@natlibfi/melinda-record-matching 5.0.4 → 5.0.5-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-and-publish.yml +4 -4
- package/dist/candidate-search/query-list/bib.js +84 -67
- package/dist/candidate-search/query-list/bib.js.map +3 -3
- package/dist/candidate-search/query-list/component.js +94 -28
- package/dist/candidate-search/query-list/component.js.map +3 -3
- package/dist/cli.js +12 -4
- package/dist/cli.js.map +2 -2
- package/dist/match-detection/features/bib/f773.js +82 -0
- package/dist/match-detection/features/bib/f773.js.map +7 -0
- package/dist/match-detection/features/bib/index.js +1 -1
- package/dist/match-detection/features/bib/index.js.map +2 -2
- package/dist/match-detection/features/bib/index.test.js +1 -1
- package/dist/match-detection/features/bib/index.test.js.map +2 -2
- package/dist/match-detection/features/bib/isbn.js +3 -13
- package/dist/match-detection/features/bib/isbn.js.map +2 -2
- package/dist/match-detection/features/bib/issn.js +2 -16
- package/dist/match-detection/features/bib/issn.js.map +2 -2
- package/dist/match-detection/features/bib/language.js +12 -22
- package/dist/match-detection/features/bib/language.js.map +2 -2
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js +6 -2
- package/dist/match-detection/features/bib/publication-time-allow-cons-years.js.map +2 -2
- package/dist/match-detection/features/bib/publication-time.js +62 -16
- package/dist/match-detection/features/bib/publication-time.js.map +2 -2
- package/dist/match-detection/features/bib/title.js +76 -26
- package/dist/match-detection/features/bib/title.js.map +3 -3
- package/dist/match-detection/index.js +8 -3
- package/dist/match-detection/index.js.map +2 -2
- package/dist/matching-utils.js +14 -0
- package/dist/matching-utils.js.map +2 -2
- package/package.json +7 -10
- package/src/candidate-search/query-list/bib.js +100 -77
- package/src/candidate-search/query-list/component.js +100 -23
- package/src/cli.js +8 -4
- package/src/match-detection/features/bib/f773.js +118 -0
- package/src/match-detection/features/bib/index.js +1 -1
- package/src/match-detection/features/bib/index.test.js +1 -1
- package/src/match-detection/features/bib/isbn.js +3 -17
- package/src/match-detection/features/bib/issn.js +3 -32
- package/src/match-detection/features/bib/language.js +15 -26
- package/src/match-detection/features/bib/publication-time-allow-cons-years.js +7 -4
- package/src/match-detection/features/bib/publication-time.js +76 -30
- package/src/match-detection/features/bib/title.js +112 -34
- package/src/match-detection/index.js +9 -3
- package/src/matching-utils.js +17 -0
- package/dist/match-detection/features/bib/title-version-original.js +0 -37
- package/dist/match-detection/features/bib/title-version-original.js.map +0 -7
- package/src/match-detection/features/bib/title-version-original.js +0 -52
|
@@ -15,7 +15,7 @@ jobs:
|
|
|
15
15
|
|
|
16
16
|
strategy:
|
|
17
17
|
matrix:
|
|
18
|
-
node-version: [
|
|
18
|
+
node-version: [24.x]
|
|
19
19
|
# See supported Node.js release schedule at https://nodejs.org/en/about/releases/
|
|
20
20
|
|
|
21
21
|
steps:
|
|
@@ -36,7 +36,7 @@ jobs:
|
|
|
36
36
|
njsscan:
|
|
37
37
|
name: Njsscan check
|
|
38
38
|
runs-on: ubuntu-latest
|
|
39
|
-
container: node:
|
|
39
|
+
container: node:24
|
|
40
40
|
|
|
41
41
|
steps:
|
|
42
42
|
- name: Checkout the code
|
|
@@ -50,7 +50,7 @@ jobs:
|
|
|
50
50
|
license-scan:
|
|
51
51
|
name: License compliance check
|
|
52
52
|
runs-on: ubuntu-latest
|
|
53
|
-
container: node:
|
|
53
|
+
container: node:24
|
|
54
54
|
|
|
55
55
|
steps:
|
|
56
56
|
- uses: actions/checkout@v6
|
|
@@ -68,7 +68,7 @@ jobs:
|
|
|
68
68
|
# Setup .npmrc file to publish to npm
|
|
69
69
|
- uses: actions/setup-node@v6
|
|
70
70
|
with:
|
|
71
|
-
node-version: '
|
|
71
|
+
node-version: '24.x'
|
|
72
72
|
registry-url: 'https://registry.npmjs.org'
|
|
73
73
|
- name: Update npm
|
|
74
74
|
run: npm install -g npm@latest
|
|
@@ -1,7 +1,11 @@
|
|
|
1
1
|
import createDebugLogger from "debug";
|
|
2
2
|
import { toQueries } from "../candidate-search-utils.js";
|
|
3
3
|
import { getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber } from "../../matching-utils.js";
|
|
4
|
+
import { extractHostMelindaIdsFromField, extractPublicationYearFrom773, getHostMelindaFields } from "./component.js";
|
|
5
|
+
import { fieldToString } from "@natlibfi/marc-record-validators-melinda";
|
|
6
|
+
import { getTitle } from "../../match-detection/features/bib/title.js";
|
|
4
7
|
const debug = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query");
|
|
8
|
+
const IS_OK_ALONE_THRESHOLD = 5;
|
|
5
9
|
export function bibSourceIds(record) {
|
|
6
10
|
const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:source-ids");
|
|
7
11
|
const debugData = debug2.extend("data");
|
|
@@ -65,23 +69,18 @@ export function bibHostComponents(record) {
|
|
|
65
69
|
const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:bibHostComponents");
|
|
66
70
|
const debugData = debug2.extend("data");
|
|
67
71
|
debug2(`Creating queries for hostIds`);
|
|
68
|
-
const
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
if (testStringOrNumber(value) && /^\(FIN01\)/u.test(String(value))) {
|
|
79
|
-
return String(value).replace(/^\(FIN01\)/u, "");
|
|
80
|
-
}
|
|
81
|
-
return false;
|
|
82
|
-
}
|
|
83
|
-
return false;
|
|
72
|
+
const f773s = getHostMelindaFields(record);
|
|
73
|
+
if (f773s.length !== 1) {
|
|
74
|
+
return [];
|
|
75
|
+
}
|
|
76
|
+
const [id] = extractHostMelindaIdsFromField(f773s[0]);
|
|
77
|
+
debug2(`Found id: ${JSON.stringify(id)}`);
|
|
78
|
+
const year = extractPublicationYearFrom773(f773s[0]);
|
|
79
|
+
debug2(`${fieldToString(f773s[0])} => ${year ? year : "N/A"}`);
|
|
80
|
+
if (year) {
|
|
81
|
+
return [`dc.date=${year} AND melinda.partsofhost=${id}`];
|
|
84
82
|
}
|
|
83
|
+
return [`melinda.partsofhost=${id}`];
|
|
85
84
|
}
|
|
86
85
|
export function bibTitle(record) {
|
|
87
86
|
return bibTitleAuthorPublisher({ record, onlyTitleLength: 5 });
|
|
@@ -100,74 +99,86 @@ export function bibTitleAuthorYearAlternates(record) {
|
|
|
100
99
|
debug(`${JSON.stringify(origQueryList)}`);
|
|
101
100
|
return { queryList: Array.from(origQueryList).reverse(), queryListType: "alternates" };
|
|
102
101
|
}
|
|
102
|
+
function getTitleForQuery(record) {
|
|
103
|
+
const title = getTitle(record, ["a"]);
|
|
104
|
+
if (title) {
|
|
105
|
+
const nWords = title.split(" ");
|
|
106
|
+
if (nWords < 3 || title.length < 12) {
|
|
107
|
+
