@natlibfi/melinda-record-matching 4.3.5 → 5.0.0-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.
Files changed (101) hide show
  1. package/.github/workflows/melinda-node-tests.yml +11 -7
  2. package/eslint.config.js +52 -0
  3. package/example.env +4 -0
  4. package/package.json +22 -90
  5. package/src/candidate-search/choose-queries.js +1 -3
  6. package/src/candidate-search/index.js +63 -89
  7. package/src/candidate-search/{index.spec.js → index.test.js} +19 -18
  8. package/src/candidate-search/query-list/bib.js +10 -4
  9. package/src/candidate-search/query-list/{bib.spec.js → bib.test.js} +8 -7
  10. package/src/candidate-search/query-list/component.js +127 -0
  11. package/src/candidate-search/query-list/index.js +23 -9
  12. package/src/cli.js +177 -0
  13. package/src/index.js +22 -37
  14. package/src/{index.spec.js → index.test.js} +11 -14
  15. package/src/match-detection/features/bib/all-source-ids.js +2 -2
  16. package/src/match-detection/features/bib/authors.js +3 -2
  17. package/src/match-detection/features/bib/index.js +16 -16
  18. package/src/match-detection/features/bib/{index.spec.js → index.test.js} +7 -6
  19. package/src/match-detection/features/bib/isbn.js +2 -2
  20. package/src/match-detection/features/bib/issn.js +1 -1
  21. package/src/match-detection/features/bib/language.js +5 -5
  22. package/src/match-detection/features/bib/media-type.js +1 -1
  23. package/src/match-detection/features/bib/melinda-id.js +1 -1
  24. package/src/match-detection/features/bib/melinda-identifier-factory.js +5 -5
  25. package/src/match-detection/features/bib/other-standard-identifier.js +1 -1
  26. package/src/match-detection/features/bib/publication-time-allow-cons-years-multi.js +6 -6
  27. package/src/match-detection/features/bib/publication-time-allow-cons-years.js +1 -1
  28. package/src/match-detection/features/bib/publication-time.js +1 -1
  29. package/src/match-detection/features/bib/standard-identifier-factory.js +1 -2
  30. package/src/match-detection/features/bib/title-version-original.js +2 -1
  31. package/src/match-detection/features/bib/title.js +6 -5
  32. package/src/match-detection/features/index.js +1 -1
  33. package/src/match-detection/index.js +1 -1
  34. package/src/match-detection/{index.spec.js → index.test.js} +9 -8
  35. package/src/matching-utils.js +8 -9
  36. package/dist/candidate-search/candidate-search-utils.js +0 -29
  37. package/dist/candidate-search/candidate-search-utils.js.map +0 -1
  38. package/dist/candidate-search/choose-queries.js +0 -122
  39. package/dist/candidate-search/choose-queries.js.map +0 -1
  40. package/dist/candidate-search/index.js +0 -260
  41. package/dist/candidate-search/index.js.map +0 -1
  42. package/dist/candidate-search/index.spec.js +0 -133
  43. package/dist/candidate-search/index.spec.js.map +0 -1
  44. package/dist/candidate-search/query-list/bib.js +0 -458
  45. package/dist/candidate-search/query-list/bib.js.map +0 -1
  46. package/dist/candidate-search/query-list/bib.spec.js +0 -46
  47. package/dist/candidate-search/query-list/bib.spec.js.map +0 -1
  48. package/dist/candidate-search/query-list/index.js +0 -47
  49. package/dist/candidate-search/query-list/index.js.map +0 -1
  50. package/dist/index.js +0 -555
  51. package/dist/index.js.map +0 -1
  52. package/dist/index.spec.js +0 -99
  53. package/dist/index.spec.js.map +0 -1
  54. package/dist/match-detection/features/bib/all-source-ids.js +0 -91
  55. package/dist/match-detection/features/bib/all-source-ids.js.map +0 -1
  56. package/dist/match-detection/features/bib/authors.js +0 -86
  57. package/dist/match-detection/features/bib/authors.js.map +0 -1
  58. package/dist/match-detection/features/bib/bibliographic-level.js +0 -15
  59. package/dist/match-detection/features/bib/bibliographic-level.js.map +0 -1
  60. package/dist/match-detection/features/bib/host-component.js +0 -15
  61. package/dist/match-detection/features/bib/host-component.js.map +0 -1
  62. package/dist/match-detection/features/bib/index.js +0 -119
  63. package/dist/match-detection/features/bib/index.js.map +0 -1
  64. package/dist/match-detection/features/bib/index.spec.js +0 -65
  65. package/dist/match-detection/features/bib/index.spec.js.map +0 -1
  66. package/dist/match-detection/features/bib/isbn.js +0 -49
  67. package/dist/match-detection/features/bib/isbn.js.map +0 -1
  68. package/dist/match-detection/features/bib/issn.js +0 -24
  69. package/dist/match-detection/features/bib/issn.js.map +0 -1
  70. package/dist/match-detection/features/bib/language.js +0 -113
  71. package/dist/match-detection/features/bib/language.js.map +0 -1
  72. package/dist/match-detection/features/bib/media-type.js +0 -56
