@natlibfi/melinda-record-matching 4.4.0-alpha.1 → 5.0.0-alpha.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.github/workflows/melinda-node-tests.yml +13 -9
- package/LICENSE.txt +21 -165
- package/README.md +2 -2
- package/dist/candidate-search/candidate-search-utils.js +6 -43
- package/dist/candidate-search/candidate-search-utils.js.map +7 -1
- package/dist/candidate-search/choose-queries.js +75 -0
- package/dist/candidate-search/choose-queries.js.map +7 -0
- package/dist/candidate-search/index.js +89 -225
- package/dist/candidate-search/index.js.map +7 -1
- package/dist/candidate-search/index.test.js +92 -0
- package/dist/candidate-search/index.test.js.map +7 -0
- package/dist/candidate-search/query-list/bib.js +141 -320
- package/dist/candidate-search/query-list/bib.js.map +7 -1
- package/dist/candidate-search/query-list/bib.test.js +34 -0
- package/dist/candidate-search/query-list/bib.test.js.map +7 -0
- package/dist/candidate-search/query-list/component.js +87 -0
- package/dist/candidate-search/query-list/component.js.map +7 -0
- package/dist/candidate-search/query-list/index.js +39 -52
- package/dist/candidate-search/query-list/index.js.map +7 -1
- package/dist/cli.js +138 -0
- package/dist/cli.js.map +7 -0
- package/dist/index.js +87 -462
- package/dist/index.js.map +7 -1
- package/dist/index.test.js +69 -0
- package/dist/index.test.js.map +7 -0
- package/dist/match-detection/features/bib/all-source-ids.js +20 -91
- package/dist/match-detection/features/bib/all-source-ids.js.map +7 -1
- package/dist/match-detection/features/bib/authors.js +20 -76
- package/dist/match-detection/features/bib/authors.js.map +7 -1
- package/dist/match-detection/features/bib/bibliographic-level.js +4 -40
- package/dist/match-detection/features/bib/bibliographic-level.js.map +7 -1
- package/dist/match-detection/features/bib/host-component.js +5 -41
- package/dist/match-detection/features/bib/host-component.js.map +7 -1
- package/dist/match-detection/features/bib/index.js +17 -119
- package/dist/match-detection/features/bib/index.js.map +7 -1
- package/dist/match-detection/features/bib/index.test.js +40 -0
- package/dist/match-detection/features/bib/index.test.js.map +7 -0
- package/dist/match-detection/features/bib/isbn.js +13 -69
- package/dist/match-detection/features/bib/isbn.js.map +7 -1
- package/dist/match-detection/features/bib/issn.js +5 -50
- package/dist/match-detection/features/bib/issn.js.map +7 -1
- package/dist/match-detection/features/bib/language.js +25 -86
- package/dist/match-detection/features/bib/language.js.map +7 -1
- package/dist/match-detection/features/bib/media-type.js +12 -64
- package/dist/match-detection/features/bib/media-type.js.map +7 -1
- package/dist/match-detection/features/bib/melinda-id.js +5 -47
- package/dist/match-detection/features/bib/melinda-id.js.map +7 -1
- package/dist/match-detection/features/bib/melinda-identifier-factory.js +16 -67
- package/dist/match-detection/features/bib/melinda-identifier-factory.js.map +7 -1
- package/dist/match-detection/features/bib/other-standard-identifier.js +5 -51
- package/dist/match-detection/features/bib/other-standard-identifier.js.map +7 -1
- package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js +49 -189
- package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js.map +7 -1
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js +10 -53
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js.map +7 -1
- package/dist/match-detection/features/bib/publication-time.js +8 -46
- package/dist/match-detection/features/bib/publication-time.js.map +7 -1
- package/dist/match-detection/features/bib/record-type.js +4 -43
- package/dist/match-detection/features/bib/record-type.js.map +7 -1
- package/dist/match-detection/features/bib/standard-identifier-factory.js +47 -163
- package/dist/match-detection/features/bib/standard-identifier-factory.js.map +7 -1
- package/dist/match-detection/features/bib/title-version-original.js +11 -53
- package/dist/match-detection/features/bib/title-version-original.js.map +7 -1
- package/dist/match-detection/features/bib/title.js +20 -72
- package/dist/match-detection/features/bib/title.js.map +7 -1
- package/dist/match-detection/features/index.js +3 -11
- package/dist/match-detection/features/index.js.map +7 -1
- package/dist/match-detection/index.js +35 -150
- package/dist/match-detection/index.js.map +7 -1
- package/dist/match-detection/index.test.js +44 -0
- package/dist/match-detection/index.test.js.map +7 -0
- package/dist/matching-utils.js +38 -102
- package/dist/matching-utils.js.map +7 -1
- package/eslint.config.js +52 -0
- package/example.env +4 -0
- package/package.json +29 -78
- package/src/candidate-search/candidate-search-utils.js +1 -28
- package/src/candidate-search/choose-queries.js +93 -0
- package/src/candidate-search/index.js +85 -119
- package/src/candidate-search/{index.spec.js → index.test.js} +30 -43
- package/src/candidate-search/query-list/bib.js +54 -55
- package/src/candidate-search/query-list/bib.test.js +38 -0
- package/src/candidate-search/query-list/component.js +127 -0
- package/src/candidate-search/query-list/index.js +31 -35
- package/src/cli.js +177 -0
- package/src/index.js +27 -68
- package/src/{index.spec.js → index.test.js} +11 -42
- package/src/match-detection/features/bib/all-source-ids.js +2 -29
- package/src/match-detection/features/bib/authors.js +5 -30
- package/src/match-detection/features/bib/bibliographic-level.js +0 -27
- package/src/match-detection/features/bib/host-component.js +0 -27
- package/src/match-detection/features/bib/index.js +16 -43
- package/src/match-detection/features/bib/index.test.js +52 -0
- package/src/match-detection/features/bib/isbn.js +2 -29
- package/src/match-detection/features/bib/issn.js +1 -28
- package/src/match-detection/features/bib/language.js +5 -32
- package/src/match-detection/features/bib/media-type.js +2 -28
- package/src/match-detection/features/bib/melinda-id.js +1 -28
- package/src/match-detection/features/bib/melinda-identifier-factory.js +5 -32
- package/src/match-detection/features/bib/other-standard-identifier.js +1 -28
- package/src/match-detection/features/bib/publication-time-allow-cons-years-multi.js +7 -34
- package/src/match-detection/features/bib/publication-time-allow-cons-years.js +1 -28
- package/src/match-detection/features/bib/publication-time.js +2 -28
- package/src/match-detection/features/bib/record-type.js +0 -27
- package/src/match-detection/features/bib/standard-identifier-factory.js +1 -28
- package/src/match-detection/features/bib/title-version-original.js +2 -28
- package/src/match-detection/features/bib/title.js +6 -34
- package/src/match-detection/features/index.js +1 -28
- package/src/match-detection/index.js +1 -28
- package/src/match-detection/{index.spec.js → index.test.js} +9 -35
- package/src/matching-utils.js +8 -37
- package/dist/candidate-search/index.spec.js +0 -144
- package/dist/candidate-search/index.spec.js.map +0 -1
- package/dist/candidate-search/query-list/bib.spec.js +0 -64
- package/dist/candidate-search/query-list/bib.spec.js.map +0 -1
- package/dist/index.spec.js +0 -127
- package/dist/index.spec.js.map +0 -1
- package/dist/match-detection/features/bib/index.spec.js +0 -93
- package/dist/match-detection/features/bib/index.spec.js.map +0 -1
- package/dist/match-detection/index.spec.js +0 -91
- package/dist/match-detection/index.spec.js.map +0 -1
- package/src/candidate-search/query-list/bib.spec.js +0 -54
- package/src/match-detection/features/bib/index.spec.js +0 -78
|
@@ -1,171 +1,77 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
})
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
var _debug = _interopRequireDefault(require("debug"));
|
|
9
|
-
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
10
|
-
/**
|
|
11
|
-
*
|
|
12
|
-
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
13
|
-
*
|
|
14
|
-
* Melinda record matching modules for Javascript
|
|
15
|
-
*
|
|
16
|
-
* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)
|
|
17
|
-
*
|
|
18
|
-
* This file is part of melinda-record-matching-js
|
|
19
|
-
*
|
|
20
|
-
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
21
|
-
* it under the terms of the GNU Lesser General Public License as
|
|
22
|
-
* published by the Free Software Foundation, either version 3 of the
|
|
23
|
-
* License, or (at your option) any later version.
|
|
24
|
-
*
|
|
25
|
-
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
26
|
-
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
27
|
-
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
28
|
-
* GNU Lesser General Public License for more details.
|
|
29
|
-
*
|
|
30
|
-
* You should have received a copy of the GNU Affero General Public License
|
|
31
|
-
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
32
|
-
*
|
|
33
|
-
* @licend The above is the entire license notice
|
|
34
|
-
* for the JavaScript code in this file.