const [f245] = record.get("245");
|
|
108
|
+
if (f245 && f245.subfields.find((sf) => sf.code === "a" && sf.value.match(/ =$/u)) && title.length >= IS_OK_ALONE_THRESHOLD) {
|
|
109
|
+
return title;
|
|
110
|
+
}
|
|
111
|
+
return getTitle(record, ["a", "b"]);
|
|
112
|
+
}
|
|
113
|
+
}
|
|
114
|
+
return title;
|
|
115
|
+
}
|
|
116
|
+
function dcTitle(record, onlyTitleLength, alternates = false) {
|
|
117
|
+
const title = getTitleForQuery(record);
|
|
118
|
+
if (!testStringOrNumber(title)) {
|
|
119
|
+
return [];
|
|
120
|
+
}
|
|
121
|
+
const formatted = getFormattedTitle(title);
|
|
122
|
+
const useWordSearch = checkUseWordSearch(formatted);
|
|
123
|
+
if (formatted.length >= onlyTitleLength && !alternates) {
|
|
124
|
+
return [`dc.title="${useWordSearch ? "" : "^"}${formatted}*"`, formatted, true];
|
|
125
|
+
}
|
|
126
|
+
const queryIsOkAlone = !useWordSearch && formatted.length >= IS_OK_ALONE_THRESHOLD;
|
|
127
|
+
return [`dc.title="${useWordSearch || !queryIsOkAlone ? "" : "^"}${formatted}${queryIsOkAlone ? "*" : ""}"`, formatted, queryIsOkAlone];
|
|
128
|
+
function getFormattedTitle(title2) {
|
|
129
|
+
const formatted2 = String(title2).replace(/[\-+ !"(){}\[\]<>;:.?/@*%=^_`~]/gu, " ").replace(/[^\w\s\p{Alphabetic}]/gu, "").replace(/ +/gu, " ").trim().replace(/^(.{30}\S*) .*$/, "$1").trim();
|
|
130
|
+
return formatted2;
|
|
131
|
+
}
|
|
132
|
+
}
|
|
103
133
|
export function bibTitleAuthorPublisher({ record, onlyTitleLength, addYear = false, alternates = false, alternateQueries = [] }) {
|
|
104
134
|
debug(`bibTitleAuthorPublisher, onlyTitleLength: ${onlyTitleLength}, addYear: ${addYear}, alternates: ${alternates}`);
|
|
105
|
-
const
|
|
106
|
-
if (
|
|
107
|
-
|
|
108
|
-
const formatted2 = String(title2).replace(/[\-+ !"(){}\[\]<>;:.?/@*%=^_`~]/gu, " ").replace(/[^\w\s\p{Alphabetic}]/gu, "").replace(/ +/gu, " ").trim().replace(/^(.{30}\S*) .*$/, "$1").trim();
|
|
109
|
-
return formatted2;
|
|
110
|
-
};
|
|
111
|
-
const formatted = getFormatted(title);
|
|
112
|
-
const useWordSearch = checkUseWordSearch(formatted);
|
|
113
|
-
if (formatted.length >= onlyTitleLength && !alternates) {
|
|
114
|
-
return [`dc.title="${useWordSearch ? "" : "^"}${formatted}*"`];
|
|
115
|
-
}
|
|
116
|
-
const queryIsOkAlone = formatted.length >= 5;
|
|
117
|
-
const query = `dc.title="${useWordSearch || !queryIsOkAlone ? "" : "^"}${formatted}${queryIsOkAlone ? "*" : ""}"`;
|
|
118
|
-
debug(`query: ${query}`);
|
|
119
|
-
const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;
|
|
120
|
-
return addAuthorsToSearch({ query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries });
|
|
135
|
+
const [query, formatted, queryIsOkAlone] = dcTitle(record, onlyTitleLength, alternates);
|
|
136
|
+
if (query === void 0) {
|
|
137
|
+
return [];
|
|
121
138
|
}
|
|
122
|
-
|
|
123
|
-
|
|
139
|
+
debug(`query: ${query}`);
|
|
140
|
+
if (formatted.length >= onlyTitleLength && !alternates) {
|
|
141
|
+
return [query];
|
|
142
|
+
}
|
|
143
|
+
const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;
|
|
144
|
+
return addAuthorsToSearch({ query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries });
|
|
145
|
+
function addAuthorsToSearch({ query: query2, queryIsOkAlone: queryIsOkAlone2 = false, addYear: addYear2 = false, alternates: alternates2 = false, alternateQueries: alternateQueries2 = [] }) {
|
|
124
146
|
debug("addAuthorsToSearch");
|
|
125
147
|
const [authorQuery] = bibAuthors(record);
|
|
126
148
|
if (authorQuery !== void 0) {
|
|
127
149
|
if (addYear2) {
|
|
128
|
-
const
|
|
129
|
-
return addYearToSearch({ query: `${authorQuery} AND ${
|
|
150
|
+
const newAlternateQueries2 = alternates2 ? [...alternateQueries2, `${authorQuery} AND ${query2}`] : alternateQueries2;
|
|
151
|
+
return addYearToSearch({ query: `${authorQuery} AND ${query2}`, queryIsOkAlone: true, alternates: alternates2, alternateQueries: newAlternateQueries2 });
|
|
130
152
|
}
|
|
131
|
-
return alternates2 ? alternateQueries2 : [`${authorQuery} AND ${
|
|
153
|
+
return alternates2 ? alternateQueries2 : [`${authorQuery} AND ${query2}`];
|
|
132
154
|
}
|
|
133
|
-
return addPublisherToSearch({ query, queryIsOkAlone, addYear: addYear2, alternates: alternates2, alternateQueries: alternateQueries2 });
|
|
155
|
+
return addPublisherToSearch({ query: query2, queryIsOkAlone: queryIsOkAlone2, addYear: addYear2, alternates: alternates2, alternateQueries: alternateQueries2 });
|
|
134
156
|
}
|
|
135
|
-
function addPublisherToSearch({ query, queryIsOkAlone = false, addYear: addYear2 = false, alternates: alternates2 = false, alternateQueries: alternateQueries2 = [] }) {
|
|
157
|
+
function addPublisherToSearch({ query: query2, queryIsOkAlone: queryIsOkAlone2 = false, addYear: addYear2 = false, alternates: alternates2 = false, alternateQueries: alternateQueries2 = [] }) {
|
|
136
158
|
const [publisherQuery] = bibPublishers(record);
|
|
137
159
|
if (publisherQuery !== void 0) {
|
|
138
160
|
if (addYear2) {
|
|
139
|
-
const
|
|
140
|
-
return addYearToSearch({ query: `${publisherQuery} AND ${
|
|
161
|
+
const newAlternateQueries2 = alternates2 ? [...alternateQueries2, `${publisherQuery} AND ${query2}`] : alternateQueries2;
|
|
162
|
+
return addYearToSearch({ query: `${publisherQuery} AND ${query2}`, queryIsOkAlone: true, alternates: alternates2, alternateQueries: newAlternateQueries2 });
|
|
141
163
|
}
|
|
142
|
-
return alternates2 ? alternateQueries2 : [`${publisherQuery} AND ${
|
|
164
|
+
return alternates2 ? alternateQueries2 : [`${publisherQuery} AND ${query2}`];
|
|
143
165
|
}
|
|
144
|
-
if (
|
|
145
|
-
return alternates2 ? alternateQueries2 : [`${
|
|
166
|
+
if (queryIsOkAlone2 && !addYear2) {
|
|
167
|
+
return alternates2 ? alternateQueries2 : [`${query2}`];
|
|
146
168
|
}
|
|
147
|
-
return addYearToSearch({ query, queryIsOkAlone, alternates: alternates2, alternateQueries: alternateQueries2 });
|
|
169
|
+
return addYearToSearch({ query: query2, queryIsOkAlone: queryIsOkAlone2, alternates: alternates2, alternateQueries: alternateQueries2 });
|
|
148
170
|
}
|
|
149
|
-
function addYearToSearch({ query, queryIsOkAlone = false, alternates: alternates2 = false, alternateQueries: alternateQueries2 = [] }) {
|
|
171
|
+
function addYearToSearch({ query: query2, queryIsOkAlone: queryIsOkAlone2 = false, alternates: alternates2 = false, alternateQueries: alternateQueries2 = [] }) {
|
|
150
172
|
const [yearQuery] = bibYear(record);
|
|
151
173
|
if (yearQuery !== void 0) {
|
|
152
|
-
const
|
|
153
|
-
return alternates2 ?