  73. package/dist/match-detection/features/bib/media-type.js.map +0 -1
  74. package/dist/match-detection/features/bib/melinda-id.js +0 -21
  75. package/dist/match-detection/features/bib/melinda-id.js.map +0 -1
  76. package/dist/match-detection/features/bib/melinda-identifier-factory.js +0 -62
  77. package/dist/match-detection/features/bib/melinda-identifier-factory.js.map +0 -1
  78. package/dist/match-detection/features/bib/other-standard-identifier.js +0 -25
  79. package/dist/match-detection/features/bib/other-standard-identifier.js.map +0 -1
  80. package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js +0 -229
  81. package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js.map +0 -1
  82. package/dist/match-detection/features/bib/publication-time-allow-cons-years.js +0 -44
  83. package/dist/match-detection/features/bib/publication-time-allow-cons-years.js.map +0 -1
  84. package/dist/match-detection/features/bib/publication-time.js +0 -21
  85. package/dist/match-detection/features/bib/publication-time.js.map +0 -1
  86. package/dist/match-detection/features/bib/record-type.js +0 -18
  87. package/dist/match-detection/features/bib/record-type.js.map +0 -1
  88. package/dist/match-detection/features/bib/standard-identifier-factory.js +0 -188
  89. package/dist/match-detection/features/bib/standard-identifier-factory.js.map +0 -1
  90. package/dist/match-detection/features/bib/title-version-original.js +0 -52
  91. package/dist/match-detection/features/bib/title-version-original.js.map +0 -1
  92. package/dist/match-detection/features/bib/title.js +0 -70
  93. package/dist/match-detection/features/bib/title.js.map +0 -1
  94. package/dist/match-detection/features/index.js +0 -11
  95. package/dist/match-detection/features/index.js.map +0 -1
  96. package/dist/match-detection/index.js +0 -174
  97. package/dist/match-detection/index.js.map +0 -1
  98. package/dist/match-detection/index.spec.js +0 -63
  99. package/dist/match-detection/index.spec.js.map +0 -1
  100. package/dist/matching-utils.js +0 -115
  101. package/dist/matching-utils.js.map +0 -1
@@ -1,174 +0,0 @@
1
- "use strict";
2
-
3
- Object.defineProperty(exports, "__esModule", {
4
- value: true
5
- });
6
- exports.features = exports.default = void 0;
7
- var _debug = _interopRequireDefault(require("debug"));
8
- var features = _interopRequireWildcard(require("./features"));
9
- exports.features = features;
10
- function _getRequireWildcardCache(e) { if ("function" != typeof WeakMap) return null; var r = new WeakMap(), t = new WeakMap(); return (_getRequireWildcardCache = function (e) { return e ? t : r; })(e); }
11
- function _interopRequireWildcard(e, r) { if (!r && e && e.__esModule) return e; if (null === e || "object" != typeof e && "function" != typeof e) return { default: e }; var t = _getRequireWildcardCache(r); if (t && t.has(e)) return t.get(e); var n = { __proto__: null }, a = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var u in e) if ("default" !== u && {}.hasOwnProperty.call(e, u)) { var i = a ? Object.getOwnPropertyDescriptor(e, u) : null; i && (i.get || i.set) ? Object.defineProperty(n, u, i) : n[u] = e[u]; } return n.default = e, t && t.set(e, n), n; }
12
- function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
13
- var _default = ({
14
- strategy,
15
- treshold = 0.9
16
- }, returnStrategy = false, localSettings = {}) => ({
17
- recordA,
18
- recordB,
19
- recordAExternal = {},
20
- recordBExternal = {}
21
- }) => {
22
- const minProbabilityQuantifier = 0.5;
23
- const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection');
24
- const debugData = debug.extend('data');
25
- debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
26
- debugData(`Records: A: ${recordA}\nB: ${recordB}`);
27
- debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);
28
- // We could add here labels for records if we didn't get external labels
29
-
30
- const featuresA = extractFeatures({
31
- record: recordA,
32
- recordExternal: recordAExternal,
33
- localSettings
34
- });
35
- debug(`We got an array of records: ${Array.isArray(recordB)}`);
36
- const recordsB = Array.isArray(recordB) ? recordB : [recordB];
37
- const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];
38
- const detectionResults = recordsB.map((record, index) => actualDetection({
39
- featuresA,
40
- recordAExternal,
41
- record,
42
- recordExternal: recordsBExternal[index],
43
- index
44
- }));
45
-
46
- // if we got array of records, we return an array of result
47
- // if we got a singular record, we return a singular result
48
- return Array.isArray(recordB) ? detectionResults : detectionResults[0];
49
- function actualDetection({
50
- featuresA,
51
- record,
52
- recordExternal,
53
- index,
54
- localSettings
55
- }) {
56