|
|
35
|
-
*
|
|
36
|
-
*/
|
|
37
|
-
// We should also get copyright time and copyright/publication times from 26x
|
|
38
|
-
// We should also get publishing time type from f008
|
|
39
|
-
// We should get reprint times from f500 $a "Lisäpainos/Lisäpainokset:"
|
|
40
|
-
var _default = () => ({
|
|
41
|
-
name: 'Publication time, allow consequent years, years from multiple sources',
|
|
42
|
-
extract: ({
|
|
43
|
-
record,
|
|
44
|
-
recordExternal
|
|
45
|
-
}) => {
|
|
46
|
-
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');
|
|
47
|
-
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
1
|
+
import createDebugLogger from "debug";
|
|
2
|
+
import { testStringOrNumber } from "../../../matching-utils.js";
|
|
3
|
+
export default () => ({
|
|
4
|
+
name: "Publication time, allow consequent years, years from multiple sources",
|
|
5
|
+
extract: ({ record, recordExternal }) => {
|
|
6
|
+
const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi");
|
|
7
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : "record";
|
|
48
8
|
const f008Values = extractF008Values(record);
|
|
49
|
-
debug(`${label} f008: ${JSON.stringify(f008Values)}`);
|
|
9
|
+
debug(`${label}: f008: ${JSON.stringify(f008Values)}`);
|
|
50
10
|
const f26xValues = extractF26xValues(record);
|
|
51
|
-
debug(`${label} f26x: ${JSON.stringify(f26xValues)}`);
|
|
11
|
+
debug(`${label}: f26x: ${JSON.stringify(f26xValues)}`);
|
|
52
12
|
const f500Values = extractF500Years(record);
|
|
53
|
-
debug(`${label} f500: ${JSON.stringify(f500Values)}`);
|
|
54
|
-
|
|
55
|
-
// We should get copyrightYear from f008Date2 to copyrightYears when f008YearType = 'r'
|
|
56
|
-
// Is the original year (f008Date2) in f008YearType === 'r' comparable to copyrightYear?
|
|
57
|
-
// We should handle unknown years (here or comparison?)
|
|
58
|
-
// We should handle year ranges for continuing resources / collections
|
|
59
|
-
|
|
60
|
-
const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter(value => value && value !== ' ' && value !== '||||'))].sort();
|
|
13
|
+
debug(`${label}: f500: ${JSON.stringify(f500Values)}`);
|
|
14
|
+
const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter((value) => value && value !== " " && value !== "||||"))].sort();
|
|
61
15
|
const copyrightYears = [...new Set(f26xValues.copyrightYears)].sort();
|
|
62
16
|
const reprintYears = [...new Set(f500Values)].sort();
|
|
63
|
-
const combined = {
|
|
64
|
-
normalYears,
|
|
65
|
-
copyrightYears,
|
|
66
|
-
reprintYears
|
|
67
|
-
};
|
|
17
|
+
const combined = { normalYears, copyrightYears, reprintYears };
|
|
68
18
|
debug(`Combined: ${JSON.stringify(combined)}`);
|
|
69
19
|
return combined;
|
|
70
|
-
function extractF008Values(
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
if (value && (0, _matchingUtils.testStringOrNumber)(value)) {
|
|
20
|
+
function extractF008Values(record2) {
|
|
21
|
+
const value = record2.get(/^008$/u)?.[0]?.value || void 0;
|
|
22
|
+
if (value && testStringOrNumber(value)) {
|
|
74
23
|
const f008Date1 = extractF008Date1(value);
|
|
75
24
|
const f008Date2 = extractF008Date2(value);
|
|
76
25
|
const f008YearType = extractF008YearType(value);
|
|
77
|
-
return {
|
|
78
|
-
f008Date1,
|
|
79
|
-
f008Date2,
|
|
80
|
-
f008YearType
|
|
81
|
-
};
|
|
26
|
+
return { f008Date1, f008Date2, f008YearType };
|
|
82
27
|
}
|
|
83
|
-
return {
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
f008YearType: undefined
|
|
87
|
-
};
|
|
88
|
-
function extractF008Date1(value) {
|
|
89
|
-
return String(value).slice(7, 11);
|
|
28
|
+
return { f008Date1: void 0, f008Date2: void 0, f008YearType: void 0 };
|
|
29
|
+
function extractF008Date1(value2) {
|
|
30
|
+
return String(value2).slice(7, 11);
|
|
90
31
|
}
|
|
91
|
-
function extractF008Date2(
|
|
92
|
-
return String(
|
|
32
|
+
function extractF008Date2(value2) {
|
|
33
|
+
return String(value2).slice(11, 15);
|
|
93
34
|
}
|
|
94
|
-
function extractF008YearType(
|
|
95
|
-
return String(
|
|
35
|
+
function extractF008YearType(value2) {
|
|
36
|
+
return String(value2).slice(6, 7);
|
|
96
37
|
}
|
|
97
38
|
}
|
|
98
|
-
function extractF26xValues(
|
|
39
|
+
function extractF26xValues(record2) {
|
|
99
40
|
const copyrightRegex = /^(?<copyrightPrefix>cop|cop.|c|©|p|℗)/u;
|
|
100
|
-
const pubNormalSubFieldValues = extractPubNormalSubfieldValues(
|
|
41
|
+
const pubNormalSubFieldValues = extractPubNormalSubfieldValues(record2, copyrightRegex);
|
|
101
42
|
debug(`Normal years: ${JSON.stringify(pubNormalSubFieldValues)}`);
|
|
102
|
-
const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(
|
|
43
|
+
const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(record2);
|
|
103
44
|
debug(`F264 copyright years: ${JSON.stringify(pubF264CopySubFieldValues)}`);
|
|
104
|
-
const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(
|
|
45
|
+
const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(record2, copyrightRegex);
|
|
105
46
|
debug(`F260 copyright years: ${JSON.stringify(pubF260CopySubFieldValues)}`);
|
|
106
|
-
return {
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
};
|
|
110
|
-
function extractPubNormalSubfieldValues(record, copyrightRegex) {
|
|
111
|
-
return record.get(/^26[04]$/u).filter(field => !(field.tag === '264' && field.ind2 === '4')).map(({
|
|
112
|
-
subfields
|
|
113
|
-
}) => subfields).flat().filter(({
|
|
114
|
-
code
|
|
115
|
-
}) => code && code === 'c').filter(({
|
|
116
|
-
value
|
|
117
|
-
}) => value && !copyrightRegex.test(value)).map(({
|
|
118
|
-
value
|
|
119
|
-
}) => value).map(value => removeNonAlphaNumeric(value));
|
|
47
|
+
return { normalYears: pubNormalSubFieldValues, copyrightYears: [...pubF260CopySubFieldValues, ...pubF264CopySubFieldValues] };
|
|
48
|
+
function extractPubNormalSubfieldValues(record3, copyrightRegex2) {
|
|
49
|
+
return record3.get(/^26[04]$/u).filter((field) => !(field.tag === "264" && field.ind2 === "4")).map(({ subfields }) => subfields).flat().filter(({ code }) => code && code === "c").filter(({ value }) => value && !copyrightRegex2.test(value)).map(({ value }) => value).map((value) => removeNonAlphaNumeric(value));
|
|
120
50
|
}
|
|
121
|
-
function extractPubF264CopySubfieldValues(
|
|
122
|
-
return
|
|
123
|
-
subfields
|
|
124
|
-
}) => subfields).flat().filter(({
|
|
125
|
-
code
|
|
126
|
-
}) => code && code === 'c').filter(({
|
|
127
|
-
value
|
|
128
|
-
}) => value).map(({
|
|
129
|
-
value
|
|
130
|
-
}) => value).map(value => value.replace(copyrightRegex, '')).map(value => removeNonAlphaNumeric(value));
|
|
51
|
+
function extractPubF264CopySubfieldValues(record3, copyrightRegex2) {
|
|
52
|
+
return record3.get(/^264$/u).filter((field) => field.ind2 === "4").map(({ subfields }) => subfields).flat().filter(({ code }) => code && code === "c").filter(({ value }) => value).map(({ value }) => value).map((value) => value.replace(copyrightRegex2, "")).map((value) => removeNonAlphaNumeric(value));
|
|
131
53
|
}
|
|
132
|
-
function extractPubF260CopySubfieldValues(
|
|
133
|
-
return
|
|
134
|
-
subfields
|
|
135
|
-
}) => subfields).flat().filter(({
|
|
136
|
-
code
|
|
137
|
-
}) => code && code === 'c').filter(({
|
|
138
|
-
value
|
|
139
|
-
}) => value && copyrightRegex.test(value)).map(({
|
|
140
|
-
value
|
|
141
|
-
}) => value).map(value => value.replace(copyrightRegex, '')).map(value => removeNonAlphaNumeric(value));
|
|
54
|
+
function extractPubF260CopySubfieldValues(record3, copyrightRegex2) {
|
|
55
|
+
return record3.get(/^260$/u).map(({ subfields }) => subfields).flat().filter(({ code }) => code && code === "c").filter(({ value }) => value && copyrightRegex2.test(value)).map(({ value }) => value).map((value) => value.replace(copyrightRegex2, "")).map((value) => removeNonAlphaNumeric(value));
