@natlibfi/melinda-record-matching 2.0.1-alpha.1 → 2.0.7-alpha.1
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/.github/CODEOWNERS +2 -0
- package/.github/dependabot.yml +40 -0
- package/.github/workflows/melinda-node-tests.yml +70 -0
- package/README.md +1 -1
- package/dist/candidate-search/candidate-search-utils.js +3 -6
- package/dist/candidate-search/candidate-search-utils.js.map +1 -1
- package/dist/candidate-search/index.js +115 -50
- package/dist/candidate-search/index.js.map +1 -1
- package/dist/candidate-search/index.spec.js +41 -48
- package/dist/candidate-search/index.spec.js.map +1 -1
- package/dist/candidate-search/query-list/bib.js +154 -64
- package/dist/candidate-search/query-list/bib.js.map +1 -1
- package/dist/candidate-search/query-list/bib.spec.js +1 -10
- package/dist/candidate-search/query-list/bib.spec.js.map +1 -1
- package/dist/candidate-search/query-list/index.js +4 -10
- package/dist/candidate-search/query-list/index.js.map +1 -1
- package/dist/index.js +498 -50
- package/dist/index.js.map +1 -1
- package/dist/index.spec.js +60 -34
- package/dist/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/all-source-ids.js +16 -19
- package/dist/match-detection/features/bib/all-source-ids.js.map +1 -1
- package/dist/match-detection/features/bib/authors.js +31 -35
- package/dist/match-detection/features/bib/authors.js.map +1 -1
- package/dist/match-detection/features/bib/bibliographic-level.js +3 -3
- package/dist/match-detection/features/bib/bibliographic-level.js.map +1 -1
- package/dist/match-detection/features/bib/host-component.js +3 -3
- package/dist/match-detection/features/bib/host-component.js.map +1 -1
- package/dist/match-detection/features/bib/index.js +28 -13
- package/dist/match-detection/features/bib/index.js.map +1 -1
- package/dist/match-detection/features/bib/index.spec.js +9 -13
- package/dist/match-detection/features/bib/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/isbn.js +28 -6
- package/dist/match-detection/features/bib/isbn.js.map +1 -1
- package/dist/match-detection/features/bib/issn.js +0 -4
- package/dist/match-detection/features/bib/issn.js.map +1 -1
- package/dist/match-detection/features/bib/language.js +69 -41
- package/dist/match-detection/features/bib/language.js.map +1 -1
- package/dist/match-detection/features/bib/media-type.js +83 -0
- package/dist/match-detection/features/bib/media-type.js.map +1 -0
- package/dist/match-detection/features/bib/melinda-id.js +0 -4
- package/dist/match-detection/features/bib/melinda-id.js.map +1 -1
- package/dist/match-detection/features/bib/melinda-identifier-factory.js +6 -17
- package/dist/match-detection/features/bib/melinda-identifier-factory.js.map +1 -1
- package/dist/match-detection/features/bib/other-standard-identifier.js +1 -4
- package/dist/match-detection/features/bib/other-standard-identifier.js.map +1 -1
- package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js +256 -0
- package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js.map +1 -0
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js +71 -0
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js.map +1 -0
- package/dist/match-detection/features/bib/publication-time.js +9 -8
- package/dist/match-detection/features/bib/publication-time.js.map +1 -1
- package/dist/match-detection/features/bib/record-type.js +6 -3
- package/dist/match-detection/features/bib/record-type.js.map +1 -1
- package/dist/match-detection/features/bib/standard-identifier-factory.js +129 -39
- package/dist/match-detection/features/bib/standard-identifier-factory.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 +29 -22
- package/dist/match-detection/features/bib/title.js.map +1 -1
- package/dist/match-detection/features/index.js +0 -4
- package/dist/match-detection/features/index.js.map +1 -1
- package/dist/match-detection/index.js +94 -47
- package/dist/match-detection/index.js.map +1 -1
- package/dist/match-detection/index.spec.js +25 -15
- package/dist/match-detection/index.spec.js.map +1 -1
- package/dist/matching-utils.js +63 -15
- package/dist/matching-utils.js.map +1 -1
- package/package.json +27 -24
- package/src/candidate-search/index.js +93 -31
- package/src/candidate-search/index.spec.js +27 -24
- package/src/candidate-search/query-list/bib.js +149 -22
- package/src/candidate-search/query-list/index.js +1 -1
- package/src/index.js +310 -38
- package/src/index.spec.js +49 -21
- package/src/match-detection/features/bib/all-source-ids.js +3 -3
- package/src/match-detection/features/bib/authors.js +23 -16
- package/src/match-detection/features/bib/bibliographic-level.js +1 -1
- package/src/match-detection/features/bib/host-component.js +1 -1
- package/src/match-detection/features/bib/index.js +4 -0
- package/src/match-detection/features/bib/index.spec.js +10 -2
- package/src/match-detection/features/bib/isbn.js +21 -2
- package/src/match-detection/features/bib/language.js +63 -22
- package/src/match-detection/features/bib/media-type.js +77 -0
- package/src/match-detection/features/bib/melinda-identifier-factory.js +2 -3
