@natlibfi/melinda-record-matching 4.0.1-alpha.1 → 4.1.0-alpha.1

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

@@ -63,6 +63,12 @@ Object.defineProperty(exports, "publicationTime", {
     return _publicationTime.default;
   }
 });
+Object.defineProperty(exports, "publicationTimeAllowConsYears", {
+  enumerable: true,
+  get: function () {
+    return _publicationTimeAllowConsYears.default;
+  }
+});
 Object.defineProperty(exports, "recordType", {
   enumerable: true,
   get: function () {
@@ -90,6 +96,7 @@ var _titleVersionOriginal = _interopRequireDefault(require("./title-version-orig
 var _authors = _interopRequireDefault(require("./authors"));
 var _recordType = _interopRequireDefault(require("./record-type"));
 var _publicationTime = _interopRequireDefault(require("./publication-time"));
+var _publicationTimeAllowConsYears = _interopRequireDefault(require("./publication-time-allow-cons-years"));
 var _language = _interopRequireDefault(require("./language"));
 var _bibliographicLevel = _interopRequireDefault(require("./bibliographic-level"));
 var _melindaId = _interopRequireDefault(require("./melinda-id"));

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

@@ -1 +1 @@
-{"version":3,"file":"index.js","names":[],"sources":["../../../../src/match-detection/features/bib/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\nexport {default as hostComponent} from './host-component';\nexport {default as isbn} from './isbn';\nexport {default as issn} from './issn';\nexport {default as otherStandardIdentifier} from './other-standard-identifier';\nexport {default as title} from './title';\nexport {default as titleVersionOriginal} from './title-version-original';\nexport {default as authors} from './authors';\nexport {default as recordType} from './record-type';\nexport {default as publicationTime} from './publication-time';\nexport {default as language} from './language';\nexport {default as bibliographicLevel} from './bibliographic-level';\nexport {default as melindaId} from './melinda-id';\nexport {default as allSourceIds} from './all-source-ids';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA4BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAyD"}
+{"version":3,"file":"index.js","names":[],"sources":["../../../../src/match-detection/features/bib/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\nexport {default as hostComponent} from './host-component';\nexport {default as isbn} from './isbn';\nexport {default as issn} from './issn';\nexport {default as otherStandardIdentifier} from './other-standard-identifier';\nexport {default as title} from './title';\nexport {default as titleVersionOriginal} from './title-version-original';\nexport {default as authors} from './authors';\nexport {default as recordType} from './record-type';\nexport {default as publicationTime} from './publication-time';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years';\nexport {default as language} from './language';\nexport {default as bibliographicLevel} from './bibliographic-level';\nexport {default as melindaId} from './melinda-id';\nexport {default as allSourceIds} from './all-source-ids';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA4BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAyD"}

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

@@ -5,6 +5,7 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.default = void 0;
 var _debug = _interopRequireDefault(require("debug"));
+var _matchingUtils = require("../../../matching-utils");
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
 /**
 *
@@ -12,7 +13,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -46,17 +47,23 @@ var _default = () => ({
     const value008 = get008Value();
     const values041 = get041Values();
     debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
+
+    /*
     if (value008 && values041.length > 0) {
       debugData(`${label} There's both 008 and 041, searching for value in both`);
       const correspondingValue = values041.find(v => v === value008);
       debugData(`${label} Corresponding value: ${correspondingValue}`);
       return correspondingValue ? [correspondingValue] : [];
     }
+    */
+
     if (!value008 && values041.length < 1) {
       debugData(`{$label} No actual values found`);
       return [];
     }
-    return value008 ? [value008] : [values041[0]];
+    const allValues = value008 === undefined ? values041 : values041.concat(value008);
+    const uniqueSortedValues = [...new Set(allValues)].sort();
+    return uniqueSortedValues;
     function get008Value() {
       var _record$get, _record$get$;
       const value = ((_record$get = record.get(/^008$/u)) === null || _record$get === void 0 ? void 0 : (_record$get$ = _record$get[0]) === null || _record$get$ === void 0 ? void 0 : _record$get$.value) || undefined;
@@ -66,11 +73,9 @@ var _default = () => ({
       }
       const code = value.slice(35, 38);
       debugData(`${label} 008 code: ${code}`);
-      return code === '|||' || code === '   ' || code === '^^^' ? undefined : code;
+      return isLangCodeForALanguage(code) ? code : undefined;
     }
 
-    // Main language for the resource: in the first f041 $a or f041 $d
-    // Should we get all $a or $d languages instead of just the first?
     // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes
 
