@natlibfi/melinda-record-matching 3.0.0-alpha.1 → 3.0.1-alpha.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/candidate-search/index.js +0 -4
- package/dist/candidate-search/index.js.map +1 -1
- package/dist/candidate-search/index.spec.js +1 -5
- package/dist/candidate-search/index.spec.js.map +1 -1
- package/dist/candidate-search/query-list/bib.js +0 -2
- package/dist/candidate-search/query-list/bib.js.map +1 -1
- package/dist/index.js +0 -1
- package/dist/index.js.map +1 -1
- package/dist/match-detection/features/bib/all-source-ids.js +0 -1
- package/dist/match-detection/features/bib/all-source-ids.js.map +1 -1
- package/dist/match-detection/features/bib/bibliographic-level.js +2 -1
- package/dist/match-detection/features/bib/bibliographic-level.js.map +1 -1
- package/dist/match-detection/features/bib/host-component.js +2 -1
- package/dist/match-detection/features/bib/host-component.js.map +1 -1
- package/dist/match-detection/features/bib/index.js +7 -0
- package/dist/match-detection/features/bib/index.js.map +1 -1
- package/dist/match-detection/features/bib/index.spec.js +0 -1
- package/dist/match-detection/features/bib/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/isbn.js +2 -1
- package/dist/match-detection/features/bib/isbn.js.map +1 -1
- package/dist/match-detection/features/bib/issn.js +2 -1
- package/dist/match-detection/features/bib/issn.js.map +1 -1
- package/dist/match-detection/features/bib/language.js +6 -6
- package/dist/match-detection/features/bib/language.js.map +1 -1
- package/dist/match-detection/features/bib/melinda-id.js +2 -1
- package/dist/match-detection/features/bib/melinda-id.js.map +1 -1
- package/dist/match-detection/features/bib/melinda-identifier-factory.js +0 -2
- package/dist/match-detection/features/bib/melinda-identifier-factory.js.map +1 -1
- package/dist/match-detection/features/bib/other-standard-identifier.js +2 -1
- package/dist/match-detection/features/bib/other-standard-identifier.js.map +1 -1
- package/dist/match-detection/features/bib/title-version-original.js +79 -0
- package/dist/match-detection/features/bib/title-version-original.js.map +1 -0
- package/dist/match-detection/features/bib/title.js +24 -6
- package/dist/match-detection/features/bib/title.js.map +1 -1
- package/dist/match-detection/index.js +11 -7
- package/dist/match-detection/index.js.map +1 -1
- package/dist/matching-utils.js +0 -2
- package/dist/matching-utils.js.map +1 -1
- package/package.json +8 -8
- package/src/candidate-search/index.js +1 -3
- package/src/candidate-search/index.spec.js +1 -3
- package/src/candidate-search/query-list/bib.js +0 -1
- package/src/index.js +0 -1
- package/src/match-detection/features/bib/all-source-ids.js +0 -1
- package/src/match-detection/features/bib/index.js +1 -0
- package/src/match-detection/features/bib/index.spec.js +1 -1
- package/src/match-detection/features/bib/language.js +6 -6
- package/src/match-detection/features/bib/melinda-identifier-factory.js +0 -1
- package/src/match-detection/features/bib/title-version-original.js +78 -0
- package/src/match-detection/features/bib/title.js +22 -4
- package/src/match-detection/index.js +7 -7
- package/src/matching-utils.js +0 -1
|
@@ -75,11 +75,18 @@ Object.defineProperty(exports, "title", {
|
|
|
75
75
|
return _title.default;
|
|
76
76
|
}
|
|
77
77
|
});
|
|
78
|
+
Object.defineProperty(exports, "titleVersionOriginal", {
|
|
79
|
+
enumerable: true,
|
|
80
|
+
get: function () {
|
|
81
|
+
return _titleVersionOriginal.default;
|
|
82
|
+
}
|
|
83
|
+
});
|
|
78
84
|
var _hostComponent = _interopRequireDefault(require("./host-component"));
|
|
79
85
|
var _isbn = _interopRequireDefault(require("./isbn"));
|
|
80
86
|
var _issn = _interopRequireDefault(require("./issn"));
|
|
81
87
|
var _otherStandardIdentifier = _interopRequireDefault(require("./other-standard-identifier"));
|
|
82
88
|
var _title = _interopRequireDefault(require("./title"));
|
|
89
|
+
var _titleVersionOriginal = _interopRequireDefault(require("./title-version-original"));
|
|
83
90
|
var _authors = _interopRequireDefault(require("./authors"));
|
|
84
91
|
var _recordType = _interopRequireDefault(require("./record-type"));
|
|
85
92
|
var _publicationTime = _interopRequireDefault(require("./publication-time"));
|
|
@@ -1 +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 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":"
|
|
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"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.spec.js","names":["describe","generateTests","path","__dirname","useMetadataFile","fixura","reader","READERS","JSON","callback","enabled","feature","options","type","expectations","expectedFeatures","inputRecord","record","MarcRecord","subfieldValues","extract","features","expect","to","eql","featuresA","featuresB","expectedPoints","compare","equal","Error"],"sources":["../../../../src/match-detection/features/bib/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 '.';\n\n\ndescribe('match-detection/features/bib/', () => {\n generateTests({\n path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n useMetadataFile: true,\n fixura: {\n reader: READERS.JSON\n },\n
|
|
1
|
+
{"version":3,"file":"index.spec.js","names":["describe","generateTests","path","__dirname","useMetadataFile","fixura","reader","READERS","JSON","callback","enabled","feature","options","type","expectations","expectedFeatures","inputRecord","record","MarcRecord","subfieldValues","extract","features","expect","to","eql","featuresA","featuresB","expectedPoints","compare","equal","Error"],"sources":["../../../../src/match-detection/features/bib/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 '.';\n\n\ndescribe('match-detection/features/bib/', () => {\n generateTests({\n path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n useMetadataFile: true,\n fixura: {\n reader: READERS.JSON\n },\n\n callback: ({enabled = true, feature, options, type, ...expectations}) => {\n\n if (!enabled) {\n return;\n }\n\n if (type === 'extract') {\n const {expectedFeatures, inputRecord} = expectations;\n const record = new MarcRecord(inputRecord, {subfieldValues: false});\n const {extract} = features[feature](options);\n\n expect(extract({record})).to.eql(expectedFeatures);\n return;\n }\n\n if (type === 'compare') {\n const {featuresA, featuresB, expectedPoints} = expectations;\n const {compare} = features[feature](options);\n\n expect(compare(featuresA, featuresB)).to.equal(expectedPoints);\n return;\n }\n\n throw new Error(`Invalid type ${type}`);\n }\n });\n});\n"],"mappings":";;AA4BA;AACA;AACA;AACA;AACA;AAA8B;AAAA;AAAA;AAhC9B;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;;AASAA,QAAQ,CAAC,+BAA+B,EAAE,MAAM;EAC9C,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,UAAU,EAAE,KAAK,CAAC;IAChGC,eAAe,EAAE,IAAI;IACrBC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IAEDC,QAAQ,EAAE,CAAC;MAACC,OAAO,GAAG,IAAI;MAAEC,OAAO;MAAEC,OAAO;MAAEC,IAAI;MAAE,GAAGC;IAAY,CAAC,KAAK;MAEvE,IAAI,CAACJ,OAAO,EAAE;QACZ;MACF;MAEA,IAAIG,IAAI,KAAK,SAAS,EAAE;QACtB,MAAM;UAACE,gBAAgB;UAAEC;QAAW,CAAC,GAAGF,YAAY;QACpD,MAAMG,MAAM,GAAG,IAAIC,sBAAU,CAACF,WAAW,EAAE;UAACG,cAAc,EAAE;QAAK,CAAC,CAAC;QACnE,MAAM;UAACC;QAAO,CAAC,GAAGC,QAAQ,CAACV,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAU,YAAM,EAACF,OAAO,CAAC;UAACH;QAAM,CAAC,CAAC,CAAC,CAACM,EAAE,CAACC,GAAG,CAACT,gBAAgB,CAAC;QAClD;MACF;MAEA,IAAIF,IAAI,KAAK,SAAS,EAAE;QACtB,MAAM;UAACY,SAAS;UAAEC,SAAS;UAAEC;QAAc,CAAC,GAAGb,YAAY;QAC3D,MAAM;UAACc;QAAO,CAAC,GAAGP,QAAQ,CAACV,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAU,YAAM,EAACM,OAAO,CAACH,SAAS,EAAEC,SAAS,CAAC,CAAC,CAACH,EAAE,CAACM,KAAK,CAACF,cAAc,CAAC;QAC9D;MACF;MAEA,MAAM,IAAIG,KAAK,CAAE,gBAAejB,IAAK,EAAC,CAAC;IACzC;EACF,CAAC,CAAC;AACJ,CAAC,CAAC"}
|
|
@@ -32,7 +32,8 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
32
32
|
* @licend The above is the entire license notice
|
|
33
33
|
* for the JavaScript code in this file.
