@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
package/dist/matching-utils.js
CHANGED
|
@@ -3,21 +3,22 @@
|
|
|
3
3
|
Object.defineProperty(exports, "__esModule", {
|
|
4
4
|
value: true
|
|
5
5
|
});
|
|
6
|
+
exports.extractSubfieldsFromField = extractSubfieldsFromField;
|
|
7
|
+
exports.getMatchCounts = getMatchCounts;
|
|
6
8
|
exports.getMelindaIdsF035 = getMelindaIdsF035;
|
|
7
9
|
exports.getSubfieldValues = getSubfieldValues;
|
|
10
|
+
exports.testStringOrNumber = testStringOrNumber;
|
|
11
|
+
exports.uniqueSubfields = uniqueSubfields;
|
|
8
12
|
exports.validateSidFieldSubfieldCounts = validateSidFieldSubfieldCounts;
|
|
9
|
-
|
|
10
13
|
var _debug = _interopRequireDefault(require("debug"));
|
|
11
|
-
|
|
12
14
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
13
|
-
|
|
14
15
|
/**
|
|
15
16
|
*
|
|
16
17
|
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
17
18
|
*
|
|
18
19
|
* Melinda record matching modules for Javascript
|
|
19
20
|
*
|
|
20
|
-
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
21
|
+
* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
|
|
21
22
|
*
|
|
22
23
|
* This file is part of melinda-record-matching-js
|
|
23
24
|
*
|
|
@@ -38,36 +39,32 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
38
39
|
* for the JavaScript code in this file.
|
|
39
40
|
*
|
|
40
41
|
*/
|
|
41
|
-
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:utils');
|
|
42
|
-
const debugData = debug.extend('data'); // eslint-disable-next-line max-statements
|
|
43
42
|
|
|
43
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:utils');
|
|
44
|
+
const debugData = debug.extend('data');
|
|
44
45
|
function getMelindaIdsF035(record) {
|
|
45
46
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:melinda-id');
|
|
46
47
|
const debugData = debug.extend('data');
|
|
47
48
|
const f035s = record.getFields('035');
|
|
48
|
-
|
|
49
49
|
if (f035s.length < 1) {
|
|
50
50
|
debug(`No f035s found.`);
|
|
51
51
|
return [];
|
|
52
52
|
}
|
|
53
|
-
|
|
54
53
|
const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));
|
|
55
54
|
const f035MelindaIds = [...new Set(allF035MelindaIds)];
|
|
56
55
|
debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);
|
|
57
56
|
debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);
|
|
58
57
|
debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);
|
|
59
58
|
return f035MelindaIds;
|
|
60
|
-
|
|
61
59
|
function toMelindaIds({
|
|
62
60
|
subfields
|
|
63
61
|
}) {
|
|
64
62
|
const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
|
|
65
|
-
return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => melindaIdRegExp.test(sub.value)).map(({
|
|
63
|
+
return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({
|
|
66
64
|
value
|
|
67
|
-
}) => value.replace(melindaIdRegExp, '$<id>'));
|
|
65
|
+
}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');
|
|
68
66
|
}
|
|
69
67
|
}
|
|
70
|
-
|
|
71
68
|
function validateSidFieldSubfieldCounts(field) {
|
|
72
69
|
// Valid SID-fields have just one $c and one $b
|
|
73
70
|
debugData(`Validating SID field ${JSON.stringify(field)}`);
|
|
@@ -76,20 +73,71 @@ function validateSidFieldSubfieldCounts(field) {
|
|
|
76
73
|
debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);
|
|
77
74
|
return countC === 1 && countB === 1;
|
|
78
75
|
}
|
|
79
|
-
|
|
80
76
|
function countSubfields(field, subfieldCode) {
|
|
81
77
|
// debug(`Counting subfields ${subfieldCode}`);
|
|
82
78
|
return field.subfields.filter(({
|
|
83
79
|
code
|
|
84
80
|
}) => code === subfieldCode).length;
|
|
85
81
|
}
|
|
86
|
-
|
|
87
82
|
function getSubfieldValues(field, subfieldCode) {
|
|
88
83
|
debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);
|
|
89
84
|
return field.subfields.filter(({
|
|
90
85
|
code
|
|
91
86
|
}) => code === subfieldCode).map(({
|
|
92
87
|
value
|
|
93
|
-
}) => value);
|
|
88
|
+
}) => testStringOrNumber(value) ? String(value) : '').filter(value => value);
|
|
89
|
+
}
|
|
90
|
+
function testStringOrNumber(value) {
|
|
91
|
+
if (typeof value === 'string' || typeof value === 'number') {
|
|
92
|
+
return true;
|
|
93
|
+
}
|
|
94
|
+
return false;
|
|
95
|
+
}
|
|
96
|
+
function extractSubfieldsFromField(field, subfieldCodes) {
|
|
97
|
+
if (field === undefined || field.subfields === undefined) {
|
|
98
|
+
return [];
|
|
99
|
+
}
|
|
100
|
+
const resultSubfields = field.subfields.filter(({
|
|
101
|
+
code
|
|
102
|
+
}) => subfieldCodes.includes(code)).map(({
