@natlibfi/melinda-record-matching 4.1.4-alpha.2 → 4.1.5-alpha.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/candidate-search/query-list/bib.js +52 -1
- package/dist/candidate-search/query-list/bib.js.map +1 -1
- package/dist/match-detection/features/bib/index.js +7 -0
- package/dist/match-detection/features/bib/index.js.map +1 -1
- package/dist/match-detection/features/bib/index.spec.js +5 -0
- package/dist/match-detection/features/bib/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/media-type.js +83 -0
- package/dist/match-detection/features/bib/media-type.js.map +1 -0
- package/dist/match-detection/features/bib/other-standard-identifier.js +1 -0
- package/dist/match-detection/features/bib/other-standard-identifier.js.map +1 -1
- package/dist/match-detection/features/bib/record-type.js +2 -0
- package/dist/match-detection/features/bib/record-type.js.map +1 -1
- package/package.json +1 -1
- package/src/candidate-search/query-list/bib.js +57 -1
- package/src/match-detection/features/bib/index.js +1 -0
- package/src/match-detection/features/bib/index.spec.js +8 -0
- package/src/match-detection/features/bib/media-type.js +77 -0
- package/src/match-detection/features/bib/other-standard-identifier.js +2 -0
- package/src/match-detection/features/bib/record-type.js +2 -0
|
@@ -6,6 +6,7 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
6
6
|
exports.bibAuthors = bibAuthors;
|
|
7
7
|
exports.bibHostComponents = bibHostComponents;
|
|
8
8
|
exports.bibMelindaIds = bibMelindaIds;
|
|
9
|
+
exports.bibPublishers = bibPublishers;
|
|
9
10
|
exports.bibSourceIds = bibSourceIds;
|
|
10
11
|
exports.bibStandardIdentifiers = bibStandardIdentifiers;
|
|
11
12
|
exports.bibTitle = bibTitle;
|
|
@@ -177,7 +178,8 @@ function bibTitle(record) {
|
|
|
177
178
|
if (formatted.length >= 5) {
|
|
178
179
|
return [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`];
|
|
179
180
|
}
|
|
180
|
-
|
|
181
|
+
// use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]
|
|
182
|
+
return addAuthorsToSearch(`dc.title="${formatted}"`);
|
|
181
183
|
}
|
|
182
184
|
return [];
|
|
183
185
|
function addAuthorsToSearch(titleQuery) {
|
|
@@ -185,6 +187,15 @@ function bibTitle(record) {
|
|
|
185
187
|
if (authorQuery !== undefined) {
|
|
186
188
|
return [`${authorQuery} AND ${titleQuery}`];
|
|
187
189
|
}
|
|
190
|
+
return addPublisherToSearch(titleQuery);
|
|
191
|
+
//return [];
|
|
192
|
+
}
|
|
193
|
+
|
|
194
|
+
function addPublisherToSearch(query) {
|
|
195
|
+
const [publisherQuery] = bibPublishers(record);
|
|
196
|
+
if (publisherQuery !== undefined) {
|
|
197
|
+
return [`${publisherQuery} AND ${query}`];
|
|
198
|
+
}
|
|
188
199
|
return [];
|
|
189
200
|
}
|
|
190
201
|
function getTitle() {
|
|
@@ -226,6 +237,9 @@ function bibAuthors(record) {
|
|
|
226
237
|
if (formatted.length >= 5) {
|
|
227
238
|
return [`dc.author="${useWordSearch ? '' : '^'}${formatted}*"`];
|
|
228
239
|
}
|
|
240
|
+
//if (formatted) {
|
|
241
|
+
// return [`dc.author="${formatted}"`];
|
|
242
|
+
//}
|
|
229
243
|
return [];
|
|
230
244
|
}
|
|
231
245
|
return [];
|
|
@@ -248,10 +262,47 @@ function bibAuthors(record) {
|
|
|
248
262
|
return false;
|
|
249
263
|
}
|
|
250
264
|
}
|
|
265
|
+
function bibPublishers(record) {
|
|
266
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');
|
|
267
|
+
const debugData = debug.extend('data');
|
|
268
|
+
debug(`Creating query for the first publisher`);
|
|
269
|
+
//debugData(record);
|
|
270
|
+
|
|
271
|
+
const publisher = getPublisher(record);
|
|
272
|
+
if ((0, _matchingUtils.testStringOrNumber)(publisher)) {
|
|
273
|
+
const formatted = String(publisher).replace(/[^\w\s\p{Alphabetic}]/gu, '')
|
|
274
|
+
// Clean up concurrent spaces from fe. subfield changes
|
|
275
|
+
.replace(/ +/gu, ' ').trim().slice(0, 30).trim();
|
|
276
|
+
debugData(`Publisher string: ${formatted}`);
|
|
277
|
+
// use non-wildcard word search from dc.publisher
|
|
278
|
+
return [`dc.publisher="${formatted}"`];
|
|
279
|
+
}
|
|
280
|
+
return [];
|
|
281
|
+
function getPublisher(record) {
|
|
282
|
+
//debugData(record);
|
|
283
|
+
const [field] = record.get(/^(?:260)|(?:264)$/u);
|
|
284
|
+
//debugData(field);
|
|
285
|
+
|
|
286
|
+
if (field) {
|
|
287
|
+
const publisherString = field.subfields.filter(({
|
|
288
|
+
code
|
|
289
|
+
}) => ['b'].includes(code)).map(({
|
|
290
|
+
value
|
|
291
|
+
}) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').filter(value => value)
|
|
292
|
+
// In Melinda's index subfield separators are indexed as ' '
|
|
293
|
+
.join(' ');
|
|
294
|
+
return publisherString;
|
|
295
|
+
}
|
|
296
|
+
return false;
|
|
297
|
+
}
|
|
298
|
+
}
|
|
251
299
|
function bibStandardIdentifiers(record) {
|
|
252
300
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');
|
|
253
301
|
const debugData = debug.extend('data');
|
|
254
302
|
debug(`Creating queries for standard identifiers`);
|
|
303
|
+
|
|
304
|
+
// DEVELOP: should we query also f015 and f028?
|
|
305
|
+
|
|
255
306
|
const fields = record.get(/^(?<def>020|022|024)$/u);
|
|
256
307
|
const identifiers = [].concat(...fields.map(toIdentifiers));
|
|
257
308
|
const uniqueIdentifiers = [...new Set(identifiers)];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"bib.js","names":["_debug","_interopRequireDefault","require","_candidateSearchUtils","_matchingUtils","obj","__esModule","default","bibSourceIds","record","debug","createDebugLogger","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","toQueries","map","toSidString","filter","nonEmptySid","field","validateSidFieldSubfieldCounts","createSidString","sfC","getSubfieldValues","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","testStringOrNumber","String","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","getMelindaIdsF035","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","booleanStartWords","formatted","trim","slice","useWordSearch","some","word","toLowerCase","startsWith","addAuthorsToSearch","titleQuery","authorQuery","bibAuthors","undefined","titleString","includes","join","author","getAuthor","authorString","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub"],"sources":["../../../src/candidate-search/query-list/bib.js"],"sourcesContent":["\n/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils';\n\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 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n const id = getHostId();\n return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];\n\n function getHostId() {\n const [field] = record.get(/^773$/u);\n\n if (field) {\n const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n if (testStringOrNumber(value) && (/^\\(FI-MELINDA\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FI-MELINDA\\)/u, '');\n }\n\n if (testStringOrNumber(value) && (/^\\(FIN01\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FIN01\\)/u, '');\n }\n\n return false;\n }\n return false;\n }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n const title = getTitle();\n const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n if (testStringOrNumber(title)) {\n const formatted = String(title)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n // use word search for titles starting with a boolean\n const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n // Prevent too many matches by having a minimum length\n // Note that currently this fails matching if there are no matches from previous matchers\n if (formatted.length >= 5) {\n return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n }\n return addAuthorsToSearch(`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`);\n }\n\n return [];\n\n function addAuthorsToSearch(titleQuery) {\n const [authorQuery] = bibAuthors(record);\n if (authorQuery !