|
|
34
34
|
*
|
|
35
|
-
*/
|
|
35
|
+
*/
|
|
36
|
+
var _default = () => {
|
|
36
37
|
const {
|
|
37
38
|
extract,
|
|
38
39
|
compare
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"isbn.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/isbn.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^020$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'ISBN'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;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,
|
|
1
|
+
{"version":3,"file":"isbn.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/isbn.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^020$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'ISBN'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;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,eA8Be,MAAM;EACnB,MAAM;IAACA,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG;EAAC,CAAC,CAAC;EAC1F,OAAO;IAACJ,OAAO;IAAEC,OAAO;IAAEI,IAAI,EAAE;EAAM,CAAC;AACzC,CAAC;AAAA"}
|
|
@@ -32,7 +32,8 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
32
32
|
* @licend The above is the entire license notice
|
|
33
33
|
* for the JavaScript code in this file.
|
|
34
34
|
*
|
|
35
|
-
*/
|
|
35
|
+
*/
|
|
36
|
+
var _default = () => {
|
|
36
37
|
const {
|
|
37
38
|
extract,
|
|
38
39
|
compare
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"issn.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/issn.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^022$/u, subfieldCodes: ['a', 'z', 'y']});\n return {extract, compare, name: 'ISSN'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;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,
|
|
1
|
+
{"version":3,"file":"issn.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/issn.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^022$/u, subfieldCodes: ['a', 'z', 'y']});\n return {extract, compare, name: 'ISSN'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,eA8Be,MAAM;EACnB,MAAM;IAACA,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG;EAAC,CAAC,CAAC;EAC/F,OAAO;IAACJ,OAAO;IAAEC,OAAO;IAAEI,IAAI,EAAE;EAAM,CAAC;AACzC,CAAC;AAAA"}
|
|
@@ -6,7 +6,6 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
6
6
|
exports.default = void 0;
|
|
7
7
|
var _debug = _interopRequireDefault(require("debug"));
|
|
8
8
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
9
|
-
/* eslint-disable max-statements */
|
|
10
9
|
/**
|
|
11
10
|
*
|
|
12
11
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -48,29 +47,30 @@ var _default = () => ({
|
|
|
48
47
|
const values041 = get041Values();
|
|
49
48
|
debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
|
|
50
49
|
if (value008 && values041.length > 0) {
|
|
51
|
-
debugData(
|
|
50
|
+
debugData(`${label} There's both 008 and 041, searching for value in both`);
|
|
52
51
|
const correspondingValue = values041.find(v => v === value008);
|
|
53
|
-
debugData(
|
|
52
|
+
debugData(`${label} Corresponding value: ${correspondingValue}`);
|
|
54
53
|
return correspondingValue ? [correspondingValue] : [];
|
|
55
54
|
}
|
|
56
55
|
if (!value008 && values041.length < 1) {
|
|
57
|
-
debugData(`No actual values found`);
|
|
56
|
+
debugData(`{$label} No actual values found`);
|
|
58
57
|
return [];
|
|
59
58
|
}
|
|
60
59
|
return value008 ? [value008] : [values041[0]];
|
|
61
60
|
function get008Value() {
|
|
62
61
|
var _record$get, _record$get$;
|
|
63
62
|
const value = ((_record$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;
|
|
64
|
-
debugData(
|
|
63
|
+
debugData(`${label} 008 value: ${value}`);
|
|
65
64
|
if (!value) {
|
|
66
65
|
return undefined;
|
|
67
66
|
}
|
|
68
67
|
const code = value.slice(35, 38);
|
|
69
|
-
debugData(
|
|
68
|
+
debugData(`${label} 008 code: ${code}`);
|
|
70
69
|
return code === '|||' || code === ' ' || code === '^^^' ? undefined : code;
|
|
71
70
|
}
|
|
72
71
|
|
|
73
72
|
// Main language for the resource: in the first f041 $a or f041 $d
|
|
73
|
+
// Should we get all $a or $d languages instead of just the first?
|
|
74
74
|
// Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes
|
|
75
75
|
|
|
76
76
|
function get041Values() {
|
|
@@ -1 +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":["
|
|
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"}
|
|
@@ -32,7 +32,8 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
32
32
|
* @licend The above is the entire license notice
|
|
33
33
|
* for the JavaScript code in this file.
