@natlibfi/melinda-record-matching 5.1.4 → 5.1.5-alpha.1

package/dist/candidate-search/index.js

@@ -3,6 +3,7 @@ import createClient, { SruSearchError } from "@natlibfi/sru-client";
 import { MarcRecord } from "@natlibfi/marc-record";
 import generateQueryList from "./query-list/index.js";
 import chooseQueries from "./choose-queries.js";
+import { preferenceSortRecords } from "../matching-utils.js";
 export { searchTypes } from "./query-list/index.js";
 export class CandidateSearchError extends Error {
 }
@@ -66,6 +67,7 @@ export default async ({ record, searchSpec, url, maxCandidates, maxRecordsPerReq
       }
       debug(`Query ${queryOffset} ${query} done.`);
       debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);
+      records.sort(preferenceSortRecords);
       return { records, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries };
     }
     debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);

package/dist/candidate-search/index.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../src/candidate-search/index.js"],
-  "sourcesContent": ["import createDebugLogger from 'debug';\nimport createClient, {SruSearchError} from '@natlibfi/sru-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\nimport generateQueryList from './query-list/index.js';\nimport chooseQueries from './choose-queries.js';\n\nexport {searchTypes} from './query-list/index.js';\n\nexport class CandidateSearchError extends Error { }\n\n// serverMaxResults : maximum size of total search result available from the server, defaults to Aleph's 20000\n\nexport default async ({record, searchSpec, url, maxCandidates, maxRecordsPerRequest = 50, serverMaxResult = 20000}) => {\n  MarcRecord.setValidationOptions({subfieldValues: false});\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search');\n  const debugData = debug.extend('data');\n\n  debugData(`SearchSpec: ${JSON.stringify(searchSpec)}`);\n  debugData(`Url: ${url}`);\n  debugData(`MaxRecordsPerRequest ${maxRecordsPerRequest}`);\n  debugData(`ServerMaxResult: ${serverMaxResult}`);\n  debugData(`MaxCandidates: ${maxCandidates}`);\n\n  // Do not retrieve more candidates than defined in maxCandidates\n  const adjustedMaxRecordsPerRequest = maxRecordsPerRequest >= maxCandidates ? maxCandidates : maxRecordsPerRequest;\n\n  const client = createClient({\n    url,\n    maxRecordsPerRequest: adjustedMaxRecordsPerRequest,\n    version: '2.0',\n    retrieveAll: false,\n    metadataFormat: 'marcJson'\n  });\n\n  const inputRecordId = getRecordId(record);\n  const queryListResult = await generateQueryList(record, searchSpec, client);\n  const queryList = queryListResult[0]?.queryList ? queryListResult[0].queryList : queryListResult;\n  const queryListType = queryListResult[0]?.queryListType ? queryListResult[0].queryListType : undefined;\n\n  // if generateQueryList errored we should throw 422\n  if (queryList.length === 0) {\n    debug(`Empty list`);\n    throw new CandidateSearchError(`Generated query list contains no queries`);\n  }\n  if (queryListType && queryListType !== 'alternates') {\n    debug(`Unknown queryListType`);\n    throw new CandidateSearchError(`Generated query list has invalid type`);\n  }\n\n  debug(`Searching matches for ${inputRecordId}`);\n  const chosenQueryList = await filterQueryList({queryList, queryListType});\n  debug(`Chosen queries: ${JSON.stringify(chosenQueryList)}`);\n\n  async function filterQueryList({queryList, queryListType, maxCandidates}) {\n    debug(`Generated queryList (type: ${queryListType}) ${JSON.stringify(queryList)}`);\n\n    if (queryListType === 'alternates' && queryList.length > 1) {\n      const queryListResult = await chooseQueries({url, queryList, queryListType, maxCandidates});\n      debug(`queryListResult: ${JSON.stringify(queryListResult)}`);\n      return queryListResult.map(elem => elem.query);\n    }\n    return queryList;\n  }\n  // state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)\n  // state.query : current query (undefined if there was no queries left)\n  // state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)\n  // state.queryCandidateCounter: amount of candidates (records+failures) retrieved from SRU for matching for current query, including the current record+failure set (undefined if there were no queries left)\n  // state.queriesLeft : amount of queries left\n  // state.queryCounter : sequence for current query\n  // state.maxedQueries : queries that resulted in more than serverMaxResults hits\n\n  return {search};\n\n  async function search({queryOffset = 0, resultSetOffset = 1, totalRecords = 0, searchCounter = 0, queryCandidateCounter = 0, queryCounter = 0, maxedQueries = []}) {\n    const query = chosenQueryList[queryOffset];\n    debug(`Running query ${JSON.stringify(query)} (${queryOffset})`);\n\n    if (query) {\n      const {records, nextOffset, total} = await retrieveRecords(client, query, resultSetOffset);\n\n      // If resultSetOffset === 1 this is the first search for the current query\n      debugData(`ResultSetOffset: ${resultSetOffset}`);\n      const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;\n      const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;\n      const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;\n      const newQueryCandidateCounter = resultSetOffset === 1 ? records.length : queryCandidateCounter + records.length;\n\n      const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;\n      const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;\n\n      if (typeof nextOffset === 'number') {\n        debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);\n        return {records, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n      }\n      debug(`Query ${queryOffset} ${query} done.`);\n      debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);\n      return {records, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n    }\n\n    debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);\n    return {records: [], queriesLeft: 0, queryCounter, maxedQueries};\n  }\n\n  function checkMaxedQuery(query, total, serverMaxResult) {\n    if (total >= serverMaxResult) {\n      debug(`WARNING: Query ${query} resulted in ${total} hits which meets the serverMaxResult (${serverMaxResult}) `);\n      return query;\n    }\n  }\n\n  function getRecordId(record) {\n    const [f001] = record.get(/^001$/u);\n    const [f003] = record.get(/^003$/u);\n    if (f001 && f003) {\n      return `${f003.value}-${f001.value}`;\n    }\n    return f001 ? f001.value : '';\n  }\n\n};\n\nexport function retrieveRecords(client, query, resultSetOffset) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:retrieveRecords');\n  const debugData = debug.extend('data');\n\n  return new Promise((resolve, reject) => {\n    const records = [];\n    let totalRecords = 0;\n\n    debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);\n\n    client.searchRetrieve(query, {startRecord: resultSetOffset})\n      .on('error', err => {\n        if (err instanceof SruSearchError) {\n          debug(`SRU SruSearchError for query: ${query}: ${err}`);\n          reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));\n        }\n\n        debug(`SRU error for query: ${query}: ${err}`);\n        reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));\n      })\n      .on('total', total => {\n        debug(`Got total: ${total}`);\n        totalRecords += total;\n      })\n      .on('end', nextOffset => {\n        try {\n          debug(`Found ${records.length} records`);\n\n          resolve({nextOffset, records, total: totalRecords});\n        } catch (err) {\n          debug(`Error caught on END`);\n          reject(err);\n        }\n      })\n      .on('record', record => {\n        // MarcRecord Error handling\n        try {\n          const marcRecord = new MarcRecord(record);\n          const [field] = marcRecord.get(/^001$/u);\n          debug(field);\n          const id = field.value ? field.value : '';\n          records.push({record: marcRecord, id});\n        } catch (error) {\n          debug(`Sru => record error: ${error}`);\n          debug(`Sru => record: ${record}`);\n        }\n      });\n  });\n}\n"],
-  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,OAAO,gBAAe,sBAAqB;AAC3C,SAAQ,kBAAiB;AAEzB,OAAO,uBAAuB;AAC9B,OAAO,mBAAmB;AAE1B,SAAQ,mBAAkB;AAEnB,aAAM,6BAA6B,MAAM;AAAE;AAIlD,eAAe,OAAO,EAAC,QAAQ,YAAY,KAAK,eAAe,uBAAuB,IAAI,kBAAkB,IAAK,MAAM;AACrH,aAAW,qBAAqB,EAAC,gBAAgB,MAAK,CAAC;AAEvD,QAAM,QAAQ,kBAAkB,oDAAoD;AACpF,QAAM,YAAY,MAAM,OAAO,MAAM;AAErC,YAAU,eAAe,KAAK,UAAU,UAAU,CAAC,EAAE;AACrD,YAAU,QAAQ,GAAG,EAAE;AACvB,YAAU,wBAAwB,oBAAoB,EAAE;AACxD,YAAU,oBAAoB,eAAe,EAAE;AAC/C,YAAU,kBAAkB,aAAa,EAAE;AAG3C,QAAM,+BAA+B,wBAAwB,gBAAgB,gBAAgB;AAE7F,QAAM,SAAS,aAAa;AAAA,IAC1B;AAAA,IACA,sBAAsB;AAAA,IACtB,SAAS;AAAA,IACT,aAAa;AAAA,IACb,gBAAgB;AAAA,EAClB,CAAC;AAED,QAAM,gBAAgB,YAAY,MAAM;AACxC,QAAM,kBAAkB,MAAM,kBAAkB,QAAQ,YAAY,MAAM;AAC1E,QAAM,YAAY,gBAAgB,CAAC,GAAG,YAAY,gBAAgB,CAAC,EAAE,YAAY;AACjF,QAAM,gBAAgB,gBAAgB,CAAC,GAAG,gBAAgB,gBAAgB,CAAC,EAAE,gBAAgB;AAG7F,MAAI,UAAU,WAAW,GAAG;AAC1B,UAAM,YAAY;AAClB,UAAM,IAAI,qBAAqB,0CAA0C;AAAA,EAC3E;AACA,MAAI,iBAAiB,kBAAkB,cAAc;AACnD,UAAM,uBAAuB;AAC7B,UAAM,IAAI,qBAAqB,uCAAuC;AAAA,EACxE;AAEA,QAAM,yBAAyB,aAAa,EAAE;AAC9C,QAAM,kBAAkB,MAAM,gBAAgB,EAAC,WAAW,cAAa,CAAC;AACxE,QAAM,mBAAmB,KAAK,UAAU,eAAe,CAAC,EAAE;AAE1D,iBAAe,gBAAgB,EAAC,WAAAA,YAAW,eAAAC,gBAAe,eAAAC,eAAa,GAAG;AACxE,UAAM,8BAA8BD,cAAa,KAAK,KAAK,UAAUD,UAAS,CAAC,EAAE;AAEjF,QAAIC,mBAAkB,gBAAgBD,WAAU,SAAS,GAAG;AAC1D,YAAMG,mBAAkB,MAAM,cAAc,EAAC,KAAK,WAAAH,YAAW,eAAAC,gBAAe,eAAAC,eAAa,CAAC;AAC1F,YAAM,oBAAoB,KAAK,UAAUC,gBAAe,CAAC,EAAE;AAC3D,aAAOA,iBAAgB,IAAI,UAAQ,KAAK,KAAK;AAAA,IAC/C;AACA,WAAOH;AAAA,EACT;AASA,SAAO,EAAC,OAAM;AAEd,iBAAe,OAAO,EAAC,cAAc,GAAG,kBAAkB,GAAG,eAAe,GAAG,gBAAgB,GAAG,wBAAwB,GAAG,eAAe,GAAG,eAAe,CAAC,EAAC,GAAG;AACjK,UAAM,QAAQ,gBAAgB,WAAW;AACzC,UAAM,iBAAiB,KAAK,UAAU,KAAK,CAAC,KAAK,WAAW,GAAG;AAE/D,QAAI,OAAO;AACT,YAAM,EAAC,SAAS,YAAY,MAAK,IAAI,MAAM,gBAAgB,QAAQ,OAAO,eAAe;AAGzF,gBAAU,oBAAoB,eAAe,EAAE;AAC/C,YAAM,kBAAkB,oBAAoB,IAAI,QAAQ;AACxD,YAAM,kBAAkB,oBAAoB,IAAI,eAAe,IAAI;AACnE,YAAM,mBAAmB,oBAAoB,IAAI,IAAI,gBAAgB;AACrE,YAAM,2BAA2B,oBAAoB,IAAI,QAAQ,SAAS,wBAAwB,QAAQ;AAE1G,YAAM,aAAa,oBAAoB,IAAI,gBAAgB,OAAO,OAAO,eAAe,IAAI;AAC5F,YAAM,kBAAkB,aAAa,aAAa,OAAO,UAAU,IAAI;AAEvE,UAAI,OAAO,eAAe,UAAU;AAClC,cAAM,iCAAiC,WAAW,IAAI,KAAK,0BAA0B,UAAU,EAAE;AACjG,eAAO,EAAC,SAAS,aAAa,iBAAiB,YAAY,aAAa,UAAU,UAAU,cAAc,IAAI,cAAc,iBAAiB,OAAO,eAAe,kBAAkB,uBAAuB,0BAA0B,cAAc,iBAAiB,cAAc,gBAAe;AAAA,MACpS;AACA,YAAM,SAAS,WAAW,IAAI,KAAK,QAAQ;AAC3C,YAAM,cAAc,UAAU,UAAU,cAAc,EAAE,gBAAgB;AACxE,aAAO,EAAC,SAAS,aAAa,cAAc,GAAG,aAAa,UAAU,UAAU,cAAc,IAAI,cAAc,iBAAiB,OAAO,eAAe,kBAAkB,uBAAuB,0BAA0B,cAAc,iBAAiB,cAAc,gBAAe;AAAA,IACxR;AAEA,UAAM,OAAO,UAAU,MAAM,uCAAuC,WAAW,EAAE;AACjF,WAAO,EAAC,SAAS,CAAC,GAAG,aAAa,GAAG,cAAc,aAAY;AAAA,EACjE;AAEA,WAAS,gBAAgB,OAAO,OAAOI,kBAAiB;AACtD,QAAI,SAASA,kBAAiB;AAC5B,YAAM,kBAAkB,KAAK,gBAAgB,KAAK,0CAA0CA,gBAAe,IAAI;AAC/G,aAAO;AAAA,IACT;AAAA,EACF;AAEA,WAAS,YAAYC,SAAQ;AAC3B,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,QAAI,QAAQ,MAAM;AAChB,aAAO,GAAG,KAAK,KAAK,IAAI,KAAK,KAAK;AAAA,IACpC;AACA,WAAO,OAAO,KAAK,QAAQ;AAAA,EAC7B;AAEF;AAEO,gBAAS,gBAAgB,QAAQ,OAAO,iBAAiB;AAC9D,QAAM,QAAQ,kBAAkB,oEAAoE;AACpG,QAAM,YAAY,MAAM,OAAO,MAAM;AAErC,SAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AACtC,UAAM,UAAU,CAAC;AACjB,QAAI,eAAe;AAEnB,UAAM,wCAAwC,KAAK,YAAY,eAAe,GAAG;AAEjF,WAAO,eAAe,OAAO,EAAC,aAAa,gBAAe,CAAC,EACxD,GAAG,SAAS,SAAO;AAClB,UAAI,eAAe,gBAAgB;AACjC,cAAM,iCAAiC,KAAK,KAAK,GAAG,EAAE;AACtD,eAAO,IAAI,qBAAqB,iCAAiC,KAAK,KAAK,GAAG,EAAE,CAAC;AAAA,MACnF;AAEA,YAAM,wBAAwB,KAAK,KAAK,GAAG,EAAE;AAC7C,aAAO,IAAI,qBAAqB,wBAAwB,KAAK,KAAK,GAAG,EAAE,CAAC;AAAA,IAC1E,CAAC,EACA,GAAG,SAAS,WAAS;AACpB,YAAM,cAAc,KAAK,EAAE;AAC3B,sBAAgB;AAAA,IAClB,CAAC,EACA,GAAG,OAAO,gBAAc;AACvB,UAAI;AACF,cAAM,SAAS,QAAQ,MAAM,UAAU;AAEvC,gBAAQ,EAAC,YAAY,SAAS,OAAO,aAAY,CAAC;AAAA,MACpD,SAAS,KAAK;AACZ,cAAM,qBAAqB;AAC3B,eAAO,GAAG;AAAA,MACZ;AAAA,IACF,CAAC,EACA,GAAG,UAAU,YAAU;AAEtB,UAAI;AACF,cAAM,aAAa,IAAI,WAAW,MAAM;AACxC,cAAM,CAAC,KAAK,IAAI,WAAW,IAAI,QAAQ;AACvC,cAAM,KAAK;AACX,cAAM,KAAK,MAAM,QAAQ,MAAM,QAAQ;AACvC,gBAAQ,KAAK,EAAC,QAAQ,YAAY,GAAE,CAAC;AAAA,MACvC,SAAS,OAAO;AACd,cAAM,wBAAwB,KAAK,EAAE;AACrC,cAAM,kBAAkB,MAAM,EAAE;AAAA,MAClC;AAAA,IACF,CAAC;AAAA,EACL,CAAC;AACH;",
