAM,EAAE;IAC5B,IAAI7C,cAAc,EAAE;MAClBO,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMuC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAErC,QAAQ,EAAEiD,cAAc,CAACjD,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HU,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAACkC,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAcA,CAACjD,QAAQ,EAAE;IAChC,MAAMkD,aAAa,GAAGlD,QAAQ,CAACwB,GAAG,CAAC2B,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,oBAAoBA,CAACL,gBAAgB,EAAE;IAC9C,MAAMI,WAAW,GAAGJ,gBAAgB,CAACQ,MAAM,CAAC,CAACC,GAAG,EAAEN,CAAC,KAAKM,GAAG,GAAGN,CAAC,EAAE,GAAG,CAAC;IACrE,OAAOC,WAAW,GAAG,GAAG,GAAG,GAAG,GAAGA,WAAW;EAC9C;EAEA,SAASH,wBAAwBA,CAACH,YAAY,EAAE;IAC9C,MAAMqB,QAAQ,GAAGrB,YAAY,CAACP,GAAG,CAAC,CAAC;MAACmB,IAAI;MAAEvD,CAAC;MAAEiE;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAGtD,QAAQ,CAACuD,IAAI,CAAC,CAAC;QAACZ,IAAI,EAAEa;MAAW,CAAC,KAAKb,IAAI,KAAKa,WAAW,CAAC;MAC9E,MAAMC,MAAM,GAAGH,OAAO,CAAClE,CAAC,EAAEiE,CAAC,CAAC;MAC5B,OAAO;QAACV,IAAI;QAAEc;MAAM,CAAC;IACvB,CAAC,CAAC;IAEF9C,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACsC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAAC5B,GAAG,CAAC,CAAC;MAACiC;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAASzB,oBAAoBA,CAACjB,SAAS,EAAEe,SAAS,EAAE;IAClD,MAAM4B,KAAK,GAAGC,aAAa,CAAC,CAAC;IAC7B,MAAMC,eAAe,GAAGC,WAAW,CAAC,CAAC;IAErCpD,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAAC8C,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAaA,CAAA,EAAG;MACvB,OAAO5C,SAAS,CACb0B,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAErB,KAAK,KAAKiB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJvD,CAAC,EAAE0D,KAAK;QACRO,CAAC,EAAEvB,SAAS,CAACL,KAAK,CAAC,CAACqB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNgB,MAAM,CAAC,CAAC;QAAC1E,CAAC;QAAEiE;MAAC,CAAC,KAAK;QAClB,IAAIjE,CAAC,CAAC2E,MAAM,KAAK,CAAC,IAAIV,CAAC,CAACU,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASF,WAAWA,CAAA,EAAG;MACrB,OAAO9C,SAAS,CACbS,GAAG,CAAC,CAAC;QAACmB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBmB,MAAM,CAAC1B,CAAC,IAAIsB,KAAK,CAACvB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA5D,OAAA,CAAAO,OAAA,GAAAgB,QAAA"}
+{
+  "version": 3,
+  "sources": ["../../src/match-detection/index.js"],
+  "sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport * as features from './features/index.js';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false, localSettings = {}) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {\n  const minProbabilityQuantifier = 0.5;\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n  const debugData = debug.extend('data');\n\n  debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n  debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n  debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n  // We could add here labels for records if we didn't get external labels\n\n  const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal, localSettings});\n\n  debug(`We got an array of records: ${Array.isArray(recordB)}`);\n  const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n  const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n  const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n  // if we got array of records, we return an array of result\n  // if we got a singular record, we return a singular result\n  return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n  function actualDetection({featuresA, record, recordExternal, index, localSettings}) {\n    const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n    const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n    debug(`Actual detection for record ${index + 1} ${labelB}`);\n    const featuresB = extractFeatures({record, recordExternal, localSettings});\n\n    debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n    debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n    const featurePairs = generateFeaturePairs(featuresA, featuresB);\n    const similarityVector = generateSimilarityVector(featurePairs);\n\n    if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n      const probability = calculateProbability(similarityVector);\n      debug(`probability: ${probability} (Treshold: ${treshold})`);\n      return returnResult({match: probability >= treshold, probability});\n    }\n\n    debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n    return returnResult({match: false, probability: 0.0});\n  }\n\n  function extractFeatures({record, recordExternal}) {\n    return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n  }\n\n  function returnResult(result) {\n    if (returnStrategy) {\n      debug(`Returning detection strategy with the result`);\n      const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n      debugData(`${JSON.stringify(resultWithStrategy)}`);\n      return resultWithStrategy;\n    }\n    return result;\n  }\n\n  function formatStrategy(strategy) {\n    const strategyNames = strategy.map(element => element.name);\n    return strategyNames || [];\n  }\n\n  function calculateProbability(similarityVector) {\n    const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n    return probability > 1.0 ? 1.0 : probability;\n  }\n\n  function generateSimilarityVector(featurePairs) {\n    const compared = featurePairs.map(({name, a, b}) => {\n      const {compare} = strategy.find(({name: featureName}) => name === featureName);\n      const points = compare(a, b);\n      return {name, points};\n    });\n\n    debugData(`Points: ${JSON.stringify(compared)}`);\n    return compared.map(({points}) => points);\n  }\n\n  function generateFeaturePairs(featuresA, featuresB) {\n    const pairs = generatePairs();\n    const missingFeatures = findMissing();\n\n    debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n    return pairs;\n\n    function generatePairs() {\n      return featuresA\n        .reduce((acc, {name, value}, index) => acc.concat({\n          name,\n          a: value,\n          b: featuresB[index].value\n        }), [])\n        .filter(({a, b}) => {\n          if (a.length === 0 || b.length === 0) {\n            return false;\n          }\n\n          return true;\n        });\n    }\n\n    function findMissing() {\n      return featuresA\n        .map(({name}) => name)\n        .filter(v => pairs.some(({name}) => name === v) === false);\n    }\n  }\n\n};\n"],
+  "mappings": "AACA,OAAO,uBAAuB;AAC9B,YAAY,cAAc;AAE1B,SAAQ;AAER,eAAe,CAAC,EAAC,UAAU,WAAW,IAAG,GAAG,iBAAiB,OAAO,gBAAgB,CAAC,MAAM,CAAC,EAAC,SAAS,SAAS,kBAAkB,CAAC,GAAG,kBAAkB,CAAC,EAAC,MAAM;AAC7J,QAAM,2BAA2B;AAEjC,QAAM,QAAQ,kBAAkB,mDAAmD;AACnF,QAAM,YAAY,MAAM,OAAO,MAAM;AAErC,YAAU,aAAa,KAAK,UAAU,QAAQ,CAAC,eAAe,KAAK,UAAU,QAAQ,CAAC,qBAAqB,KAAK,UAAU,cAAc,CAAC,EAAE;AAC3I,YAAU,eAAe,OAAO;AAAA,KAAQ,OAAO,EAAE;AACjD,QAAM,iBAAiB,KAAK,UAAU,eAAe,CAAC,QAAQ,KAAK,UAAU,eAAe,CAAC,EAAE;AAG/F,QAAM,YAAY,gBAAgB,EAAC,QAAQ,SAAS,gBAAgB,iBAAiB,cAAa,CAAC;AAEnG,QAAM,+BAA+B,MAAM,QAAQ,OAAO,CAAC,EAAE;AAC7D,QAAM,WAAW,MAAM,QAAQ,OAAO,IAAI,UAAU,CAAC,OAAO;AAC5D,QAAM,mBAAmB,MAAM,QAAQ,OAAO,IAAI,kBAAkB,CAAC,eAAe;AAEpF,QAAM,mBAAmB,SAAS,IAAI,CAAC,QAAQ,UAAU,gBAAgB,EAAC,WAAW,iBAAiB,QAAQ,gBAAgB,iBAAiB,KAAK,GAAG,MAAK,CAAC,CAAC;AAI9J,SAAO,MAAM,QAAQ,OAAO,IAAI,mBAAmB,iBAAiB,CAAC;AAErE,WAAS,gBAAgB,EAAC,WAAAA,YAAW,QAAQ,gBAAgB,OAAO,eAAAC,eAAa,GAAG;AAClF,UAAM,SAAS,mBAAmB,gBAAgB,QAAQ,gBAAgB,QAAQ;AAClF,UAAM,SAAS,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAG/E,UAAM,+BAA+B,QAAQ,CAAC,IAAI,MAAM,EAAE;AAC1D,UAAM,YAAY,gBAAgB,EAAC,QAAQ,gBAAgB,eAAAA,eAAa,CAAC;AAEzE,cAAU,gBAAgB,MAAM,MAAM,KAAK,UAAUD,UAAS,CAAC,EAAE;AACjE,cAAU,gBAAgB,MAAM,MAAM,KAAK,UAAU,SAAS,CAAC,EAAE;AAEjE,UAAM,eAAe,qBAAqBA,YAAW,SAAS;AAC9D,UAAM,mBAAmB,yBAAyB,YAAY;AAE9D,QAAI,iBAAiB,KAAK,OAAK,KAAK,wBAAwB,GAAG;AAC7D,YAAM,cAAc,qBAAqB,gBAAgB;AACzD,YAAM,gBAAgB,WAAW,eAAe,QAAQ,GAAG;AAC3D,aAAO,aAAa,EAAC,OAAO,eAAe,UAAU,YAAW,CAAC;AAAA,IACnE;AAEA,cAAU,4DAA4D,wBAAwB,GAAG;AACjG,WAAO,aAAa,EAAC,OAAO,OAAO,aAAa,EAAG,CAAC;AAAA,EACtD;AAEA,WAAS,gBAAgB,EAAC,QAAQ,eAAc,GAAG;AACjD,WAAO,SAAS,OAAO,CAAC,KAAK,EAAC,MAAM,QAAO,MAAM,IAAI,OAAO,EAAC,MAAM,OAAO,QAAQ,EAAC,QAAQ,eAAc,CAAC,EAAC,CAAC,GAAG,CAAC,CAAC;AAAA,EACnH;AAEA,WAAS,aAAa,QAAQ;AAC5B,QAAI,gBAAgB;AAClB,YAAM,8CAA8C;AACpD,YAAM,qBAAqB,EAAC,OAAO,OAAO,OAAO,aAAa,OAAO,aAAa,UAAU,eAAe,QAAQ,GAAG,SAAQ;AAC9H,gBAAU,GAAG,KAAK,UAAU,kBAAkB,CAAC,EAAE;AACjD,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AAEA,WAAS,eAAeE,WAAU;AAChC,UAAM,gBAAgBA,UAAS,IAAI,aAAW,QAAQ,IAAI;AAC1D,WAAO,iBAAiB,CAAC;AAAA,EAC3B;AAEA,WAAS,qBAAqB,kBAAkB;AAC9C,UAAM,cAAc,iBAAiB,OAAO,CAAC,KAAK,MAAM,MAAM,GAAG,CAAG;AACpE,WAAO,cAAc,IAAM,IAAM;AAAA,EACnC;AAEA,WAAS,yBAAyB,cAAc;AAC9C,UAAM,WAAW,aAAa,IAAI,CAAC,EAAC,MAAM,GAAG,EAAC,MAAM;AAClD,YAAM,EAAC,QAAO,IAAI,SAAS,KAAK,CAAC,EAAC,MAAM,YAAW,MAAM,SAAS,WAAW;AAC7E,YAAM,SAAS,QAAQ,GAAG,CAAC;AAC3B,aAAO,EAAC,MAAM,OAAM;AAAA,IACtB,CAAC;AAED,cAAU,WAAW,KAAK,UAAU,QAAQ,CAAC,EAAE;AAC/C,WAAO,SAAS,IAAI,CAAC,EAAC,OAAM,MAAM,MAAM;AAAA,EAC1C;AAEA,WAAS,qBAAqBF,YAAW,WAAW;AAClD,UAAM,QAAQ,cAAc;AAC5B,UAAM,kBAAkB,YAAY;AAEpC,UAAM,2FAA2F,KAAK,UAAU,eAAe,CAAC,EAAE;AAClI,WAAO;AAEP,aAAS,gBAAgB;AACvB,aAAOA,WACJ,OAAO,CAAC,KAAK,EAAC,MAAM,MAAK,GAAG,UAAU,IAAI,OAAO;AAAA,QAChD;AAAA,QACA,GAAG;AAAA,QACH,GAAG,UAAU,KAAK,EAAE;AAAA,MACtB,CAAC,GAAG,CAAC,CAAC,EACL,OAAO,CAAC,EAAC,GAAG,EAAC,MAAM;AAClB,YAAI,EAAE,WAAW,KAAK,EAAE,WAAW,GAAG;AACpC,iBAAO;AAAA,QACT;AAEA,eAAO;AAAA,MACT,CAAC;AAAA,IACL;AAEA,aAAS,cAAc;AACrB,aAAOA,WACJ,IAAI,CAAC,EAAC,KAAI,MAAM,IAAI,EACpB,OAAO,OAAK,MAAM,KAAK,CAAC,EAAC,KAAI,MAAM,SAAS,CAAC,MAAM,KAAK;AAAA,IAC7D;AAAA,EACF;AAEF;",
+  "names": ["featuresA", "localSettings", "strategy"]
+}

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

@@ -0,0 +1,44 @@
+import assert from "node:assert";
+import { describe } from "node:test";
+import createDebugLogger from "debug";
+import { inspect } from "util";
+import generateTests from "@natlibfi/fixugen";
+import { READERS } from "@natlibfi/fixura";
+import { MarcRecord } from "@natlibfi/marc-record";
+import createDetectionInterface from "./index.js";
+import * as features from "./features/index.js";
+const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:test");
+const debugData = debug.extend("data");
+describe("match-detection", () => {
+  generateTests({
+    path: [import.meta.dirname, "..", "..", "test-fixtures", "match-detection", "index"],
+    useMetadataFile: true,
+    recurse: false,
+    fixura: {
+      reader: READERS.JSON
+    },
+    callback: ({ getFixture, options, expectedResults, array, enabled = true }) => {
+      if (!enabled) {
+        debug(`*** DISABLED TEST! ***`);
+        return;
+      }
+      const detect = createDetectionInterface(formatOptions());
+      const recordA = new MarcRecord(getFixture("recordA.json"), { subfieldValues: false });
+      debugData(inspect(recordA));
+      debug(`Our recordB is an array of records: ${array}`);
+      const recordB = array ? getFixture("recordB.json").map((recordJson) => new MarcRecord(recordJson, { subfieldValues: false })) : new MarcRecord(getFixture("recordB.json"), { subfieldValues: false });
+      debugData(inspect(recordB));
+      const results = detect({ recordA, recordB });
+      debugData(`${JSON.stringify(results)}`);
+      assert.deepStrictEqual(results, expectedResults);
+      function formatOptions() {
+        const contextFeatures = features[options.strategy.type];
+        return {
+          ...options,
+          strategy: options.strategy.features.map((v) => contextFeatures[v]())
+        };
+      }
+    }
+  });
+});
+//# sourceMappingURL=index.test.js.map