== undefined) {\n return [`${authorQuery} AND ${titleQuery}`];\n }\n return [];\n }\n\n function getTitle() {\n const [field] = record.get(/^245$/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 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 const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\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 = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\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))\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\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 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"],"mappings":";;;;;;;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,qBAAA,GAAAD,OAAA;AACA,IAAAE,cAAA,GAAAF,OAAA;AAA8H,SAAAD,uBAAAI,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA7B9H;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;;AAMO,SAASG,YAAYA,CAACC,MAAM,EAAE;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtC;;EAEAH,KAAK,CAAE,iCAAgC,CAAC;EAExC,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAG,CAAC,KAAK,CAAC;EAC/BH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACJ,KAAK,CAAE,EAAC,CAAC;EAEnE,OAAOA,KAAK,CAACE,MAAM,GAAG,CAAC,GAAGG,YAAY,CAACL,KAAK,CAAC,GAAG,EAAE;EAElD,SAASK,YAAYA,CAACL,KAAK,EAAE;IAC3BJ,KAAK,CAAE,wCAAuC,CAAC;IAE/C,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAK,CAAC;IAEvC,IAAIM,UAAU,CAACJ,MAAM,GAAG,CAAC,EAAE;MACzBN,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO,EAAE;IACX;IAEA,MAAMY,UAAU,GAAG,IAAAC,+BAAS,EAACH,UAAU,EAAE,kBAAkB,CAAC;IAE5D,OAAOE,UAAU;IAEjB,SAASD,aAAaA,CAACP,KAAK,EAAE;MAC5BJ,KAAK,CAAE,qCAAoC,CAAC;;MAE5C;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAG,CAACC,WAAW,CAAC,CAACC,MAAM,CAACC,WAAW,IAAIA,WAAW,CAAC;MAC5E,OAAOP,UAAU;MAEjB,SAASK,WAAWA,CAACG,KAAK,EAAE;QAC1BlB,KAAK,CAAE,6BAA4B,CAAC;QAEpC,OAAO,IAAAmB,6CAA8B,EAACD,KAAK,CAAC,GAAGE,eAAe,CAACF,KAAK,CAAC,GAAG,EAAE;QAE1E,SAASE,eAAeA,CAACF,KAAK,EAAE;UAC9BlB,KAAK,CAAE,8BAA6B,CAAC;UACrC,MAAM,CAACqB,GAAG,CAAC,GAAG,IAAAC,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAC3C,MAAM,CAACK,GAAG,CAAC,GAAG,IAAAD,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAE3C,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAG,CAAC,CAAC;UACrE,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAG,CAAC;UAEjDrB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACa,GAAG,CAAE,MAAKd,IAAI,CAACC,SAAS,CAACe,GAAG,CAAE,EAAC,CAAC;UAC5D,OAAOC,UAAU,CAACI,MAAM,CAACD,UAAU,CAAC;QACtC;QAEA,SAASF,kBAAkBA,CAACI,aAAa,EAAE;UACzC,MAAMC,iBAAiB,GAAI,gDAAiD;UAC5E,MAAMC,eAAe,GAAG,IAAAC,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACJ,iBAAiB,EAAE,OAAO,CAAC,GAAG,EAAE;UAC1H5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAC,CAAC;UAC9D,OAAOA,eAAe;QACxB;QAEA,SAASL,yBAAyBA,CAACG,aAAa,EAAE;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAAC,CAAC;UACzC,MAAMM,kBAAkB,GAAI,OAAQ;UACpC,OAAO,IAAAH,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACC,kBAAkB,EAAE,EAAE,CAAC,GAAG,EAAE;QACvG;MAEF;IACF;EACF;AACF;AAEO,SAASC,aAAaA,CAACrC,MAAM,EAAE;EACpC;;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wEAAwE,CAAC;EACzG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,iCAAgC,CAAC;;EAExC;EACA;EACA;EACA,MAAMqC,UAAU,GAAG,IAAAC,gCAAiB,EAACvC,MAAM,CAAC;EAE5CG,SAAS,CAAE,uBAAsBmC,UAAU,CAAC/B,MAAO,MAAKC,IAAI,CAACC,SAAS,CAAC6B,UAAU,CAAE,EAAC,CAAC;EAErF,IAAIA,UAAU,CAAC/B,MAAM,GAAG,CAAC,EAAE;IACzBN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACwB,UAAU,EAAE,mBAAmB,CAAC;AACnD;;AAGA;AACA;AACO,SAASE,iBAAiBA,CAACxC,MAAM,EAAE;EACxC,MAAMyC,EAAE,GAAGC,SAAS,CAAC,CAAC;EACtB,OAAO,IAAAT,iCAAkB,EAACQ,EAAE,CAAC,GAAG,CAAE,uBAAsBA,EAAG,EAAC,CAAC,GAAG,EAAE;EAElE,SAASC,SAASA,CAAA,EAAG;IACnB,MAAM,CAACvB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM;QAACwB;MAAK,CAAC,GAAGxB,KAAK,CAACyB,SAAS,CAACC,IAAI,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC;MAEpE,IAAI,IAAAb,iCAAkB,EAACU,KAAK,CAAC,IAAK,kBAAkB,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACzE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,kBAAkB,EAAE,EAAE,CAAC;MACtD;MAEA,IAAI,IAAAF,iCAAkB,EAACU,KAAK,CAAC,IAAK,aAAa,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACpE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC;MACjD;MAEA,OAAO,KAAK;IACd;IACA,OAAO,KAAK;EACd;AACF;;AAEA;AACA;AACA;;AAEA;AACA;AACA;;AAEO,SAASa,QAAQA,CAAChD,MAAM,EAAE;EAC/B,MAAMiD,KAAK,GAAGC,QAAQ,CAAC,CAAC;EACxB,MAAMC,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAErD,IAAI,IAAAlB,iCAAkB,EAACgB,KAAK,CAAC,EAAE;IAC7B,MAAMG,SAAS,GAAGlB,MAAM,CAACe,KAAK,CAAC,CAC5Bd,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,CAAC,CAAC,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,CAAC,CAAC;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,CAAC,CAAC,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACA;IACA,IAAIL,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,aAAYgD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IAChE;IACA,OAAOQ,kBAAkB,CAAE,aAAYL,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;EAClF;EAEA,OAAO,EAAE;EAET,SAASQ,kBAAkBA,CAACC,UAAU,EAAE;IACtC,MAAM,CAACC,WAAW,CAAC,GAAGC,UAAU,CAAC/D,MAAM,CAAC;IACxC,IAAI8D,WAAW,KAAKE,SAAS,EAAE;MAC7B,OAAO,CAAE,GAAEF,WAAY,QAAOD,UAAW,EAAC,CAAC;IAC7C;IACA,OAAO,EAAE;EACX;EAEA,SAASX,QAAQA,CAAA,EAAG;IAClB,MAAM,CAAC/B,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM8C,WAAW,GAAG9C,KAAK,CAACyB;MACxB;MAAA,CACC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACoB,QAAQ,CAACpB,IAAI,CAAC;MAC7C;MAAA,CACC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACCwB,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOF,WAAW;IACpB;IACA,OAAO,KAAK;EACd;AACF;AAEO,SAASF,UAAUA,CAAC/D,MAAM,EAAE;EACjC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,qCAAoC,CAAC;EAC5C;;EAEA,MAAMmE,MAAM,GAAGC,SAAS,CAACrE,MAAM,CAAC;EAChC,MAAMmD,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAGrD,IAAI,IAAAlB,iCAAkB,EAACmC,MAAM,CAAC,EAAE;IAC9B,MAAMhB,SAAS,GAAGlB,MAAM,CAACkC,MAAM,CAAC,CAC7BjC,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,CAAC,CAAC,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,CAAC,CAAC;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,CAAC,CAAC,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACAtD,SAAS,CAAE,kBAAiBiD,SAAU,EAAC,CAAC;IACxC,IAAIA,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,cAAagD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IACjE;IACA,OAAO,EAAE;EACX;EAEA,OAAO,EAAE;EAET,SAASiB,SAASA,CAACrE,MAAM,EAAE;IACzB;IACA;IACA,MAAM,CAACmB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,wCAAwC,CAAC;IACpE;;IAEA,IAAIa,KAAK,EAAE;MACT,MAAMmD,YAAY,GAAGnD,KAAK,CAACyB,SAAS,CACjC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAACoB,QAAQ,CAACpB,IAAI,CAAC,CAAC,CACxC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACCwB,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOG,YAAY;IACrB;IACA,OAAO,KAAK;EACd;AACF;AAGO,SAASC,sBAAsBA,CAACvE,MAAM,EAAE;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;EAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,2CAA0C,CAAC;EAElD,MAAMuE,MAAM,GAAGxE,MAAM,CAACM,GAAG,CAAC,wBAAwB,CAAC;EACnD,MAAMmE,WAAW,GAAG,EAAE,CAAC5C,MAAM,CAAC,GAAG2C,MAAM,CAACzD,GAAG,CAAC2D,aAAa,CAAC,CAAC;EAC3D,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACH,WAAW,CAAC,CAAC;EAEnDtE,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAS,CAAC+D,MAAM,CAAE,EAAC,CAAC;EAClErE,SAAS,CAAE,gBAAesE,WAAW,CAAClE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACgE,WAAW,CAAE,EAAC,CAAC;EAChFtE,SAAS,CAAE,uBAAsBwE,iBAAiB,CAACpE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACkE,iBAAiB,CAAE,EAAC,CAAC;EAEnG,IAAIA,iBAAiB,CAACpE,MAAM,GAAG,CAAC,EAAE;IAChCN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAAC6D,iBAAiB,EAAE,eAAe,CAAC;EAEpD,SAASD,aAAaA,CAAC;IAACG,GAAG;IAAEjC;EAAS,CAAC,EAAE;IACvC,MAAMkC,cAAc,GAAI,kBAAmB;IAC3C,MAAMC,aAAa,GAAI,mBAAoB;IAE3C,IAAIF,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOjC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAImC,cAAc,CAAC/B,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CAC5H5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,IAAIkC,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOjC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAImC,cAAc,CAAC/B,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CACvH5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,OAAOC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAIoC,aAAa,CAAChC,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CACtH5B,GAAG,CAAC,CAAC;MAAC4B;IAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;EACpC;AACF"}
|
|
1
|
+
{"version":3,"file":"bib.js","names":["_debug","_interopRequireDefault","require","_candidateSearchUtils","_matchingUtils","obj","__esModule","default","bibSourceIds","record","debug","createDebugLogger","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","toQueries","map","toSidString","filter","nonEmptySid","field","validateSidFieldSubfieldCounts","createSidString","sfC","getSubfieldValues","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","testStringOrNumber","String","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","getMelindaIdsF035","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","booleanStartWords","formatted","trim","slice","useWordSearch","some","word","toLowerCase","startsWith","addAuthorsToSearch","titleQuery","authorQuery","bibAuthors","undefined","addPublisherToSearch","query","publisherQuery","bibPublishers","titleString","includes","join","author","getAuthor","authorString","publisher","getPublisher","publisherString","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub"],"sources":["../../../src/candidate-search/query-list/bib.js"],"sourcesContent":["\n/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils';\n\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 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n const id = getHostId();\n return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];\n\n function getHostId() {\n const [field] = record.get(/^773$/u);\n\n if (field) {\n const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n if (testStringOrNumber(value) && (/^\\(FI-MELINDA\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FI-MELINDA\\)/u, '');\n }\n\n if (testStringOrNumber(value) && (/^\\(FIN01\\)/u).test(String(value))) {\n return String(value).replace(/^\\(FIN01\\)/u, '');\n }\n\n return false;\n }\n return false;\n }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n const title = getTitle();\n const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n if (testStringOrNumber(title)) {\n const formatted = String(title)\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n // use word search for titles starting with a boolean\n const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n // Prevent too many matches by having a minimum length\n // Note that currently this fails matching if there are no matches from previous matchers\n if (formatted.length >= 5) {\n return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n }\n // use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]\n return addAuthorsToSearch(`dc.title=\"${formatted}\"`);\n }\n\n return [];\n\n function addAuthorsToSearch(titleQuery) {\n const [authorQuery] = bibAuthors(record);\n if (authorQuery !== undefined) {\n return [`${authorQuery} AND ${titleQuery}`];\n }\n return addPublisherToSearch(titleQuery);\n //return [];\n }\n\n function addPublisherToSearch(query) {\n const [publisherQuery] = bibPublishers(record);\n if (publisherQuery !== undefined) {\n return [`${publisherQuery} AND ${query}`];\n }\n return [];\n }\n\n function getTitle() {\n const [field] = record.get(/^245$/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 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 const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\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 = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\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 ? '' : '^'}${formatted}*\"`];\n }\n //if (formatted) {\n // return [`dc.author=\"${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))\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)$/u);\n //debugData(field);\n\n if (field) {\n const publisherString = field.subfields\n .filter(({code}) => ['b'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .filter(value => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n return publisherString;\n }\n return false;\n }\n}\n\n\nexport function bibStandardIdentifiers(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n const debugData = debug.extend('data');\n debug(`Creating queries for standard identifiers`);\n\n // DEVELOP: should we query also f015 and f028?\n\n const fields = record.get(/^(?<def>020|022|024)$/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"],"mappings":";;;;;;;;;;;;AA4BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,qBAAA,GAAAD,OAAA;AACA,IAAAE,cAAA,GAAAF,OAAA;AAA8H,SAAAD,uBAAAI,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA7B9H;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;;AAMO,SAASG,YAAYA,CAACC,MAAM,EAAE;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtC;;EAEAH,KAAK,CAAE,iCAAgC,CAAC;EAExC,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAG,CAAC,KAAK,CAAC;EAC/BH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACJ,KAAK,CAAE,EAAC,CAAC;EAEnE,OAAOA,KAAK,CAACE,MAAM,GAAG,CAAC,GAAGG,YAAY,CAACL,KAAK,CAAC,GAAG,EAAE;EAElD,SAASK,YAAYA,CAACL,KAAK,EAAE;IAC3BJ,KAAK,CAAE,wCAAuC,CAAC;IAE/C,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAK,CAAC;IAEvC,IAAIM,UAAU,CAACJ,MAAM,GAAG,CAAC,EAAE;MACzBN,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO,EAAE;IACX;IAEA,MAAMY,UAAU,GAAG,IAAAC,+BAAS,EAACH,UAAU,EAAE,kBAAkB,CAAC;IAE5D,OAAOE,UAAU;IAEjB,SAASD,aAAaA,CAACP,KAAK,EAAE;MAC5BJ,KAAK,CAAE,qCAAoC,CAAC;;MAE5C;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAG,CAACC,WAAW,CAAC,CAACC,MAAM,CAACC,WAAW,IAAIA,WAAW,CAAC;MAC5E,OAAOP,UAAU;MAEjB,SAASK,WAAWA,CAACG,KAAK,EAAE;QAC1BlB,KAAK,CAAE,6BAA4B,CAAC;QAEpC,OAAO,IAAAmB,6CAA8B,EAACD,KAAK,CAAC,GAAGE,eAAe,CAACF,KAAK,CAAC,GAAG,EAAE;QAE1E,SAASE,eAAeA,CAACF,KAAK,EAAE;UAC9BlB,KAAK,CAAE,8BAA6B,CAAC;UACrC,MAAM,CAACqB,GAAG,CAAC,GAAG,IAAAC,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAC3C,MAAM,CAACK,GAAG,CAAC,GAAG,IAAAD,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAE3C,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAG,CAAC,CAAC;UACrE,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAG,CAAC;UAEjDrB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACa,GAAG,CAAE,MAAKd,IAAI,CAACC,SAAS,CAACe,GAAG,CAAE,EAAC,CAAC