@natlibfi/melinda-record-matching 5.1.4 → 5.1.6-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/cli.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../src/cli.js"],
-  "sourcesContent": ["import fs from 'fs';\nimport yargs from 'yargs';\nimport createMatchOperator, {candidateSearch, matchDetection} from './index.js';\nimport createDebugLogger from 'debug';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\ncli();\n\nasync function cli() {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:cli');\n  const args = yargs(process.argv.slice(2))\n    .scriptName('melinda-record-matching-js')\n    .epilog('Copyright (C) 2022-2023 University Of Helsinki (The National Library Of Finland)')\n    .usage('$0 <file> [options] and env variable info in README')\n    .usage('Installed globally: $0 <file> [options] and env variable info in README')\n    .usage('Not installed: npx $0 <file> [options] and env variable info in README')\n    .usage('Build from source: node dist/index.js <file> [options] and env variable info in README')\n    .showHelpOnFail(true)\n    .example([['$ node /dist/cli.js record.json']])\n    .env('MELINDA_RECORD_MATCH')\n    .version()\n    .positional('file', {type: 'string', describe: 'Json file of records to match'})\n    .options({\n      t: {type: 'string', default: 'IDS', alias: 'searchType', describe: 'IDS, STANDARD_IDS, COMPONENT, CONTENT or CONTENTALT'},\n      m: {type: 'number', default: 1, alias: 'maxMatches', describe: ''},\n      c: {type: 'number', default: 1000, alias: 'maxCandidates', describe: ''},\n      s: {type: 'boolean', default: false, alias: 'returnStrategy', describe: ''},\n      q: {type: 'boolean', default: false, alias: 'returnQuery', describe: ''},\n      n: {type: 'boolean', default: false, alias: 'returnNonMatches', describe: ''}\n    })\n    .check((args) => {\n      const [file] = args._;\n      if (file === undefined) {\n        throw new Error('No file argument given');\n      }\n\n      if (!fs.existsSync(file)) {\n        throw new Error(`File ${file} does not exist`);\n      }\n\n      if (args.sruUrl === undefined) {\n        throw new Error('Setup sru url');\n      }\n\n      if (!['IDS', 'STANDARD_IDS', 'COMPONENT', 'CONTENT', 'CONTENTALT', 'YSO'].includes(args.searchType)) {\n        throw new Error('Invalid search type');\n      }\n\n      return true;\n    })\n    .parseSync();\n\n  const [file] = args._;\n  const {searchType} = args;\n  debug(JSON.stringify(args));\n\n  const detection = {\n    threshold: 0.8999,\n    strategy: generateStrategy(searchType)\n  };\n\n  const search = {\n    url: args.sruUrl, searchSpec: generateSearchSpec(searchType)\n  };\n\n  const matchOperator = await createMatchOperator({detection, search, ...args});\n\n  const fileRaw = fs.readFileSync(file, 'utf8');\n  const record = new MarcRecord(JSON.parse(fileRaw), {subfieldValues: false});\n\n  const result = await matchOperator({record});\n  debug(JSON.stringify(result));\n\n\n  function generateStrategy(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.melindaId(),\n        matchDetection.features.bib.allSourceIds(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    // We could have differing strategy for STANDARD_IDS\n    // Let's not run title in strategy when we found the candidates through standard_ids search\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.publisherNumber(),\n        // matchDetection.features.bib.sid(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        // Let's not use the same title matchDetection here\n        //matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        // We probably should have some leeway here for notated music as BK etc.\n        matchDetection.features.bib.recordType(),\n        // Use publicationTimeAllowConsYearsMulti to\n        //  - ignore one year differences in publicationTime\n        //  - extract publicationTimes from f008, f26x and reprint notes in f500\n        //  - do not substract points for mismatching (normal) publicationTime, if there's a match between\n        //       normal publicationTime and a reprintPublication time\n        matchDetection.features.bib.publicationTimeAllowConsYearsMulti(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['CONTENT', 'CONTENTALT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.publisherNumber(),\n        // matchDetection.features.bib.sid(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.publicationTime(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['YSO']) {\n      return [\n        // matchDetection.features.bib.sid(), // SIDs might be relevant on the auth side as well?\n        matchDetection.features.auth.yso()\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n\n\n  function generateSearchSpec(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.melindaId,\n        candidateSearch.searchTypes.bib.sourceIds\n      ];\n    }\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [candidateSearch.searchTypes.bib.standardIdentifiers];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        //candidateSearch.searchTypes.bib.sourceIds,\n        candidateSearch.searchTypes.component.hostIdMelinda,\n        candidateSearch.searchTypes.component.hostIdOtherSource\n      ];\n    }\n\n    if (['CONTENT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        //candidateSearch.searchTypes.bib.titleAuthor,\n        candidateSearch.searchTypes.bib.title\n      ];\n    }\n    if (['CONTENTALT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        // titleAuthorYearAlternates searches for matchCandidates\n        // with alternate queries, starting from more tight searches\n        candidateSearch.searchTypes.bib.titleAuthorYearAlternates\n      ];\n    }\n\n    if (searchType === 'YSO') {\n      return [\n        candidateSearch.searchTypes.auth.authURX\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n}\n"],