|
|
103
|
+
code,
|
|
104
|
+
value
|
|
105
|
+
}) => ({
|
|
106
|
+
code,
|
|
107
|
+
value: testStringOrNumber(value) ? String(value) : ''
|
|
108
|
+
})).filter(value => value);
|
|
109
|
+
return resultSubfields;
|
|
110
|
+
}
|
|
111
|
+
function uniqueSubfields(subfields) {
|
|
112
|
+
return subfields.reduce((arr, e) => {
|
|
113
|
+
if (!arr.find(item => item.code === e.code && item.value === e.value)) {
|
|
114
|
+
const newArr = arr.concat(e);
|
|
115
|
+
return newArr;
|
|
116
|
+
}
|
|
117
|
+
return arr;
|
|
118
|
+
}, []);
|
|
119
|
+
}
|
|
120
|
+
function getMatchCounts(aValues, bValues) {
|
|
121
|
+
const matchingValues = getMatchingValuesAmount(aValues, bValues);
|
|
122
|
+
return {
|
|
123
|
+
maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
|
|
124
|
+
// possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
|
|
125
|
+
possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
|
|
126
|
+
matchingValues
|
|
127
|
+
};
|
|
128
|
+
function getMatchingValuesAmount(aValues, bValues) {
|
|
129
|
+
if (bValues.length > aValues.length) {
|
|
130
|
+
return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
|
|
131
|
+
}
|
|
132
|
+
if (aValues.length > bValues.length) {
|
|
133
|
+
return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
|
|
134
|
+
}
|
|
135
|
+
|
|
136
|
+
// If we have same amount of values, we'll check matches both ways, to avoid mixups in cases
|
|
137
|
+
// there would be duplicate values
|
|
138
|
+
const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
|
|
139
|
+
const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
|
|
140
|
+
return aToB < bToA ? aToB : bToA;
|
|
141
|
+
}
|
|
94
142
|
}
|
|
95
143
|
//# sourceMappingURL=matching-utils.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/matching-utils.js"],"names":["debug","debugData","extend","getMelindaIdsF035","record","f035s","getFields","length","allF035MelindaIds","concat","map","toMelindaIds","f035MelindaIds","Set","JSON","stringify","subfields","melindaIdRegExp","filter","sub","includes","code","test","value","replace","validateSidFieldSubfieldCounts","field","countC","countSubfields","countB","subfieldCode","getSubfieldValues"],"mappings":";;;;;;;;;AA4BA;;;;AA5BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA,MAAMA,KAAK,GAAG,oBAAkB,yCAAlB,CAAd;AACA,MAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB,C,CAEA;;AACO,SAASC,iBAAT,CAA2BC,MAA3B,EAAmC;AAExC,QAAMJ,KAAK,GAAG,oBAAkB,8CAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AAEA,QAAMG,KAAK,GAAGD,MAAM,CAACE,SAAP,CAAiB,KAAjB,CAAd;;AAEA,MAAID,KAAK,CAACE,MAAN,GAAe,CAAnB,EAAsB;AACpBP,IAAAA,KAAK,CAAE,iBAAF,CAAL;AACA,WAAO,EAAP;AACD;;AAED,QAAMQ,iBAAiB,GAAG,GAAGC,MAAH,CAAU,GAAGJ,KAAK,CAACK,GAAN,CAAUC,YAAV,CAAb,CAA1B;AACA,QAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQL,iBAAR,CAAJ,CAAvB;AAEAP,EAAAA,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeV,KAAf,CAAsB,EAApD,CAAT;AACAJ,EAAAA,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeP,iBAAf,CAAkC,EAAzE,CAAT;AACAP,EAAAA,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeH,cAAf,CAA+B,EAA1E,CAAT;AAEA,SAAOA,cAAP;;AAEA,WAASD,YAAT,CAAsB;AAACK,IAAAA;AAAD,GAAtB,EAAmC;AACjC,UAAMC,eAAe,GAAG,8CAAxB;AAEA,WAAOD,SAAS,CACbE,MADI,CACGC,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWC,QAAX,CAAoBD,GAAG,CAACE,IAAxB,CADV,EAEJH,MAFI,CAEGC,GAAG,IAAIF,eAAe,CAACK,IAAhB,CAAqBH,GAAG,CAACI,KAAzB,CAFV,EAGJb,GAHI,CAGA,CAAC;AAACa,MAAAA;AAAD,KAAD,KAAaA,KAAK,CAACC,OAAN,CAAcP,eAAd,EAA+B,OAA/B,CAHb,CAAP;AAKD;AACF;;AAEM,SAASQ,8BAAT,CAAwCC,KAAxC,EAA+C;AACpD;AACAzB,EAAAA,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,EAA/C,CAAT;AACA,QAAMC,MAAM,GAAGC,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;AACA,QAAMG,MAAM,GAAGD,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;AACA1B,EAAAA,KAAK,CAAE,SAAQ2B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAE,EAAvF,CAAL;AAEA,SAAOF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAlC;AACD;;AAED,SAASD,cAAT,CAAwBF,KAAxB,EAA+BI,YAA/B,EAA6C;AAC3C;AACA,SAAOJ,KAAK,CAACV,SAAN,CAAgBE,MAAhB,CAAuB,CAAC;AAACG,IAAAA;AAAD,GAAD,KAAYA,IAAI,KAAKS,YAA5C,EAA0DvB,MAAjE;AACD;;AAEM,SAASwB,iBAAT,CAA2BL,KAA3B,EAAkCI,YAAlC,EAAgD;AACrD7B,EAAAA,SAAS,CAAE,oBAAmB6B,YAAa,SAAQhB,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,EAAhE,CAAT;AACA,SAAOA,KAAK,CAACV,SAAN,CAAgBE,MAAhB,CAAuB,CAAC;AAACG,IAAAA;AAAD,GAAD,KAAYA,IAAI,KAAKS,YAA5C,EAA0DpB,GAA1D,CAA8D,CAAC;AAACa,IAAAA;AAAD,GAAD,KAAaA,KAA3E,CAAP;AACD","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';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\n// eslint-disable-next-line max-statements\nexport function getMelindaIdsF035(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n const debugData = debug.extend('data');\n\n const f035s = record.getFields('035');\n\n if (f035s.length < 1) {\n debug(`No f035s found.