|
|
34
34
|
*
|
|
35
|
-
*/
|
|
35
|
+
*/
|
|
36
|
+
var _default = () => {
|
|
36
37
|
const {
|
|
37
38
|
extract,
|
|
38
39
|
compare
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"melinda-id.js","names":["extract","compare","createInterface","name"],"sources":["../../../../src/match-detection/features/bib/melinda-id.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 createInterface from './melinda-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface();\n return {extract, compare, name: 'melinda-id'};\n};\n"],"mappings":";;;;;;AA4BA;AAA2D;AA5B3D;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,
|
|
1
|
+
{"version":3,"file":"melinda-id.js","names":["extract","compare","createInterface","name"],"sources":["../../../../src/match-detection/features/bib/melinda-id.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 createInterface from './melinda-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface();\n return {extract, compare, name: 'melinda-id'};\n};\n"],"mappings":";;;;;;AA4BA;AAA2D;AA5B3D;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,eA8Be,MAAM;EACnB,MAAM;IAACA,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,iCAAe,GAAE;EAC5C,OAAO;IAACF,OAAO;IAAEC,OAAO;IAAEE,IAAI,EAAE;EAAY,CAAC;AAC/C,CAAC;AAAA"}
|
|
@@ -64,8 +64,6 @@ var _default = () => {
|
|
|
64
64
|
f035MelindaIds
|
|
65
65
|
};
|
|
66
66
|
}
|
|
67
|
-
|
|
68
|
-
// eslint-disable-next-line max-statements
|
|
69
67
|
function compare(a, b) {
|
|
70
68
|
if (a.isMelindaRecord && b.isMelindaRecord && a.f001 === b.f001) {
|
|
71
69
|
debugData(`Melinda record's A f001 ${a.f001} matches Melinda record's B f001 ${a.f001}`);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"melinda-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","extract","compare","record","recordExternal","isMelindaRecord","get","some","f003","value","f001","map","field","f035MelindaIds","getMelindaIdsF035","undefined","length","label","a","b","id","JSON","stringify","idA","idB"],"sources":["../../../../src/match-detection/features/bib/melinda-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 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 {getMelindaIdsF035} from '../../../matching-utils';\n\n// 003+001 FI-MELINDA <melinda-id>\n// 035 $a (FI-MELINDA)<melinda-id>\n// 035 $z (FI-MELINDA)<melinda-id>\n// 035 $a FCC<melinda-id>\n// 035 $z FCC<melinda-id>\n// melinda-id = 001234567\n\nexport default () => {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/melinda-id');\n const debugData = debug.extend('data');\n\n return {extract, compare};\n\n function extract({record, recordExternal}) {\n\n const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');\n const [f001] = record.get('001').map(field => field.value);\n const f035MelindaIds = getMelindaIdsF035(record);\n\n if (\n isMelindaRecord === undefined &&\n f001 === undefined &&\n f035MelindaIds.length < 1) {\n\n debug(`${recordExternal.label} No Melinda-IDs found`);\n return {};\n }\n\n return {isMelindaRecord, f001, f035MelindaIds};\n\n }\n\n
|
|
1
|
+
{"version":3,"file":"melinda-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","extract","compare","record","recordExternal","isMelindaRecord","get","some","f003","value","f001","map","field","f035MelindaIds","getMelindaIdsF035","undefined","length","label","a","b","id","JSON","stringify","idA","idB"],"sources":["../../../../src/match-detection/features/bib/melinda-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 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 {getMelindaIdsF035} from '../../../matching-utils';\n\n// 003+001 FI-MELINDA <melinda-id>\n// 035 $a (FI-MELINDA)<melinda-id>\n// 035 $z (FI-MELINDA)<melinda-id>\n// 035 $a FCC<melinda-id>\n// 035 $z FCC<melinda-id>\n// melinda-id = 001234567\n\nexport default () => {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/melinda-id');\n const debugData = debug.extend('data');\n\n return {extract, compare};\n\n function extract({record, recordExternal}) {\n\n const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');\n const [f001] = record.get('001').map(field => field.value);\n const f035MelindaIds = getMelindaIdsF035(record);\n\n if (\n isMelindaRecord === undefined &&\n f001 === undefined &&\n f035MelindaIds.length < 1) {\n\n debug(`${recordExternal.label} No Melinda-IDs found`);\n return {};\n }\n\n return {isMelindaRecord, f001, f035MelindaIds};\n\n }\n\n function compare(a, b) {\n\n if (a.isMelindaRecord && b.isMelindaRecord &&\n a.f001 === b.f001) {\n debugData(`Melinda record's A f001 ${a.f001} matches Melinda record's B f001 ${a.f001}`);\n return 1;\n }\n\n if (a.isMelindaRecord && typeof b.f035MelindaIds !== 'undefined' &&\n b.f035MelindaIds.some(id => id === a.f001)) {\n debugData(`Melinda record's A f001 ${a.f001} matches record B f035 ${JSON.stringify(b.f035MelindaIds)}`);\n return 1;\n }\n\n if (b.isMelindaRecord && typeof a.f035MelindaIds !== 'undefined' &&\n a.f035MelindaIds.some(id => id === b.f001)) {\n debugData(`Melinda record's B f001 ${b.f001} matches record A f035 ${JSON.stringify(a.f035MelindaIds)}`);\n return 1;\n }\n\n if (typeof a.f035MelindaIds !== 'undefined' && typeof b.f035MelindaIds !== 'undefined' &&\n a.f035MelindaIds.some(idA => b.f035MelindaIds.some(idB => idB === idA))) {\n debugData(`Record A f035 ${JSON.stringify(a.f035MelindaIds)} matches record B f035 ${JSON.stringify(b.f035MelindaIds)}`);\n return 1;\n }\n debug(`No matching Melinda-IDs.`);\n return 0;\n }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA0D;AA7B1D;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;AACA;AACA;AACA;AACA;AACA;AAAA,eAEe,MAAM;EAEnB,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,2EAA2E,CAAC;EAC5G,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAAC;IAACE,MAAM;IAAEC;EAAc,CAAC,EAAE;IAEzC,MAAMC,eAAe,GAAGF,MAAM,CAACG,GAAG,CAAC,KAAK,CAAC,CAACC,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACC,KAAK,KAAK,YAAY,CAAC;IACnF,MAAM,CAACC,IAAI,CAAC,GAAGP,MAAM,CAACG,GAAG,CAAC,KAAK,CAAC,CAACK,GAAG,CAACC,KAAK,IAAIA,KAAK,CAACH,KAAK,CAAC;IAC1D,MAAMI,cAAc,GAAG,IAAAC,gCAAiB,EAACX,MAAM,CAAC;IAEhD,IACEE,eAAe,KAAKU,SAAS,IAC7BL,IAAI,KAAKK,SAAS,IAClBF,cAAc,CAACG,MAAM,GAAG,CAAC,EAAE;MAE3BnB,KAAK,CAAE,GAAEO,cAAc,CAACa,KAAM,uBAAsB,CAAC;MACrD,OAAO,CAAC,CAAC;IACX;IAEA,OAAO;MAACZ,eAAe;MAAEK,IAAI;MAAEG;IAAc,CAAC;EAEhD;EAEA,SAASX,OAAO,CAACgB,CAAC,EAAEC,CAAC,EAAE;IAErB,IAAID,CAAC,CAACb,eAAe,IAAIc,CAAC,CAACd,eAAe,IACtCa,CAAC,CAACR,IAAI,KAAKS,CAAC,CAACT,IAAI,EAAE;MACrBX,SAAS,CAAE,2BAA0BmB,CAAC,CAACR,IAAK,oCAAmCQ,CAAC,CAACR,IAAK,EAAC,CAAC;MACxF,OAAO,CAAC;IACV;IAEA,IAAIQ,CAAC,CAACb,eAAe,IAAI,OAAOc,CAAC,CAACN,cAAc,KAAK,WAAW,IAC5DM,CAAC,CAACN,cAAc,CAACN,IAAI,CAACa,EAAE,IAAIA,EAAE,KAAKF,CAAC,CAACR,IAAI,CAAC,EAAE;MAC9CX,SAAS,CAAE,2BAA0BmB,CAAC,CAACR,IAAK,0BAAyBW,IAAI,CAACC,SAAS,CAACH,CAAC,CAACN,cAAc,CAAE,EAAC,CAAC;MACxG,OAAO,CAAC;IACV;IAEA,IAAIM,CAAC,CAACd,eAAe,IAAI,OAAOa,CAAC,CAACL,cAAc,KAAK,WAAW,IAC5DK,CAAC,CAACL,cAAc,CAACN,IAAI,CAACa,EAAE,IAAIA,EAAE,KAAKD,CAAC,CAACT,IAAI,CAAC,EAAE;MAC9CX,SAAS,CAAE,2BAA0BoB,CAAC,CAACT,IAAK,0BAAyBW,IAAI,CAACC,SAAS,CAACJ,CAAC,CAACL,cAAc,CAAE,EAAC,CAAC;MACxG,OAAO,CAAC;IACV;IAEA,IAAI,OAAOK,CAAC,CAACL,cAAc,KAAK,WAAW,IAAI,OAAOM,CAAC,CAACN,cAAc,KAAK,WAAW,IACjFK,CAAC,CAACL,cAAc,CAACN,IAAI,CAACgB,GAAG,IAAIJ,CAAC,CAACN,cAAc,CAACN,IAAI,CAACiB,GAAG,IAAIA,GAAG,KAAKD,GAAG,CAAC,CAAC,EAAE;MAC5ExB,SAAS,CAAE,iBAAgBsB,IAAI,CAACC,SAAS,CAACJ,CAAC,CAACL,cAAc,CAAE,0BAAyBQ,IAAI,CAACC,SAAS,CAACH,CAAC,CAACN,cAAc,CAAE,EAAC,CAAC;MACxH,OAAO,CAAC;IACV;IACAhB,KAAK,CAAE,0BAAyB,CAAC;IACjC,OAAO,CAAC;EACV;AACF,CAAC;AAAA"}
|
|
@@ -32,7 +32,8 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
32
32
|
* @licend The above is the entire license notice
|
|
33
33
|
* for the JavaScript code in this file.