-  "mappings": "AAAA,OAAO,QAAQ;AACf,OAAO,WAAW;AAClB,OAAO,uBAAsB,iBAAiB,sBAAqB;AACnE,OAAO,uBAAuB;AAC9B,SAAQ,kBAAiB;AAEzB,IAAI;AAEJ,eAAe,MAAM;AACnB,QAAM,QAAQ,kBAAkB,uCAAuC;AACvE,QAAM,OAAO,MAAM,QAAQ,KAAK,MAAM,CAAC,CAAC,EACrC,WAAW,4BAA4B,EACvC,OAAO,kFAAkF,EACzF,MAAM,qDAAqD,EAC3D,MAAM,yEAAyE,EAC/E,MAAM,wEAAwE,EAC9E,MAAM,wFAAwF,EAC9F,eAAe,IAAI,EACnB,QAAQ,CAAC,CAAC,iCAAiC,CAAC,CAAC,EAC7C,IAAI,sBAAsB,EAC1B,QAAQ,EACR,WAAW,QAAQ,EAAC,MAAM,UAAU,UAAU,gCAA+B,CAAC,EAC9E,QAAQ;AAAA,IACP,GAAG,EAAC,MAAM,UAAU,SAAS,OAAO,OAAO,cAAc,UAAU,sDAAqD;AAAA,IACxH,GAAG,EAAC,MAAM,UAAU,SAAS,GAAG,OAAO,cAAc,UAAU,GAAE;AAAA,IACjE,GAAG,EAAC,MAAM,UAAU,SAAS,KAAM,OAAO,iBAAiB,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,kBAAkB,UAAU,GAAE;AAAA,IAC1E,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,eAAe,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,oBAAoB,UAAU,GAAE;AAAA,EAC9E,CAAC,EACA,MAAM,CAACA,UAAS;AACf,UAAM,CAACC,KAAI,IAAID,MAAK;AACpB,QAAIC,UAAS,QAAW;AACtB,YAAM,IAAI,MAAM,wBAAwB;AAAA,IAC1C;AAEA,QAAI,CAAC,GAAG,WAAWA,KAAI,GAAG;AACxB,YAAM,IAAI,MAAM,QAAQA,KAAI,iBAAiB;AAAA,IAC/C;AAEA,QAAID,MAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,eAAe;AAAA,IACjC;AAEA,QAAI,CAAC,CAAC,OAAO,gBAAgB,aAAa,WAAW,cAAc,KAAK,EAAE,SAASA,MAAK,UAAU,GAAG;AACnG,YAAM,IAAI,MAAM,qBAAqB;AAAA,IACvC;AAEA,WAAO;AAAA,EACT,CAAC,EACA,UAAU;AAEb,QAAM,CAAC,IAAI,IAAI,KAAK;AACpB,QAAM,EAAC,WAAU,IAAI;AACrB,QAAM,KAAK,UAAU,IAAI,CAAC;AAE1B,QAAM,YAAY;AAAA,IAChB,WAAW;AAAA,IACX,UAAU,iBAAiB,UAAU;AAAA,EACvC;AAEA,QAAM,SAAS;AAAA,IACb,KAAK,KAAK;AAAA,IAAQ,YAAY,mBAAmB,UAAU;AAAA,EAC7D;AAEA,QAAM,gBAAgB,MAAM,oBAAoB,EAAC,WAAW,QAAQ,GAAG,KAAI,CAAC;AAE5E,QAAM,UAAU,GAAG,aAAa,MAAM,MAAM;AAC5C,QAAM,SAAS,IAAI,WAAW,KAAK,MAAM,OAAO,GAAG,EAAC,gBAAgB,MAAK,CAAC;AAE1E,QAAM,SAAS,MAAM,cAAc,EAAC,OAAM,CAAC;AAC3C,QAAM,KAAK,UAAU,MAAM,CAAC;AAG5B,WAAS,iBAAiBE,aAAY;AACpC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,UAAU;AAAA,QACtC,eAAe,SAAS,IAAI,aAAa;AAAA,QACzC,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAKA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,gBAAgB;AAAA;AAAA,QAE5C,eAAe,SAAS,IAAI,wBAAwB;AAAA;AAAA;AAAA,QAGpD,eAAe,SAAS,IAAI,QAAQ;AAAA;AAAA,QAEpC,eAAe,SAAS,IAAI,WAAW;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,QAMvC,eAAe,SAAS,IAAI,mCAAmC;AAAA,QAC/D,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,YAAY,EAAE,SAASA,WAAU,GAAG;AAClD,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,gBAAgB;AAAA;AAAA,QAE5C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,gBAAgB;AAAA,QAC5C,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,KAAK,GAAG;AACX,aAAO;AAAA;AAAA,QAEL,eAAe,SAAS,KAAK,IAAI;AAAA,MACnC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AAGA,WAAS,mBAAmBA,aAAY;AACtC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA,QAChC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO,CAAC,gBAAgB,YAAY,IAAI,mBAAmB;AAAA,IAC7D;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA;AAAA,QAEL,gBAAgB,YAAY,UAAU;AAAA,QACtC,gBAAgB,YAAY,UAAU;AAAA,MACxC;AAAA,IACF;AAEA,QAAI,CAAC,SAAS,EAAE,SAASA,WAAU,GAAG;AACpC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA,QAEhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AACA,QAAI,CAAC,YAAY,EAAE,SAASA,WAAU,GAAG;AACvC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA;AAAA,QAGhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAIA,gBAAe,OAAO;AACxB,aAAO;AAAA,QACL,gBAAgB,YAAY,KAAK;AAAA,MACnC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AACF;",
+  "sourcesContent": ["import fs from 'fs';\nimport yargs from 'yargs';\nimport createMatchOperator, {candidateSearch, matchDetection} from './index.js';\nimport createDebugLogger from 'debug';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\ncli();\n\n// eslint-disable-next-line max-lines-per-function\nasync function cli() {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:cli');\n  const args = yargs(process.argv.slice(2))\n    .scriptName('melinda-record-matching-js')\n    .epilog('Copyright (C) 2022-2023 University Of Helsinki (The National Library Of Finland)')\n    .usage('$0 <file> [options] and env variable info in README')\n    .usage('Installed globally: $0 <file> [options] and env variable info in README')\n    .usage('Not installed: npx $0 <file> [options] and env variable info in README')\n    .usage('Build from source: node dist/index.js <file> [options] and env variable info in README')\n    .showHelpOnFail(true)\n    .example([['$ node /dist/cli.js record.json']])\n    .env('MELINDA_RECORD_MATCH')\n    .version()\n    .positional('file', {type: 'string', describe: 'Json file of records to match'})\n    .options({\n      t: {type: 'string', default: 'IDS', alias: 'searchType', describe: 'IDS, STANDARD_IDS, COMPONENT, CONTENT or CONTENTALT'},\n      m: {type: 'number', default: 1, alias: 'maxMatches', describe: ''},\n      c: {type: 'number', default: 1000, alias: 'maxCandidates', describe: ''},\n      s: {type: 'boolean', default: false, alias: 'returnStrategy', describe: ''},\n      q: {type: 'boolean', default: false, alias: 'returnQuery', describe: ''},\n      n: {type: 'boolean', default: false, alias: 'returnNonMatches', describe: ''}\n    })\n    .check((args) => {\n      const [file] = args._;\n      if (file === undefined) {\n        throw new Error('No file argument given');\n      }\n\n      if (!fs.existsSync(file)) {\n        throw new Error(`File ${file} does not exist`);\n      }\n\n      if (args.sruUrl === undefined) {\n        throw new Error('Setup sru url');\n      }\n\n      if (!