@natlibfi/melinda-record-matching 2.2.1-alpha.1 → 2.2.1-alpha.3

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

@@ -0,0 +1,112 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.default = void 0;
+var _debug = _interopRequireDefault(require("debug"));
+var _matchingUtils = require("../../../matching-utils");
+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:match-detection:features:standard-identifiers');
+const debugData = debug.extend('data');
+var _default = ({
+  pattern,
+  subfieldCodes
+}) => {
+  return {
+    extract,
+    compare
+  };
+  function extract(record) {
+    const [field] = record.get(pattern);
+    if (field) {
+      return field.subfields.filter(({
+        code
+      }) => subfieldCodes.includes(code)).map(({
+        code,
+        value
+      }) => ({
+        code,
+        value: (0, _matchingUtils.testStringOrNumber)(value) ? String(value).replace(/-/ug, '') : ''
+      })).filter(value => value);
+    }
+    return [];
+  }
+  function compare(a, b) {
+    if (a.length === 0 || b.length === 0) {
+      debugData(`No standardidentifiers to compare`);
+      return 0;
+    }
+    if (bothHaveValidIdentifiers()) {
+      const {
+        maxValues,
+        matchingValues
+      } = getValueCount(true);
+      return matchingValues / maxValues * 0.75;
+    }
+    const {
+      maxValues,
+      matchingValues
+    } = getValueCount();
+    return matchingValues / maxValues * 0.2;
+    function bothHaveValidIdentifiers() {
+      const aValues = a.filter(({
+        code
+      }) => code === 'a');
+      const bValues = a.filter(({
+        code
+      }) => code === 'a');
+      return aValues.length > 0 && bValues.length > 0;
+    }
+    function getValueCount(validOnly = false) {
+      const aValues = getIdentifiers(a);
+      const bValues = getIdentifiers(b);
+      return {
+        maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
+        matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length
+      };
+      function getIdentifiers(values) {
+        if (validOnly) {
+          return values.filter(({
+            code
+          }) => code === 'a').map(({
+            value
+          }) => value);
+        }
+        return values.map(({
+          value
+        }) => value);
+      }
+    }
+  }
+};
+exports.default = _default;
+//# sourceMappingURL=standard-identifier-factory.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"standard-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","pattern","subfieldCodes","extract","compare","record","field","get","subfields","filter","code","includes","map","value","testStringOrNumber","String","replace","a","b","length","bothHaveValidIdentifiers","maxValues","matchingValues","getValueCount","aValues","bValues","validOnly","getIdentifiers","aValue","some","bValue","values"],"sources":["../../../../src/match-detection/features/bib/standard-identifier-factory.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';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');\nconst debugData = debug.extend('data');\n\nexport default ({pattern, subfieldCodes}) => {\n  return {extract, compare};\n\n  function extract(record) {\n    const [field] = record.get(pattern);\n\n    if (field) {\n      return field.subfields\n        .filter(({code}) => subfieldCodes.includes(code))\n        .