|
|
34
34
|
*
|
|
35
|
-
*/
|
|
35
|
+
*/
|
|
36
|
+
var _default = () => {
|
|
36
37
|
const {
|
|
37
38
|
extract,
|
|
38
39
|
compare
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"other-standard-identifier.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/other-standard-identifier.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^024$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'Other standard identifier'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;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,
|
|
1
|
+
{"version":3,"file":"other-standard-identifier.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/other-standard-identifier.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 createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^024$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'Other standard identifier'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;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,eA8Be,MAAM;EACnB,MAAM;IAACA,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG;EAAC,CAAC,CAAC;EAC1F,OAAO;IAACJ,OAAO;IAAEC,OAAO;IAAEI,IAAI,EAAE;EAA2B,CAAC;AAC9D,CAAC;AAAA"}
|
|
@@ -0,0 +1,79 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
|
|
3
|
+
Object.defineProperty(exports, "__esModule", {
|
|
4
|
+
value: true
|
|
5
|
+
});
|
|
6
|
+
exports.default = void 0;
|
|
7
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
8
|
+
var _natural = require("natural");
|
|
9
|
+
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
10
|
+
/**
|
|
11
|
+
*
|
|
12
|
+
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
13
|
+
*
|
|
14
|
+
* Melinda record matching modules for Javascript
|
|
15
|
+
*
|
|
16
|
+
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
17
|
+
*
|
|
18
|
+
* This file is part of melinda-record-matching-js
|
|
19
|
+
*
|
|
20
|
+
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
21
|
+
* it under the terms of the GNU Lesser General Public License as
|
|
22
|
+
* published by the Free Software Foundation, either version 3 of the
|
|
23
|
+
* License, or (at your option) any later version.
|
|
24
|
+
*
|
|
25
|
+
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
26
|
+
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
27
|
+
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
28
|
+
* GNU Lesser General Public License for more details.
|
|
29
|
+
*
|
|
30
|
+
* You should have received a copy of the GNU Affero General Public License
|
|
31
|
+
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
32
|
+
*
|
|
33
|
+
* @licend The above is the entire license notice
|
|
34
|
+
* for the JavaScript code in this file.
|
|
35
|
+
*
|
|
36
|
+
*/
|
|
37
|
+
var _default = ({
|
|
38
|
+
treshold = 10
|
|
39
|
+
} = {}) => ({
|
|
40
|
+
name: 'titleVersionOriginal',
|
|
41
|
+
extract: ({
|
|
42
|
+
record
|
|
43
|
+
}) => {
|
|
44
|
+
const title = getTitle();
|
|
45
|
+
if (title) {
|
|
46
|
+
return [title.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
|
|
47
|
+
}
|
|
48
|
+
return [];
|
|
49
|
+
function getTitle() {
|
|
50
|
+
const [field] = record.get(/^245$/u);
|
|
51
|
+
if (field) {
|
|
52
|
+
return field.subfields.filter(({
|
|
53
|
+
code
|
|
54
|
+
}) => ['a', 'b'].includes(code)).map(({
|
|
55
|
+
value
|
|
56
|
+
}) => value).join('');
|
|
57
|
+
}
|
|
58
|
+
}
|
|
59
|
+
},
|
|
60
|
+
compare: (a, b) => {
|
|
61
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib/title-version-original');
|
|
62
|
+
const distance = (0, _natural.LevenshteinDistance)(a[0], b[0]);
|
|
63
|
+
if (distance === 0) {
|
|
64
|
+
return 0.5;
|
|
65
|
+
}
|
|
66
|
+
const maxLength = getMaxLength();
|
|
67
|
+
const percentage = distance / maxLength * 100;
|
|
68
|
+
debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);
|
|
69
|
+
if (percentage <= treshold) {
|
|
70
|
+
return 0.3;
|
|
71
|
+
}
|
|
72
|
+
return -0.5;
|
|
73
|
+
function getMaxLength() {
|
|
74
|
+
return a[0].length > b[0].length ? a[0].length : b[0].length;
|
|
75
|
+
}
|
|
76
|
+
}
|
|
77
|
+
});
|
|
78
|
+
exports.default = _default;
|
|
79
|
+
//# sourceMappingURL=title-version-original.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"title-version-original.js","names":["treshold","name","extract","record","title","getTitle","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-version-original.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {LevenshteinDistance as leven} from 'natural';\n\nexport default ({treshold = 10} = {}) => ({\n name: 'titleVersionOriginal',\n extract: ({record}) => {\n const title = getTitle();\n\n if (title) {\n return [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}) => value)\n .join('');\n }\n }\n },\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title-version-original');\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;AAAqD;AA7BrD;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,eA+Be,CAAC;EAACA,QAAQ,GAAG;AAAE,CAAC,GAAG,CAAC,CAAC,MAAM;EACxCC,IAAI,EAAE,sBAAsB;EAC5BC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IACrB,MAAMC,KAAK,GAAGC,QAAQ,EAAE;IAExB,IAAID,KAAK,EAAE;MACT,OAAO,CAACA,KAAK,CAACE,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CAACC,WAAW,EAAE,CAAC;IACvE;IAEA,OAAO,EAAE;IAET,SAASF,QAAQ,GAAG;MAClB,MAAM,CAACG,KAAK,CAAC,GAAGL,MAAM,CAACM,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,KAAKA,KAAK,CAAC,CACvBC,IAAI,CAAC,EAAE,CAAC;MACb;IACF;EACF,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,uFAAuF,CAAC;IACxH,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,IAAI1B,QAAQ,EAAE;MAC1B,OAAO,GAAG;IACZ;IAEA,OAAO,CAAC,GAAG;IAEX,SAASyB,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"}
|
|
@@ -4,9 +4,9 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
6
|
exports.default = void 0;
|
|
7
|
-
var _debug = _interopRequireDefault(require("debug"));
|
|
8
7
|
var _natural = require("natural");
|
|
9
8
|
var _matchingUtils = require("../../../matching-utils");
|
|
9
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
10
10
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
11
11
|
/**
|
|
12
12
|
*
|
|
@@ -34,24 +34,42 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
34
34
|
* @licend The above is the entire license notice
|
|
35
35
|
* for the JavaScript code in this file.