['IDS', 'STANDARD_IDS', 'COMPONENT', 'CONTENT', 'CONTENTALT', 'YSO'].includes(args.searchType)) {\n        throw new Error('Invalid search type');\n      }\n\n      return true;\n    })\n    .parseSync();\n\n  const [file] = args._;\n  const {searchType} = args;\n  debug(JSON.stringify(args));\n\n  const detection = {\n    threshold: 0.8999,\n    strategy: generateStrategy(searchType)\n  };\n\n  const search = {\n    url: args.sruUrl, searchSpec: generateSearchSpec(searchType)\n  };\n\n  const matchOperator = await createMatchOperator({detection, search, ...args});\n\n  const fileRaw = fs.readFileSync(file, 'utf8');\n  const record = new MarcRecord(JSON.parse(fileRaw), {subfieldValues: false});\n\n  const result = await matchOperator({record});\n  debug(JSON.stringify(result));\n\n\n  function generateStrategy(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.melindaId(),\n        matchDetection.features.bib.allSourceIds(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    // We could have differing strategy for STANDARD_IDS\n    // Let's not run title in strategy when we found the candidates through standard_ids search\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.publisherNumber(),\n        // matchDetection.features.bib.sid(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        // Let's not use the same title matchDetection here\n        //matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        // We probably should have some leeway here for notated music as BK etc.\n        matchDetection.features.bib.recordType(),\n        // Use publicationTimeAllowConsYearsMulti to\n        //  - ignore one year differences in publicationTime\n        //  - extract publicationTimes from f008, f26x and reprint notes in f500\n        //  - do not substract points for mismatching (normal) publicationTime, if there's a match between\n        //       normal publicationTime and a reprintPublication time\n        matchDetection.features.bib.publicationTimeAllowConsYearsMulti(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['CONTENT', 'CONTENTALT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.publisherNumber(),\n        // matchDetection.features.bib.sid(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.publicationTime(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['YSO']) {\n      return [\n        // matchDetection.features.bib.sid(), // SIDs might be relevant on the auth side as well?\n        matchDetection.features.auth.yso()\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n\n\n  function generateSearchSpec(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.melindaId,\n        candidateSearch.searchTypes.bib.sourceIds\n      ];\n    }\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [candidateSearch.searchTypes.bib.standardIdentifiers];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        //candidateSearch.searchTypes.bib.sourceIds,\n        candidateSearch.searchTypes.component.hostIdMelinda,\n        candidateSearch.searchTypes.component.hostIdOtherSource\n      ];\n    }\n\n    if (['CONTENT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        //candidateSearch.searchTypes.bib.titleAuthor,\n        candidateSearch.searchTypes.bib.title\n      ];\n    }\n    if (['CONTENTALT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        // titleAuthorYearAlternates searches for matchCandidates\n        // with alternate queries, starting from more tight searches\n        candidateSearch.searchTypes.bib.titleAuthorYearAlternates\n      ];\n    }\n\n    if (searchType === 'YSO') {\n      return [\n        candidateSearch.searchTypes.auth.authURX\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n}\n"],
+  "mappings": "AAAA,OAAO,QAAQ;AACf,OAAO,WAAW;AAClB,OAAO,uBAAsB,iBAAiB,sBAAqB;AACnE,OAAO,uBAAuB;AAC9B,SAAQ,kBAAiB;AAEzB,IAAI;AAGJ,eAAe,MAAM;AACnB,QAAM,QAAQ,kBAAkB,uCAAuC;AACvE,QAAM,OAAO,MAAM,QAAQ,KAAK,MAAM,CAAC,CAAC,EACrC,WAAW,4BAA4B,EACvC,OAAO,kFAAkF,EACzF,MAAM,qDAAqD,EAC3D,MAAM,yEAAyE,EAC/E,MAAM,wEAAwE,EAC9E,MAAM,wFAAwF,EAC9F,eAAe,IAAI,EACnB,QAAQ,CAAC,CAAC,iCAAiC,CAAC,CAAC,EAC7C,IAAI,sBAAsB,EAC1B,QAAQ,EACR,WAAW,QAAQ,EAAC,MAAM,UAAU,UAAU,gCAA+B,CAAC,EAC9E,QAAQ;AAAA,IACP,GAAG,EAAC,MAAM,UAAU,SAAS,OAAO,OAAO,cAAc,UAAU,sDAAqD;AAAA,IACxH,GAAG,EAAC,MAAM,UAAU,SAAS,GAAG,OAAO,cAAc,UAAU,GAAE;AAAA,IACjE,GAAG,EAAC,MAAM,UAAU,SAAS,KAAM,OAAO,iBAAiB,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,kBAAkB,UAAU,GAAE;AAAA,IAC1E,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,eAAe,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,oBAAoB,UAAU,GAAE;AAAA,EAC9E,CAAC,EACA,MAAM,CAACA,UAAS;AACf,UAAM,CAACC,KAAI,IAAID,MAAK;AACpB,QAAIC,UAAS,QAAW;AACtB,YAAM,IAAI,MAAM,wBAAwB;AAAA,IAC1C;AAEA,QAAI,CAAC,GAAG,WAAWA,KAAI,GAAG;AACxB,YAAM,IAAI,MAAM,QAAQA,KAAI,iBAAiB;AAAA,IAC/C;AAEA,QAAID,MAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,eAAe;AAAA,IACjC;AAEA,QAAI,CAAC,CAAC,OAAO,gBAAgB,aAAa,WAAW,cAAc,KAAK,EAAE,SAASA,MAAK,UAAU,GAAG;AACnG,YAAM,IAAI,MAAM,qBAAqB;AAAA,IACvC;AAEA,WAAO;AAAA,EACT,CAAC,EACA,UAAU;AAEb,QAAM,CAAC,IAAI,IAAI,KAAK;AACpB,QAAM,EAAC,WAAU,IAAI;AACrB,QAAM,KAAK,UAAU,IAAI,CAAC;AAE1B,QAAM,YAAY;AAAA,IAChB,WAAW;AAAA,IACX,UAAU,iBAAiB,UAAU;AAAA,EACvC;AAEA,QAAM,SAAS;AAAA,IACb,KAAK,KAAK;AAAA,IAAQ,YAAY,mBAAmB,UAAU;AAAA,EAC7D;AAEA,QAAM,gBAAgB,MAAM,oBAAoB,EAAC,WAAW,QAAQ,GAAG,KAAI,CAAC;AAE5E,QAAM,UAAU,GAAG,aAAa,MAAM,MAAM;AAC5C,QAAM,SAAS,IAAI,WAAW,KAAK,MAAM,OAAO,GAAG,EAAC,gBAAgB,MAAK,CAAC;AAE1E,QAAM,SAAS,MAAM,cAAc,EAAC,OAAM,CAAC;AAC3C,QAAM,KAAK,UAAU,MAAM,CAAC;AAG5B,WAAS,iBAAiBE,aAAY;AACpC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,UAAU;AAAA,QACtC,eAAe,SAAS,IAAI,aAAa;AAAA,QACzC,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAKA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,gBAAgB;AAAA;AAAA,QAE5C,eAAe,SAAS,IAAI,wBAAwB;AAAA;AAAA;AAAA,QAGpD,eAAe,SAAS,IAAI,QAAQ;AAAA;AAAA,QAEpC,eAAe,SAAS,IAAI,WAAW;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,QAMvC,eAAe,SAAS,IAAI,mCAAmC;AAAA,QAC/D,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,YAAY,EAAE,SAASA,WAAU,GAAG;AAClD,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,gBAAgB;AAAA;AAAA,QAE5C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,gBAAgB;AAAA,QAC5C,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,KAAK,GAAG;AACX,aAAO;AAAA;AAAA,QAEL,eAAe,SAAS,KAAK,IAAI;AAAA,MACnC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AAGA,WAAS,mBAAmBA,aAAY;AACtC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA,QAChC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO,CAAC,gBAAgB,YAAY,IAAI,mBAAmB;AAAA,IAC7D;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA;AAAA,QAEL,gBAAgB,YAAY,UAAU;AAAA,QACtC,gBAAgB,YAAY,UAAU;AAAA,MACxC;AAAA,IACF;AAEA,QAAI,CAAC,SAAS,EAAE,SAASA,WAAU,GAAG;AACpC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA,QAEhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AACA,QAAI,CAAC,YAAY,EAAE,SAASA,WAAU,GAAG;AACvC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA;AAAA,QAGhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAIA,gBAAe,OAAO;AACxB,aAAO;AAAA,QACL,gBAAgB,YAAY,KAAK;AAAA,MACnC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AACF;",
   "names": ["args", "file", "searchType"]
 }

package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js

@@ -109,6 +109,13 @@ export default () => ({
     if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {
       return 0;
     }
+    const bNormalInACopyright = a.copyrightYears.filter((aValue) => b.normalYears.some((bValue) => aValue === bValue));
+    const aNormalInBCopyright = b.copyrightYears.filter((bValue) => a.normalYears.some((aValue) => bValue === aValue));
+    debug(`BNorm in ACopyright: ${JSON.stringify(bNormalInACopyright)}`);
+    debug(`ANorm in BCopyright: ${JSON.stringify(aNormalInBCopyright)}`);
+    if (bNormalInACopyright > 0 || aNormalInBCopyright > 0) {
+      return 0;
+    }
     return -1;
   }
 });

package/dist/match-detection/features/bib/publication-time-allow-cons-years-multi.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/publication-time-allow-cons-years-multi.js"],
-  "sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We should also get publishing time type from f008\n// We should get reprint times from f500 $a \"Lis\u00E4painos/Lis\u00E4painokset:\"\n\nexport default () => ({\n  name: 'Publication time, allow consequent years, years from multiple sources',\n  extract: ({record, recordExternal}) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const f008Values = extractF008Values(record);\n    debug(`${label}: f008: ${JSON.stringify(f008Values)}`);\n\n    const f26xValues = extractF26xValues(record);\n    debug(`${label}: f26x: ${JSON.stringify(f26xValues)}`);\n\n    const f500Values = extractF500Years(record);\n    debug(`${label}: f500: ${JSON.stringify(f500Values)}`);\n\n    // We should get copyrightYear from f008Date2 to copyrightYears when f008YearType = 'r'\n    // Is the original year (f008Date2) in f008YearType === 'r' comparable to copyrightYear?\n    // We should handle unknown years (here or comparison?)\n    // We should handle year ranges for continuing resources / collections\n\n    const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter(value => value && value !== '    ' && value !== '||||'))].sort();\n    const copyrightYears = [...new Set(f26xValues.copyrightYears)].sort();\n    const reprintYears = [...new Set(f500Values)].sort();\n\n    const combined = {normalYears, copyrightYears, reprintYears};\n\n    debug(`Combined: ${JSON.stringify(combined)}`);\n\n    return combined;\n\n    function extractF008Values(record) {\n      // Record should have only one f008 - in case of several, we handle the first one\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      if (value && testStringOrNumber(value)) {\n        const f008Date1 = extractF008Date1(value);\n        const f008Date2 = extractF008Date2(value);\n        const f008YearType = extractF008YearType(value);\n        return {f008Date1, f008Date2, f008YearType};\n      }\n      return {f008Date1: undefined, f008Date2: undefined, f008YearType: undefined};\n\n      function extractF008Date1(value) {\n        return String(value).slice(7, 11);\n      }\n\n      function extractF008Date2(value) {\n        return String(value).slice(11, 15);\n      }\n\n      function extractF008YearType(value) {\n        return String(value).slice(6, 7);\n      }\n    }\n\n    function extractF26xValues(record) {\n      const copyrightRegex = /^(?