|
|
142
56
|
}
|
|
143
57
|
function removeNonAlphaNumeric(value) {
|
|
144
58
|
debug(`Cleaning: ${JSON.stringify(value)}`);
|
|
145
59
|
const nonAlphaNumericRegex = /[^A-Za-z0-9]/ug;
|
|
146
|
-
return value ? value.replace(nonAlphaNumericRegex,
|
|
60
|
+
return value ? value.replace(nonAlphaNumericRegex, "") : value;
|
|
147
61
|
}
|
|
148
62
|
}
|
|
149
|
-
function extractF500Years(
|
|
63
|
+
function extractF500Years(record2) {
|
|
150
64
|
const reprintRegex = /(?<reprint>Lisäpainokset:|Lisäpainos:)/u;
|
|
151
|
-
const reprintFieldContents =
|
|
152
|
-
subfields
|
|
153
|
-
}) => subfields).flat().filter(({
|
|
154
|
-
code
|
|
155
|
-
}) => code === 'a').map(({
|
|
156
|
-
value
|
|
157
|
-
}) => value);
|
|
158
|
-
//.filter(value => value.test(reprintRegex));
|
|
159
|
-
|
|
65
|
+
const reprintFieldContents = record2.get(/^500$/u).map(({ subfields }) => subfields).flat().filter(({ code }) => code === "a").map(({ value }) => value);
|
|
160
66
|
debug(`f500 reprint field contents: ${JSON.stringify(reprintFieldContents)}`);
|
|
161
|
-
const filteredF500 = reprintFieldContents.filter(content => content && content.match(reprintRegex));
|
|
67
|
+
const filteredF500 = reprintFieldContents.filter((content) => content && content.match(reprintRegex));
|
|
162
68
|
debug(`f500 reprint field contents (filtered): ${JSON.stringify(filteredF500)}`);
|
|
163
|
-
const
|
|
164
|
-
return
|
|
69
|
+
const reprintYears2 = extractReprintYears(filteredF500);
|
|
70
|
+
return reprintYears2;
|
|
165
71
|
}
|
|
166
72
|
function extractReprintYears(contents) {
|
|
167
73
|
const yearRegex = /[0-9][0-9][0-9][0-9]/gu;
|
|
168
|
-
const years = contents.map(content => content.match(yearRegex)).flat();
|
|
74
|
+
const years = contents.map((content) => content.match(yearRegex)).flat();
|
|
169
75
|
debug(`${JSON.stringify(years)}`);
|
|
170
76
|
const uniqYears = [...new Set(years)].sort();
|
|
171
77
|
debug(`${JSON.stringify(uniqYears)}`);
|
|
@@ -174,7 +80,7 @@ var _default = () => ({
|
|
|
174
80
|
},
|
|
175
81
|
// eslint-disable-next-line max-statements
|
|
176
82
|
compare: (a, b) => {
|
|
177
|
-
const debug = (
|
|
83
|
+
const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi");
|
|
178
84
|
debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);
|
|
179
85
|
const [firstA] = a.normalYears ? a.normalYears : a;
|
|
180
86
|
const [firstB] = b.normalYears ? b.normalYears : b;
|
|
@@ -182,34 +88,23 @@ var _default = () => ({
|
|
|
182
88
|
if (firstA === firstB) {
|
|
183
89
|
return 0.1;
|
|
184
90
|
}
|
|
185
|
-
|
|
186
|
-
// If either of years is a non string/number, values are not comparable
|
|
187
|
-
if (!(0, _matchingUtils.testStringOrNumber)(firstA) || !(0, _matchingUtils.testStringOrNumber)(firstB)) {
|
|
91
|
+
if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {
|
|
188
92
|
return 0;
|
|
189
93
|
}
|
|
190
94
|
const firstANumber = parseInt(firstA, 10);
|
|
191
95
|
const firstBNumber = parseInt(firstB, 10);
|
|
192
|
-
|
|
193
|
-
// Handle consequent years as a match
|
|
194
|
-
// see publication-time for a version that does not handle consequent years as a match
|
|
195
96
|
if (!(isNaN(firstANumber) || isNaN(firstBNumber)) && (firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber)) {
|
|
196
97
|
return 0.1;
|
|
197
98
|
}
|
|
198
|
-
|
|
199
|
-
|
|
200
|
-
|
|
201
|
-
// Do not give minus points if a normal publishing year is found in normal years
|
|
202
|
-
const bNormalInANormal = a.normalYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));
|
|
203
|
-
const aNormalInBNormal = b.normalYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));
|
|
99
|
+
const bNormalInANormal = a.normalYears.filter((aValue) => b.normalYears.some((bValue) => aValue === bValue));
|
|
100
|
+
const aNormalInBNormal = b.normalYears.filter((bValue) => a.normalYears.some((aValue) => bValue === aValue));
|
|
204
101
|
debug(`BNorm in ANorm: ${JSON.stringify(bNormalInANormal)}`);
|
|
205
102
|
debug(`ANorm in BNorm: ${JSON.stringify(aNormalInBNormal)}`);
|
|
206
103
|
if (bNormalInANormal > 0 || aNormalInBNormal > 0) {
|
|
207
104
|
return 0;
|
|
208
105
|
}
|
|
209
|
-
|
|
210
|
-
|
|
211
|
-
const bNormalInAReprint = a.reprintYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));
|
|
212
|
-
const aNormalInBReprint = b.reprintYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));
|
|
106
|
+
const bNormalInAReprint = a.reprintYears.filter((aValue) => b.normalYears.some((bValue) => aValue === bValue));
|
|
107
|
+
const aNormalInBReprint = b.reprintYears.filter((bValue) => a.normalYears.some((aValue) => bValue === aValue));
|
|
213
108
|
debug(`BNorm in AReprint: ${JSON.stringify(bNormalInAReprint)}`);
|
|
214
109
|
debug(`ANorm in BReprint: ${JSON.stringify(aNormalInBReprint)}`);
|
|
215
110
|
if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {
|
|
@@ -217,40 +112,5 @@ var _default = () => ({
|
|
|
217
112
|
}
|
|
218
113
|
return -1;
|
|
219
114
|
}
|
|
220
|
-
});
|
|
221
|
-
|
|
222
|
-
// 06 - Type of date/Publication status
|
|
223
|
-
// 07-10 - Date 1
|
|
224
|
-
// 11-14 - Date 2
|
|
225
|
-
//
|
|
226
|
-
// 06 - Type of date/Publication status
|
|
227
|
-
// b - No dates given; B.C. date involved
|
|
228
|
-
// c - Continuing resource currently published
|
|
229
|
-
// d - Continuing resource ceased publication
|
|
230
|
-
// e - Detailed date
|
|
231
|
-
// i - Inclusive dates of collection
|
|
232
|
-
// k - Range of years of bulk of collection
|
|
233
|
-
// m - Multiple dates
|
|
234
|
-
// n - Dates unknown
|
|
235
|
-
// p - Date of distribution/release/issue and production/recording session when different
|
|
236
|
-
// q - Questionable date
|
|
237
|
-
// r - Reprint/reissue date and original date
|
|
238
|
-
// s - Single known date/probable date
|
|
239
|
-
// t - Publication date and copyright date
|
|
240
|
-
// u - Continuing resource status unknown
|
|
241
|
-
// | - No attempt to code
|
|
242
|
-
//
|
|
243
|
-
// 07-10 - Date 1
|
|
244
|
-
// 1-9 - Date digit
|
|
245
|
-
// # - Date element is not applicable
|
|
246
|
-
// u - Date element is totally or partially unknown
|
|
247
|
-
// |||| - No attempt to code
|
|
248
|
-
//
|
|
249
|
-
// 11-14 - Date 2
|
|
250
|
-
// 1-9 - Date digit
|
|
251
|
-
// # - Date element is not applicable
|
|
252
|
-
// u - Date element is totally or partially unknown
|
|
253
|
-
// |||| - No attempt to code
|
|
254
|
-
//
|
|
255
|
-
exports.default = _default;
|
|
256
|
-
//# sourceMappingURL=publication-time-allow-cons-years-multi.js.map
|
|
115
|
+
});
|
|
116
|
+
//# sourceMappingURL=publication-time-allow-cons-years-multi.js.map
|
|
@@ -1 +1,7 @@
|
|
|
1
|
-