`);\n return [];\n }\n\n const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));\n const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n return f035MelindaIds;\n\n function toMelindaIds({subfields}) {\n const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code))\n .filter(sub => melindaIdRegExp.test(sub.value))\n .map(({value}) => value.replace(melindaIdRegExp, '$<id>'));\n\n }\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n // Valid SID-fields have just one $c and one $b\n debugData(`Validating SID field ${JSON.stringify(field)}`);\n const countC = countSubfields(field, 'c');\n const countB = countSubfields(field, 'b');\n debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n // debug(`Counting subfields ${subfieldCode}`);\n return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n return field.subfields.filter(({code}) => code === subfieldCode).map(({value}) => value);\n}\n\n"],"file":"matching-utils.js"}
|
|
1
|
+
{"version":3,"file":"matching-utils.js","names":["_debug","_interopRequireDefault","require","obj","__esModule","default","debug","createDebugLogger","debugData","extend","getMelindaIdsF035","record","f035s","getFields","length","allF035MelindaIds","concat","map","toMelindaIds","f035MelindaIds","Set","JSON","stringify","subfields","melindaIdRegExp","filter","sub","includes","code","testStringOrNumber","value","test","String","replace","validateSidFieldSubfieldCounts","field","countC","countSubfields","countB","subfieldCode","getSubfieldValues","extractSubfieldsFromField","subfieldCodes","undefined","resultSubfields","uniqueSubfields","reduce","arr","e","find","item","newArr","getMatchCounts","aValues","bValues","matchingValues","getMatchingValuesAmount","maxValues","possibleMatchValues","aValue","some","bValue","aToB","bToA"],"sources":["../src/matching-utils.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n const debugData = debug.extend('data');\n\n const f035s = record.getFields('035');\n\n if (f035s.length < 1) {\n debug(`No f035s found.`);\n return [];\n }\n\n const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));\n const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n return f035MelindaIds;\n\n function toMelindaIds({subfields}) {\n const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code))\n .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n\n }\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n // Valid SID-fields have just one $c and one $b\n debugData(`Validating SID field ${JSON.stringify(field)}`);\n const countC = countSubfields(field, 'c');\n const countB = countSubfields(field, 'b');\n debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n // debug(`Counting subfields ${subfieldCode}`);\n return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n return field.subfields\n .filter(({code}) => code === subfieldCode)\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n if (typeof value === 'string' || typeof value === 'number') {\n return true;\n }\n return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n if (field === undefined || field.subfields === undefined) {\n return [];\n }\n const resultSubfields = field.subfields\n .filter(({code}) => subfieldCodes.includes(code))\n .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n .filter(value => value);\n return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n return subfields.reduce((arr, e) => {\n if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n const newArr = arr.concat(e);\n return newArr;\n }\n return arr;\n }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n return {\n maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n matchingValues\n };\n\n function getMatchingValuesAmount(aValues, bValues) {\n if (bValues.length > aValues.length) {\n return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n }\n if (aValues.length > bValues.length) {\n return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n }\n\n // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n // there would be duplicate values\n const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n return