|
|
174
|
+
const yearAndOtherQueries = `${yearQuery} AND ${query2}`;
|
|
175
|
+
return alternates2 ? [...alternateQueries2, yearAndOtherQueries] : [yearAndOtherQueries];
|
|
154
176
|
}
|
|
155
|
-
if (
|
|
156
|
-
return alternates2 ? alternateQueries2 : [`${
|
|
177
|
+
if (queryIsOkAlone2) {
|
|
178
|
+
return alternates2 ? alternateQueries2 : [`${query2}`];
|
|
157
179
|
}
|
|
158
180
|
return [];
|
|
159
181
|
}
|
|
160
|
-
function getTitle() {
|
|
161
|
-
const [field] = record.get(/^245$/u);
|
|
162
|
-
if (field) {
|
|
163
|
-
const titleString = field.subfields.filter(({ code }) => ["a", "b"].includes(code)).map(({ value }) => testStringOrNumber(value) ? String(value) : "").filter((value) => value).join(" ");
|
|
164
|
-
if (/^[1-9]$/u.test(field.ind2)) {
|
|
165
|
-
return titleString.slice(parseInt(field.ind2));
|
|
166
|
-
}
|
|
167
|
-
return titleString;
|
|
168
|
-
}
|
|
169
|
-
return false;
|
|
170
|
-
}
|
|
171
182
|
}
|
|
172
183
|
export function bibAuthors(record) {
|
|
173
184
|
const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors");
|
|
@@ -200,14 +211,17 @@ export function bibPublishers(record) {
|
|
|
200
211
|
const publisher = getPublisher(record);
|
|
201
212
|
if (testStringOrNumber(publisher)) {
|
|
202
213
|
const formatted = String(publisher).replace(/[^\w\s\p{Alphabetic}]/gu, "").replace(/ +/gu, " ").trim().slice(0, 30).trim();
|
|
203
|
-
debugData(`Publisher string: ${formatted}`);
|
|
214
|
+
debugData(`Publisher string: '${formatted}'`);
|
|
215
|
+
if (formatted.length === 0) {
|
|
216
|
+
return [];
|
|
217
|
+
}
|
|
204
218
|
return [`dc.publisher="${formatted}"`];
|
|
205
219
|
}
|
|
206
220
|
return [];
|
|
207
221
|
function getPublisher(record2) {
|
|
208
|
-
const [field] = record2.get(/^(
|
|
222
|
+
const [field] = record2.get(/^26[04]$/u).filter((f) => f.subfields.some((sf) => sf.code === "b"));
|
|
209
223
|
if (field) {
|
|
210
|
-
const publisherString = field.subfields.filter(({ code }) => ["b"].includes(code)).map(({ value }) => testStringOrNumber(value) ? String(value) : "").filter((value) => value).join(" ");
|
|
224
|
+
const publisherString = field.subfields.filter(({ code }) => ["b"].includes(code)).map(({ value }) => testStringOrNumber(value) ? String(value) : "").filter((value) => value).join(" ").trim();
|
|
211
225
|
return publisherString;
|
|
212
226
|
}
|
|
213
227
|
return false;
|
|
@@ -224,13 +238,16 @@ export function bibYear(record) {
|
|
|
224
238
|
return [];
|
|
225
239
|
function getYear(record2) {
|
|
226
240
|
const [f008] = record2.get(/^008$/u);
|
|
227
|
-
if (f008
|
|
228
|
-
debug2("f008 missing");
|
|
241
|
+
if (!f008 || !f008.value || f008.value.length < 11) {
|
|
242
|
+
debug2("f008 missing/crappy");
|
|
229
243
|
return false;
|
|
230
244
|
}
|
|
231
245
|
debugData(`f008: ${JSON.stringify(f008)}`);
|
|
232
246
|
const { value } = f008;
|
|
233
|
-
|
|
247
|
+
if (value === "||||" || value === " ") {
|
|
248
|
+
return false;
|
|
249
|
+
}
|
|
250
|
+
return value.slice(7, 11);
|
|
234
251
|
}
|
|
235
252
|
}
|
|
236
253
|
export function checkUseWordSearch(formatted) {
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"version": 3,
|
|
3
3
|
"sources": ["../../../src/candidate-search/query-list/bib.js"],
|
|
4
|
-
"sourcesContent": ["import createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query');\n\nexport function bibSourceIds(record) {\n\n /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n SID__ $c 123457 $b helka\n SID__ $c (ANDL100020)1077305 $b sata\n SID__ $c VER999999 $ FI-KV\n SID__ $c /10024/508126 $ REPO_THESEUS\n\n In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n 1234567helka\n 1077305sata\n VER999999FIKV\n /10024/508126REPO_THESEUS\n\n Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n after the matching record is removed from the local db.\n\n */\n\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n const debugData = debug.extend('data');\n //const debugInfo = debug.extend('info');\n\n debug(`Creating queries for sourceid's`);\n\n const fSids = record.get('SID');\n debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n function toSidQueries(fSids) {\n debug(`Creating actual queries for sourceid's`);\n\n const sidStrings = getSidStrings(fSids);\n\n if (sidStrings.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n return sidQueries;\n\n function getSidStrings(fSids) {\n debug(`Getting Sid strings from SID fields`);\n\n // Map SID fields to valid sidStrings, filter out empty strings\n const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n return sidStrings;\n\n function toSidString(field) {\n debug(`Getting string from a field`);\n\n return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n function createSidString(field) {\n debug(`Creating string from a field`);\n const [sfC] = getSubfieldValues(field, 'c');\n const [sfB] = getSubfieldValues(field, 'b');\n\n const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n return cleanedSfC.concat(cleanedSfB);\n }\n\n function removeSourcePrefix(subfieldValue) {\n const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n return normalizedValue;\n }\n\n function normalizeSidSubfieldValue(subfieldValue) {\n debugData(`Normalizing ${subfieldValue}`);\n const normalizeAwayRegex = (/[- ]/u);\n return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n }\n\n }\n }\n }\n}\n\nexport function bibMelindaIds(record) {\n // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n const debugData = debug.extend('data');\n debug(`Creating queries for MelindaIds`);\n\n // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n // Both (FI-MELINDA)- and FCC-prefixed forms are found\n // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n const melindaIds = getMelindaIdsF035(record);\n\n debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n if (melindaIds.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibHostComponents');\n const debugData = debug.extend('data');\n debug(`Creating queries for hostIds`);\n\n const id = getHostId();\n debugData(`Found id: ${JSON.stringify(id)}`);\n\n return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];\n\n function getHostId() {\n const [field] = record.get(/^773$/u);\n\n if (field) {\n const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n if (testStringOrNumber(value) && (/^\\(FI-MELINDA\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FI-MELINDA\\)/u, '');\n }\n\n if (testStringOrNumber(value) && (/^\\(FIN01\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FIN01\\)/u, '');\n }\n\n return false;\n }\n return false;\n }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\nexport function bibTitle(record) {\n // We get author/publisher only when formatted title is shorter than 5 chars\n return bibTitleAuthorPublisher({record, onlyTitleLength: 5});\n}\n\nexport function bibTitleAuthor(record) {\n debug('bibTitleAuthor');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n return bibTitleAuthorPublisher({record, onlyTitleLength: 100});\n}\n\nexport function bibTitleAuthorYear(record) {\n debug('bibTitleAuthorYearAlternates');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n\n return bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true});\n}\n\nexport function bibTitleAuthorYearAlternates(record) {\n debug('bibTitleAuthorYearAlternates');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n const origQueryList = bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true, alternates: true, alternateQueries: []});\n debug(`${JSON.stringify(origQueryList)}`);\n return {queryList: Array.from(origQueryList).reverse(), queryListType: 'alternates'};\n}\n\nexport