package/dist/match-detection/index.test.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../src/match-detection/index.test.js"],
+  "sourcesContent": ["import assert from 'node:assert';\nimport {describe} from 'node:test';\nimport createDebugLogger from 'debug';\nimport {inspect} from 'util';\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\nimport createDetectionInterface from './index.js';\nimport * as features from './features/index.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:test');\nconst debugData = debug.extend('data');\n\ndescribe('match-detection', () => {\n  generateTests({\n    path: [import.meta.dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],\n    useMetadataFile: true,\n    recurse: false,\n    fixura: {\n      reader: READERS.JSON\n    },\n    callback: ({getFixture, options, expectedResults, array, enabled = true}) => {\n\n      if (!enabled) {\n        debug(`*** DISABLED TEST! ***`);\n        return;\n      }\n\n      const detect = createDetectionInterface(formatOptions());\n      const recordA = new MarcRecord(getFixture('recordA.json'), {subfieldValues: false});\n      debugData(inspect(recordA));\n\n      debug(`Our recordB is an array of records: ${array}`);\n      const recordB = array\n        ? getFixture('recordB.json').map(recordJson => new MarcRecord(recordJson, {subfieldValues: false}))\n        : new MarcRecord(getFixture('recordB.json'), {subfieldValues: false});\n      debugData(inspect(recordB));\n\n      const results = detect({recordA, recordB});\n      debugData(`${JSON.stringify(results)}`);\n\n      assert.deepStrictEqual(results, expectedResults);\n\n      function formatOptions() {\n        const contextFeatures = features[options.strategy.type];\n\n        return {\n          ...options,\n          strategy: options.strategy.features.map(v => contextFeatures[v]())\n        };\n      }\n    }\n  });\n});\n"],
+  "mappings": "AAAA,OAAO,YAAY;AACnB,SAAQ,gBAAe;AACvB,OAAO,uBAAuB;AAC9B,SAAQ,eAAc;AACtB,OAAO,mBAAmB;AAC1B,SAAQ,eAAc;AACtB,SAAQ,kBAAiB;AAEzB,OAAO,8BAA8B;AACrC,YAAY,cAAc;AAE1B,MAAM,QAAQ,kBAAkB,wDAAwD;AACxF,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,SAAS,mBAAmB,MAAM;AAChC,gBAAc;AAAA,IACZ,MAAM,CAAC,YAAY,SAAS,MAAM,MAAM,iBAAiB,mBAAmB,OAAO;AAAA,IACnF,iBAAiB;AAAA,IACjB,SAAS;AAAA,IACT,QAAQ;AAAA,MACN,QAAQ,QAAQ;AAAA,IAClB;AAAA,IACA,UAAU,CAAC,EAAC,YAAY,SAAS,iBAAiB,OAAO,UAAU,KAAI,MAAM;AAE3E,UAAI,CAAC,SAAS;AACZ,cAAM,wBAAwB;AAC9B;AAAA,MACF;AAEA,YAAM,SAAS,yBAAyB,cAAc,CAAC;AACvD,YAAM,UAAU,IAAI,WAAW,WAAW,cAAc,GAAG,EAAC,gBAAgB,MAAK,CAAC;AAClF,gBAAU,QAAQ,OAAO,CAAC;AAE1B,YAAM,uCAAuC,KAAK,EAAE;AACpD,YAAM,UAAU,QACZ,WAAW,cAAc,EAAE,IAAI,gBAAc,IAAI,WAAW,YAAY,EAAC,gBAAgB,MAAK,CAAC,CAAC,IAChG,IAAI,WAAW,WAAW,cAAc,GAAG,EAAC,gBAAgB,MAAK,CAAC;AACtE,gBAAU,QAAQ,OAAO,CAAC;AAE1B,YAAM,UAAU,OAAO,EAAC,SAAS,QAAO,CAAC;AACzC,gBAAU,GAAG,KAAK,UAAU,OAAO,CAAC,EAAE;AAEtC,aAAO,gBAAgB,SAAS,eAAe;AAE/C,eAAS,gBAAgB;AACvB,cAAM,kBAAkB,SAAS,QAAQ,SAAS,IAAI;AAEtD,eAAO;AAAA,UACL,GAAG;AAAA,UACH,UAAU,QAAQ,SAAS,SAAS,IAAI,OAAK,gBAAgB,CAAC,EAAE,CAAC;AAAA,QACnE;AAAA,MACF;AAAA,IACF;AAAA,EACF,CAAC;AACH,CAAC;",
+  "names": []
+}

package/dist/matching-utils.js

@@ -1,123 +1,62 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.extractSubfieldsFromField = extractSubfieldsFromField;
-exports.getMatchCounts = getMatchCounts;
-exports.getMelindaIdsF035 = getMelindaIdsF035;
-exports.getSubfieldValues = getSubfieldValues;
-exports.testStringOrNumber = testStringOrNumber;
-exports.uniqueSubfields = uniqueSubfields;
-exports.validateSidFieldSubfieldCounts = validateSidFieldSubfieldCounts;
-var _debug = _interopRequireDefault(require("debug"));
-function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
-/**
-*
-* @licstart  The following is the entire license notice for the JavaScript code in this file.