aToB < bToA ? aToB : bToA;\n }\n}\n"],"mappings":";;;;;;;;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AAAsC,SAAAD,uBAAAE,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA5BtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAIA,MAAMG,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiBA,CAACC,MAAM,EAAE;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAiB,EAAC,8CAA8C,CAAC;EAC/E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAS,CAAC,KAAK,CAAC;EAErC,IAAID,KAAK,CAACE,MAAM,GAAG,CAAC,EAAE;IACpBR,KAAK,CAAE,iBAAgB,CAAC;IACxB,OAAO,EAAE;EACX;EAEA,MAAMS,iBAAiB,GAAG,EAAE,CAACC,MAAM,CAAC,GAAGJ,KAAK,CAACK,GAAG,CAACC,YAAY,CAAC,CAAC;EAC/D,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACL,iBAAiB,CAAC,CAAC;EAEtDP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACV,KAAK,CAAE,EAAC,CAAC;EAC/DJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACP,iBAAiB,CAAE,EAAC,CAAC;EACpFP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACH,cAAc,CAAE,EAAC,CAAC;EAErF,OAAOA,cAAc;EAErB,SAASD,YAAYA,CAAC;IAACK;EAAS,CAAC,EAAE;IACjC,MAAMC,eAAe,GAAG,8CAA8C;IAEtE,OAAOD,SAAS,CACbE,MAAM,CAACC,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,GAAG,CAACE,IAAI,CAAC,CAAC,CAC5CH,MAAM,CAACC,GAAG,IAAIG,kBAAkB,CAACH,GAAG,CAACI,KAAK,CAAC,IAAIN,eAAe,CAACO,IAAI,CAACC,MAAM,CAACN,GAAG,CAACI,KAAK,CAAC,CAAC,CAAC,CACvFb,GAAG,CAAC,CAAC;MAACa;IAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAACT,eAAe,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC;EAEvG;AACF;AAEO,SAASU,8BAA8BA,CAACC,KAAK,EAAE;EACpD;EACA3B,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC1D,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC7B,KAAK,CAAE,SAAQ8B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAE,EAAC,CAAC;EAE9F,OAAOF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAC;AACrC;AAEA,SAASD,cAAcA,CAACF,KAAK,EAAEI,YAAY,EAAE;EAC3C;EACA,OAAOJ,KAAK,CAACZ,SAAS,CAACE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CAACzB,MAAM;AACzE;AAEO,SAAS0B,iBAAiBA,CAACL,KAAK,EAAEI,YAAY,EAAE;EACrD/B,SAAS,CAAE,oBAAmB+B,YAAa,SAAQlB,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC3E,OAAOA,KAAK,CAACZ,SAAS,CACnBE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CACzCtB,GAAG,CAAC,CAAC;IAACa;EAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;AAC3B;AAEO,SAASD,kBAAkBA,CAACC,KAAK,EAAE;EACxC,IAAI,OAAOA,KAAK,KAAK,QAAQ,IAAI,OAAOA,KAAK,KAAK,QAAQ,EAAE;IAC1D,OAAO,IAAI;EACb;EACA,OAAO,KAAK;AACd;AAEO,SAASW,yBAAyBA,CAACN,KAAK,EAAEO,aAAa,EAAE;EAC9D,IAAIP,KAAK,KAAKQ,SAAS,IAAIR,KAAK,CAACZ,SAAS,KAAKoB,SAAS,EAAE;IACxD,OAAO,EAAE;EACX;EACA,MAAMC,eAAe,GAAGT,KAAK,CAACZ,SAAS,CACpCE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKc,aAAa,CAACf,QAAQ,CAACC,IAAI,CAAC,CAAC,CAChDX,GAAG,CAAC,CAAC;IAACW,IAAI;IAAEE;EAAK,CAAC,MAAM;IAACF,IAAI;IAAEE,KAAK,EAAED,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG;EAAE,CAAC,CAAC,CAAC,CACvFL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;EACzB,OAAOc,eAAe;AACxB;AAEO,SAASC,eAAeA,CAACtB,SAAS,EAAE;EACzC,OAAOA,SAAS,CAACuB,MAAM,CAAC,CAACC,GAAG,EAAEC,CAAC,KAAK;IAClC,IAAI,CAACD,GAAG,CAACE,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACtB,IAAI,KAAKoB,CAAC,CAACpB,IAAI,IAAIsB,IAAI,CAACpB,KAAK,KAAKkB,CAAC,CAAClB,KAAK,CAAC,EAAE;MACrE,MAAMqB,MAAM,GAAGJ,GAAG,CAAC/B,MAAM,CAACgC,CAAC,CAAC;MAC5B,OAAOG,MAAM;IACf;IACA,OAAOJ,GAAG;EACZ,CAAC,EAAE,EAAE,CAAC;AACR;AAEO,SAASK,cAAcA,CAACC,OAAO,EAAEC,OAAO,EAAE;EAE/C,MAAMC,cAAc,GAAGC,uBAAuB,CAACH,OAAO,EAAEC,OAAO,CAAC;EAEhE,OAAO;IACLG,SAAS,EAAEJ,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM;IAC5E;IACA4C,mBAAmB,EAAEL,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM;IACtFyC;EACF,CAAC;EAED,SAASC,uBAAuBA,CAACH,OAAO,EAAEC,OAAO,EAAE;IACjD,IAAIA,OAAO,CAACxC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,EAAE;MACnC,OAAOuC,OAAO,CAAC5B,MAAM,CAACkC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC/C,MAAM;IACnF;IACA,IAAIuC,OAAO,CAACvC,MAAM,GAAGwC,OAAO,CAACxC,MAAM,EAAE;MACnC,OAAOwC,OAAO,CAAC7B,MAAM,CAACoC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC7C,MAAM;IACnF;;IAEA;IACA;IACA,MAAMgD,IAAI,GAAGT,OAAO,CAAC5B,MAAM,CAACkC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC/C,MAAM;IACvF,MAAMiD,IAAI,GAAGT,OAAO,CAAC7B,MAAM,CAACoC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC7C,MAAM;IAEvF,OAAOgD,IAAI,GAAGC,IAAI,GAAGD,IAAI,GAAGC,IAAI;EAClC;AACF"}
|
package/package.json
CHANGED
|
@@ -14,15 +14,16 @@
|
|
|
14
14
|
"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
|
|
15
15
|
},
|
|
16
16
|
"license": "LGPL-3.0+",
|
|
17
|
-
"version": "2.0.