     function get041Values() {
@@ -80,22 +85,67 @@ var _default = () => ({
         subfields
       }) => subfields).flat().filter(({
         code
-      }) => code === 'a' || code === 'd').map(({
+      }) => code === 'a' || code === 'd').filter(({
+        value
+      }) => value && isLangCodeForALanguage(value)).map(({
         value
       }) => value);
     }
+
+    // Check if a string is a possible, validly formed language code for a single language
+    // Currently accept also codes in capitals
+    function isLangCodeForALanguage(code) {
+      if (code.length !== 3) {
+        debugData(`Code ${code} is not correct length (3) for a language code.`);
+        return false;
+      }
+      if (code === '|||' || code === '   ' || code === '^^^' || code === 'mul' || code === 'zxx') {
+        debugData(`Code ${code} is not code for a spesific language.`);
+        return false;
+      }
+      const langCodePattern = /^[a-z][a-z][a-z]$/ui;
+      if (!langCodePattern.test(code)) {
+        debugData(`Code ${code} is not valid as a language code`);
+        return false;
+      }
+      return true;
+    }
   },
+  // eslint-disable-next-line max-statements
   compare: (a, b) => {
-    debugData(`Comparing ${JSON.stringify(a[0])} and ${JSON.stringify(b[0])}`);
+    debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);
     if (a.length === 0 || b.length === 0) {
       debugData(`No language to compare`);
       return 0;
     }
-    debugData(`There are languages to compare`);
-    if (a[0] === b[0]) {
+    if (a.length === b.length && a.every((element, index) => element === b[index])) {
+      debugData(`All languages match`);
       return 0.1;
     }
-    return a[0] === 'und' || b[0] === 'und' ? 0.0 : -1.0;
+    const {
+      matchingValues,
+      possibleMatchValues,
+      maxValues
+    } = (0, _matchingUtils.getMatchCounts)(a, b);
+    if (matchingValues < 1) {
+      debug(`Both have languages, but none of these match.`);
+      return -1.0;
+    }
+    debug(`Both have languages, ${matchingValues}/${possibleMatchValues} valid languages match.`);
+    // ignore non-matches if there is mismatching amount of values
+    debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);
+    //we give some kind of penalty for mismatching amount of values instead of simple divide?
+    const missingCount = maxValues - possibleMatchValues;
+    const misMatchCount = possibleMatchValues - matchingValues;
+    debug(`\t missing: ${missingCount}`);
+    debug(`\t mismatches: ${misMatchCount}`);
+    const penaltyForMissing = 0.02 * (maxValues - possibleMatchValues);
+    const penaltyForMisMatch = 0.05 * (possibleMatchValues - matchingValues);
+    debug(`\t points: penaltyForMissing: ${penaltyForMissing}`);
+    debug(`\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);
+    const points = Number(Number(0.1 - penaltyForMisMatch - penaltyForMissing).toFixed(2));
+    debug(`Total points: ${points}`);
+    return points;
   }
 });
 exports.default = _default;