-*
-* Melinda record matching modules for Javascript
-*
-* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
-*
-* This file is part of melinda-record-matching-js
-*
-* melinda-record-matching-js program is free software: you can redistribute it and/or modify
-* it under the terms of the GNU Lesser General Public License as
-* published by the Free Software Foundation, either version 3 of the
-* License, or (at your option) any later version.
-*
-* melinda-record-matching-js is distributed in the hope that it will be useful,
-* but WITHOUT ANY WARRANTY; without even the implied warranty of
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-* GNU Lesser General Public License for more details.
-*
-* You should have received a copy of the GNU Affero General Public License
-* along with this program.  If not, see <http://www.gnu.org/licenses/>.
-*
-* @licend  The above is the entire license notice
-* for the JavaScript code in this file.
-*
-*/
-
-const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:utils');
-const debugData = debug.extend('data');
-function getMelindaIdsF035(record) {
-  const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:melinda-id');
-  const debugData = debug.extend('data');
-  const f035s = record.getFields('035');
+import createDebugLogger from "debug";
+const debug = createDebugLogger("@natlibfi/melinda-record-matching:utils");
+const debugData = debug.extend("data");
+export function getMelindaIdsF035(record) {
+  const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:melinda-id");
+  const debugData2 = debug2.extend("data");
+  const f035s = record.getFields("035");
   if (f035s.length < 1) {
-    debug(`No f035s found.`);
+    debug2(`No f035s found.`);
     return [];
   }
-  const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));
+  const allF035MelindaIds = [].concat(...f035s.map((field) => toMelindaIds(field)));
   const f035MelindaIds = [...new Set(allF035MelindaIds)];
-  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);
-  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);
-  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);
+  debugData2(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);
+  debugData2(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);
+  debugData2(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);
   return f035MelindaIds;
-  function toMelindaIds({
-    subfields
-  }) {
-    const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
-    return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({
-      value
-    }) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');
-  }
 }
-function validateSidFieldSubfieldCounts(field) {
-  // Valid SID-fields have just one $c and one $b
+export function toMelindaIds({ subfields }, subfieldsToParse = ["a", "z"]) {
+  const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
+  return subfields.filter((sub) => subfieldsToParse.includes(sub.code)).filter((sub) => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({ value }) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, "$<id>") : "");
+}
+export function validateSidFieldSubfieldCounts(field) {
   debugData(`Validating SID field ${JSON.stringify(field)}`);
-  const countC = countSubfields(field, 'c');
-  const countB = countSubfields(field, 'b');
+  const countC = countSubfields(field, "c");
+  const countB = countSubfields(field, "b");
   debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);
   return countC === 1 && countB === 1;
 }
 function countSubfields(field, subfieldCode) {
-  // debug(`Counting subfields ${subfieldCode}`);
-  return field.subfields.filter(({
-    code
-  }) => code === subfieldCode).length;
+  return field.subfields.filter(({ code }) => code === subfieldCode).length;
 }
-function getSubfieldValues(field, subfieldCode) {
+export function getSubfieldValues(field, subfieldCode) {
   debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);
-  return field.subfields.filter(({
-    code
-  }) => code === subfieldCode).map(({
-    value
-  }) => testStringOrNumber(value) ? String(value) : '').filter(value => value);
+  return field.subfields.filter(({ code }) => code === subfieldCode).map(({ value }) => testStringOrNumber(value) ? String(value) : "").filter((value) => value);
 }
-function testStringOrNumber(value) {
-  if (typeof value === 'string' || typeof value === 'number') {