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))\n        .filter(value => value);\n    }\n\n    return [];\n  }\n\n  function compare(a, b) {\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No standardidentifiers to compare`);\n      return 0;\n    }\n\n    if (bothHaveValidIdentifiers()) {\n      const {maxValues, matchingValues} = getValueCount(true);\n      return matchingValues / maxValues * 0.75;\n    }\n\n    const {maxValues, matchingValues} = getValueCount();\n    return matchingValues / maxValues * 0.2;\n\n    function bothHaveValidIdentifiers() {\n      const aValues = a.filter(({code}) => code === 'a');\n      const bValues = a.filter(({code}) => code === 'a');\n      return aValues.length > 0 && bValues.length > 0;\n    }\n\n    function getValueCount(validOnly = false) {\n      const aValues = getIdentifiers(a);\n      const bValues = getIdentifiers(b);\n\n      return {\n        maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n        matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length\n      };\n\n      function getIdentifiers(values) {\n        if (validOnly) {\n          return values\n            .filter(({code}) => code === 'a')\n            .map(({value}) => value);\n        }\n\n        return values.map(({value}) => value);\n      }\n    }\n  }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA2D;AA7B3D;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,iFAAiF,CAAC;AAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,CAAC;EAACC,OAAO;EAAEC;AAAa,CAAC,KAAK;EAC3C,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAACE,MAAM,EAAE;IACvB,MAAM,CAACC,KAAK,CAAC,GAAGD,MAAM,CAACE,GAAG,CAACN,OAAO,CAAC;IAEnC,IAAIK,KAAK,EAAE;MACT,OAAOA,KAAK,CAACE,SAAS,CACnBC,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKR,aAAa,CAACS,QAAQ,CAACD,IAAI,CAAC,CAAC,CAChDE,GAAG,CAAC,CAAC;QAACF,IAAI;QAAEG;MAAK,CAAC,MAAM;QAACH,IAAI;QAAEG,KAAK,EAAE,IAAAC,iCAAkB,EAACD,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG;MAAE,CAAC,CAAC,CAAC,CAC1GP,MAAM,CAACI,KAAK,IAAIA,KAAK,CAAC;IAC3B;IAEA,OAAO,EAAE;EACX;EAEA,SAAST,OAAO,CAACa,CAAC,EAAEC,CAAC,EAAE;IACrB,IAAID,CAAC,CAACE,MAAM,KAAK,CAAC,IAAID,CAAC,CAACC,MAAM,KAAK,CAAC,EAAE;MACpCpB,SAAS,CAAE,mCAAkC,CAAC;MAC9C,OAAO,CAAC;IACV;IAEA,IAAIqB,wBAAwB,EAAE,EAAE;MAC9B,MAAM;QAACC,SAAS;QAAEC;MAAc,CAAC,GAAGC,aAAa,CAAC,IAAI,CAAC;MACvD,OAAOD,cAAc,GAAGD,SAAS,GAAG,IAAI;IAC1C;IAEA,MAAM;MAACA,SAAS;MAAEC;IAAc,CAAC,GAAGC,aAAa,EAAE;IACnD,OAAOD,cAAc,GAAGD,SAAS,GAAG,GAAG;IAEvC,SAASD,wBAAwB,GAAG;MAClC,MAAMI,OAAO,GAAGP,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,MAAMe,OAAO,GAAGR,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,OAAOc,OAAO,CAACL,MAAM,GAAG,CAAC,IAAIM,OAAO,CAACN,MAAM,GAAG,CAAC;IACjD;IAEA,SAASI,aAAa,CAACG,SAAS,GAAG,KAAK,EAAE;MACxC,MAAMF,OAAO,GAAGG,cAAc,CAACV,CAAC,CAAC;MACjC,MAAMQ,OAAO,GAAGE,cAAc,CAACT,CAAC,CAAC;MAEjC,OAAO;QACLG,SAAS,EAAEG,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM,GAAGK,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM;QAC5EG,cAAc,EAAEE,OAAO,CAACf,MAAM,CAACmB,MAAM,IAAIH,OAAO,CAACI,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAACX;MACtF,CAAC;MAED,SAASQ,cAAc,CAACI,MAAM,EAAE;QAC9B,IAAIL,SAAS,EAAE;UACb,OAAOK,MAAM,CACVtB,MAAM,CAAC,CAAC;YAACC;UAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCE,GAAG,CAAC,CAAC;YAACC;UAAK,CAAC,KAAKA,KAAK,CAAC;QAC5B;QAEA,OAAOkB,MAAM,CAACnB,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,CAAC;MACvC;IACF;EACF;AACF,CAAC;AAAA"}

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

@@ -0,0 +1,78 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.default = void 0;
+var _debug = _interopRequireDefault(require("debug"));
+var _natural = require("natural");
+var _matchingUtils = require("../../../matching-utils");
+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.