+  "sourcesContent": ["import createDebugLogger from 'debug';\nimport createClient, {SruSearchError} from '@natlibfi/sru-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\nimport generateQueryList from './query-list/index.js';\nimport chooseQueries from './choose-queries.js';\nimport {preferenceSortRecords} from '../matching-utils.js';\n\nexport {searchTypes} from './query-list/index.js';\n\nexport class CandidateSearchError extends Error { }\n\n// serverMaxResults : maximum size of total search result available from the server, defaults to Aleph's 20000\n\nexport default async ({record, searchSpec, url, maxCandidates, maxRecordsPerRequest = 50, serverMaxResult = 20000}) => {\n  MarcRecord.setValidationOptions({subfieldValues: false});\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search');\n  const debugData = debug.extend('data');\n\n  debugData(`SearchSpec: ${JSON.stringify(searchSpec)}`);\n  debugData(`Url: ${url}`);\n  debugData(`MaxRecordsPerRequest ${maxRecordsPerRequest}`);\n  debugData(`ServerMaxResult: ${serverMaxResult}`);\n  debugData(`MaxCandidates: ${maxCandidates}`);\n\n  // Do not retrieve more candidates than defined in maxCandidates\n  const adjustedMaxRecordsPerRequest = maxRecordsPerRequest >= maxCandidates ? maxCandidates : maxRecordsPerRequest;\n\n  const client = createClient({\n    url,\n    maxRecordsPerRequest: adjustedMaxRecordsPerRequest,\n    version: '2.0',\n    retrieveAll: false,\n    metadataFormat: 'marcJson'\n  });\n\n  const inputRecordId = getRecordId(record);\n  const queryListResult = await generateQueryList(record, searchSpec, client);\n  const queryList = queryListResult[0]?.queryList ? queryListResult[0].queryList : queryListResult;\n  const queryListType = queryListResult[0]?.queryListType ? queryListResult[0].queryListType : undefined;\n\n  // if generateQueryList errored we should throw 422\n  if (queryList.length === 0) {\n    debug(`Empty list`);\n    throw new CandidateSearchError(`Generated query list contains no queries`);\n  }\n  if (queryListType && queryListType !== 'alternates') {\n    debug(`Unknown queryListType`);\n    throw new CandidateSearchError(`Generated query list has invalid type`);\n  }\n\n  debug(`Searching matches for ${inputRecordId}`);\n  const chosenQueryList = await filterQueryList({queryList, queryListType});\n  debug(`Chosen queries: ${JSON.stringify(chosenQueryList)}`);\n\n  async function filterQueryList({queryList, queryListType, maxCandidates}) {\n    debug(`Generated queryList (type: ${queryListType}) ${JSON.stringify(queryList)}`);\n\n    if (queryListType === 'alternates' && queryList.length > 1) {\n      const queryListResult = await chooseQueries({url, queryList, queryListType, maxCandidates});\n      debug(`queryListResult: ${JSON.stringify(queryListResult)}`);\n      return queryListResult.map(elem => elem.query);\n    }\n    return queryList;\n  }\n  // state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)\n  // state.query : current query (undefined if there was no queries left)\n  // state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)\n  // state.queryCandidateCounter: amount of candidates (records+failures) retrieved from SRU for matching for current query, including the current record+failure set (undefined if there were no queries left)\n  // state.queriesLeft : amount of queries left\n  // state.queryCounter : sequence for current query\n  // state.maxedQueries : queries that resulted in more than serverMaxResults hits\n\n  return {search};\n\n  async function search({queryOffset = 0, resultSetOffset = 1, totalRecords = 0, searchCounter = 0, queryCandidateCounter = 0, queryCounter = 0, maxedQueries = []}) {\n    const query = chosenQueryList[queryOffset];\n    debug(`Running query ${JSON.stringify(query)} (${queryOffset})`);\n\n    if (query) {\n      const {records, nextOffset, total} = await retrieveRecords(client, query, resultSetOffset);\n\n      // If resultSetOffset === 1 this is the first search for the current query\n      debugData(`ResultSetOffset: ${resultSetOffset}`);\n      const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;\n      const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;\n      const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;\n      const newQueryCandidateCounter = resultSetOffset === 1 ? records.length : queryCandidateCounter + records.length;\n\n      const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;\n      const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;\n\n      if (typeof nextOffset === 'number') {\n        debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);\n        return {records, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n      }\n      debug(`Query ${queryOffset} ${query} done.`);\n      debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);\n\n      records.sort(preferenceSortRecords);\n      return {records, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n    }\n\n    debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);\n    return {records: [], queriesLeft: 0, queryCounter, maxedQueries};\n  }\n\n  function checkMaxedQuery(query, total, serverMaxResult) {\n    if (total >= serverMaxResult) {\n      debug(`WARNING: Query ${query} resulted in ${total} hits which meets the serverMaxResult (${serverMaxResult}) `);\n      return query;\n    }\n  }\n\n  function getRecordId(record) {\n    const [f001] = record.get(/^001$/u);\n    const [f003] = record.get(/^003$/u);\n    if (f001 && f003) {\n      return `${f003.value}-${f001.value}`;\n    }\n    return f001 ? f001.value : '';\n  }\n\n};\n\nexport function retrieveRecords(client, query, resultSetOffset) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:retrieveRecords');\n  const debugData = debug.extend('data');\n\n  return new Promise((resolve, reject) => {\n    const records = [];\n    let totalRecords = 0;\n\n    debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);\n\n    client.searchRetrieve(query, {startRecord: resultSetOffset})\n      .on('error', err => {\n        if (err instanceof SruSearchError) {\n          debug(`SRU SruSearchError for query: ${query}: ${err}`);\n          reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));\n        }\n\n        debug(`SRU error for query: ${query}: ${err}`);\n        reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));\n      })\n      .on('total', total => {\n        debug(`Got total: ${total}`);\n        totalRecords += total;\n      })\n      .on('end', nextOffset => {\n        try {\n          debug(`Found ${records.length} records`);\n\n          resolve({nextOffset, records, total: totalRecords});\n        } catch (err) {\n          debug(`Error caught on END`);\n          reject(err);\n        }\n      })\n      .on('record', record => {\n        // MarcRecord Error handling\n        try {\n          const marcRecord = new MarcRecord(record);\n          const [field] = marcRecord.get(/^001$/u);\n          debug(field);\n          const id = field.value ? field.value : '';\n          records.push({record: marcRecord, id});\n        } catch (error) {\n          debug(`Sru => record error: ${error}`);\n          debug(`Sru => record: ${record}`);\n        }\n      });\n  });\n}\n"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,OAAO,gBAAe,sBAAqB;AAC3C,SAAQ,kBAAiB;AAEzB,OAAO,uBAAuB;AAC9B,OAAO,mBAAmB;AAC1B,SAAQ,6BAA4B;AAEpC,SAAQ,mBAAkB;AAEnB,aAAM,6BAA6B,MAAM;AAAE;AAIlD,eAAe,OAAO,EAAC,QAAQ,YAAY,KAAK,eAAe,uBAAuB,IAAI,kBAAkB,IAAK,MAAM;AACrH,aAAW,qBAAqB,EAAC,gBAAgB,MAAK,CAAC;AAEvD,QAAM,QAAQ,kBAAkB,oDAAoD;AACpF,QAAM,YAAY,MAAM,OAAO,MAAM;AAErC,YAAU,eAAe,KAAK,UAAU,UAAU,CAAC,EAAE;AACrD,YAAU,QAAQ,GAAG,EAAE;AACvB,YAAU,wBAAwB,oBAAoB,EAAE;AACxD,YAAU,oBAAoB,eAAe,EAAE;AAC/C,YAAU,kBAAkB,aAAa,EAAE;AAG3C,QAAM,+BAA+B,wBAAwB,gBAAgB,gBAAgB;AAE7F,QAAM,SAAS,aAAa;AAAA,IAC1B;AAAA,IACA,sBAAsB;AAAA,IACtB,SAAS;AAAA,IACT,aAAa;AAAA,IACb,gBAAgB;AAAA,EAClB,CAAC;AAED,QAAM,gBAAgB,YAAY,MAAM;AACxC,QAAM,kBAAkB,MAAM,kBAAkB,QAAQ,YAAY,MAAM;AAC1E,QAAM,YAAY,gBAAgB,CAAC,GAAG,YAAY,gBAAgB,CAAC,EAAE,YAAY;AACjF,QAAM,gBAAgB,gBAAgB,CAAC,GAAG,gBAAgB,gBAAgB,CAAC,EAAE,gBAAgB;AAG7F,MAAI,UAAU,WAAW,GAAG;AAC1B,UAAM,YAAY;AAClB,UAAM,IAAI,qBAAqB,0CAA0C;AAAA,EAC3E;AACA,MAAI,iBAAiB,kBAAkB,cAAc;AACnD,UAAM,uBAAuB;AAC7B,UAAM,IAAI,qBAAqB,uCAAuC;AAAA,EACxE;AAEA,QAAM,yBAAyB,aAAa,EAAE;AAC9C,QAAM,kBAAkB,MAAM,gBAAgB,EAAC,WAAW,cAAa,CAAC;AACxE,QAAM,mBAAmB,KAAK,UAAU,eAAe,CAAC,EAAE;AAE1D,iBAAe,gBAAgB,EAAC,WAAAA,YAAW,eAAAC,gBAAe,eAAAC,eAAa,GAAG;AACxE,UAAM,8BAA8BD,cAAa,KAAK,KAAK,UAAUD,UAAS,CAAC,EAAE;AAEjF,QAAIC,mBAAkB,gBAAgBD,WAAU,SAAS,GAAG;AAC1D,YAAMG,mBAAkB,MAAM,cAAc,EAAC,KAAK,WAAAH,YAAW,eAAAC,gBAAe,eAAAC,eAAa,CAAC;AAC1F,YAAM,oBAAoB,KAAK,UAAUC,gBAAe,CAAC,EAAE;AAC3D,aAAOA,iBAAgB,IAAI,UAAQ,KAAK,KAAK;AAAA,IAC/C;AACA,WAAOH;AAAA,EACT;AASA,SAAO,EAAC,OAAM;AAEd,iBAAe,OAAO,EAAC,cAAc,GAAG,kBAAkB,GAAG,eAAe,GAAG,gBAAgB,GAAG,wBAAwB,GAAG,eAAe,GAAG,eAAe,CAAC,EAAC,GAAG;AACjK,UAAM,QAAQ,gBAAgB,WAAW;AACzC,UAAM,iBAAiB,KAAK,UAAU,KAAK,CAAC,KAAK,WAAW,GAAG;AAE/D,QAAI,OAAO;AACT,YAAM,EAAC,SAAS,YAAY,MAAK,IAAI,MAAM,gBAAgB,QAAQ,OAAO,eAAe;AAGzF,gBAAU,oBAAoB,eAAe,EAAE;AAC/C,YAAM,kBAAkB,oBAAoB,IAAI,QAAQ;AACxD,YAAM,kBAAkB,oBAAoB,IAAI,eAAe,IAAI;AACnE,YAAM,mBAAmB,oBAAoB,IAAI,IAAI,gBAAgB;AACrE,YAAM,2BAA2B,oBAAoB,IAAI,QAAQ,SAAS,wBAAwB,QAAQ;AAE1G,YAAM,aAAa,oBAAoB,IAAI,gBAAgB,OAAO,OAAO,eAAe,IAAI;AAC5F,YAAM,kBAAkB,aAAa,aAAa,OAAO,UAAU,IAAI;AAEvE,UAAI,OAAO,eAAe,UAAU;AAClC,cAAM,iCAAiC,WAAW,IAAI,KAAK,0BAA0B,UAAU,EAAE;AACjG,eAAO,EAAC,SAAS,aAAa,iBAAiB,YAAY,aAAa,UAAU,UAAU,cAAc,IAAI,cAAc,iBAAiB,OAAO,eAAe,kBAAkB,uBAAuB,0BAA0B,cAAc,iBAAiB,cAAc,gBAAe;AAAA,MACpS;AACA,YAAM,SAAS,WAAW,IAAI,KAAK,QAAQ;AAC3C,YAAM,cAAc,UAAU,UAAU,cAAc,EAAE,gBAAgB;AAExE,cAAQ,KAAK,qBAAqB;AAClC,aAAO,EAAC,SAAS,aAAa,cAAc,GAAG,aAAa,UAAU,UAAU,cAAc,IAAI,cAAc,iBAAiB,OAAO,eAAe,kBAAkB,uBAAuB,0BAA0B,cAAc,iBAAiB,cAAc,gBAAe;AAAA,IACxR;AAEA,UAAM,OAAO,UAAU,MAAM,uCAAuC,WAAW,EAAE;AACjF,WAAO,EAAC,SAAS,CAAC,GAAG,aAAa,GAAG,cAAc,aAAY;AAAA,EACjE;AAEA,WAAS,gBAAgB,OAAO,OAAOI,kBAAiB;AACtD,QAAI,SAASA,kBAAiB;AAC5B,YAAM,kBAAkB,KAAK,gBAAgB,KAAK,0CAA0CA,gBAAe,IAAI;AAC/G,aAAO;AAAA,IACT;AAAA,EACF;AAEA,WAAS,YAAYC,SAAQ;AAC3B,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,QAAI,QAAQ,MAAM;AAChB,aAAO,GAAG,KAAK,KAAK,IAAI,KAAK,KAAK;AAAA,IACpC;AACA,WAAO,OAAO,KAAK,QAAQ;AAAA,EAC7B;AAEF;AAEO,gBAAS,gBAAgB,QAAQ,OAAO,iBAAiB;AAC9D,QAAM,QAAQ,kBAAkB,oEAAoE;AACpG,QAAM,YAAY,MAAM,OAAO,MAAM;AAErC,SAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AACtC,UAAM,UAAU,CAAC;AACjB,QAAI,eAAe;AAEnB,UAAM,wCAAwC,KAAK,YAAY,eAAe,GAAG;AAEjF,WAAO,eAAe,OAAO,EAAC,aAAa,gBAAe,CAAC,EACxD,GAAG,SAAS,SAAO;AAClB,UAAI,eAAe,gBAAgB;AACjC,cAAM,iCAAiC,KAAK,KAAK,GAAG,EAAE;AACtD,eAAO,IAAI,qBAAqB,iCAAiC,KAAK,KAAK,GAAG,EAAE,CAAC;AAAA,MACnF;AAEA,YAAM,wBAAwB,KAAK,KAAK,GAAG,EAAE;AAC7C,aAAO,IAAI,qBAAqB,wBAAwB,KAAK,KAAK,GAAG,EAAE,CAAC;AAAA,IAC1E,CAAC,EACA,GAAG,SAAS,WAAS;AACpB,YAAM,cAAc,KAAK,EAAE;AAC3B,sBAAgB;AAAA,IAClB,CAAC,EACA,GAAG,OAAO,gBAAc;AACvB,UAAI;AACF,cAAM,SAAS,QAAQ,MAAM,UAAU;AAEvC,gBAAQ,EAAC,YAAY,SAAS,OAAO,aAAY,CAAC;AAAA,MACpD,SAAS,KAAK;AACZ,cAAM,qBAAqB;AAC3B,eAAO,GAAG;AAAA,MACZ;AAAA,IACF,CAAC,EACA,GAAG,UAAU,YAAU;AAEtB,UAAI;AACF,cAAM,aAAa,IAAI,WAAW,MAAM;AACxC,cAAM,CAAC,KAAK,IAAI,WAAW,IAAI,QAAQ;AACvC,cAAM,KAAK;AACX,cAAM,KAAK,MAAM,QAAQ,MAAM,QAAQ;AACvC,gBAAQ,KAAK,EAAC,QAAQ,YAAY,GAAE,CAAC;AAAA,MACvC,SAAS,OAAO;AACd,cAAM,wBAAwB,KAAK,EAAE;AACrC,cAAM,kBAAkB,MAAM,EAAE;AAAA,MAClC;AAAA,IACF,CAAC;AAAA,EACL,CAAC;AACH;",
   "names": ["queryList", "queryListType", "maxCandidates", "queryListResult", "serverMaxResult", "record"]
 }

package/dist/candidate-search/query-list/bib.js

@@ -259,7 +259,7 @@ export function bibStandardIdentifiers(record) {
   const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers");
   const debugData = debug2.extend("data");
   debug2(`Creating queries for standard identifiers`);
-  const fields = record.get(/^(?<def>020|022|024)$/u);
+  const fields = record.get(/^(?<def>015|020|022|024|028)$/u);
   const identifiers = [].concat(...fields.map(toIdentifiers));
   const uniqueIdentifiers = [...new Set(identifiers)];
   debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);
@@ -273,12 +273,25 @@ export function bibStandardIdentifiers(record) {
   function toIdentifiers({ tag, subfields }) {
     const issnIsbnReqExp = /^[A-Za-z0-9-]+$/u;
     const otherIdReqExp = /^[A-Za-z0-9-:]+$/u;
-    if (tag === "022") {
-      return subfields.filter((sub) => ["a", "z", "y"].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value))).map(({ value }) => String(value));
+    if (tag === "015") {
+      return subfields.filter((sf) => sf.code === "a" && sf.value.match(/^[A-Z0-9\-]+$/ui)).map((sf) => sf.value);
     }
     if (tag === "020") {
       return subfields.filter((sub) => ["a", "z"].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value))).map(({ value }) => String(value));
     }
+    if (tag === "022") {
+      return subfields.filter((sub) => ["a", "z", "y"].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value))).map(({ value }) => String(value));
+    }
+    if (tag === "028") {
+      const a = subfields.find((sf) => sf.code === "a");
+      if (a) {
+        const b = subfields.find((sf) => sf.code === "b");
+        if (b) {
+          return [a.value, `${b.value} ${a.value}`];
+        }
+        return [a.value];
+      }
+    }
     return subfields.filter((sub) => ["a", "z"].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value))).map(({ value }) => String(value));
   }
 }

package/dist/candidate-search/query-list/bib.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../src/candidate-search/query-list/bib.js"],
-  "sourcesContent": ["import createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils.js';\nimport {extractHostMelindaIdsFromField, extractPublicationYearFrom773, getHostMelindaFields} from './component.js';\nimport {fieldToString} from '@natlibfi/marc-record-validators-melinda';\nimport {getTitle} from '../../match-detection/features/bib/title.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query');\n\nconst IS_OK_ALONE_THRESHOLD = 5;\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibHostComponents');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for hostIds`);\n  const f773s = getHostMelindaFields(record);\n  if (f773s.length !== 1) { // This needs some thinking...\n    return [];\n  }\n\n  const [id] = extractHostMelindaIdsFromField(f773s[0]);\n  debug(`Found id: ${JSON.stringify(id)}`);\n\n  // NB! record.isCR() does not work if LDR/07=a, so we get year from 773$g!\n  const year = extractPublicationYearFrom773(f773s[0]);\n  debug(`${fieldToString(f773s[0])} => ${year ? year : 'N/A'}`);\n  if (year) {\n    return [`dc.date=${year} AND melinda.partsofhost=${id}`];\n  }\n\n  return [`melinda.partsofhost=${id}`];\n\n}\n\n\n\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\nexport function bibTitle(record) {\n  // We get author/publisher only when formatted title is shorter than 5 chars\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 5});\n}\n\nexport function bibTitleAuthor(record) {\n  debug('bibTitleAuthor');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 100});\n}\n\nexport function bibTitleAuthorYear(record) {\n  debug('bibTitleAuthorYearAlternates');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true});\n}\n\nexport function bibTitleAuthorYearAlternates(record) {\n  debug('bibTitleAuthorYearAlternates');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n  const origQueryList = bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true, alternates: true, alternateQueries: []});\n  debug(`${JSON.stringify(origQueryList)}`);\n  return {queryList: Array.from(origQueryList).reverse(), queryListType: 'alternates'};\n}\n\nfunction getTitleForQuery(record) {\n  const title = getTitle(record, ['a']);\n  if (title) {\n    const nWords = title.split(' ');\n    // If lone $a is deemed too short, fetch $b as well:\n    // (NV: I've seen pairs with 245$a-only vs 245$a$b, so I'd like to use $a only if it is long enough)\n\n    if (nWords < 3 || title.length < 12) {\n      const [f245] = record.get('245');\n      // If punctuation is ' =' I think that f245$a is good enough despite shortness. Trying to balance between two bad situations...\n      // \"Suo (= short title) siell\u00E4, vetel\u00E4 (= missing $b name in another language) t\u00E4\u00E4ll\u00E4...