+export function testStringOrNumber(value) {
+  if (typeof value === "string" || typeof value === "number") {
     return true;
   }
   return false;
 }
-function extractSubfieldsFromField(field, subfieldCodes) {
-  if (field === undefined || field.subfields === undefined) {
+export function extractSubfieldsFromField(field, subfieldCodes) {
+  if (field === void 0 || field.subfields === void 0) {
     return [];
   }
-  const resultSubfields = field.subfields.filter(({
-    code
-  }) => subfieldCodes.includes(code)).map(({
-    code,
-    value
-  }) => ({
-    code,
-    value: testStringOrNumber(value) ? String(value) : ''
-  })).filter(value => value);
+  const resultSubfields = field.subfields.filter(({ code }) => subfieldCodes.includes(code)).map(({ code, value }) => ({ code, value: testStringOrNumber(value) ? String(value) : "" })).filter((value) => value);
   return resultSubfields;
 }
-function uniqueSubfields(subfields) {
+export function uniqueSubfields(subfields) {
   return subfields.reduce((arr, e) => {
-    if (!arr.find(item => item.code === e.code && item.value === e.value)) {
+    if (!arr.find((item) => item.code === e.code && item.value === e.value)) {
       const newArr = arr.concat(e);
       return newArr;
     }
     return arr;
   }, []);
 }
-function getMatchCounts(aValues, bValues) {
+export function getMatchCounts(aValues, bValues) {
   const matchingValues = getMatchingValuesAmount(aValues, bValues);
   return {
     maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
@@ -125,19 +64,16 @@ function getMatchCounts(aValues, bValues) {
     possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
     matchingValues
   };
-  function getMatchingValuesAmount(aValues, bValues) {
-    if (bValues.length > aValues.length) {
-      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
+  function getMatchingValuesAmount(aValues2, bValues2) {
+    if (bValues2.length > aValues2.length) {
+      return aValues2.filter((aValue) => bValues2.some((bValue) => aValue === bValue)).length;
     }
-    if (aValues.length > bValues.length) {
-      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
+    if (aValues2.length > bValues2.length) {
+      return bValues2.filter((bValue) => aValues2.some((aValue) => bValue === aValue)).length;
     }
-
-    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases
-    // there would be duplicate values
-    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
-    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
+    const aToB = aValues2.filter((aValue) => bValues2.some((bValue) => aValue === bValue)).length;
+    const bToA = bValues2.filter((bValue) => aValues2.some((aValue) => bValue === aValue)).length;
     return aToB < bToA ? aToB : bToA;
   }
 }
-//# sourceMappingURL=matching-utils.js.map
+//# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -1 +1,7 @@
-{"version":3,"file":"matching-utils.js","names":["_debug","_interopRequireDefault","require","obj","__esModule","default","debug","createDebugLogger","debugData","extend","getMelindaIdsF035","record","f035s","getFields","length","allF035MelindaIds","concat","map","toMelindaIds","f035MelindaIds","Set","JSON","stringify","subfields","melindaIdRegExp","filter","sub","includes","code","testStringOrNumber","value","test","String","replace","validateSidFieldSubfieldCounts","field","countC","countSubfields","countB","subfieldCode","getSubfieldValues","extractSubfieldsFromField","subfieldCodes","undefined","resultSubfields","uniqueSubfields","reduce","arr","e","find","item","newArr","getMatchCounts","aValues","bValues","matchingValues","getMatchingValuesAmount","maxValues","possibleMatchValues","aValue","some","bValue","aToB","bToA"],"sources":["../src/matching-utils.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n\n  function toMelindaIds({subfields}) {\n    const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code))\n      .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n      .