function bibTitleAuthorPublisher({record, onlyTitleLength, addYear = false, alternates = false, alternateQueries = []}) {\n debug(`bibTitleAuthorPublisher, onlyTitleLength: ${onlyTitleLength}, addYear: ${addYear}, alternates: ${alternates}`);\n const title = getTitle();\n if (testStringOrNumber(title)) {\n const formatted = getFormatted(title);\n\n // use word search for titles starting with a boolean\n const useWordSearch = checkUseWordSearch(formatted);\n // Prevent too many matches / SRU crashing by having a minimum length\n // Note that currently this fails matching if there are no matches from previous matchers\n if (formatted.length >= onlyTitleLength && !alternates) {\n return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n }\n const queryIsOkAlone = formatted.length >= 5;\n\n // use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]\n const query = `dc.title=\"${useWordSearch || !queryIsOkAlone ? '' : '^'}${formatted}${queryIsOkAlone ? '*' : ''}\"`;\n debug(`query: ${query}`);\n const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;\n\n return addAuthorsToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries});\n\n function getFormatted(title) {\n const formatted = String(title)\n .replace(/[\\-+ !\"(){}\\[\\]<>;:.?/@*%=^_`~]/gu, ' ')\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .replace(/^(.{30}\\S*) .*$/, \"$1\")\n .trim();\n\n return formatted;\n }\n }\n\n return [];\n\n function addAuthorsToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n debug('addAuthorsToSearch');\n const [authorQuery] = bibAuthors(record);\n if (authorQuery !== undefined) {\n if (addYear) {\n const newAlternateQueries = alternates ? [...alternateQueries, `${authorQuery} AND ${query}`] : alternateQueries;\n return addYearToSearch({query: `${authorQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n }\n return alternates ? alternateQueries : [`${authorQuery} AND ${query}`];\n }\n return addPublisherToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries});\n //return [];\n }\n\n function addPublisherToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n const [publisherQuery] = bibPublishers(record);\n if (publisherQuery !== undefined) {\n if (addYear) {\n const newAlternateQueries = alternates ? [...alternateQueries, `${publisherQuery} AND ${query}`] : alternateQueries;\n return addYearToSearch({query: `${publisherQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n }\n return alternates ? alternateQueries : [`${publisherQuery} AND ${query}`];\n }\n if (queryIsOkAlone && !addYear) {\n return alternates ? alternateQueries : [`${query}`];\n }\n return addYearToSearch({query, queryIsOkAlone, alternates, alternateQueries});\n }\n\n function addYearToSearch({query, queryIsOkAlone = false, alternates = false, alternateQueries = []}) {\n const [yearQuery] = bibYear(record);\n if (yearQuery !== undefined) {\n const newAlternateQueries = alternates ? [...alternateQueries, `${yearQuery} AND ${query}`] : alternateQueries;\n return alternates ? newAlternateQueries : [`${yearQuery} AND ${query}`];\n }\n if (queryIsOkAlone) {\n return alternates ? alternateQueries : [`${query}`];\n }\n return [];\n }\n\n function getTitle() {\n const [field] = record.get(/^245$/u);\n\n if (field) {\n const titleString = field.subfields\n //.filter(({code}) => ['a', 'b', 'n', 'p'].includes(code))\n .filter(({code}) => ['a', 'b'].includes(code))\n //.filter(({code}) => ['a'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n\n if (/^[1-9]$/u.test(field.ind2)) { // Skip non-filing characters\n return titleString.slice(parseInt(field.ind2));\n }\n return titleString;\n }\n return false;\n }\n}\n\nexport function bibAuthors(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');\n const debugData = debug.extend('data');\n debug(`Creating query for the first author`);\n //debugData(record);\n\n const author = getAuthor(record);\n\n if (testStringOrNumber(author)) {\n const formatted = String(author)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n // use word search for authors starting with a boolean\n const useWordSearch = checkUseWordSearch(formatted);\n // Prevent too many matches by having a minimum length\n debugData(`Author string: ${formatted}`);\n if (formatted.length >= 5) {\n return [`dc.author=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n }\n return [];\n }\n\n return [];\n\n function getAuthor(record) {\n //debugData(record);\n // eslint-disable-next-line prefer-named-capture-group\n const [field] = record.get(/^(100)|(110)|(111)|(700)|(710)|(711)$/u);\n //debugData(field);\n\n if (field) {\n const authorString = field.subfields\n .filter(({code}) => ['a'].includes(code)) // We might use different subfield code sets for X00, X10 and X11?\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n return authorString;\n }\n return false;\n }\n}\n\nexport function bibPublishers(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');\n const debugData = debug.extend('data');\n debug(`Creating query for the first publisher`);\n //debugData(record);\n\n const publisher = getPublisher(record);\n if (testStringOrNumber(publisher)) {\n const formatted = String(publisher)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n debugData(`Publisher string: ${formatted}`);\n // use non-wildcard word search from dc.publisher\n return [`dc.publisher=\"${formatted}\"`];\n }\n\n return [];\n\n function getPublisher(record) {\n //debugData(record);\n const [field] = record.get(/^(?:260)|(?:264)$/u);\n //debugData(field);\n\n if (field) {\n const publisherString = field.subfields\n .filter(({code}) => ['b'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n return publisherString;\n }\n return false;\n }\n}\n\nexport function bibYear(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibYear');\n const debugData = debug.extend('data');\n debug(`Creating query for the publishing year`);\n\n const year = getYear(record);\n if (year) {\n return [`dc.date=\"${year}\"`];\n }\n return [];\n\n function getYear(record) {\n const [f008] = record.get(/^008$/u);\n if (f008 === undefined) {\n debug('f008 missing');\n return false;\n }\n\n debugData(`f008: ${JSON.stringify(f008)}`);\n const {value} = f008;\n return testStringOrNumber(value) ? String(value).slice(7, 11) : undefined;\n }\n}\n\nexport function checkUseWordSearch(formatted) {\n const lowercased = formatted.toLowerCase();\n // Note: add a space to startWords to catch just actual boolean words\n const booleanStartWords = ['and ', 'or ', 'nor ', 'not '];\n return booleanStartWords.some(word => lowercased.startsWith(word));\n}\n\nexport function bibStandardIdentifiers(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n const debugData = debug.extend('data');\n debug(`Creating queries for standard identifiers`);\n\n // DEVELOP: should we query also f015 and f028?\n\n const fields = record.get(/^(?<def>020|022|024)$/u);\n const identifiers = [].concat(...fields.map(toIdentifiers));\n const uniqueIdentifiers = [...new Set(identifiers)];\n\n debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n if (uniqueIdentifiers.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n function toIdentifiers({tag, subfields}) {\n const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n if (tag === '022') {\n return subfields\n .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n\n if (tag === '020') {\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n}\n"],
|
|
5
|
-
"mappings": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,gCAAgC,mBAAmB,0BAAyB;
|
|
6
|
-
"names": ["debug", "fSids", "sidStrings", "field", "title", "formatted", "addYear", "alternates", "alternateQueries", "record"]