\"\n      if (f245 && f245.subfields.find(sf => sf.code === 'a' && sf.value.match(/ =$/u)) && title.length >= IS_OK_ALONE_THRESHOLD) {\n        return title;\n      }\n      return getTitle(record, ['a', 'b']); // Try to get a longer title\n    }\n  }\n  return title;\n}\n\nfunction dcTitle(record, onlyTitleLength, alternates = false) {\n  const title = getTitleForQuery(record);\n  if (!testStringOrNumber(title)) {\n    return [];\n  }\n\n  const formatted = getFormattedTitle(title);\n\n  // use word search for titles starting with a boolean\n  const useWordSearch = checkUseWordSearch(formatted);\n  // Prevent too many matches / SRU crashing by having a minimum length\n  // Note that currently this fails matching if there are no matches from previous matchers\n  if (formatted.length >= onlyTitleLength && !alternates) {\n    return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`, formatted, true];\n  }\n\n  const queryIsOkAlone = !useWordSearch && formatted.length >= IS_OK_ALONE_THRESHOLD;\n\n  // use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]\n  return [`dc.title=\"${useWordSearch || !queryIsOkAlone ? '' : '^'}${formatted}${queryIsOkAlone ? '*' : ''}\"`, formatted, queryIsOkAlone];\n\n  function getFormattedTitle(title) {\n    const formatted = String(title)\n      .replace(/[\\-+ !\"(){}\\[\\]<>;:.?/@*%=^_`~]/gu, ' ')\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '') // Apparently matches to/works with non-aplhabetic scripts such as Chinese as well\n      .replace(/  +/gu, ' ') // Clean up concurrent spaces from eg. subfield changes\n      .trim()\n      .replace(/^(.{30}\\S*) .*$/, \"$1\")\n      .trim();\n\n    return formatted;\n  }\n}\n\nexport function bibTitleAuthorPublisher({record, onlyTitleLength, addYear = false, alternates = false, alternateQueries = []}) {\n  debug(`bibTitleAuthorPublisher, onlyTitleLength: ${onlyTitleLength}, addYear: ${addYear}, alternates: ${alternates}`);\n  const [query, formatted, queryIsOkAlone] = dcTitle(record, onlyTitleLength, alternates);\n  if (query === undefined) {\n    return [];\n  }\n\n  debug(`query: ${query}`);\n\n  // Prevent too many matches / SRU crashing by having a minimum length\n  // Note that currently this fails matching if there are no matches from previous matchers\n  if (formatted.length >= onlyTitleLength && !alternates) {\n    return [query];\n  }\n\n  const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;\n\n  return addAuthorsToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries});\n\n  function addAuthorsToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n    debug('addAuthorsToSearch');\n    const [authorQuery] = bibAuthors(record);\n    if (authorQuery !== undefined) {\n      if (addYear) {\n        const newAlternateQueries = alternates ? [...alternateQueries, `${authorQuery} AND ${query}`] : alternateQueries;\n        return addYearToSearch({query: `${authorQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n      }\n      return alternates ? alternateQueries : [`${authorQuery} AND ${query}`];\n    }\n    return addPublisherToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries});\n    //return [];\n  }\n\n  function addPublisherToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n    const [publisherQuery] = bibPublishers(record);\n    if (publisherQuery !== undefined) {\n      if (addYear) {\n        const newAlternateQueries = alternates ? [...alternateQueries, `${publisherQuery} AND ${query}`] : alternateQueries;\n        return addYearToSearch({query: `${publisherQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n      }\n      return alternates ? alternateQueries : [`${publisherQuery} AND ${query}`];\n    }\n    if (queryIsOkAlone && !addYear) {\n      return alternates ? alternateQueries : [`${query}`];\n    }\n    return addYearToSearch({query, queryIsOkAlone, alternates, alternateQueries});\n  }\n\n  function addYearToSearch({query, queryIsOkAlone = false, alternates = false, alternateQueries = []}) {\n    const [yearQuery] = bibYear(record);\n    if (yearQuery !== undefined) {\n      const yearAndOtherQueries = `${yearQuery} AND ${query}`;\n      return alternates ? [...alternateQueries, yearAndOtherQueries] : [yearAndOtherQueries];\n    }\n    if (queryIsOkAlone) {\n      return alternates ? alternateQueries : [`${query}`];\n    }\n    return [];\n  }\n\n}\n\nexport function bibAuthors(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the first author`);\n  //debugData(record);\n\n  const author = getAuthor(record);\n\n  if (testStringOrNumber(author)) {\n    const formatted = String(author)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for authors starting with a boolean\n    const useWordSearch = checkUseWordSearch(formatted);\n    // Prevent too many matches by having a minimum length\n    debugData(`Author string: ${formatted}`);\n    if (formatted.length >= 5) {\n      return [`dc.author=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n    }\n    return [];\n  }\n\n  return [];\n\n  function getAuthor(record) {\n    //debugData(record);\n    const [field] = record.get(/^(?:100|110|111|700|710|711)$/u);\n    //debugData(field);\n\n    if (field) {\n      const authorString = field.subfields\n        .filter(({code}) => ['a'].includes(code)) // We might use different subfield code sets for X00, X10 and X11?\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n      return authorString;\n    }\n    return false;\n  }\n}\n\nexport function bibPublishers(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the first publisher`);\n  //debugData(record);\n\n  const publisher = getPublisher(record);\n  if (testStringOrNumber(publisher)) {\n    const formatted = String(publisher)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    debugData(`Publisher string: '${formatted}'`);\n    if (formatted.length === 0) {\n      return [];\n    }\n    // use non-wildcard word search from dc.publisher\n    return [`dc.publisher=\"${formatted}\"`];\n  }\n\n  return [];\n\n  function getPublisher(record) {\n    //debugData(record);\n    const [field] = record.get(/^26[04]$/u).filter(f => f.subfields.some(sf => sf.code === 'b')); // Filter removes copyright-only fields\n    //debugData(field);\n\n    if (field) {\n      const publisherString = field.subfields\n        .filter(({code}) => ['b'].includes(code))\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ')\n        .trim();\n      return publisherString;\n    }\n    return false;\n  }\n}\n\nexport function bibYear(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibYear');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the publishing year`);\n\n  const year = getYear(record);\n  if (year) {\n    return [`dc.date=\"${year}\"`];\n  }\n  return [];\n\n  function getYear(record) {\n    const [f008] = record.get(/^008$/u);\n    if (!f008 || !f008.value || f008.value.length < 11) {\n      debug('f008 missing/crappy');\n      return false;\n    }\n\n    debugData(`f008: ${JSON.stringify(f008)}`);\n    const {value} = f008;\n    if (value === '||||' || value === '    ') { // Meaningless values\n      return false;\n    }\n    return value.slice(7, 11);\n  }\n}\n\nexport function checkUseWordSearch(formatted) {\n  const lowercased = formatted.toLowerCase();\n  // Note: add a space to startWords to catch just actual boolean words\n  const booleanStartWords = ['and ', 'or ', 'nor ', 'not '];\n  return booleanStartWords.some(word => lowercased.startsWith(word));\n}\n\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  // DEVELOP: should we query also f015 and f028?\n\n  const fields = record.get(/^(?<def>020|022|024)$/u);\n  const identifiers = [].concat(...fields.map(toIdentifiers));\n  const uniqueIdentifiers = [...new Set(identifiers)];\n\n  debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n  debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n  debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n  if (uniqueIdentifiers.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n  function toIdentifiers({tag, subfields}) {\n    const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n    const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n    if (tag === '022') {\n      return subfields\n        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))\n      .map(({value}) => String(value));\n  }\n}\n"],
-  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,gCAAgC,mBAAmB,0BAAyB;AACvG,SAAQ,gCAAgC,+BAA+B,4BAA2B;AAClG,SAAQ,qBAAoB;AAC5B,SAAQ,gBAAe;AAEvB,MAAM,QAAQ,kBAAkB,0DAA0D;AAE1F,MAAM,wBAAwB;AAEvB,gBAAS,aAAa,QAAQ;AA0BnC,QAAMA,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AAGrC,EAAAA,OAAM,iCAAiC;AAEvC,QAAM,QAAQ,OAAO,IAAI,KAAK;AAC9B,YAAU,eAAe,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAElE,SAAO,MAAM,SAAS,IAAI,aAAa,KAAK,IAAI,CAAC;AAEjD,WAAS,aAAaC,QAAO;AAC3B,IAAAD,OAAM,wCAAwC;AAE9C,UAAM,aAAa,cAAcC,MAAK;AAEtC,QAAI,WAAW,SAAS,GAAG;AACzB,MAAAD,OAAM,2CAA2C;AACjD,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,aAAa,UAAU,YAAY,kBAAkB;AAE3D,WAAO;AAEP,aAAS,cAAcC,QAAO;AAC5B,MAAAD,OAAM,qCAAqC;AAG3C,YAAME,cAAaD,OAAM,IAAI,WAAW,EAAE,OAAO,iBAAe,WAAW;AAC3E,aAAOC;AAEP,eAAS,YAAY,OAAO;AAC1B,QAAAF,OAAM,6BAA6B;AAEnC,eAAO,+BAA+B,KAAK,IAAI,gBAAgB,KAAK,IAAI;AAExE,iBAAS,gBAAgBG,QAAO;AAC9B,UAAAH,OAAM,8BAA8B;AACpC,gBAAM,CAAC,GAAG,IAAI,kBAAkBG,QAAO,GAAG;AAC1C,gBAAM,CAAC,GAAG,IAAI,kBAAkBA,QAAO,GAAG;AAE1C,gBAAM,aAAa,mBAAmB,0BAA0B,GAAG,CAAC;AACpE,gBAAM,aAAa,0BAA0B,GAAG;AAEhD,oBAAU,GAAG,KAAK,UAAU,GAAG,CAAC,MAAM,KAAK,UAAU,GAAG,CAAC,EAAE;AAC3D,iBAAO,WAAW,OAAO,UAAU;AAAA,QACrC;AAEA,iBAAS,mBAAmB,eAAe;AACzC,gBAAM,oBAAqB;AAC3B,gBAAM,kBAAkB,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,mBAAmB,OAAO,IAAI;AACxH,oBAAU,cAAc,aAAa,OAAO,eAAe,EAAE;AAC7D,iBAAO;AAAA,QACT;AAEA,iBAAS,0BAA0B,eAAe;AAChD,oBAAU,eAAe,aAAa,EAAE;AACxC,gBAAM,qBAAsB;AAC5B,iBAAO,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,oBAAoB,EAAE,IAAI;AAAA,QACrG;AAAA,MAEF;AAAA,IACF;AAAA,EACF;AACF;AAEO,gBAAS,cAAc,QAAQ;AAGpC,QAAMH,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,iCAAiC;AAKvC,QAAM,aAAa,kBAAkB,MAAM;AAE3C,YAAU,uBAAuB,WAAW,MAAM,MAAM,KAAK,UAAU,UAAU,CAAC,EAAE;AAEpF,MAAI,WAAW,SAAS,GAAG;AACzB,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,YAAY,mBAAmB;AAClD;AAKO,gBAAS,kBAAkB,QAAQ;AACxC,QAAMA,SAAQ,kBAAkB,4EAA4E;AAC5G,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,8BAA8B;AACpC,QAAM,QAAQ,qBAAqB,MAAM;AACzC,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,CAAC,EAAE,IAAI,+BAA+B,MAAM,CAAC,CAAC;AACpD,EAAAA,OAAM,aAAa,KAAK,UAAU,EAAE,CAAC,EAAE;AAGvC,QAAM,OAAO,8BAA8B,MAAM,CAAC,CAAC;AACnD,EAAAA,OAAM,GAAG,cAAc,MAAM,CAAC,CAAC,CAAC,OAAO,OAAO,OAAO,KAAK,EAAE;AAC5D,MAAI,MAAM;AACR,WAAO,CAAC,WAAW,IAAI,4BAA4B,EAAE,EAAE;AAAA,EACzD;AAEA,SAAO,CAAC,uBAAuB,EAAE,EAAE;AAErC;AAQO,gBAAS,SAAS,QAAQ;AAE/B,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,EAAC,CAAC;AAC7D;AAEO,gBAAS,eAAe,QAAQ;AACrC,QAAM,gBAAgB;AAGtB,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,IAAG,CAAC;AAC/D;AAEO,gBAAS,mBAAmB,QAAQ;AACzC,QAAM,8BAA8B;AAIpC,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,KAAI,CAAC;AAC9E;AAEO,gBAAS,6BAA6B,QAAQ;AACnD,QAAM,8BAA8B;AAGpC,QAAM,gBAAgB,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,MAAM,YAAY,MAAM,kBAAkB,CAAC,EAAC,CAAC;AACnI,QAAM,GAAG,KAAK,UAAU,aAAa,CAAC,EAAE;AACxC,SAAO,EAAC,WAAW,MAAM,KAAK,aAAa,EAAE,QAAQ,GAAG,eAAe,aAAY;AACrF;AAEA,SAAS,iBAAiB,QAAQ;AAChC,QAAM,QAAQ,SAAS,QAAQ,CAAC,GAAG,CAAC;AACpC,MAAI,OAAO;AACT,UAAM,SAAS,MAAM,MAAM,GAAG;AAI9B,QAAI,SAAS,KAAK,MAAM,SAAS,IAAI;AACnC,YAAM,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK;AAG/B,UAAI,QAAQ,KAAK,UAAU,KAAK,QAAM,GAAG,SAAS,OAAO,GAAG,MAAM,MAAM,MAAM,CAAC,KAAK,MAAM,UAAU,uBAAuB;AACzH,eAAO;AAAA,MACT;AACA,aAAO,SAAS,QAAQ,CAAC,KAAK,GAAG,CAAC;AAAA,IACpC;AAAA,EACF;AACA,SAAO;AACT;AAEA,SAAS,QAAQ,QAAQ,iBAAiB,aAAa,OAAO;AAC5D,QAAM,QAAQ,iBAAiB,MAAM;AACrC,MAAI,CAAC,mBAAmB,KAAK,GAAG;AAC9B,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,YAAY,kBAAkB,KAAK;AAGzC,QAAM,gBAAgB,mBAAmB,SAAS;AAGlD,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,aAAa,gBAAgB,KAAK,GAAG,GAAG,SAAS,MAAM,WAAW,IAAI;AAAA,EAChF;AAEA,QAAM,iBAAiB,CAAC,iBAAiB,UAAU,UAAU;AAG7D,SAAO,CAAC,aAAa,iBAAiB,CAAC,iBAAiB,KAAK,GAAG,GAAG,SAAS,GAAG,iBAAiB,MAAM,EAAE,KAAK,WAAW,cAAc;AAEtI,WAAS,kBAAkBI,QAAO;AAChC,UAAMC,aAAY,OAAOD,MAAK,EAC3B,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,2BAA2B,EAAE,EACrC,QAAQ,SAAS,GAAG,EACpB,KAAK,EACL,QAAQ,mBAAmB,IAAI,EAC/B,KAAK;AAER,WAAOC;AAAA,EACT;AACF;AAEO,gBAAS,wBAAwB,EAAC,QAAQ,iBAAiB,UAAU,OAAO,aAAa,OAAO,mBAAmB,CAAC,EAAC,GAAG;AAC7H,QAAM,6CAA6C,eAAe,cAAc,OAAO,iBAAiB,UAAU,EAAE;AACpH,QAAM,CAAC,OAAO,WAAW,cAAc,IAAI,QAAQ,QAAQ,iBAAiB,UAAU;AACtF,MAAI,UAAU,QAAW;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,UAAU,KAAK,EAAE;AAIvB,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,KAAK;AAAA,EACf;AAEA,QAAM,sBAAsB,aAAa,CAAC,GAAG,kBAAkB,KAAK,IAAI;AAExE,SAAO,mBAAmB,EAAC,OAAO,gBAAgB,SAAS,YAAY,kBAAkB,oBAAmB,CAAC;AAE7G,WAAS,mBAAmB,EAAC,OAAAC,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACvH,UAAM,oBAAoB;AAC1B,UAAM,CAAC,WAAW,IAAI,WAAW,MAAM;AACvC,QAAI,gBAAgB,QAAW;AAC7B,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,WAAW,QAAQJ,MAAK,EAAE,IAAII;AAChG,eAAO,gBAAgB,EAAC,OAAO,GAAG,WAAW,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MACxI;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,WAAW,QAAQJ,MAAK,EAAE;AAAA,IACvE;AACA,WAAO,qBAAqB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,SAAAC,UAAS,YAAAC,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAE5F;AAEA,WAAS,qBAAqB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACzH,UAAM,CAAC,cAAc,IAAI,cAAc,MAAM;AAC7C,QAAI,mBAAmB,QAAW;AAChC,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,cAAc,QAAQJ,MAAK,EAAE,IAAII;AACnG,eAAO,gBAAgB,EAAC,OAAO,GAAG,cAAc,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MAC3I;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,cAAc,QAAQJ,MAAK,EAAE;AAAA,IAC1E;AACA,QAAIC,mBAAkB,CAACC,UAAS;AAC9B,aAAOC,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,gBAAgB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,YAAAE,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAC9E;AAEA,WAAS,gBAAgB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,YAAAE,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACnG,UAAM,CAAC,SAAS,IAAI,QAAQ,MAAM;AAClC,QAAI,cAAc,QAAW;AAC3B,YAAM,sBAAsB,GAAG,SAAS,QAAQJ,MAAK;AACrD,aAAOG,cAAa,CAAC,GAAGC,mBAAkB,mBAAmB,IAAI,CAAC,mBAAmB;AAAA,IACvF;AACA,QAAIH,iBAAgB;AAClB,aAAOE,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,CAAC;AAAA,EACV;AAEF;AAEO,gBAAS,WAAW,QAAQ;AACjC,QAAMN,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,qCAAqC;AAG3C,QAAM,SAAS,UAAU,MAAM;AAE/B,MAAI,mBAAmB,MAAM,GAAG;AAC9B,UAAM,YAAY,OAAO,MAAM,EAC5B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAGR,UAAM,gBAAgB,mBAAmB,SAAS;AAElD,cAAU,kBAAkB,SAAS,EAAE;AACvC,QAAI,UAAU,UAAU,GAAG;AACzB,aAAO,CAAC,cAAc,gBAAgB,KAAK,GAAG,GAAG,SAAS,IAAI;AAAA,IAChE;AACA,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,CAAC;AAER,WAAS,UAAUY,SAAQ;AAEzB,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,gCAAgC;AAG3D,QAAI,OAAO;AACT,YAAM,eAAe,MAAM,UACxB,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG;AACX,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,cAAc,QAAQ;AACpC,QAAMZ,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAG9C,QAAM,YAAY,aAAa,MAAM;AACrC,MAAI,mBAAmB,SAAS,GAAG;AACjC,UAAM,YAAY,OAAO,SAAS,EAC/B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAER,cAAU,sBAAsB,SAAS,GAAG;AAC5C,QAAI,UAAU,WAAW,GAAG;AAC1B,aAAO,CAAC;AAAA,IACV;AAEA,WAAO,CAAC,iBAAiB,SAAS,GAAG;AAAA,EACvC;AAEA,SAAO,CAAC;AAER,WAAS,aAAaY,SAAQ;AAE5B,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,WAAW,EAAE,OAAO,OAAK,EAAE,UAAU,KAAK,QAAM,GAAG,SAAS,GAAG,CAAC;AAG3F,QAAI,OAAO;AACT,YAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG,EACR,KAAK;AACR,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,QAAQ,QAAQ;AAC9B,QAAMZ,SAAQ,kBAAkB,kEAAkE;AAClG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAE9C,QAAM,OAAO,QAAQ,MAAM;AAC3B,MAAI,MAAM;AACR,WAAO,CAAC,YAAY,IAAI,GAAG;AAAA,EAC7B;AACA,SAAO,CAAC;AAER,WAAS,QAAQY,SAAQ;AACvB,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,QAAI,CAAC,QAAQ,CAAC,KAAK,SAAS,KAAK,MAAM,SAAS,IAAI;AAClD,MAAAZ,OAAM,qBAAqB;AAC3B,aAAO;AAAA,IACT;AAEA,cAAU,SAAS,KAAK,UAAU,IAAI,CAAC,EAAE;AACzC,UAAM,EAAC,MAAK,IAAI;AAChB,QAAI,UAAU,UAAU,UAAU,QAAQ;AACxC,aAAO;AAAA,IACT;AACA,WAAO,MAAM,MAAM,GAAG,EAAE;AAAA,EAC1B;AACF;AAEO,gBAAS,mBAAmB,WAAW;AAC5C,QAAM,aAAa,UAAU,YAAY;AAEzC,QAAM,oBAAoB,CAAC,QAAQ,OAAO,QAAQ,MAAM;AACxD,SAAO,kBAAkB,KAAK,UAAQ,WAAW,WAAW,IAAI,CAAC;AACnE;AAEO,gBAAS,uBAAuB,QAAQ;AAE7C,QAAMA,SAAQ,kBAAkB,iFAAiF;AACjH,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,2CAA2C;AAIjD,QAAM,SAAS,OAAO,IAAI,wBAAwB;AAClD,QAAM,cAAc,CAAC,EAAE,OAAO,GAAG,OAAO,IAAI,aAAa,CAAC;AAC1D,QAAM,oBAAoB,CAAC,GAAG,IAAI,IAAI,WAAW,CAAC;AAElD,YAAU,+BAA+B,KAAK,UAAU,MAAM,CAAC,EAAE;AACjE,YAAU,gBAAgB,YAAY,MAAM,MAAM,KAAK,UAAU,WAAW,CAAC,EAAE;AAC/E,YAAU,uBAAuB,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAElG,MAAI,kBAAkB,SAAS,GAAG;AAChC,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,mBAAmB,eAAe;AAEnD,WAAS,cAAc,EAAC,KAAK,UAAS,GAAG;AACvC,UAAM,iBAAkB;AACxB,UAAM,gBAAiB;AAEvB,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EAC3H,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,WAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,cAAc,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACrH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,EACnC;AACF;",
+  "sourcesContent": ["import createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils.js';\nimport {extractHostMelindaIdsFromField, extractPublicationYearFrom773, getHostMelindaFields} from './component.js';\nimport {fieldToString} from '@natlibfi/marc-record-validators-melinda';\nimport {getTitle} from '../../match-detection/features/bib/title.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query');\n\nconst IS_OK_ALONE_THRESHOLD = 5;\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibHostComponents');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for hostIds`);\n  const f773s = getHostMelindaFields(record);\n  if (f773s.length !== 1) { // This needs some thinking...\n    return [];\n  }\n\n  const [id] = extractHostMelindaIdsFromField(f773s[0]);\n  debug(`Found id: ${JSON.stringify(id)}`);\n\n  // NB! record.isCR() does not work if LDR/07=a, so we get year from 773$g!\n  const year = extractPublicationYearFrom773(f773s[0]);\n  debug(`${fieldToString(f773s[0])} => ${year ? year : 'N/A'}`);\n  if (year) {\n    return [`dc.date=${year} AND melinda.partsofhost=${id}`];\n  }\n\n  return [`melinda.partsofhost=${id}`];\n\n}\n\n\n\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\nexport function bibTitle(record) {\n  // We get author/publisher only when formatted title is shorter than 5 chars\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 5});\n}\n\nexport function bibTitleAuthor(record) {\n  debug('bibTitleAuthor');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 100});\n}\n\nexport function bibTitleAuthorYear(record) {\n  debug('bibTitleAuthorYearAlternates');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n\n  return bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true});\n}\n\nexport function bibTitleAuthorYearAlternates(record) {\n  debug('bibTitleAuthorYearAlternates');\n  // We use onlyTitleLength that is longer than our formatted length to\n  // get an author or an publisher always\n  const origQueryList = bibTitleAuthorPublisher({record, onlyTitleLength: 100, addYear: true, alternates: true, alternateQueries: []});\n  debug(`${JSON.stringify(origQueryList)}`);\n  return {queryList: Array.from(origQueryList).reverse(), queryListType: 'alternates'};\n}\n\nfunction getTitleForQuery(record) {\n  const title = getTitle(record, ['a']);\n  if (title) {\n    const nWords = title.split(' ');\n    // If lone $a is deemed too short, fetch $b as well:\n    // (NV: I've seen pairs with 245$a-only vs 245$a$b, so I'd like to use $a only if it is long enough)\n\n    if (nWords < 3 || title.length < 12) {\n      const [f245] = record.get('245');\n      // If punctuation is ' =' I think that f245$a is good enough despite shortness. Trying to balance between two bad situations...\n      // \"Suo (= short title) siell\u00E4, vetel\u00E4 (= missing $b name in another language) t\u00E4\u00E4ll\u00E4...\"\n      if (f245 && f245.subfields.find(sf => sf.code === 'a' && sf.value.match(/ =$/u)) && title.length >= IS_OK_ALONE_THRESHOLD) {\n        return title;\n      }\n      return getTitle(record, ['a', 'b']); // Try to get a longer title\n    }\n  }\n  return title;\n}\n\nfunction dcTitle(record, onlyTitleLength, alternates = false) {\n  const title = getTitleForQuery(record);\n  if (!testStringOrNumber(title)) {\n    return [];\n  }\n\n  const formatted = getFormattedTitle(title);\n\n  // use word search for titles starting with a boolean\n  const useWordSearch = checkUseWordSearch(formatted);\n  // Prevent too many matches / SRU crashing by having a minimum length\n  // Note that currently this fails matching if there are no matches from previous matchers\n  if (formatted.length >= onlyTitleLength && !alternates) {\n    return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`, formatted, true];\n  }\n\n  const queryIsOkAlone = !useWordSearch && formatted.length >= IS_OK_ALONE_THRESHOLD;\n\n  // use word search without ending * also in combination searches to avoid SRU-server crashes [MRA-189]\n  return [`dc.title=\"${useWordSearch || !queryIsOkAlone ? '' : '^'}${formatted}${queryIsOkAlone ? '*' : ''}\"`, formatted, queryIsOkAlone];\n\n  function getFormattedTitle(title) {\n    const formatted = String(title)\n      .replace(/[\\-+ !\"(){}\\[\\]<>;:.?/@*%=^_`~]/gu, ' ')\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '') // Apparently matches to/works with non-aplhabetic scripts such as Chinese as well\n      .replace(/  +/gu, ' ') // Clean up concurrent spaces from eg. subfield changes\n      .trim()\n      .replace(/^(.{30}\\S*) .*$/, \"$1\")\n      .trim();\n\n    return formatted;\n  }\n}\n\nexport function bibTitleAuthorPublisher({record, onlyTitleLength, addYear = false, alternates = false, alternateQueries = []}) {\n  debug(`bibTitleAuthorPublisher, onlyTitleLength: ${onlyTitleLength}, addYear: ${addYear}, alternates: ${alternates}`);\n  const [query, formatted, queryIsOkAlone] = dcTitle(record, onlyTitleLength, alternates);\n  if (query === undefined) {\n    return [];\n  }\n\n  debug(`query: ${query}`);\n\n  // Prevent too many matches / SRU crashing by having a minimum length\n  // Note that currently this fails matching if there are no matches from previous matchers\n  if (formatted.length >= onlyTitleLength && !alternates) {\n    return [query];\n  }\n\n  const newAlternateQueries = alternates ? [...alternateQueries, query] : alternateQueries;\n\n  return addAuthorsToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries: newAlternateQueries});\n\n  function addAuthorsToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n    debug('addAuthorsToSearch');\n    const [authorQuery] = bibAuthors(record);\n    if (authorQuery !== undefined) {\n      if (addYear) {\n        const newAlternateQueries = alternates ? [...alternateQueries, `${authorQuery} AND ${query}`] : alternateQueries;\n        return addYearToSearch({query: `${authorQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n      }\n      return alternates ? alternateQueries : [`${authorQuery} AND ${query}`];\n    }\n    return addPublisherToSearch({query, queryIsOkAlone, addYear, alternates, alternateQueries});\n    //return [];\n  }\n\n  function addPublisherToSearch({query, queryIsOkAlone = false, addYear = false, alternates = false, alternateQueries = []}) {\n    const [publisherQuery] = bibPublishers(record);\n    if (publisherQuery !== undefined) {\n      if (addYear) {\n        const newAlternateQueries = alternates ? [...alternateQueries, `${publisherQuery} AND ${query}`] : alternateQueries;\n        return addYearToSearch({query: `${publisherQuery} AND ${query}`, queryIsOkAlone: true, alternates, alternateQueries: newAlternateQueries});\n      }\n      return alternates ? alternateQueries : [`${publisherQuery} AND ${query}`];\n    }\n    if (queryIsOkAlone && !addYear) {\n      return alternates ? alternateQueries : [`${query}`];\n    }\n    return addYearToSearch({query, queryIsOkAlone, alternates, alternateQueries});\n  }\n\n  function addYearToSearch({query, queryIsOkAlone = false, alternates = false, alternateQueries = []}) {\n    const [yearQuery] = bibYear(record);\n    if (yearQuery !== undefined) {\n      const yearAndOtherQueries = `${yearQuery} AND ${query}`;\n      return alternates ? [...alternateQueries, yearAndOtherQueries] : [yearAndOtherQueries];\n    }\n    if (queryIsOkAlone) {\n      return alternates ? alternateQueries : [`${query}`];\n    }\n    return [];\n  }\n\n}\n\nexport function bibAuthors(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the first author`);\n  //debugData(record);\n\n  const author = getAuthor(record);\n\n  if (testStringOrNumber(author)) {\n    const formatted = String(author)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for authors starting with a boolean\n    const useWordSearch = checkUseWordSearch(formatted);\n    // Prevent too many matches by having a minimum length\n    debugData(`Author string: ${formatted}`);\n    if (formatted.length >= 5) {\n      return [`dc.author=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n    }\n    return [];\n  }\n\n  return [];\n\n  function getAuthor(record) {\n    //debugData(record);\n    const [field] = record.get(/^(?:100|110|111|700|710|711)$/u);\n    //debugData(field);\n\n    if (field) {\n      const authorString = field.subfields\n        .filter(({code}) => ['a'].includes(code)) // We might use different subfield code sets for X00, X10 and X11?\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n      return authorString;\n    }\n    return false;\n  }\n}\n\nexport function bibPublishers(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibPublishers');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the first publisher`);\n  //debugData(record);\n\n  const publisher = getPublisher(record);\n  if (testStringOrNumber(publisher)) {\n    const formatted = String(publisher)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    debugData(`Publisher string: '${formatted}'`);\n    if (formatted.length === 0) {\n      return [];\n    }\n    // use non-wildcard word search from dc.publisher\n    return [`dc.publisher=\"${formatted}\"`];\n  }\n\n  return [];\n\n  function getPublisher(record) {\n    //debugData(record);\n    const [field] = record.get(/^26[04]$/u).filter(f => f.subfields.some(sf => sf.code === 'b')); // Filter removes copyright-only fields\n    //debugData(field);\n\n    if (field) {\n      const publisherString = field.subfields\n        .filter(({code}) => ['b'].includes(code))\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ')\n        .trim();\n      return publisherString;\n    }\n    return false;\n  }\n}\n\nexport function bibYear(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibYear');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the publishing year`);\n\n  const year = getYear(record);\n  if (year) {\n    return [`dc.date=\"${year}\"`];\n  }\n  return [];\n\n  function getYear(record) {\n    const [f008] = record.get(/^008$/u);\n    if (!f008 || !f008.value || f008.value.length < 11) {\n      debug('f008 missing/crappy');\n      return false;\n    }\n\n    debugData(`f008: ${JSON.stringify(f008)}`);\n    const {value} = f008;\n    if (value === '||||' || value === '    ') { // Meaningless values\n      return false;\n    }\n    return value.slice(7, 11);\n  }\n}\n\nexport function checkUseWordSearch(formatted) {\n  const lowercased = formatted.toLowerCase();\n  // Note: add a space to startWords to catch just actual boolean words\n  const booleanStartWords = ['and ', 'or ', 'nor ', 'not '];\n  return booleanStartWords.some(word => lowercased.startsWith(word));\n}\n\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  // DEVELOP: should we query also f015 and f028?\n\n  // const fields = record.get(/^(?<def>020|022|024)$/u);\n  const fields = record.get(/^(?<def>015|020|022|024|028)$/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-:]+$/u);\n\n    if (tag === '015') { // TODO: test (does this use the right index etc.)\n      return subfields\n        .filter(sf => sf.code === 'a' && sf.value.match(/^[A-Z0-9\\-]+$/ui))\n        .map(sf => sf.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    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 === '028') { // TODO: test\n      const a = subfields.find(sf => sf.code === 'a');\n      if (a) {\n        const b = subfields.find(sf => sf.code === 'b');\n        // TODO: normalize?