- package/src/match-detection/features/bib/other-standard-identifier.js +2 -0
- package/src/match-detection/features/bib/publication-time-allow-cons-years-multi.js +268 -0
- package/src/match-detection/features/bib/publication-time-allow-cons-years.js +67 -0
- package/src/match-detection/features/bib/publication-time.js +9 -4
- package/src/match-detection/features/bib/record-type.js +5 -1
- package/src/match-detection/features/bib/standard-identifier-factory.js +107 -22
- package/src/match-detection/features/bib/title-version-original.js +78 -0
- package/src/match-detection/features/bib/title.js +27 -7
- package/src/match-detection/index.js +60 -22
- package/src/match-detection/index.spec.js +18 -4
- package/src/matching-utils.js +62 -5
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"bibliographic-level.js","names":["_default","name","extract","record","leader","compare","a","b","exports","default"],"sources":["../../../../src/match-detection/features/bib/bibliographic-level.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 () => ({\n name: 'Bibliographic level',\n extract: ({record}) => record.leader[7] ? [record.leader[7]] : [],\n compare: (a, b) => a[0] === b[0] ? 0.1 : -0.2\n});\n"],"mappings":";;;;;;AAAA;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,IAAAA,QAAA,GA4BeA,CAAA,MAAO;EACpBC,IAAI,EAAE,qBAAqB;EAC3BC,OAAO,EAAEA,CAAC;IAACC;EAAM,CAAC,KAAKA,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,GAAG,CAACD,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE;EACjEC,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAAC,OAAA,CAAAC,OAAA,GAAAT,QAAA"}
|
|
@@ -4,7 +4,6 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
6
|
exports.default = void 0;
|
|
7
|
-
|
|
8
7
|
/**
|
|
9
8
|
*
|
|
10
9
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -34,9 +33,10 @@ exports.default = void 0;
|
|
|
34
33
|
*/
|
|
35
34
|
var _default = () => ({
|
|
36
35
|
name: 'Host/Component record',
|
|
37
|
-
extract:
|
|
36
|
+
extract: ({
|
|
37
|
+
record
|
|
38
|
+
}) => record.get(/^773$/u).length > 0 ? ['component'] : ['host'],
|
|
38
39
|
compare: (a, b) => a[0] === b[0] ? 0.0 : -1.0
|
|
39
40
|
});
|
|
40
|
-
|
|
41
41
|
exports.default = _default;
|
|
42
42
|
//# sourceMappingURL=host-component.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"host-component.js","names":["_default","name","extract","record","get","length","compare","a","b","exports","default"],"sources":["../../../../src/match-detection/features/bib/host-component.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport default () => ({\n name: 'Host/Component record',\n extract: ({record}) => record.get(/^773$/u).length > 0 ? ['component'] : ['host'],\n compare: (a, b) => a[0] === b[0] ? 0.0 : -1.0\n});\n"],"mappings":";;;;;;AAAA;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,IAAAA,QAAA,GA4BeA,CAAA,MAAO;EACpBC,IAAI,EAAE,uBAAuB;EAC7BC,OAAO,EAAEA,CAAC;IAACC;EAAM,CAAC,KAAKA,MAAM,CAACC,GAAG,CAAC,QAAQ,CAAC,CAACC,MAAM,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC;EACjFC,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAAC,OAAA,CAAAC,OAAA,GAAAV,QAAA"}
|
|
@@ -45,6 +45,12 @@ Object.defineProperty(exports, "language", {
|
|
|
45
45
|
return _language.default;
|
|
46
46
|
}
|
|
47
47
|
});
|
|
48
|
+
Object.defineProperty(exports, "mediaType", {
|
|
49
|
+
enumerable: true,
|
|
50
|
+
get: function () {
|
|
51
|
+
return _mediaType.default;
|
|
52
|
+
}
|
|
53
|
+
});
|
|
48
54
|
Object.defineProperty(exports, "melindaId", {
|
|
49
55
|
enumerable: true,
|
|
50
56
|
get: function () {
|
|
@@ -63,6 +69,18 @@ Object.defineProperty(exports, "publicationTime", {
|
|
|
63
69
|
return _publicationTime.default;
|
|
64
70
|
}
|
|
65
71
|
});
|
|
72
|
+
Object.defineProperty(exports, "publicationTimeAllowConsYears", {
|
|
73
|
+
enumerable: true,
|
|
74
|
+
get: function () {
|
|
75
|
+
return _publicationTimeAllowConsYears.default;
|
|
76
|
+
}
|
|
77
|
+
});
|
|
78
|
+
Object.defineProperty(exports, "publicationTimeAllowConsYearsMulti", {
|
|
79
|
+
enumerable: true,
|
|
80
|
+
get: function () {
|
|
81
|
+
return _publicationTimeAllowConsYearsMulti.default;
|
|
82
|
+
}
|
|
83
|
+
});
|
|
66
84
|
Object.defineProperty(exports, "recordType", {
|
|
67
85
|
enumerable: true,
|
|
68
86
|
get: function () {
|
|
@@ -75,30 +93,27 @@ Object.defineProperty(exports, "title", {
|
|
|
75
93
|
return _title.default;
|
|
76
94
|
}
|
|
77
95
|
});
|
|
78
|
-
|
|
96
|
+
Object.defineProperty(exports, "titleVersionOriginal", {
|
|
97
|
+
enumerable: true,
|
|
98
|
+
get: function () {
|
|
99
|
+
return _titleVersionOriginal.default;
|
|
100
|
+
}
|
|
101
|
+
});
|
|
79
102
|
var _hostComponent = _interopRequireDefault(require("./host-component"));
|
|
80
|
-
|
|
81
103
|
var _isbn = _interopRequireDefault(require("./isbn"));
|
|
82
|
-
|
|
83
104
|
var _issn = _interopRequireDefault(require("./issn"));
|
|
84
|
-
|
|
85
105
|
var _otherStandardIdentifier = _interopRequireDefault(require("./other-standard-identifier"));
|
|
86
|
-
|
|
87
106
|
var _title = _interopRequireDefault(require("./title"));
|
|
88
|
-
|
|
107
|
+