+*
+*/var _default = ({
+  treshold = 10
+} = {}) => ({
+  name: 'Title',
+  extract: record => {
+    const title = getTitle();
+    if ((0, _matchingUtils.testStringOrNumber)(title)) {
+      return [String(title).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
+    }
+    return [];
+    function getTitle() {
+      const [field] = record.get(/^245$/u);
+      if (field) {
+        return field.subfields.filter(({
+          code
+        }) => ['a', 'b'].includes(code)).map(({
+          value
+        }) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').join('');
+      }
+      return false;
+    }
+  },
+  compare: (a, b) => {
+    const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib/title');
+    const distance = (0, _natural.LevenshteinDistance)(a[0], b[0]);
+    if (distance === 0) {
+      return 0.5;
+    }
+    const maxLength = getMaxLength();
+    const percentage = distance / maxLength * 100;
+    debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);
+    if (percentage <= treshold) {
+      return 0.3;
+    }
+    return -0.5;
+    function getMaxLength() {
+      return a[0].length > b[0].length ? a[0].length : b[0].length;
+    }
+  }
+});
+exports.default = _default;
+//# sourceMappingURL=title.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"title.js","names":["treshold","name","extract","record","title","getTitle","testStringOrNumber","String","replace","toLowerCase","field","get","subfields","filter","code","includes","map","value","join","compare","a","b","debug","createDebugLogger","distance","leven","maxLength","getMaxLength","percentage","length"],"sources":["../../../../src/match-detection/features/bib/title.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';\nimport {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nexport default ({treshold = 10} = {}) => ({\n  name: 'Title',\n  extract: record => {\n    const title = getTitle();\n\n    if (testStringOrNumber(title)) {\n      return [String(title).replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase()];\n    }\n\n    return [];\n\n    function getTitle() {\n      const [field] = record.get(/^245$/u);\n\n      if (field) {\n        return field.subfields\n          .filter(({code}) => ['a', 'b'].includes(code))\n          .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n          .join('');\n      }\n      return false;\n    }\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n    const distance = leven(a[0], b[0]);\n\n    if (distance === 0) {\n      return 0.5;\n    }\n\n    const maxLength = getMaxLength();\n    const percentage = distance / maxLength * 100;\n\n    debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);\n\n    if (percentage <= treshold) {\n      return 0.3;\n    }\n\n    return -0.5;\n\n    function getMaxLength() {\n      return a[0].length > b[0].length ? a[0].length : b[0].length;\n    }\n\n  }\n});\n"],"mappings":";;;;;;AA4BA;AACA;AACA;AAA2D;AA9B3D;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,EA1BA,eAgCe,CAAC;EAACA,QAAQ,GAAG;AAAE,CAAC,GAAG,CAAC,CAAC,MAAM;EACxCC,IAAI,EAAE,OAAO;EACbC,OAAO,EAAEC,MAAM,IAAI;IACjB,MAAMC,KAAK,GAAGC,QAAQ,EAAE;IAExB,IAAI,IAAAC,iCAAkB,EAACF,KAAK,CAAC,EAAE;MAC7B,OAAO,CAACG,MAAM,CAACH,KAAK,CAAC,CAACI,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CAACC,WAAW,EAAE,CAAC;IAC/E;IAEA,OAAO,EAAE;IAET,SAASJ,QAAQ,GAAG;MAClB,MAAM,CAACK,KAAK,CAAC,GAAGP,MAAM,CAACQ,GAAG,CAAC,QAAQ,CAAC;MAEpC,IAAID,KAAK,EAAE;QACT,OAAOA,KAAK,CAACE,SAAS,CACnBC,MAAM,CAAC,CAAC;UAACC;QAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,IAAI,CAAC,CAAC,CAC7CE,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAK,IAAAX,iCAAkB,EAACW,KAAK,CAAC,GAAGV,MAAM,CAACU,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEC,IAAI,CAAC,EAAE,CAAC;MACb;MACA,OAAO,KAAK;IACd;EACF,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,sEAAsE,CAAC;IACvG,MAAMC,QAAQ,GAAG,IAAAC,4BAAK,EAACL,CAAC,CAAC,CAAC,CAAC,EAAEC,CAAC,CAAC,CAAC,CAAC,CAAC;IAElC,IAAIG,QAAQ,KAAK,CAAC,EAAE;MAClB,OAAO,GAAG;IACZ;IAEA,MAAME,SAAS,GAAGC,YAAY,EAAE;IAChC,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAS,GAAG,GAAG;IAE7CJ,KAAK,CAAE,IAAGF,CAAE,SAAQC,CAAE,mBAAkBK,SAAU,gBAAeF,QAAS,kBAAiBI,UAAW,EAAC,CAAC;IAExG,IAAIA,UAAU,IAAI5B,QAAQ,EAAE;MAC1B,OAAO,GAAG;IACZ;IAEA,OAAO,CAAC,GAAG;IAEX,SAAS2B,YAAY,GAAG;MACtB,OAAOP,CAAC,CAAC,CAAC,CAAC,CAACS,MAAM,GAAGR,CAAC,CAAC,CAAC,CAAC,CAACQ,MAAM,GAAGT,CAAC,CAAC,CAAC,CAAC,CAACS,MAAM,GAAGR,CAAC,CAAC,CAAC,CAAC,CAACQ,MAAM;IAC9D;EAEF;AACF,CAAC,CAAC;AAAA"}