\n        if (b) {\n          return [a.value, `${b.value} ${a.value}`];\n        }\n        return [a.value];\n      }\n    }\n\n    // Default seems to be 024:\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": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,gCAAgC,mBAAmB,0BAAyB;AACvG,SAAQ,gCAAgC,+BAA+B,4BAA2B;AAClG,SAAQ,qBAAoB;AAC5B,SAAQ,gBAAe;AAEvB,MAAM,QAAQ,kBAAkB,0DAA0D;AAE1F,MAAM,wBAAwB;AAEvB,gBAAS,aAAa,QAAQ;AA0BnC,QAAMA,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AAGrC,EAAAA,OAAM,iCAAiC;AAEvC,QAAM,QAAQ,OAAO,IAAI,KAAK;AAC9B,YAAU,eAAe,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAElE,SAAO,MAAM,SAAS,IAAI,aAAa,KAAK,IAAI,CAAC;AAEjD,WAAS,aAAaC,QAAO;AAC3B,IAAAD,OAAM,wCAAwC;AAE9C,UAAM,aAAa,cAAcC,MAAK;AAEtC,QAAI,WAAW,SAAS,GAAG;AACzB,MAAAD,OAAM,2CAA2C;AACjD,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,aAAa,UAAU,YAAY,kBAAkB;AAE3D,WAAO;AAEP,aAAS,cAAcC,QAAO;AAC5B,MAAAD,OAAM,qCAAqC;AAG3C,YAAME,cAAaD,OAAM,IAAI,WAAW,EAAE,OAAO,iBAAe,WAAW;AAC3E,aAAOC;AAEP,eAAS,YAAY,OAAO;AAC1B,QAAAF,OAAM,6BAA6B;AAEnC,eAAO,+BAA+B,KAAK,IAAI,gBAAgB,KAAK,IAAI;AAExE,iBAAS,gBAAgBG,QAAO;AAC9B,UAAAH,OAAM,8BAA8B;AACpC,gBAAM,CAAC,GAAG,IAAI,kBAAkBG,QAAO,GAAG;AAC1C,gBAAM,CAAC,GAAG,IAAI,kBAAkBA,QAAO,GAAG;AAE1C,gBAAM,aAAa,mBAAmB,0BAA0B,GAAG,CAAC;AACpE,gBAAM,aAAa,0BAA0B,GAAG;AAEhD,oBAAU,GAAG,KAAK,UAAU,GAAG,CAAC,MAAM,KAAK,UAAU,GAAG,CAAC,EAAE;AAC3D,iBAAO,WAAW,OAAO,UAAU;AAAA,QACrC;AAEA,iBAAS,mBAAmB,eAAe;AACzC,gBAAM,oBAAqB;AAC3B,gBAAM,kBAAkB,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,mBAAmB,OAAO,IAAI;AACxH,oBAAU,cAAc,aAAa,OAAO,eAAe,EAAE;AAC7D,iBAAO;AAAA,QACT;AAEA,iBAAS,0BAA0B,eAAe;AAChD,oBAAU,eAAe,aAAa,EAAE;AACxC,gBAAM,qBAAsB;AAC5B,iBAAO,mBAAmB,aAAa,IAAI,OAAO,aAAa,EAAE,QAAQ,oBAAoB,EAAE,IAAI;AAAA,QACrG;AAAA,MAEF;AAAA,IACF;AAAA,EACF;AACF;AAEO,gBAAS,cAAc,QAAQ;AAGpC,QAAMH,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,iCAAiC;AAKvC,QAAM,aAAa,kBAAkB,MAAM;AAE3C,YAAU,uBAAuB,WAAW,MAAM,MAAM,KAAK,UAAU,UAAU,CAAC,EAAE;AAEpF,MAAI,WAAW,SAAS,GAAG;AACzB,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,YAAY,mBAAmB;AAClD;AAKO,gBAAS,kBAAkB,QAAQ;AACxC,QAAMA,SAAQ,kBAAkB,4EAA4E;AAC5G,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,8BAA8B;AACpC,QAAM,QAAQ,qBAAqB,MAAM;AACzC,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,CAAC,EAAE,IAAI,+BAA+B,MAAM,CAAC,CAAC;AACpD,EAAAA,OAAM,aAAa,KAAK,UAAU,EAAE,CAAC,EAAE;AAGvC,QAAM,OAAO,8BAA8B,MAAM,CAAC,CAAC;AACnD,EAAAA,OAAM,GAAG,cAAc,MAAM,CAAC,CAAC,CAAC,OAAO,OAAO,OAAO,KAAK,EAAE;AAC5D,MAAI,MAAM;AACR,WAAO,CAAC,WAAW,IAAI,4BAA4B,EAAE,EAAE;AAAA,EACzD;AAEA,SAAO,CAAC,uBAAuB,EAAE,EAAE;AAErC;AAQO,gBAAS,SAAS,QAAQ;AAE/B,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,EAAC,CAAC;AAC7D;AAEO,gBAAS,eAAe,QAAQ;AACrC,QAAM,gBAAgB;AAGtB,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,IAAG,CAAC;AAC/D;AAEO,gBAAS,mBAAmB,QAAQ;AACzC,QAAM,8BAA8B;AAIpC,SAAO,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,KAAI,CAAC;AAC9E;AAEO,gBAAS,6BAA6B,QAAQ;AACnD,QAAM,8BAA8B;AAGpC,QAAM,gBAAgB,wBAAwB,EAAC,QAAQ,iBAAiB,KAAK,SAAS,MAAM,YAAY,MAAM,kBAAkB,CAAC,EAAC,CAAC;AACnI,QAAM,GAAG,KAAK,UAAU,aAAa,CAAC,EAAE;AACxC,SAAO,EAAC,WAAW,MAAM,KAAK,aAAa,EAAE,QAAQ,GAAG,eAAe,aAAY;AACrF;AAEA,SAAS,iBAAiB,QAAQ;AAChC,QAAM,QAAQ,SAAS,QAAQ,CAAC,GAAG,CAAC;AACpC,MAAI,OAAO;AACT,UAAM,SAAS,MAAM,MAAM,GAAG;AAI9B,QAAI,SAAS,KAAK,MAAM,SAAS,IAAI;AACnC,YAAM,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK;AAG/B,UAAI,QAAQ,KAAK,UAAU,KAAK,QAAM,GAAG,SAAS,OAAO,GAAG,MAAM,MAAM,MAAM,CAAC,KAAK,MAAM,UAAU,uBAAuB;AACzH,eAAO;AAAA,MACT;AACA,aAAO,SAAS,QAAQ,CAAC,KAAK,GAAG,CAAC;AAAA,IACpC;AAAA,EACF;AACA,SAAO;AACT;AAEA,SAAS,QAAQ,QAAQ,iBAAiB,aAAa,OAAO;AAC5D,QAAM,QAAQ,iBAAiB,MAAM;AACrC,MAAI,CAAC,mBAAmB,KAAK,GAAG;AAC9B,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,YAAY,kBAAkB,KAAK;AAGzC,QAAM,gBAAgB,mBAAmB,SAAS;AAGlD,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,aAAa,gBAAgB,KAAK,GAAG,GAAG,SAAS,MAAM,WAAW,IAAI;AAAA,EAChF;AAEA,QAAM,iBAAiB,CAAC,iBAAiB,UAAU,UAAU;AAG7D,SAAO,CAAC,aAAa,iBAAiB,CAAC,iBAAiB,KAAK,GAAG,GAAG,SAAS,GAAG,iBAAiB,MAAM,EAAE,KAAK,WAAW,cAAc;AAEtI,WAAS,kBAAkBI,QAAO;AAChC,UAAMC,aAAY,OAAOD,MAAK,EAC3B,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,2BAA2B,EAAE,EACrC,QAAQ,SAAS,GAAG,EACpB,KAAK,EACL,QAAQ,mBAAmB,IAAI,EAC/B,KAAK;AAER,WAAOC;AAAA,EACT;AACF;AAEO,gBAAS,wBAAwB,EAAC,QAAQ,iBAAiB,UAAU,OAAO,aAAa,OAAO,mBAAmB,CAAC,EAAC,GAAG;AAC7H,QAAM,6CAA6C,eAAe,cAAc,OAAO,iBAAiB,UAAU,EAAE;AACpH,QAAM,CAAC,OAAO,WAAW,cAAc,IAAI,QAAQ,QAAQ,iBAAiB,UAAU;AACtF,MAAI,UAAU,QAAW;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,UAAU,KAAK,EAAE;AAIvB,MAAI,UAAU,UAAU,mBAAmB,CAAC,YAAY;AACtD,WAAO,CAAC,KAAK;AAAA,EACf;AAEA,QAAM,sBAAsB,aAAa,CAAC,GAAG,kBAAkB,KAAK,IAAI;AAExE,SAAO,mBAAmB,EAAC,OAAO,gBAAgB,SAAS,YAAY,kBAAkB,oBAAmB,CAAC;AAE7G,WAAS,mBAAmB,EAAC,OAAAC,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACvH,UAAM,oBAAoB;AAC1B,UAAM,CAAC,WAAW,IAAI,WAAW,MAAM;AACvC,QAAI,gBAAgB,QAAW;AAC7B,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,WAAW,QAAQJ,MAAK,EAAE,IAAII;AAChG,eAAO,gBAAgB,EAAC,OAAO,GAAG,WAAW,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MACxI;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,WAAW,QAAQJ,MAAK,EAAE;AAAA,IACvE;AACA,WAAO,qBAAqB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,SAAAC,UAAS,YAAAC,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAE5F;AAEA,WAAS,qBAAqB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,SAAAC,WAAU,OAAO,YAAAC,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACzH,UAAM,CAAC,cAAc,IAAI,cAAc,MAAM;AAC7C,QAAI,mBAAmB,QAAW;AAChC,UAAIF,UAAS;AACX,cAAMG,uBAAsBF,cAAa,CAAC,GAAGC,mBAAkB,GAAG,cAAc,QAAQJ,MAAK,EAAE,IAAII;AACnG,eAAO,gBAAgB,EAAC,OAAO,GAAG,cAAc,QAAQJ,MAAK,IAAI,gBAAgB,MAAM,YAAAG,aAAY,kBAAkBE,qBAAmB,CAAC;AAAA,MAC3I;AACA,aAAOF,cAAaC,oBAAmB,CAAC,GAAG,cAAc,QAAQJ,MAAK,EAAE;AAAA,IAC1E;AACA,QAAIC,mBAAkB,CAACC,UAAS;AAC9B,aAAOC,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,gBAAgB,EAAC,OAAAA,QAAO,gBAAAC,iBAAgB,YAAAE,aAAY,kBAAAC,kBAAgB,CAAC;AAAA,EAC9E;AAEA,WAAS,gBAAgB,EAAC,OAAAJ,QAAO,gBAAAC,kBAAiB,OAAO,YAAAE,cAAa,OAAO,kBAAAC,oBAAmB,CAAC,EAAC,GAAG;AACnG,UAAM,CAAC,SAAS,IAAI,QAAQ,MAAM;AAClC,QAAI,cAAc,QAAW;AAC3B,YAAM,sBAAsB,GAAG,SAAS,QAAQJ,MAAK;AACrD,aAAOG,cAAa,CAAC,GAAGC,mBAAkB,mBAAmB,IAAI,CAAC,mBAAmB;AAAA,IACvF;AACA,QAAIH,iBAAgB;AAClB,aAAOE,cAAaC,oBAAmB,CAAC,GAAGJ,MAAK,EAAE;AAAA,IACpD;AACA,WAAO,CAAC;AAAA,EACV;AAEF;AAEO,gBAAS,WAAW,QAAQ;AACjC,QAAMN,SAAQ,kBAAkB,qEAAqE;AACrG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,qCAAqC;AAG3C,QAAM,SAAS,UAAU,MAAM;AAE/B,MAAI,mBAAmB,MAAM,GAAG;AAC9B,UAAM,YAAY,OAAO,MAAM,EAC5B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAGR,UAAM,gBAAgB,mBAAmB,SAAS;AAElD,cAAU,kBAAkB,SAAS,EAAE;AACvC,QAAI,UAAU,UAAU,GAAG;AACzB,aAAO,CAAC,cAAc,gBAAgB,KAAK,GAAG,GAAG,SAAS,IAAI;AAAA,IAChE;AACA,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,CAAC;AAER,WAAS,UAAUY,SAAQ;AAEzB,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,gCAAgC;AAG3D,QAAI,OAAO;AACT,YAAM,eAAe,MAAM,UACxB,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG;AACX,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,cAAc,QAAQ;AACpC,QAAMZ,SAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAG9C,QAAM,YAAY,aAAa,MAAM;AACrC,MAAI,mBAAmB,SAAS,GAAG;AACjC,UAAM,YAAY,OAAO,SAAS,EAC/B,QAAQ,2BAA2B,EAAE,EAErC,QAAQ,QAAQ,GAAG,EACnB,KAAK,EACL,MAAM,GAAG,EAAE,EACX,KAAK;AAER,cAAU,sBAAsB,SAAS,GAAG;AAC5C,QAAI,UAAU,WAAW,GAAG;AAC1B,aAAO,CAAC;AAAA,IACV;AAEA,WAAO,CAAC,iBAAiB,SAAS,GAAG;AAAA,EACvC;AAEA,SAAO,CAAC;AAER,WAAS,aAAaY,SAAQ;AAE5B,UAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,WAAW,EAAE,OAAO,OAAK,EAAE,UAAU,KAAK,QAAM,GAAG,SAAS,GAAG,CAAC;AAG3F,QAAI,OAAO;AACT,YAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,CAAC,GAAG,EAAE,SAAS,IAAI,CAAC,EACvC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK,EAErB,KAAK,GAAG,EACR,KAAK;AACR,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,QAAQ,QAAQ;AAC9B,QAAMZ,SAAQ,kBAAkB,kEAAkE;AAClG,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,wCAAwC;AAE9C,QAAM,OAAO,QAAQ,MAAM;AAC3B,MAAI,MAAM;AACR,WAAO,CAAC,YAAY,IAAI,GAAG;AAAA,EAC7B;AACA,SAAO,CAAC;AAER,WAAS,QAAQY,SAAQ;AACvB,UAAM,CAAC,IAAI,IAAIA,QAAO,IAAI,QAAQ;AAClC,QAAI,CAAC,QAAQ,CAAC,KAAK,SAAS,KAAK,MAAM,SAAS,IAAI;AAClD,MAAAZ,OAAM,qBAAqB;AAC3B,aAAO;AAAA,IACT;AAEA,cAAU,SAAS,KAAK,UAAU,IAAI,CAAC,EAAE;AACzC,UAAM,EAAC,MAAK,IAAI;AAChB,QAAI,UAAU,UAAU,UAAU,QAAQ;AACxC,aAAO;AAAA,IACT;AACA,WAAO,MAAM,MAAM,GAAG,EAAE;AAAA,EAC1B;AACF;AAEO,gBAAS,mBAAmB,WAAW;AAC5C,QAAM,aAAa,UAAU,YAAY;AAEzC,QAAM,oBAAoB,CAAC,QAAQ,OAAO,QAAQ,MAAM;AACxD,SAAO,kBAAkB,KAAK,UAAQ,WAAW,WAAW,IAAI,CAAC;AACnE;AAEO,gBAAS,uBAAuB,QAAQ;AAE7C,QAAMA,SAAQ,kBAAkB,iFAAiF;AACjH,QAAM,YAAYA,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,2CAA2C;AAKjD,QAAM,SAAS,OAAO,IAAI,gCAAgC;AAC1D,QAAM,cAAc,CAAC,EAAE,OAAO,GAAG,OAAO,IAAI,aAAa,CAAC;AAC1D,QAAM,oBAAoB,CAAC,GAAG,IAAI,IAAI,WAAW,CAAC;AAElD,YAAU,+BAA+B,KAAK,UAAU,MAAM,CAAC,EAAE;AACjE,YAAU,gBAAgB,YAAY,MAAM,MAAM,KAAK,UAAU,WAAW,CAAC,EAAE;AAC/E,YAAU,uBAAuB,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAElG,MAAI,kBAAkB,SAAS,GAAG;AAChC,IAAAA,OAAM,2CAA2C;AACjD,WAAO,CAAC;AAAA,EACV;AAEA,SAAO,UAAU,mBAAmB,eAAe;AAEnD,WAAS,cAAc,EAAC,KAAK,UAAS,GAAG;AACvC,UAAM,iBAAkB;AACxB,UAAM,gBAAiB;AAEvB,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,QAAM,GAAG,SAAS,OAAO,GAAG,MAAM,MAAM,iBAAiB,CAAC,EACjE,IAAI,QAAM,GAAG,KAAK;AAAA,IACvB;AAEA,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,QAAI,QAAQ,OAAO;AACjB,aAAO,UACJ,OAAO,SAAO,CAAC,KAAK,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,eAAe,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EAC3H,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,IACnC;AAEA,QAAI,QAAQ,OAAO;AACjB,YAAM,IAAI,UAAU,KAAK,QAAM,GAAG,SAAS,GAAG;AAC9C,UAAI,GAAG;AACL,cAAM,IAAI,UAAU,KAAK,QAAM,GAAG,SAAS,GAAG;AAE9C,YAAI,GAAG;AACL,iBAAO,CAAC,EAAE,OAAO,GAAG,EAAE,KAAK,IAAI,EAAE,KAAK,EAAE;AAAA,QAC1C;AACA,eAAO,CAAC,EAAE,KAAK;AAAA,MACjB;AAAA,IACF;AAGA,WAAO,UACJ,OAAO,SAAO,CAAC,KAAK,GAAG,EAAE,SAAS,IAAI,IAAI,KAAK,mBAAmB,IAAI,KAAK,KAAK,cAAc,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACrH,IAAI,CAAC,EAAC,MAAK,MAAM,OAAO,KAAK,CAAC;AAAA,EACnC;AACF;",
   "names": ["debug", "fSids", "sidStrings", "field", "title", "formatted", "query", "queryIsOkAlone", "addYear", "alternates", "alternateQueries", "newAlternateQueries", "record"]
 }

package/dist/match-detection/features/bib/title.js

@@ -20,7 +20,7 @@ export default ({ threshold = 0.9 } = {}) => ({
     }
     return ["", "", "", ""];
     function normalizeTitle(title) {
-      return title.normalize("NFD").replace(/\b(?:ein|ett|I|one|uno|yksi)\b/uig, "1").replace(/\b(?:dos|kaksi|II|två|two|zwei)\b/uig, "2").replace(/\b(?:drei|III|kolme|three|tre|tres)\b/uig, "3").replace(/\b(?:four|fyra|IV|neljä|quat+ro|vier)\b/uig, "4").replace(/\b(?:five|fünf|fyra|viisi|V)\b/uig, "5").replace(/\b(?:kuusi|sechts|sex|six|VI)\b/uig, "6").replace(/\b(and|ja|och|und)\b/uig, "&").replace(/[^\p{Letter}\p{Number}&]/gu, "").toLowerCase();
+      return title.normalize("NFD").replace(/\b(?:ein|ett|I|one|uno|yksi)\b/uig, "1").replace(/\b(?:dos|kaksi|II|två|two|zwei)\b/uig, "2").replace(/\b(?:drei|III|kolme|three|tre|tres)\b/uig, "3").replace(/\b(?:four|fyra|IV|neljä|quat+ro|vier)\b/uig, "4").replace(/\b(?:five|fünf|fyra|viisi|V)\b/uig, "5").replace(/\b(?:kuusi|sechts|sex|six|VI)\b/uig, "6").replace(/\b(and|ja|och|und)\b/uig, "&").replace(/[^\p{Letter}\p{Number}& ]/gu, "").replace(/\s+/ug, " ").replace(/ $/u, "").replace(/^ /u, "").toLowerCase();
     }
   },
   compare: (origA, origB) => {
@@ -28,17 +28,21 @@ export default ({ threshold = 0.9 } = {}) => ({
     function compare2(a, b) {
       const [aa, ab, an, ap] = a;
       const [ba, bb, bn, bp] = b;
+      debugData(`COMPARE:
+['${aa}', '${ab}', '${an}', '${ap}'] VS
+['${ba}', '${bb}', '${bn}', '${bp}']`);
       const aFull = toFullTitle(a);
       const bFull = toFullTitle(b);
-      debug(`COMPARE:
-  '${aFull}' vs  
-  '${bFull}'`);
       if (isEmpty(aa) || isEmpty(ba)) {
         return -1;
       }
       if (aFull === bFull) {
         return TITLE_MAX;
       }
+      const [altAa, altAn, altBa, altBn] = reconstructTitleAndNumber(aa, an, ba, bn);
+      if (altAa) {
+        return compare2([altAa, ab, altAn, ap], [altBa, bb, altBn, bp]);
+      }
       if (ab && ab !== "" && ab === bb) {
         debug(`Ignore $b ${ab}`);
         return compare2([aa, "", an, ap], [ba, "", bn, bp]);
@@ -69,7 +73,7 @@ export default ({ threshold = 0.9 } = {}) => ({
       if (an && an !== bn) {
         return -1;
       }
-      const [distance, maxLength, correctness] = doLevenshtein(aFull, bFull);
+      const [distance, maxLength, correctness] = doLevenshtein(aFull.replace(/ /ug, ""), bFull.replace(/ /ug, ""));
       debug(`'${aFull}' vs '${bFull}': Max length = ${maxLength}, distance = ${distance}, correctness = ${correctness}`);
       if (correctness >= threshold) {
         return TITLE_MAX * 0.8;
@@ -97,6 +101,15 @@ export default ({ threshold = 0.9 } = {}) => ({
           return candScore;
         }
       }
+      if (an === bn && ap === bp) {
+        const candScore = TITLE_MAX / 2;
+        if (aa === bb && !ab) {
+          return candScore;
+        }
+        if (ba === ab && !bb) {
+          return candScore;
+        }
+      }
       return -0.5;
     }
     function isEmpty(x) {
@@ -121,6 +134,39 @@ export default ({ threshold = 0.9 } = {}) => ({
     function getMaxLength(str1, str2) {
       return str1.length > str2.length ? str1.length : str2.length;
     }
+    function reconstructTitleAndNumber(a1, n1, a2, n2) {
+      if (!n1 && n2) {
+        const [altA1, altN1] = splitToBaseAndNumber(a1);
+        if (altN1) {
+          debugData(`A: Reconstruct:
+ $a '${a1}' =>
+ $a '${altA1}' +
+ $n '${altN1}`);
+          return [altA1, altN1, a2, n2];
+        }
+      }
+      if (n1 && !n2) {
+        const [altA2, altN2] = splitToBaseAndNumber(a2);
+        if (altN2) {
+          debugData(`B: Reconstruct $a '${a2}' as $a '${altA2}' and $n '${altN2}`);
+          return [a1, n1, altA2, altN2];
+        }
+      }
+      return [null, null, null, null];
+    }
+    function splitToBaseAndNumber(val) {
+      const words = val.split(" ");
+      const len = words.length;
+      if (len < 2 || !words[len - 1].match(/^[1-9][0-9]*$/)) {
+        return [val, null];
+      }
+      if (len > 2 && words[len - 2].match(/^(del|osa|vol|volume)$/u)) {