var _titleVersionOriginal = _interopRequireDefault(require("./title-version-original"));
|
|
89
108
|
var _authors = _interopRequireDefault(require("./authors"));
|
|
90
|
-
|
|
91
109
|
var _recordType = _interopRequireDefault(require("./record-type"));
|
|
92
|
-
|
|
93
110
|
var _publicationTime = _interopRequireDefault(require("./publication-time"));
|
|
94
|
-
|
|
111
|
+
var _publicationTimeAllowConsYears = _interopRequireDefault(require("./publication-time-allow-cons-years"));
|
|
112
|
+
var _publicationTimeAllowConsYearsMulti = _interopRequireDefault(require("./publication-time-allow-cons-years-multi"));
|
|
95
113
|
var _language = _interopRequireDefault(require("./language"));
|
|
96
|
-
|
|
97
114
|
var _bibliographicLevel = _interopRequireDefault(require("./bibliographic-level"));
|
|
98
|
-
|
|
99
115
|
var _melindaId = _interopRequireDefault(require("./melinda-id"));
|
|
100
|
-
|
|
101
116
|
var _allSourceIds = _interopRequireDefault(require("./all-source-ids"));
|
|
102
|
-
|
|
117
|
+
var _mediaType = _interopRequireDefault(require("./media-type"));
|
|
103
118
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
104
119
|
//# sourceMappingURL=index.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../../../../src/match-detection/features/bib/index.js"],"
|
|
1
|
+
{"version":3,"file":"index.js","names":["_hostComponent","_interopRequireDefault","require","_isbn","_issn","_otherStandardIdentifier","_title","_titleVersionOriginal","_authors","_recordType","_publicationTime","_publicationTimeAllowConsYears","_publicationTimeAllowConsYearsMulti","_language","_bibliographicLevel","_melindaId","_allSourceIds","_mediaType","obj","__esModule","default"],"sources":["../../../../src/match-detection/features/bib/index.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport {default as hostComponent} from './host-component';\nexport {default as isbn} from './isbn';\nexport {default as issn} from './issn';\nexport {default as otherStandardIdentifier} from './other-standard-identifier';\nexport {default as title} from './title';\nexport {default as titleVersionOriginal} from './title-version-original';\nexport {default as authors} from './authors';\nexport {default as recordType} from './record-type';\nexport {default as publicationTime} from './publication-time';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years';\nexport {default as publicationTimeAllowConsYearsMulti} from './publication-time-allow-cons-years-multi';\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';\nexport {default as mediaType} from './media-type';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA4BA,IAAAA,cAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,KAAA,GAAAF,sBAAA,CAAAC,OAAA;AACA,IAAAE,KAAA,GAAAH,sBAAA,CAAAC,OAAA;AACA,IAAAG,wBAAA,GAAAJ,sBAAA,CAAAC,OAAA;AACA,IAAAI,MAAA,GAAAL,sBAAA,CAAAC,OAAA;AACA,IAAAK,qBAAA,GAAAN,sBAAA,CAAAC,OAAA;AACA,IAAAM,QAAA,GAAAP,sBAAA,CAAAC,OAAA;AACA,IAAAO,WAAA,GAAAR,sBAAA,CAAAC,OAAA;AACA,IAAAQ,gBAAA,GAAAT,sBAAA,CAAAC,OAAA;AACA,IAAAS,8BAAA,GAAAV,sBAAA,CAAAC,OAAA;AACA,IAAAU,mCAAA,GAAAX,sBAAA,CAAAC,OAAA;AACA,IAAAW,SAAA,GAAAZ,sBAAA,CAAAC,OAAA;AACA,IAAAY,mBAAA,GAAAb,sBAAA,CAAAC,OAAA;AACA,IAAAa,UAAA,GAAAd,sBAAA,CAAAC,OAAA;AACA,IAAAc,aAAA,GAAAf,sBAAA,CAAAC,OAAA;AACA,IAAAe,UAAA,GAAAhB,sBAAA,CAAAC,OAAA;AAAkD,SAAAD,uBAAAiB,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA"}
|
|
@@ -1,21 +1,14 @@
|
|
|
1
1
|
"use strict";
|
|
2
2
|
|
|
3
3
|
var _fixugen = _interopRequireDefault(require("@natlibfi/fixugen"));
|
|
4
|
-
|
|
5
4
|
var _fixura = require("@natlibfi/fixura");
|
|
6
|
-
|
|
7
5
|
var _chai = require("chai");
|
|
8
|
-
|
|
9
6
|
var _marcRecord = require("@natlibfi/marc-record");
|
|
10
|
-
|
|
11
7
|
var features = _interopRequireWildcard(require("."));
|
|
12
|
-
|
|
8
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
13
9
|
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); }
|
|
14
|
-
|
|
15
10
|
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; }
|
|
16
|
-
|
|
17
11
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
18
|
-
|
|
19
12
|
/**
|
|
20
13
|
*
|
|
21
14
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -43,6 +36,9 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
43
36
|
* for the JavaScript code in this file.
|
|
44
37
|
*
|
|
45
38
|
*/
|
|
39
|
+
|
|
40
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib:test');
|
|
41
|
+
const debugData = debug.extend('data');
|
|
46
42
|
describe('match-detection/features/bib/', () => {
|
|
47
43
|
(0, _fixugen.default)({
|
|
48
44
|
path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],
|
|
@@ -50,7 +46,6 @@ describe('match-detection/features/bib/', () => {
|
|
|
50
46
|
fixura: {
|
|
51
47
|
reader: _fixura.READERS.JSON
|
|
52
48
|
},
|
|
53
|
-
// eslint-disable-next-line max-statements
|
|
54
49
|
callback: ({
|
|
55
50
|
enabled = true,
|
|
56
51
|
feature,
|
|
@@ -61,7 +56,7 @@ describe('match-detection/features/bib/', () => {
|
|
|
61
56
|
if (!enabled) {
|
|
62
57
|
return;
|
|
63
58
|
}
|
|
64
|
-
|
|
59
|
+
debug(`Testing: ${feature} ${type}`);
|
|
65
60
|
if (type === 'extract') {
|
|
66
61
|
const {