|
|
4
|
+
"sourcesContent": ["import createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils.js';\nimport {extractHostMelindaIdsFromField, extractPublicationYearFrom773, getHostMelindaFields} from './component.js';\nimport {fieldToString} from '@natlibfi/marc-record-validators-melinda';\nimport {getTitle} from '../../match-detection/features/bib/title.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query');\n\nconst IS_OK_ALONE_THRESHOLD = 5;\n\nexport function bibSourceIds(record) {\n\n /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n SID__ $c 123457 $b helka\n SID__ $c (ANDL100020)1077305 $b sata\n SID__ $c VER999999 $ FI-KV\n SID__ $c /10024/508126 $ REPO_THESEUS\n\n In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n 1234567helka\n 1077305sata\n VER999999FIKV\n /10024/508126REPO_THESEUS\n\n Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n after the matching record is removed from the local db.\n\n */\n\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n const debugData = debug.extend('data');\n //const debugInfo = debug.extend('info');\n\n debug(`Creating queries for sourceid's`);\n\n const fSids = record.get('SID');\n debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n function toSidQueries(fSids) {\n debug(`Creating actual queries for sourceid's`);\n\n const sidStrings = getSidStrings(fSids);\n\n if (sidStrings.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n return sidQueries;\n\n function getSidStrings(fSids) {\n debug(`Getting Sid strings from SID fields`);\n\n // Map SID fields to valid sidStrings, filter out empty strings\n const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n return sidStrings;\n\n function toSidString(field) {\n debug(`Getting string from a field`);\n\n return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n function createSidString(field) {\n debug(`Creating string from a field`);\n const [sfC] = getSubfieldValues(field, 'c');\n const [sfB] = getSubfieldValues(field, 'b');\n\n const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n return cleanedSfC.concat(cleanedSfB);\n }\n\n function removeSourcePrefix(subfieldValue) {\n const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n return normalizedValue;\n }\n\n function normalizeSidSubfieldValue(subfieldValue) {\n debugData(`Normalizing ${subfieldValue}`);\n const normalizeAwayRegex = (/[- ]/u);\n return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n }\n\n }\n }\n }\n}\n\nexport function bibMelindaIds(record) {\n // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n const debugData = debug.extend('data');\n debug(`Creating queries for MelindaIds`);\n\n // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n // Both (FI-MELINDA)- and FCC-prefixed forms are found\n // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n const melindaIds = getMelindaIdsF035(record);\n\n debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n if (melindaIds.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibHostComponents');\n const debugData = debug.extend('data');\n debug(`Creating queries for hostIds`);\n const f773s = getHostMelindaFields(record);\n if (f773s.length !== 1) { // This needs some thinking...\n return [];\n }\n\n const [id] = extractHostMelindaIdsFromField(f773s[0]);\n debug(`Found id: ${JSON.stringify(id)}`);\n\n // NB! record.isCR() does not work if LDR/07=a, so we get year from 773$g!\n const year = extractPublicationYearFrom773(f773s[0]);\n debug(`${fieldToString(f773s[0])} => ${year ? year : 'N/A'}`);\n if (year) {\n return [`dc.date=${year} AND melinda.partsofhost=${id}`];\n }\n\n return [`melinda.partsofhost=${id}`];\n\n}\n\n\n\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\nexport function bibTitle(record) {\n // We get author/publisher only when formatted title is shorter than 5 chars\n return bibTitleAuthorPublisher({record, onlyTitleLength: 5});\n}\n\nexport function bibTitleAuthor(record) {\n debug('bibTitleAuthor');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n return bibTitleAuthorPublisher({record, onlyTitleLength: 100});\n}\n\nexport function bibTitleAuthorYear(record) {\n debug('bibTitleAuthorYearAlternates');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n\n return bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true});\n}\n\nexport function bibTitleAuthorYearAlternates(record) {\n debug('bibTitleAuthorYearAlternates');\n // We use onlyTitleLength that is longer than our formatted length to\n // get an author or an publisher always\n const origQueryList = bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true, alternates: true, alternateQueries: []});\n debug(`${JSON.stringify(origQueryList)}`);\n return {queryList: Array.from(origQueryList).reverse(), queryListType: 'alternates'};\n}\n\nfunction getTitleForQuery(record) {\n const title = getTitle(record, ['a']);\n if (title) {\n const nWords = title.split(' ');\n // If lone $a is deemed too short, fetch $b as well:\n // (NV: I've seen pairs with 245$a-only vs 245$a$b, so I'd like to use $a only if it is long enough)\n\n if (nWords < 3 || title.length < 12) {\n const [f245] = record.get('245');\n // If punctuation is ' =' I think that f245$a is good enough despite shortness. Trying to balance between two bad situations...\n // \"Suo (= short title) siell\u00E4, vetel\u00E4 (= missing $b name in another language) t\u00E4\u00E4ll\u00E4...\"\n if (f245 && f245.subfields.find(sf => sf.code === 'a' && sf.value.match(/ =$/u)) && title.length >= IS_OK_ALONE_THRESHOLD) {\n return title;\n }\n return getTitle(record, ['a', 'b']); // Try to get a longer title\n }\n }\n return title;\n}\n\nfunction dcTitle(record, onlyTitleLength, alternates = false) {\n const title = getTitleForQuery(record);\n if (!testStringOrNumber(title)) {\n return [];\n }\n\n const formatted = getFormattedTitle(title);\n\n // use word search for titles starting with a boolean\n const useWordSearch = checkUseWordSearch(formatted);\n // Prevent too many matches / SRU crashing by having a minimum length\n // Note that currently this fails matching if there are no matches from previous matchers\n if (formatted.length >= onlyTitleLength && !alternates) {\n return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`, formatted, true];\n }\n\n const queryIsOkAlone = !useWordSearch && formatted.length >= IS_OK_ALONE_THRESHOLD;\n\n // use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]\n return [`dc.title=\"${useWordSearch || !queryIsOkAlone ? '' : '^'}${formatted}${queryIsOkAlone ? '*' : ''}\"`, formatted, queryIsOkAlone];\n\n function getFormattedTitle(title) {\n const formatted = String(title)\n .replace(/[\\-+ !\"(){}\\[\\]<>;:.?/@*%=^_`~]/gu, ' ')\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '') // Apparently matches to/works with non-aplhabetic scripts such as Chinese as well\n .replace(/ +/gu, ' ') // Clean up concurrent spaces from eg. subfield changes\n .trim()\n .replace(/^(.{30}\\S*) .*$/, \"$1\")\n .trim();\n\n return formatted;\n }\n}\n\nexport function bibTitleAuthorPublisher({record, onlyTitleLength, addYear = false, alternates = false, alternateQueries = []}) {\n debug(`bibTitleAuthorPublisher, onlyTitleLength: ${onlyTitleLength}, addYear: ${addYear}, alternates: ${alternates}`);\n const [query, formatted, queryIsOkAlone] = dcTitle(record, onlyTitleLength, alternates);\n if (query === undefined) {\n return [];\n }\n\n debug(`query: ${query}`);\n\n // Prevent too many matches / SRU crashing by having a minimum length\n // Note that currently this fails matching if there are no matches from previous matchers\n if (formatted.length >= onlyTitleLength && !alternates) {\n return [query];\n }\n\n const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;\n\n return addAuthorsToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries});\n\n function addAuthorsToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n debug('addAuthorsToSearch');\n const [authorQuery] = bibAuthors(record);\n if (authorQuery !