;UAC5D,OAAOC,UAAU,CAACI,MAAM,CAACD,UAAU,CAAC;QACtC;QAEA,SAASF,kBAAkBA,CAACI,aAAa,EAAE;UACzC,MAAMC,iBAAiB,GAAI,gDAAiD;UAC5E,MAAMC,eAAe,GAAG,IAAAC,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACJ,iBAAiB,EAAE,OAAO,CAAC,GAAG,EAAE;UAC1H5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAC,CAAC;UAC9D,OAAOA,eAAe;QACxB;QAEA,SAASL,yBAAyBA,CAACG,aAAa,EAAE;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAAC,CAAC;UACzC,MAAMM,kBAAkB,GAAI,OAAQ;UACpC,OAAO,IAAAH,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACC,kBAAkB,EAAE,EAAE,CAAC,GAAG,EAAE;QACvG;MAEF;IACF;EACF;AACF;AAEO,SAASC,aAAaA,CAACrC,MAAM,EAAE;EACpC;;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wEAAwE,CAAC;EACzG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,iCAAgC,CAAC;;EAExC;EACA;EACA;EACA,MAAMqC,UAAU,GAAG,IAAAC,gCAAiB,EAACvC,MAAM,CAAC;EAE5CG,SAAS,CAAE,uBAAsBmC,UAAU,CAAC/B,MAAO,MAAKC,IAAI,CAACC,SAAS,CAAC6B,UAAU,CAAE,EAAC,CAAC;EAErF,IAAIA,UAAU,CAAC/B,MAAM,GAAG,CAAC,EAAE;IACzBN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACwB,UAAU,EAAE,mBAAmB,CAAC;AACnD;;AAGA;AACA;AACO,SAASE,iBAAiBA,CAACxC,MAAM,EAAE;EACxC,MAAMyC,EAAE,GAAGC,SAAS,CAAC,CAAC;EACtB,OAAO,IAAAT,iCAAkB,EAACQ,EAAE,CAAC,GAAG,CAAE,uBAAsBA,EAAG,EAAC,CAAC,GAAG,EAAE;EAElE,SAASC,SAASA,CAAA,EAAG;IACnB,MAAM,CAACvB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM;QAACwB;MAAK,CAAC,GAAGxB,KAAK,CAACyB,SAAS,CAACC,IAAI,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC;MAEpE,IAAI,IAAAb,iCAAkB,EAACU,KAAK,CAAC,IAAK,kBAAkB,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACzE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,kBAAkB,EAAE,EAAE,CAAC;MACtD;MAEA,IAAI,IAAAF,iCAAkB,EAACU,KAAK,CAAC,IAAK,aAAa,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACpE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC;MACjD;MAEA,OAAO,KAAK;IACd;IACA,OAAO,KAAK;EACd;AACF;;AAEA;AACA;AACA;;AAEA;AACA;AACA;;AAEO,SAASa,QAAQA,CAAChD,MAAM,EAAE;EAC/B,MAAMiD,KAAK,GAAGC,QAAQ,CAAC,CAAC;EACxB,MAAMC,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAErD,IAAI,IAAAlB,iCAAkB,EAACgB,KAAK,CAAC,EAAE;IAC7B,MAAMG,SAAS,GAAGlB,MAAM,CAACe,KAAK,CAAC,CAC5Bd,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,CAAC,CAAC,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,CAAC,CAAC;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,CAAC,CAAC,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACA;IACA,IAAIL,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,aAAYgD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IAChE;IACA;IACA,OAAOQ,kBAAkB,CAAE,aAAYR,SAAU,GAAE,CAAC;EACtD;EAEA,OAAO,EAAE;EAET,SAASQ,kBAAkBA,CAACC,UAAU,EAAE;IACtC,MAAM,CAACC,WAAW,CAAC,GAAGC,UAAU,CAAC/D,MAAM,CAAC;IACxC,IAAI8D,WAAW,KAAKE,SAAS,EAAE;MAC7B,OAAO,CAAE,GAAEF,WAAY,QAAOD,UAAW,EAAC,CAAC;IAC7C;IACA,OAAOI,oBAAoB,CAACJ,UAAU,CAAC;IACvC;EACF;;EAEA,SAASI,oBAAoBA,CAACC,KAAK,EAAE;IACnC,MAAM,CAACC,cAAc,CAAC,GAAGC,aAAa,CAACpE,MAAM,CAAC;IAC9C,IAAImE,cAAc,KAAKH,SAAS,EAAE;MAChC,OAAO,CAAE,GAAEG,cAAe,QAAOD,KAAM,EAAC,CAAC;IAC3C;IACA,OAAO,EAAE;EACX;EAEA,SAAShB,QAAQA,CAAA,EAAG;IAClB,MAAM,CAAC/B,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAMkD,WAAW,GAAGlD,KAAK,CAACyB;MACxB;MAAA,CACC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACwB,QAAQ,CAACxB,IAAI,CAAC;MAC7C;MAAA,CACC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACC4B,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOF,WAAW;IACpB;IACA,OAAO,KAAK;EACd;AACF;AAEO,SAASN,UAAUA,CAAC/D,MAAM,EAAE;EACjC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,qCAAoC,CAAC;EAC5C;;EAEA,MAAMuE,MAAM,GAAGC,SAAS,CAACzE,MAAM,CAAC;EAChC,MAAMmD,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAGrD,IAAI,IAAAlB,iCAAkB,EAACuC,MAAM,CAAC,EAAE;IAC9B,MAAMpB,SAAS,GAAGlB,MAAM,CAACsC,MAAM,CAAC,CAC7BrC,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,CAAC,CAAC,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,CAAC,CAAC;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,CAAC,CAAC,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACAtD,SAAS,CAAE,kBAAiBiD,SAAU,EAAC,CAAC;IACxC,IAAIA,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,cAAagD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IACjE;IACA;IACA;IACA;IACA,OAAO,EAAE;EACX;EAEA,OAAO,EAAE;EAET,SAASqB,SAASA,CAACzE,MAAM,EAAE;IACzB;IACA;IACA,MAAM,CAACmB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,wCAAwC,CAAC;IACpE;;IAEA,IAAIa,KAAK,EAAE;MACT,MAAMuD,YAAY,GAAGvD,KAAK,CAACyB,SAAS,CACjC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAACwB,QAAQ,CAACxB,IAAI,CAAC,CAAC,CACxC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACC4B,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOG,YAAY;IACrB;IACA,OAAO,KAAK;EACd;AACF;AAEO,SAASN,aAAaA,CAACpE,MAAM,EAAE;EACpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wEAAwE,CAAC;EACzG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,wCAAuC,CAAC;EAC/C;;EAEA,MAAM0E,SAAS,GAAGC,YAAY,CAAC5E,MAAM,CAAC;EACtC,IAAI,IAAAiC,iCAAkB,EAAC0C,SAAS,CAAC,EAAE;IACjC,MAAMvB,SAAS,GAAGlB,MAAM,CAACyC,SAAS,CAAC,CAChCxC,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,CAAC,CAAC,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,CAAC,CAAC;IAETlD,SAAS,CAAE,qBAAoBiD,SAAU,EAAC,CAAC;IAC3C;IACA,OAAO,CAAE,iBAAgBA,SAAU,GAAE,CAAC;EACxC;EAEA,OAAO,EAAE;EAET,SAASwB,YAAYA,CAAC5E,MAAM,EAAE;IAC5B;IACA,MAAM,CAACmB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,oBAAoB,CAAC;IAChD;;IAEA,IAAIa,KAAK,EAAE;MACT,MAAM0D,eAAe,GAAG1D,KAAK,CAACyB,SAAS,CACpC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAACwB,QAAQ,CAACxB,IAAI,CAAC,CAAC,CACxC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACC4B,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOM,eAAe;IACxB;IACA,OAAO,KAAK;EACd;AACF;AAGO,SAASC,sBAAsBA,CAAC9E,MAAM,EAAE;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;EAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,2CAA0C,CAAC;;EAElD;;EAEA,MAAM8E,MAAM,GAAG/E,MAAM,CAACM,GAAG,CAAC,wBAAwB,CAAC;EACnD,MAAM0E,WAAW,GAAG,EAAE,CAACnD,MAAM,CAAC,GAAGkD,MAAM,CAAChE,GAAG,CAACkE,aAAa,CAAC,CAAC;EAC3D,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACH,WAAW,CAAC,CAAC;EAEnD7E,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAS,CAACsE,MAAM,CAAE,EAAC,CAAC;EAClE5E,SAAS,CAAE,gBAAe6E,WAAW,CAACzE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACuE,WAAW,CAAE,EAAC,CAAC;EAChF7E,SAAS,CAAE,uBAAsB+E,iBAAiB,CAAC3E,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACyE,iBAAiB,CAAE,EAAC,CAAC;EAEnG,IAAIA,iBAAiB,CAAC3E,MAAM,GAAG,CAAC,EAAE;IAChCN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACoE,iBAAiB,EAAE,eAAe,CAAC;EAEpD,SAASD,aAAaA,CAAC;IAACG,GAAG;IAAExC;EAAS,CAAC,EAAE;IACvC,MAAMyC,cAAc,GAAI,kBAAmB;IAC3C,MAAMC,aAAa,GAAI,mBAAoB;IAE3C,IAAIF,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOxC,SAAS,CACb3B,MAAM,CAACsE,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAACjB,QAAQ,CAACiB,GAAG,CAACzC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsD,GAAG,CAAC5C,KAAK,CAAC,IAAI0C,cAAc,CAACtC,IAAI,CAACb,MAAM,CAACqD,GAAG,CAAC5C,KAAK,CAAC,CAAC,CAAC,CAC5H5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,IAAIyC,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOxC,SAAS,CACb3B,MAAM,CAACsE,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACjB,QAAQ,CAACiB,GAAG,CAACzC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsD,GAAG,CAAC5C,KAAK,CAAC,IAAI0C,cAAc,CAACtC,IAAI,CAACb,MAAM,CAACqD,GAAG,CAAC5C,KAAK,CAAC,CAAC,CAAC,CACvH5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,OAAOC,SAAS,CACb3B,MAAM,CAACsE,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACjB,QAAQ,CAACiB,GAAG,CAACzC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsD,GAAG,CAAC5C,KAAK,CAAC,IAAI2C,aAAa,CAACvC,IAAI,CAACb,MAAM,CAACqD,GAAG,CAAC5C,KAAK,CAAC,CAAC,CAAC,CACtH5B,GAAG,CAAC,CAAC;MAAC4B;IAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;EACpC;AACF"}