package/dist/match-detection/features/bib/language.js.map

@@ -1 +1 @@
-{"version":3,"file":"language.js","names":["debug","createDebugLogger","debugData","extend","name","extract","record","recordExternal","label","value008","get008Value","values041","get041Values","JSON","stringify","length","correspondingValue","find","v","value","get","undefined","code","slice","filter","ind2","map","subfields","flat","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/language.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:match-detection:features:language');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'Language',\n  extract: ({record, recordExternal}) => {\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const value008 = get008Value();\n    const values041 = get041Values();\n    debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);\n\n    if (value008 && values041.length > 0) {\n      debugData(`${label} There's both 008 and 041, searching for value in both`);\n      const correspondingValue = values041.find(v => v === value008);\n      debugData(`${label} Corresponding value: ${correspondingValue}`);\n      return correspondingValue ? [correspondingValue] : [];\n    }\n\n    if (!value008 && values041.length < 1) {\n      debugData(`{$label} No actual values found`);\n      return [];\n    }\n\n    return value008 ? [value008] : [values041[0]];\n\n    function get008Value() {\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      debugData(`${label} 008 value: ${value}`);\n\n      if (!value) {\n        return undefined;\n      }\n\n      const code = value.slice(35, 38);\n      debugData(`${label} 008 code: ${code}`);\n      return code === '|||' || code === '   ' || code === '^^^' ? undefined : code;\n    }\n\n    // Main language for the resource: in the first f041 $a or f041 $d\n    // Should we get all $a or $d languages instead of just the first?\n    // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes\n\n    function get041Values() {\n      return record.get(/^041$/u)\n        .filter(({ind2}) => ind2 === ' ')\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a' || code === 'd')\n        .map(({value}) => value);\n    }\n  },\n  compare: (a, b) => {\n    debugData(`Comparing ${JSON.stringify(a[0])} and ${JSON.stringify(b[0])}`);\n\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No language to compare`);\n      return 0;\n    }\n\n    debugData(`There are languages to compare`);\n\n    if (a[0] === b[0]) {\n      return 0.1;\n    }\n\n    return a[0] === 'und' || b[0] === 'und' ? 0.0 : -1.0;\n  }\n});\n"],"mappings":";;;;;;AA4BA;AAAsC;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,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;AACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,OAAO;EACpBC,IAAI,EAAE,UAAU;EAChBC,OAAO,EAAE,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,KAAK;IACrC,MAAMC,KAAK,GAAGD,cAAc,IAAIA,cAAc,CAACC,KAAK,GAAGD,cAAc,CAACC,KAAK,GAAG,QAAQ;IAEtF,MAAMC,QAAQ,GAAGC,WAAW,EAAE;IAC9B,MAAMC,SAAS,GAAGC,YAAY,EAAE;IAChCV,SAAS,CAAE,GAAEM,KAAM,SAAQK,IAAI,CAACC,SAAS,CAACL,QAAQ,CAAE,UAASI,IAAI,CAACC,SAAS,CAACH,SAAS,CAAE,EAAC,CAAC;IAEzF,IAAIF,QAAQ,IAAIE,SAAS,CAACI,MAAM,GAAG,CAAC,EAAE;MACpCb,SAAS,CAAE,GAAEM,KAAM,wDAAuD,CAAC;MAC3E,MAAMQ,kBAAkB,GAAGL,SAAS,CAACM,IAAI,CAACC,CAAC,IAAIA,CAAC,KAAKT,QAAQ,CAAC;MAC9DP,SAAS,CAAE,GAAEM,KAAM,yBAAwBQ,kBAAmB,EAAC,CAAC;MAChE,OAAOA,kBAAkB,GAAG,CAACA,kBAAkB,CAAC,GAAG,EAAE;IACvD;IAEA,IAAI,CAACP,QAAQ,IAAIE,SAAS,CAACI,MAAM,GAAG,CAAC,EAAE;MACrCb,SAAS,CAAE,iCAAgC,CAAC;MAC5C,OAAO,EAAE;IACX;IAEA,OAAOO,QAAQ,GAAG,CAACA,QAAQ,CAAC,GAAG,CAACE,SAAS,CAAC,CAAC,CAAC,CAAC;IAE7C,SAASD,WAAW,GAAG;MAAA;MACrB,MAAMS,KAAK,GAAG,gBAAAb,MAAM,CAACc,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIE,SAAS;MAC3DnB,SAAS,CAAE,GAAEM,KAAM,eAAcW,KAAM,EAAC,CAAC;MAEzC,IAAI,CAACA,KAAK,EAAE;QACV,OAAOE,SAAS;MAClB;MAEA,MAAMC,IAAI,GAAGH,KAAK,CAACI,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;MAChCrB,SAAS,CAAE,GAAEM,KAAM,cAAac,IAAK,EAAC,CAAC;MACvC,OAAOA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,GAAGD,SAAS,GAAGC,IAAI;IAC9E;;IAEA;IACA;IACA;;IAEA,SAASV,YAAY,GAAG;MACtB,OAAON,MAAM,CAACc,GAAG,CAAC,QAAQ,CAAC,CACxBI,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCC,GAAG,CAAC,CAAC;QAACC;MAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,EAAE,CACNJ,MAAM,CAAC,CAAC;QAACF;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,IAAIA,IAAI,KAAK,GAAG,CAAC,CAChDI,GAAG,CAAC,CAAC;QAACP;MAAK,CAAC,KAAKA,KAAK,CAAC;IAC5B;EACF,CAAC;EACDU,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB7B,SAAS,CAAE,aAAYW,IAAI,CAACC,SAAS,CAACgB,CAAC,CAAC,CAAC,CAAC,CAAE,QAAOjB,IAAI,CAACC,SAAS,CAACiB,CAAC,CAAC,CAAC,CAAC,CAAE,EAAC,CAAC;IAE1E,IAAID,CAAC,CAACf,MAAM,KAAK,CAAC,IAAIgB,CAAC,CAAChB,MAAM,KAAK,CAAC,EAAE;MACpCb,SAAS,CAAE,wBAAuB,CAAC;MACnC,OAAO,CAAC;IACV;IAEAA,SAAS,CAAE,gCAA+B,CAAC;IAE3C,IAAI4B,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,EAAE;MACjB,OAAO,GAAG;IACZ;IAEA,OAAOD,CAAC,CAAC,CAAC,CAAC,KAAK,KAAK,IAAIC,CAAC,CAAC,CAAC,CAAC,KAAK,KAAK,GAAG,GAAG,GAAG,CAAC,GAAG;EACtD;AACF,CAAC,CAAC;AAAA"}