+        const number2 = `${words[len - 2]} ${words[len - 1]}`;
+        return [words.slice(0, len - 2).join(" "), number2];
+      }
+      const number = words.pop();
+      return [words.join(" "), number];
+    }
   }
 });
 export function getTitle(record, targetSubfieldCodes = ["a", "b", "n", "p"]) {

package/dist/match-detection/features/bib/title.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/title.js"],
-  "sourcesContent": ["import createDebugLogger from 'debug';\nimport naturalPkg from 'natural';\nconst {LevenshteinDistance: leven} = naturalPkg;\nimport {testStringOrNumber} from '../../../matching-utils.js';\nimport { subfieldToString } from '@natlibfi/marc-record-validators-melinda/dist/utils.js';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\nconst debugData = debug.extend('data');\n\nconst TITLE_MAX = 0.5;\nexport default ({threshold = 0.9} = {}) => ({\n  name: 'Title',\n  extract: ({record, recordExternal}) => {\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n    const a = getTitle(record, ['a']);\n    debug(`${label}: title: ${a}`);\n\n    if (a) {\n      const b = normalizeTitle(getTitle(record, ['b'])) || '';\n      const n = normalizeTitle(getTitle(record, ['n'])) || '';\n      const p = normalizeTitle(getTitle(record, ['p'])) || '';\n      return [normalizeTitle(a), b, n, p];\n    }\n\n    return ['', '', '', ''];\n\n    function normalizeTitle(title) {\n      return title\n        // decompose unicode diacritics\n        .normalize('NFD')\n        .replace(/\\b(?:ein|ett|I|one|uno|yksi)\\b/uig, '1')\n        .replace(/\\b(?:dos|kaksi|II|tv\u00E5|two|zwei)\\b/uig, '2')\n        .replace(/\\b(?:drei|III|kolme|three|tre|tres)\\b/uig, '3')\n        .replace(/\\b(?:four|fyra|IV|nelj\u00E4|quat+ro|vier)\\b/uig, '4')\n        .replace(/\\b(?:five|f\u00FCnf|fyra|viisi|V)\\b/uig, '5')\n        .replace(/\\b(?:kuusi|sechts|sex|six|VI)\\b/uig, '6')\n        .replace(/\\b(and|ja|och|und)\\b/uig, '&')\n        // strip non-letters/numbers\n        // - note: combined with decomposing unicode diacritics this normalizes both 'saa' and 's\u00E4\u00E4' as 'saa'\n        // - we could precompose the Finnish letters back to avoid this\n        // - see validator normalize-utf8-diacritics for details\n        .replace(/[^\\p{Letter}\\p{Number}&]/gu, '')\n        .toLowerCase();\n    }\n\n  },\n  compare: (origA, origB) => {\n    return compare2(origA, origB);\n\n    function compare2(a, b) { // Added this wrapper function as I had issues with recursion\n\n      const [aa, ab, an, ap] = a;\n      const [ba, bb, bn, bp] = b;\n\n      const aFull = toFullTitle(a);\n      const bFull = toFullTitle(b);\n\n      debug(`COMPARE:\\n  '${aFull}' vs  \\n  '${bFull}'`);\n\n      if (isEmpty(aa) || isEmpty(ba)) {\n        return -1.0;\n      }\n\n      if (aFull === bFull) {\n        return TITLE_MAX;\n      }\n\n\n      if (ab && ab !== '' && ab === bb) { // Remove $b from equation (MELKEHITYS-3494)\n        debug(`Ignore \\$b ${ab}`);\n        return compare2([aa, '', an, ap], [ba, '', bn, bp]);\n      }\n      if (an && an !== '' && an === bn) { // Remove $n from equation\n        debug(`Ignore \\$n ${an}`);\n        return compare2([aa, ab, '', ap], [ba, bb, '', bp]);\n      }\n      if (ap && ap !== '' && ap === bp) { // Remove $p from equation\n        debug(`Ignore \\$p ${ap}`);\n        return compare2([aa, ab, an, ''], [ba, bb, bn, '']);\n      }\n      // There seems to be wobble between $b and $p\n      if (!isEmpty(ab) && ab === bp) {\n        return compare2([aa, '', an, ap], [ba, bb, bn, '']);\n      }\n      if (!isEmpty(ap) && ap === bp) {\n        return compare2([aa, ab, an, ''], [ba, '', bn, bp]);\n      }\n\n      if (an && bn && an.match(/[0-9]/u)) {\n        debug(`NUMBER FOUND. Compare ${an} and ${bn}`);\n        const atmp = an.replace(/[^0-9]/ug, '');\n        const btmp = bn.replace(/[^0-9]/ug, '');\n        if (atmp === btmp) {\n          debug(`Ignore \\$n '${an}' vs '${bn}'`);\n          return compare2([aa, ab, '', ap], [ba, bb, '', bp]);\n        }\n      }\n\n\n      // f245$n information is critical; it can not mismatch at all  (exceptions have already been handled above):\n      if (an && an !== bn) {\n        return -1.0;\n      }\n\n      const [distance, maxLength, correctness] = doLevenshtein(aFull, bFull);\n\n      debug(`'${aFull}' vs '${bFull}': Max length = ${maxLength}, distance = ${distance}, correctness = ${correctness}`);\n\n\n      if (correctness >= threshold) {\n        return TITLE_MAX * 0.8;\n      }\n\n      // Subset removal (MELKEHITYS-3498-ish)\n      if (ab && bb) { // At this point ab and bb are never equal...\n        // If X is a subset of Y, then remove X and shorten Y (= remove the shared content)\n        if (ab.indexOf(bb) === 0) {\n          return compare2([aa, ab.substring(bb.length), an, ap], [ba, '', bn, bp]);\n        }\n        if (bb.indexOf(ab) === 0) {\n          return compare2([aa, '', an, ap], [ba, bb.substring(ab.length), bn, bp]);\n        }\n      }\n  \n      // Try the same without $p:\n      if (localXor(ap, bp)) {\n        const result = compare2([aa, ab, an, ''], [ba, bb, bn, '']);\n        return result > 0.0 ? result * 0.8 : result;\n      }\n    \n      if (aa === ba && an === bn) {\n        // No $n: Don't score $a vs $ab. Return 0, and let other match detectors decide\n        const candScore = TITLE_MAX/2;\n        if (ap === bp && localXor(ab, bb)) {\n          debug(`Handle omitted \\$b using x ${candScore}`);\n          return candScore;\n        }\n        if (ab === bb && localXor(ap, bp)) {\n          debug(`Handle omitted \\$p using x ${candScore}`);\n          return candScore;\n        }\n      }\n\n      return -0.5; // Not likely\n    }\n\n    function isEmpty(x) {\n      return !x || x.length === 0;\n    }\n\n    function localXor(x, y) {\n      if (isEmpty(x)) {\n        return !isEmpty(y);\n      }\n      // 'x' exists, thus 'y' can not exist:\n      return isEmpty(y);\n    }\n\n    function doLevenshtein(string1, string2) {\n      const distance = leven(string1, string2);\n      const len = getMaxLength(string1, string2);\n      const correctness = 1.0 - (distance / len);\n      return [distance, len, correctness];\n    }\n\n    function toFullTitle(arr) {\n      const relevant = arr.filter(val => typeof val === 'string' && val.length);\n      // No longer add space between subfields. Due to heavy normalization there are no spaces left in array elements either!\n      return relevant.join('');\n    }\n\n    function getMaxLength(str1, str2) {\n      return str1.length > str2.length ? str1.length : str2.length;\n    }\n\n  }\n});\n\nexport function getTitle(record, targetSubfieldCodes = ['a', 'b', 'n', 'p']) {\n  const [field] = record.get(/^245$/u);\n  debugData(`titleField: ${JSON.stringify(field)}`);\n\n  if (field) {\n    const title = field.subfields\n      // get also $n:s and $p:s here\n      .filter(({code}) => targetSubfieldCodes.includes(code))\n      // Would be nice to normalize $n values...\n      .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n      .join(' ')\n      .replace(/\\[[^\\]]*\\]/ug, ' ') // Remove [whatever] stuff. ADD TEST FOR THIS\n      .replace(/ [=\\/:](?:$| )/ug, ' ') // Strip punctuation\n      // Also \u0111 vs d pairs seen:\n      //.replace(/(?:\u0111|\u0227|t\u0304|k\u030C|\u01E7|s\u0306|c\u0306)/ug, '') // Hack. Saamelaisbibliografia has often dropped this wierd characters (oft old articles, no longer used in sami either)\n      // trim:\n      .replace(/ +/ug, ' ')\n      .replace(/^ +/u, '')\n      .replace(/ +$/u, '');\n\n    // Skip non-filing indicator (note that '9' is a magic indicator value, so we don't do it):\n    if (/^[1-8]$/u.test(field.ind2)) { // Skip non-filing characters\n      return title.slice(parseInt(field.ind2));\n    }\n    return title;\n\n  }\n  return false;\n}"],
-  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,OAAO,gBAAgB;AACvB,MAAM,EAAC,qBAAqB,MAAK,IAAI;AACrC,SAAQ,0BAAyB;AACjC,SAAS,wBAAwB;AAGjC,MAAM,QAAQ,kBAAkB,sEAAsE;AACtG,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,MAAM,YAAY;AAClB,eAAe,CAAC,EAAC,YAAY,IAAG,IAAI,CAAC,OAAO;AAAA,EAC1C,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,QAAQ,eAAc,MAAM;AACrC,UAAM,QAAQ,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAC9E,UAAM,IAAI,SAAS,QAAQ,CAAC,GAAG,CAAC;AAChC,UAAM,GAAG,KAAK,YAAY,CAAC,EAAE;AAE7B,QAAI,GAAG;AACL,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,aAAO,CAAC,eAAe,CAAC,GAAG,GAAG,GAAG,CAAC;AAAA,IACpC;AAEA,WAAO,CAAC,IAAI,IAAI,IAAI,EAAE;AAEtB,aAAS,eAAe,OAAO;AAC7B,aAAO,MAEJ,UAAU,KAAK,EACf,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,wCAAwC,GAAG,EACnD,QAAQ,4CAA4C,GAAG,EACvD,QAAQ,8CAA8C,GAAG,EACzD,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,sCAAsC,GAAG,EACjD,QAAQ,2BAA2B,GAAG,EAKtC,QAAQ,8BAA8B,EAAE,EACxC,YAAY;AAAA,IACjB;AAAA,EAEF;AAAA,EACA,SAAS,CAAC,OAAO,UAAU;AACzB,WAAO,SAAS,OAAO,KAAK;AAE5B,aAAS,SAAS,GAAG,GAAG;AAEtB,YAAM,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI;AACzB,YAAM,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI;AAEzB,YAAM,QAAQ,YAAY,CAAC;AAC3B,YAAM,QAAQ,YAAY,CAAC;AAE3B,YAAM;AAAA,KAAgB,KAAK;AAAA,KAAc,KAAK,GAAG;AAEjD,UAAI,QAAQ,EAAE,KAAK,QAAQ,EAAE,GAAG;AAC9B,eAAO;AAAA,MACT;AAEA,UAAI,UAAU,OAAO;AACnB,eAAO;AAAA,MACT;AAGA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AAEA,UAAI,CAAC,QAAQ,EAAE,KAAK,OAAO,IAAI;AAC7B,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,CAAC,QAAQ,EAAE,KAAK,OAAO,IAAI;AAC7B,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AAEA,UAAI,MAAM,MAAM,GAAG,MAAM,QAAQ,GAAG;AAClC,cAAM,yBAAyB,EAAE,QAAQ,EAAE,EAAE;AAC7C,cAAM,OAAO,GAAG,QAAQ,YAAY,EAAE;AACtC,cAAM,OAAO,GAAG,QAAQ,YAAY,EAAE;AACtC,YAAI,SAAS,MAAM;AACjB,gBAAM,cAAe,EAAE,SAAS,EAAE,GAAG;AACrC,iBAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,QACpD;AAAA,MACF;AAIA,UAAI,MAAM,OAAO,IAAI;AACnB,eAAO;AAAA,MACT;AAEA,YAAM,CAAC,UAAU,WAAW,WAAW,IAAI,cAAc,OAAO,KAAK;AAErE,YAAM,IAAI,KAAK,SAAS,KAAK,mBAAmB,SAAS,gBAAgB,QAAQ,mBAAmB,WAAW,EAAE;AAGjH,UAAI,eAAe,WAAW;AAC5B,eAAO,YAAY;AAAA,MACrB;AAGA,UAAI,MAAM,IAAI;AAEZ,YAAI,GAAG,QAAQ,EAAE,MAAM,GAAG;AACxB,iBAAO,SAAS,CAAC,IAAI,GAAG,UAAU,GAAG,MAAM,GAAG,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,QACzE;AACA,YAAI,GAAG,QAAQ,EAAE,MAAM,GAAG;AACxB,iBAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,GAAG,UAAU,GAAG,MAAM,GAAG,IAAI,EAAE,CAAC;AAAA,QACzE;AAAA,MACF;AAGA,UAAI,SAAS,IAAI,EAAE,GAAG;AACpB,cAAM,SAAS,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAC1D,eAAO,SAAS,IAAM,SAAS,MAAM;AAAA,MACvC;AAEA,UAAI,OAAO,MAAM,OAAO,IAAI;AAE1B,cAAM,YAAY,YAAU;AAC5B,YAAI,OAAO,MAAM,SAAS,IAAI,EAAE,GAAG;AACjC,gBAAM,6BAA8B,SAAS,EAAE;AAC/C,iBAAO;AAAA,QACT;AACA,YAAI,OAAO,MAAM,SAAS,IAAI,EAAE,GAAG;AACjC,gBAAM,6BAA8B,SAAS,EAAE;AAC/C,iBAAO;AAAA,QACT;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAEA,aAAS,QAAQ,GAAG;AAClB,aAAO,CAAC,KAAK,EAAE,WAAW;AAAA,IAC5B;AAEA,aAAS,SAAS,GAAG,GAAG;AACtB,UAAI,QAAQ,CAAC,GAAG;AACd,eAAO,CAAC,QAAQ,CAAC;AAAA,MACnB;AAEA,aAAO,QAAQ,CAAC;AAAA,IAClB;AAEA,aAAS,cAAc,SAAS,SAAS;AACvC,YAAM,WAAW,MAAM,SAAS,OAAO;AACvC,YAAM,MAAM,aAAa,SAAS,OAAO;AACzC,YAAM,cAAc,IAAO,WAAW;AACtC,aAAO,CAAC,UAAU,KAAK,WAAW;AAAA,IACpC;AAEA,aAAS,YAAY,KAAK;AACxB,YAAM,WAAW,IAAI,OAAO,SAAO,OAAO,QAAQ,YAAY,IAAI,MAAM;AAExE,aAAO,SAAS,KAAK,EAAE;AAAA,IACzB;AAEA,aAAS,aAAa,MAAM,MAAM;AAChC,aAAO,KAAK,SAAS,KAAK,SAAS,KAAK,SAAS,KAAK;AAAA,IACxD;AAAA,EAEF;AACF;AAEO,gBAAS,SAAS,QAAQ,sBAAsB,CAAC,KAAK,KAAK,KAAK,GAAG,GAAG;AAC3E,QAAM,CAAC,KAAK,IAAI,OAAO,IAAI,QAAQ;AACnC,YAAU,eAAe,KAAK,UAAU,KAAK,CAAC,EAAE;AAEhD,MAAI,OAAO;AACT,UAAM,QAAQ,MAAM,UAEjB,OAAO,CAAC,EAAC,KAAI,MAAM,oBAAoB,SAAS,IAAI,CAAC,EAErD,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,KAAK,GAAG,EACR,QAAQ,gBAAgB,GAAG,EAC3B,QAAQ,oBAAoB,GAAG,EAI/B,QAAQ,QAAQ,GAAG,EACnB,QAAQ,QAAQ,EAAE,EAClB,QAAQ,QAAQ,EAAE;AAGrB,QAAI,WAAW,KAAK,MAAM,IAAI,GAAG;AAC/B,aAAO,MAAM,MAAM,SAAS,MAAM,IAAI,CAAC;AAAA,IACzC;AACA,WAAO;AAAA,EAET;AACA,SAAO;AACT;",
-  "names": []
+  "sourcesContent": ["import createDebugLogger from 'debug';\nimport naturalPkg from 'natural';\nconst {LevenshteinDistance: leven} = naturalPkg;\nimport {testStringOrNumber} from '../../../matching-utils.js';\nimport { subfieldToString } from '@natlibfi/marc-record-validators-melinda/dist/utils.js';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\nconst debugData = debug.extend('data');\n\nconst TITLE_MAX = 0.5;\nexport default ({threshold = 0.9} = {}) => ({\n  name: 'Title',\n  extract: ({record, recordExternal}) => {\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n    const a = getTitle(record, ['a']);\n    debug(`${label}: title: ${a}`);\n\n    if (a) {\n      const b = normalizeTitle(getTitle(record, ['b'])) || '';\n      const n = normalizeTitle(getTitle(record, ['n'])) || '';\n      const p = normalizeTitle(getTitle(record, ['p'])) || '';\n      return [normalizeTitle(a), b, n, p];\n    }\n\n    return ['', '', '', ''];\n\n    function normalizeTitle(title) {\n      return title\n        // decompose unicode diacritics\n        .normalize('NFD')\n        .replace(/\\b(?:ein|ett|I|one|uno|yksi)\\b/uig, '1')\n        .replace(/\\b(?:dos|kaksi|II|tv\u00E5|two|zwei)\\b/uig, '2')\n        .replace(/\\b(?:drei|III|kolme|three|tre|tres)\\b/uig, '3')\n        .replace(/\\b(?:four|fyra|IV|nelj\u00E4|quat+ro|vier)\\b/uig, '4')\n        .replace(/\\b(?:five|f\u00FCnf|fyra|viisi|V)\\b/uig, '5')\n        .replace(/\\b(?:kuusi|sechts|sex|six|VI)\\b/uig, '6')\n        .replace(/\\b(and|ja|och|und)\\b/uig, '&')\n        // strip non-letters/numbers\n        // - note: combined with decomposing unicode diacritics this normalizes both 'saa' and 's\u00E4\u00E4' as 'saa'\n        // - we could precompose the Finnish letters back to avoid this\n        // - see validator normalize-utf8-diacritics for details\n        .replace(/[^\\p{Letter}\\p{Number}& ]/gu, '') // 2026-07-01: keep ' '. We need it for splitToBaseAndNumber()\n        .replace(/\\s+/ug, ' ')\n        .replace(/ $/u, '')\n        .replace(/^ /u, '')\n        .toLowerCase();\n    }\n\n  },\n  compare: (origA, origB) => {\n    return compare2(origA, origB);\n\n    function compare2(a, b) { // Added this wrapper function as I had issues with recursion\n\n      const [aa, ab, an, ap] = a;\n      const [ba, bb, bn, bp] = b;\n      debugData(`COMPARE:\\n['${aa}', '${ab}', '${an}', '${ap}'] VS\\n['${ba}', '${bb}', '${bn}', '${bp}']`);\n\n      const aFull = toFullTitle(a);\n      const bFull = toFullTitle(b);\n\n      // debugData(`JOINED COMPARE:\\n  '${aFull}' vs  \\n  '${bFull}'`);\n\n      if (isEmpty(aa) || isEmpty(ba)) {\n        return -1.0;\n      }\n\n      if (aFull === bFull) {\n        return TITLE_MAX;\n      }\n\n      // MELKEHITYS-3496-ish: move part of $a to $n:\n      const [altAa, altAn, altBa, altBn] = reconstructTitleAndNumber(aa, an, ba, bn);\n      if (altAa) {\n        return compare2([altAa, ab, altAn, ap], [altBa, bb, altBn, bp]);\n      }\n\n      if (ab && ab !