<copyrightPrefix>cop|cop.|c|\u00A9|p|\u2117)/u;\n\n      const pubNormalSubFieldValues = extractPubNormalSubfieldValues(record, copyrightRegex);\n      debug(`Normal years: ${JSON.stringify(pubNormalSubFieldValues)}`);\n\n      const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(record);\n      debug(`F264 copyright years: ${JSON.stringify(pubF264CopySubFieldValues)}`);\n\n      const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(record, copyrightRegex);\n      debug(`F260 copyright years: ${JSON.stringify(pubF260CopySubFieldValues)}`);\n\n      return {normalYears: pubNormalSubFieldValues, copyrightYears: [...pubF260CopySubFieldValues, ...pubF264CopySubFieldValues]};\n\n      function extractPubNormalSubfieldValues(record, copyrightRegex) {\n        return record.get(/^26[04]$/u)\n          .filter((field) => !(field.tag === '264' && field.ind2 === '4'))\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value && !copyrightRegex.test(value))\n          .map(({value}) => value)\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function extractPubF264CopySubfieldValues(record, copyrightRegex) {\n        return record.get(/^264$/u)\n          .filter((field) => field.ind2 === '4')\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value)\n          .map(({value}) => value)\n          .map((value) => value.replace(copyrightRegex, ''))\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function extractPubF260CopySubfieldValues(record, copyrightRegex) {\n        return record.get(/^260$/u)\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value && copyrightRegex.test(value))\n          .map(({value}) => value)\n          .map((value) => value.replace(copyrightRegex, ''))\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function removeNonAlphaNumeric(value) {\n        debug(`Cleaning: ${JSON.stringify(value)}`);\n        const nonAlphaNumericRegex = /[^A-Za-z0-9]/ug;\n        return value ? value.replace(nonAlphaNumericRegex, '') : value;\n      }\n    }\n\n    function extractF500Years(record) {\n\n      const reprintRegex = /(?<reprint>Lis\u00E4painokset:|Lis\u00E4painos:)/u;\n      const reprintFieldContents = record.get(/^500$/u)\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a')\n        .map(({value}) => value);\n      //.filter(value => value.test(reprintRegex));\n\n      debug(`f500 reprint field contents: ${JSON.stringify(reprintFieldContents)}`);\n\n      const filteredF500 = reprintFieldContents.filter((content) => content && content.match(reprintRegex));\n\n      debug(`f500 reprint field contents (filtered): ${JSON.stringify(filteredF500)}`);\n\n      const reprintYears = extractReprintYears(filteredF500);\n\n      return reprintYears;\n    }\n\n    function extractReprintYears(contents) {\n      const yearRegex = /[0-9][0-9][0-9][0-9]/gu;\n      const years = contents.map(content => content.match(yearRegex))\n        .flat();\n      debug(`${JSON.stringify(years)}`);\n      const uniqYears = [...new Set(years)].sort();\n      debug(`${JSON.stringify(uniqYears)}`);\n      return uniqYears;\n    }\n\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n    debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);\n\n    const [firstA] = a.normalYears ? a.normalYears : a;\n    const [firstB] = b.normalYears ? b.normalYears : b;\n\n    debug(`Comparing ${JSON.stringify(firstA)} to ${JSON.stringify(firstB)}`);\n\n    if (firstA === firstB) {\n      return 0.1;\n    }\n\n    // If either of years is a non string/number, values are not comparable\n    if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {\n      return 0;\n    }\n\n    const firstANumber = parseInt(firstA, 10);\n    const firstBNumber = parseInt(firstB, 10);\n\n    // Handle consequent years as a match\n    // see publication-time for a version that does not handle consequent years as a match\n    if (!(isNaN(firstANumber) || isNaN(firstBNumber)) &&\n      (firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber)) {\n      return 0.1;\n    }\n\n    // We should do something with copyrightYears, too\n\n    // Do not give minus points if a normal publishing year is found in normal years\n    const bNormalInANormal = a.normalYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n    const aNormalInBNormal = b.normalYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n    debug(`BNorm in ANorm: ${JSON.stringify(bNormalInANormal)}`);\n    debug(`ANorm in BNorm: ${JSON.stringify(aNormalInBNormal)}`);\n\n    if (bNormalInANormal > 0 || aNormalInBNormal > 0) {\n      return 0;\n    }\n\n    // Do not give minus points if a normal publishing year is found in reprint years\n    const bNormalInAReprint = a.reprintYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n    const aNormalInBReprint = b.reprintYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n    debug(`BNorm