+{"version":3,"file":"language.js","names":["debug","createDebugLogger","debugData","extend","name","extract","record","recordExternal","label","value008","get008Value","values041","get041Values","JSON","stringify","length","allValues","undefined","concat","uniqueSortedValues","Set","sort","value","get","code","slice","isLangCodeForALanguage","filter","ind2","map","subfields","flat","langCodePattern","test","compare","a","b","every","element","index","matchingValues","possibleMatchValues","maxValues","getMatchCounts","missingCount","misMatchCount","penaltyForMissing","penaltyForMisMatch","points","Number","toFixed"],"sources":["../../../../src/match-detection/features/bib/language.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) 2020-2023 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 {getMatchCounts} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:language');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'Language',\n  extract: ({record, recordExternal}) => {\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const value008 = get008Value();\n    const values041 = get041Values();\n    debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);\n\n    /*\n    if (value008 && values041.length > 0) {\n      debugData(`${label} There's both 008 and 041, searching for value in both`);\n      const correspondingValue = values041.find(v => v === value008);\n      debugData(`${label} Corresponding value: ${correspondingValue}`);\n      return correspondingValue ? [correspondingValue] : [];\n    }\n    */\n\n    if (!value008 && values041.length < 1) {\n      debugData(`{$label} No actual values found`);\n      return [];\n    }\n\n    const allValues = value008 === undefined ? values041 : values041.concat(value008);\n    const uniqueSortedValues = [...new Set(allValues)].sort();\n\n    return uniqueSortedValues;\n\n    function get008Value() {\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      debugData(`${label} 008 value: ${value}`);\n\n      if (!value) {\n        return undefined;\n      }\n\n      const code = value.slice(35, 38);\n      debugData(`${label} 008 code: ${code}`);\n      return isLangCodeForALanguage(code) ? code : undefined;\n    }\n\n    // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes\n\n    function get041Values() {\n      return record.get(/^041$/u)\n        .filter(({ind2}) => ind2 === ' ')\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a' || code === 'd')\n        .filter(({value}) => value && isLangCodeForALanguage(value))\n        .map(({value}) => value);\n    }\n\n    // Check if a string is a possible, validly formed language code for a single language\n    // Currently accept also codes in capitals\n    function isLangCodeForALanguage(code) {\n      if (code.length !== 3) {\n        debugData(`Code ${code} is not correct length (3) for a language code.`);\n        return false;\n      }\n      if (code === '|||' || code === '   ' || code === '^^^' || code === 'mul' || code === 'zxx') {\n        debugData(`Code ${code} is not code for a spesific language.`);\n        return false;\n      }\n      const langCodePattern = /^[a-z][a-z][a-z]$/ui;\n      if (!langCodePattern.test(code)) {\n        debugData(`Code ${code} is not valid as a language code`);\n        return false;\n      }\n      return true;\n    }\n\n  },\n  // eslint-disable-next-line max-statements\n  compare: (a, b) => {\n    debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);\n\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No language to compare`);\n      return 0;\n    }\n\n    if (a.length === b.length && a.every((element, index) => element === b[index])) {\n      debugData(`All languages match`);\n      return 0.1;\n    }\n\n    const {matchingValues, possibleMatchValues, maxValues} = getMatchCounts(a, b);\n\n    if (matchingValues < 1) {\n      debug(`Both have languages, but none of these match.`);\n      return -1.0;\n    }\n    debug(`Both have languages, ${matchingValues}/${possibleMatchValues} valid languages match.