|
|
@@ -45,6 +45,12 @@ Object.defineProperty(exports, "language", {
|
|
|
45
45
|
return _language.default;
|
|
46
46
|
}
|
|
47
47
|
});
|
|
48
|
+
Object.defineProperty(exports, "mediaType", {
|
|
49
|
+
enumerable: true,
|
|
50
|
+
get: function () {
|
|
51
|
+
return _mediaType.default;
|
|
52
|
+
}
|
|
53
|
+
});
|
|
48
54
|
Object.defineProperty(exports, "melindaId", {
|
|
49
55
|
enumerable: true,
|
|
50
56
|
get: function () {
|
|
@@ -108,5 +114,6 @@ var _language = _interopRequireDefault(require("./language"));
|
|
|
108
114
|
var _bibliographicLevel = _interopRequireDefault(require("./bibliographic-level"));
|
|
109
115
|
var _melindaId = _interopRequireDefault(require("./melinda-id"));
|
|
110
116
|
var _allSourceIds = _interopRequireDefault(require("./all-source-ids"));
|
|
117
|
+
var _mediaType = _interopRequireDefault(require("./media-type"));
|
|
111
118
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
112
119
|
//# sourceMappingURL=index.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","names":["_hostComponent","_interopRequireDefault","require","_isbn","_issn","_otherStandardIdentifier","_title","_titleVersionOriginal","_authors","_recordType","_publicationTime","_publicationTimeAllowConsYears","_publicationTimeAllowConsYearsMulti","_language","_bibliographicLevel","_melindaId","_allSourceIds","obj","__esModule","default"],"sources":["../../../../src/match-detection/features/bib/index.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport {default as hostComponent} from './host-component';\nexport {default as isbn} from './isbn';\nexport {default as issn} from './issn';\nexport {default as otherStandardIdentifier} from './other-standard-identifier';\nexport {default as title} from './title';\nexport {default as titleVersionOriginal} from './title-version-original';\nexport {default as authors} from './authors';\nexport {default as recordType} from './record-type';\nexport {default as publicationTime} from './publication-time';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years';\nexport {default as publicationTimeAllowConsYearsMulti} from './publication-time-allow-cons-years-multi';\nexport {default as language} from './language';\nexport {default as bibliographicLevel} from './bibliographic-level';\nexport {default as melindaId} from './melinda-id';\nexport {default as allSourceIds} from './all-source-ids';\n"],"mappings":"
|
|
1
|
+
{"version":3,"file":"index.js","names":["_hostComponent","_interopRequireDefault","require","_isbn","_issn","_otherStandardIdentifier","_title","_titleVersionOriginal","_authors","_recordType","_publicationTime","_publicationTimeAllowConsYears","_publicationTimeAllowConsYearsMulti","_language","_bibliographicLevel","_melindaId","_allSourceIds","_mediaType","obj","__esModule","default"],"sources":["../../../../src/match-detection/features/bib/index.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport {default as hostComponent} from './host-component';\nexport {default as isbn} from './isbn';\nexport {default as issn} from './issn';\nexport {default as otherStandardIdentifier} from './other-standard-identifier';\nexport {default as title} from './title';\nexport {default as titleVersionOriginal} from './title-version-original';\nexport {default as authors} from './authors';\nexport {default as recordType} from './record-type';\nexport {default as publicationTime} from './publication-time';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years';\nexport {default as publicationTimeAllowConsYearsMulti} from './publication-time-allow-cons-years-multi';\nexport {default as language} from './language';\nexport {default as bibliographicLevel} from './bibliographic-level';\nexport {default as melindaId} from './melinda-id';\nexport {default as allSourceIds} from './all-source-ids';\nexport {default as mediaType} from './media-type';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA4BA,IAAAA,cAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,KAAA,GAAAF,sBAAA,CAAAC,OAAA;AACA,IAAAE,KAAA,GAAAH,sBAAA,CAAAC,OAAA;AACA,IAAAG,wBAAA,GAAAJ,sBAAA,CAAAC,OAAA;AACA,IAAAI,MAAA,GAAAL,sBAAA,CAAAC,OAAA;AACA,IAAAK,qBAAA,GAAAN,sBAAA,CAAAC,OAAA;AACA,IAAAM,QAAA,GAAAP,sBAAA,CAAAC,OAAA;AACA,IAAAO,WAAA,GAAAR,sBAAA,CAAAC,OAAA;AACA,IAAAQ,gBAAA,GAAAT,sBAAA,CAAAC,OAAA;AACA,IAAAS,8BAAA,GAAAV,sBAAA,CAAAC,OAAA;AACA,IAAAU,mCAAA,GAAAX,sBAAA,CAAAC,OAAA;AACA,IAAAW,SAAA,GAAAZ,sBAAA,CAAAC,OAAA;AACA,IAAAY,mBAAA,GAAAb,sBAAA,CAAAC,OAAA;AACA,IAAAa,UAAA,GAAAd,sBAAA,CAAAC,OAAA;AACA,IAAAc,aAAA,GAAAf,sBAAA,CAAAC,OAAA;AACA,IAAAe,UAAA,GAAAhB,sBAAA,CAAAC,OAAA;AAAkD,SAAAD,uBAAAiB,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA"}
|
|
@@ -5,6 +5,7 @@ var _fixura = require("@natlibfi/fixura");
|
|
|
5
5
|
var _chai = require("chai");
|
|
6
6
|
var _marcRecord = require("@natlibfi/marc-record");
|
|
7
7
|
var features = _interopRequireWildcard(require("."));
|
|
8
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
8
9
|
function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
|
|
9
10
|
function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
|
|
10
11
|
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
@@ -36,6 +37,8 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
36
37
|
*
|
|
37
38
|
*/
|
|
38
39
|
|
|
40
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features/bib:test');
|
|
41
|
+
const debugData = debug.extend('data');
|
|
39
42
|
describe('match-detection/features/bib/', () => {
|
|
40
43
|
(0, _fixugen.default)({
|
|
41
44
|
path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],
|
|
@@ -53,6 +56,7 @@ describe('match-detection/features/bib/', () => {
|
|
|
53
56
|
if (!enabled) {
|
|
54
57
|
return;
|
|
55
58
|
}
|
|
59
|
+
debug(`Testing: ${feature} ${type}`);
|
|
56
60
|
if (type === 'extract') {
|
|
57
61
|
const {
|
|
58
62
|
expectedFeatures,
|
|
@@ -61,6 +65,7 @@ describe('match-detection/features/bib/', () => {
|
|
|
61
65
|
const record = new _marcRecord.MarcRecord(inputRecord, {
|
|
62
66
|
subfieldValues: false
|
|
63
67
|
});
|
|
68
|
+
debugData(`Record: ${record}`);
|
|
64
69
|
const {
|
|
65
70
|
extract
|
|
66
71
|
} = features[feature](options);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.spec.js","names":["_fixugen","_interopRequireDefault","require","_fixura","_chai","_marcRecord","features","_interopRequireWildcard","_getRequireWildcardCache","nodeInterop","WeakMap","cacheBabelInterop","cacheNodeInterop","obj","__esModule","default","cache","has","get","newObj","hasPropertyDescriptor","Object","defineProperty","getOwnPropertyDescriptor","key","prototype","hasOwnProperty","call","desc","set","describe","generateTests","path","__dirname","useMetadataFile","fixura","reader","READERS","JSON","callback","enabled","feature","options","type","expectations","expectedFeatures","inputRecord","record","MarcRecord","subfieldValues","extract","expect","to","eql","featuresA","featuresB","expectedPoints","compare","equal","Error"],"sources":["../../../../src/match-detection/features/bib/index.spec.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from '.';\n\n\ndescribe('match-detection/features/bib/', () => {\n generateTests({\n path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n useMetadataFile: true,\n fixura: {\n reader: READERS.JSON\n },\n\n callback: ({enabled = true, feature, options, type, ...expectations}) => {\n\n if (!enabled) {\n return;\n }\n\n if (type === 'extract') {\n const {expectedFeatures, inputRecord} = expectations;\n const record = new MarcRecord(inputRecord, {subfieldValues: false});\n const {extract} = features[feature](options);\n\n expect(extract({record})).to.eql(expectedFeatures);\n return;\n }\n\n if (type === 'compare') {\n const {featuresA, featuresB, expectedPoints} = expectations;\n const {compare} = features[feature](options);\n\n expect(compare(featuresA, featuresB)).to.equal(expectedPoints);\n return;\n }\n\n throw new Error(`Invalid type ${type}`);\n }\n });\n});\n"],"mappings":";;AA4BA,IAAAA,QAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,OAAA,GAAAD,OAAA;AACA,IAAAE,KAAA,GAAAF,OAAA;AACA,IAAAG,WAAA,GAAAH,OAAA;AACA,IAAAI,QAAA,GAAAC,uBAAA,CAAAL,OAAA;