in AReprint: ${JSON.stringify(bNormalInAReprint)}`);\n    debug(`ANorm in BReprint: ${JSON.stringify(aNormalInBReprint)}`);\n\n    if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {\n      return 0;\n    }\n\n    return -1;\n\n  }\n});\n\n// https://www.loc.gov/marc/bibliographic/bd008.html\n// field 008\n// 06 - Type of date/Publication status\n// 07-10 - Date 1\n// 11-14 - Date 2\n//\n// 06 - Type of date/Publication status\n// b - No dates given; B.C. date involved\n// c - Continuing resource currently published\n// d - Continuing resource ceased publication\n// e - Detailed date\n// i - Inclusive dates of collection\n// k - Range of years of bulk of collection\n// m - Multiple dates\n// n - Dates unknown\n// p - Date of distribution/release/issue and production/recording session when different\n// q - Questionable date\n// r - Reprint/reissue date and original date\n// s - Single known date/probable date\n// t - Publication date and copyright date\n// u - Continuing resource status unknown\n// | - No attempt to code\n//\n// 07-10 - Date 1\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n// 11-14 - Date 2\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n\n"],
-  "mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,0BAAyB;AAMjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,QAAQ,eAAc,MAAM;AACrC,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,QAAQ,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAE9E,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,iBAAiB,MAAM;AAC1C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAOrD,UAAM,cAAc,CAAC,GAAG,IAAI,IAAI,WAAW,YAAY,OAAO,WAAW,SAAS,EAAE,OAAO,WAAS,SAAS,UAAU,UAAU,UAAU,MAAM,CAAC,CAAC,EAAE,KAAK;AAC1J,UAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,WAAW,cAAc,CAAC,EAAE,KAAK;AACpE,UAAM,eAAe,CAAC,GAAG,IAAI,IAAI,UAAU,CAAC,EAAE,KAAK;AAEnD,UAAM,WAAW,EAAC,aAAa,gBAAgB,aAAY;AAE3D,UAAM,aAAa,KAAK,UAAU,QAAQ,CAAC,EAAE;AAE7C,WAAO;AAEP,aAAS,kBAAkBA,SAAQ;AAEjC,YAAM,QAAQA,QAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,UAAI,SAAS,mBAAmB,KAAK,GAAG;AACtC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,eAAe,oBAAoB,KAAK;AAC9C,eAAO,EAAC,WAAW,WAAW,aAAY;AAAA,MAC5C;AACA,aAAO,EAAC,WAAW,QAAW,WAAW,QAAW,cAAc,OAAS;AAE3E,eAAS,iBAAiBC,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,EAAE;AAAA,MAClC;AAEA,eAAS,iBAAiBA,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,IAAI,EAAE;AAAA,MACnC;AAEA,eAAS,oBAAoBA,QAAO;AAClC,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,CAAC;AAAA,MACjC;AAAA,IACF;AAEA,aAAS,kBAAkBD,SAAQ;AACjC,YAAM,iBAAiB;AAEvB,YAAM,0BAA0B,+BAA+BA,SAAQ,cAAc;AACrF,YAAM,iBAAiB,KAAK,UAAU,uBAAuB,CAAC,EAAE;AAEhE,YAAM,4BAA4B,iCAAiCA,OAAM;AACzE,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,YAAM,4BAA4B,iCAAiCA,SAAQ,cAAc;AACzF,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,aAAO,EAAC,aAAa,yBAAyB,gBAAgB,CAAC,GAAG,2BAA2B,GAAG,yBAAyB,EAAC;AAE1H,eAAS,+BAA+BA,SAAQE,iBAAgB;AAC9D,eAAOF,QAAO,IAAI,WAAW,EAC1B,OAAO,CAAC,UAAU,EAAE,MAAM,QAAQ,SAAS,MAAM,SAAS,IAAI,EAC9D,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAAS,CAACE,gBAAe,KAAK,KAAK,CAAC,EACxD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,OAAO,CAAC,UAAU,MAAM,SAAS,GAAG,EACpC,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,KAAK,EACzB,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQE,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAASE,gBAAe,KAAK,KAAK,CAAC,EACvD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQA,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,sBAAsB,OAAO;AACpC,cAAM,aAAa,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1C,cAAM,uBAAuB;AAC7B,eAAO,QAAQ,MAAM,QAAQ,sBAAsB,EAAE,IAAI;AAAA,MAC3D;AAAA,IACF;AAEA,aAAS,iBAAiBF,SAAQ;AAEhC,YAAM,eAAe;AACrB,YAAM,uBAAuBA,QAAO,IAAI,QAAQ,EAC7C,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,GAAG,EAC/B,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK;AAGzB,YAAM,gCAAgC,KAAK,UAAU,oBAAoB,CAAC,EAAE;AAE5E,YAAM,eAAe,qBAAqB,OAAO,CAAC,YAAY,WAAW,QAAQ,MAAM,YAAY,CAAC;AAEpG,YAAM,2CAA2C,KAAK,UAAU,YAAY,CAAC,EAAE;AAE/E,YAAMG,gBAAe,oBAAoB,YAAY;AAErD,aAAOA;AAAA,IACT;AAEA,aAAS,oBAAoB,UAAU;AACrC,YAAM,YAAY;AAClB,YAAM,QAAQ,SAAS,IAAI,aAAW,QAAQ,MAAM,SAAS,CAAC,EAC3D,KAAK;AACR,YAAM,GAAG,KAAK,UAAU,KAAK,CAAC,EAAE;AAChC,YAAM,YAAY,CAAC,GAAG,IAAI,IAAI,KAAK,CAAC,EAAE,KAAK;AAC3C,YAAM,GAAG,KAAK,UAAU,SAAS,CAAC,EAAE;AACpC,aAAO;AAAA,IACT;AAAA,EAEF;AAAA,EACA,SAAS,CAAC,GAAG,MAAM;AACjB,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,aAAa,KAAK,UAAU,CAAC,CAAC,OAAO,KAAK,UAAU,CAAC,CAAC,EAAE;AAE9D,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AACjD,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AAEjD,UAAM,aAAa,KAAK,UAAU,MAAM,CAAC,OAAO,KAAK,UAAU,MAAM,CAAC,EAAE;AAExE,QAAI,WAAW,QAAQ;AACrB,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,mBAAmB,MAAM,KAAK,CAAC,mBAAmB,MAAM,GAAG;AAC9D,aAAO;AAAA,IACT;AAEA,UAAM,eAAe,SAAS,QAAQ,EAAE;AACxC,UAAM,eAAe,SAAS,QAAQ,EAAE;AAIxC,QAAI,EAAE,MAAM,YAAY,KAAK,MAAM,YAAY,OAC5C,eAAe,MAAM,gBAAgB,eAAe,MAAM,eAAe;AAC1E,aAAO;AAAA,IACT;AAKA,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAC3D,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAE3D,QAAI,mBAAmB,KAAK,mBAAmB,GAAG;AAChD,aAAO;AAAA,IACT;AAGA,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAC/D,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAE/D,QAAI,oBAAoB,KAAK,oBAAoB,GAAG;AAClD,aAAO;AAAA,IACT;AAEA,WAAO;AAAA,EAET;AACF;",