|
|
17
|
+
"version": "2.0.7-alpha.1",
|
|
18
18
|
"main": "./dist/index.js",
|
|
19
19
|
"engines": {
|
|
20
|
-
"node": ">=
|
|
20
|
+
"node": ">=18"
|
|
21
21
|
},
|
|
22
22
|
"publishConfig": {
|
|
23
23
|
"access": "public"
|
|
24
24
|
},
|
|
25
25
|
"scripts": {
|
|
26
|
+
"prepare": "npm run build",
|
|
26
27
|
"start": "node dist/index.js",
|
|
27
28
|
"cli": "node dist/cli.js",
|
|
28
29
|
"lint": "eslint src",
|
|
@@ -35,34 +36,36 @@
|
|
|
35
36
|
"watch:test": "cross-env DEBUG=1 NODE_ENV=test nodemon -w src -w test-fixtures --exec 'npm run test:dev'"
|
|
36
37
|
},
|
|
37
38
|
"dependencies": {
|
|
38
|
-
"@natlibfi/marc-record": "^7.
|
|
39
|
-
"@natlibfi/marc-record-serializers": "^
|
|
40
|
-
"@natlibfi/
|
|
41
|
-
"
|
|
42
|
-
"
|
|
43
|
-
"
|
|
44
|
-
"
|
|
45
|
-
"
|
|
39
|
+
"@natlibfi/marc-record": "^7.3.0",
|
|
40
|
+
"@natlibfi/marc-record-serializers": "^10.1.0",
|
|
41
|
+
"@natlibfi/melinda-commons": "^13.0.5",
|
|
42
|
+
"@natlibfi/sru-client": "^6.0.4",
|
|
43
|
+
"debug": "^4.3.4",
|
|
44
|
+
"isbn3": "^1.1.40",
|
|
45
|
+
"moment": "^2.29.4",
|
|
46
|
+
"natural": "^6.5.0",
|
|
47
|
+
"uuid": "^9.0.0",
|
|
48
|
+
"winston": "^3.10.0"
|
|
46
49
|
},
|
|
47
50
|
"devDependencies": {
|
|
48
|
-
"@babel/cli": "^7.
|
|
49
|
-
"@babel/core": "^7.
|
|
50
|
-
"@babel/
|
|
51
|
-
"@babel/
|
|
52
|
-
"@babel/
|
|
53
|
-
"@babel/
|
|
54
|
-
"@natlibfi/eslint-config-melinda-backend": "^
|
|
55
|
-
"@natlibfi/fixugen": "^
|
|
56
|
-
"@natlibfi/fixugen-http-client": "^
|
|
57
|
-
"@natlibfi/fixura": "^
|
|
51
|
+
"@babel/cli": "^7.22.10",
|
|
52
|
+
"@babel/core": "^7.22.10",
|
|
53
|
+
"@babel/eslint-parser": "^7.22.10",
|
|
54
|
+
"@babel/node": "^7.22.10",
|
|
55
|
+
"@babel/preset-env": "^7.22.10",
|
|
56
|
+
"@babel/register": "^7.22.5",
|
|
57
|
+
"@natlibfi/eslint-config-melinda-backend": "^3.0.1",
|
|
58
|
+
"@natlibfi/fixugen": "^2.0.1",
|
|
59
|
+
"@natlibfi/fixugen-http-client": "^3.0.0",
|
|
60
|
+
"@natlibfi/fixura": "^3.0.1",
|
|
58
61
|
"babel-plugin-istanbul": "^6.1.1",
|
|
59
62
|
"babel-plugin-rewire": "^1.2.0",
|
|
60
|
-
"chai": "^4.3.
|
|
63
|
+
"chai": "^4.3.7",
|
|
61
64
|
"chai-as-promised": "^7.1.1",
|
|
62
65
|
"cross-env": "^7.0.3",
|
|
63
|
-
"eslint": "^
|
|
64
|
-
"mocha": "^
|
|
65
|
-
"nodemon": "^
|
|
66
|
+
"eslint": "^8.47.0",
|
|
67
|
+
"mocha": "^10.2.0",
|
|
68
|
+
"nodemon": "^3.0.1",
|
|
66
69
|
"nyc": "^15.1.0"
|
|
67
70
|
},
|
|
68
71
|
"eslintConfig": {
|
|
@@ -4,7 +4,7 @@
|
|
|
4
4
|
*
|
|
5
5
|
* Melinda record matching modules for Javascript
|
|
6
6
|
*
|
|
7
|
-
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
7
|
+
* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
|
|
8
8
|
*
|
|
9
9
|
* This file is part of melinda-record-matching-js
|
|
10
10
|
*
|
|
@@ -31,56 +31,95 @@ import createClient, {SruSearchError} from '@natlibfi/sru-client';
|
|
|
31
31
|
import {MarcRecord} from '@natlibfi/marc-record';
|
|
32
32
|
import {MARCXML} from '@natlibfi/marc-record-serializers';
|
|
33
33
|
import generateQueryList from './query-list';
|
|
34
|
+
import {Error as MatchingError} from '@natlibfi/melinda-commons';
|
|
34
35
|
|
|
35
36
|
export {searchTypes} from './query-list';
|
|
36
37
|
|
|
37
38
|
export class CandidateSearchError extends Error {}
|
|
38
39
|
|
|
39
|
-
|
|
40
|
+
// serverMaxResults : maximum size of total search result available from the server, defaults to Aleph's 20000
|
|
41
|
+
|
|
42
|
+
export default ({record, searchSpec, url, maxCandidates, maxRecordsPerRequest = 50, serverMaxResult = 20000}) => {
|
|
40
43
|
MarcRecord.setValidationOptions({subfieldValues: false});
|
|
41
44
|
|
|
42
45
|
const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search');
|
|
46
|
+
const debugData = debug.extend('data');
|
|
47
|
+
|
|
48
|
+
debugData(`SearchSpec: ${JSON.stringify(searchSpec)}`);
|
|
49
|
+
debugData(`Url: ${url}`);
|
|
50
|
+
debugData(`MaxRecordsPerRequest ${maxRecordsPerRequest}`);
|
|
51
|
+
debugData(`ServerMaxResult: ${serverMaxResult}`);
|
|
52
|
+
debugData(`MaxCandidates: ${maxCandidates}`);
|
|
53
|
+
|
|
54
|
+
// Do not retrieve more candidates than defined in maxCandidates
|
|
55
|
+
const adjustedMaxRecordsPerRequest = maxRecordsPerRequest >= maxCandidates ? maxCandidates : maxRecordsPerRequest;