|
|
36
36
|
*
|
|
37
|
-
*/
|
|
37
|
+
*/
|
|
38
|
+
|
|
39
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features:title');
|
|
40
|
+
const debugData = debug.extend('data');
|
|
41
|
+
var _default = ({
|
|
38
42
|
treshold = 10
|
|
39
43
|
} = {}) => ({
|
|
40
44
|
name: 'Title',
|
|
41
45
|
extract: ({
|
|
42
|
-
record
|
|
46
|
+
record,
|
|
47
|
+
recordExternal
|
|
43
48
|
}) => {
|
|
49
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
44
50
|
const title = getTitle();
|
|
51
|
+
debug(`${label} title: ${title}`);
|
|
45
52
|
if ((0, _matchingUtils.testStringOrNumber)(title)) {
|
|
46
|
-
|
|
53
|
+
const titleAsNormalizedString = String(title)
|
|
54
|
+
// decompose unicode diacritics
|
|
55
|
+
.normalize('NFD')
|
|
56
|
+
// strip non-letters/numbers
|
|
57
|
+
// - note: combined with decomposing unicode diactics this normalizes both 'saa' and 'sää' as 'saa'
|
|
58
|
+
// - we could precompose the finnish letters back to avoid this
|
|
59
|
+
.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase();
|
|
60
|
+
debug(`${label} titleString: ${titleAsNormalizedString}`);
|
|
61
|
+
return [titleAsNormalizedString];
|
|
47
62
|
}
|
|
48
63
|
return [];
|
|
49
64
|
function getTitle() {
|
|
50
65
|
const [field] = record.get(/^245$/u);
|
|
66
|
+
debugData(`${label} titleField: ${JSON.stringify(field)}`);
|
|
51
67
|
if (field) {
|
|
52
|
-
return field.subfields
|
|
68
|
+
return field.subfields
|
|
69
|
+
// get also $n:s and $p:s here
|
|
70
|
+
.filter(({
|
|
53
71
|
code
|
|
54
|
-
}) => ['a', 'b'].includes(code)).map(({
|
|
72
|
+
}) => ['a', 'b', 'n', 'p'].includes(code)).map(({
|
|
55
73
|
value
|
|
56
74
|
}) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').join('');
|
|
57
75
|
}
|
|
@@ -1 +1 @@
|
|
|
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","
|
|
1
|
+
{"version":3,"file":"title.js","names":["debug","createDebugLogger","debugData","extend","treshold","name","extract","record","recordExternal","label","title","getTitle","testStringOrNumber","titleAsNormalizedString","String","normalize","replace","toLowerCase","field","get","JSON","stringify","subfields","filter","code","includes","map","value","join","compare","a","b","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\n\nimport {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:title');\nconst debugData = debug.extend('data');\n\n\nexport default ({treshold = 10} = {}) => ({\n name: 'Title',\n extract: ({record, recordExternal}) => {\n const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n const title = getTitle();\n debug(`${label} title: ${title}`);\n\n if (testStringOrNumber(title)) {\n const titleAsNormalizedString = String(title)\n // decompose unicode diacritics\n .normalize('NFD')\n // strip non-letters/numbers\n // - note: combined with decomposing unicode diactics this normalizes both 'saa' and 'sää' as 'saa'\n // - we could precompose the finnish letters back to avoid this\n .replace(/[^\\p{Letter}\\p{Number}]/gu, '')\n .toLowerCase();\n debug(`${label} titleString: ${titleAsNormalizedString}`);\n return [titleAsNormalizedString];\n }\n\n return [];\n\n function getTitle() {\n const [field] = record.get(/^245$/u);\n debugData(`${label} titleField: ${JSON.stringify(field)}`);\n\n if (field) {\n return field.subfields\n // get also $n:s and $p:s here\n .filter(({code}) => ['a', 'b', 'n', 'p'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .join('');\n }\n return false;\n }\n },\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n const distance = leven(a[0], b[0]);\n\n if (distance === 0) {\n return 0.5;\n }\n\n const maxLength = getMaxLength();\n const percentage = distance / maxLength * 100;\n\n debug(`'${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":";;;;;;AA6BA;AACA;AACA;AAAsC;AA/BtC;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;;AAOA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,kEAAkE,CAAC;AACnG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAGxB,CAAC;EAACC,QAAQ,GAAG;AAAE,CAAC,GAAG,CAAC,CAAC,MAAM;EACxCC,IAAI,EAAE,OAAO;EACbC,OAAO,EAAE,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,KAAK;IACrC,MAAMC,KAAK,GAAGD,cAAc,IAAIA,cAAc,CAACC,KAAK,GAAGD,cAAc,CAACC,KAAK,GAAG,QAAQ;IACtF,MAAMC,KAAK,GAAGC,QAAQ,EAAE;IACxBX,KAAK,CAAE,GAAES,KAAM,WAAUC,KAAM,EAAC,CAAC;IAEjC,IAAI,IAAAE,iCAAkB,EAACF,KAAK,CAAC,EAAE;MAC7B,MAAMG,uBAAuB,GAAGC,MAAM,CAACJ,KAAK;MAC1C;MAAA,CACCK,SAAS,CAAC,KAAK;MAChB;MACA;MACA;MAAA,CACCC,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CACxCC,WAAW,EAAE;MAChBjB,KAAK,CAAE,GAAES,KAAM,iBAAgBI,uBAAwB,EAAC,CAAC;MACzD,OAAO,CAACA,uBAAuB,CAAC;IAClC;IAEA,OAAO,EAAE;IAET,SAASF,QAAQ,GAAG;MAClB,MAAM,CAACO,KAAK,CAAC,GAAGX,MAAM,CAACY,GAAG,CAAC,QAAQ,CAAC;MACpCjB,SAAS,CAAE,GAAEO,KAAM,gBAAeW,IAAI,CAACC,SAAS,CAACH,KAAK,CAAE,EAAC,CAAC;MAE1D,IAAIA,KAAK,EAAE;QACT,OAAOA,KAAK,CAACI;QACX;QAAA,CACCC,MAAM,CAAC,CAAC;UAACC;QAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,IAAI,CAAC,CAAC,CACvDE,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAK,IAAAf,iCAAkB,EAACe,KAAK,CAAC,GAAGb,MAAM,CAACa,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,MAAM/B,KAAK,GAAG,IAAAC,cAAiB,EAAC,sEAAsE,CAAC;IACvG,MAAM+B,QAAQ,GAAG,IAAAC,4BAAK,EAACH,CAAC,CAAC,CAAC,CAAC,EAAEC,CAAC,CAAC,CAAC,CAAC,CAAC;IAElC,IAAIC,QAAQ,KAAK,CAAC,EAAE;MAClB,OAAO,GAAG;IACZ;IAEA,MAAME,SAAS,GAAGC,YAAY,EAAE;IAChC,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAS,GAAG,GAAG;IAE7ClC,KAAK,CAAE,IAAG8B,CAAE,SAAQC,CAAE,mBAAkBG,SAAU,gBAAeF,QAAS,kBAAiBI,UAAW,EAAC,CAAC;IAExG,IAAIA,UAAU,IAAIhC,QAAQ,EAAE;MAC1B,OAAO,GAAG;IACZ;IAEA,OAAO,CAAC,GAAG;IAEX,SAAS+B,YAAY,GAAG;MACtB,OAAOL,CAAC,CAAC,CAAC,CAAC,CAACO,MAAM,GAAGN,CAAC,CAAC,CAAC,CAAC,CAACM,MAAM,GAAGP,CAAC,CAAC,CAAC,CAAC,CAACO,MAAM,GAAGN,CAAC,CAAC,CAAC,CAAC,CAACM,MAAM;IAC9D;EAEF;AACF,CAAC,CAAC;AAAA"}
|
|
@@ -10,7 +10,6 @@ exports.features = features;
|
|
|
10
10
|
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); }
|
|
11
11
|
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; }
|
|
12
12
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
13
|
-
/* eslint-disable no-console */
|
|
14
13
|
/**
|
|
15
14
|
*
|
|
16
15
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -41,7 +40,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
41
40
|
var _default = ({
|
|
42
41
|
strategy,
|
|
43
42
|
treshold = 0.9
|
|
44
|
-
}, returnStrategy = false) => ({
|
|
43
|
+
}, returnStrategy = false, localSettings = {}) => ({