+  "sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We should also get publishing time type from f008\n// We should get reprint times from f500 $a \"Lis\u00E4painos/Lis\u00E4painokset:\"\n\nexport default () => ({\n  name: 'Publication time, allow consequent years, years from multiple sources',\n  extract: ({record, recordExternal}) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const f008Values = extractF008Values(record);\n    debug(`${label}: f008: ${JSON.stringify(f008Values)}`);\n\n    const f26xValues = extractF26xValues(record);\n    debug(`${label}: f26x: ${JSON.stringify(f26xValues)}`);\n\n    const f500Values = extractF500Years(record);\n    debug(`${label}: f500: ${JSON.stringify(f500Values)}`);\n\n    // We should get copyrightYear from f008Date2 to copyrightYears when f008YearType = 'r'\n    // Is the original year (f008Date2) in f008YearType === 'r' comparable to copyrightYear?\n    // We should handle unknown years (here or comparison?)\n    // We should handle year ranges for continuing resources / collections\n\n    const normalYears = [...new Set(f26xValues.normalYears.concat(f008Values.f008Date1).filter(value => value && value !== '    ' && value !== '||||'))].sort();\n    const copyrightYears = [...new Set(f26xValues.copyrightYears)].sort();\n    const reprintYears = [...new Set(f500Values)].sort();\n\n    const combined = {normalYears, copyrightYears, reprintYears};\n\n    debug(`Combined: ${JSON.stringify(combined)}`);\n\n    return combined;\n\n    function extractF008Values(record) {\n      // Record should have only one f008 - in case of several, we handle the first one\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      if (value && testStringOrNumber(value)) {\n        const f008Date1 = extractF008Date1(value);\n        const f008Date2 = extractF008Date2(value);\n        const f008YearType = extractF008YearType(value);\n        return {f008Date1, f008Date2, f008YearType};\n      }\n      return {f008Date1: undefined, f008Date2: undefined, f008YearType: undefined};\n\n      function extractF008Date1(value) {\n        return String(value).slice(7, 11);\n      }\n\n      function extractF008Date2(value) {\n        return String(value).slice(11, 15);\n      }\n\n      function extractF008YearType(value) {\n        return String(value).slice(6, 7);\n      }\n    }\n\n    function extractF26xValues(record) {\n      const copyrightRegex = /^(?<copyrightPrefix>cop|cop.|c|\u00A9|p|\u2117)/u;\n\n      const pubNormalSubFieldValues = extractPubNormalSubfieldValues(record, copyrightRegex);\n      debug(`Normal years: ${JSON.stringify(pubNormalSubFieldValues)}`);\n\n      const pubF264CopySubFieldValues = extractPubF264CopySubfieldValues(record);\n      debug(`F264 copyright years: ${JSON.stringify(pubF264CopySubFieldValues)}`);\n\n      const pubF260CopySubFieldValues = extractPubF260CopySubfieldValues(record, copyrightRegex);\n      debug(`F260 copyright years: ${JSON.stringify(pubF260CopySubFieldValues)}`);\n\n      return {normalYears: pubNormalSubFieldValues, copyrightYears: [...pubF260CopySubFieldValues, ...pubF264CopySubFieldValues]};\n\n      function extractPubNormalSubfieldValues(record, copyrightRegex) {\n        return record.get(/^26[04]$/u)\n          .filter((field) => !(field.tag === '264' && field.ind2 === '4'))\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value && !copyrightRegex.test(value))\n          .map(({value}) => value)\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function extractPubF264CopySubfieldValues(record, copyrightRegex) {\n        return record.get(/^264$/u)\n          .filter((field) => field.ind2 === '4')\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value)\n          .map(({value}) => value)\n          .map((value) => value.replace(copyrightRegex, ''))\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function extractPubF260CopySubfieldValues(record, copyrightRegex) {\n        return record.get(/^260$/u)\n          .map(({subfields}) => subfields)\n          .flat()\n          .filter(({code}) => code && code === 'c')\n          .filter(({value}) => value && copyrightRegex.test(value))\n          .map(({value}) => value)\n          .map((value) => value.replace(copyrightRegex, ''))\n          .map((value) => removeNonAlphaNumeric(value));\n      }\n\n      function removeNonAlphaNumeric(value) {\n        debug(`Cleaning: ${JSON.stringify(value)}`);\n        const nonAlphaNumericRegex = /[^A-Za-z0-9]/ug;\n        return value ? value.replace(nonAlphaNumericRegex, '') : value;\n      }\n    }\n\n    function extractF500Years(record) {\n\n      const reprintRegex = /(?