`);\n    // ignore non-matches if there is mismatching amount of values\n    debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);\n    //we give some kind of penalty for mismatching amount of values instead of simple divide?\n    const missingCount = maxValues - possibleMatchValues;\n    const misMatchCount = possibleMatchValues - matchingValues;\n    debug(`\\t missing: ${missingCount}`);\n    debug(`\\t mismatches: ${misMatchCount}`);\n\n    const penaltyForMissing = 0.02 * (maxValues - possibleMatchValues);\n    const penaltyForMisMatch = 0.05 * (possibleMatchValues - matchingValues);\n    debug(`\\t points: penaltyForMissing: ${penaltyForMissing}`);\n    debug(`\\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);\n\n    const points = Number(Number(0.1 - penaltyForMisMatch - penaltyForMissing).toFixed(2));\n    debug(`Total points: ${points}`);\n\n    return points;\n  }\n});\n"],"mappings":";;;;;;AA4BA;AACA;AAAuD;AA7BvD;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;;AAKA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;AACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,OAAO;EACpBC,IAAI,EAAE,UAAU;EAChBC,OAAO,EAAE,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,KAAK;IACrC,MAAMC,KAAK,GAAGD,cAAc,IAAIA,cAAc,CAACC,KAAK,GAAGD,cAAc,CAACC,KAAK,GAAG,QAAQ;IAEtF,MAAMC,QAAQ,GAAGC,WAAW,EAAE;IAC9B,MAAMC,SAAS,GAAGC,YAAY,EAAE;IAChCV,SAAS,CAAE,GAAEM,KAAM,SAAQK,IAAI,CAACC,SAAS,CAACL,QAAQ,CAAE,UAASI,IAAI,CAACC,SAAS,CAACH,SAAS,CAAE,EAAC,CAAC;;IAEzF;AACJ;AACA;AACA;AACA;AACA;AACA;AACA;;IAEI,IAAI,CAACF,QAAQ,IAAIE,SAAS,CAACI,MAAM,GAAG,CAAC,EAAE;MACrCb,SAAS,CAAE,iCAAgC,CAAC;MAC5C,OAAO,EAAE;IACX;IAEA,MAAMc,SAAS,GAAGP,QAAQ,KAAKQ,SAAS,GAAGN,SAAS,GAAGA,SAAS,CAACO,MAAM,CAACT,QAAQ,CAAC;IACjF,MAAMU,kBAAkB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACJ,SAAS,CAAC,CAAC,CAACK,IAAI,EAAE;IAEzD,OAAOF,kBAAkB;IAEzB,SAAST,WAAW,GAAG;MAAA;MACrB,MAAMY,KAAK,GAAG,gBAAAhB,MAAM,CAACiB,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIL,SAAS;MAC3Df,SAAS,CAAE,GAAEM,KAAM,eAAcc,KAAM,EAAC,CAAC;MAEzC,IAAI,CAACA,KAAK,EAAE;QACV,OAAOL,SAAS;MAClB;MAEA,MAAMO,IAAI,GAAGF,KAAK,CAACG,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;MAChCvB,SAAS,CAAE,GAAEM,KAAM,cAAagB,IAAK,EAAC,CAAC;MACvC,OAAOE,sBAAsB,CAACF,IAAI,CAAC,GAAGA,IAAI,GAAGP,SAAS;IACxD;;IAEA;;IAEA,SAASL,YAAY,GAAG;MACtB,OAAON,MAAM,CAACiB,GAAG,CAAC,QAAQ,CAAC,CACxBI,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCC,GAAG,CAAC,CAAC;QAACC;MAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,EAAE,CACNJ,MAAM,CAAC,CAAC;QAACH;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,IAAIA,IAAI,KAAK,GAAG,CAAC,CAChDG,MAAM,CAAC,CAAC;QAACL;MAAK,CAAC,KAAKA,KAAK,IAAII,sBAAsB,CAACJ,KAAK,CAAC,CAAC,CAC3DO,GAAG,CAAC,CAAC;QAACP;MAAK,CAAC,KAAKA,KAAK,CAAC;IAC5B;;IAEA;IACA;IACA,SAASI,sBAAsB,CAACF,IAAI,EAAE;MACpC,IAAIA,IAAI,CAACT,MAAM,KAAK,CAAC,EAAE;QACrBb,SAAS,CAAE,QAAOsB,IAAK,iDAAgD,CAAC;QACxE,OAAO,KAAK;MACd;MACA,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,EAAE;QAC1FtB,SAAS,CAAE,QAAOsB,IAAK,uCAAsC,CAAC;QAC9D,OAAO,KAAK;MACd;MACA,MAAMQ,eAAe,GAAG,qBAAqB;MAC7C,IAAI,CAACA,eAAe,CAACC,IAAI,CAACT,IAAI,CAAC,EAAE;QAC/BtB,SAAS,CAAE,QAAOsB,IAAK,kCAAiC,CAAC;QACzD,OAAO,KAAK;MACd;MACA,OAAO,IAAI;IACb;EAEF,CAAC;EACD;EACAU,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjBlC,SAAS,CAAE,aAAYW,IAAI,CAACC,SAAS,CAACqB,CAAC,CAAE,QAAOtB,IAAI,CAACC,SAAS,CAACsB,CAAC,CAAE,EAAC,CAAC;IAEpE,IAAID,CAAC,CAACpB,MAAM,KAAK,CAAC,IAAIqB,CAAC,CAACrB,MAAM,KAAK,CAAC,EAAE;MACpCb,SAAS,CAAE,wBAAuB,CAAC;MACnC,OAAO,CAAC;IACV;IAEA,IAAIiC,CAAC,CAACpB,MAAM,KAAKqB,CAAC,CAACrB,MAAM,IAAIoB,CAAC,CAACE,KAAK,CAAC,CAACC,OAAO,EAAEC,KAAK,KAAKD,OAAO,KAAKF,CAAC,CAACG,KAAK,CAAC,CAAC,EAAE;MAC9ErC,SAAS,CAAE,qBAAoB,CAAC;MAChC,OAAO,GAAG;IACZ;IAEA,MAAM;MAACsC,cAAc;MAAEC,mBAAmB;MAAEC;IAAS,CAAC,GAAG,IAAAC,6BAAc,EAACR,CAAC,EAAEC,CAAC,CAAC;IAE7E,IAAII,cAAc,GAAG,CAAC,EAAE;MACtBxC,KAAK,CAAE,+CAA8C,CAAC;MACtD,OAAO,CAAC,GAAG;IACb;IACAA,KAAK,CAAE,wBAAuBwC,cAAe,IAAGC,mBAAoB,yBAAwB,CAAC;IAC7F;IACAzC,KAAK,CAAE,qBAAoByC,mBAAoB,IAAGC,SAAU,EAAC,CAAC;IAC9D;IACA,MAAME,YAAY,GAAGF,SAAS,GAAGD,mBAAmB;IACpD,MAAMI,aAAa,GAAGJ,mBAAmB,GAAGD,cAAc;IAC1DxC,KAAK,CAAE,eAAc4C,YAAa,EAAC,CAAC;IACpC5C,KAAK,CAAE,kBAAiB6C,aAAc,EAAC,CAAC;IAExC,MAAMC,iBAAiB,GAAG,IAAI,IAAIJ,SAAS,GAAGD,mBAAmB,CAAC;IAClE,MAAMM,kBAAkB,GAAG,IAAI,IAAIN,mBAAmB,GAAGD,cAAc,CAAC;IACxExC,KAAK,CAAE,iCAAgC8C,iBAAkB,EAAC,CAAC;IAC3D9C,KAAK,CAAE,kCAAiC+C,kBAAmB,EAAC,CAAC;IAE7D,MAAMC,MAAM,GAAGC,MAAM,CAACA,MAAM,CAAC,GAAG,GAAGF,kBAAkB,GAAGD,iBAAiB,CAAC,CAACI,OAAO,CAAC,CAAC,CAAC,CAAC;IACtFlD,KAAK,CAAE,iBAAgBgD,MAAO,EAAC,CAAC;IAEhC,OAAOA,MAAM;EACf;AACF,CAAC,CAAC;AAAA"}