|
|
1
|
+
{"version":3,"file":"index.spec.js","names":["_fixugen","_interopRequireDefault","require","_fixura","_chai","_marcRecord","features","_interopRequireWildcard","_debug","_getRequireWildcardCache","nodeInterop","WeakMap","cacheBabelInterop","cacheNodeInterop","obj","__esModule","default","cache","has","get","newObj","hasPropertyDescriptor","Object","defineProperty","getOwnPropertyDescriptor","key","prototype","hasOwnProperty","call","desc","set","debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","fixura","reader","READERS","JSON","callback","enabled","feature","options","type","expectations","expectedFeatures","inputRecord","record","MarcRecord","subfieldValues","extract","expect","to","eql","featuresA","featuresB","expectedPoints","compare","equal","Error"],"sources":["../../../../src/match-detection/features/bib/index.spec.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from '.';\nimport createDebugLogger from 'debug';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib:test');\nconst debugData = debug.extend('data');\n\n\ndescribe('match-detection/features/bib/', () => {\n generateTests({\n path: [__dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n useMetadataFile: true,\n fixura: {\n reader: READERS.JSON\n },\n\n callback: ({enabled = true, feature, options, type, ...expectations}) => {\n\n if (!enabled) {\n return;\n }\n\n debug(`Testing: ${feature} ${type}`);\n\n if (type === 'extract') {\n const {expectedFeatures, inputRecord} = expectations;\n const record = new MarcRecord(inputRecord, {subfieldValues: false});\n debugData(`Record: ${record}`);\n const {extract} = features[feature](options);\n\n expect(extract({record})).to.eql(expectedFeatures);\n return;\n }\n\n if (type === 'compare') {\n const {featuresA, featuresB, expectedPoints} = expectations;\n const {compare} = features[feature](options);\n\n expect(compare(featuresA, featuresB)).to.equal(expectedPoints);\n return;\n }\n\n throw new Error(`Invalid type ${type}`);\n }\n });\n});\n"],"mappings":";;AA4BA,IAAAA,QAAA,GAAAC,sBAAA,CAAAC,OAAA;AACA,IAAAC,OAAA,GAAAD,OAAA;AACA,IAAAE,KAAA,GAAAF,OAAA;AACA,IAAAG,WAAA,GAAAH,OAAA;AACA,IAAAI,QAAA,GAAAC,uBAAA,CAAAL,OAAA;AACA,IAAAM,MAAA,GAAAP,sBAAA,CAAAC,OAAA;AAAsC,SAAAO,yBAAAC,WAAA,eAAAC,OAAA,kCAAAC,iBAAA,OAAAD,OAAA,QAAAE,gBAAA,OAAAF,OAAA,YAAAF,wBAAA,YAAAA,CAAAC,WAAA,WAAAA,WAAA,GAAAG,gBAAA,GAAAD,iBAAA,KAAAF,WAAA;AAAA,SAAAH,wBAAAO,GAAA,EAAAJ,WAAA,SAAAA,WAAA,IAAAI,GAAA,IAAAA,GAAA,CAAAC,UAAA,WAAAD,GAAA,QAAAA,GAAA,oBAAAA,GAAA,wBAAAA,GAAA,4BAAAE,OAAA,EAAAF,GAAA,UAAAG,KAAA,GAAAR,wBAAA,CAAAC,WAAA,OAAAO,KAAA,IAAAA,KAAA,CAAAC,GAAA,CAAAJ,GAAA,YAAAG,KAAA,CAAAE,GAAA,CAAAL,GAAA,SAAAM,MAAA,WAAAC,qBAAA,GAAAC,MAAA,CAAAC,cAAA,IAAAD,MAAA,CAAAE,wBAAA,WAAAC,GAAA,IAAAX,GAAA,QAAAW,GAAA,kBAAAH,MAAA,CAAAI,SAAA,CAAAC,cAAA,CAAAC,IAAA,CAAAd,GAAA,EAAAW,GAAA,SAAAI,IAAA,GAAAR,qBAAA,GAAAC,MAAA,CAAAE,wBAAA,CAAAV,GAAA,EAAAW,GAAA,cAAAI,IAAA,KAAAA,IAAA,CAAAV,GAAA,IAAAU,IAAA,CAAAC,GAAA,KAAAR,MAAA,CAAAC,cAAA,CAAAH,MAAA,EAAAK,GAAA,EAAAI,IAAA,YAAAT,MAAA,CAAAK,GAAA,IAAAX,GAAA,CAAAW,GAAA,SAAAL,MAAA,CAAAJ,OAAA,GAAAF,GAAA,MAAAG,KAAA,IAAAA,KAAA,CAAAa,GAAA,CAAAhB,GAAA,EAAAM,MAAA,YAAAA,MAAA;AAAA,SAAAnB,uBAAAa,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AAjCtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAUA,MAAMiB,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;AACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAGtCC,QAAQ,CAAC,+BAA+B,EAAE,MAAM;EAC9C,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,UAAU,EAAE,KAAK,CAAC;IAChGC,eAAe,EAAE,IAAI;IACrBC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IAEDC,QAAQ,EAAEA,CAAC;MAACC,OAAO,GAAG,IAAI;MAAEC,OAAO;MAAEC,OAAO;MAAEC,IAAI;MAAE,GAAGC;IAAY,CAAC,KAAK;MAEvE,IAAI,CAACJ,OAAO,EAAE;QACZ;MACF;MAEAd,KAAK,CAAE,YAAWe,OAAQ,IAAGE,IAAK,EAAC,CAAC;MAEpC,IAAIA,IAAI,KAAK,SAAS,EAAE;QACtB,MAAM;UAACE,gBAAgB;UAAEC;QAAW,CAAC,GAAGF,YAAY;QACpD,MAAMG,MAAM,GAAG,IAAIC,sBAAU,CAACF,WAAW,EAAE;UAACG,cAAc,EAAE;QAAK,CAAC,CAAC;QACnErB,SAAS,CAAE,WAAUmB,MAAO,EAAC,CAAC;QAC9B,MAAM;UAACG;QAAO,CAAC,GAAGjD,QAAQ,CAACwC,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAS,YAAM,EAACD,OAAO,CAAC;UAACH;QAAM,CAAC,CAAC,CAAC,CAACK,EAAE,CAACC,GAAG,CAACR,gBAAgB,CAAC;QAClD;MACF;MAEA,IAAIF,IAAI,KAAK,SAAS,EAAE;QACtB,MAAM;UAACW,SAAS;UAAEC,SAAS;UAAEC;QAAc,CAAC,GAAGZ,YAAY;QAC3D,MAAM;UAACa;QAAO,CAAC,GAAGxD,QAAQ,CAACwC,OAAO,CAAC,CAACC,OAAO,CAAC;QAE5C,IAAAS,YAAM,EAACM,OAAO,CAACH,SAAS,EAAEC,SAAS,CAAC,CAAC,CAACH,EAAE,CAACM,KAAK,CAACF,cAAc,CAAC;QAC9D;MACF;MAEA,MAAM,IAAIG,KAAK,CAAE,gBAAehB,IAAK,EAAC,CAAC;IACzC;EACF,CAAC,CAAC;AACJ,CAAC,CAAC"}
|
|
@@ -0,0 +1,83 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
|
|
3
|
+
Object.defineProperty(exports, "__esModule", {
|
|
4
|
+
value: true
|
|
5
|
+
});
|
|
6
|
+
exports.default = void 0;
|
|
7
|
+
var _debug = _interopRequireDefault(require("debug"));
|
|
8
|
+
function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
|
|
9
|
+
/* eslint-disable max-statements */
|
|
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) 2020-2022 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
|
+
|
|
38
|
+
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features:media-type');
|
|
39
|
+
const debugData = debug.extend('data');
|
|
40
|
+
var _default = () => ({
|
|
41
|
+
name: 'Media type',
|
|
42
|
+
extract: ({
|
|
43
|
+
record,
|
|
44
|
+
recordExternal
|
|
45
|
+
}) => {
|
|
46
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
47
|
+
debugData(`Record (${label}): ${JSON.stringify(record)}`);
|
|
48
|
+
debugData(`RecordExternal: ${JSON.stringify(recordExternal)}`);
|
|
49
|
+
const values337 = get337Values();
|
|
50
|
+
debug(`${label} 337 $b values: ${JSON.stringify(values337)}`);
|
|
51
|
+
return values337;
|
|
52
|
+
function get337Values() {
|
|
53
|
+
return record.get(/^337$/u).filter(f => f.subfields.some(subfield => subfield.code === '2' && subfield.value === 'rdamedia')).map(({
|
|
54
|
+
subfields
|
|
55
|
+
}) => subfields).flat().filter(({
|
|
56
|
+
code
|
|
57
|
+
}) => code === 'b').map(({
|
|
58
|
+
value
|
|
59
|
+
}) => value);
|
|
60
|
+
}
|
|
61
|
+
},
|
|
62
|
+
compare: (a, b) => {
|
|
63
|
+
debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);
|
|
64
|
+
|
|
65
|
+
// Should we give extra good points if all mediaTypes match?
|
|
66
|
+
// Should we give partial points for partially matching mediaTypes?
|
|
67
|
+
// Should we check whether recordType is 'mixedMaterials'
|
|
68
|
+
// Should we okay typical cases of not totally matching mediaTypes? What would these be?