== undefined) {\n if (addYear) {\n const newAlternateQueries = alternates ? [...alternateQueries, `${authorQuery} AND ${query}`] : alternateQueries;\n return addYearToSearch({query: `${authorQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n }\n return alternates ? alternateQueries : [`${authorQuery} AND ${query}`];\n }\n return addPublisherToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries});\n //return [];\n }\n\n function addPublisherToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n const [publisherQuery] = bibPublishers(record);\n if (publisherQuery !== undefined) {\n if (addYear) {\n const newAlternateQueries = alternates ? [...alternateQueries, `${publisherQuery} AND ${query}`] : alternateQueries;\n return addYearToSearch({query: `${publisherQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n }\n return alternates ? alternateQueries : [`${publisherQuery} AND ${query}`];\n }\n if (queryIsOkAlone && !addYear) {\n return alternates ? alternateQueries : [`${query}`];\n }\n return addYearToSearch({query, queryIsOkAlone, alternates, alternateQueries});\n }\n\n function addYearToSearch({query, queryIsOkAlone = false, alternates = false, alternateQueries = []}) {\n const [yearQuery] = bibYear(record);\n if (yearQuery !== undefined) {\n const yearAndOtherQueries = `${yearQuery} AND ${query}`;\n return alternates ? [...alternateQueries, yearAndOtherQueries] : [yearAndOtherQueries];\n }\n if (queryIsOkAlone) {\n return alternates ? alternateQueries : [`${query}`];\n }\n return [];\n }\n\n}\n\nexport function bibAuthors(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');\n const debugData = debug.extend('data');\n debug(`Creating query for the first author`);\n //debugData(record);\n\n const author = getAuthor(record);\n\n if (testStringOrNumber(author)) {\n const formatted = String(author)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n // use word search for authors starting with a boolean\n const useWordSearch = checkUseWordSearch(formatted);\n // Prevent too many matches by having a minimum length\n debugData(`Author string: ${formatted}`);\n if (formatted.length >= 5) {\n return [`dc.author=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n }\n return [];\n }\n\n return [];\n\n function getAuthor(record) {\n //debugData(record);\n // eslint-disable-next-line prefer-named-capture-group\n const [field] = record.get(/^(100)|(110)|(111)|(700)|(710)|(711)$/u);\n //debugData(field);\n\n if (field) {\n const authorString = field.subfields\n .filter(({code}) => ['a'].includes(code)) // We might use different subfield code sets for X00, X10 and X11?\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n return authorString;\n }\n return false;\n }\n}\n\nexport function bibPublishers(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');\n const debugData = debug.extend('data');\n debug(`Creating query for the first publisher`);\n //debugData(record);\n\n const publisher = getPublisher(record);\n if (testStringOrNumber(publisher)) {\n const formatted = String(publisher)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n debugData(`Publisher string: '${formatted}'`);\n if (formatted.length === 0) {\n return [];\n }\n // use non-wildcard word search from dc.publisher\n return [`dc.publisher=\"${formatted}\"`];\n }\n\n return [];\n\n function getPublisher(record) {\n //debugData(record);\n const [field] = record.get(/^26[04]$/u).filter(f => f.subfields.some(sf => sf.code === 'b')); // Filter removes copyright-only fields\n //debugData(field);\n\n if (field) {\n const publisherString = field.subfields\n .filter(({code}) => ['b'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ')\n .trim();\n return publisherString;\n }\n return false;\n }\n}\n\nexport function bibYear(record) {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibYear');\n const debugData = debug.extend('data');\n debug(`Creating query for the publishing year`);\n\n const year = getYear(record);\n if (year) {\n return [`dc.date=\"${year}\"`];\n }\n return [];\n\n function getYear(record) {\n const [f008] = record.get(/^008$/u);\n if (!f008 || !f008.value || f008.value.length < 11) {\n debug('f008 missing/crappy');\n return false;\n }\n\n debugData(`f008: ${JSON.stringify(f008)}`);\n const {value} = f008;\n if (value === '||||' || value === ' ') { // Meaningless values\n return false;\n }\n return value.slice(7, 11);\n }\n}\n\nexport function checkUseWordSearch(formatted) {\n const lowercased = formatted.toLowerCase();\n // Note: add a space to startWords to catch just actual boolean words\n const booleanStartWords = ['and ', 'or ', 'nor ', 'not '];\n return booleanStartWords.some(word => lowercased.startsWith(word));\n}\n\nexport function bibStandardIdentifiers(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n const debugData = debug.extend('data');\n debug(`Creating queries for standard identifiers`);\n\n // DEVELOP: should we query also f015 and f028?\n\n const fields = record.get(/^(?<def>020|022|024)$/u);\n const identifiers = [].concat(...fields.map(toIdentifiers));\n const uniqueIdentifiers = [...new Set(identifiers)];\n\n debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n if (uniqueIdentifiers.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n function toIdentifiers({tag, subfields}) {\n const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n if (tag === '022') {\n return subfields\n .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n\n if (tag === '020') {\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))\n .map(({value}) => String(value));\n }\n}\n"],
|
|
5
|
+
"mappings": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,gCAAgC,mBAAmB,0BAAyB;AACvG,SAAQ,gCAAgC,+BAA+B,4BAA2B;AAClG,SAAQ,qBAAoB;AAC5B,SAAQ,gBAAe;AAEvB,MAAM,QAAQ,kBAAkB,0DAA0D;AAE1F,MAAM,wBAAwB;AAEvB,gBAAS,aAAa,QAAQ;AA0BnC,QAAMA,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AAGrC,EAAAA,OAAM,iCAAiC;AAEvC,QAAM,QAAQ,OAAO,IAAI,KAAK;AAC9B,YAAU,eAAe,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAElE,SAAO,MAAM,SAAS,IAAI,aAAa,KAAK,IAAI,CAAC;AAEjD,WAAS,aAAaC,QAAO;AAC3B,IAAAD,OAAM,wCAAwC;AAE9C,UAAM,aAAa,cAAcC,MAAK;AAEtC,QAAI,WAAW,SAAS,GAAG;AACzB,MAAAD,OAAM,2CAA2C;AACjD,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,aAAa,UAAU,YAAY,kBAAkB;AAE3D,WAAO;AAEP,aAAS,cAAcC,QAAO;AAC5B,MAAAD,OAAM,qCAAqC;AAG3C,YAAME,cAAaD,OAAM,IAAI,WAAW,EAAE,OAAO,iBAAe,WAAW;AAC3E,aAAOC;AAEP,eAAS,YAAY,OAAO;AAC1B,QAAAF,OAAM,6BAA6B;AAEnC,eAAO,+BAA+B,KAAK,IAAI,gBAAgB,KAAK,IAAI;AAExE,iBAAS,gBAAgBG,QAAO;AAC9B,UAAAH,OAAM,8BAA8B;AACpC,gBAAM,CAAC,GAAG,IAAI,kBAAkBG,QAAO,GAAG;AAC1C,gBAAM,CAAC,GAAG,IAAI,kBAAkBA,QAAO,GAAG;AAE1C,gBAAM,aAAa,mBAAmB,0BAA0B,GAAG,CAAC;AACpE,gBAAM,aAAa,0BAA0B,GAAG;AAEhD,oBAAU,GAAG,KAAK,UAAU,GAAG,CAAC,MAAM,KAAK,UAAU,GAAG,CAAC,EAAE;AAC3D,iBAAO,WAAW,OAAO,UAAU;AAAA,QACrC;AAEA,iBAAS,mBAAmB,eAAe;AACzC,gBAAM,oBAAqB;AAC3B,gBAAM,kBAAkB,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,mBAAmB,OAAO,IAAI;AACxH,oBAAU,cAAc,aAAa,OAAO,eAAe,EAAE;AAC7D,iBAAO;AAAA,QACT;AAEA,iBAAS,0BAA0B,eAAe;AAChD,oBAAU,eAAe,aAAa,EAAE;AACxC,gBAAM,qBAAsB;AAC5B,iBAAO,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,oBAAoB,EAAE,IAAI;AAAA,QACrG;AAAA,MAEF;AAAA,IACF;AAAA,EACF;AACF;AAEO,gBAAS,cAAc,QAAQ;AAGpC,QAAMH,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,iCAAiC;AAKvC,QAAM,aAAa,kBAAkB,MAAM;AAE3C,YAAU,uBAAuB,WAAW,MAAM,MAAM,KAAK,UAAU,UAAU,CAAC,EAAE;AAEpF,MAAI,WAAW,SAAS,GAAG;AACzB,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,YAAY,mBAAmB;AAClD;AAKO,gBAAS,kBAAkB,QAAQ;AACxC,QAAMA,SAAQ,kBAAkB,4EAA4E;AAC5G,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,8BAA8B;AACpC,QAAM,QAAQ,qBAAqB,MAAM;AACzC,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,CAAC,EAAE,IAAI,+BAA+B,MAAM,CAAC,CAAC;AACpD,EAAAA,OAAM,aAAa,KAAK,UAAU,EAAE,CAAC,EAAE;AAGvC,QAAM,OAAO,8BAA8B,MAAM,CAAC,CAAC;AACnD,EAAAA,OAAM,GAAG,cAAc,MAAM,CAAC,CAAC,CAAC,OAAO,OAAO,OAAO,KAAK,EAAE;AAC5D,MAAI,MAAM;AACR,WAAO,CAAC,WAAW,IAAI,4BAA4B,EAAE,EAAE;AAAA,EACzD;AAEA,SAAO,CAAC,uBAAuB,EAAE,EAAE;AAErC;AAQO,gBAAS,SAAS,QAAQ;AAE/B,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,EAAC,CAAC;AAC7D;AAEO,gBAAS,eAAe,QAAQ;AACrC,QAAM,gBAAgB;AAGtB,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,IAAG,CAAC;AAC/D;AAEO,gBAAS,mBAAmB,QAAQ;AACzC,QAAM,8BAA8B;AAIpC,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,KAAI,CAAC;AAC9E;AAEO,gBAAS,6BAA6B,QAAQ;AACnD,QAAM,8BAA8B;AAGpC,QAAM,gBAAgB,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,MAAM,YAAY,MAAM,kBAAkB,CAAC,EAAC,CAAC;AACnI,QAAM,GAAG,KAAK,UAAU,aAAa,CAAC,EAAE;AACxC,SAAO,EAAC,WAAW,MAAM,KAAK,aAAa,EAAE,QAAQ,GAAG,eAAe,aAAY;AACrF;AAEA,SAAS,iBAAiB,QAAQ;AAChC,QAAM,QAAQ,SAAS,QAAQ,CAAC,GAAG,CAAC;AACpC,MAAI,OAAO;AACT,UAAM,SAAS,MAAM,MAAM,GAAG;AAI9B,QAAI,SAAS,KAAK,MAAM,SAAS,IAAI;AACnC,YAAM,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK;AAG/B,UAAI,QAAQ,KAAK,UAAU,KAAK,QAAM,GAAG,SAAS,OAAO,GAAG,MAAM,MAAM,MAAM,CAAC,KAAK,MAAM,UAAU,uBAAuB;AACzH,eAAO;AAAA,MACT;AACA,aAAO,SAAS,QAAQ,CAAC,KAAK,GAAG,CAAC;AAAA,IACpC;AAAA,EACF;AACA,SAAO;AACT;AAEA,SAAS,QAAQ,QAAQ,iBAAiB,aAAa,OAAO;AAC5D,QAAM,QAAQ,iBAAiB,MAAM;AACrC,MAAI,CAAC,mBAAmB,KAAK,GAAG;AAC9B,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,YAAY,kBAAkB,KAAK;AAGzC,QAAM,gBAAgB,mBAAmB,SAAS;AAGlD,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,aAAa,gBAAgB,KAAK,GAAG,GAAG,SAAS,MAAM,WAAW,IAAI;AAAA,EAChF;AAEA,QAAM,iBAAiB,CAAC,iBAAiB,UAAU,UAAU;AAG7D,SAAO,CAAC,aAAa,iBAAiB,CAAC,iBAAiB,KAAK,GAAG,GAAG,SAAS,GAAG,iBAAiB,MAAM,EAAE,KAAK,WAAW,cAAc;AAEtI,WAAS,kBAAkBI,QAAO;AAChC,UAAMC,aAAY,OAAOD,MAAK,EAC3B,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,2BAA2B,EAAE,EACrC,QAAQ,SAAS,GAAG,EACpB,KAAK,EACL,QAAQ,mBAAmB,IAAI,EAC/B,KAAK;AAER,WAAOC;AAAA,EACT;AACF;AAEO,gBAAS,wBAAwB,EAAC,QAAQ,iBAAiB,UAAU,OAAO,aAAa,OAAO,mBAAmB,CAAC,EAAC,GAAG;AAC7H,QAAM,6CAA6C,eAAe,cAAc,OAAO,iBAAiB,UAAU,EAAE;AACpH,QAAM,CAAC,OAAO,WAAW,cAAc,IAAI,QAAQ,QAAQ,iBAAiB,UAAU;AACtF,MAAI,UAAU,QAAW;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,UAAU,KAAK,EAAE;AAIvB,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,KAAK;AAAA,EACf;AAEA,QAAM,sBAAsB,aAAa,CAAC,GAAG,kBAAkB,KAAK,IAAI;AAExE,SAAO,mBAAmB,EAAC,OAAO,gBAAgB,SAAS,YAAY,kBAAkB,oBAAmB,CAAC;AAE7G,WAAS,mBAAmB,EAAC,OAAAC,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACvH,UAAM,oBAAoB;AAC1B,UAAM,CAAC,WAAW,IAAI,WAAW,MAAM;AACvC,QAAI,gBAAgB,QAAW;AAC7B,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,WAAW,QAAQJ,MAAK,EAAE,IAAII;AAChG,eAAO,gBAAgB,EAAC,OAAO,GAAG,WAAW,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MACxI;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,WAAW,QAAQJ,MAAK,EAAE;AAAA,IACvE;AACA,WAAO,qBAAqB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,SAAAC,UAAS,YAAAC,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAE5F;AAEA,WAAS,qBAAqB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACzH,UAAM,CAAC,cAAc,IAAI,cAAc,MAAM;AAC7C,QAAI,mBAAmB,QAAW;AAChC,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,cAAc,QAAQJ,MAAK,EAAE,IAAII;AACnG,eAAO,gBAAgB,EAAC,OAAO,GAAG,cAAc,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MAC3I;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,cAAc,QAAQJ,MAAK,EAAE;AAAA,IAC1E;AACA,QAAIC,mBAAkB,CAACC,UAAS;AAC9B,aAAOC,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,gBAAgB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,YAAAE,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAC9E;AAEA,WAAS,gBAAgB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,YAAAE,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACnG,UAAM,CAAC,SAAS,IAAI,QAAQ,MAAM;AAClC,QAAI,cAAc,QAAW;AAC3B,YAAM,sBAAsB,GAAG,SAAS,QAAQJ,MAAK;AACrD,aAAOG,cAAa,CAAC,GAAGC,mBAAkB,mBAAmB,IAAI,CAAC,mBAAmB;AAAA,IACvF;AACA,QAAIH,iBAAgB;AAClB,aAAOE,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,CAAC;AAAA,EACV;AAEF;AAEO,gBAAS,WAAW,QAAQ;AACjC,QAAMN,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,qCAAqC;AAG3C,QAAM,SAAS,UAAU,MAAM;AAE/B,MAAI,mBAAmB,MAAM,GAAG;AAC9B,UAAM,YAAY,OAAO,MAAM,EAC5B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAGR,UAAM,gBAAgB,mBAAmB,SAAS;AAElD,cAAU,kBAAkB,SAAS,EAAE;AACvC,QAAI,UAAU,UAAU,GAAG;AACzB,aAAO,CAAC,cAAc,gBAAgB,KAAK,GAAG,GAAG,SAAS,IAAI;AAAA,IAChE;AACA,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,CAAC;AAER,WAAS,UAAUY,SAAQ;AAGzB,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,wCAAwC;AAGnE,QAAI,OAAO;AACT,YAAM,eAAe,MAAM,UACxB,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG;AACX,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,cAAc,QAAQ;AACpC,QAAMZ,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAG9C,QAAM,YAAY,aAAa,MAAM;AACrC,MAAI,mBAAmB,SAAS,GAAG;AACjC,UAAM,YAAY,OAAO,SAAS,EAC/B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAER,cAAU,sBAAsB,SAAS,GAAG;AAC5C,QAAI,UAAU,WAAW,GAAG;AAC1B,aAAO,CAAC;AAAA,IACV;AAEA,WAAO,CAAC,iBAAiB,SAAS,GAAG;AAAA,EACvC;AAEA,SAAO,CAAC;AAER,WAAS,aAAaY,SAAQ;AAE5B,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,WAAW,EAAE,OAAO,OAAK,EAAE,UAAU,KAAK,QAAM,GAAG,SAAS,GAAG,CAAC;AAG3F,QAAI,OAAO;AACT,YAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG,EACR,KAAK;AACR,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,QAAQ,QAAQ;AAC9B,QAAMZ,SAAQ,kBAAkB,kEAAkE;AAClG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAE9C,QAAM,OAAO,QAAQ,MAAM;AAC3B,MAAI,MAAM;AACR,WAAO,CAAC,YAAY,IAAI,GAAG;AAAA,EAC7B;AACA,SAAO,CAAC;AAER,WAAS,QAAQY,SAAQ;AACvB,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,QAAI,CAAC,QAAQ,CAAC,KAAK,SAAS,KAAK,MAAM,SAAS,IAAI;AAClD,MAAAZ,OAAM,qBAAqB;AAC3B,aAAO;AAAA,IACT;AAEA,cAAU,SAAS,KAAK,UAAU,IAAI,CAAC,EAAE;AACzC,UAAM,EAAC,MAAK,IAAI;AAChB,QAAI,UAAU,UAAU,UAAU,QAAQ;AACxC,aAAO;AAAA,IACT;AACA,WAAO,MAAM,MAAM,GAAG,EAAE;AAAA,EAC1B;AACF;AAEO,gBAAS,mBAAmB,WAAW;AAC5C,QAAM,aAAa,UAAU,YAAY;AAEzC,QAAM,oBAAoB,CAAC,QAAQ,OAAO,QAAQ,MAAM;AACxD,SAAO,kBAAkB,KAAK,UAAQ,WAAW,WAAW,IAAI,CAAC;AACnE;AAEO,gBAAS,uBAAuB,QAAQ;AAE7C,QAAMA,SAAQ,kBAAkB,iFAAiF;AACjH,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,2CAA2C;AAIjD,QAAM,SAAS,OAAO,IAAI,wBAAwB;AAClD,QAAM,cAAc,CAAC,EAAE,OAAO,GAAG,OAAO,IAAI,aAAa,CAAC;AAC1D,QAAM,oBAAoB,CAAC,GAAG,IAAI,IAAI,WAAW,CAAC;AAElD,YAAU,+BAA+B,KAAK,UAAU,MAAM,CAAC,EAAE;AACjE,YAAU,gBAAgB,YAAY,MAAM,MAAM,KAAK,UAAU,WAAW,CAAC,EAAE;AAC/E,YAAU,uBAAuB,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAElG,MAAI,kBAAkB,SAAS,GAAG;AAChC,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,mBAAmB,eAAe;AAEnD,WAAS,cAAc,EAAC,KAAK,UAAS,GAAG;AACvC,UAAM,iBAAkB;AACxB,UAAM,gBAAiB;AAEvB,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EAC3H,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,WAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,cAAc,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACrH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,EACnC;AACF;",
|
|
6
|
+
"names": ["debug", "fSids", "sidStrings", "field", "title", "formatted", "query", "queryIsOkAlone", "addYear", "alternates", "alternateQueries", "newAlternateQueries", "record"]
|
|
7
7
|
}
|
|
@@ -1,40 +1,106 @@
|
|
|
1
1
|
import createDebugLogger from "debug";
|
|
2
2
|
import { promisify } from "util";
|
|
3
3
|