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

@@ -0,0 +1,67 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.default = void 0;
+var _matchingUtils = require("../../../matching-utils");
+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) 2020-2023 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.
+*
+*/
+// We should also get copyright time and copyright/publication times from 26x
+var _default = () => ({
+  name: 'Publication time, allow consequent years',
+  extract: ({
+    record
+  }) => {
+    var _record$get, _record$get$;
+    const value = ((_record$get = record.get(/^008$/u)) === null || _record$get === void 0 ? void 0 : (_record$get$ = _record$get[0]) === null || _record$get$ === void 0 ? void 0 : _record$get$.value) || undefined;
+    return (0, _matchingUtils.testStringOrNumber)(value) ? [String(value).slice(7, 11)] : [];
+  },
+  compare: (a, b) => {
+    const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years');
+    debug(`Comparing ${a[0]} to ${b[0]}`);
+    const [firstA] = a;
+    const [firstB] = b;
+    if (firstA === firstB) {
+      return 0.1;
+    }
+    const firstANumber = parseInt(firstA, 10);
+    const firstBNumber = parseInt(firstB, 10);
+    if (isNaN(firstANumber) || isNaN(firstBNumber)) {
+      return -1;
+    }
+
+    // Handle consequent years as a match
+    // see publication-time for a version that does not handle consequent years as a match
+    return firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber ? 0.1 : -1;
+  }
+});
+exports.default = _default;
+//# sourceMappingURL=publication-time-allow-cons-years.js.map

package/dist/match-detection/features/bib/publication-time-allow-cons-years.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"publication-time-allow-cons-years.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b","debug","createDebugLogger","firstA","firstB","firstANumber","parseInt","firstBNumber","isNaN"],"sources":["../../../../src/match-detection/features/bib/publication-time-allow-cons-years.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) 2020-2023 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 {testStringOrNumber} from '../../../matching-utils';\nimport createDebugLogger from 'debug';\n\n// We should also get copyright time and copyright/publication times from 26x\n\nexport default () => ({\n  name: 'Publication time, allow consequent years',\n  extract: ({record}) => {\n    const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years');\n    debug(`Comparing ${a[0]} to ${b[0]}`);\n\n    const [firstA] = a;\n    const [firstB] = b;\n\n    if (firstA === firstB) {\n      return 0.1;\n    }\n\n    const firstANumber = parseInt(firstA, 10);\n    const firstBNumber = parseInt(firstB, 10);\n\n    if (isNaN(firstANumber) || isNaN(firstBNumber)) {\n      return -1;\n    }\n\n    // Handle consequent years as a match\n    // see publication-time for a version that does not handle consequent years as a match\n    return firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber ? 0.1 : -1;\n  }\n});\n"],"mappings":";;;;;;AA4BA;AACA;AAAsC;AA7BtC;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;AAKA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,0CAA0C;EAChDC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IAAA;IACrB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIE,SAAS;IAC3D,OAAO,IAAAC,iCAAkB,EAACH,KAAK,CAAC,GAAG,CAACI,MAAM,CAACJ,KAAK,CAAC,CAACK,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE;EACtE,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,kGAAkG,CAAC;IACnID,KAAK,CAAE,aAAYF,CAAC,CAAC,CAAC,CAAE,OAAMC,CAAC,CAAC,CAAC,CAAE,EAAC,CAAC;IAErC,MAAM,CAACG,MAAM,CAAC,GAAGJ,CAAC;IAClB,MAAM,CAACK,MAAM,CAAC,GAAGJ,CAAC;IAElB,IAAIG,MAAM,KAAKC,MAAM,EAAE;MACrB,OAAO,GAAG;IACZ;IAEA,MAAMC,YAAY,GAAGC,QAAQ,CAACH,MAAM,EAAE,EAAE,CAAC;IACzC,MAAMI,YAAY,GAAGD,QAAQ,CAACF,MAAM,EAAE,EAAE,CAAC;IAEzC,IAAII,KAAK,CAACH,YAAY,CAAC,IAAIG,KAAK,CAACD,YAAY,CAAC,EAAE;MAC9C,OAAO,CAAC,CAAC;IACX;;IAEA;IACA;IACA,OAAOF,YAAY,GAAG,CAAC,KAAKE,YAAY,IAAIF,YAAY,GAAG,CAAC,KAAKE,YAAY,GAAG,GAAG,GAAG,CAAC,CAAC;EAC1F;AACF,CAAC,CAAC;AAAA"}

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

@@ -33,7 +33,7 @@ var _matchingUtils = require("../../../matching-utils");
 *
 */
 // We should also get copyright time and copyright/publication times from 26x
-// We could also enable matching of records which have consequent publication years
+// see publication-time-allow-cons-years for a version allowing consequent years to match
 var _default = () => ({
   name: 'Publication time',
   extract: ({

package/dist/match-detection/features/bib/publication-time.js.map

@@ -1 +1 @@
-{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.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 {testStringOrNumber} from '../../../matching-utils';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We could also enable matching of records which have consequent publication years\n\nexport default () => ({\n  name: 'Publication time',\n  extract: ({record}) => {\n    const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n  },\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;AA4BA;AA5BA;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;AACA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,kBAAkB;EACxBC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IAAA;IACrB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIE,SAAS;IAC3D,OAAO,IAAAC,iCAAkB,EAACH,KAAK,CAAC,GAAG,CAACI,MAAM,CAACJ,KAAK,CAAC,CAACK,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE;EACtE,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAA"}
+{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.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 {testStringOrNumber} from '../../../matching-utils';\n\n// We should also get copyright time and copyright/publication times from 26x\n// see publication-time-allow-cons-years for a version allowing consequent years to match\n\nexport default () => ({\n  name: 'Publication time',\n  extract: ({record}) => {\n    const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n  },\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;AA4BA;AA5BA;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;AACA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,kBAAkB;EACxBC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IAAA;IACrB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIE,SAAS;IAC3D,OAAO,IAAAC,iCAAkB,EAACH,KAAK,CAAC,GAAG,CAACI,MAAM,CAACJ,KAAK,CAAC,CAACK,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE;EACtE,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAA"}