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n\n  }\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n  const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n  return {\n    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n    matchingValues\n  };\n\n  function getMatchingValuesAmount(aValues, bValues) {\n    if (bValues.length > aValues.length) {\n      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    }\n    if (aValues.length > bValues.length) {\n      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n    }\n\n    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n    // there would be duplicate values\n    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n    return aToB < bToA ? aToB : bToA;\n  }\n}\n"],"mappings":";;;;;;;;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AAAsC,SAAAD,uBAAAE,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA5BtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAIA,MAAMG,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiBA,CAACC,MAAM,EAAE;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAiB,EAAC,8CAA8C,CAAC;EAC/E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAS,CAAC,KAAK,CAAC;EAErC,IAAID,KAAK,CAACE,MAAM,GAAG,CAAC,EAAE;IACpBR,KAAK,CAAE,iBAAgB,CAAC;IACxB,OAAO,EAAE;EACX;EAEA,MAAMS,iBAAiB,GAAG,EAAE,CAACC,MAAM,CAAC,GAAGJ,KAAK,CAACK,GAAG,CAACC,YAAY,CAAC,CAAC;EAC/D,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACL,iBAAiB,CAAC,CAAC;EAEtDP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACV,KAAK,CAAE,EAAC,CAAC;EAC/DJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACP,iBAAiB,CAAE,EAAC,CAAC;EACpFP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACH,cAAc,CAAE,EAAC,CAAC;EAErF,OAAOA,cAAc;EAErB,SAASD,YAAYA,CAAC;IAACK;EAAS,CAAC,EAAE;IACjC,MAAMC,eAAe,GAAG,8CAA8C;IAEtE,OAAOD,SAAS,CACbE,MAAM,CAACC,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,GAAG,CAACE,IAAI,CAAC,CAAC,CAC5CH,MAAM,CAACC,GAAG,IAAIG,kBAAkB,CAACH,GAAG,CAACI,KAAK,CAAC,IAAIN,eAAe,CAACO,IAAI,CAACC,MAAM,CAACN,GAAG,CAACI,KAAK,CAAC,CAAC,CAAC,CACvFb,GAAG,CAAC,CAAC;MAACa;IAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAACT,eAAe,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC;EAEvG;AACF;AAEO,SAASU,8BAA8BA,CAACC,KAAK,EAAE;EACpD;EACA3B,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC1D,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC7B,KAAK,CAAE,SAAQ8B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAE,EAAC,CAAC;EAE9F,OAAOF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAC;AACrC;AAEA,SAASD,cAAcA,CAACF,KAAK,EAAEI,YAAY,EAAE;EAC3C;EACA,OAAOJ,KAAK,CAACZ,SAAS,CAACE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CAACzB,MAAM;AACzE;AAEO,SAAS0B,iBAAiBA,CAACL,KAAK,EAAEI,YAAY,EAAE;EACrD/B,SAAS,CAAE,oBAAmB+B,YAAa,SAAQlB,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC3E,OAAOA,KAAK,CAACZ,SAAS,CACnBE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CACzCtB,GAAG,CAAC,CAAC;IAACa;EAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;AAC3B;AAEO,SAASD,kBAAkBA,CAACC,KAAK,EAAE;EACxC,IAAI,OAAOA,KAAK,KAAK,QAAQ,IAAI,OAAOA,KAAK,KAAK,QAAQ,EAAE;IAC1D,OAAO,IAAI;EACb;EACA,OAAO,KAAK;AACd;AAEO,SAASW,yBAAyBA,CAACN,KAAK,EAAEO,aAAa,EAAE;EAC9D,IAAIP,KAAK,KAAKQ,SAAS,IAAIR,KAAK,CAACZ,SAAS,KAAKoB,SAAS,EAAE;IACxD,OAAO,EAAE;EACX;EACA,MAAMC,eAAe,GAAGT,KAAK,CAACZ,SAAS,CACpCE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKc,aAAa,CAACf,QAAQ,CAACC,IAAI,CAAC,CAAC,CAChDX,GAAG,CAAC,CAAC;IAACW,IAAI;IAAEE;EAAK,CAAC,MAAM;IAACF,IAAI;IAAEE,KAAK,EAAED,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG;EAAE,CAAC,CAAC,CAAC,CACvFL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;EACzB,OAAOc,eAAe;AACxB;AAEO,SAASC,eAAeA,CAACtB,SAAS,EAAE;EACzC,OAAOA,SAAS,CAACuB,MAAM,CAAC,CAACC,GAAG,EAAEC,CAAC,KAAK;IAClC,IAAI,CAACD,GAAG,CAACE,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACtB,IAAI,KAAKoB,CAAC,CAACpB,IAAI,IAAIsB,IAAI,CAACpB,KAAK,KAAKkB,CAAC,CAAClB,KAAK,CAAC,EAAE;MACrE,MAAMqB,MAAM,GAAGJ,GAAG,CAAC/B,MAAM,CAACgC,CAAC,CAAC;MAC5B,OAAOG,MAAM;IACf;IACA,OAAOJ,GAAG;EACZ,CAAC,EAAE,EAAE,CAAC;AACR;AAEO,SAASK,cAAcA,CAACC,OAAO,EAAEC,OAAO,EAAE;EAE/C,MAAMC,cAAc,GAAGC,uBAAuB,CAACH,OAAO,EAAEC,OAAO,CAAC;EAEhE,OAAO;IACLG,SAAS,EAAEJ,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM;IAC5E;IACA4C,mBAAmB,EAAEL,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM;IACtFyC;EACF,CAAC;EAED,SAASC,uBAAuBA,CAACH,OAAO,EAAEC,OAAO,EAAE;IACjD,IAAIA,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,EAAE;MACnC,OAAOuC,OAAO,CAAC5B,MAAM,CAACkC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC/C,MAAM;IACnF;IACA,IAAIuC,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,EAAE;MACnC,OAAOwC,OAAO,CAAC7B,MAAM,CAACoC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC7C,MAAM;IACnF;;IAEA;IACA;IACA,MAAMgD,IAAI,GAAGT,OAAO,CAAC5B,MAAM,CAACkC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC/C,MAAM;IACvF,MAAMiD,IAAI,GAAGT,OAAO,CAAC7B,MAAM,CAACoC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC7C,MAAM;IAEvF,OAAOgD,IAAI,GAAGC,IAAI,GAAGD,IAAI,GAAGC,IAAI;EAClC;AACF"}