- const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';
57
- const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';
58
- debug(`Actual detection for record ${index + 1} ${labelB}`);
59
- const featuresB = extractFeatures({
60
- record,
61
- recordExternal,
62
- localSettings
63
- });
64
- debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);
65
- debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);
66
- const featurePairs = generateFeaturePairs(featuresA, featuresB);
67
- const similarityVector = generateSimilarityVector(featurePairs);
68
- if (similarityVector.some(v => v >= minProbabilityQuantifier)) {
69
- const probability = calculateProbability(similarityVector);
70
- debug(`probability: ${probability} (Treshold: ${treshold})`);
71
- return returnResult({
72
- match: probability >= treshold,
73
- probability
74
- });
75
- }
76
- debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);
77
- return returnResult({
78
- match: false,
79
- probability: 0.0
80
- });
81
- }
82
- function extractFeatures({
83
- record,
84
- recordExternal
85
- }) {
86
- return strategy.reduce((acc, {
87
- name,
88
- extract
89
- }) => acc.concat({
90
- name,
91
- value: extract({
92
- record,
93
- recordExternal
94
- })
95
- }), []);
96
- }
97
- function returnResult(result) {
98
- if (returnStrategy) {
99
- debug(`Returning detection strategy with the result`);
100
- const resultWithStrategy = {
101
- match: result.match,
102
- probability: result.probability,
103
- strategy: formatStrategy(strategy),
104
- treshold
105
- };
106
- debugData(`${JSON.stringify(resultWithStrategy)}`);
107
- return resultWithStrategy;
108
- }
109
- return result;
110
- }
111
- function formatStrategy(strategy) {
112
- const strategyNames = strategy.map(element => element.name);
113
- return strategyNames || [];
114
- }
115
- function calculateProbability(similarityVector) {
116
- const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);
117
- return probability > 1.0 ? 1.0 : probability;
118
- }
119
- function generateSimilarityVector(featurePairs) {
120
- const compared = featurePairs.map(({
121
- name,
122
- a,
123
- b
124
- }) => {
125
- const {
126
- compare
127
- } = strategy.find(({
128
- name: featureName
129
- }) => name === featureName);
130
- const points = compare(a, b);
131
- return {
132
- name,
133
- points
134
- };
135
- });
136
- debugData(`Points: ${JSON.stringify(compared)}`);
137
- return compared.map(({
138
- points
139
- }) => points);
140
- }
141
- function generateFeaturePairs(featuresA, featuresB) {
142
- const pairs = generatePairs();
143
- const missingFeatures = findMissing();
144
- debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);
145
- return pairs;
146
- function generatePairs() {
147
- return featuresA.reduce((acc, {
148
- name,
149
- value
150
- }, index) => acc.concat({
151
- name,
152
- a: value,
153
- b: featuresB[index].value
154
- }), []).filter(({
155
- a,
156
- b
157
- }) => {
158
- if (a.length === 0 || b.length === 0) {
159
- return false;
160
- }
161
- return true;
162
- });
163
- }
164
- function findMissing() {
165
- return featuresA.map(({
166
- name
167
- }) => name).filter(v => pairs.some(({
168
- name
169
- }) => name === v) === false);
170
- }
171
- }
172
- };
173
- exports.default = _default;
174
- //# sourceMappingURL=index.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"index.js","names":["_debug","_interopRequireDefault","require","features","_interopRequireWildcard","exports","_getRequireWildcardCache","e","WeakMap","r","t","__esModule","default","has","get","n","__proto__","a","Object","defineProperty","getOwnPropertyDescriptor","u","hasOwnProperty","call","i","set","_default","strategy","treshold","returnStrategy","localSettings","recordA","recordB","recordAExternal","recordBExternal","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","record","recordExternal","Array","isArray","recordsB","recordsBExternal","detectionResults","map","index","actualDetection","labelA","label","labelB","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateProbability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["\nimport createDebugLogger from 'debug';\nimport * as features from './features';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false, localSettings = {}) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {\n const minProbabilityQuantifier = 0.5;\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n const debugData = debug.extend('data');\n\n debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n // We could add here labels for records if we didn't get external labels\n\n const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal, localSettings});\n\n debug(`We got an array of records: ${Array.isArray(recordB)}`);\n const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n // if we got array of records, we return an array of result\n // if we got a singular record, we return a singular result\n return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n function actualDetection({featuresA, record, recordExternal, index, localSettings}) {\n const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n debug(`Actual detection for record ${index + 1} ${labelB}`);\n const featuresB = extractFeatures({record, recordExternal, localSettings});\n\n debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n const featurePairs = generateFeaturePairs(featuresA, featuresB);\n const similarityVector = generateSimilarityVector(featurePairs);\n\n if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n const probability = calculateProbability(similarityVector);\n debug(`probability: ${probability} (Treshold: ${treshold})`);\n return returnResult({match: probability >= treshold, probability});\n }\n\n debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n return returnResult({match: false, probability: 0.0});\n }\n\n function extractFeatures({record, recordExternal}) {\n return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n }\n\n function returnResult(result) {\n if (returnStrategy) {\n debug(`Returning detection strategy with the result`);\n const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n debugData(`${JSON.stringify(resultWithStrategy)}`);\n return resultWithStrategy;\n }\n return result;\n }\n\n function formatStrategy(strategy) {\n const strategyNames = strategy.map(element => element.name);\n return strategyNames || [];\n }\n\n function calculateProbability(similarityVector) {\n const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n return probability > 1.0 ? 1.0 : probability;\n }\n\n function generateSimilarityVector(featurePairs) {\n const compared = featurePairs.map(({name, a, b}) => {\n const {compare} = strategy.find(({name: featureName}) => name === featureName);\n const points = compare(a, b);\n return {name, points};\n });\n\n debugData(`Points: ${JSON.stringify(compared)}`);\n return compared.map(({points}) => points);\n }\n\n function generateFeaturePairs(featuresA, featuresB) {\n const pairs = generatePairs();\n const missingFeatures = findMissing();\n\n debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n return pairs;\n\n function generatePairs() {\n return featuresA\n .reduce((acc, {name, value}, index) => acc.concat({\n name,\n a: value,\n b: featuresB[index].value\n }), [])\n .filter(({a, b}) => {\n if (a.length === 0 || b.length === 0) {\n return false;\n }\n\n return true;\n });\n }\n\n function findMissing() {\n return featuresA\n .map(({name}) => name)\n .filter(v => pairs.some(({name}) => name === v) === false);\n }\n }\n\n};\n"],"mappings":";;;;;;AACA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,QAAA,GAAAC,uBAAA,CAAAF,OAAA;AAAuCG,OAAA,CAAAF,QAAA,GAAAA,QAAA;AAAA,SAAAG,yBAAAC,CAAA,6BAAAC,OAAA,mBAAAC,CAAA,OAAAD,OAAA,IAAAE,CAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,CAAA,WAAAA,CAAA,GAAAG,CAAA,GAAAD,CAAA,KAAAF,CAAA;AAAA,SAAAH,wBAAAG,CAAA,EAAAE,CAAA,SAAAA,CAAA,IAAAF,CAAA,IAAAA,CAAA,CAAAI,UAAA,SAAAJ,CAAA,eAAAA,CAAA,uBAAAA,CAAA,yBAAAA,CAAA,WAAAK,OAAA,EAAAL,CAAA,QAAAG,CAAA,GAAAJ,wBAAA,CAAAG,CAAA,OAAAC,CAAA,IAAAA,CAAA,CAAAG,GAAA,CAAAN,CAAA,UAAAG,CAAA,CAAAI,GAAA,CAAAP,CAAA,OAAAQ,CAAA,KAAAC,SAAA,UAAAC,CAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,CAAA,IAAAd,CAAA,oBAAAc,CAAA,OAAAC,cAAA,CAAAC,IAAA,CAAAhB,CAAA,EAAAc,CAAA,SAAAG,CAAA,GAAAP,CAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAb,CAAA,EAAAc,CAAA,UAAAG,CAAA,KAAAA,CAAA,CAAAV,GAAA,IAAAU,CAAA,CAAAC,GAAA,IAAAP,MAAA,CAAAC,cAAA,CAAAJ,CAAA,EAAAM,CAAA,EAAAG,CAAA,IAAAT,CAAA,CAAAM,CAAA,IAAAd,CAAA,CAAAc,CAAA,YAAAN,CAAA,CAAAH,OAAA,GAAAL,CAAA,EAAAG,CAAA,IAAAA,CAAA,CAAAe,GAAA,CAAAlB,CAAA,EAAAQ,CAAA,GAAAA,CAAA;AAAA,SAAAd,uBAAAM,CAAA,WAAAA,CAAA,IAAAA,CAAA,CAAAI,UAAA,GAAAJ,CAAA,KAAAK,OAAA,EAAAL,CAAA;AAAA,IAAAmB,QAAA,GAIxBA,CAAC;EAACC,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,EAAEC,aAAa,GAAG,CAAC,CAAC,KAAK,CAAC;EAACC,OAAO;EAAEC,OAAO;EAAEC,eAAe,GAAG,CAAC,CAAC;EAAEC,eAAe,GAAG,CAAC;AAAC,CAAC,KAAK;EAC7J,MAAMC,wBAAwB,GAAG,GAAG;EAEpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,mDAAmD,CAAC;EACpF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtCD,SAAS,CAAC,aAAaE,IAAI,CAACC,SAAS,CAACd,QAAQ,CAAC,eAAea,IAAI,CAACC,SAAS,CAACb,QAAQ,CAAC,qBAAqBY,IAAI,CAACC,SAAS,CAACZ,cAAc,CAAC,EAAE,CAAC;EAC5IS,SAAS,CAAC,eAAeP,OAAO,QAAQC,OAAO,EAAE,CAAC;EAClDI,KAAK,CAAC,iBAAiBI,IAAI,CAACC,SAAS,CAACR,eAAe,CAAC,QAAQO,IAAI,CAACC,SAAS,CAACP,eAAe,CAAC,EAAE,CAAC;EAChG;;EAEA,MAAMQ,SAAS,GAAGC,eAAe,CAAC;IAACC,MAAM,EAAEb,OAAO;IAAEc,cAAc,EAAEZ,eAAe;IAAEH;EAAa,CAAC,CAAC;EAEpGM,KAAK,CAAC,+BAA+BU,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,EAAE,CAAC;EAC9D,MAAMgB,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMiB,gBAAgB,GAAGH,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGE,eAAe,GAAG,CAACA,eAAe,CAAC;EAErF,MAAMgB,gBAAgB,GAAGF,QAAQ,CAACG,GAAG,CAAC,CAACP,MAAM,EAAEQ,KAAK,KAAKC,eAAe,CAAC;IAACX,SAAS;IAAET,eAAe;IAAEW,MAAM;IAAEC,cAAc,EAAEI,gBAAgB,CAACG,KAAK,CAAC;IAAEA;EAAK,CAAC,CAAC,CAAC;;EAE/J;EACA;EACA,OAAON,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGkB,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAeA,CAAC;IAACX,SAAS;IAAEE,MAAM;IAAEC,cAAc;IAAEO,KAAK;IAAEtB;EAAa,CAAC,EAAE;IAClF,MAAMwB,MAAM,GAAGrB,eAAe,IAAIA,eAAe,CAACsB,KAAK,GAAGtB,eAAe,CAACsB,KAAK,GAAG,GAAG;IACrF,MAAMC,MAAM,GAAGX,cAAc,IAAIA,cAAc,CAACU,KAAK,GAAGV,cAAc,CAACU,KAAK,GAAG,GAAG;IAGlFnB,KAAK,CAAC,+BAA+BgB,KAAK,GAAG,CAAC,IAAII,MAAM,EAAE,CAAC;IAC3D,MAAMC,SAAS,GAAGd,eAAe,CAAC;MAACC,MAAM;MAAEC,cAAc;MAAEf;IAAa,CAAC,CAAC;IAE1EQ,SAAS,CAAC,gBAAgBgB,MAAM,MAAMd,IAAI,CAACC,SAAS,CAACC,SAAS,CAAC,EAAE,CAAC;IAClEJ,SAAS,CAAC,gBAAgBkB,MAAM,MAAMhB,IAAI,CAACC,SAAS,CAACgB,SAAS,CAAC,EAAE,CAAC;IAElE,MAAMC,YAAY,GAAGC,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAI5B,wBAAwB,CAAC,EAAE;MAC7D,MAAM6B,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DxB,KAAK,CAAC,gBAAgB4B,WAAW,eAAepC,QAAQ,GAAG,CAAC;MAC5D,OAAOsC,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAIpC,QAAQ;QAAEoC;MAAW,CAAC,CAAC;IACpE;IAEA1B,SAAS,CAAC,4DAA4DH,wBAAwB,GAAG,CAAC;IAClG,OAAO+B,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASrB,eAAeA,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,EAAE;IACjD,OAAOlB,QAAQ,CAACyC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAC;QAAC3B,MAAM;QAAEC;MAAc,CAAC;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EACpH;EAEA,SAASqB,YAAYA,CAACQ,MAAM,EAAE;IAC5B,IAAI7C,cAAc,EAAE;MAClBO,KAAK,CAAC,8CAA8C,CAAC;MACrD,MAAMuC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAErC,QAAQ,EAAEiD,cAAc,CAACjD,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HU,SAAS,CAAC,GAAGE,IAAI,CAACC,SAAS,CAACkC,kBAAkB,CAAC,EAAE,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAcA,CAACjD,QAAQ,EAAE;IAChC,MAAMkD,aAAa,GAAGlD,QAAQ,CAACwB,GAAG,CAAC2B,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,oBAAoBA,CAACL,gBAAgB,EAAE;IAC9C,MAAMI,WAAW,GAAGJ,gBAAgB,CAACQ,MAAM,CAAC,CAACC,GAAG,EAAEN,CAAC,KAAKM,GAAG,GAAGN,CAAC,EAAE,GAAG,CAAC;IACrE,OAAOC,WAAW,GAAG,GAAG,GAAG,GAAG,GAAGA,WAAW;EAC9C;EAEA,SAASH,wBAAwBA,CAACH,YAAY,EAAE;IAC9C,MAAMqB,QAAQ,GAAGrB,YAAY,CAACP,GAAG,CAAC,CAAC;MAACmB,IAAI;MAAErD,CAAC;MAAE+D;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAGtD,QAAQ,CAACuD,IAAI,CAAC,CAAC;QAACZ,IAAI,EAAEa;MAAW,CAAC,KAAKb,IAAI,KAAKa,WAAW,CAAC;MAC9E,MAAMC,MAAM,GAAGH,OAAO,CAAChE,CAAC,EAAE+D,CAAC,CAAC;MAC5B,OAAO;QAACV,IAAI;QAAEc;MAAM,CAAC;IACvB,CAAC,CAAC;IAEF9C,SAAS,CAAC,WAAWE,IAAI,CAACC,SAAS,CAACsC,QAAQ,CAAC,EAAE,CAAC;IAChD,OAAOA,QAAQ,CAAC5B,GAAG,CAAC,CAAC;MAACiC;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAASzB,oBAAoBA,CAACjB,SAAS,EAAEe,SAAS,EAAE;IAClD,MAAM4B,KAAK,GAAGC,aAAa,CAAC,CAAC;IAC7B,MAAMC,eAAe,GAAGC,WAAW,CAAC,CAAC;IAErCpD,KAAK,CAAC,2FAA2FI,IAAI,CAACC,SAAS,CAAC8C,eAAe,CAAC,EAAE,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAaA,CAAA,EAAG;MACvB,OAAO5C,SAAS,CACb0B,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAErB,KAAK,KAAKiB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJrD,CAAC,EAAEwD,KAAK;QACRO,CAAC,EAAEvB,SAAS,CAACL,KAAK,CAAC,CAACqB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNgB,MAAM,CAAC,CAAC;QAACxE,CAAC;QAAE+D;MAAC,CAAC,KAAK;QAClB,IAAI/D,CAAC,CAACyE,MAAM,KAAK,CAAC,IAAIV,CAAC,CAACU,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASF,WAAWA,CAAA,EAAG;MACrB,OAAO9C,SAAS,CACbS,GAAG,CAAC,CAAC;QAACmB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBmB,MAAM,CAAC1B,CAAC,IAAIsB,KAAK,CAACvB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA1D,OAAA,CAAAO,OAAA,GAAAc,QAAA","ignoreList":[]}
@@ -1,63 +0,0 @@
1
- "use strict";
2
-
3
- var _fixugen = _interopRequireDefault(require("@natlibfi/fixugen"));
4
- var _fixura = require("@natlibfi/fixura");
5
- var _chai = require("chai");
6
- var _marcRecord = require("@natlibfi/marc-record");
7
- var features = _interopRequireWildcard(require("./features"));
8
- var _ = _interopRequireDefault(require("."));
9
- var _util = require("util");
10
- var _debug = _interopRequireDefault(require("debug"));
11
- function _getRequireWildcardCache(e) { if ("function" != typeof WeakMap) return null; var r = new WeakMap(), t = new WeakMap(); return (_getRequireWildcardCache = function (e) { return e ? t : r; })(e); }
12