package/dist/matching-utils.js

@@ -4,6 +4,7 @@ Object.defineProperty(exports, "__esModule", {
   value: true
 });
 exports.extractSubfieldsFromField = extractSubfieldsFromField;
+exports.getMatchCounts = getMatchCounts;
 exports.getMelindaIdsF035 = getMelindaIdsF035;
 exports.getSubfieldValues = getSubfieldValues;
 exports.testStringOrNumber = testStringOrNumber;
@@ -116,4 +117,27 @@ function uniqueSubfields(subfields) {
     return arr;
   }, []);
 }
+function getMatchCounts(aValues, bValues) {
+  const matchingValues = getMatchingValuesAmount(aValues, bValues);
+  return {
+    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
+    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
+    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;
+    }
+    if (aValues.length > bValues.length) {
+      return bValues.filter(bValue => aValues.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;
+    return aToB < bToA ? aToB : bToA;
+  }
+}
 //# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -1 +1 @@
-{"version":3,"file":"matching-utils.js","names":["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"],"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"],"mappings":";;;;;;;;;;;AA4BA;AAAsC;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,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiB,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,YAAY,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,8BAA8B,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,cAAc,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,iBAAiB,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,kBAAkB,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,yBAAyB,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,eAAe,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"}
+{"version":3,"file":"matching-utils.js","names":["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;AAAsC;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,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiB,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,YAAY,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,8BAA8B,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,cAAc,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,iBAAiB,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,kBAAkB,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,yBAAyB,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,eAAe,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,cAAc,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,uBAAuB,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"}

package/package.json

@@ -14,7 +14,7 @@
 		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
 	},
 	"license": "LGPL-3.0+",
-	"version": "4.0.1-alpha.1",
+	"version": "4.1.0-alpha.1",
 	"main": "./dist/index.js",
 	"engines": {
 		"node": ">=18"
@@ -38,12 +38,12 @@
 	"dependencies": {
 		"@natlibfi/marc-record": "^7.2.2",
 		"@natlibfi/marc-record-serializers": "^9.0.4",
-		"@natlibfi/melinda-commons": "^12.0.9",
+		"@natlibfi/melinda-commons": "^13.0.2",
 		"@natlibfi/sru-client": "^5.0.4",
 		"debug": "^4.3.4",
-		"isbn3": "^1.1.30",
+		"isbn3": "^1.1.33",
 		"moment": "^2.29.4",
-		"natural": "^6.1.4",
+		"natural": "^6.2.0",
 		"uuid": "^9.0.0",
 		"winston": "^3.8.2"
 	},
@@ -51,7 +51,7 @@
 		"@babel/cli": "^7.20.7",
 		"@babel/core": "^7.20.12",
 		"@babel/eslint-parser": "^7.19.1",
-		"@babel/node": "^7.20.0",
+		"@babel/node": "^7.20.7",
 		"@babel/preset-env": "^7.20.2",
 		"@babel/register": "^7.18.9",
 		"@natlibfi/eslint-config-melinda-backend": "^2.0.2",
@@ -63,7 +63,7 @@
 		"chai": "^4.3.7",
 		"chai-as-promised": "^7.1.1",
 		"cross-env": "^7.0.3",
-		"eslint": "^8.32.0",
+		"eslint": "^8.34.0",
 		"mocha": "^10.2.0",
 		"nodemon": "^2.0.20",
 		"nyc": "^15.1.0"

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

@@ -35,6 +35,7 @@ export {default as titleVersionOriginal} from './title-version-original';
 export {default as authors} from './authors';
 export {default as recordType} from './record-type';
 export {default as publicationTime} from './publication-time';
+export {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years';
 export {default as language} from './language';
 export {default as bibliographicLevel} from './bibliographic-level';
 export {default as melindaId} from './melinda-id';

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

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -27,6 +27,7 @@
 */
 
 import createDebugLogger from 'debug';
+import {getMatchCounts} from '../../../matching-utils';
 
 const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:language');
 const debugData = debug.extend('data');
@@ -40,19 +41,24 @@ export default () => ({
     const values041 = get041Values();
     debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
 
+    /*
     if (value008 && values041.length > 0) {
       debugData(`${label} There's both 008 and 041, searching for value in both`);
       const correspondingValue = values041.find(v => v === value008);
       debugData(`${label} Corresponding value: ${correspondingValue}`);
       return correspondingValue ? [correspondingValue] : [];
     }
+    */
 
     if (!value008 && values041.length < 1) {
       debugData(`{$label} No actual values found`);
       return [];
     }
 
-    return value008 ? [value008] : [values041[0]];
+    const allValues = value008 === undefined ? values041 : values041.concat(value008);
+    const uniqueSortedValues = [...new Set(allValues)].sort();
+
+    return uniqueSortedValues;
 
     function get008Value() {
       const value = record.get(/^008$/u)?.[0]?.value || undefined;
@@ -64,11 +70,9 @@ export default () => ({
 
       const code = value.slice(35, 38);
       debugData(`${label} 008 code: ${code}`);
-      return code === '|||' || code === '   ' || code === '^^^' ? undefined : code;
+      return isLangCodeForALanguage(code) ? code : undefined;
     }
 