|
|
45
44
|
recordA,
|
|
46
45
|
recordB,
|
|
47
46
|
recordAExternal = {},
|
|
@@ -53,9 +52,12 @@ var _default = ({
|
|
|
53
52
|
debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
|
|
54
53
|
debugData(`Records: A: ${recordA}\nB: ${recordB}`);
|
|
55
54
|
debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);
|
|
55
|
+
// We could add here labels for records if we didn't get external labels
|
|
56
|
+
|
|
56
57
|
const featuresA = extractFeatures({
|
|
57
58
|
record: recordA,
|
|
58
|
-
recordExternal: recordAExternal
|
|
59
|
+
recordExternal: recordAExternal,
|
|
60
|
+
localSettings
|
|
59
61
|
});
|
|
60
62
|
debug(`We got an array of records: ${Array.isArray(recordB)}`);
|
|
61
63
|
const recordsB = Array.isArray(recordB) ? recordB : [recordB];
|
|
@@ -75,21 +77,23 @@ var _default = ({
|
|
|
75
77
|
featuresA,
|
|
76
78
|
record,
|
|
77
79
|
recordExternal,
|
|
78
|
-
index
|
|
80
|
+
index,
|
|
81
|
+
localSettings
|
|
79
82
|
}) {
|
|
80
83
|
const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';
|
|
81
84
|
const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';
|
|
82
85
|
debug(`Actual detection for record ${index + 1} ${labelB}`);
|
|
83
86
|
const featuresB = extractFeatures({
|
|
84
87
|
record,
|
|
85
|
-
recordExternal
|
|
88
|
+
recordExternal,
|
|
89
|
+
localSettings
|
|
86
90
|
});
|
|
87
91
|
debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);
|
|
88
92
|
debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);
|
|
89
93
|
const featurePairs = generateFeaturePairs(featuresA, featuresB);
|
|
90
94
|
const similarityVector = generateSimilarityVector(featurePairs);
|
|
91
95
|
if (similarityVector.some(v => v >= minProbabilityQuantifier)) {
|
|
92
|
-
const probability =
|
|
96
|
+
const probability = calculateProbability(similarityVector);
|
|
93
97
|
debug(`probability: ${probability} (Treshold: ${treshold})`);
|
|
94
98
|
return returnResult({
|
|
95
99
|
match: probability >= treshold,
|
|
@@ -135,7 +139,7 @@ var _default = ({
|
|
|
135
139
|
const strategyNames = strategy.map(element => element.name);
|
|
136
140
|
return strategyNames || [];
|
|
137
141
|
}
|
|
138
|
-
function
|
|
142
|
+
function calculateProbability(similarityVector) {
|
|
139
143
|
const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);
|
|
140
144
|
return probability > 1.0 ? 1.0 : probability;
|
|
141
145
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","names":["strategy","treshold","returnStrategy","recordA","recordB","recordAExternal","recordBExternal","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","record","recordExternal","Array","isArray","recordsB","recordsBExternal","detectionResults","map","index","actualDetection","labelA","label","labelB","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateprobability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","a","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","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, recordAExternal = {}, recordBExternal = {}}) => {\n const minProbabilityQuantifier = 0.5;\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n const debugData = debug.extend('data');\n\n debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n\n const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal});\n\n debug(`We got an array of records: ${Array.isArray(recordB)}`);\n const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n // if we got array of records, we return an array of result\n // if we got a singular record, we return a singular result\n return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n function actualDetection({featuresA, record, recordExternal, index}) {\n const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n debug(`Actual detection for record ${index + 1} ${labelB}`);\n const featuresB = extractFeatures({record, recordExternal});\n\n debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n const featurePairs = generateFeaturePairs(featuresA, featuresB);\n const similarityVector = generateSimilarityVector(featurePairs);\n\n if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n const probability = calculateprobability(similarityVector);\n debug(`probability: ${probability} (Treshold: ${treshold})`);\n return returnResult({match: probability >= treshold, probability});\n }\n\n debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n return returnResult({match: false, probability: 0.0});\n }\n\n function extractFeatures({record, recordExternal}) {\n return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n }\n\n function returnResult(result) {\n if (returnStrategy) {\n debug(`Returning detection strategy with the result`);\n const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n debugData(`${JSON.stringify(resultWithStrategy)}`);\n return resultWithStrategy;\n }\n return result;\n }\n\n function formatStrategy(strategy) {\n const strategyNames = strategy.map(element => element.name);\n return strategyNames || [];\n }\n\n function calculateprobability(similarityVector) {\n const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n return probability > 1.0 ? 1.0 : probability;\n }\n\n function generateSimilarityVector(featurePairs) {\n const compared = featurePairs.map(({name, a, b}) => {\n const {compare} = strategy.find(({name: featureName}) => name === featureName);\n const points = compare(a, b);\n return {name, points};\n });\n\n debugData(`Points: ${JSON.stringify(compared)}`);\n return compared.map(({points}) => points);\n }\n\n function generateFeaturePairs(featuresA, featuresB) {\n const pairs = generatePairs();\n const missingFeatures = findMissing();\n\n debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n return pairs;\n\n function generatePairs() {\n return featuresA\n .reduce((acc, {name, value}, index) => acc.concat({\n name,\n a: value,\n b: featuresB[index].value\n }), [])\n .filter(({a, b}) => {\n if (a.length === 0 || b.length === 0) {\n return false;\n }\n\n return true;\n });\n }\n\n function findMissing() {\n return featuresA\n .map(({name}) => name)\n .filter(v => pairs.some(({name}) => name === v) === false);\n }\n }\n\n};\n"],"mappings":";;;;;;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,CAAC;EAACC,OAAO;EAAEC,OAAO;EAAEC,eAAe,GAAG,CAAC,CAAC;EAAEC,eAAe,GAAG,CAAC;AAAC,CAAC,KAAK;EACzI,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,CAACb,QAAQ,CAAE,eAAcY,IAAI,CAACC,SAAS,CAACZ,QAAQ,CAAE,qBAAoBW,IAAI,CAACC,SAAS,CAACX,cAAc,CAAE,EAAC,CAAC;EAC5IQ,SAAS,CAAE,eAAcP,OAAQ,QAAOC,OAAQ,EAAC,CAAC;EAClDI,KAAK,CAAE,iBAAgBI,IAAI,CAACC,SAAS,CAACR,eAAe,CAAE,QAAOO,IAAI,CAACC,SAAS,CAACP,eAAe,CAAE,EAAC,CAAC;EAEhG,MAAMQ,SAAS,GAAGC,eAAe,CAAC;IAACC,MAAM,EAAEb,OAAO;IAAEc,cAAc,EAAEZ;EAAe,CAAC,CAAC;EAErFG,KAAK,CAAE,+BAA8BU,KAAK,CAACC,OAAO,CAACf,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMgB,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMiB,gBAAgB,GAAGH,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGE,eAAe,GAAG,CAACA,eAAe,CAAC;EAErF,MAAMgB,gBAAgB,GAAGF,QAAQ,CAACG,GAAG,CAAC,CAACP,MAAM,EAAEQ,KAAK,KAAKC,eAAe,CAAC;IAACX,SAAS;IAAET,eAAe;IAAEW,MAAM;IAAEC,cAAc,EAAEI,gBAAgB,CAACG,KAAK,CAAC;IAAEA;EAAK,CAAC,CAAC,CAAC;;EAE/J;EACA;EACA,OAAON,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGkB,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAe,CAAC;IAACX,SAAS;IAAEE,MAAM;IAAEC,cAAc;IAAEO;EAAK,CAAC,EAAE;IACnE,MAAME,MAAM,GAAGrB,eAAe,IAAIA,eAAe,CAACsB,KAAK,GAAGtB,eAAe,CAACsB,KAAK,GAAG,GAAG;IACrF,MAAMC,MAAM,GAAGX,cAAc,IAAIA,cAAc,CAACU,KAAK,GAAGV,cAAc,CAACU,KAAK,GAAG,GAAG;IAGlFnB,KAAK,CAAE,+BAA8BgB,KAAK,GAAG,CAAE,IAAGI,MAAO,EAAC,CAAC;IAC3D,MAAMC,SAAS,GAAGd,eAAe,CAAC;MAACC,MAAM;MAAEC;IAAc,CAAC,CAAC;IAE3DP,SAAS,CAAE,gBAAegB,MAAO,MAAKd,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IAClEJ,SAAS,CAAE,gBAAekB,MAAO,MAAKhB,IAAI,CAACC,SAAS,CAACgB,SAAS,CAAE,EAAC,CAAC;IAElE,MAAMC,YAAY,GAAGC,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAI5B,wBAAwB,CAAC,EAAE;MAC7D,MAAM6B,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DxB,KAAK,CAAE,gBAAe4B,WAAY,eAAcnC,QAAS,GAAE,CAAC;MAC5D,OAAOqC,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAInC,QAAQ;QAAEmC;MAAW,CAAC,CAAC;IACpE;IAEA1B,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAO+B,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASrB,eAAe,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,EAAE;IACjD,OAAOjB,QAAQ,CAACwC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAC;QAAC3B,MAAM;QAAEC;MAAc,CAAC;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EACpH;EAEA,SAASqB,YAAY,CAACQ,MAAM,EAAE;IAC5B,IAAI5C,cAAc,EAAE;MAClBM,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMuC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAEpC,QAAQ,EAAEgD,cAAc,CAAChD,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HS,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAACkC,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAc,CAAChD,QAAQ,EAAE;IAChC,MAAMiD,aAAa,GAAGjD,QAAQ,CAACuB,GAAG,CAAC2B,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,CAACP,GAAG,CAAC,CAAC;MAACmB,IAAI;MAAEU,CAAC;MAAEC;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAGtD,QAAQ,CAACuD,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;IAEF/C,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACsC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAAC5B,GAAG,CAAC,CAAC;MAACkC;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAAS1B,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,EAAE;IAClD,MAAM6B,KAAK,GAAGC,aAAa,EAAE;IAC7B,MAAMC,eAAe,GAAGC,WAAW,EAAE;IAErCrD,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAAC+C,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAa,GAAG;MACvB,OAAO7C,SAAS,CACb0B,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAErB,KAAK,KAAKiB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJU,CAAC,EAAEP,KAAK;QACRQ,CAAC,EAAExB,SAAS,CAACL,KAAK,CAAC,CAACqB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNiB,MAAM,CAAC,CAAC;QAACV,CAAC;QAAEC;MAAC,CAAC,KAAK;QAClB,IAAID,CAAC,CAACW,MAAM,KAAK,CAAC,IAAIV,CAAC,CAACU,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASF,WAAW,GAAG;MACrB,OAAO/C,SAAS,CACbS,GAAG,CAAC,CAAC;QAACmB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBoB,MAAM,CAAC3B,CAAC,IAAIuB,KAAK,CAACxB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA"}
|
|
1
|
+
{"version":3,"file":"index.js","names":["strategy","treshold","returnStrategy","localSettings","recordA","recordB","recordAExternal","recordBExternal","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","record","recordExternal","Array","isArray","recordsB","recordsBExternal","detectionResults","map","index","actualDetection","labelA","label","labelB","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateProbability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","a","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport * as features from './features';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false, localSettings = {}) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {\n const minProbabilityQuantifier = 0.5;\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n const debugData = debug.extend('data');\n\n debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n // We could add here labels for records if we didn't get external labels\n\n const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal, localSettings});\n\n debug(`We got an array of records: ${Array.isArray(recordB)}`);\n const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n // if we got array of records, we return an array of result\n // if we got a singular record, we return a singular result\n return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n function actualDetection({featuresA, record, recordExternal, index, localSettings}) {\n const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n debug(`Actual detection for record ${index + 1} ${labelB}`);\n const featuresB = extractFeatures({record, recordExternal, localSettings});\n\n debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n const featurePairs = generateFeaturePairs(featuresA, featuresB);\n const similarityVector = generateSimilarityVector(featurePairs);\n\n if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n const probability = calculateProbability(similarityVector);\n debug(`probability: ${probability} (Treshold: ${treshold})`);\n return returnResult({match: probability >= treshold, probability});\n }\n\n debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n return returnResult({match: false, probability: 0.0});\n }\n\n function extractFeatures({record, recordExternal}) {\n return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n }\n\n function returnResult(result) {\n if (returnStrategy) {\n debug(`Returning detection strategy with the result`);\n const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n debugData(`${JSON.stringify(resultWithStrategy)}`);\n return resultWithStrategy;\n }\n return result;\n }\n\n function formatStrategy(strategy) {\n const strategyNames = strategy.map(element => element.name);\n return strategyNames || [];\n }\n\n function calculateProbability(similarityVector) {\n const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n return probability > 1.0 ? 