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

@@ -0,0 +1,11 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.bib = void 0;
+var bib = _interopRequireWildcard(require("./bib"));
+exports.bib = bib;
+function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
+function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
+//# sourceMappingURL=index.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":[],"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;AAA6B;AAAA;AAAA"}

package/dist/match-detection/index.js

@@ -0,0 +1,162 @@
+"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(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
+function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
+function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
+/* eslint-disable no-console */
+/**
+*
+* @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) => (recordA, recordB) => {
+  const minProbabilityQuantifier = 0.5;
+  const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection');
+  const debugData = debug.extend('data');
+  debugData(`Strategy: ${JSON.stringify(strategy)}`);
+  debugData(`Treshold: ${JSON.stringify(treshold)}`);
+  debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
+  const featuresA = extractFeatures(recordA);
+  debug(`We got an array of records: ${Array.isArray(recordB)}`);
+  const recordsB = Array.isArray(recordB) ? recordB : [recordB];
+  const detectionResults = recordsB.map(record => actualDetection(featuresA, record));
+  return Array.isArray(recordB) ? detectionResults : detectionResults[0];
+  function actualDetection(featuresA, record) {
+    const featuresB = extractFeatures(record);
+    debugData(`Features (a): ${JSON.stringify(featuresA)}`);
+    debugData(`Features (b): ${JSON.stringify(featuresB)}`);
+    const featurePairs = generateFeaturePairs(featuresA, featuresB);
+    const similarityVector = generateSimilarityVector(featurePairs);
+    if (similarityVector.some(v => v >= minProbabilityQuantifier)) {
+      const probability = calculateprobability(similarityVector);
+      debug(`probability: ${probability} (Treshold: ${treshold})`);
+      return returnResult({
+        match: probability >= treshold,
+        probability
+      });
+    }
+    debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);
+    return returnResult({
+      match: false,
+      probability: 0.0
+    });
+  }
+  function extractFeatures(record) {
+    return strategy.reduce((acc, {
+      name,
+      extract
+    }) => acc.concat({
+      name,
+      value: extract(record)
+    }), []);
+  }
+  function returnResult(result) {
+    if (returnStrategy) {
+      debug(`Returning detection strategy with the result`);
+      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);
+    return strategyNames || [];
+  }
+  function calculateprobability(similarityVector) {
+    const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);
+    return probability > 1.0 ? 1.0 : probability;
+  }
+  function generateSimilarityVector(featurePairs) {
+    const compared = featurePairs.map(({
+      name,
+      a,
+      b
+    }) => {
+      const {
+        compare
+      } = strategy.find(({
+        name: featureName
+      }) => name === featureName);
+      const points = compare(a, b);
+      return {
+        name,
+        points
+      };
+    });
+    debugData(`Points: ${JSON.stringify(compared)}`);
+    return compared.map(({
+      points
+    }) => points);
+  }
+  function generateFeaturePairs(featuresA, 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({
+        name,
+        a: value,