== '' && ab === bb) { // Remove $b from equation (MELKEHITYS-3494)\n        debug(`Ignore \\$b ${ab}`);\n        return compare2([aa, '', an, ap], [ba, '', bn, bp]);\n      }\n      if (an && an !== '' && an === bn) { // Remove $n from equation\n        debug(`Ignore \\$n ${an}`);\n        return compare2([aa, ab, '', ap], [ba, bb, '', bp]);\n      }\n      if (ap && ap !== '' && ap === bp) { // Remove $p from equation\n        debug(`Ignore \\$p ${ap}`);\n        return compare2([aa, ab, an, ''], [ba, bb, bn, '']);\n      }\n      // There seems to be wobble between $b and $p:\n      if (!isEmpty(ab) && ab === bp) {\n        return compare2([aa, '', an, ap], [ba, bb, bn, '']);\n      }\n      if (!isEmpty(ap) && ap === bp) {\n        return compare2([aa, ab, an, ''], [ba, '', bn, bp]);\n      }\n\n      if (an && bn && an.match(/[0-9]/u)) {\n        debug(`NUMBER FOUND. Compare ${an} and ${bn}`);\n        const atmp = an.replace(/[^0-9]/ug, '');\n        const btmp = bn.replace(/[^0-9]/ug, '');\n        if (atmp === btmp) {\n          debug(`Ignore \\$n '${an}' vs '${bn}'`);\n          return compare2([aa, ab, '', ap], [ba, bb, '', bp]);\n        }\n      }\n\n\n      // f245$n information is critical; it can not mismatch at all  (exceptions have already been handled above):\n      if (an && an !== bn) {\n        return -1.0;\n      }\n\n      const [distance, maxLength, correctness] = doLevenshtein(aFull.replace(/ /ug, ''), bFull.replace(/ /ug, ''));\n\n      debug(`'${aFull}' vs '${bFull}': Max length = ${maxLength}, distance = ${distance}, correctness = ${correctness}`);\n\n\n      if (correctness >= threshold) {\n        return TITLE_MAX * 0.8;\n      }\n\n      // Subset removal (MELKEHITYS-3498-ish)\n      if (ab && bb) { // At this point ab and bb are never equal...\n        // If X is a subset of Y, then remove X and shorten Y (= remove the shared content)\n        if (ab.indexOf(bb) === 0) {\n          return compare2([aa, ab.substring(bb.length), an, ap], [ba, '', bn, bp]);\n        }\n        if (bb.indexOf(ab) === 0) {\n          return compare2([aa, '', an, ap], [ba, bb.substring(ab.length), bn, bp]);\n        }\n      }\n\n      // Try the same without $p:\n      if (localXor(ap, bp)) {\n        const result = compare2([aa, ab, an, ''], [ba, bb, bn, '']);\n        return result > 0.0 ? result * 0.8 : result;\n      }\n\n      if (aa === ba && an === bn) {\n        const candScore = TITLE_MAX/2;\n        // $b is missing from one:\n        if (ap === bp && localXor(ab, bb)) {\n          debug(`Handle omitted \\$b using x ${candScore}`);\n          return candScore;\n        }\n        if (ab === bb && localXor(ap, bp)) {\n          debug(`Handle omitted \\$p using x ${candScore}`);\n          return candScore;\n        }\n      }\n\n      if (an === bn && ap === bp ) {\n        // $b-less $a is other record's $b, see MELKEHITYS-3485 for motivation (though we skip 246 and 490 here):\n        const candScore = TITLE_MAX/2;\n        if (aa === bb && !ab) {\n          return candScore;\n        }\n        if (ba === ab && !bb) {\n          return candScore;\n        }\n      }\n\n      return -0.5; // Not likely\n    }\n\n    function isEmpty(x) {\n      return !x || x.length === 0;\n    }\n\n    function localXor(x, y) {\n      if (isEmpty(x)) {\n        return !isEmpty(y);\n      }\n      // 'x' exists, thus 'y' can not exist:\n      return isEmpty(y);\n    }\n\n    function doLevenshtein(string1, string2) {\n      const distance = leven(string1, string2);\n      const len = getMaxLength(string1, string2);\n      const correctness = 1.0 - (distance / len);\n      return [distance, len, correctness];\n    }\n\n    function toFullTitle(arr) {\n      const relevant = arr.filter(val => typeof val === 'string' && val.length);\n      // No longer add space between subfields. Due to heavy normalization there are no spaces left in array elements either!\n      return relevant.join('');\n    }\n\n    function getMaxLength(str1, str2) {\n      return str1.length > str2.length ? str1.length : str2.length;\n    }\n\n    function reconstructTitleAndNumber(a1, n1, a2, n2) {\n      if (!n1 && n2) {\n        const [altA1, altN1] = splitToBaseAndNumber(a1);\n        if (altN1) {\n          debugData(`A: Reconstruct:\\n $a '${a1}' =>\\n $a '${altA1}' +\\n $n '${altN1}`);\n          return [altA1, altN1, a2, n2];\n        }\n      }\n      if (n1 && !n2) {\n        const [altA2, altN2] = splitToBaseAndNumber(a2);\n        if (altN2) {\n          debugData(`B: Reconstruct $a '${a2}' as $a '${altA2}' and $n '${altN2}`);\n          return [a1, n1, altA2, altN2];\n        }\n\n      }\n      return [null, null, null, null]; // Fail\n    }\n\n\n    function splitToBaseAndNumber(val) {\n      const words = val.split(' ');\n      const len = words.length;\n      if (len < 2 || !words[len-1].match(/^[1-9][0-9]*$/)) {\n        return [val, null];\n      }\n      if (len > 2 && words[len-2].match(/^(del|osa|vol|volume)$/u)) { // NB! \"vol.\" has lost it's '.' during normalization!\n        const number = `${words[len-2]} ${words[len-1]}`;\n        return [words.slice(0, len-2).join(' '), number];\n      }\n\n      const number = words.pop();\n      return [words.join(' '), number];\n    }\n\n  }\n});\n\nexport function getTitle(record, targetSubfieldCodes = ['a', 'b', 'n', 'p']) {\n  const [field] = record.get(/^245$/u);\n  debugData(`titleField: ${JSON.stringify(field)}`);\n\n  if (field) {\n    const title = field.subfields\n      // get also $n:s and $p:s here\n      .filter(({code}) => targetSubfieldCodes.includes(code))\n      // Would be nice to normalize $n values...\n      .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n      .join(' ')\n      .replace(/\\[[^\\]]*\\]/ug, ' ') // Remove [whatever] stuff. ADD TEST FOR THIS\n      .replace(/ [=\\/:](?:$| )/ug, ' ') // Strip punctuation\n      // Also \u0111 vs d pairs seen:\n      //.replace(/(?:\u0111|\u0227|t\u0304|k\u030C|\u01E7|s\u0306|c\u0306)/ug, '') // Hack. Saamelaisbibliografia has often dropped this wierd characters (oft old articles, no longer used in sami either)\n      // trim:\n      .replace(/ +/ug, ' ')\n      .replace(/^ +/u, '')\n      .replace(/ +$/u, '');\n\n    // Skip non-filing indicator (note that '9' is a magic indicator value, so we don't do it):\n    if (/^[1-8]$/u.test(field.ind2)) { // Skip non-filing characters\n      return title.slice(parseInt(field.ind2));\n    }\n    return title;\n\n  }\n  return false;\n}"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,OAAO,gBAAgB;AACvB,MAAM,EAAC,qBAAqB,MAAK,IAAI;AACrC,SAAQ,0BAAyB;AACjC,SAAS,wBAAwB;AAGjC,MAAM,QAAQ,kBAAkB,sEAAsE;AACtG,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,MAAM,YAAY;AAClB,eAAe,CAAC,EAAC,YAAY,IAAG,IAAI,CAAC,OAAO;AAAA,EAC1C,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,QAAQ,eAAc,MAAM;AACrC,UAAM,QAAQ,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAC9E,UAAM,IAAI,SAAS,QAAQ,CAAC,GAAG,CAAC;AAChC,UAAM,GAAG,KAAK,YAAY,CAAC,EAAE;AAE7B,QAAI,GAAG;AACL,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,YAAM,IAAI,eAAe,SAAS,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK;AACrD,aAAO,CAAC,eAAe,CAAC,GAAG,GAAG,GAAG,CAAC;AAAA,IACpC;AAEA,WAAO,CAAC,IAAI,IAAI,IAAI,EAAE;AAEtB,aAAS,eAAe,OAAO;AAC7B,aAAO,MAEJ,UAAU,KAAK,EACf,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,wCAAwC,GAAG,EACnD,QAAQ,4CAA4C,GAAG,EACvD,QAAQ,8CAA8C,GAAG,EACzD,QAAQ,qCAAqC,GAAG,EAChD,QAAQ,sCAAsC,GAAG,EACjD,QAAQ,2BAA2B,GAAG,EAKtC,QAAQ,+BAA+B,EAAE,EACzC,QAAQ,SAAS,GAAG,EACpB,QAAQ,OAAO,EAAE,EACjB,QAAQ,OAAO,EAAE,EACjB,YAAY;AAAA,IACjB;AAAA,EAEF;AAAA,EACA,SAAS,CAAC,OAAO,UAAU;AACzB,WAAO,SAAS,OAAO,KAAK;AAE5B,aAAS,SAAS,GAAG,GAAG;AAEtB,YAAM,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI;AACzB,YAAM,CAAC,IAAI,IAAI,IAAI,EAAE,IAAI;AACzB,gBAAU;AAAA,IAAe,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE;AAAA,IAAY,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI;AAEnG,YAAM,QAAQ,YAAY,CAAC;AAC3B,YAAM,QAAQ,YAAY,CAAC;AAI3B,UAAI,QAAQ,EAAE,KAAK,QAAQ,EAAE,GAAG;AAC9B,eAAO;AAAA,MACT;AAEA,UAAI,UAAU,OAAO;AACnB,eAAO;AAAA,MACT;AAGA,YAAM,CAAC,OAAO,OAAO,OAAO,KAAK,IAAI,0BAA0B,IAAI,IAAI,IAAI,EAAE;AAC7E,UAAI,OAAO;AACT,eAAO,SAAS,CAAC,OAAO,IAAI,OAAO,EAAE,GAAG,CAAC,OAAO,IAAI,OAAO,EAAE,CAAC;AAAA,MAChE;AAEA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,MAAM,OAAO,MAAM,OAAO,IAAI;AAChC,cAAM,aAAc,EAAE,EAAE;AACxB,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AAEA,UAAI,CAAC,QAAQ,EAAE,KAAK,OAAO,IAAI;AAC7B,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AACA,UAAI,CAAC,QAAQ,EAAE,KAAK,OAAO,IAAI;AAC7B,eAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,MACpD;AAEA,UAAI,MAAM,MAAM,GAAG,MAAM,QAAQ,GAAG;AAClC,cAAM,yBAAyB,EAAE,QAAQ,EAAE,EAAE;AAC7C,cAAM,OAAO,GAAG,QAAQ,YAAY,EAAE;AACtC,cAAM,OAAO,GAAG,QAAQ,YAAY,EAAE;AACtC,YAAI,SAAS,MAAM;AACjB,gBAAM,cAAe,EAAE,SAAS,EAAE,GAAG;AACrC,iBAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,QACpD;AAAA,MACF;AAIA,UAAI,MAAM,OAAO,IAAI;AACnB,eAAO;AAAA,MACT;AAEA,YAAM,CAAC,UAAU,WAAW,WAAW,IAAI,cAAc,MAAM,QAAQ,OAAO,EAAE,GAAG,MAAM,QAAQ,OAAO,EAAE,CAAC;AAE3G,YAAM,IAAI,KAAK,SAAS,KAAK,mBAAmB,SAAS,gBAAgB,QAAQ,mBAAmB,WAAW,EAAE;AAGjH,UAAI,eAAe,WAAW;AAC5B,eAAO,YAAY;AAAA,MACrB;AAGA,UAAI,MAAM,IAAI;AAEZ,YAAI,GAAG,QAAQ,EAAE,MAAM,GAAG;AACxB,iBAAO,SAAS,CAAC,IAAI,GAAG,UAAU,GAAG,MAAM,GAAG,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAAA,QACzE;AACA,YAAI,GAAG,QAAQ,EAAE,MAAM,GAAG;AACxB,iBAAO,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,GAAG,UAAU,GAAG,MAAM,GAAG,IAAI,EAAE,CAAC;AAAA,QACzE;AAAA,MACF;AAGA,UAAI,SAAS,IAAI,EAAE,GAAG;AACpB,cAAM,SAAS,SAAS,CAAC,IAAI,IAAI,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,IAAI,EAAE,CAAC;AAC1D,eAAO,SAAS,IAAM,SAAS,MAAM;AAAA,MACvC;AAEA,UAAI,OAAO,MAAM,OAAO,IAAI;AAC1B,cAAM,YAAY,YAAU;AAE5B,YAAI,OAAO,MAAM,SAAS,IAAI,EAAE,GAAG;AACjC,gBAAM,6BAA8B,SAAS,EAAE;AAC/C,iBAAO;AAAA,QACT;AACA,YAAI,OAAO,MAAM,SAAS,IAAI,EAAE,GAAG;AACjC,gBAAM,6BAA8B,SAAS,EAAE;AAC/C,iBAAO;AAAA,QACT;AAAA,MACF;AAEA,UAAI,OAAO,MAAM,OAAO,IAAK;AAE3B,cAAM,YAAY,YAAU;AAC5B,YAAI,OAAO,MAAM,CAAC,IAAI;AACpB,iBAAO;AAAA,QACT;AACA,YAAI,OAAO,MAAM,CAAC,IAAI;AACpB,iBAAO;AAAA,QACT;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAEA,aAAS,QAAQ,GAAG;AAClB,aAAO,CAAC,KAAK,EAAE,WAAW;AAAA,IAC5B;AAEA,aAAS,SAAS,GAAG,GAAG;AACtB,UAAI,QAAQ,CAAC,GAAG;AACd,eAAO,CAAC,QAAQ,CAAC;AAAA,MACnB;AAEA,aAAO,QAAQ,CAAC;AAAA,IAClB;AAEA,aAAS,cAAc,SAAS,SAAS;AACvC,YAAM,WAAW,MAAM,SAAS,OAAO;AACvC,YAAM,MAAM,aAAa,SAAS,OAAO;AACzC,YAAM,cAAc,IAAO,WAAW;AACtC,aAAO,CAAC,UAAU,KAAK,WAAW;AAAA,IACpC;AAEA,aAAS,YAAY,KAAK;AACxB,YAAM,WAAW,IAAI,OAAO,SAAO,OAAO,QAAQ,YAAY,IAAI,MAAM;AAExE,aAAO,SAAS,KAAK,EAAE;AAAA,IACzB;AAEA,aAAS,aAAa,MAAM,MAAM;AAChC,aAAO,KAAK,SAAS,KAAK,SAAS,KAAK,SAAS,KAAK;AAAA,IACxD;AAEA,aAAS,0BAA0B,IAAI,IAAI,IAAI,IAAI;AACjD,UAAI,CAAC,MAAM,IAAI;AACb,cAAM,CAAC,OAAO,KAAK,IAAI,qBAAqB,EAAE;AAC9C,YAAI,OAAO;AACT,oBAAU;AAAA,OAAyB,EAAE;AAAA,OAAc,KAAK;AAAA,OAAa,KAAK,EAAE;AAC5E,iBAAO,CAAC,OAAO,OAAO,IAAI,EAAE;AAAA,QAC9B;AAAA,MACF;AACA,UAAI,MAAM,CAAC,IAAI;AACb,cAAM,CAAC,OAAO,KAAK,IAAI,qBAAqB,EAAE;AAC9C,YAAI,OAAO;AACT,oBAAU,sBAAsB,EAAE,YAAY,KAAK,aAAa,KAAK,EAAE;AACvE,iBAAO,CAAC,IAAI,IAAI,OAAO,KAAK;AAAA,QAC9B;AAAA,MAEF;AACA,aAAO,CAAC,MAAM,MAAM,MAAM,IAAI;AAAA,IAChC;AAGA,aAAS,qBAAqB,KAAK;AACjC,YAAM,QAAQ,IAAI,MAAM,GAAG;AAC3B,YAAM,MAAM,MAAM;AAClB,UAAI,MAAM,KAAK,CAAC,MAAM,MAAI,CAAC,EAAE,MAAM,eAAe,GAAG;AACnD,eAAO,CAAC,KAAK,IAAI;AAAA,MACnB;AACA,UAAI,MAAM,KAAK,MAAM,MAAI,CAAC,EAAE,MAAM,yBAAyB,GAAG;AAC5D,cAAMA,UAAS,GAAG,MAAM,MAAI,CAAC,CAAC,IAAI,MAAM,MAAI,CAAC,CAAC;AAC9C,eAAO,CAAC,MAAM,MAAM,GAAG,MAAI,CAAC,EAAE,KAAK,GAAG,GAAGA,OAAM;AAAA,MACjD;AAEA,YAAM,SAAS,MAAM,IAAI;AACzB,aAAO,CAAC,MAAM,KAAK,GAAG,GAAG,MAAM;AAAA,IACjC;AAAA,EAEF;AACF;AAEO,gBAAS,SAAS,QAAQ,sBAAsB,CAAC,KAAK,KAAK,KAAK,GAAG,GAAG;AAC3E,QAAM,CAAC,KAAK,IAAI,OAAO,IAAI,QAAQ;AACnC,YAAU,eAAe,KAAK,UAAU,KAAK,CAAC,EAAE;AAEhD,MAAI,OAAO;AACT,UAAM,QAAQ,MAAM,UAEjB,OAAO,CAAC,EAAC,KAAI,MAAM,oBAAoB,SAAS,IAAI,CAAC,EAErD,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,KAAK,GAAG,EACR,QAAQ,gBAAgB,GAAG,EAC3B,QAAQ,oBAAoB,GAAG,EAI/B,QAAQ,QAAQ,GAAG,EACnB,QAAQ,QAAQ,EAAE,EAClB,QAAQ,QAAQ,EAAE;AAGrB,QAAI,WAAW,KAAK,MAAM,IAAI,GAAG;AAC/B,aAAO,MAAM,MAAM,SAAS,MAAM,IAAI,CAAC;AAAA,IACzC;AACA,WAAO;AAAA,EAET;AACA,SAAO;AACT;",
+  "names": ["number"]
 }

package/dist/matching-utils.js

@@ -90,4 +90,43 @@ export function stringBefore(string, substring) {
   }
   return string.substring(0, pos);
 }
+export function preferenceSortRecords(x, y) {
+  debug(`sortRecords: RECORD COMPARISON
+${JSON.stringify(x)} vs
+${JSON.stringify(y)}`);
+  const xFikka = hasLow(x.record, "FIKKA");
+  const yFikka = hasLow(y.record, "FIKKA");
+  if (xFikka && !yFikka) {
+    debug("sortRecords: prefer FIKKA");
+    return -1;
+  }
+  if (yFikka && !xFikka) {
+    return 1;
+  }
+  const xAllCaps = hasAllCapsTitle(x.record);
+  const yAllCaps = hasAllCapsTitle(y.record);
+  if (!xAllCaps && yAllCaps) {
+    debug("sortRecords: prefer non-all caps");
+    return -1;
+  }
+  if (!yAllCaps && xAllCaps) {
+    return 1;
+  }
+  return 0;
+  function hasLow(record, lowTag) {
+    return record.fields?.some((f) => f.tag === "LOW" && f.subfields?.some((sf) => sf.code === "a" && sf.value === lowTag));
+  }
+  function hasAllCapsTitle(record) {
+    const f245 = record.fields?.find((f) => f.tag === "245");
+    if (!f245) {
+      debugData(`WARNING: No f245 found for ALL CAPS check`);
+      return false;
+    }
+    const title = f245.subfields.filter((sf) => ["a", "b", "c"].includes(sf.code)).map((sf) => sf.value).join(" ");
+    if (title.match(/[A-Z]/u) && !title.match(/[a-z]/u)) {
+      return true;
+    }
+    return false;
+  }
+}
 //# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../src/matching-utils.js"],