import { toQueries } from "../candidate-search-utils.js";
|
|
4
|
-
import { getSubfieldValues, testStringOrNumber, toMelindaIds } from "../../matching-utils.js";
|
|
4
|
+
import { getSubfieldValues, stringAfter, stringBefore, testStringOrNumber, toMelindaIds } from "../../matching-utils.js";
|
|
5
|
+
import { fieldToString, uniqArray } from "@natlibfi/marc-record-validators-melinda/dist/utils.js";
|
|
5
6
|
const setTimeoutPromise = promisify(setTimeout);
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
return toQueries(ids, "melinda.partsofhost");
|
|
7
|
+
const debug = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:component");
|
|
8
|
+
const debugData = debug.extend("data");
|
|
9
|
+
export function parse773g(field) {
|
|
10
|
+
const g = field?.subfields?.find((sf) => sf.code === "g");
|
|
11
|
+
if (!g) {
|
|
12
|
+
return {};
|
|
13
13
|
}
|
|
14
|
-
|
|
15
|
-
return
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
14
|
+
const value = normalizeValue(g.value);
|
|
15
|
+
return {
|
|
16
|
+
// NB! We are not currently interested in volume (vuosikerta)
|
|
17
|
+
year: gToYear(value) || null,
|
|
18
|
+
number: gToNumber(value) || null,
|
|
19
|
+
pages: gToPages(value) || null
|
|
20
|
+
// Might also be eg. Raita 5. However, returns only the number part
|
|
21
|
+
};
|
|
22
|
+
function normalizeValue(value2) {
|
|
23
|
+
return value2.replace(/(?:\. -|\.)$/u, "");
|
|
24
|
+
}
|
|
25
|
+
function gToNumber(value2) {
|
|
26
|
+
const extract1 = value2.replace(/^.*: ([1-9][0-9]?[0-9]), (?:p\.|page|pp\.|s\.|Seite|sida|sidor|sivu).*$/ui, "$1");
|
|
27
|
+
if (extract1 !== value2) {
|
|
28
|
+
return extract1;
|
|
29
|
+
}
|
|
30
|
+
return void 0;
|
|
31
|
+
}
|
|
32
|
+
function gToPages(value2) {
|
|
33
|
+
if (value2.match(/^[0-9]+(?:-[0-9]+)?(?:, [0-9]+(?:-[0-9]+)?)*$/u)) {
|
|
34
|
+
return value2;
|
|
35
|
+
}
|
|
36
|
+
const numberPartOnly = value2.replace(/^.*(?:p\.|raidat|raita|s\.|Seite|sivut?|pages?) /ui, "");
|
|
37
|
+
if (numberPartOnly !== value2) {
|
|
38
|
+
return gToPages(numberPartOnly);
|
|
39
|
+
}
|
|
40
|
+
return void 0;
|
|
41
|
+
}
|
|
42
|
+
function gToYear(value2) {
|
|
43
|
+
debug(`EXTRACT YEAR FROM '${value2}`);
|
|
44
|
+
if (value2.match(/^[1-9][0-9]?[0-9]? ?\((?:20[012][0-9]|19[0-9][0-9])\)/u) || value2.match(/^\((?:20[012][0-9]|19[0-9][0-9])\)/u)) {
|
|
45
|
+
return stringBefore(stringAfter(value2, "("), ")");
|
|
46
|
+
}
|
|
47
|
+
if (value2.match(/^(?:20[012][0-9]|19[0-9][0-9]) :/u)) {
|
|
48
|
+
return stringBefore(value2, " ");
|
|
49
|
+
}
|
|
50
|
+
if (value2.match(/^\([1-9][0-9]?\) ?(?:20[012][0-9]|19[0-9][0-9])[^0-9]/)) {
|
|
51
|
+
return value2.replace(/^\([0-9]+\) ?/u, "").replace(/[^0-9].*$/u, "");
|
|
20
52
|
}
|
|
21
|
-
|
|
22
|
-
|
|
53
|
+
if (value2.match(/^[^0-9]*(?:[1-9]|[12][0-9]|30|31)\.(?:1[012]|[1-9])\. ?(?:20[012][0-9]|19[0-9][0-9])[^0-9]/u)) {
|
|
54
|
+
return value2.replace(/^[^0-9]*[0-9]+\.[0-9]+\. ?/u, "").replace(/[^0-9].*$/u, "");
|
|
55
|
+
}
|
|
56
|
+
return void 0;
|
|
57
|
+
}
|
|
58
|
+
}
|
|
59
|
+
export function extractPublicationYearFrom773(field) {
|
|
60
|
+
const data = parse773g(field);
|
|
61
|
+
debug(`Extracted year ${data.year} from f773$g ${fieldToString(field)}`);
|
|
62
|
+
return data.year;
|
|
63
|
+
}
|
|
64
|
+
function isMelindaSubfieldW(subfield) {
|
|
65
|
+
if (subfield.code !== "w") {
|
|
66
|
+
return false;
|
|
67
|
+
}
|
|
68
|
+
return /^\((?:FI-MELINDA|FIN01)\)0[0-9]{8}$/ui.test(subfield.value);
|
|
69
|
+
}
|
|
70
|
+
function removeMelindaPrefixFromValue(value) {
|
|
71
|
+
return value.replace(/^(?:\(FI-MELINDA\)|\(FIN01\))/, "");
|
|
72
|
+
}
|
|
73
|
+
export function getHostMelindaFields(record) {
|
|
74
|
+
return record.get(/^773$/u).filter((f) => f.subfields.some((sf) => isMelindaSubfieldW(sf)));
|
|
75
|
+
}
|
|
76
|
+
export function extractHostMelindaIdsFromField(field) {
|
|
77
|
+
return field.subfields.filter((sf) => isMelindaSubfieldW(sf)).map((sf) => removeMelindaPrefixFromValue(sf.value));
|
|
78
|
+
}
|
|
79
|
+
export function getHostMelindaIds(record) {
|
|
80
|
+
return uniqArray(getHostMelindaFields(record).map((f) => extractHostMelindaIdsFromField(f)).flat());
|
|
81
|
+
}
|
|
82
|
+
export function hostIdMelinda(record) {
|
|
83
|
+
debug(`Creating query for the Melinda Id host`);
|
|
84
|
+
const ids = getHostMelindaIds(record);
|
|
85
|
+
if (ids.length === 0) {
|
|
86
|
+
debug(`No valid Melinda Id host found`);
|
|
87
|
+
return [];
|
|
23
88
|
}
|
|
89
|
+
return toQueries(getHostMelindaIds(record), "melinda.partsofhost");
|
|
24
90
|
}
|
|
25
91
|
export async function hostIdOtherSource(record, client) {
|
|
26
|
-
const
|
|
27
|
-
const
|
|
28
|
-
|
|
92
|
+
const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:hostIdOtherSource");
|
|
93
|
+
const debugData2 = debug2.extend("data");
|
|
94
|
+
debug2(`Creating query for the Other source Id host`);
|
|
29
95
|
const ids = getHostIds(record);
|
|
30
96
|
const otherSources = getOtherSources(record);
|
|
31
97
|
if (ids.length > 0 && otherSources.length > 0) {
|
|
32
98
|
const otherSourceIds = ids.map((id) => otherSources.map((source) => `${id}${source}`)).flat();
|
|
33
99
|
const melindaHostQueries = await handleSruCalls(otherSourceIds);
|
|
34
|
-
|
|
100
|
+
debug2(JSON.stringify(melindaHostQueries));
|
|
35
101
|
return melindaHostQueries;
|
|
36
102
|
}
|
|
37
|
-
|
|
103
|
+
debug2(`No valid other source hosts found`);
|
|
38
104
|
return [];
|
|
39
105
|
function getHostIds(record2) {
|
|
40
106
|
const f773s = record2.get(/^773$/u).filter((f773) => f773.subfields.some((sub) => sub.code === "w" && !/\(FI-MELINDA\).*/ui.test(sub.value)));
|
|
@@ -42,7 +108,7 @@ export async function hostIdOtherSource(record, client) {
|
|
|
42
108
|
return false;
|
|
43
109
|
}
|
|
44
110
|
const subfieldWs = f773s.map((f773) => {
|
|
45
|
-
|
|
111
|
+
debugData2(`f773: ${JSON.stringify(f773)}`);
|
|
46
112
|
return getSubfieldValues(f773, "w").flat();
|
|
47
113
|
}).flat();
|
|
48
114
|
const ids2 = subfieldWs.map((value) => `${value}`.replace(/\(FI-.*\)/ui, "")).filter((value) => testStringOrNumber(value)).filter((value, index, array) => array.indexOf(value) === index);
|
|
@@ -55,29 +121,29 @@ export async function hostIdOtherSource(record, client) {
|
|
|
55
121
|
async function handleSruCalls(otherSourceIds, ids2 = []) {
|
|
56
122
|
const [otherSourceId, ...rest] = otherSourceIds;
|
|
57
123
|
if (otherSourceId === void 0) {
|
|
58
|
-
|
|
124
|
+
debug2(`host ids: ${ids2}`);
|
|
59
125
|
const validIds = ids2.filter((id2) => id2);
|
|
60
126
|
return toQueries(validIds, "melinda.partsofhost");
|
|
61
127
|
}
|
|
62
128
|
const otherSourceHostQuery = await toQueries([otherSourceId], "melinda.sourceid");
|
|
63
129
|
const id = await new Promise((resolve, reject) => {
|
|
64
|
-
|
|
130
|
+
debug2(`Searching for hosts with query: ${otherSourceHostQuery}`);
|
|
65
131
|
let recordId;
|
|
66
132
|
client.searchRetrieve(otherSourceHostQuery).on("error", (err) => {
|
|
67
|
-
|
|
133
|
+
debug2(`SRU error for query: ${otherSourceHostQuery}: ${err}`);
|
|
68
134
|
reject(err);
|
|
69
135
|
}).on("end", async () => {
|
|
70
136
|
try {
|
|
71
|
-
|
|
137
|
+
debug2(`Searching for hosts: done`);
|
|
72
138
|
await setTimeoutPromise(10);
|
|
73
139
|
resolve(recordId);
|
|
74
140
|
} catch (err) {
|
|
75
|
-
|
|
141
|
+
debug2(`Error caught on END`);
|
|
76
142
|
reject(err);
|
|
77
143
|
}
|
|
78
144
|
}).on("record", (record2) => {
|
|
79
145
|
const [field] = record2.get(/^001$/u);
|
|
80
|
-
|
|
146
|
+
debug2(field);
|
|
81
147
|
recordId = field.value ? field.value : "";
|
|
82
148
|
});
|
|
83
149
|
});
|