|
|
69
|
+
|
|
70
|
+
if (a.every(elem => b.includes(elem))) {
|
|
71
|
+
debug(`All mediaTypes from A are in B`);
|
|
72
|
+
return 1;
|
|
73
|
+
}
|
|
74
|
+
if (b.every(elem => a.includes(elem))) {
|
|
75
|
+
debug(`All mediaTypes from B are in A`);
|
|
76
|
+
return 1;
|
|
77
|
+
}
|
|
78
|
+
debug(`Mismatch in mediaTypes between A and B`);
|
|
79
|
+
return -1;
|
|
80
|
+
}
|
|
81
|
+
});
|
|
82
|
+
exports.default = _default;
|
|
83
|
+
//# sourceMappingURL=media-type.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"media-type.js","names":["_debug","_interopRequireDefault","require","obj","__esModule","default","debug","createDebugLogger","debugData","extend","_default","name","extract","record","recordExternal","label","JSON","stringify","values337","get337Values","get","filter","f","subfields","some","subfield","code","value","map","flat","compare","a","b","every","elem","includes","exports"],"sources":["../../../../src/match-detection/features/bib/media-type.js"],"sourcesContent":["/* eslint-disable max-statements */\n/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:media-type');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n name: 'Media type',\n extract: ({record, recordExternal}) => {\n const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n debugData(`Record (${label}): ${JSON.stringify(record)}`);\n debugData(`RecordExternal: ${JSON.stringify(recordExternal)}`);\n const values337 = get337Values();\n debug(`${label} 337 $b values: ${JSON.stringify(values337)}`);\n\n return values337;\n\n function get337Values() {\n return record.get(/^337$/u)\n .filter(f => f.subfields.some((subfield) => subfield.code === '2' && subfield.value === 'rdamedia'))\n .map(({subfields}) => subfields)\n .flat()\n .filter(({code}) => code === 'b')\n .map(({value}) => value);\n }\n },\n compare: (a, b) => {\n debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);\n\n // Should we give extra good points if all mediaTypes match?\n // Should we give partial points for partially matching mediaTypes?\n // Should we check whether recordType is 'mixedMaterials'\n // Should we okay typical cases of not totally matching mediaTypes? What would these be?\n\n if (a.every(elem => b.includes(elem))) {\n debug(`All mediaTypes from A are in B`);\n return 1;\n }\n\n if (b.every(elem => a.includes(elem))) {\n debug(`All mediaTypes from B are in A`);\n return 1;\n }\n\n debug(`Mismatch in mediaTypes between A and B`);\n return -1;\n\n }\n});\n"],"mappings":";;;;;;AA6BA,IAAAA,MAAA,GAAAC,sBAAA,CAAAC,OAAA;AAAsC,SAAAD,uBAAAE,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;AACA;;AAIA,MAAMG,KAAK,GAAG,IAAAC,cAAiB,EAAC,uEAAuE,CAAC;AACxG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,IAAAC,QAAA,GAExBA,CAAA,MAAO;EACpBC,IAAI,EAAE,YAAY;EAClBC,OAAO,EAAEA,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,KAAK;IACrC,MAAMC,KAAK,GAAGD,cAAc,IAAIA,cAAc,CAACC,KAAK,GAAGD,cAAc,CAACC,KAAK,GAAG,QAAQ;IACtFP,SAAS,CAAE,WAAUO,KAAM,MAAKC,IAAI,CAACC,SAAS,CAACJ,MAAM,CAAE,EAAC,CAAC;IACzDL,SAAS,CAAE,mBAAkBQ,IAAI,CAACC,SAAS,CAACH,cAAc,CAAE,EAAC,CAAC;IAC9D,MAAMI,SAAS,GAAGC,YAAY,CAAC,CAAC;IAChCb,KAAK,CAAE,GAAES,KAAM,mBAAkBC,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IAE7D,OAAOA,SAAS;IAEhB,SAASC,YAAYA,CAAA,EAAG;MACtB,OAAON,MAAM,CAACO,GAAG,CAAC,QAAQ,CAAC,CACxBC,MAAM,CAACC,CAAC,IAAIA,CAAC,CAACC,SAAS,CAACC,IAAI,CAAEC,QAAQ,IAAKA,QAAQ,CAACC,IAAI,KAAK,GAAG,IAAID,QAAQ,CAACE,KAAK,KAAK,UAAU,CAAC,CAAC,CACnGC,GAAG,CAAC,CAAC;QAACL;MAAS,CAAC,KAAKA,SAAS,CAAC,CAC/BM,IAAI,CAAC,CAAC,CACNR,MAAM,CAAC,CAAC;QAACK;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCE,GAAG,CAAC,CAAC;QAACD;MAAK,CAAC,KAAKA,KAAK,CAAC;IAC5B;EACF,CAAC;EACDG,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjBxB,SAAS,CAAE,aAAYQ,IAAI,CAACC,SAAS,CAACc,CAAC,CAAE,QAAOf,IAAI,CAACC,SAAS,CAACe,CAAC,CAAE,EAAC,CAAC;;IAEpE;IACA;IACA;IACA;;IAEA,IAAID,CAAC,CAACE,KAAK,CAACC,IAAI,IAAIF,CAAC,CAACG,QAAQ,CAACD,IAAI,CAAC,CAAC,EAAE;MACrC5B,KAAK,CAAE,gCAA+B,CAAC;MACvC,OAAO,CAAC;IACV;IAEA,IAAI0B,CAAC,CAACC,KAAK,CAACC,IAAI,IAAIH,CAAC,CAACI,QAAQ,CAACD,IAAI,CAAC,CAAC,EAAE;MACrC5B,KAAK,CAAE,gCAA+B,CAAC;MACvC,OAAO,CAAC;IACV;IAEAA,KAAK,CAAE,wCAAuC,CAAC;IAC/C,OAAO,CAAC,CAAC;EAEX;AACF,CAAC,CAAC;AAAA8B,OAAA,CAAA/B,OAAA,GAAAK,QAAA"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"other-standard-identifier.js","names":["_standardIdentifierFactory","_interopRequireDefault","require","obj","__esModule","default","_default","extract","compare","createInterface","pattern","subfieldCodes","name","exports"],"sources":["../../../../src/match-detection/features/bib/other-standard-identifier.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createInterface from './standard-identifier-factory';\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^024$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'Other standard identifier'};\n};\n"],"mappings":";;;;;;AA4BA,IAAAA,0BAAA,GAAAC,sBAAA,CAAAC,OAAA;AAA4D,SAAAD,uBAAAE,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA5B5D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;
|
|
1
|
+
{"version":3,"file":"other-standard-identifier.js","names":["_standardIdentifierFactory","_interopRequireDefault","require","obj","__esModule","default","_default","extract","compare","createInterface","pattern","subfieldCodes","name","exports"],"sources":["../../../../src/match-detection/features/bib/other-standard-identifier.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createInterface from './standard-identifier-factory';\n\n// DEVELOP: we should compare indicators and sf2 for f024\n\nexport default () => {\n const {extract, compare} = createInterface({pattern: /^024$/u, subfieldCodes: ['a', 'z']});\n return {extract, compare, name: 'Other standard identifier'};\n};\n"],"mappings":";;;;;;AA4BA,IAAAA,0BAAA,GAAAC,sBAAA,CAAAC,OAAA;AAA4D,SAAAD,uBAAAE,GAAA,WAAAA,GAAA,IAAAA,GAAA,CAAAC,UAAA,GAAAD,GAAA,KAAAE,OAAA,EAAAF,GAAA;AA5B5D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA;AAAA,IAAAG,QAAA,GAEeA,CAAA,KAAM;EACnB,MAAM;IAACC,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG;EAAC,CAAC,CAAC;EAC1F,OAAO;IAACJ,OAAO;IAAEC,OAAO;IAAEI,IAAI,EAAE;EAA2B,CAAC;AAC9D,CAAC;AAAAC,OAAA,CAAAR,OAAA,GAAAC,QAAA"}
|
|
@@ -32,6 +32,8 @@ exports.default = void 0;
|
|
|
32
32
|
*
|
|
33
33
|
*/
|
|
34
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
|
|
35
37
|
var _default = () => ({
|
|
36
38
|
name: 'Record type',
|
|
37
39
|
extract: ({
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"record-type.js","names":["_default","name","extract","record","leader","compare","a","b","exports","default"],"sources":["../../../../src/match-detection/features/bib/record-type.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\n// we could handle the case of books/notes\n\nexport default () => ({\n name: 'Record type',\n extract: ({record}) => record.leader[6] ? [record.leader[6]] : [],\n compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5\n});\n"],"mappings":";;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;AAAA,IAAAA,QAAA,GAEeA,CAAA,MAAO;EACpBC,IAAI,EAAE,aAAa;EACnBC,OAAO,EAAEA,CAAC;IAACC;EAAM,CAAC,KAAKA,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,GAAG,CAACD,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE;EACjEC,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAAC,OAAA,CAAAC,OAAA,GAAAT,QAAA"}
|
|
1
|
+
{"version":3,"file":"record-type.js","names":["_default","name","extract","record","leader","compare","a","b","exports","default"],"sources":["../../../../src/match-detection/features/bib/record-type.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\n// we could handle the case of books/notes\n// Recordtype: LDR/06 - Type of Record\n// Note: currently matchValidator fails all mismatching recordTypes, so this won't actually do much\n\nexport default () => ({\n name: 'Record type',\n extract: ({record}) => record.leader[6] ? [record.leader[6]] : [],\n compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5\n});\n"],"mappings":";;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;AACA;AACA;AAAA,IAAAA,QAAA,GAEeA,CAAA,MAAO;EACpBC,IAAI,EAAE,aAAa;EACnBC,OAAO,EAAEA,CAAC;IAACC;EAAM,CAAC,KAAKA,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,GAAG,CAACD,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE;EACjEC,OAAO,EAAEA,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAAC,OAAA,CAAAC,OAAA,GAAAT,QAAA"}
|
package/package.json
CHANGED
|
@@ -201,7 +201,8 @@ export function bibTitle(record) {
|
|
|
201
201
|
if (formatted.length >= 5) {
|
|
202
202
|
return [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`];
|
|
203
203
|
}
|
|
204
|
-
|
|
204
|
+
// use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]
|
|
205
|
+
return addAuthorsToSearch(`dc.title="${formatted}"`);
|
|
205
206
|
}
|
|
206
207
|
|
|
207
208
|
return [];
|
|
@@ -211,6 +212,15 @@ export function bibTitle(record) {
|
|
|
211
212
|
if (authorQuery !== undefined) {
|
|
212
213
|
return [`${authorQuery} AND ${titleQuery}`];
|
|
213
214
|
}
|
|
215
|
+
return addPublisherToSearch(titleQuery);
|
|
216
|
+
//return [];
|
|
217
|
+
}
|
|
218
|
+
|
|
219
|
+
function addPublisherToSearch(query) {
|
|
220
|
+
const [publisherQuery] = bibPublishers(record);
|
|
221
|
+
if (publisherQuery !== undefined) {
|
|
222
|
+
return [`${publisherQuery} AND ${query}`];
|
|
223
|
+
}
|
|
214
224
|
return [];
|
|
215
225
|
}
|
|
216
226
|
|
|
@@ -258,6 +268,9 @@ export function bibAuthors(record) {
|
|
|
258
268
|
if (formatted.length >= 5) {
|
|
259
269
|
return [`dc.author="${useWordSearch ? '' : '^'}${formatted}*"`];
|
|
260
270
|
}
|
|
271
|
+
//if (formatted) {
|
|
272
|
+
// return [`dc.author="${formatted}"`];
|
|
273
|
+
//}
|
|
261
274
|
return [];
|
|
262
275
|
}
|
|
263
276
|
|
|
@@ -282,6 +295,47 @@ export function bibAuthors(record) {
|
|
|
282
295
|
}
|
|
283
296
|
}
|
|
284
297
|
|
|
298
|
+
export function bibPublishers(record) {
|
|
299
|
+
const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');
|
|
300
|
+
const debugData = debug.extend('data');
|
|
301
|
+
debug(`Creating query for the first publisher`);
|
|
302
|
+
//debugData(record);
|
|
303
|
+
|
|
304
|
+
const publisher = getPublisher(record);
|
|
305
|
+
if (testStringOrNumber(publisher)) {
|
|
306
|
+
const formatted = String(publisher)
|
|
307
|
+
.replace(/[^\w\s\p{Alphabetic}]/gu, '')
|
|
308
|
+
// Clean up concurrent spaces from fe. subfield changes
|
|
309
|
+
.replace(/ +/gu, ' ')
|
|
310
|
+
.trim()
|
|
311
|
+
.slice(0, 30)
|
|
312
|
+
.trim();
|
|
313
|
+
|
|
314
|
+
debugData(`Publisher string: ${formatted}`);
|
|
315
|
+
// use non-wildcard word search from dc.publisher
|
|
316
|
+
return [`dc.publisher="${formatted}"`];
|
|
317
|
+
}
|
|
318
|
+
|
|
319
|
+
return [];
|
|
320
|
+
|
|
321
|
+
function getPublisher(record) {
|
|
322
|
+
//debugData(record);
|
|
323
|
+
const [field] = record.get(/^(?:260)|(?:264)$/u);
|
|
324
|
+
//debugData(field);
|
|
325
|
+
|
|
326
|
+
if (field) {
|
|
327
|
+
const publisherString = field.subfields
|
|
328
|
+
.filter(({code}) => ['b'].includes(code))
|
|
329
|
+
.map(({value}) => testStringOrNumber(value) ? String(value) : '')
|
|
330
|
+
.filter(value => value)
|
|
331
|
+
// In Melinda's index subfield separators are indexed as ' '
|
|
332
|
+
.join(' ');
|
|
333
|
+
return publisherString;
|
|
334
|
+
}
|
|
335
|
+
return false;
|
|
336
|
+
}
|
|
337
|
+
}
|
|
338
|
+
|
|
285
339
|
|
|
286
340
|
export function bibStandardIdentifiers(record) {
|
|
287
341
|
|
|
@@ -289,6 +343,8 @@ export function bibStandardIdentifiers(record) {
|
|
|
289
343
|
const debugData = debug.extend('data');
|
|
290
344
|
debug(`Creating queries for standard identifiers`);
|
|
291
345
|
|
|
346
|
+
// DEVELOP: should we query also f015 and f028?