+        b: featuresB[index].value
+      }), []).filter(({
+        a,
+        b
+      }) => {
+        if (a.length === 0 || b.length === 0) {
+          return false;
+        }
+        return true;
+      });
+    }
+    function findMissing() {
+      return featuresA.map(({
+        name
+      }) => name).filter(v => pairs.some(({
+        name
+      }) => name === v) === false);
+    }
+  }
+};
+exports.default = _default;
+//# sourceMappingURL=index.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","names":["strategy","treshold","returnStrategy","recordA","recordB","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","Array","isArray","recordsB","detectionResults","map","record","actualDetection","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateprobability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","a","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","index","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["/* eslint-disable no-console */\n/**\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) => (recordA, recordB) => {\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)}`);\n  debugData(`Treshold: ${JSON.stringify(treshold)}`);\n  debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n\n  const featuresA = extractFeatures(recordA);\n\n  debug(`We got an array of records: ${Array.isArray(recordB)}`);\n  const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n  const detectionResults = recordsB.map(record => actualDetection(featuresA, record));\n\n  return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n  function actualDetection(featuresA, record) {\n\n    const featuresB = extractFeatures(record);\n\n    debugData(`Features (a): ${JSON.stringify(featuresA)}`);\n    debugData(`Features (b): ${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) {\n    return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract(record)}), []);\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":";;;;;;AA6BA;AACA;AAAuC;AAAA;AAAA;AAAA;AA9BvC;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;AACA;AA1BA,eAiCe,CAAC;EAACA,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,KAAK,CAACC,OAAO,EAAEC,OAAO,KAAK;EAC3F,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,CAACX,QAAQ,CAAE,EAAC,CAAC;EAClDQ,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACV,QAAQ,CAAE,EAAC,CAAC;EAClDO,SAAS,CAAE,mBAAkBE,IAAI,CAACC,SAAS,CAACT,cAAc,CAAE,EAAC,CAAC;EAE9D,MAAMU,SAAS,GAAGC,eAAe,CAACV,OAAO,CAAC;EAE1CG,KAAK,CAAE,+BAA8BQ,KAAK,CAACC,OAAO,CAACX,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMY,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACX,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMa,gBAAgB,GAAGD,QAAQ,CAACE,GAAG,CAACC,MAAM,IAAIC,eAAe,CAACR,SAAS,EAAEO,MAAM,CAAC,CAAC;EAEnF,OAAOL,KAAK,CAACC,OAAO,CAACX,OAAO,CAAC,GAAGa,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAe,CAACR,SAAS,EAAEO,MAAM,EAAE;IAE1C,MAAME,SAAS,GAAGR,eAAe,CAACM,MAAM,CAAC;IAEzCX,SAAS,CAAE,iBAAgBE,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IACvDJ,SAAS,CAAE,iBAAgBE,IAAI,CAACC,SAAS,CAACU,SAAS,CAAE,EAAC,CAAC;IAEvD,MAAMC,YAAY,GAAGC,oBAAoB,CAACX,SAAS,EAAES,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAItB,wBAAwB,CAAC,EAAE;MAC7D,MAAMuB,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DlB,KAAK,CAAE,gBAAesB,WAAY,eAAc3B,QAAS,GAAE,CAAC;MAC5D,OAAO6B,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAI3B,QAAQ;QAAE2B;MAAW,CAAC,CAAC;IACpE;IAEApB,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAOyB,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASf,eAAe,CAACM,MAAM,EAAE;IAC/B,OAAOnB,QAAQ,CAACgC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAChB,MAAM;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EAClG;EAEA,SAASW,YAAY,CAACQ,MAAM,EAAE;IAC5B,IAAIpC,cAAc,EAAE;MAClBI,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMiC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAE5B,QAAQ,EAAEwC,cAAc,CAACxC,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HO,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAAC4B,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAc,CAACxC,QAAQ,EAAE;IAChC,MAAMyC,aAAa,GAAGzC,QAAQ,CAACkB,GAAG,CAACwB,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,oBAAoB,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,wBAAwB,CAACH,YAAY,EAAE;IAC9C,MAAMqB,QAAQ,GAAGrB,YAAY,CAACJ,GAAG,CAAC,CAAC;MAACgB,IAAI;MAAEU,CAAC;MAAEC;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAG9C,QAAQ,CAAC+C,IAAI,CAAC,CAAC;QAACb,IAAI,EAAEc;MAAW,CAAC,KAAKd,IAAI,KAAKc,WAAW,CAAC;MAC9E,MAAMC,MAAM,GAAGH,OAAO,CAACF,CAAC,EAAEC,CAAC,CAAC;MAC5B,OAAO;QAACX,IAAI;QAAEe;MAAM,CAAC;IACvB,CAAC,CAAC;IAEFzC,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACgC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAACzB,GAAG,CAAC,CAAC;MAAC+B;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAAS1B,oBAAoB,CAACX,SAAS,EAAES,SAAS,EAAE;IAClD,MAAM6B,KAAK,GAAGC,aAAa,EAAE;IAC7B,MAAMC,eAAe,GAAGC,WAAW,EAAE;IAErC/C,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAACyC,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAa,GAAG;MACvB,OAAOvC,SAAS,CACboB,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAEiB,KAAK,KAAKrB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJU,CAAC,EAAEP,KAAK;QACRQ,CAAC,EAAExB,SAAS,CAACiC,KAAK,CAAC,CAACjB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNkB,MAAM,CAAC,CAAC;QAACX,CAAC;QAAEC;MAAC,CAAC,KAAK;QAClB,IAAID,CAAC,CAACY,MAAM,KAAK,CAAC,IAAIX,CAAC,CAACW,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASH,WAAW,GAAG;MACrB,OAAOzC,SAAS,CACbM,GAAG,CAAC,CAAC;QAACgB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBqB,MAAM,CAAC5B,CAAC,IAAIuB,KAAK,CAACxB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA"}

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

@@ -0,0 +1,88 @@
+"use strict";
+
+var _fixugen = _interopRequireDefault(require("@natlibfi/fixugen"));
+var _fixura = require("@natlibfi/fixura");
+var _chai = require("chai");
+var _marcRecord = require("@natlibfi/marc-record");
+var features = _interopRequireWildcard(require("./features"));
+var _ = _interopRequireDefault(require("."));
+var _util = require("util");
+var _debug = _interopRequireDefault(require("debug"));
+function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
+function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
+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.
+*
+*/
+
+const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:test');
+const debugData = debug.extend('data');
+describe('match-detection', () => {
+  (0, _fixugen.default)({
+    path: [__dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],
+    useMetadataFile: true,
+    recurse: false,
+    fixura: {
+      reader: _fixura.READERS.JSON
+    },
+    callback: ({
+      getFixture,
+      options,
+      expectedResults,
+      array,
+      enabled = true
+    }) => {
+      if (!enabled) {
+        debug(`*** DISABLED TEST! ***`);
+        return;
+      }
+      const detect = (0, _.default)(formatOptions());
+      const recordA = new _marcRecord.MarcRecord(getFixture('recordA.json'), {
+        subfieldValues: false
+      });
+      debugData((0, _util.inspect)(recordA));
+      debug(`Our recordB is an array of records: ${array}`);
+      const recordB = array ? getFixture('recordB.json').map(recordJson => new _marcRecord.MarcRecord(recordJson, {
+        subfieldValues: false
+      })) : new _marcRecord.MarcRecord(getFixture('recordB.json'), {
+        subfieldValues: false
+      });
+      debugData((0, _util.inspect)(recordB));
+      const results = detect(recordA, recordB);
+      debugData(`${JSON.stringify(results)}`);
+      (0, _chai.expect)(results).to.eql(expectedResults);
+      function formatOptions() {
+        const contextFeatures = features[options.strategy.type];
+        return {
+          ...options,
+          strategy: options.strategy.features.map(v => contextFeatures[v]())
+        };
+      }
+    }
+  });
+});