1.0 : probability;\n }\n\n function generateSimilarityVector(featurePairs) {\n const compared = featurePairs.map(({name, a, b}) => {\n const {compare} = strategy.find(({name: featureName}) => name === featureName);\n const points = compare(a, b);\n return {name, points};\n });\n\n debugData(`Points: ${JSON.stringify(compared)}`);\n return compared.map(({points}) => points);\n }\n\n function generateFeaturePairs(featuresA, featuresB) {\n const pairs = generatePairs();\n const missingFeatures = findMissing();\n\n debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n return pairs;\n\n function generatePairs() {\n return featuresA\n .reduce((acc, {name, value}, index) => acc.concat({\n name,\n a: value,\n b: featuresB[index].value\n }), [])\n .filter(({a, b}) => {\n if (a.length === 0 || b.length === 0) {\n return false;\n }\n\n return true;\n });\n }\n\n function findMissing() {\n return featuresA\n .map(({name}) => name)\n .filter(v => pairs.some(({name}) => name === v) === false);\n }\n }\n\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAAuC;AAAA;AAAA;AAAA;AA7BvC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,eAiCe,CAAC;EAACA,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,EAAEC,aAAa,GAAG,CAAC,CAAC,KAAK,CAAC;EAACC,OAAO;EAAEC,OAAO;EAAEC,eAAe,GAAG,CAAC,CAAC;EAAEC,eAAe,GAAG,CAAC;AAAC,CAAC,KAAK;EAC7J,MAAMC,wBAAwB,GAAG,GAAG;EAEpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,mDAAmD,CAAC;EACpF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtCD,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACd,QAAQ,CAAE,eAAca,IAAI,CAACC,SAAS,CAACb,QAAQ,CAAE,qBAAoBY,IAAI,CAACC,SAAS,CAACZ,cAAc,CAAE,EAAC,CAAC;EAC5IS,SAAS,CAAE,eAAcP,OAAQ,QAAOC,OAAQ,EAAC,CAAC;EAClDI,KAAK,CAAE,iBAAgBI,IAAI,CAACC,SAAS,CAACR,eAAe,CAAE,QAAOO,IAAI,CAACC,SAAS,CAACP,eAAe,CAAE,EAAC,CAAC;EAChG;;EAEA,MAAMQ,SAAS,GAAGC,eAAe,CAAC;IAACC,MAAM,EAAEb,OAAO;IAAEc,cAAc,EAAEZ,eAAe;IAAEH;EAAa,CAAC,CAAC;EAEpGM,KAAK,CAAE,+BAA8BU,KAAK,CAACC,OAAO,CAACf,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMgB,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMiB,gBAAgB,GAAGH,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGE,eAAe,GAAG,CAACA,eAAe,CAAC;EAErF,MAAMgB,gBAAgB,GAAGF,QAAQ,CAACG,GAAG,CAAC,CAACP,MAAM,EAAEQ,KAAK,KAAKC,eAAe,CAAC;IAACX,SAAS;IAAET,eAAe;IAAEW,MAAM;IAAEC,cAAc,EAAEI,gBAAgB,CAACG,KAAK,CAAC;IAAEA;EAAK,CAAC,CAAC,CAAC;;EAE/J;EACA;EACA,OAAON,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGkB,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAe,CAAC;IAACX,SAAS;IAAEE,MAAM;IAAEC,cAAc;IAAEO,KAAK;IAAEtB;EAAa,CAAC,EAAE;IAClF,MAAMwB,MAAM,GAAGrB,eAAe,IAAIA,eAAe,CAACsB,KAAK,GAAGtB,eAAe,CAACsB,KAAK,GAAG,GAAG;IACrF,MAAMC,MAAM,GAAGX,cAAc,IAAIA,cAAc,CAACU,KAAK,GAAGV,cAAc,CAACU,KAAK,GAAG,GAAG;IAGlFnB,KAAK,CAAE,+BAA8BgB,KAAK,GAAG,CAAE,IAAGI,MAAO,EAAC,CAAC;IAC3D,MAAMC,SAAS,GAAGd,eAAe,CAAC;MAACC,MAAM;MAAEC,cAAc;MAAEf;IAAa,CAAC,CAAC;IAE1EQ,SAAS,CAAE,gBAAegB,MAAO,MAAKd,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IAClEJ,SAAS,CAAE,gBAAekB,MAAO,MAAKhB,IAAI,CAACC,SAAS,CAACgB,SAAS,CAAE,EAAC,CAAC;IAElE,MAAMC,YAAY,GAAGC,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAI5B,wBAAwB,CAAC,EAAE;MAC7D,MAAM6B,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DxB,KAAK,CAAE,gBAAe4B,WAAY,eAAcpC,QAAS,GAAE,CAAC;MAC5D,OAAOsC,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAIpC,QAAQ;QAAEoC;MAAW,CAAC,CAAC;IACpE;IAEA1B,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAO+B,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASrB,eAAe,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,EAAE;IACjD,OAAOlB,QAAQ,CAACyC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAC;QAAC3B,MAAM;QAAEC;MAAc,CAAC;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EACpH;EAEA,SAASqB,YAAY,CAACQ,MAAM,EAAE;IAC5B,IAAI7C,cAAc,EAAE;MAClBO,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMuC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAErC,QAAQ,EAAEiD,cAAc,CAACjD,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HU,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAACkC,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAc,CAACjD,QAAQ,EAAE;IAChC,MAAMkD,aAAa,GAAGlD,QAAQ,CAACwB,GAAG,CAAC2B,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,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,CAACP,GAAG,CAAC,CAAC;MAACmB,IAAI;MAAEU,CAAC;MAAEC;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAGvD,QAAQ,CAACwD,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;IAEF/C,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACsC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAAC5B,GAAG,CAAC,CAAC;MAACkC;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAAS1B,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,EAAE;IAClD,MAAM6B,KAAK,GAAGC,aAAa,EAAE;IAC7B,MAAMC,eAAe,GAAGC,WAAW,EAAE;IAErCrD,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAAC+C,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAa,GAAG;MACvB,OAAO7C,SAAS,CACb0B,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAErB,KAAK,KAAKiB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJU,CAAC,EAAEP,KAAK;QACRQ,CAAC,EAAExB,SAAS,CAACL,KAAK,CAAC,CAACqB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNiB,MAAM,CAAC,CAAC;QAACV,CAAC;QAAEC;MAAC,CAAC,KAAK;QAClB,IAAID,CAAC,CAACW,MAAM,KAAK,CAAC,IAAIV,CAAC,CAACU,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASF,WAAW,GAAG;MACrB,OAAO/C,SAAS,CACbS,GAAG,CAAC,CAAC;QAACmB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBoB,MAAM,CAAC3B,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/matching-utils.js
CHANGED
|
@@ -39,8 +39,6 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
39
39
|
|
|
40
40
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:utils');
|
|
41
41
|
const debugData = debug.extend('data');
|
|
42
|
-
|
|
43
|
-
// eslint-disable-next-line max-statements
|
|
44
42
|
function getMelindaIdsF035(record) {
|
|
45
43
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:melinda-id');
|
|
46
44
|
const debugData = debug.extend('data');
|