|
|
56
|
+
|
|
43
57
|
const inputRecordId = getRecordId(record);
|
|
44
58
|
const queryList = generateQueryList(record, searchSpec);
|
|
45
59
|
const client = createClient({
|
|
46
|
-
url,
|
|
60
|
+
url,
|
|
61
|
+
maxRecordsPerRequest: adjustedMaxRecordsPerRequest,
|
|
47
62
|
version: '2.0',
|
|
48
63
|
retrieveAll: false
|
|
49
64
|
});
|
|
50
65
|
|
|
51
66
|
debug(`Searching matches for ${inputRecordId}`);
|
|
52
67
|
debug(`Generated queryList ${JSON.stringify(queryList)}`);
|
|
53
|
-
|
|
68
|
+
|
|
69
|
+
// if generateQueryList errored we should throw 422
|
|
70
|
+
|
|
71
|
+
if (queryList.length === 0) {
|
|
54
72
|
throw new CandidateSearchError(`Generated query list contains no queries`);
|
|
55
73
|
}
|
|
56
74
|
|
|
57
|
-
//
|
|
58
|
-
|
|
75
|
+
// state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)
|
|
76
|
+
// state.query : current query (undefined if there was no queries left)
|
|
77
|
+
// state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)
|
|
78
|
+
// state.queryCandidateCounter: amount of candidates (records+failures) retrieved from SRU for matching for current query, including the current record+failure set (undefined if there were no queries left)
|
|
79
|
+
// state.queriesLeft : amount of queries left
|
|
80
|
+
// state.queryCounter : sequence for current query
|
|
81
|
+
// state.maxedQueries : queries that resulted in more than serverMaxResults hits
|
|
82
|
+
|
|
83
|
+
|
|
84
|
+
return async ({queryOffset = 0, resultSetOffset = 1, totalRecords = 0, searchCounter = 0, queryCandidateCounter = 0, queryCounter = 0, maxedQueries = []}) => {
|
|
59
85
|
const query = queryList[queryOffset];
|
|
60
86
|
|
|
61
87
|
if (query) {
|
|
62
|
-
const {records, nextOffset} = await retrieveRecords();
|
|
88
|
+
const {records, failures, nextOffset, total} = await retrieveRecords();
|
|
89
|
+
|
|
90
|
+
// If resultSetOffset === 1 this is the first search for the current query
|
|
91
|
+
debugData(`ResultSetOffset: ${resultSetOffset}`);
|
|
92
|
+
const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;
|
|
93
|
+
const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;
|
|
94
|
+
const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;
|
|
95
|
+
const newQueryCandidateCounter = resultSetOffset === 1 ? records.length + failures.length : queryCandidateCounter + records.length + failures.length;
|
|
96
|
+
|
|
97
|
+
const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;
|
|
98
|
+
const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;
|
|
63
99
|
|
|
64
100
|
if (typeof nextOffset === 'number') {
|
|
65
|
-
debug(`
|
|
66
|
-
return {records, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1)};
|
|
101
|
+
debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);
|
|
102
|
+
return {records, failures, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};
|
|
67
103
|
}
|
|
68
|
-
debug(`Query ${queryOffset} ${query} done
|
|
69
|
-
|
|
104
|
+
debug(`Query ${queryOffset} ${query} done.`);
|
|
105
|
+
debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);
|
|
106
|
+
return {records, failures, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};
|
|
70
107
|
}
|
|
71
108
|
|
|
72
109
|
debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);
|
|
73
|
-
return {records: []};
|
|
110
|
+
return {records: [], failures: [], queriesLeft: 0, queryCounter, maxedQueries};
|
|
74
111
|
|
|
75
112
|
function retrieveRecords() {
|
|
76
113
|
return new Promise((resolve, reject) => {
|
|
77
114
|
const promises = [];
|
|
115
|
+
// eslint-disable-next-line functional/no-let
|
|
116
|
+
let totalRecords = 0;
|
|
78
117
|
|
|
79
118
|
debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);
|
|
80
119
|
|
|
81
120
|
client.searchRetrieve(query, {startRecord: resultSetOffset})
|
|
82
121
|
.on('error', err => {
|
|
83
|
-
// eslint-disable-next-line functional/no-conditional-
|
|
122
|
+
// eslint-disable-next-line functional/no-conditional-statements
|
|
84
123
|
if (err instanceof SruSearchError) {
|
|
85
124
|
debug(`SRU SruSearchError for query: ${query}: ${err}`);
|
|
86
125
|
reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));
|
|