|
|
67
62
|
expectedFeatures,
|
|
@@ -70,13 +65,15 @@ describe('match-detection/features/bib/', () => {
|
|
|
70
65
|
const record = new _marcRecord.MarcRecord(inputRecord, {
|
|
71
66
|
subfieldValues: false
|
|
72
67
|
});
|
|
68
|
+
debugData(`Record: ${record}`);
|
|
73
69
|
const {
|
|
74
70
|
extract
|
|
75
71
|
} = features[feature](options);
|
|
76
|
-
(0, _chai.expect)(extract(
|
|
72
|
+
(0, _chai.expect)(extract({
|
|
73
|
+
record
|
|
74
|
+
})).to.eql(expectedFeatures);
|
|
77
75
|
return;
|
|
78
76
|
}
|
|
79
|
-
|
|
80
77
|
if (type === 'compare') {
|
|
81
78
|
const {
|
|
82
79
|
featuresA,
|
|
@@ -89,7 +86,6 @@ describe('match-detection/features/bib/', () => {
|
|
|
89
86
|
(0, _chai.expect)(compare(featuresA, featuresB)).to.equal(expectedPoints);
|
|
90
87
|
return;
|
|
91
88
|
}
|
|
92
|
-
|
|
93
89
|
throw new Error(`Invalid type ${type}`);
|
|
94
90
|
}
|
|
95
91
|
});
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"index.spec.js","names":["_fixugen","_interopRequireDefault","require","_fixura","_chai","_marcRecord","features","_interopRequireWildcard","_debug","_getRequireWildcardCache","nodeInterop","WeakMap","cacheBabelInterop","cacheNodeInterop","obj","__esModule","default","cache","has","get","newObj","hasPropertyDescriptor","Object","defineProperty","getOwnPropertyDescriptor","key","prototype","hasOwnProperty","call","desc","set","debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","fixura","reader","READERS","JSON","callback","enabled","feature","options","type","expectations","expectedFeatures","inputRecord","record","MarcRecord","subfieldValues","extract","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 '.';\nimport createDebugLogger from 'debug';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib:test');\nconst debugData = debug.extend('data');\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 debug(`Testing: ${feature} ${type}`);\n\n if (type === 'extract') {\n const {expectedFeatures, inputRecord} = expectations;\n const record = new MarcRecord(inputRecord, {subfieldValues: false});\n debugData(`Record: ${record}`);\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,IAAAA,QAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,OAAA,GAAAD,OAAA;AACA,IAAAE,KAAA,GAAAF,OAAA;AACA,IAAAG,WAAA,GAAAH,OAAA;AACA,IAAAI,QAAA,GAAAC,uBAAA,CAAAL,OAAA;AACA,IAAAM,MAAA,GAAAP,sBAAA,CAAAC,OAAA;AAAsC,SAAAO,yBAAAC,WAAA,eAAAC,OAAA,kCAAAC,iBAAA,OAAAD,OAAA,QAAAE,gBAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,WAAA,WAAAA,WAAA,GAAAG,gBAAA,GAAAD,iBAAA,KAAAF,WAAA;AAAA,SAAAH,wBAAAO,GAAA,EAAAJ,WAAA,SAAAA,WAAA,IAAAI,GAAA,IAAAA,GAAA,CAAAC,UAAA,WAAAD,GAAA,QAAAA,GAAA,oBAAAA,GAAA,wBAAAA,GAAA,4BAAAE,OAAA,EAAAF,GAAA,UAAAG,KAAA,GAAAR,wBAAA,CAAAC,WAAA,OAAAO,KAAA,IAAAA,KAAA,CAAAC,GAAA,CAAAJ,GAAA,YAAAG,KAAA,CAAAE,GAAA,CAAAL,GAAA,SAAAM,MAAA,WAAAC,qBAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,GAAA,IAAAX,GAAA,QAAAW,GAAA,kBAAAH,MAAA,CAAAI,SAAA,CAAAC,cAAA,CAAAC,IAAA,CAAAd,GAAA,EAAAW,GAAA,SAAAI,IAAA,GAAAR,qBAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAV,GAAA,EAAAW,GAAA,cAAAI,IAAA,KAAAA,IAAA,CAAAV,GAAA,IAAAU,IAAA,CAAAC,GAAA,KAAAR,MAAA,CAAAC,cAAA,CAAAH,MAAA,EAAAK,GAAA,EAAAI,IAAA,YAAAT,MAAA,CAAAK,GAAA,IAAAX,GAAA,CAAAW,GAAA,SAAAL,MAAA,CAAAJ,OAAA,GAAAF,GAAA,MAAAG,KAAA,IAAAA,KAAA,CAAAa,GAAA,CAAAhB,GAAA,EAAAM,MAAA,YAAAA,MAAA;AAAA,SAAAnB,uBAAAa,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AAjCtC;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;;AAUA,MAAMiB,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;AACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAGtCC,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,EAAEA,CAAC;MAACC,OAAO,GAAG,IAAI;MAAEC,OAAO;MAAEC,OAAO;MAAEC,IAAI;MAAE,GAAGC;IAAY,CAAC,KAAK;MAEvE,IAAI,CAACJ,OAAO,EAAE;QACZ;MACF;MAEAd,KAAK,CAAE,YAAWe,OAAQ,IAAGE,IAAK,EAAC,CAAC;MAEpC,IAAIA,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;QACnErB,SAAS,CAAE,WAAUmB,MAAO,EAAC,CAAC;QAC9B,MAAM;UAACG;QAAO,CAAC,GAAGjD,QAAQ,CAACwC,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAS,YAAM,EAACD,OAAO,CAAC;UAACH;QAAM,CAAC,CAAC,CAAC,CAACK,EAAE,CAACC,GAAG,CAACR,gBAAgB,CAAC;QAClD;MACF;MAEA,IAAIF,IAAI,KAAK,SAAS,EAAE;QACtB,MAAM;UAACW,SAAS;UAAEC,SAAS;UAAEC;QAAc,CAAC,GAAGZ,YAAY;QAC3D,MAAM;UAACa;QAAO,CAAC,GAAGxD,QAAQ,CAACwC,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAS,YAAM,EAACM,OAAO,CAACH,SAAS,EAAEC,SAAS,CAAC,CAAC,CAACH,EAAE,CAACM,KAAK,CAACF,cAAc,CAAC;QAC9D;MACF;MAEA,MAAM,IAAIG,KAAK,CAAE,gBAAehB,IAAK,EAAC,CAAC;IACzC;EACF,CAAC,CAAC;AACJ,CAAC,CAAC"}
|
|
@@ -4,11 +4,10 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
6
|
exports.default = void 0;
|
|
7
|
-
|
|
8
7
|
var _standardIdentifierFactory = _interopRequireDefault(require("./standard-identifier-factory"));
|
|
9
|
-
|
|
8
|
+
var _isbn = require("isbn3");
|
|
9
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
10
10
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
11
|
-
|
|
12
11
|
/**
|
|
13
12
|
*
|
|
14
13
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -36,20 +35,43 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
36
35
|
* for the JavaScript code in this file.