- function _interopRequireWildcard(e, r) { if (!r && e && e.__esModule) return e; if (null === e || "object" != typeof e && "function" != typeof e) return { default: e }; var t = _getRequireWildcardCache(r); if (t && t.has(e)) return t.get(e); var n = { __proto__: null }, a = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var u in e) if ("default" !== u && {}.hasOwnProperty.call(e, u)) { var i = a ? Object.getOwnPropertyDescriptor(e, u) : null; i && (i.get || i.set) ? Object.defineProperty(n, u, i) : n[u] = e[u]; } return n.default = e, t && t.set(e, n), n; }
13
- function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
14
- const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:test');
15
- const debugData = debug.extend('data');
16
- describe('match-detection', () => {
17
- (0, _fixugen.default)({
18
- path: [__dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],
19
- useMetadataFile: true,
20
- recurse: false,
21
- fixura: {
22
- reader: _fixura.READERS.JSON
23
- },
24
- callback: ({
25
- getFixture,
26
- options,
27
- expectedResults,
28
- array,
29
- enabled = true
30
- }) => {
31
- if (!enabled) {
32
- debug(`*** DISABLED TEST! ***`);
33
- return;
34
- }
35
- const detect = (0, _.default)(formatOptions());
36
- const recordA = new _marcRecord.MarcRecord(getFixture('recordA.json'), {
37
- subfieldValues: false
38
- });
39
- debugData((0, _util.inspect)(recordA));
40
- debug(`Our recordB is an array of records: ${array}`);
41
- const recordB = array ? getFixture('recordB.json').map(recordJson => new _marcRecord.MarcRecord(recordJson, {
42
- subfieldValues: false
43
- })) : new _marcRecord.MarcRecord(getFixture('recordB.json'), {
44
- subfieldValues: false
45
- });
46
- debugData((0, _util.inspect)(recordB));
47
- const results = detect({
48
- recordA,
49
- recordB
50
- });
51
- debugData(`${JSON.stringify(results)}`);
52
- (0, _chai.expect)(results).to.eql(expectedResults);
53
- function formatOptions() {
54
- const contextFeatures = features[options.strategy.type];
55
- return {
56
- ...options,
57
- strategy: options.strategy.features.map(v => contextFeatures[v]())
58
- };
59
- }
60
- }
61
- });
62
- });
63
- //# sourceMappingURL=index.spec.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"index.spec.js","names":["_fixugen","_interopRequireDefault","require","_fixura","_chai","_marcRecord","features","_interopRequireWildcard","_","_util","_debug","_getRequireWildcardCache","e","WeakMap","r","t","__esModule","default","has","get","n","__proto__","a","Object","defineProperty","getOwnPropertyDescriptor","u","hasOwnProperty","call","i","set","debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","recurse","fixura","reader","READERS","JSON","callback","getFixture","options","expectedResults","array","enabled","detect","createDetectionInterface","formatOptions","recordA","MarcRecord","subfieldValues","inspect","recordB","map","recordJson","results","stringify","expect","to","eql","contextFeatures","strategy","type","v"],"sources":["../../src/match-detection/index.spec.js"],"sourcesContent":["\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './features';\nimport createDetectionInterface from '.';\nimport {inspect} from 'util';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:test');\nconst debugData = debug.extend('data');\n\ndescribe('match-detection', () => {\n generateTests({\n path: [__dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],\n useMetadataFile: true,\n recurse: false,\n fixura: {\n reader: READERS.JSON\n },\n callback: ({getFixture, options, expectedResults, array, enabled = true}) => {\n\n if (!enabled) {\n debug(`*** DISABLED TEST! ***`);\n return;\n }\n\n const detect = createDetectionInterface(formatOptions());\n const recordA = new MarcRecord(getFixture('recordA.json'), {subfieldValues: false});\n debugData(inspect(recordA));\n\n debug(`Our recordB is an array of records: ${array}`);\n const recordB = array\n ? getFixture('recordB.json').map(recordJson => new MarcRecord(recordJson, {subfieldValues: false}))\n : new MarcRecord(getFixture('recordB.json'), {subfieldValues: false});\n debugData(inspect(recordB));\n\n const results = detect({recordA, recordB});\n debugData(`${JSON.stringify(results)}`);\n\n expect(results).to.eql(expectedResults);\n\n function formatOptions() {\n const contextFeatures = features[options.strategy.type];\n\n return {\n ...options,\n strategy: options.strategy.features.map(v => contextFeatures[v]())\n };\n }\n }\n });\n});\n"],"mappings":";;AACA,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,CAAA,GAAAP,sBAAA,CAAAC,OAAA;AACA,IAAAO,KAAA,GAAAP,OAAA;AACA,IAAAQ,MAAA,GAAAT,sBAAA,CAAAC,OAAA;AAAsC,SAAAS,yBAAAC,CAAA,6BAAAC,OAAA,mBAAAC,CAAA,OAAAD,OAAA,IAAAE,CAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,CAAA,WAAAA,CAAA,GAAAG,CAAA,GAAAD,CAAA,KAAAF,CAAA;AAAA,SAAAL,wBAAAK,CAAA,EAAAE,CAAA,SAAAA,CAAA,IAAAF,CAAA,IAAAA,CAAA,CAAAI,UAAA,SAAAJ,CAAA,eAAAA,CAAA,uBAAAA,CAAA,yBAAAA,CAAA,WAAAK,OAAA,EAAAL,CAAA,QAAAG,CAAA,GAAAJ,wBAAA,CAAAG,CAAA,OAAAC,CAAA,IAAAA,CAAA,CAAAG,GAAA,CAAAN,CAAA,UAAAG,CAAA,CAAAI,GAAA,CAAAP,CAAA,OAAAQ,CAAA,KAAAC,SAAA,UAAAC,CAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,CAAA,IAAAd,CAAA,oBAAAc,CAAA,OAAAC,cAAA,CAAAC,IAAA,CAAAhB,CAAA,EAAAc,CAAA,SAAAG,CAAA,GAAAP,CAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAb,CAAA,EAAAc,CAAA,UAAAG,CAAA,KAAAA,CAAA,CAAAV,GAAA,IAAAU,CAAA,CAAAC,GAAA,IAAAP,MAAA,CAAAC,cAAA,CAAAJ,CAAA,EAAAM,CAAA,EAAAG,CAAA,IAAAT,CAAA,CAAAM,CAAA,IAAAd,CAAA,CAAAc,CAAA,YAAAN,CAAA,CAAAH,OAAA,GAAAL,CAAA,EAAAG,CAAA,IAAAA,CAAA,CAAAe,GAAA,CAAAlB,CAAA,EAAAQ,CAAA,GAAAA,CAAA;AAAA,SAAAnB,uBAAAW,CAAA,WAAAA,CAAA,IAAAA,CAAA,CAAAI,UAAA,GAAAJ,CAAA,KAAAK,OAAA,EAAAL,CAAA;AAEtC,MAAMmB,KAAK,GAAG,IAAAC,cAAiB,EAAC,wDAAwD,CAAC;AACzF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAEtCC,QAAQ,CAAC,iBAAiB,EAAE,MAAM;EAChC,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,OAAO,CAAC;IAC1EC,eAAe,EAAE,IAAI;IACrBC,OAAO,EAAE,KAAK;IACdC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IACDC,QAAQ,EAAEA,CAAC;MAACC,UAAU;MAAEC,OAAO;MAAEC,eAAe;MAAEC,KAAK;MAAEC,OAAO,GAAG;IAAI,CAAC,KAAK;MAE3E,IAAI,CAACA,OAAO,EAAE;QACZnB,KAAK,CAAC,wBAAwB,CAAC;QAC/B;MACF;MAEA,MAAMoB,MAAM,GAAG,IAAAC,SAAwB,EAACC,aAAa,CAAC,CAAC,CAAC;MACxD,MAAMC,OAAO,GAAG,IAAIC,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACnFvB,SAAS,CAAC,IAAAwB,aAAO,EAACH,OAAO,CAAC,CAAC;MAE3BvB,KAAK,CAAC,uCAAuCkB,KAAK,EAAE,CAAC;MACrD,MAAMS,OAAO,GAAGT,KAAK,GACjBH,UAAU,CAAC,cAAc,CAAC,CAACa,GAAG,CAACC,UAAU,IAAI,IAAIL,sBAAU,CAACK,UAAU,EAAE;QAACJ,cAAc,EAAE;MAAK,CAAC,CAAC,CAAC,GACjG,IAAID,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACvEvB,SAAS,CAAC,IAAAwB,aAAO,EAACC,OAAO,CAAC,CAAC;MAE3B,MAAMG,OAAO,GAAGV,MAAM,CAAC;QAACG,OAAO;QAAEI;MAAO,CAAC,CAAC;MAC1CzB,SAAS,CAAC,GAAGW,IAAI,CAACkB,SAAS,CAACD,OAAO,CAAC,EAAE,CAAC;MAEvC,IAAAE,YAAM,EAACF,OAAO,CAAC,CAACG,EAAE,CAACC,GAAG,CAACjB,eAAe,CAAC;MAEvC,SAASK,aAAaA,CAAA,EAAG;QACvB,MAAMa,eAAe,GAAG5D,QAAQ,CAACyC,OAAO,CAACoB,QAAQ,CAACC,IAAI,CAAC;QAEvD,OAAO;UACL,GAAGrB,OAAO;UACVoB,QAAQ,EAAEpB,OAAO,CAACoB,QAAQ,CAAC7D,QAAQ,CAACqD,GAAG,CAACU,CAAC,IAAIH,eAAe,CAACG,CAAC,CAAC,CAAC,CAAC;QACnE,CAAC;MACH;IACF;EACF,CAAC,CAAC;AACJ,CAAC,CAAC","ignoreList":[]}
@@ -1,115 +0,0 @@
1
- "use strict";
2
-
3
- Object.defineProperty(exports, "__esModule", {
4
- value: true
5
- });
6
- exports.extractSubfieldsFromField = extractSubfieldsFromField;
7
- exports.getMatchCounts = getMatchCounts;
8
- exports.getMelindaIdsF035 = getMelindaIdsF035;
9
- exports.getSubfieldValues = getSubfieldValues;
10
- exports.testStringOrNumber = testStringOrNumber;
11
- exports.uniqueSubfields = uniqueSubfields;
12
- exports.validateSidFieldSubfieldCounts = validateSidFieldSubfieldCounts;
13
- var _debug = _interopRequireDefault(require("debug"));
14
- function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
15
- const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:utils');
16
- const debugData = debug.extend('data');
17
- function getMelindaIdsF035(record) {
18
- const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:melinda-id');
19
- const debugData = debug.extend('data');
20
- const f035s = record.getFields('035');
21
- if (f035s.length < 1) {
22
- debug(`No f035s found.`);
23
- return [];
24
- }
25
- const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));
26
- const f035MelindaIds = [...new Set(allF035MelindaIds)];
27
- debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);
28
- debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);
29
- debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);
30
- return f035MelindaIds;
31
- function toMelindaIds({
32
- subfields
33
- }) {
34
- const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
35
- return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({
36
- value
37
- }) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');
38
- }
39
- }
40
- function validateSidFieldSubfieldCounts(field) {
41
- // Valid SID-fields have just one $c and one $b
42
- debugData(`Validating SID field ${JSON.stringify(field)}`);
43
- const countC = countSubfields(field, 'c');
44
- const countB = countSubfields(field, 'b');
45
- debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);
46
- return countC === 1 && countB === 1;
47
- }
48
- function countSubfields(field, subfieldCode) {
49
- // debug(`Counting subfields ${subfieldCode}`);
50
- return field.subfields.filter(({
51
- code
52
- }) => code === subfieldCode).length;
53
- }
54
- function getSubfieldValues(field, subfieldCode) {
55
- debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);
56
- return field.subfields.filter(({
57
- code
58
- }) => code === subfieldCode).map(({
59
- value
60
- }) => testStringOrNumber(value) ? String(value) : '').filter(value => value);