|
|
347
|
+
|
|
292
348
|
const fields = record.get(/^(?<def>020|022|024)$/u);
|
|
293
349
|
const identifiers = [].concat(...fields.map(toIdentifiers));
|
|
294
350
|
const uniqueIdentifiers = [...new Set(identifiers)];
|
|
@@ -41,3 +41,4 @@ export {default as language} from './language';
|
|
|
41
41
|
export {default as bibliographicLevel} from './bibliographic-level';
|
|
42
42
|
export {default as melindaId} from './melinda-id';
|
|
43
43
|
export {default as allSourceIds} from './all-source-ids';
|
|
44
|
+
export {default as mediaType} from './media-type';
|
|
@@ -31,6 +31,11 @@ import {READERS} from '@natlibfi/fixura';
|
|
|
31
31
|
import {expect} from 'chai';
|
|
32
32
|
import {MarcRecord} from '@natlibfi/marc-record';
|
|
33
33
|
import * as features from '.';
|
|
34
|
+
import createDebugLogger from 'debug';
|
|
35
|
+
|
|
36
|
+
|
|
37
|
+
const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib:test');
|
|
38
|
+
const debugData = debug.extend('data');
|
|
34
39
|
|
|
35
40
|
|
|
36
41
|
describe('match-detection/features/bib/', () => {
|
|
@@ -47,9 +52,12 @@ describe('match-detection/features/bib/', () => {
|
|
|
47
52
|
return;
|
|
48
53
|
}
|
|
49
54
|
|
|
55
|
+
debug(`Testing: ${feature} ${type}`);
|
|
56
|
+
|
|
50
57
|
if (type === 'extract') {
|
|
51
58
|
const {expectedFeatures, inputRecord} = expectations;
|
|
52
59
|
const record = new MarcRecord(inputRecord, {subfieldValues: false});
|
|
60
|
+
debugData(`Record: ${record}`);
|
|
53
61
|
const {extract} = features[feature](options);
|
|
54
62
|
|
|
55
63
|
expect(extract({record})).to.eql(expectedFeatures);
|
|
@@ -0,0 +1,77 @@
|
|
|
1
|
+
/* eslint-disable max-statements */
|
|
2
|
+
/**
|
|
3
|
+
*
|
|
4
|
+
* @licstart The following is the entire license notice for the JavaScript code in this file.
|
|
5
|
+
*
|
|
6
|
+
* Melinda record matching modules for Javascript
|
|
7
|
+
*
|
|
8
|
+
* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
|
|
9
|
+
*
|
|
10
|
+
* This file is part of melinda-record-matching-js
|
|
11
|
+
*
|
|
12
|
+
* melinda-record-matching-js program is free software: you can redistribute it and/or modify
|
|
13
|
+
* it under the terms of the GNU Lesser General Public License as
|
|
14
|
+
* published by the Free Software Foundation, either version 3 of the
|
|
15
|
+
* License, or (at your option) any later version.
|
|
16
|
+
*
|
|
17
|
+
* melinda-record-matching-js is distributed in the hope that it will be useful,
|
|
18
|
+
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
19
|
+
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
20
|
+
* GNU Lesser General Public License for more details.
|
|
21
|
+
*
|
|
22
|
+
* You should have received a copy of the GNU Affero General Public License
|
|
23
|
+
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
24
|
+
*
|
|
25
|
+
* @licend The above is the entire license notice
|
|
26
|
+
* for the JavaScript code in this file.
|
|
27
|
+
*
|
|
28
|
+
*/
|
|
29
|
+
|
|
30
|
+
import createDebugLogger from 'debug';
|
|
31
|
+
|
|
32
|
+
const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:media-type');
|
|
33
|
+
const debugData = debug.extend('data');
|
|
34
|
+
|
|
35
|
+
export default () => ({
|
|
36
|
+
name: 'Media type',
|
|
37
|
+
extract: ({record, recordExternal}) => {
|
|
38
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
39
|
+
debugData(`Record (${label}): ${JSON.stringify(record)}`);
|
|
40
|
+
debugData(`RecordExternal: ${JSON.stringify(recordExternal)}`);
|
|
41
|
+
const values337 = get337Values();
|
|
42
|
+
debug(`${label} 337 $b values: ${JSON.stringify(values337)}`);
|
|
43
|
+
|
|
44
|
+
return values337;
|
|
45
|
+
|
|
46
|
+
function get337Values() {
|
|
47
|
+
return record.get(/^337$/u)
|
|
48
|
+
.filter(f => f.subfields.some((subfield) => subfield.code === '2' && subfield.value === 'rdamedia'))
|
|
49
|
+
.map(({subfields}) => subfields)
|
|
50
|
+
.flat()
|
|
51
|
+
.filter(({code}) => code === 'b')
|
|
52
|
+
.map(({value}) => value);
|
|
53
|
+
}
|
|
54
|
+
},
|
|
55
|
+
compare: (a, b) => {
|
|
56
|
+
debugData(`Comparing ${JSON.stringify(a)} and ${JSON.stringify(b)}`);
|
|
57
|
+
|
|
58
|
+
// Should we give extra good points if all mediaTypes match?
|
|
59
|
+
// Should we give partial points for partially matching mediaTypes?
|
|
60
|
+
// Should we check whether recordType is 'mixedMaterials'
|
|
61
|
+
// Should we okay typical cases of not totally matching mediaTypes? What would these be?
|
|
62
|
+
|
|
63
|
+
if (a.every(elem => b.includes(elem))) {
|
|
64
|
+
debug(`All mediaTypes from A are in B`);
|
|
65
|
+
return 1;
|
|
66
|
+
}
|
|
67
|
+
|
|
68
|
+
if (b.every(elem => a.includes(elem))) {
|
|
69
|
+
debug(`All mediaTypes from B are in A`);
|
|
70
|
+
return 1;
|
|
71
|
+
}
|
|
72
|
+
|
|
73
|
+
debug(`Mismatch in mediaTypes between A and B`);
|
|
74
|
+
return -1;
|
|
75
|
+
|
|
76
|
+
}
|
|
77
|
+
});
|
|
@@ -28,6 +28,8 @@
|
|
|
28
28
|
|
|
29
29
|
import createInterface from './standard-identifier-factory';
|
|
30
30
|
|
|
31
|
+
// DEVELOP: we should compare indicators and sf2 for f024
|
|
32
|
+
|
|
31
33
|
export default () => {
|
|
32
34
|
const {extract, compare} = createInterface({pattern: /^024$/u, subfieldCodes: ['a', 'z']});
|
|
33
35
|
return {extract, compare, name: 'Other standard identifier'};
|