+{
+  "version": 3,
+  "sources": ["../src/matching-utils.js"],
+  "sourcesContent": ["import createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(field => toMelindaIds(field)));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n}\n\nexport function toMelindaIds({subfields}, subfieldsToParse = ['a', 'z']) {\n  const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n  return subfields\n    .filter(sub => subfieldsToParse.includes(sub.code))\n    .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n    .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n  const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n  return {\n    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n    matchingValues\n  };\n\n  function getMatchingValuesAmount(aValues, bValues) {\n    if (bValues.length > aValues.length) {\n      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    }\n    if (aValues.length > bValues.length) {\n      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n    }\n\n    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n    // there would be duplicate values\n    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n    return aToB < bToA ? aToB : bToA;\n  }\n}\n"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAE9B,MAAM,QAAQ,kBAAkB,yCAAyC;AACzE,MAAM,YAAY,MAAM,OAAO,MAAM;AAE9B,gBAAS,kBAAkB,QAAQ;AAExC,QAAMA,SAAQ,kBAAkB,8CAA8C;AAC9E,QAAMC,aAAYD,OAAM,OAAO,MAAM;AAErC,QAAM,QAAQ,OAAO,UAAU,KAAK;AAEpC,MAAI,MAAM,SAAS,GAAG;AACpB,IAAAA,OAAM,iBAAiB;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,oBAAoB,CAAC,EAAE,OAAO,GAAG,MAAM,IAAI,WAAS,aAAa,KAAK,CAAC,CAAC;AAC9E,QAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,iBAAiB,CAAC;AAErD,EAAAC,WAAU,WAAW,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAC9D,EAAAA,WAAU,QAAQ,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AACnF,EAAAA,WAAU,eAAe,eAAe,MAAM,MAAM,KAAK,UAAU,cAAc,CAAC,EAAE;AAEpF,SAAO;AACT;AAEO,gBAAS,aAAa,EAAC,UAAS,GAAG,mBAAmB,CAAC,KAAK,GAAG,GAAG;AACvE,QAAM,kBAAkB;AAExB,SAAO,UACJ,OAAO,SAAO,iBAAiB,SAAS,IAAI,IAAI,CAAC,EACjD,OAAO,SAAO,mBAAmB,IAAI,KAAK,KAAK,gBAAgB,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtF,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,EAAE,QAAQ,iBAAiB,OAAO,IAAI,EAAE;AACtG;AAEO,gBAAS,+BAA+B,OAAO;AAEpD,YAAU,wBAAwB,KAAK,UAAU,KAAK,CAAC,EAAE;AACzD,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,MAAM,eAAe,MAAM,qBAAqB,WAAW,KAAK,WAAW,CAAC,EAAE;AAE7F,SAAO,WAAW,KAAK,WAAW;AACpC;AAEA,SAAS,eAAe,OAAO,cAAc;AAE3C,SAAO,MAAM,UAAU,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EAAE;AACnE;AAEO,gBAAS,kBAAkB,OAAO,cAAc;AACrD,YAAU,oBAAoB,YAAY,SAAS,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1E,SAAO,MAAM,UACV,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EACxC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK;AAC1B;AAEO,gBAAS,mBAAmB,OAAO;AACxC,MAAI,OAAO,UAAU,YAAY,OAAO,UAAU,UAAU;AAC1D,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAEO,gBAAS,0BAA0B,OAAO,eAAe;AAC9D,MAAI,UAAU,UAAa,MAAM,cAAc,QAAW;AACxD,WAAO,CAAC;AAAA,EACV;AACA,QAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,cAAc,SAAS,IAAI,CAAC,EAC/C,IAAI,CAAC,EAAC,MAAM,MAAK,OAAO,EAAC,MAAM,OAAO,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,GAAE,EAAE,EACtF,OAAO,WAAS,KAAK;AACxB,SAAO;AACT;AAEO,gBAAS,gBAAgB,WAAW;AACzC,SAAO,UAAU,OAAO,CAAC,KAAK,MAAM;AAClC,QAAI,CAAC,IAAI,KAAK,UAAQ,KAAK,SAAS,EAAE,QAAQ,KAAK,UAAU,EAAE,KAAK,GAAG;AACrE,YAAM,SAAS,IAAI,OAAO,CAAC;AAC3B,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT,GAAG,CAAC,CAAC;AACP;AAEO,gBAAS,eAAe,SAAS,SAAS;AAE/C,QAAM,iBAAiB,wBAAwB,SAAS,OAAO;AAE/D,SAAO;AAAA,IACL,WAAW,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA;AAAA,IAEtE,qBAAqB,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA,IAChF;AAAA,EACF;AAEA,WAAS,wBAAwBC,UAASC,UAAS;AACjD,QAAIA,SAAQ,SAASD,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AACA,QAAID,SAAQ,SAASC,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AAIA,UAAM,OAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AACjF,UAAM,OAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAEjF,WAAO,OAAO,OAAO,OAAO;AAAA,EAC9B;AACF;",
+  "names": ["debug", "debugData", "aValues", "bValues"]
+}