@@ -88,37 +127,45 @@ export default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {
|
|
|
88
127
|
debug(`SRU error for query: ${query}: ${err}`);
|
|
89
128
|
reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));
|
|
90
129
|
})
|
|
130
|
+
.on('total', total => {
|
|
131
|
+
debug(`Got total: ${total}`);
|
|
132
|
+
totalRecords += total;
|
|
133
|
+
})
|
|
91
134
|
.on('end', async nextOffset => {
|
|
92
135
|
try {
|
|
93
|
-
const
|
|
94
|
-
|
|
136
|
+
const recordPromises = await Promise.allSettled(promises);
|
|
137
|
+
debugData(`All recordPromises: ${JSON.stringify(recordPromises)}`);
|
|
138
|
+
const filtered = recordPromises.filter(r => r.status === 'fulfilled').map(r => r.value);
|
|
139
|
+
const failures = recordPromises.filter(r => r.status === 'rejected').map(r => ({status: r.reason.status, payload: r.reason.payload}));
|
|
140
|
+
|
|
141
|
+
debug(`Found ${recordPromises.length} records`);
|
|
142
|
+
debug(`Found ${filtered.length} convertable candidates`);
|
|
143
|
+
debug(`Found ${failures.length} NON-convertable candidates`);
|
|
144
|
+
debugData(`Converted: ${JSON.stringify(filtered)}.`);
|
|
145
|
+
debugData(`Not converted: ${JSON.stringify(failures)}.`);
|
|
95
146
|
|
|
96
|
-
debug(`Found ${filtered.length} candidates`);
|
|
97
147
|
|
|
98
|
-
resolve({nextOffset, records: filtered});
|
|
148
|
+
resolve({nextOffset, records: filtered, failures, total: totalRecords});
|
|
99
149
|
} catch (err) {
|
|
150
|
+
debug(`Error caught on END`);
|
|
100
151
|
reject(err);
|
|
101
152
|
}
|
|
102
153
|
})
|
|
103
|
-
.on('record',
|
|
154
|
+
.on('record', recordXML => {
|
|
104
155
|
promises.push(handleRecord()); // eslint-disable-line functional/immutable-data
|
|
105
156
|
|
|
106
157
|
async function handleRecord() {
|
|
107
158
|
try {
|
|
108
|
-
const
|
|
109
|
-
const
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
// Record *should* match itself AND in REST the input record is given id 000000001 always
|
|
113
|
-
debug(`Checking ${inputRecordId} vs ${foundRecordId}`);
|
|
114
|
-
if (inputRecordId === foundRecordId) {
|
|
115
|
-
debug(`Input and candidate are the same record per 001. Discarding candidate`);
|
|
116
|
-
return;
|
|
117
|
-
}
|
|
118
|
-
|
|
119
|
-
return {record: foundRecordMarc, id: foundRecordId};
|
|
159
|
+
const recordMarc = await MARCXML.from(recordXML, {subfieldValues: false});
|
|
160
|
+
const recordId = getRecordId(recordMarc);
|
|
161
|
+
|
|
162
|
+
return {record: recordMarc, id: recordId};
|
|
120
163
|
} catch (err) {
|
|
121
|
-
|
|
164
|
+
// What should this do?
|
|
165
|
+
const idFromXML = getRecordIdFromXML(recordXML);
|
|
166
|
+
debugData(`Failed converting record: ${err.message}, id: ${idFromXML}, data: ${recordXML}`);
|
|
167
|
+
//return {message: `Failed converting record: ${err.message}`, id: idFromXML, data: recordXML};
|
|
168
|
+
throw new MatchingError(422, {message: `Failed converting record: ${err.message}`, id: idFromXML || '000000000', data: recordXML});
|
|
122
169
|
}
|
|
123
170
|
}
|
|
124
171
|
});
|
|
@@ -126,8 +173,23 @@ export default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {
|
|
|
126
173
|
}
|
|
127
174
|
};
|
|
128
175
|
|
|
176
|
+
function checkMaxedQuery(query, total, serverMaxResult) {
|
|
177
|
+
if (total >= serverMaxResult) {
|
|
178
|
+
debug(`WARNING: Query ${query} resulted in ${total} hits which meets the serverMaxResult (${serverMaxResult}) `);
|
|
179
|
+
return query;
|
|
180
|
+
}
|
|
181
|
+
}
|
|
182
|
+
|
|
183
|
+
|
|
129
184
|
function getRecordId(record) {
|
|
130
185
|
const [field] = record.get(/^001$/u);
|
|
131
186
|
return field ? field.value : '';
|
|
132
187
|
}
|
|
188
|
+
|
|
189
|
+
function getRecordIdFromXML(recordXML) {
|
|
190
|
+
//<controlfield tag=\"001\">015376846</controlfield
|
|
191
|
+
debug(`Cannot yet find possible database record id from recordXML (length ${recordXML.length})`);
|
|
192
|
+
return undefined;
|
|
193
|
+
}
|
|
194
|
+
|
|
133
195
|
};
|
|
@@ -4,7 +4,7 @@
|
|
|
4
4
|
*
|
|
5
5
|
* Melinda record matching modules for Javascript
|
|
6
6
|
*
|
|
7
|
-
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
7
|
+
* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
|
|
8
8
|
*
|
|
9
9
|