|
|
37
36
|
*
|
|
38
37
|
*/
|
|
38
|
+
|
|
39
|
+
const debug = (0, _debug.default)(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);
|
|
40
|
+
const debugData = debug.extend('data');
|
|
39
41
|
var _default = () => {
|
|
42
|
+
const IDENTIFIER_NAME = 'ISBN';
|
|
43
|
+
function validatorAndNormalizer(string) {
|
|
44
|
+
const isbnParseResult = (0, _isbn.parse)(string);
|
|
45
|
+
debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);
|
|
46
|
+
if (isbnParseResult === null) {
|
|
47
|
+
debug(`Not parseable ISBN, just removing hyphens`);
|
|
48
|
+
return {
|
|
49
|
+
valid: false,
|
|
50
|
+
value: string.replace(/-/ug, '')
|
|
51
|
+
};
|
|
52
|
+
}
|
|
53
|
+
debug(`Parseable ISBN, normalizing to ISBN-13`);
|
|
54
|
+
return {
|
|
55
|
+
valid: true,
|
|
56
|
+
value: isbnParseResult.isbn13
|
|
57
|
+
};
|
|
58
|
+
}
|
|
40
59
|
const {
|
|
41
60
|
extract,
|
|
42
61
|
compare
|
|
43
62
|
} = (0, _standardIdentifierFactory.default)({
|
|
63
|
+
identifier: IDENTIFIER_NAME,
|
|
44
64
|
pattern: /^020$/u,
|
|
45
|
-
subfieldCodes: ['a', 'z']
|
|
65
|
+
subfieldCodes: ['a', 'z'],
|
|
66
|
+
validIdentifierSubfieldCodes: ['a'],
|
|
67
|
+
invalidIdentifierSubfieldCodes: ['z'],
|
|
68
|
+
validatorAndNormalizer
|
|
46
69
|
});
|
|
47
70
|
return {
|
|
48
71
|
extract,
|
|
49
72
|
compare,
|
|
50
|
-
name:
|
|
73
|
+
name: IDENTIFIER_NAME
|
|
51
74
|
};
|
|
52
75
|
};
|
|
53
|
-
|
|
54
76
|
exports.default = _default;
|
|
55
77
|
//# sourceMappingURL=isbn.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"isbn.js","names":["_standardIdentifierFactory","_interopRequireDefault","require","_isbn","_debug","obj","__esModule","default","debug","createDebugLogger","debugData","extend","_default","IDENTIFIER_NAME","validatorAndNormalizer","string","isbnParseResult","isbnParse","JSON","stringify","valid","value","replace","isbn13","extract","compare","createInterface","identifier","pattern","subfieldCodes","validIdentifierSubfieldCodes","invalidIdentifierSubfieldCodes","name","exports"],"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';\nimport {parse as isbnParse} from 'isbn3';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);\nconst debugData = debug.extend('data');\n\nexport default () => {\n const IDENTIFIER_NAME = 'ISBN';\n\n function validatorAndNormalizer(string) {\n const isbnParseResult = isbnParse(string);\n debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);\n if (isbnParseResult === null) {\n debug(`Not parseable ISBN, just removing hyphens`);\n return {valid: false, value: string.replace(/-/ug, '')};\n }\n debug(`Parseable ISBN, normalizing to ISBN-13`);\n return {valid: true, value: isbnParseResult.isbn13};\n }\n\n const {extract, compare} = createInterface({identifier: IDENTIFIER_NAME, pattern: /^020$/u, subfieldCodes: ['a', 'z'], validIdentifierSubfieldCodes: ['a'], invalidIdentifierSubfieldCodes: ['z'], validatorAndNormalizer});\n return {extract, compare, name: IDENTIFIER_NAME};\n};\n\n"],"mappings":";;;;;;AA4BA,IAAAA,0BAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,KAAA,GAAAD,OAAA;AACA,IAAAE,MAAA,GAAAH,sBAAA,CAAAC,OAAA;AAAsC,SAAAD,uBAAAI,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA9BtC;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;;AAMA,MAAMG,KAAK,GAAG,IAAAC,cAAiB,EAAE,sFAAqF,CAAC;AACvH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,IAAAC,QAAA,GAExBA,CAAA,KAAM;EACnB,MAAMC,eAAe,GAAG,MAAM;EAE9B,SAASC,sBAAsBA,CAACC,MAAM,EAAE;IACtC,MAAMC,eAAe,GAAG,IAAAC,WAAS,EAACF,MAAM,CAAC;IACzCL,SAAS,CAAE,oBAAmBQ,IAAI,CAACC,SAAS,CAACH,eAAe,CAAE,EAAC,CAAC;IAChE,IAAIA,eAAe,KAAK,IAAI,EAAE;MAC5BR,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO;QAACY,KAAK,EAAE,KAAK;QAAEC,KAAK,EAAEN,MAAM,CAACO,OAAO,CAAC,KAAK,EAAE,EAAE;MAAC,CAAC;IACzD;IACAd,KAAK,CAAE,wCAAuC,CAAC;IAC/C,OAAO;MAACY,KAAK,EAAE,IAAI;MAAEC,KAAK,EAAEL,eAAe,CAACO;IAAM,CAAC;EACrD;EAEA,MAAM;IAACC,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,UAAU,EAAEd,eAAe;IAAEe,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IAAEC,4BAA4B,EAAE,CAAC,GAAG,CAAC;IAAEC,8BAA8B,EAAE,CAAC,GAAG,CAAC;IAAEjB;EAAsB,CAAC,CAAC;EAC3N,OAAO;IAACU,OAAO;IAAEC,OAAO;IAAEO,IAAI,EAAEnB;EAAe,CAAC;AAClD,CAAC;AAAAoB,OAAA,CAAA1B,OAAA,GAAAK,QAAA"}
|
|
@@ -4,11 +4,8 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
6
|
exports.default = void 0;
|
|
7
|
-
|
|
8
7
|
var _standardIdentifierFactory = _interopRequireDefault(require("./standard-identifier-factory"));
|
|
9
|
-
|
|
10
8
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
11
|
-
|
|
12
9
|
/**
|
|
13
10
|
*
|
|
14
11
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
@@ -50,6 +47,5 @@ var _default = () => {
|
|
|
50
47
|
name: 'ISSN'
|
|
51
48
|
};
|
|
52
49
|
};
|
|
53
|
-
|
|
54
50
|
exports.default = _default;
|
|
55
51
|
//# sourceMappingURL=issn.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"issn.js","names":["_standardIdentifierFactory","_interopRequireDefault","require","obj","__esModule","default","_default","extract","compare","createInterface","pattern","subfieldCodes","name","exports"],"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,IAAAA,0BAAA,GAAAC,sBAAA,CAAAC,OAAA;AAA4D,SAAAD,uBAAAE,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;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,IAAAG,QAAA,GA8BeA,CAAA,KAAM;EACnB,MAAM;IAACC,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;AAAAC,OAAA,CAAAR,OAAA,GAAAC,QAAA"}
|
|
@@ -4,20 +4,16 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
6
|
exports.default = void 0;
|
|
7
|
-
|
|
8
7
|
var _debug = _interopRequireDefault(require("debug"));
|
|
9
|
-
|
|
8
|
+
var _matchingUtils = require("../../../matching-utils");
|
|
10
9
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
11
|
-
|
|
12
|
-
/* eslint-disable max-statements */
|
|
13
|
-
|
|
14
10
|
/**
|
|
15
11
|
*
|
|
16
12
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
17
13
|
*
|
|
18
14
|
* Melinda record matching modules for Javascript
|
|
19
15
|
*
|
|
20
|
-
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
16
|
+
* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)
|
|
21
17
|
*
|
|
22
18
|
* This file is part of melinda-record-matching-js
|
|
23
19
|
*
|
|
@@ -38,46 +34,38 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
38
34
|
* for the JavaScript code in this file.