+//# sourceMappingURL=index.spec.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"index.spec.js","names":["debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","recurse","fixura","reader","READERS","JSON","callback","getFixture","options","expectedResults","array","enabled","detect","createDetectionInterface","formatOptions","recordA","MarcRecord","subfieldValues","inspect","recordB","map","recordJson","results","stringify","expect","to","eql","contextFeatures","features","strategy","type","v"],"sources":["../../src/match-detection/index.spec.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 generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './features';\nimport createDetectionInterface from '.';\nimport {inspect} from 'util';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:test');\nconst debugData = debug.extend('data');\n\ndescribe('match-detection', () => {\n  generateTests({\n    path: [__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      expect(results).to.eql(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":";;AA4BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAsC;AAAA;AAAA;AAnCtC;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;;AAWA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,wDAAwD,CAAC;AACzF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAEtCC,QAAQ,CAAC,iBAAiB,EAAE,MAAM;EAChC,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,OAAO,CAAC;IAC1EC,eAAe,EAAE,IAAI;IACrBC,OAAO,EAAE,KAAK;IACdC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IACDC,QAAQ,EAAE,CAAC;MAACC,UAAU;MAAEC,OAAO;MAAEC,eAAe;MAAEC,KAAK;MAAEC,OAAO,GAAG;IAAI,CAAC,KAAK;MAE3E,IAAI,CAACA,OAAO,EAAE;QACZnB,KAAK,CAAE,wBAAuB,CAAC;QAC/B;MACF;MAEA,MAAMoB,MAAM,GAAG,IAAAC,SAAwB,EAACC,aAAa,EAAE,CAAC;MACxD,MAAMC,OAAO,GAAG,IAAIC,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACnFvB,SAAS,CAAC,IAAAwB,aAAO,EAACH,OAAO,CAAC,CAAC;MAE3BvB,KAAK,CAAE,uCAAsCkB,KAAM,EAAC,CAAC;MACrD,MAAMS,OAAO,GAAGT,KAAK,GACjBH,UAAU,CAAC,cAAc,CAAC,CAACa,GAAG,CAACC,UAAU,IAAI,IAAIL,sBAAU,CAACK,UAAU,EAAE;QAACJ,cAAc,EAAE;MAAK,CAAC,CAAC,CAAC,GACjG,IAAID,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACvEvB,SAAS,CAAC,IAAAwB,aAAO,EAACC,OAAO,CAAC,CAAC;MAE3B,MAAMG,OAAO,GAAGV,MAAM,CAACG,OAAO,EAAEI,OAAO,CAAC;MACxCzB,SAAS,CAAE,GAAEW,IAAI,CAACkB,SAAS,CAACD,OAAO,CAAE,EAAC,CAAC;MAEvC,IAAAE,YAAM,EAACF,OAAO,CAAC,CAACG,EAAE,CAACC,GAAG,CAACjB,eAAe,CAAC;MAEvC,SAASK,aAAa,GAAG;QACvB,MAAMa,eAAe,GAAGC,QAAQ,CAACpB,OAAO,CAACqB,QAAQ,CAACC,IAAI,CAAC;QAEvD,OAAO;UACL,GAAGtB,OAAO;UACVqB,QAAQ,EAAErB,OAAO,CAACqB,QAAQ,CAACD,QAAQ,CAACR,GAAG,CAACW,CAAC,IAAIJ,eAAe,CAACI,CAAC,CAAC,EAAE;QACnE,CAAC;MACH;IACF;EACF,CAAC,CAAC;AACJ,CAAC,CAAC"}

package/dist/matching-utils.js

@@ -0,0 +1,95 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.getMelindaIdsF035 = getMelindaIdsF035;
+exports.getSubfieldValues = getSubfieldValues;
+exports.testStringOrNumber = testStringOrNumber;
+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');
+
+// eslint-disable-next-line max-statements
+function getMelindaIdsF035(record) {
+  const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:melinda-id');
+  const debugData = debug.extend('data');
+  const f035s = record.getFields('035');
+  if (f035s.length < 1) {
+    debug(`No f035s found.`);
+    return [];
+  }
+  const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));
+  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)}`);
+  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
+  debugData(`Validating SID field ${JSON.stringify(field)}`);
+  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;
+}
+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);
+}
+function testStringOrNumber(value) {
+  if (typeof value === 'string' || typeof value === 'number') {
+    return true;
+  }
+  return false;
+}
+//# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -0,0 +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"],"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\n// eslint-disable-next-line max-statements\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"],"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;;AAEtC;AACO,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"}