{"version":3,"file":"publication-time-allow-cons-years-multi.js","names":["_matchingUtils","require","_debug","_interopRequireDefault","obj","__esModule","default","_default","name","extract","record","recordExternal","debug","createDebugLogger","label","f008Values","extractF008Values","JSON","stringify","f26xValues","extractF26xValues","f500Values","extractF500Years","normalYears","Set","concat","f008Date1","filter","value","sort","copyrightYears","reprintYears","combined","get","undefined","testStringOrNumber","extractF008Date1","f008Date2","extractF008Date2","f008YearType","extractF008YearType","String","slice","copyrightRegex","pubNormalSubFieldValues","extractPubNormalSubfieldValues","pubF264CopySubFieldValues","extractPubF264CopySubfieldValues","pubF260CopySubFieldValues","extractPubF260CopySubfieldValues","field","tag","ind2","map","subfields","flat","code","test","removeNonAlphaNumeric","replace","nonAlphaNumericRegex","reprintRegex","reprintFieldContents","filteredF500","content","match","extractReprintYears","contents","yearRegex","years","uniqYears","compare","a","b","firstA","firstB","firstANumber","parseInt","firstBNumber","isNaN","bNormalInANormal","aValue","some","bValue","aNormalInBNormal","bNormalInAReprint","aNormalInBReprint","exports"],"sources":["../../../../src/match-detection/features/bib/publication-time-allow-cons-years-multi.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport {testStringOrNumber} from '../../../matching-utils';\nimport createDebugLogger from 'debug';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We should also get publishing time type from f008\n// We should get reprint times from f500 $a \"Lisäpainos/Lisäpainokset:\"\n\nexport default () => ({\n name: 'Publication time, allow consequent years, years from multiple sources',\n extract: ({record, recordExternal}) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n const f008Values = extractF008Values(record);\n debug(`${label} f008: ${JSON.stringify(f008Values)}`);\n\n const f26xValues = extractF26xValues(record);\n debug(`${label} f26x: ${JSON.stringify(f26xValues)}`);\n\n const f500Values = extractF500Years(record);\n debug(`${label} f500: ${JSON.stringify(f500Values)}`);\n\n // We should get copyrightYear from f008Date2 to copyrightYears when f008YearType = 'r'\n // Is the original year (f008Date2) in f008YearType === 'r' comparable to copyrightYear?\n // We should handle unknown years (here or comparison?)\n // We should handle year ranges for continuing resources / collections\n\n const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter(value => value && value !== ' ' && value !== '||||'))].sort();\n const copyrightYears = [...new Set(f26xValues.copyrightYears)].sort();\n const reprintYears = [...new Set(f500Values)].sort();\n\n const combined = {normalYears, copyrightYears, reprintYears};\n\n debug(`Combined: ${JSON.stringify(combined)}`);\n\n return combined;\n\n function extractF008Values(record) {\n // Record should have only one f008 - in case of several, we handle the first one\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n if (value && testStringOrNumber(value)) {\n const f008Date1 = extractF008Date1(value);\n const f008Date2 = extractF008Date2(value);\n const f008YearType = extractF008YearType(value);\n return {f008Date1, f008Date2, f008YearType};\n }\n return {f008Date1: undefined, f008Date2: undefined, f008YearType: undefined};\n\n function extractF008Date1(value) {\n return String(value).slice(7, 11);\n }\n\n function extractF008Date2(value) {\n return String(value).slice(11, 15);\n }\n\n function extractF008YearType(value) {\n return String(value).slice(6, 7);\n }\n }\n\n function extractF26xValues(record) {\n const copyrightRegex = /^(?<copyrightPrefix>cop|cop.|c|©|p|℗)/u;\n\n const pubNormalSubFieldValues = extractPubNormalSubfieldValues(record, copyrightRegex);\n debug(`Normal years: ${JSON.stringify(pubNormalSubFieldValues)}`);\n\n const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(record);\n debug(`F264 copyright years: ${JSON.stringify(pubF264CopySubFieldValues)}`);\n\n const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(record, copyrightRegex);\n debug(`F260 copyright years: ${JSON.stringify(pubF260CopySubFieldValues)}`);\n\n return {normalYears: pubNormalSubFieldValues, copyrightYears: [...pubF260CopySubFieldValues, ...pubF264CopySubFieldValues]};\n\n function extractPubNormalSubfieldValues(record, copyrightRegex) {\n return record.get(/^26[04]$/u)\n .filter((field) => !(field.tag === '264' && field.ind2 === '4'))\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value && !copyrightRegex.test(value))\n .map(({value}) => value)\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function extractPubF264CopySubfieldValues(record, copyrightRegex) {\n return record.get(/^264$/u)\n .filter((field) => field.ind2 === '4')\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value)\n .map(({value}) => value)\n .map((value) => value.replace(copyrightRegex, ''))\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function extractPubF260CopySubfieldValues(record, copyrightRegex) {\n return record.get(/^260$/u)\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value && copyrightRegex.test(value))\n .map(({value}) => value)\n .map((value) => value.replace(copyrightRegex, ''))\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function removeNonAlphaNumeric(value) {\n debug(`Cleaning: ${JSON.stringify(value)}`);\n const nonAlphaNumericRegex = /[^A-Za-z0-9]/ug;\n return value ? value.replace(nonAlphaNumericRegex, '') : value;\n }\n }\n\n function extractF500Years(record) {\n\n const reprintRegex = /(?<reprint>Lisäpainokset:|Lisäpainos:)/u;\n const reprintFieldContents = record.get(/^500$/u)\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code === 'a')\n .map(({value}) => value);\n //.filter(value => value.test(reprintRegex));\n\n debug(`f500 reprint field contents: ${JSON.stringify(reprintFieldContents)}`);\n\n const filteredF500 = reprintFieldContents.filter((content) => content && content.match(reprintRegex));\n\n debug(`f500 reprint field contents (filtered): ${JSON.stringify(filteredF500)}`);\n\n const reprintYears = extractReprintYears(filteredF500);\n\n return reprintYears;\n }\n\n function extractReprintYears(contents) {\n const yearRegex = /[0-9][0-9][0-9][0-9]/gu;\n const years = contents.map(content => content.match(yearRegex))\n .flat();\n debug(`${JSON.stringify(years)}`);\n const uniqYears = [...new Set(years)].sort();\n debug(`${JSON.stringify(uniqYears)}`);\n return uniqYears;\n }\n\n },\n // eslint-disable-next-line max-statements\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);\n\n const [firstA] = a.normalYears ? a.normalYears : a;\n const [firstB] = b.normalYears ? b.normalYears : b;\n\n debug(`Comparing ${JSON.stringify(firstA)} to ${JSON.stringify(firstB)}`);\n\n if (firstA === firstB) {\n return 0.1;\n }\n\n // If either of years is a non string/number, values are not comparable\n if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {\n return 0;\n }\n\n const firstANumber = parseInt(firstA, 10);\n const firstBNumber = parseInt(firstB, 10);\n\n // Handle consequent years as a match\n // see publication-time for a version that does not handle consequent years as a match\n if (!(isNaN(firstANumber) || isNaN(firstBNumber)) &&\n (firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber)) {\n return 0.1;\n }\n\n // We should do something with copyrightYears, too\n\n // Do not give minus points if a normal publishing year is found in normal years\n const bNormalInANormal = a.normalYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n const aNormalInBNormal = b.normalYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n debug(`BNorm in ANorm: ${JSON.stringify(bNormalInANormal)}`);\n debug(`ANorm in BNorm: ${JSON.stringify(aNormalInBNormal)}`);\n\n if (bNormalInANormal > 0 || aNormalInBNormal > 0) {\n return 0;\n }\n\n // Do not give minus points if a normal publishing year is found in reprint years\n const bNormalInAReprint = a.reprintYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n const aNormalInBReprint = b.reprintYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n debug(`BNorm in AReprint: ${JSON.stringify(bNormalInAReprint)}`);\n debug(`ANorm in BReprint: ${JSON.stringify(aNormalInBReprint)}`);\n\n if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {\n return 0;\n }\n\n return -1;\n\n }\n});\n\n// https://www.loc.gov/marc/bibliographic/bd008.html\n// field 008\n// 06 - Type of date/Publication status\n// 07-10 - Date 1\n// 11-14 - Date 2\n//\n// 06 - Type of date/Publication status\n// b - No dates given; B.C. date involved\n// c - Continuing resource currently published\n// d - Continuing resource ceased publication\n// e - Detailed date\n// i - Inclusive dates of collection\n// k - Range of years of bulk of collection\n// m - Multiple dates\n// n - Dates unknown\n// p - Date of distribution/release/issue and production/recording session when different\n// q - Questionable date\n// r - Reprint/reissue date and original date\n// s - Single known date/probable date\n// t - Publication date and copyright date\n// u - Continuing resource status unknown\n// | - No attempt to code\n//\n// 07-10 - Date 1\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n// 11-14 - Date 2\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n\n"],"mappings":";;;;;;AA4BA,IAAAA,cAAA,GAAAC,OAAA;AACA,IAAAC,MAAA,GAAAC,sBAAA,CAAAF,OAAA;AAAsC,SAAAE,uBAAAC,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA7BtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAKA;AACA;AACA;AAAA,IAAAG,QAAA,GAEeA,CAAA,MAAO;EACpBC,IAAI,EAAE,uEAAuE;EAC7EC,OAAO,EAAEA,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,KAAK;IACrC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wGAAwG,CAAC;IACzI,MAAMC,KAAK,GAAGH,cAAc,IAAIA,cAAc,CAACG,KAAK,GAAGH,cAAc,CAACG,KAAK,GAAG,QAAQ;IAEtF,MAAMC,UAAU,GAAGC,iBAAiB,CAACN,MAAM,CAAC;IAC5CE,KAAK,CAAE,GAAEE,KAAM,UAASG,IAAI,CAACC,SAAS,CAACH,UAAU,CAAE,EAAC,CAAC;IAErD,MAAMI,UAAU,GAAGC,iBAAiB,CAACV,MAAM,CAAC;IAC5CE,KAAK,CAAE,GAAEE,KAAM,UAASG,IAAI,CAACC,SAAS,CAACC,UAAU,CAAE,EAAC,CAAC;IAErD,MAAME,UAAU,GAAGC,gBAAgB,CAACZ,MAAM,CAAC;IAC3CE,KAAK,CAAE,GAAEE,KAAM,UAASG,IAAI,CAACC,SAAS,CAACG,UAAU,CAAE,EAAC,CAAC;;IAErD;IACA;IACA;IACA;;IAEA,MAAME,WAAW,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACL,UAAU,CAACI,WAAW,CAACE,MAAM,CAACV,UAAU,CAACW,SAAS,CAAC,CAACC,MAAM,CAACC,KAAK,IAAIA,KAAK,IAAIA,KAAK,KAAK,MAAM,IAAIA,KAAK,KAAK,MAAM,CAAC,CAAC,CAAC,CAACC,IAAI,CAAC,CAAC;IAC3J,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIN,GAAG,CAACL,UAAU,CAACW,cAAc,CAAC,CAAC,CAACD,IAAI,CAAC,CAAC;IACrE,MAAME,YAAY,GAAG,CAAC,GAAG,IAAIP,GAAG,CAACH,UAAU,CAAC,CAAC,CAACQ,IAAI,CAAC,CAAC;IAEpD,MAAMG,QAAQ,GAAG;MAACT,WAAW;MAAEO,cAAc;MAAEC;IAAY,CAAC;IAE5DnB,KAAK,CAAE,aAAYK,IAAI,CAACC,SAAS,CAACc,QAAQ,CAAE,EAAC,CAAC;IAE9C,OAAOA,QAAQ;IAEf,SAAShB,iBAAiBA,CAACN,MAAM,EAAE;MACjC;MACA,MAAMkB,KAAK,GAAGlB,MAAM,CAACuB,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,EAAEL,KAAK,IAAIM,SAAS;MAC3D,IAAIN,KAAK,IAAI,IAAAO,iCAAkB,EAACP,KAAK,CAAC,EAAE;QACtC,MAAMF,SAAS,GAAGU,gBAAgB,CAACR,KAAK,CAAC;QACzC,MAAMS,SAAS,GAAGC,gBAAgB,CAACV,KAAK,CAAC;QACzC,MAAMW,YAAY,GAAGC,mBAAmB,CAACZ,KAAK,CAAC;QAC/C,OAAO;UAACF,SAAS;UAAEW,SAAS;UAAEE;QAAY,CAAC;MAC7C;MACA,OAAO;QAACb,SAAS,EAAEQ,SAAS;QAAEG,SAAS,EAAEH,SAAS;QAAEK,YAAY,EAAEL;MAAS,CAAC;MAE5E,SAASE,gBAAgBA,CAACR,KAAK,EAAE;QAC/B,OAAOa,MAAM,CAACb,KAAK,CAAC,CAACc,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC;MACnC;MAEA,SAASJ,gBAAgBA,CAACV,KAAK,EAAE;QAC/B,OAAOa,MAAM,CAACb,KAAK,CAAC,CAACc,KAAK,CAAC,EAAE,EAAE,EAAE,CAAC;MACpC;MAEA,SAASF,mBAAmBA,CAACZ,KAAK,EAAE;QAClC,OAAOa,MAAM,CAACb,KAAK,CAAC,CAACc,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC;MAClC;IACF;IAEA,SAAStB,iBAAiBA,CAACV,MAAM,EAAE;MACjC,MAAMiC,cAAc,GAAG,wCAAwC;MAE/D,MAAMC,uBAAuB,GAAGC,8BAA8B,CAACnC,MAAM,EAAEiC,cAAc,CAAC;MACtF/B,KAAK,CAAE,iBAAgBK,IAAI,CAACC,SAAS,CAAC0B,uBAAuB,CAAE,EAAC,CAAC;MAEjE,MAAME,yBAAyB,GAAGC,gCAAgC,CAACrC,MAAM,CAAC;MAC1EE,KAAK,CAAE,yBAAwBK,IAAI,CAACC,SAAS,CAAC4B,yBAAyB,CAAE,EAAC,CAAC;MAE3E,MAAME,yBAAyB,GAAGC,gCAAgC,CAACvC,MAAM,EAAEiC,cAAc,CAAC;MAC1F/B,KAAK,CAAE,yBAAwBK,IAAI,CAACC,SAAS,CAAC8B,yBAAyB,CAAE,EAAC,CAAC;MAE3E,OAAO;QAACzB,WAAW,EAAEqB,uBAAuB;QAAEd,cAAc,EAAE,CAAC,GAAGkB,yBAAyB,EAAE,GAAGF,yBAAyB;MAAC,CAAC;MAE3H,SAASD,8BAA8BA,CAACnC,MAAM,EAAEiC,cAAc,EAAE;QAC9D,OAAOjC,MAAM,CAACuB,GAAG,CAAC,WAAW,CAAC,CAC3BN,MAAM,CAAEuB,KAAK,IAAK,EAAEA,KAAK,CAACC,GAAG,KAAK,KAAK,IAAID,KAAK,CAACE,IAAI,KAAK,GAAG,CAAC,CAAC,CAC/DC,GAAG,CAAC,CAAC;UAACC;QAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,CAAC,CAAC,CACN5B,MAAM,CAAC,CAAC;UAAC6B;QAAI,CAAC,KAAKA,IAAI,IAAIA,IAAI,KAAK,GAAG,CAAC,CACxC7B,MAAM,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,IAAI,CAACe,cAAc,CAACc,IAAI,CAAC7B,KAAK,CAAC,CAAC,CACzDyB,GAAG,CAAC,CAAC;UAACzB;QAAK,CAAC,KAAKA,KAAK,CAAC,CACvByB,GAAG,CAAEzB,KAAK,IAAK8B,qBAAqB,CAAC9B,KAAK,CAAC,CAAC;MACjD;MAEA,SAASmB,gCAAgCA,CAACrC,MAAM,EAAEiC,cAAc,EAAE;QAChE,OAAOjC,MAAM,CAACuB,GAAG,CAAC,QAAQ,CAAC,CACxBN,MAAM,CAAEuB,KAAK,IAAKA,KAAK,CAACE,IAAI,KAAK,GAAG,CAAC,CACrCC,GAAG,CAAC,CAAC;UAACC;QAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,CAAC,CAAC,CACN5B,MAAM,CAAC,CAAC;UAAC6B;QAAI,CAAC,KAAKA,IAAI,IAAIA,IAAI,KAAK,GAAG,CAAC,CACxC7B,MAAM,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,CAAC,CAC1ByB,GAAG,CAAC,CAAC;UAACzB;QAAK,CAAC,KAAKA,KAAK,CAAC,CACvByB,GAAG,CAAEzB,KAAK,IAAKA,KAAK,CAAC+B,OAAO,CAAChB,cAAc,EAAE,EAAE,CAAC,CAAC,CACjDU,GAAG,CAAEzB,KAAK,IAAK8B,qBAAqB,CAAC9B,KAAK,CAAC,CAAC;MACjD;MAEA,SAASqB,gCAAgCA,CAACvC,MAAM,EAAEiC,cAAc,EAAE;QAChE,OAAOjC,MAAM,CAACuB,GAAG,CAAC,QAAQ,CAAC,CACxBoB,GAAG,CAAC,CAAC;UAACC;QAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,CAAC,CAAC,CACN5B,MAAM,CAAC,CAAC;UAAC6B;QAAI,CAAC,KAAKA,IAAI,IAAIA,IAAI,KAAK,GAAG,CAAC,CACxC7B,MAAM,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,IAAIe,cAAc,CAACc,IAAI,CAAC7B,KAAK,CAAC,CAAC,CACxDyB,GAAG,CAAC,CAAC;UAACzB;QAAK,CAAC,KAAKA,KAAK,CAAC,CACvByB,GAAG,CAAEzB,KAAK,IAAKA,KAAK,CAAC+B,OAAO,CAAChB,cAAc,EAAE,EAAE,CAAC,CAAC,CACjDU,GAAG,CAAEzB,KAAK,IAAK8B,qBAAqB,CAAC9B,KAAK,CAAC,CAAC;MACjD;MAEA,SAAS8B,qBAAqBA,CAAC9B,KAAK,EAAE;QACpChB,KAAK,CAAE,aAAYK,IAAI,CAACC,SAAS,CAACU,KAAK,CAAE,EAAC,CAAC;QAC3C,MAAMgC,oBAAoB,GAAG,gBAAgB;QAC7C,OAAOhC,KAAK,GAAGA,KAAK,CAAC+B,OAAO,CAACC,oBAAoB,EAAE,EAAE,CAAC,GAAGhC,KAAK;MAChE;IACF;IAEA,SAASN,gBAAgBA,CAACZ,MAAM,EAAE;MAEhC,MAAMmD,YAAY,GAAG,yCAAyC;MAC9D,MAAMC,oBAAoB,GAAGpD,MAAM,CAACuB,GAAG,CAAC,QAAQ,CAAC,CAC9CoB,GAAG,CAAC,CAAC;QAACC;MAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BC,IAAI,CAAC,CAAC,CACN5B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCH,GAAG,CAAC,CAAC;QAACzB;MAAK,CAAC,KAAKA,KAAK,CAAC;MACxB;;MAEFhB,KAAK,CAAE,gCAA+BK,IAAI,CAACC,SAAS,CAAC4C,oBAAoB,CAAE,EAAC,CAAC;MAE7E,MAAMC,YAAY,GAAGD,oBAAoB,CAACnC,MAAM,CAAEqC,OAAO,IAAKA,OAAO,IAAIA,OAAO,CAACC,KAAK,CAACJ,YAAY,CAAC,CAAC;MAErGjD,KAAK,CAAE,2CAA0CK,IAAI,CAACC,SAAS,CAAC6C,YAAY,CAAE,EAAC,CAAC;MAEhF,MAAMhC,YAAY,GAAGmC,mBAAmB,CAACH,YAAY,CAAC;MAEtD,OAAOhC,YAAY;IACrB;IAEA,SAASmC,mBAAmBA,CAACC,QAAQ,EAAE;MACrC,MAAMC,SAAS,GAAG,wBAAwB;MAC1C,MAAMC,KAAK,GAAGF,QAAQ,CAACd,GAAG,CAACW,OAAO,IAAIA,OAAO,CAACC,KAAK,CAACG,SAAS,CAAC,CAAC,CAC5Db,IAAI,CAAC,CAAC;MACT3C,KAAK,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACmD,KAAK,CAAE,EAAC,CAAC;MACjC,MAAMC,SAAS,GAAG,CAAC,GAAG,IAAI9C,GAAG,CAAC6C,KAAK,CAAC,CAAC,CAACxC,IAAI,CAAC,CAAC;MAC5CjB,KAAK,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACoD,SAAS,CAAE,EAAC,CAAC;MACrC,OAAOA,SAAS;IAClB;EAEF,CAAC;EACD;EACAC,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB,MAAM7D,KAAK,GAAG,IAAAC,cAAiB,EAAC,wGAAwG,CAAC;IACzID,KAAK,CAAE,aAAYK,IAAI,CAACC,SAAS,CAACsD,CAAC,CAAE,OAAMvD,IAAI,CAACC,SAAS,CAACuD,CAAC,CAAE,EAAC,CAAC;IAE/D,MAAM,CAACC,MAAM,CAAC,GAAGF,CAAC,CAACjD,WAAW,GAAGiD,CAAC,CAACjD,WAAW,GAAGiD,CAAC;IAClD,MAAM,CAACG,MAAM,CAAC,GAAGF,CAAC,CAAClD,WAAW,GAAGkD,CAAC,CAAClD,WAAW,GAAGkD,CAAC;IAElD7D,KAAK,CAAE,aAAYK,IAAI,CAACC,SAAS,CAACwD,MAAM,CAAE,OAAMzD,IAAI,CAACC,SAAS,CAACyD,MAAM,CAAE,EAAC,CAAC;IAEzE,IAAID,MAAM,KAAKC,MAAM,EAAE;MACrB,OAAO,GAAG;IACZ;;IAEA;IACA,IAAI,CAAC,IAAAxC,iCAAkB,EAACuC,MAAM,CAAC,IAAI,CAAC,IAAAvC,iCAAkB,EAACwC,MAAM,CAAC,EAAE;MAC9D,OAAO,CAAC;IACV;IAEA,MAAMC,YAAY,GAAGC,QAAQ,CAACH,MAAM,EAAE,EAAE,CAAC;IACzC,MAAMI,YAAY,GAAGD,QAAQ,CAACF,MAAM,EAAE,EAAE,CAAC;;IAEzC;IACA;IACA,IAAI,EAAEI,KAAK,CAACH,YAAY,CAAC,IAAIG,KAAK,CAACD,YAAY,CAAC,CAAC,KAC7CF,YAAY,GAAG,CAAC,KAAKE,YAAY,IAAIF,YAAY,GAAG,CAAC,KAAKE,YAAY,CAAC,EAAE;MAC3E,OAAO,GAAG;IACZ;;IAEA;;IAEA;IACA,MAAME,gBAAgB,GAAGR,CAAC,CAACjD,WAAW,CAACI,MAAM,CAACsD,MAAM,IAAIR,CAAC,CAAClD,WAAW,CAAC2D,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC;IACxG,MAAMC,gBAAgB,GAAGX,CAAC,CAAClD,WAAW,CAACI,MAAM,CAACwD,MAAM,IAAIX,CAAC,CAACjD,WAAW,CAAC2D,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC;IACxGrE,KAAK,CAAE,mBAAkBK,IAAI,CAACC,SAAS,CAAC8D,gBAAgB,CAAE,EAAC,CAAC;IAC5DpE,KAAK,CAAE,mBAAkBK,IAAI,CAACC,SAAS,CAACkE,gBAAgB,CAAE,EAAC,CAAC;IAE5D,IAAIJ,gBAAgB,GAAG,CAAC,IAAII,gBAAgB,GAAG,CAAC,EAAE;MAChD,OAAO,CAAC;IACV;;IAEA;IACA,MAAMC,iBAAiB,GAAGb,CAAC,CAACzC,YAAY,CAACJ,MAAM,CAACsD,MAAM,IAAIR,CAAC,CAAClD,WAAW,CAAC2D,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC;IAC1G,MAAMG,iBAAiB,GAAGb,CAAC,CAAC1C,YAAY,CAACJ,MAAM,CAACwD,MAAM,IAAIX,CAAC,CAACjD,WAAW,CAAC2D,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC;IAC1GrE,KAAK,CAAE,sBAAqBK,IAAI,CAACC,SAAS,CAACmE,iBAAiB,CAAE,EAAC,CAAC;IAChEzE,KAAK,CAAE,sBAAqBK,IAAI,CAACC,SAAS,CAACoE,iBAAiB,CAAE,EAAC,CAAC;IAEhE,IAAID,iBAAiB,GAAG,CAAC,IAAIC,iBAAiB,GAAG,CAAC,EAAE;MAClD,OAAO,CAAC;IACV;IAEA,OAAO,CAAC,CAAC;EAEX;AACF,CAAC,CAAC,EAEF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAAC,OAAA,CAAAjF,OAAA,GAAAC,QAAA"}
|
|
1
|
+
{
|
|
2
|
+
"version": 3,
|
|
3
|
+
"sources": ["../../../../src/match-detection/features/bib/publication-time-allow-cons-years-multi.js"],
|
|
4
|
+
"sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We should also get publishing time type from f008\n// We should get reprint times from f500 $a \"Lis\u00E4painos/Lis\u00E4painokset:\"\n\nexport default () => ({\n name: 'Publication time, allow consequent years, years from multiple sources',\n extract: ({record, recordExternal}) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n const f008Values = extractF008Values(record);\n debug(`${label}: f008: ${JSON.stringify(f008Values)}`);\n\n const f26xValues = extractF26xValues(record);\n debug(`${label}: f26x: ${JSON.stringify(f26xValues)}`);\n\n const f500Values = extractF500Years(record);\n debug(`${label}: f500: ${JSON.stringify(f500Values)}`);\n\n // We should get copyrightYear from f008Date2 to copyrightYears when f008YearType = 'r'\n // Is the original year (f008Date2) in f008YearType === 'r' comparable to copyrightYear?\n // We should handle unknown years (here or comparison?)\n // We should handle year ranges for continuing resources / collections\n\n const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter(value => value && value !== ' ' && value !== '||||'))].sort();\n const copyrightYears = [...new Set(f26xValues.copyrightYears)].sort();\n const reprintYears = [...new Set(f500Values)].sort();\n\n const combined = {normalYears, copyrightYears, reprintYears};\n\n debug(`Combined: ${JSON.stringify(combined)}`);\n\n return combined;\n\n function extractF008Values(record) {\n // Record should have only one f008 - in case of several, we handle the first one\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n if (value && testStringOrNumber(value)) {\n const f008Date1 = extractF008Date1(value);\n const f008Date2 = extractF008Date2(value);\n const f008YearType = extractF008YearType(value);\n return {f008Date1, f008Date2, f008YearType};\n }\n return {f008Date1: undefined, f008Date2: undefined, f008YearType: undefined};\n\n function extractF008Date1(value) {\n return String(value).slice(7, 11);\n }\n\n function extractF008Date2(value) {\n return String(value).slice(11, 15);\n }\n\n function extractF008YearType(value) {\n return String(value).slice(6, 7);\n }\n }\n\n function extractF26xValues(record) {\n const copyrightRegex = /^(?<copyrightPrefix>cop|cop.|c|\u00A9|p|\u2117)/u;\n\n const pubNormalSubFieldValues = extractPubNormalSubfieldValues(record, copyrightRegex);\n debug(`Normal years: ${JSON.stringify(pubNormalSubFieldValues)}`);\n\n const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(record);\n debug(`F264 copyright years: ${JSON.stringify(pubF264CopySubFieldValues)}`);\n\n const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(record, copyrightRegex);\n debug(`F260 copyright years: ${JSON.stringify(pubF260CopySubFieldValues)}`);\n\n return {normalYears: pubNormalSubFieldValues, copyrightYears: [...pubF260CopySubFieldValues, ...pubF264CopySubFieldValues]};\n\n function extractPubNormalSubfieldValues(record, copyrightRegex) {\n return record.get(/^26[04]$/u)\n .filter((field) => !(field.tag === '264' && field.ind2 === '4'))\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value && !copyrightRegex.test(value))\n .map(({value}) => value)\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function extractPubF264CopySubfieldValues(record, copyrightRegex) {\n return record.get(/^264$/u)\n .filter((field) => field.ind2 === '4')\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value)\n .map(({value}) => value)\n .map((value) => value.replace(copyrightRegex, ''))\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function extractPubF260CopySubfieldValues(record, copyrightRegex) {\n return record.get(/^260$/u)\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code && code === 'c')\n .filter(({value}) => value && copyrightRegex.test(value))\n .map(({value}) => value)\n .map((value) => value.replace(copyrightRegex, ''))\n .map((value) => removeNonAlphaNumeric(value));\n }\n\n function removeNonAlphaNumeric(value) {\n debug(`Cleaning: ${JSON.stringify(value)}`);\n const nonAlphaNumericRegex = /[^A-Za-z0-9]/ug;\n return value ? value.replace(nonAlphaNumericRegex, '') : value;\n }\n }\n\n function extractF500Years(record) {\n\n const reprintRegex = /(?<reprint>Lis\u00E4painokset:|Lis\u00E4painos:)/u;\n const reprintFieldContents = record.get(/^500$/u)\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code === 'a')\n .map(({value}) => value);\n //.filter(value => value.test(reprintRegex));\n\n debug(`f500 reprint field contents: ${JSON.stringify(reprintFieldContents)}`);\n\n const filteredF500 = reprintFieldContents.filter((content) => content && content.match(reprintRegex));\n\n debug(`f500 reprint field contents (filtered): ${JSON.stringify(filteredF500)}`);\n\n const reprintYears = extractReprintYears(filteredF500);\n\n return reprintYears;\n }\n\n function extractReprintYears(contents) {\n const yearRegex = /[0-9][0-9][0-9][0-9]/gu;\n const years = contents.map(content => content.match(yearRegex))\n .flat();\n debug(`${JSON.stringify(years)}`);\n const uniqYears = [...new Set(years)].sort();\n debug(`${JSON.stringify(uniqYears)}`);\n return uniqYears;\n }\n\n },\n // eslint-disable-next-line max-statements\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);\n\n const [firstA] = a.normalYears ? a.normalYears : a;\n const [firstB] = b.normalYears ? b.normalYears : b;\n\n debug(`Comparing ${JSON.stringify(firstA)} to ${JSON.stringify(firstB)}`);\n\n if (firstA === firstB) {\n return 0.1;\n }\n\n // If either of years is a non string/number, values are not comparable\n if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {\n return 0;\n }\n\n const firstANumber = parseInt(firstA, 10);\n const firstBNumber = parseInt(firstB, 10);\n\n // Handle consequent years as a match\n // see publication-time for a version that does not handle consequent years as a match\n if (!