|
|
39
35
|
*
|
|
40
36
|
*/
|
|
37
|
+
|
|
41
38
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features:language');
|
|
42
39
|
const debugData = debug.extend('data');
|
|
43
|
-
|
|
44
40
|
var _default = () => ({
|
|
45
41
|
name: 'Language',
|
|
46
|
-
extract:
|
|
42
|
+
extract: ({
|
|
43
|
+
record,
|
|
44
|
+
recordExternal
|
|
45
|
+
}) => {
|
|
46
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
47
47
|
const value008 = get008Value();
|
|
48
48
|
const values041 = get041Values();
|
|
49
|
-
debugData(
|
|
50
|
-
|
|
51
|
-
if (value008 && values041.length > 0) {
|
|
52
|
-
debugData(`There's both 008 and 041, searching for value in both`);
|
|
53
|
-
const correspondingValue = values041.find(v => v === value008);
|
|
54
|
-
debugData(`Corresponding value: ${correspondingValue}`);
|
|
55
|
-
return correspondingValue ? [correspondingValue] : [];
|
|
56
|
-
}
|
|
57
|
-
|
|
49
|
+
debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
|
|
58
50
|
if (!value008 && values041.length < 1) {
|
|
59
|
-
debugData(`No actual values found`);
|
|
51
|
+
debugData(`{$label} No actual values found`);
|
|
60
52
|
return [];
|
|
61
53
|
}
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
54
|
+
const allValues = value008 === undefined ? values041 : values041.concat(value008);
|
|
55
|
+
const uniqueSortedValues = [...new Set(allValues)].sort();
|
|
56
|
+
return uniqueSortedValues;
|
|
65
57
|
function get008Value() {
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
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;
|
|
69
|
-
debugData(`008 value: ${value}`);
|
|
70
|
-
|
|
58
|
+
const value = record.get(/^008$/u)?.[0]?.value || undefined;
|
|
59
|
+
debugData(`${label} 008 value: ${value}`);
|
|
71
60
|
if (!value) {
|
|
72
61
|
return undefined;
|
|
73
62
|
}
|
|
74
|
-
|
|
75
63
|
const code = value.slice(35, 38);
|
|
76
|
-
debugData(
|
|
77
|
-
return code
|
|
78
|
-
}
|
|
79
|
-
// Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes
|
|
64
|
+
debugData(`${label} 008 code: ${code}`);
|
|
65
|
+
return isLangCodeForALanguage(code) ? code : undefined;
|
|
66
|
+
}
|
|
80
67
|
|
|
68
|
+
// Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes
|
|
81
69
|
|
|
82
70
|
function get041Values() {
|
|
83
71
|
return record.get(/^041$/u).filter(({
|
|
@@ -86,28 +74,68 @@ var _default = () => ({
|
|
|
86
74
|
subfields
|
|
87
75
|
}) => subfields).flat().filter(({
|
|
88
76
|
code
|
|
89
|
-
}) => code === 'a' || code === 'd').
|
|
77
|
+
}) => code === 'a' || code === 'd').filter(({
|
|
78
|
+
value
|
|
79
|
+
}) => value && isLangCodeForALanguage(value)).map(({
|
|
90
80
|
value
|
|
91
81
|
}) => value);
|
|
92
82
|
}
|
|
83
|
+
|
|
84
|
+
// Check if a string is a possible, validly formed language code for a single language
|
|
85
|
+
// Currently accept also codes in capitals
|
|
86
|
+
function isLangCodeForALanguage(code) {
|
|
87
|
+
if (code.length !== 3) {
|
|
88
|
+
debugData(`Code ${code} is not correct length (3) for a language code.`);
|
|
89
|
+
return false;
|
|
90
|
+
}
|
|
91
|
+
if (code === '|||' || code === ' ' || code === '^^^' || code === 'mul' || code === 'zxx') {
|
|
92
|
+
debugData(`Code ${code} is not code for a spesific language.`);
|
|
93
|
+
return false;
|
|
94
|
+
}
|
|
95
|
+
const langCodePattern = /^[a-z][a-z][a-z]$/ui;
|
|
96
|
+
if (!langCodePattern.test(code)) {
|
|
97
|
+
debugData(`Code ${code} is not valid as a language code`);
|
|
98
|
+
return false;
|
|
99
|
+
}
|
|
100
|
+
return true;
|
|
101
|
+
}
|
|
93
102
|
},
|
|
103
|
+
// eslint-disable-next-line max-statements
|
|
94
104
|
compare: (a, b) => {
|
|
95
|
-
debugData(`Comparing ${JSON.stringify(a
|
|
96
|
-
|
|
105
|
+
debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);
|
|
97
106
|
if (a.length === 0 || b.length === 0) {
|
|
98
107
|
debugData(`No language to compare`);
|
|
99
108
|
return 0;
|
|
100
109
|
}
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
if (a[0] === b[0]) {
|
|
110
|
+
if (a.length === b.length && a.every((element, index) => element === b[index])) {
|
|
111
|
+
debugData(`All languages match`);
|
|
105
112
|
return 0.1;
|
|
106
113
|
}
|
|
107
|
-
|
|
108
|
-
|
|
114
|
+
const {
|
|
115
|
+
matchingValues,
|
|
116
|
+
possibleMatchValues,
|
|
117
|
+
maxValues
|
|
118
|
+
} = (0, _matchingUtils.getMatchCounts)(a, b);
|
|
119
|
+
if (matchingValues < 1) {
|
|
120
|
+
debug(`Both have languages, but none of these match.`);
|
|
121
|
+
return -1.0;
|
|
122
|
+
}
|
|
123
|
+
debug(`Both have languages, ${matchingValues}/${possibleMatchValues} valid languages match.`);
|
|
124
|
+
// ignore non-matches if there is mismatching amount of values
|
|
125
|
+
debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);
|
|
126
|
+
//we give some kind of penalty for mismatching amount of values instead of simple divide?