61
- }
62
- function testStringOrNumber(value) {
63
- if (typeof value === 'string' || typeof value === 'number') {
64
- return true;
65
- }
66
- return false;
67
- }
68
- function extractSubfieldsFromField(field, subfieldCodes) {
69
- if (field === undefined || field.subfields === undefined) {
70
- return [];
71
- }
72
- const resultSubfields = field.subfields.filter(({
73
- code
74
- }) => subfieldCodes.includes(code)).map(({
75
- code,
76
- value
77
- }) => ({
78
- code,
79
- value: testStringOrNumber(value) ? String(value) : ''
80
- })).filter(value => value);
81
- return resultSubfields;
82
- }
83
- function uniqueSubfields(subfields) {
84
- return subfields.reduce((arr, e) => {
85
- if (!arr.find(item => item.code === e.code && item.value === e.value)) {
86
- const newArr = arr.concat(e);
87
- return newArr;
88
- }
89
- return arr;
90
- }, []);
91
- }
92
- function getMatchCounts(aValues, bValues) {
93
- const matchingValues = getMatchingValuesAmount(aValues, bValues);
94
- return {
95
- maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
96
- // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
97
- possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
98
- matchingValues
99
- };
100
- function getMatchingValuesAmount(aValues, bValues) {
101
- if (bValues.length > aValues.length) {
102
- return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
103
- }
104
- if (aValues.length > bValues.length) {
105
- return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
106
- }
107
-
108
- // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases
109
- // there would be duplicate values
110
- const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
111
- const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
112
- return aToB < bToA ? aToB : bToA;
113
- }
114
- }
115
- //# sourceMappingURL=matching-utils.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"matching-utils.js","names":["_debug","_interopRequireDefault","require","e","__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","find","item","newArr","getMatchCounts","aValues","bValues","matchingValues","getMatchingValuesAmount","maxValues","possibleMatchValues","aValue","some","bValue","aToB","bToA"],"sources":["../src/matching-utils.js"],"sourcesContent":["import 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":";;;;;;;;;;;;AAAA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AAAsC,SAAAD,uBAAAE,CAAA,WAAAA,CAAA,IAAAA,CAAA,CAAAC,UAAA,GAAAD,CAAA,KAAAE,OAAA,EAAAF,CAAA;AAEtC,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,CAAC,iBAAiB,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,CAAC,WAAWI,KAAK,CAACE,MAAM,MAAMO,IAAI,CAACC,SAAS,CAACV,KAAK,CAAC,EAAE,CAAC;EAC/DJ,SAAS,CAAC,QAAQO,iBAAiB,CAACD,MAAM,MAAMO,IAAI,CAACC,SAAS,CAACP,iBAAiB,CAAC,EAAE,CAAC;EACpFP,SAAS,CAAC,eAAeW,cAAc,CAACL,MAAM,MAAMO,IAAI,CAACC,SAAS,CAACH,cAAc,CAAC,EAAE,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,CAAC,wBAAwBa,IAAI,CAACC,SAAS,CAACa,KAAK,CAAC,EAAE,CAAC;EAC1D,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC7B,KAAK,CAAC,SAAS8B,MAAM,eAAeE,MAAM,qBAAqBF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAC,EAAE,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,CAAC,oBAAoB+B,YAAY,SAASlB,IAAI,CAACC,SAAS,CAACa,KAAK,CAAC,EAAE,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,EAAE5C,CAAC,KAAK;IAClC,IAAI,CAAC4C,GAAG,CAACC,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACrB,IAAI,KAAKzB,CAAC,CAACyB,IAAI,IAAIqB,IAAI,CAACnB,KAAK,KAAK3B,CAAC,CAAC2B,KAAK,CAAC,EAAE;MACrE,MAAMoB,MAAM,GAAGH,GAAG,CAAC/B,MAAM,CAACb,CAAC,CAAC;MAC5B,OAAO+C,MAAM;IACf;IACA,OAAOH,GAAG;EACZ,CAAC,EAAE,EAAE,CAAC;AACR;AAEO,SAASI,cAAcA,CAACC,OAAO,EAAEC,OAAO,EAAE;EAE/C,MAAMC,cAAc,GAAGC,uBAAuB,CAACH,OAAO,EAAEC,OAAO,CAAC;EAEhE,OAAO;IACLG,SAAS,EAAEJ,OAAO,CAACtC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,GAAGsC,OAAO,CAACtC,MAAM,GAAGuC,OAAO,CAACvC,MAAM;IAC5E;IACA2C,mBAAmB,EAAEL,OAAO,CAACtC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,GAAGsC,OAAO,CAACtC,MAAM;IACtFwC;EACF,CAAC;EAED,SAASC,uBAAuBA,CAACH,OAAO,EAAEC,OAAO,EAAE;IACjD,IAAIA,OAAO,CAACvC,MAAM,GAAGsC,OAAO,CAACtC,MAAM,EAAE;MACnC,OAAOsC,OAAO,CAAC3B,MAAM,CAACiC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC9C,MAAM;IACnF;IACA,IAAIsC,OAAO,CAACtC,MAAM,GAAGuC,OAAO,CAACvC,MAAM,EAAE;MACnC,OAAOuC,OAAO,CAAC5B,MAAM,CAACmC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC5C,MAAM;IACnF;;IAEA;IACA;IACA,MAAM+C,IAAI,GAAGT,OAAO,CAAC3B,MAAM,CAACiC,MAAM,IAAIL,OAAO,CAACM,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAAC9C,MAAM;IACvF,MAAMgD,IAAI,GAAGT,OAAO,CAAC5B,MAAM,CAACmC,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAAC5C,MAAM;IAEvF,OAAO+C,IAAI,GAAGC,IAAI,GAAGD,IAAI,GAAGC,IAAI;EAClC;AACF","ignoreList":[]}