-  "sourcesContent": ["import createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(field => toMelindaIds(field)));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n}\n\nexport function toMelindaIds({subfields}, subfieldsToParse = ['a', 'z']) {\n  const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n  return subfields\n    .filter(sub => subfieldsToParse.includes(sub.code))\n    .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n    .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n  const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n  return {\n    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n    matchingValues\n  };\n\n  function getMatchingValuesAmount(aValues, bValues) {\n    if (bValues.length > aValues.length) {\n      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    }\n    if (aValues.length > bValues.length) {\n      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n    }\n\n    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n    // there would be duplicate values\n    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n    return aToB < bToA ? aToB : bToA;\n  }\n}\n\n\nexport function stringAfter(string, substring) {\n  const pos = string.indexOf(substring);\n  if (pos === -1) {\n    return string;\n  }\n  return string.substring(pos+substring.length);\n}\n\nexport function stringBefore(string, substring) {\n  const pos = string.indexOf(substring);\n  if (pos === -1) {\n    return string;\n  }\n  return string.substring(0, pos);\n}\n"],
-  "mappings": "AAAA,OAAO,uBAAuB;AAE9B,MAAM,QAAQ,kBAAkB,yCAAyC;AACzE,MAAM,YAAY,MAAM,OAAO,MAAM;AAE9B,gBAAS,kBAAkB,QAAQ;AAExC,QAAMA,SAAQ,kBAAkB,8CAA8C;AAC9E,QAAMC,aAAYD,OAAM,OAAO,MAAM;AAErC,QAAM,QAAQ,OAAO,UAAU,KAAK;AAEpC,MAAI,MAAM,SAAS,GAAG;AACpB,IAAAA,OAAM,iBAAiB;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,oBAAoB,CAAC,EAAE,OAAO,GAAG,MAAM,IAAI,WAAS,aAAa,KAAK,CAAC,CAAC;AAC9E,QAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,iBAAiB,CAAC;AAErD,EAAAC,WAAU,WAAW,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAC9D,EAAAA,WAAU,QAAQ,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AACnF,EAAAA,WAAU,eAAe,eAAe,MAAM,MAAM,KAAK,UAAU,cAAc,CAAC,EAAE;AAEpF,SAAO;AACT;AAEO,gBAAS,aAAa,EAAC,UAAS,GAAG,mBAAmB,CAAC,KAAK,GAAG,GAAG;AACvE,QAAM,kBAAkB;AAExB,SAAO,UACJ,OAAO,SAAO,iBAAiB,SAAS,IAAI,IAAI,CAAC,EACjD,OAAO,SAAO,mBAAmB,IAAI,KAAK,KAAK,gBAAgB,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtF,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,EAAE,QAAQ,iBAAiB,OAAO,IAAI,EAAE;AACtG;AAEO,gBAAS,+BAA+B,OAAO;AAEpD,YAAU,wBAAwB,KAAK,UAAU,KAAK,CAAC,EAAE;AACzD,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,MAAM,eAAe,MAAM,qBAAqB,WAAW,KAAK,WAAW,CAAC,EAAE;AAE7F,SAAO,WAAW,KAAK,WAAW;AACpC;AAEA,SAAS,eAAe,OAAO,cAAc;AAE3C,SAAO,MAAM,UAAU,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EAAE;AACnE;AAEO,gBAAS,kBAAkB,OAAO,cAAc;AACrD,YAAU,oBAAoB,YAAY,SAAS,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1E,SAAO,MAAM,UACV,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EACxC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK;AAC1B;AAEO,gBAAS,mBAAmB,OAAO;AACxC,MAAI,OAAO,UAAU,YAAY,OAAO,UAAU,UAAU;AAC1D,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAEO,gBAAS,0BAA0B,OAAO,eAAe;AAC9D,MAAI,UAAU,UAAa,MAAM,cAAc,QAAW;AACxD,WAAO,CAAC;AAAA,EACV;AACA,QAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,cAAc,SAAS,IAAI,CAAC,EAC/C,IAAI,CAAC,EAAC,MAAM,MAAK,OAAO,EAAC,MAAM,OAAO,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,GAAE,EAAE,EACtF,OAAO,WAAS,KAAK;AACxB,SAAO;AACT;AAEO,gBAAS,gBAAgB,WAAW;AACzC,SAAO,UAAU,OAAO,CAAC,KAAK,MAAM;AAClC,QAAI,CAAC,IAAI,KAAK,UAAQ,KAAK,SAAS,EAAE,QAAQ,KAAK,UAAU,EAAE,KAAK,GAAG;AACrE,YAAM,SAAS,IAAI,OAAO,CAAC;AAC3B,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT,GAAG,CAAC,CAAC;AACP;AAEO,gBAAS,eAAe,SAAS,SAAS;AAE/C,QAAM,iBAAiB,wBAAwB,SAAS,OAAO;AAE/D,SAAO;AAAA,IACL,WAAW,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA;AAAA,IAEtE,qBAAqB,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA,IAChF;AAAA,EACF;AAEA,WAAS,wBAAwBC,UAASC,UAAS;AACjD,QAAIA,SAAQ,SAASD,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AACA,QAAID,SAAQ,SAASC,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AAIA,UAAM,OAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AACjF,UAAM,OAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAEjF,WAAO,OAAO,OAAO,OAAO;AAAA,EAC9B;AACF;AAGO,gBAAS,YAAY,QAAQ,WAAW;AAC7C,QAAM,MAAM,OAAO,QAAQ,SAAS;AACpC,MAAI,QAAQ,IAAI;AACd,WAAO;AAAA,EACT;AACA,SAAO,OAAO,UAAU,MAAI,UAAU,MAAM;AAC9C;AAEO,gBAAS,aAAa,QAAQ,WAAW;AAC9C,QAAM,MAAM,OAAO,QAAQ,SAAS;AACpC,MAAI,QAAQ,IAAI;AACd,WAAO;AAAA,EACT;AACA,SAAO,OAAO,UAAU,GAAG,GAAG;AAChC;",
+  "sourcesContent": ["import createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(field => toMelindaIds(field)));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n}\n\nexport function toMelindaIds({subfields}, subfieldsToParse = ['a', 'z']) {\n  const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n  return subfields\n    .filter(sub => subfieldsToParse.includes(sub.code))\n    .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n    .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n  const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n  return {\n    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n    matchingValues\n  };\n\n  function getMatchingValuesAmount(aValues, bValues) {\n    if (bValues.length > aValues.length) {\n      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    }\n    if (aValues.length > bValues.length) {\n      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n    }\n\n    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n    // there would be duplicate values\n    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n    return aToB < bToA ? aToB : bToA;\n  }\n}\n\n\nexport function stringAfter(string, substring) {\n  const pos = string.indexOf(substring);\n  if (pos === -1) {\n    return string;\n  }\n  return string.substring(pos+substring.length);\n}\n\nexport function stringBefore(string, substring) {\n  const pos = string.indexOf(substring);\n  if (pos === -1) {\n    return string;\n  }\n  return string.substring(0, pos);\n}\n\n// We sort records here (put FIKKA and low-cased records first, ref MELINDA-3492)\n// NB! this only sorts the records the current set. If the best stuff would come in a later set and this set contains good enough record, we'll never reach the best...\n// Still this would alleviate the issue when maxMatches === 1...\nexport function preferenceSortRecords(x, y) {\n  debug(`sortRecords: RECORD COMPARISON\\n${JSON.stringify(x)} vs\\n${JSON.stringify(y)}`);\n  // Prefer the record with a FIKKA LOW tag:\n  const xFikka = hasLow(x.record, 'FIKKA');\n  const yFikka = hasLow(y.record, 'FIKKA');\n\n  if (xFikka && !yFikka) {\n    debug('sortRecords: prefer FIKKA');\n    return -1;\n  }\n  if (yFikka && !xFikka) {\n    return 1;\n  }\n\n  // Prefer downcased:\n  const xAllCaps = hasAllCapsTitle(x.record);\n  const yAllCaps = hasAllCapsTitle(y.record);\n\n  if (!xAllCaps && yAllCaps) {\n    debug('sortRecords: prefer non-all caps');\n    return -1; // x is better\n  }\n  if (!yAllCaps && xAllCaps) {\n    return 1; // y is better\n  }\n  // Default:\n  return 0;\n\n  function hasLow(record, lowTag) {\n    return record.fields?.some(f => f.tag === 'LOW' && f.subfields?.some(sf => sf.code === 'a' && sf.value === lowTag));\n  }\n\n  function hasAllCapsTitle(record) {\n    const f245 = record.fields?.find(f => f.tag === '245');\n    if (!f245) {\n      debugData(`WARNING: No f245 found for ALL CAPS check`);\n      return false;\n    }\n\n    const title = f245.subfields.filter(sf => ['a', 'b', 'c'].includes(sf.code)).map(sf => sf.value).join(' ');\n    if (title.match(/[A-Z]/u) && !title.match(/[a-z]/u)) {\n      return true;\n    }\n    return false;\n  }\n}"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAE9B,MAAM,QAAQ,kBAAkB,yCAAyC;AACzE,MAAM,YAAY,MAAM,OAAO,MAAM;AAE9B,gBAAS,kBAAkB,QAAQ;AAExC,QAAMA,SAAQ,kBAAkB,8CAA8C;AAC9E,QAAMC,aAAYD,OAAM,OAAO,MAAM;AAErC,QAAM,QAAQ,OAAO,UAAU,KAAK;AAEpC,MAAI,MAAM,SAAS,GAAG;AACpB,IAAAA,OAAM,iBAAiB;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,oBAAoB,CAAC,EAAE,OAAO,GAAG,MAAM,IAAI,WAAS,aAAa,KAAK,CAAC,CAAC;AAC9E,QAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,iBAAiB,CAAC;AAErD,EAAAC,WAAU,WAAW,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAC9D,EAAAA,WAAU,QAAQ,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AACnF,EAAAA,WAAU,eAAe,eAAe,MAAM,MAAM,KAAK,UAAU,cAAc,CAAC,EAAE;AAEpF,SAAO;AACT;AAEO,gBAAS,aAAa,EAAC,UAAS,GAAG,mBAAmB,CAAC,KAAK,GAAG,GAAG;AACvE,QAAM,kBAAkB;AAExB,SAAO,UACJ,OAAO,SAAO,iBAAiB,SAAS,IAAI,IAAI,CAAC,EACjD,OAAO,SAAO,mBAAmB,IAAI,KAAK,KAAK,gBAAgB,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtF,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,EAAE,QAAQ,iBAAiB,OAAO,IAAI,EAAE;AACtG;AAEO,gBAAS,+BAA+B,OAAO;AAEpD,YAAU,wBAAwB,KAAK,UAAU,KAAK,CAAC,EAAE;AACzD,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,MAAM,eAAe,MAAM,qBAAqB,WAAW,KAAK,WAAW,CAAC,EAAE;AAE7F,SAAO,WAAW,KAAK,WAAW;AACpC;AAEA,SAAS,eAAe,OAAO,cAAc;AAE3C,SAAO,MAAM,UAAU,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EAAE;AACnE;AAEO,gBAAS,kBAAkB,OAAO,cAAc;AACrD,YAAU,oBAAoB,YAAY,SAAS,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1E,SAAO,MAAM,UACV,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EACxC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK;AAC1B;AAEO,gBAAS,mBAAmB,OAAO;AACxC,MAAI,OAAO,UAAU,YAAY,OAAO,UAAU,UAAU;AAC1D,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAEO,gBAAS,0BAA0B,OAAO,eAAe;AAC9D,MAAI,UAAU,UAAa,MAAM,cAAc,QAAW;AACxD,WAAO,CAAC;AAAA,EACV;AACA,QAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,cAAc,SAAS,IAAI,CAAC,EAC/C,IAAI,CAAC,EAAC,MAAM,MAAK,OAAO,EAAC,MAAM,OAAO,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,GAAE,EAAE,EACtF,OAAO,WAAS,KAAK;AACxB,SAAO;AACT;AAEO,gBAAS,gBAAgB,WAAW;AACzC,SAAO,UAAU,OAAO,CAAC,KAAK,MAAM;AAClC,QAAI,CAAC,IAAI,KAAK,UAAQ,KAAK,SAAS,EAAE,QAAQ,KAAK,UAAU,EAAE,KAAK,GAAG;AACrE,YAAM,SAAS,IAAI,OAAO,CAAC;AAC3B,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT,GAAG,CAAC,CAAC;AACP;AAEO,gBAAS,eAAe,SAAS,SAAS;AAE/C,QAAM,iBAAiB,wBAAwB,SAAS,OAAO;AAE/D,SAAO;AAAA,IACL,WAAW,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA;AAAA,IAEtE,qBAAqB,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA,IAChF;AAAA,EACF;AAEA,WAAS,wBAAwBC,UAASC,UAAS;AACjD,QAAIA,SAAQ,SAASD,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AACA,QAAID,SAAQ,SAASC,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AAIA,UAAM,OAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AACjF,UAAM,OAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAEjF,WAAO,OAAO,OAAO,OAAO;AAAA,EAC9B;AACF;AAGO,gBAAS,YAAY,QAAQ,WAAW;AAC7C,QAAM,MAAM,OAAO,QAAQ,SAAS;AACpC,MAAI,QAAQ,IAAI;AACd,WAAO;AAAA,EACT;AACA,SAAO,OAAO,UAAU,MAAI,UAAU,MAAM;AAC9C;AAEO,gBAAS,aAAa,QAAQ,WAAW;AAC9C,QAAM,MAAM,OAAO,QAAQ,SAAS;AACpC,MAAI,QAAQ,IAAI;AACd,WAAO;AAAA,EACT;AACA,SAAO,OAAO,UAAU,GAAG,GAAG;AAChC;AAKO,gBAAS,sBAAsB,GAAG,GAAG;AAC1C,QAAM;AAAA,EAAmC,KAAK,UAAU,CAAC,CAAC;AAAA,EAAQ,KAAK,UAAU,CAAC,CAAC,EAAE;AAErF,QAAM,SAAS,OAAO,EAAE,QAAQ,OAAO;AACvC,QAAM,SAAS,OAAO,EAAE,QAAQ,OAAO;AAEvC,MAAI,UAAU,CAAC,QAAQ;AACrB,UAAM,2BAA2B;AACjC,WAAO;AAAA,EACT;AACA,MAAI,UAAU,CAAC,QAAQ;AACrB,WAAO;AAAA,EACT;AAGA,QAAM,WAAW,gBAAgB,EAAE,MAAM;AACzC,QAAM,WAAW,gBAAgB,EAAE,MAAM;AAEzC,MAAI,CAAC,YAAY,UAAU;AACzB,UAAM,kCAAkC;AACxC,WAAO;AAAA,EACT;AACA,MAAI,CAAC,YAAY,UAAU;AACzB,WAAO;AAAA,EACT;AAEA,SAAO;AAEP,WAAS,OAAO,QAAQ,QAAQ;AAC9B,WAAO,OAAO,QAAQ,KAAK,OAAK,EAAE,QAAQ,SAAS,EAAE,WAAW,KAAK,QAAM,GAAG,SAAS,OAAO,GAAG,UAAU,MAAM,CAAC;AAAA,EACpH;AAEA,WAAS,gBAAgB,QAAQ;AAC/B,UAAM,OAAO,OAAO,QAAQ,KAAK,OAAK,EAAE,QAAQ,KAAK;AACrD,QAAI,CAAC,MAAM;AACT,gBAAU,2CAA2C;AACrD,aAAO;AAAA,IACT;AAEA,UAAM,QAAQ,KAAK,UAAU,OAAO,QAAM,CAAC,KAAK,KAAK,GAAG,EAAE,SAAS,GAAG,IAAI,CAAC,EAAE,IAAI,QAAM,GAAG,KAAK,EAAE,KAAK,GAAG;AACzG,QAAI,MAAM,MAAM,QAAQ,KAAK,CAAC,MAAM,MAAM,QAAQ,GAAG;AACnD,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AACF;",
   "names": ["debug", "debugData", "aValues", "bValues"]
 }

package/package.json

@@ -13,12 +13,12 @@
 		"url": "https://github.com/NatLibFi/melinda-record-matching-js"
 	},
 	"license": "MIT",
-	"version": "5.1.4",
+	"version": "5.1.5-alpha.1",
 	"type": "module",
 	"main": "./dist/index.js",
 	"bin": "./dist/cli.js",
 	"engines": {
-		"node": ">=22"
+		"node": ">=24"
 	},
 	"publishConfig": {
 		"access": "public"
@@ -30,11 +30,12 @@
 		"cli": "node dist/cli.js",
 		"lint": "eslint ./src",
 		"lint:dev": "eslint --fix ./src",
-		"test:base": "node --test --test-force-exit --experimental-test-coverage --test-reporter=spec './src/**/*.test.js'",
+		"test:base": "node --test --test-force-exit --experimental-test-coverage --test-reporter=spec 'test/**/*.test.js'",
 		"test": "npm run lint && npm run test:base",
 		"test:only": "npm run lint && npm run test:base:only",
-		"test:base:only": "cross-env DEBUG=@natlibfi/*matching* NODE_ENV=test node --test --test-only --test-force-exit --experimental-test-coverage --test-reporter=spec './src/**/*.test.js'",
-		"watch:test": "cross-env DEBUG=@natlibfi/* NODE_ENV=test node --watch --test --experimental-test-coverage --test-reporter=spec './src/**/*.test.js'",
+		"test:base:only": "cross-env DEBUG=@natlibfi/* NODE_ENV=test node --test --test-only --test-force-exit --experimental-test-coverage --test-reporter=spec 'test/**/*.test.js'",
+		"test:base:only:index": "cross-env DEBUG=@natlibfi/* NODE_ENV=test node --test --test-only --test-force-exit --experimental-test-coverage --test-reporter=spec './test/index.test.js'",
+		"watch:test": "cross-env DEBUG=@natlibfi/* NODE_ENV=test node --watch --test --experimental-test-coverage --test-reporter=spec 'test/**/*.test.js'",
 		"dev": "npm run watch:test",
 		"dev:debug": "cross-env LOG_LEVEL=debug DEBUG=@natlibfi/* NODE_ENV=test npm run watch:test"
 	},
@@ -42,9 +43,9 @@
 		"@natlibfi/marc-record": "^10.0.2",
 		"@natlibfi/marc-record-validators-melinda": "^12.0.15",
 		"@natlibfi/melinda-commons": "^14.0.2",
-		"@natlibfi/sru-client": "^7.0.1",
+		"@natlibfi/sru-client": "^7.0.2",
 		"debug": "^4.4.3",
-		"isbn3": "^2.0.8",
+		"isbn3": "^2.0.9",
 		"natural": "^8.1.1",
 		"yargs": "^18.0.0"
 	},
@@ -54,9 +55,6 @@
 		"@natlibfi/fixura": "^4.0.1",
 		"cross-env": "^10.1.0",
 		"esbuild": "^0.28.1",
-		"eslint": "^10.5.0"
-	},
-	"overrides": {
-		"uuid": "^14.0.0"
+		"eslint": "^10.7.0"
 	}
 }