|
|
127
|
+
const missingCount = maxValues - possibleMatchValues;
|
|
128
|
+
const misMatchCount = possibleMatchValues - matchingValues;
|
|
129
|
+
debug(`\t missing: ${missingCount}`);
|
|
130
|
+
debug(`\t mismatches: ${misMatchCount}`);
|
|
131
|
+
const penaltyForMissing = 0.02 * (maxValues - possibleMatchValues);
|
|
132
|
+
const penaltyForMisMatch = 0.05 * (possibleMatchValues - matchingValues);
|
|
133
|
+
debug(`\t points: penaltyForMissing: ${penaltyForMissing}`);
|
|
134
|
+
debug(`\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);
|
|
135
|
+
const points = Number(Number(0.1 - penaltyForMisMatch - penaltyForMissing).toFixed(2));
|
|
136
|
+
debug(`Total points: ${points}`);
|
|
137
|
+
return points;
|
|
109
138
|
}
|
|
110
139
|
});
|
|
111
|
-
|
|
112
140
|
exports.default = _default;
|
|
113
141
|
//# sourceMappingURL=language.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"
|
|
1
|
+
{"version":3,"file":"language.js","names":["_debug","_interopRequireDefault","require","_matchingUtils","obj","__esModule","default","debug","createDebugLogger","debugData","extend","_default","name","extract","record","recordExternal","label","value008","get008Value","values041","get041Values","JSON","stringify","length","allValues","undefined","concat","uniqueSortedValues","Set","sort","value","get","code","slice","isLangCodeForALanguage","filter","ind2","map","subfields","flat","langCodePattern","test","compare","a","b","every","element","index","matchingValues","possibleMatchValues","maxValues","getMatchCounts","missingCount","misMatchCount","penaltyForMissing","penaltyForMisMatch","points","Number","toFixed","exports"],"sources":["../../../../src/match-detection/features/bib/language.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {getMatchCounts} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:language');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n name: 'Language',\n extract: ({record, recordExternal}) => {\n const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n const value008 = get008Value();\n const values041 = get041Values();\n debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);\n\n if (!value008 && values041.length < 1) {\n debugData(`{$label} No actual values found`);\n return [];\n }\n\n const allValues = value008 === undefined ? values041 : values041.concat(value008);\n const uniqueSortedValues = [...new Set(allValues)].sort();\n\n return uniqueSortedValues;\n\n function get008Value() {\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n debugData(`${label} 008 value: ${value}`);\n\n if (!value) {\n return undefined;\n }\n\n const code = value.slice(35, 38);\n debugData(`${label} 008 code: ${code}`);\n return isLangCodeForALanguage(code) ? code : undefined;\n }\n\n // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes\n\n function get041Values() {\n return record.get(/^041$/u)\n .filter(({ind2}) => ind2 === ' ')\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code === 'a' || code === 'd')\n .filter(({value}) => value && isLangCodeForALanguage(value))\n .map(({value}) => value);\n }\n\n // Check if a string is a possible, validly formed language code for a single language\n // Currently accept also codes in capitals\n function isLangCodeForALanguage(code) {\n if (code.length !== 3) {\n debugData(`Code ${code} is not correct length (3) for a language code.`);\n return false;\n }\n if (code === '|||' || code === ' ' || code === '^^^' || code === 'mul' || code === 'zxx') {\n debugData(`Code ${code} is not code for a spesific language.`);\n return false;\n }\n const langCodePattern = /^[a-z][a-z][a-z]$/ui;\n if (!langCodePattern.test(code)) {\n debugData(`Code ${code} is not valid as a language code`);\n return false;\n }\n return true;\n }\n\n },\n // eslint-disable-next-line max-statements\n compare: (a, b) => {\n debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);\n\n if (a.length === 0 || b.length === 0) {\n debugData(`No language to compare`);\n return 0;\n }\n\n if (a.length === b.length && a.every((element, index) => element === b[index])) {\n debugData(`All languages match`);\n return 0.1;\n }\n\n const {matchingValues, possibleMatchValues, maxValues} = getMatchCounts(a, b);\n\n if (matchingValues < 1) {\n debug(`Both have languages, but none of these match.`);\n return -1.0;\n }\n debug(`Both have languages, ${matchingValues}/${possibleMatchValues} valid languages match.