(isNaN(firstANumber) || isNaN(firstBNumber)) &&\n (firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber)) {\n return 0.1;\n }\n\n // We should do something with copyrightYears, too\n\n // Do not give minus points if a normal publishing year is found in normal years\n const bNormalInANormal = a.normalYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n const aNormalInBNormal = b.normalYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n debug(`BNorm in ANorm: ${JSON.stringify(bNormalInANormal)}`);\n debug(`ANorm in BNorm: ${JSON.stringify(aNormalInBNormal)}`);\n\n if (bNormalInANormal > 0 || aNormalInBNormal > 0) {\n return 0;\n }\n\n // Do not give minus points if a normal publishing year is found in reprint years\n const bNormalInAReprint = a.reprintYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n const aNormalInBReprint = b.reprintYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n debug(`BNorm in AReprint: ${JSON.stringify(bNormalInAReprint)}`);\n debug(`ANorm in BReprint: ${JSON.stringify(aNormalInBReprint)}`);\n\n if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {\n return 0;\n }\n\n return -1;\n\n }\n});\n\n// https://www.loc.gov/marc/bibliographic/bd008.html\n// field 008\n// 06 - Type of date/Publication status\n// 07-10 - Date 1\n// 11-14 - Date 2\n//\n// 06 - Type of date/Publication status\n// b - No dates given; B.C. date involved\n// c - Continuing resource currently published\n// d - Continuing resource ceased publication\n// e - Detailed date\n// i - Inclusive dates of collection\n// k - Range of years of bulk of collection\n// m - Multiple dates\n// n - Dates unknown\n// p - Date of distribution/release/issue and production/recording session when different\n// q - Questionable date\n// r - Reprint/reissue date and original date\n// s - Single known date/probable date\n// t - Publication date and copyright date\n// u - Continuing resource status unknown\n// | - No attempt to code\n//\n// 07-10 - Date 1\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n// 11-14 - Date 2\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n\n"],
|
|
5
|
+
"mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,0BAAyB;AAMjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,QAAQ,eAAc,MAAM;AACrC,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,QAAQ,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAE9E,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,iBAAiB,MAAM;AAC1C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAOrD,UAAM,cAAc,CAAC,GAAG,IAAI,IAAI,WAAW,YAAY,OAAO,WAAW,SAAS,EAAE,OAAO,WAAS,SAAS,UAAU,UAAU,UAAU,MAAM,CAAC,CAAC,EAAE,KAAK;AAC1J,UAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,WAAW,cAAc,CAAC,EAAE,KAAK;AACpE,UAAM,eAAe,CAAC,GAAG,IAAI,IAAI,UAAU,CAAC,EAAE,KAAK;AAEnD,UAAM,WAAW,EAAC,aAAa,gBAAgB,aAAY;AAE3D,UAAM,aAAa,KAAK,UAAU,QAAQ,CAAC,EAAE;AAE7C,WAAO;AAEP,aAAS,kBAAkBA,SAAQ;AAEjC,YAAM,QAAQA,QAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,UAAI,SAAS,mBAAmB,KAAK,GAAG;AACtC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,eAAe,oBAAoB,KAAK;AAC9C,eAAO,EAAC,WAAW,WAAW,aAAY;AAAA,MAC5C;AACA,aAAO,EAAC,WAAW,QAAW,WAAW,QAAW,cAAc,OAAS;AAE3E,eAAS,iBAAiBC,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,EAAE;AAAA,MAClC;AAEA,eAAS,iBAAiBA,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,IAAI,EAAE;AAAA,MACnC;AAEA,eAAS,oBAAoBA,QAAO;AAClC,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,CAAC;AAAA,MACjC;AAAA,IACF;AAEA,aAAS,kBAAkBD,SAAQ;AACjC,YAAM,iBAAiB;AAEvB,YAAM,0BAA0B,+BAA+BA,SAAQ,cAAc;AACrF,YAAM,iBAAiB,KAAK,UAAU,uBAAuB,CAAC,EAAE;AAEhE,YAAM,4BAA4B,iCAAiCA,OAAM;AACzE,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,YAAM,4BAA4B,iCAAiCA,SAAQ,cAAc;AACzF,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,aAAO,EAAC,aAAa,yBAAyB,gBAAgB,CAAC,GAAG,2BAA2B,GAAG,yBAAyB,EAAC;AAE1H,eAAS,+BAA+BA,SAAQE,iBAAgB;AAC9D,eAAOF,QAAO,IAAI,WAAW,EAC1B,OAAO,CAAC,UAAU,EAAE,MAAM,QAAQ,SAAS,MAAM,SAAS,IAAI,EAC9D,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAAS,CAACE,gBAAe,KAAK,KAAK,CAAC,EACxD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,OAAO,CAAC,UAAU,MAAM,SAAS,GAAG,EACpC,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,KAAK,EACzB,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQE,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAASE,gBAAe,KAAK,KAAK,CAAC,EACvD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQA,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,sBAAsB,OAAO;AACpC,cAAM,aAAa,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1C,cAAM,uBAAuB;AAC7B,eAAO,QAAQ,MAAM,QAAQ,sBAAsB,EAAE,IAAI;AAAA,MAC3D;AAAA,IACF;AAEA,aAAS,iBAAiBF,SAAQ;AAEhC,YAAM,eAAe;AACrB,YAAM,uBAAuBA,QAAO,IAAI,QAAQ,EAC7C,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,GAAG,EAC/B,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK;AAGzB,YAAM,gCAAgC,KAAK,UAAU,oBAAoB,CAAC,EAAE;AAE5E,YAAM,eAAe,qBAAqB,OAAO,CAAC,YAAY,WAAW,QAAQ,MAAM,YAAY,CAAC;AAEpG,YAAM,2CAA2C,KAAK,UAAU,YAAY,CAAC,EAAE;AAE/E,YAAMG,gBAAe,oBAAoB,YAAY;AAErD,aAAOA;AAAA,IACT;AAEA,aAAS,oBAAoB,UAAU;AACrC,YAAM,YAAY;AAClB,YAAM,QAAQ,SAAS,IAAI,aAAW,QAAQ,MAAM,SAAS,CAAC,EAC3D,KAAK;AACR,YAAM,GAAG,KAAK,UAAU,KAAK,CAAC,EAAE;AAChC,YAAM,YAAY,CAAC,GAAG,IAAI,IAAI,KAAK,CAAC,EAAE,KAAK;AAC3C,YAAM,GAAG,KAAK,UAAU,SAAS,CAAC,EAAE;AACpC,aAAO;AAAA,IACT;AAAA,EAEF;AAAA;AAAA,EAEA,SAAS,CAAC,GAAG,MAAM;AACjB,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,aAAa,KAAK,UAAU,CAAC,CAAC,OAAO,KAAK,UAAU,CAAC,CAAC,EAAE;AAE9D,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AACjD,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AAEjD,UAAM,aAAa,KAAK,UAAU,MAAM,CAAC,OAAO,KAAK,UAAU,MAAM,CAAC,EAAE;AAExE,QAAI,WAAW,QAAQ;AACrB,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,mBAAmB,MAAM,KAAK,CAAC,mBAAmB,MAAM,GAAG;AAC9D,aAAO;AAAA,IACT;AAEA,UAAM,eAAe,SAAS,QAAQ,EAAE;AACxC,UAAM,eAAe,SAAS,QAAQ,EAAE;AAIxC,QAAI,EAAE,MAAM,YAAY,KAAK,MAAM,YAAY,OAC5C,eAAe,MAAM,gBAAgB,eAAe,MAAM,eAAe;AAC1E,aAAO;AAAA,IACT;AAKA,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAC3D,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAE3D,QAAI,mBAAmB,KAAK,mBAAmB,GAAG;AAChD,aAAO;AAAA,IACT;AAGA,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAC/D,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAE/D,QAAI,oBAAoB,KAAK,oBAAoB,GAAG;AAClD,aAAO;AAAA,IACT;AAEA,WAAO;AAAA,EAET;AACF;",
|
|
6
|
+
"names": ["record", "value", "copyrightRegex", "reprintYears"]
|
|
7
|
+
}
|
|
@@ -1,59 +1,20 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
})
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
var _debug = _interopRequireDefault(require("debug"));
|
|
9
|
-
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
10
|
-
/**
|
|
11
|
-
*
|
|
12
|
-
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
13
|
-
*
|
|
14
|
-
* Melinda record matching modules for Javascript
|
|
15
|
-
*
|
|
16
|
-
* Copyright (C) 2020-2023 University Of Helsinki (The National Library Of Finland)
|
|
17
|
-
*
|
|
18
|
-
* This file is part of melinda-record-matching-js
|
|
19
|
-
*
|
|
20
|
-
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
21
|
-
* it under the terms of the GNU Lesser General Public License as
|
|
22
|
-
* published by the Free Software Foundation, either version 3 of the
|
|
23
|
-
* License, or (at your option) any later version.
|
|
24
|
-
*
|
|
25
|
-
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
26
|
-
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
27
|
-
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
28
|
-
* GNU Lesser General Public License for more details.
|
|
29
|
-
*
|
|
30
|
-
* You should have received a copy of the GNU Affero General Public License
|
|
31
|
-
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
32
|
-
*
|
|
33
|
-
* @licend The above is the entire license notice
|
|
34
|
-
* for the JavaScript code in this file.