* This file is part of melinda-record-matching-js
|
|
10
10
|
*
|
|
@@ -30,7 +30,11 @@ import {expect} from 'chai';
|
|
|
30
30
|
import {READERS} from '@natlibfi/fixura';
|
|
31
31
|
import generateTests from '@natlibfi/fixugen-http-client';
|
|
32
32
|
import {MarcRecord} from '@natlibfi/marc-record';
|
|
33
|
+
import {Error as MatchingError} from '@natlibfi/melinda-commons';
|
|
33
34
|
import createSearchInterface, {CandidateSearchError} from '.';
|
|
35
|
+
import createDebugLogger from 'debug';
|
|
36
|
+
|
|
37
|
+
const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:test');
|
|
34
38
|
|
|
35
39
|
describe('candidate-search', () => {
|
|
36
40
|
generateTests({
|
|
@@ -42,8 +46,7 @@ describe('candidate-search', () => {
|
|
|
42
46
|
}
|
|
43
47
|
});
|
|
44
48
|
|
|
45
|
-
|
|
46
|
-
async function callback({getFixture, factoryOptions, searchOptions, expectedFactoryError, expectedSearchError, enabled = true}) {
|
|
49
|
+
async function callback({getFixture, factoryOptions, searchOptions, expectedFactoryError = false, expectedSearchError = false, enabled = true}) {
|
|
47
50
|
const url = 'http://foo.bar';
|
|
48
51
|
|
|
49
52
|
if (!enabled) {
|
|
@@ -61,55 +64,55 @@ describe('candidate-search', () => {
|
|
|
61
64
|
}
|
|
62
65
|
|
|
63
66
|
const search = createSearchInterface({...formatFactoryOptions(), url});
|
|
64
|
-
await iterate({searchOptions});
|
|
67
|
+
await iterate({searchOptions, expectedSearchError});
|
|
65
68
|
|
|
66
69
|
function formatFactoryOptions() {
|
|
70
|
+
debug(`Using factoryOptions: ${JSON.stringify(factoryOptions)}`);
|
|
67
71
|
return {
|
|
68
72
|
...factoryOptions,
|
|
69
|
-
maxRecordsPerRequest: 1,
|
|
73
|
+
maxRecordsPerRequest: factoryOptions.maxRecordsPerRequest || 1,
|
|
74
|
+
maxServerResults: factoryOptions.maxServerResults || undefined,
|
|
70
75
|
record: new MarcRecord(factoryOptions.record, {subfieldValues: false})
|
|
71
76
|
};
|
|
72
77
|
}
|
|
73
78
|
|
|
74
|
-
|
|
75
|
-
async function iterate({searchOptions, count = 1}) {
|
|
79
|
+
async function iterate({searchOptions, expectedSearchError, expectedErrorStatus, count = 1}) {
|
|
76
80
|
const expectedResults = getFixture(`expectedResults${count}.json`);
|
|
77
81
|
|
|
78
|
-
if (expectedSearchError) { // eslint-disable-line functional/no-conditional-
|
|
82
|
+
if (expectedSearchError) { // eslint-disable-line functional/no-conditional-statements
|
|
79
83
|
try {
|
|
80
84
|
await search(searchOptions);
|
|
81
85
|
throw new Error('Expected an error');
|
|
82
86
|
} catch (err) {
|
|
87
|
+
debug(`Got an error: ${err}`);
|
|
83
88
|
expect(err).to.be.an('error');
|
|
84
|
-
|
|
89
|
+
const errorMessage = err instanceof MatchingError ? err.payload.message : err.message;
|
|
90
|
+
const errorStatus = err instanceof MatchingError ? err.status : undefined;
|
|
91
|
+
debug(`errorMessage: ${errorMessage}, errorStatus: ${errorStatus}`);
|
|
92
|
+
expect(errorMessage).to.match(new RegExp(expectedSearchError, 'u'));
|
|
93
|
+
|
|
94
|
+
if (expectedErrorStatus) {
|
|
95
|
+
expect(errorStatus).to.be(expectedErrorStatus);
|
|
96
|
+
return;
|
|
97
|
+
}
|
|
85
98
|
return;
|
|
86
99
|
}
|
|
87
100
|
}
|
|
88
101
|
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
if (results.records.length > 0) {
|
|
94
|
-
return iterate({
|
|
95
|
-
searchOptions: resultsToOptions(results),
|
|
96
|
-
count: count + 1
|
|
97
|
-
});
|
|
102
|
+
// eslint-disable-next-line functional/no-conditional-statements
|
|
103
|
+
if (!expectedSearchError) {
|
|
104
|
+
const results = await search(searchOptions);
|
|
105
|
+
expect(formatResults(results)).to.eql(expectedResults);
|
|
98
106
|
}
|
|
99
107
|
|
|
100
108
|
function formatResults(results) {
|
|
101
|
-
|
|
109
|
+
debug(results);
|
|
102
110
|
return {
|
|
103
111
|
...results,
|
|
104
112
|
records: results.records.map(({record, id}) => ({id, record: record.toObject()}))
|
|
105
113
|
};
|
|
106
114
|
}
|
|
107
115
|
|
|
108
|
-
function resultsToOptions(results) {
|
|
109
|
-
return Object.entries(results)
|
|
110
|
-
.filter(([k]) => k === 'records' === false) // If key is 'records' return false
|
|
111
|
-
.reduce((acc, [k, v]) => ({...acc, [k]: v}), {});
|
|
112
|
-
}
|
|
113
116
|
}
|
|
114
117
|
}
|
|
115
118
|
});
|