`);\n // ignore non-matches if there is mismatching amount of values\n debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);\n //we give some kind of penalty for mismatching amount of values instead of simple divide?\n const missingCount = maxValues - possibleMatchValues;\n const misMatchCount = possibleMatchValues - matchingValues;\n debug(`\\t missing: ${missingCount}`);\n debug(`\\t mismatches: ${misMatchCount}`);\n\n const penaltyForMissing = 0.02 * (maxValues - possibleMatchValues);\n const penaltyForMisMatch = 0.05 * (possibleMatchValues - matchingValues);\n debug(`\\t points: penaltyForMissing: ${penaltyForMissing}`);\n debug(`\\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);\n\n const points = Number(Number(0.1 - penaltyForMisMatch - penaltyForMissing).toFixed(2));\n debug(`Total points: ${points}`);\n\n return points;\n }\n});\n"],"mappings":";;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,cAAA,GAAAD,OAAA;AAAuD,SAAAD,uBAAAG,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA7BvD;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAKA,MAAMG,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;AACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,IAAAC,QAAA,GAExBA,CAAA,MAAO;EACpBC,IAAI,EAAE,UAAU;EAChBC,OAAO,EAAEA,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,CAAC,CAAC;IAC9B,MAAMC,SAAS,GAAGC,YAAY,CAAC,CAAC;IAChCX,SAAS,CAAE,GAAEO,KAAM,SAAQK,IAAI,CAACC,SAAS,CAACL,QAAQ,CAAE,UAASI,IAAI,CAACC,SAAS,CAACH,SAAS,CAAE,EAAC,CAAC;IAEzF,IAAI,CAACF,QAAQ,IAAIE,SAAS,CAACI,MAAM,GAAG,CAAC,EAAE;MACrCd,SAAS,CAAE,iCAAgC,CAAC;MAC5C,OAAO,EAAE;IACX;IAEA,MAAMe,SAAS,GAAGP,QAAQ,KAAKQ,SAAS,GAAGN,SAAS,GAAGA,SAAS,CAACO,MAAM,CAACT,QAAQ,CAAC;IACjF,MAAMU,kBAAkB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACJ,SAAS,CAAC,CAAC,CAACK,IAAI,CAAC,CAAC;IAEzD,OAAOF,kBAAkB;IAEzB,SAAST,WAAWA,CAAA,EAAG;MACrB,MAAMY,KAAK,GAAGhB,MAAM,CAACiB,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,EAAED,KAAK,IAAIL,SAAS;MAC3DhB,SAAS,CAAE,GAAEO,KAAM,eAAcc,KAAM,EAAC,CAAC;MAEzC,IAAI,CAACA,KAAK,EAAE;QACV,OAAOL,SAAS;MAClB;MAEA,MAAMO,IAAI,GAAGF,KAAK,CAACG,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;MAChCxB,SAAS,CAAE,GAAEO,KAAM,cAAagB,IAAK,EAAC,CAAC;MACvC,OAAOE,sBAAsB,CAACF,IAAI,CAAC,GAAGA,IAAI,GAAGP,SAAS;IACxD;;IAEA;;IAEA,SAASL,YAAYA,CAAA,EAAG;MACtB,OAAON,MAAM,CAACiB,GAAG,CAAC,QAAQ,CAAC,CACxBI,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCC,GAAG,CAAC,CAAC;QAACC;MAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,CAAC,CAAC,CACNJ,MAAM,CAAC,CAAC;QAACH;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,IAAIA,IAAI,KAAK,GAAG,CAAC,CAChDG,MAAM,CAAC,CAAC;QAACL;MAAK,CAAC,KAAKA,KAAK,IAAII,sBAAsB,CAACJ,KAAK,CAAC,CAAC,CAC3DO,GAAG,CAAC,CAAC;QAACP;MAAK,CAAC,KAAKA,KAAK,CAAC;IAC5B;;IAEA;IACA;IACA,SAASI,sBAAsBA,CAACF,IAAI,EAAE;MACpC,IAAIA,IAAI,CAACT,MAAM,KAAK,CAAC,EAAE;QACrBd,SAAS,CAAE,QAAOuB,IAAK,iDAAgD,CAAC;QACxE,OAAO,KAAK;MACd;MACA,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,IAAIA,IAAI,KAAK,KAAK,EAAE;QAC1FvB,SAAS,CAAE,QAAOuB,IAAK,uCAAsC,CAAC;QAC9D,OAAO,KAAK;MACd;MACA,MAAMQ,eAAe,GAAG,qBAAqB;MAC7C,IAAI,CAACA,eAAe,CAACC,IAAI,CAACT,IAAI,CAAC,EAAE;QAC/BvB,SAAS,CAAE,QAAOuB,IAAK,kCAAiC,CAAC;QACzD,OAAO,KAAK;MACd;MACA,OAAO,IAAI;IACb;EAEF,CAAC;EACD;EACAU,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjBnC,SAAS,CAAE,aAAYY,IAAI,CAACC,SAAS,CAACqB,CAAC,CAAE,QAAOtB,IAAI,CAACC,SAAS,CAACsB,CAAC,CAAE,EAAC,CAAC;IAEpE,IAAID,CAAC,CAACpB,MAAM,KAAK,CAAC,IAAIqB,CAAC,CAACrB,MAAM,KAAK,CAAC,EAAE;MACpCd,SAAS,CAAE,wBAAuB,CAAC;MACnC,OAAO,CAAC;IACV;IAEA,IAAIkC,CAAC,CAACpB,MAAM,KAAKqB,CAAC,CAACrB,MAAM,IAAIoB,CAAC,CAACE,KAAK,CAAC,CAACC,OAAO,EAAEC,KAAK,KAAKD,OAAO,KAAKF,CAAC,CAACG,KAAK,CAAC,CAAC,EAAE;MAC9EtC,SAAS,CAAE,qBAAoB,CAAC;MAChC,OAAO,GAAG;IACZ;IAEA,MAAM;MAACuC,cAAc;MAAEC,mBAAmB;MAAEC;IAAS,CAAC,GAAG,IAAAC,6BAAc,EAACR,CAAC,EAAEC,CAAC,CAAC;IAE7E,IAAII,cAAc,GAAG,CAAC,EAAE;MACtBzC,KAAK,CAAE,+CAA8C,CAAC;MACtD,OAAO,CAAC,GAAG;IACb;IACAA,KAAK,CAAE,wBAAuByC,cAAe,IAAGC,mBAAoB,yBAAwB,CAAC;IAC7F;IACA1C,KAAK,CAAE,qBAAoB0C,mBAAoB,IAAGC,SAAU,EAAC,CAAC;IAC9D;IACA,MAAME,YAAY,GAAGF,SAAS,GAAGD,mBAAmB;IACpD,MAAMI,aAAa,GAAGJ,mBAAmB,GAAGD,cAAc;IAC1DzC,KAAK,CAAE,eAAc6C,YAAa,EAAC,CAAC;IACpC7C,KAAK,CAAE,kBAAiB8C,aAAc,EAAC,CAAC;IAExC,MAAMC,iBAAiB,GAAG,IAAI,IAAIJ,SAAS,GAAGD,mBAAmB,CAAC;IAClE,MAAMM,kBAAkB,GAAG,IAAI,IAAIN,mBAAmB,GAAGD,cAAc,CAAC;IACxEzC,KAAK,CAAE,iCAAgC+C,iBAAkB,EAAC,CAAC;IAC3D/C,KAAK,CAAE,kCAAiCgD,kBAAmB,EAAC,CAAC;IAE7D,MAAMC,MAAM,GAAGC,MAAM,CAACA,MAAM,CAAC,GAAG,GAAGF,kBAAkB,GAAGD,iBAAiB,CAAC,CAACI,OAAO,CAAC,CAAC,CAAC,CAAC;IACtFnD,KAAK,CAAE,iBAAgBiD,MAAO,EAAC,CAAC;IAEhC,OAAOA,MAAM;EACf;AACF,CAAC,CAAC;AAAAG,OAAA,CAAArD,OAAA,GAAAK,QAAA"}
|