-    // Main language for the resource: in the first f041 $a or f041 $d
-    // Should we get all $a or $d languages instead of just the first?
     // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes
 
     function get041Values() {
@@ -77,23 +81,67 @@ export default () => ({
         .map(({subfields}) => subfields)
         .flat()
         .filter(({code}) => code === 'a' || code === 'd')
+        .filter(({value}) => value && isLangCodeForALanguage(value))
         .map(({value}) => value);
     }
+
+    // Check if a string is a possible, validly formed language code for a single language
+    // Currently accept also codes in capitals
+    function isLangCodeForALanguage(code) {
+      if (code.length !== 3) {
+        debugData(`Code ${code} is not correct length (3) for a language code.`);
+        return false;
+      }
+      if (code === '|||' || code === '   ' || code === '^^^' || code === 'mul' || code === 'zxx') {
+        debugData(`Code ${code} is not code for a spesific language.`);
+        return false;
+      }
+      const langCodePattern = /^[a-z][a-z][a-z]$/ui;
+      if (!langCodePattern.test(code)) {
+        debugData(`Code ${code} is not valid as a language code`);
+        return false;
+      }
+      return true;
+    }
+
   },
+  // eslint-disable-next-line max-statements
   compare: (a, b) => {
-    debugData(`Comparing ${JSON.stringify(a[0])} and ${JSON.stringify(b[0])}`);
+    debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);
 
     if (a.length === 0 || b.length === 0) {
       debugData(`No language to compare`);
       return 0;
     }
 
-    debugData(`There are languages to compare`);
-
-    if (a[0] === b[0]) {
+    if (a.length === b.length && a.every((element, index) => element === b[index])) {
+      debugData(`All languages match`);
       return 0.1;
     }
 
-    return a[0] === 'und' || b[0] === 'und' ? 0.0 : -1.0;
+    const {matchingValues, possibleMatchValues, maxValues} = getMatchCounts(a, b);
+
+    if (matchingValues < 1) {
+      debug(`Both have languages, but none of these match.`);
+      return -1.0;
+    }
+    debug(`Both have languages, ${matchingValues}/${possibleMatchValues} valid languages match.`);
+    // ignore non-matches if there is mismatching amount of values
+    debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);
+    //we give some kind of penalty for mismatching amount of values instead of simple divide?
+    const missingCount = maxValues - possibleMatchValues;
+    const misMatchCount = possibleMatchValues - matchingValues;
+    debug(`\t missing: ${missingCount}`);
+    debug(`\t mismatches: ${misMatchCount}`);
+
+    const penaltyForMissing = 0.02 * (maxValues - possibleMatchValues);
+    const penaltyForMisMatch = 0.05 * (possibleMatchValues - matchingValues);
+    debug(`\t points: penaltyForMissing: ${penaltyForMissing}`);
+    debug(`\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);
+
+    const points = Number(Number(0.1 - penaltyForMisMatch - penaltyForMissing).toFixed(2));
+    debug(`Total points: ${points}`);
+
+    return points;
   }
 });

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

@@ -0,0 +1,62 @@
+/**
+*
+* @licstart  The following is the entire license notice for the JavaScript code in this file.
+*
+* Melinda record matching modules for Javascript
+*
+* Copyright (C) 2020-2023 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.
+*
+*/
+
+import {testStringOrNumber} from '../../../matching-utils';
+import createDebugLogger from 'debug';
+
+// We should also get copyright time and copyright/publication times from 26x
+
+export default () => ({
+  name: 'Publication time, allow consequent years',
+  extract: ({record}) => {
+    const value = record.get(/^008$/u)?.[0]?.value || undefined;
+    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
+  },
+  compare: (a, b) => {
+    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years');
+    debug(`Comparing ${a[0]} to ${b[0]}`);
+
+    const [firstA] = a;
+    const [firstB] = b;
+
+    if (firstA === firstB) {
+      return 0.1;
+    }
+
+    const firstANumber = parseInt(firstA, 10);
+    const firstBNumber = parseInt(firstB, 10);
+
+    if (isNaN(firstANumber) || isNaN(firstBNumber)) {
+      return -1;
+    }
+
+    // Handle consequent years as a match
+    // see publication-time for a version that does not handle consequent years as a match
+    return firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber ? 0.1 : -1;
+  }
+});

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

@@ -29,7 +29,7 @@
 import {testStringOrNumber} from '../../../matching-utils';
 
 // We should also get copyright time and copyright/publication times from 26x
-// We could also enable matching of records which have consequent publication years
+// see publication-time-allow-cons-years for a version allowing consequent years to match
 
 export default () => ({
   name: 'Publication time',

package/src/matching-utils.js

@@ -113,3 +113,31 @@ export function uniqueSubfields(subfields) {
     return arr;
   }, []);
 }
+
+export function getMatchCounts(aValues, bValues) {
+
+  const matchingValues = getMatchingValuesAmount(aValues, bValues);
+
+  return {
+    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
+    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
+    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;
+    }
+    if (aValues.length > bValues.length) {
+      return bValues.filter(bValue => aValues.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;
+
+    return aToB < bToA ? aToB : bToA;
+  }
+}