|
|
35
|
-
*
|
|
36
|
-
*/
|
|
37
|
-
// We should also get copyright time and copyright/publication times from 26x
|
|
38
|
-
var _default = () => ({
|
|
39
|
-
name: 'Publication time, allow consequent years',
|
|
40
|
-
extract: ({
|
|
41
|
-
record
|
|
42
|
-
}) => {
|
|
43
|
-
const value = record.get(/^008$/u)?.[0]?.value || undefined;
|
|
44
|
-
return (0, _matchingUtils.testStringOrNumber)(value) ? [String(value).slice(7, 11)] : [];
|
|
1
|
+
import createDebugLogger from "debug";
|
|
2
|
+
import { testStringOrNumber } from "../../../matching-utils.js";
|
|
3
|
+
export default () => ({
|
|
4
|
+
name: "Publication time, allow consequent years",
|
|
5
|
+
extract: ({ record }) => {
|
|
6
|
+
const value = record.get(/^008$/u)?.[0]?.value || void 0;
|
|
7
|
+
return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
|
|
45
8
|
},
|
|
46
9
|
compare: (a, b) => {
|
|
47
|
-
const debug = (
|
|
10
|
+
const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years");
|
|
48
11
|
debug(`Comparing ${a[0]} to ${b[0]}`);
|
|
49
12
|
const [firstA] = a;
|
|
50
13
|
const [firstB] = b;
|
|
51
14
|
if (firstA === firstB) {
|
|
52
15
|
return 0.1;
|
|
53
16
|
}
|
|
54
|
-
|
|
55
|
-
// If either of years is a non string/number, values are not comparable
|
|
56
|
-
if (!(0, _matchingUtils.testStringOrNumber)(firstA) || !(0, _matchingUtils.testStringOrNumber)(firstB)) {
|
|
17
|
+
if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {
|
|
57
18
|
return 0;
|
|
58
19
|
}
|
|
59
20
|
const firstANumber = parseInt(firstA, 10);
|
|
@@ -61,11 +22,7 @@ var _default = () => ({
|
|
|
61
22
|
if (isNaN(firstANumber) || isNaN(firstBNumber)) {
|
|
62
23
|
return -1;
|
|
63
24
|
}
|
|
64
|
-
|
|
65
|
-
// Handle consequent years as a match
|
|
66
|
-
// see publication-time for a version that does not handle consequent years as a match
|
|
67
25
|
return firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber ? 0.1 : -1;
|
|
68
26
|
}
|
|
69
27
|
});
|
|
70
|
-
|
|
71
|
-
//# sourceMappingURL=publication-time-allow-cons-years.js.map
|
|
28
|
+
//# sourceMappingURL=publication-time-allow-cons-years.js.map
|
|
@@ -1 +1,7 @@
|
|
|
1
|
-
{
|
|
1
|
+
{
|
|
2
|
+
"version": 3,
|
|
3
|
+
"sources": ["../../../../src/match-detection/features/bib/publication-time-allow-cons-years.js"],
|
|
4
|
+
"sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n\nexport default () => ({\n name: 'Publication time, allow consequent years',\n extract: ({record}) => {\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n },\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years');\n debug(`Comparing ${a[0]} to ${b[0]}`);\n\n const [firstA] = a;\n const [firstB] = b;\n\n if (firstA === firstB) {\n return 0.1;\n }\n\n // If either of years is a non string/number, values are not comparable\n if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {\n return 0;\n }\n\n const firstANumber = parseInt(firstA, 10);\n const firstBNumber = parseInt(firstB, 10);\n\n if (isNaN(firstANumber) || isNaN(firstBNumber)) {\n return -1;\n }\n\n // Handle consequent years as a match\n // see publication-time for a version that does not handle consequent years as a match\n return firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber ? 0.1 : -1;\n }\n});\n"],
|
|
5
|
+
"mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,0BAAyB;AAIjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,OAAM,MAAM;AACrB,UAAM,QAAQ,OAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,WAAO,mBAAmB,KAAK,IAAI,CAAC,OAAO,KAAK,EAAE,MAAM,GAAG,EAAE,CAAC,IAAI,CAAC;AAAA,EACrE;AAAA,EACA,SAAS,CAAC,GAAG,MAAM;AACjB,UAAM,QAAQ,kBAAkB,kGAAkG;AAClI,UAAM,aAAa,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC,EAAE;AAEpC,UAAM,CAAC,MAAM,IAAI;AACjB,UAAM,CAAC,MAAM,IAAI;AAEjB,QAAI,WAAW,QAAQ;AACrB,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,mBAAmB,MAAM,KAAK,CAAC,mBAAmB,MAAM,GAAG;AAC9D,aAAO;AAAA,IACT;AAEA,UAAM,eAAe,SAAS,QAAQ,EAAE;AACxC,UAAM,eAAe,SAAS,QAAQ,EAAE;AAExC,QAAI,MAAM,YAAY,KAAK,MAAM,YAAY,GAAG;AAC9C,aAAO;AAAA,IACT;AAIA,WAAO,eAAe,MAAM,gBAAgB,eAAe,MAAM,eAAe,MAAM;AAAA,EACxF;AACF;",
|
|
6
|
+
"names": []
|
|
7
|
+
}
|
|
@@ -1,48 +1,10 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
var _matchingUtils = require("../../../matching-utils");
|
|
8
|
-
/**
|
|
9
|
-
*
|
|
10
|
-
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
11
|
-
*
|
|
12
|
-
* Melinda record matching modules for Javascript
|
|
13
|
-
*
|
|
14
|
-
* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
|
|
15
|
-
*
|
|
16
|
-
* This file is part of melinda-record-matching-js
|
|
17
|
-
*
|
|
18
|
-
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
19
|
-
* it under the terms of the GNU Lesser General Public License as
|
|
20
|
-
* published by the Free Software Foundation, either version 3 of the
|
|
21
|
-
* License, or (at your option) any later version.
|
|
22
|
-
*
|
|
23
|
-
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
24
|
-
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
25
|
-
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
26
|
-
* GNU Lesser General Public License for more details.
|
|
27
|
-
*
|
|
28
|
-
* You should have received a copy of the GNU Affero General Public License
|
|
29
|
-
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
30
|
-
*
|
|
31
|
-
* @licend The above is the entire license notice
|
|
32
|
-
* for the JavaScript code in this file.
|
|
33
|
-
*
|
|
34
|
-
*/
|
|
35
|
-
// We should also get copyright time and copyright/publication times from 26x
|
|
36
|
-
// see publication-time-allow-cons-years for a version allowing consequent years to match
|
|
37
|
-
var _default = () => ({
|
|
38
|
-
name: 'Publication time',
|
|
39
|
-
extract: ({
|
|
40
|
-
record
|
|
41
|
-
}) => {
|
|
42
|
-
const value = record.get(/^008$/u)?.[0]?.value || undefined;
|
|
43
|
-
return (0, _matchingUtils.testStringOrNumber)(value) ? [String(value).slice(7, 11)] : [];
|
|
1
|
+
import { testStringOrNumber } from "../../../matching-utils.js";
|
|
2
|
+
export default () => ({
|
|
3
|
+
name: "Publication time",
|
|
4
|
+
extract: ({ record }) => {
|
|
5
|
+
const value = record.get(/^008$/u)?.[0]?.value || void 0;
|
|
6
|
+
return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
|
|
44
7
|
},
|
|
45
|
-
compare: (a, b) => a[0] === b[0] ? 0.1 : -1
|
|
8
|
+
compare: (a, b) => a[0] === b[0] ? 0.1 : -1
|
|
46
9
|
});
|
|
47
|
-
|
|
48
|
-
//# sourceMappingURL=publication-time.js.map
|
|
10
|
+
//# sourceMappingURL=publication-time.js.map
|
|
@@ -1 +1,7 @@
|
|
|
1
|
-
{
|
|
1
|
+
{
|
|
2
|
+
"version": 3,
|
|
3
|
+
"sources": ["../../../../src/match-detection/features/bib/publication-time.js"],
|
|
4
|
+
"sourcesContent": ["\n\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// see publication-time-allow-cons-years for a version allowing consequent years to match\n\nexport default () => ({\n name: 'Publication time',\n extract: ({record}) => {\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n },\n compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],
|
|
5
|
+
"mappings": "AAEA,SAAQ,0BAAyB;AAKjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,OAAM,MAAM;AACrB,UAAM,QAAQ,OAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,WAAO,mBAAmB,KAAK,IAAI,CAAC,OAAO,KAAK,EAAE,MAAM,GAAG,EAAE,CAAC,IAAI,CAAC;AAAA,EACrE;AAAA,EACA,SAAS,CAAC,GAAG,MAAM,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,MAAM;AAC3C;",
|
|
6
|
+
"names": []
|
|
7
|
+
}
|
|
@@ -1,45 +1,6 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
value: true
|
|
5
|
-
});
|
|
6
|
-
exports.default = void 0;
|
|
7
|
-
/**
|
|
8
|
-
*
|
|
9
|
-
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
10
|
-
*
|
|
11
|
-
* Melinda record matching modules for Javascript
|
|
12
|
-
*
|
|
13
|
-
* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
|
|
14
|
-
*
|
|
15
|
-
* This file is part of melinda-record-matching-js
|
|
16
|
-
*
|
|
17
|
-
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
18
|
-
* it under the terms of the GNU Lesser General Public License as
|
|
19
|
-
* published by the Free Software Foundation, either version 3 of the
|
|
20
|
-
* License, or (at your option) any later version.
|
|
21
|
-
*
|
|
22
|
-
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
23
|
-
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
24
|
-
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
25
|
-
* GNU Lesser General Public License for more details.
|
|
26
|
-
*
|
|
27
|
-
* You should have received a copy of the GNU Affero General Public License
|
|
28
|
-
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
29
|
-
*
|
|
30
|
-
* @licend The above is the entire license notice
|
|
31
|
-
* for the JavaScript code in this file.
|
|
32
|
-
*
|
|
33
|
-
*/
|
|
34
|
-
// we could handle the case of books/notes
|
|
35
|
-
// Recordtype: LDR/06 - Type of Record
|
|
36
|
-
// Note: currently matchValidator fails all mismatching recordTypes, so this won't actually do much
|
|
37
|
-
var _default = () => ({
|
|
38
|
-
name: 'Record type',
|
|
39
|
-
extract: ({
|
|
40
|
-
record
|
|
41
|
-
}) => record.leader[6] ? [record.leader[6]] : [],
|
|
1
|
+
export default () => ({
|
|
2
|
+
name: "Record type",
|
|
3
|
+
extract: ({ record }) => record.leader[6] ? [record.leader[6]] : [],
|
|
42
4
|
compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5
|
|
43
5
|
});
|
|
44
|
-
|
|
45
|
-
//# sourceMappingURL=record-type.js.map
|
|
6
|
+
//# sourceMappingURL=record-type.js.map
|