<reprint>Lis\u00E4painokset:|Lis\u00E4painos:)/u;\n      const reprintFieldContents = record.get(/^500$/u)\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a')\n        .map(({value}) => value);\n      //.filter(value => value.test(reprintRegex));\n\n      debug(`f500 reprint field contents: ${JSON.stringify(reprintFieldContents)}`);\n\n      const filteredF500 = reprintFieldContents.filter((content) => content && content.match(reprintRegex));\n\n      debug(`f500 reprint field contents (filtered): ${JSON.stringify(filteredF500)}`);\n\n      const reprintYears = extractReprintYears(filteredF500);\n\n      return reprintYears;\n    }\n\n    function extractReprintYears(contents) {\n      const yearRegex = /[0-9][0-9][0-9][0-9]/gu;\n      const years = contents.map(content => content.match(yearRegex))\n        .flat();\n      debug(`${JSON.stringify(years)}`);\n      const uniqYears = [...new Set(years)].sort();\n      debug(`${JSON.stringify(uniqYears)}`);\n      return uniqYears;\n    }\n\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/publication-time-allow-cons-years-multi');\n    debug(`Comparing ${JSON.stringify(a)} to ${JSON.stringify(b)}`);\n\n    const [firstA] = a.normalYears ? a.normalYears : a;\n    const [firstB] = b.normalYears ? b.normalYears : b;\n\n    debug(`Comparing ${JSON.stringify(firstA)} to ${JSON.stringify(firstB)}`);\n\n    if (firstA === firstB) {\n      return 0.1;\n    }\n\n    // If either of years is a non string/number, values are not comparable\n    if (!testStringOrNumber(firstA) || !testStringOrNumber(firstB)) {\n      return 0;\n    }\n\n    const firstANumber = parseInt(firstA, 10);\n    const firstBNumber = parseInt(firstB, 10);\n\n    // Handle consequent years as a match\n    // see publication-time for a version that does not handle consequent years as a match\n    if (!(isNaN(firstANumber) || isNaN(firstBNumber)) &&\n      (firstANumber + 1 === firstBNumber || firstANumber - 1 === firstBNumber)) {\n      return 0.1;\n    }\n\n    // We should do something with copyrightYears, too\n\n    // Do not give minus points if a normal publishing year is found in normal years\n    const bNormalInANormal = a.normalYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n    const aNormalInBNormal = b.normalYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n    debug(`BNorm in ANorm: ${JSON.stringify(bNormalInANormal)}`);\n    debug(`ANorm in BNorm: ${JSON.stringify(aNormalInBNormal)}`);\n\n    if (bNormalInANormal > 0 || aNormalInBNormal > 0) {\n      return 0;\n    }\n\n    // Do not give minus points if a normal publishing year is found in reprint years\n    const bNormalInAReprint = a.reprintYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n    const aNormalInBReprint = b.reprintYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n    debug(`BNorm in AReprint: ${JSON.stringify(bNormalInAReprint)}`);\n    debug(`ANorm in BReprint: ${JSON.stringify(aNormalInBReprint)}`);\n\n    if (bNormalInAReprint > 0 || aNormalInBReprint > 0) {\n      return 0;\n    }\n\n    // Do not give minus points if a normal publishing year is found in copyright years\n    const bNormalInACopyright = a.copyrightYears.filter(aValue => b.normalYears.some(bValue => aValue === bValue));\n    const aNormalInBCopyright= b.copyrightYears.filter(bValue => a.normalYears.some(aValue => bValue === aValue));\n    debug(`BNorm in ACopyright: ${JSON.stringify(bNormalInACopyright)}`);\n    debug(`ANorm in BCopyright: ${JSON.stringify(aNormalInBCopyright)}`);\n\n    if (bNormalInACopyright > 0 || aNormalInBCopyright > 0) {\n      return 0;\n    }\n\n\n\n\n\n    return -1;\n\n  }\n});\n\n// https://www.loc.gov/marc/bibliographic/bd008.html\n// field 008\n// 06 - Type of date/Publication status\n// 07-10 - Date 1\n// 11-14 - Date 2\n//\n// 06 - Type of date/Publication status\n// b - No dates given; B.C. date involved\n// c - Continuing resource currently published\n// d - Continuing resource ceased publication\n// e - Detailed date\n// i - Inclusive dates of collection\n// k - Range of years of bulk of collection\n// m - Multiple dates\n// n - Dates unknown\n// p - Date of distribution/release/issue and production/recording session when different\n// q - Questionable date\n// r - Reprint/reissue date and original date\n// s - Single known date/probable date\n// t - Publication date and copyright date\n// u - Continuing resource status unknown\n// | - No attempt to code\n//\n// 07-10 - Date 1\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n// 11-14 - Date 2\n// 1-9 - Date digit\n// # - Date element is not applicable\n// u - Date element is totally or partially unknown\n// |||| - No attempt to code\n//\n\n"],
+  "mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,0BAAyB;AAMjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,QAAQ,eAAc,MAAM;AACrC,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,QAAQ,kBAAkB,eAAe,QAAQ,eAAe,QAAQ;AAE9E,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,kBAAkB,MAAM;AAC3C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAErD,UAAM,aAAa,iBAAiB,MAAM;AAC1C,UAAM,GAAG,KAAK,WAAW,KAAK,UAAU,UAAU,CAAC,EAAE;AAOrD,UAAM,cAAc,CAAC,GAAG,IAAI,IAAI,WAAW,YAAY,OAAO,WAAW,SAAS,EAAE,OAAO,WAAS,SAAS,UAAU,UAAU,UAAU,MAAM,CAAC,CAAC,EAAE,KAAK;AAC1J,UAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,WAAW,cAAc,CAAC,EAAE,KAAK;AACpE,UAAM,eAAe,CAAC,GAAG,IAAI,IAAI,UAAU,CAAC,EAAE,KAAK;AAEnD,UAAM,WAAW,EAAC,aAAa,gBAAgB,aAAY;AAE3D,UAAM,aAAa,KAAK,UAAU,QAAQ,CAAC,EAAE;AAE7C,WAAO;AAEP,aAAS,kBAAkBA,SAAQ;AAEjC,YAAM,QAAQA,QAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,UAAI,SAAS,mBAAmB,KAAK,GAAG;AACtC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,YAAY,iBAAiB,KAAK;AACxC,cAAM,eAAe,oBAAoB,KAAK;AAC9C,eAAO,EAAC,WAAW,WAAW,aAAY;AAAA,MAC5C;AACA,aAAO,EAAC,WAAW,QAAW,WAAW,QAAW,cAAc,OAAS;AAE3E,eAAS,iBAAiBC,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,EAAE;AAAA,MAClC;AAEA,eAAS,iBAAiBA,QAAO;AAC/B,eAAO,OAAOA,MAAK,EAAE,MAAM,IAAI,EAAE;AAAA,MACnC;AAEA,eAAS,oBAAoBA,QAAO;AAClC,eAAO,OAAOA,MAAK,EAAE,MAAM,GAAG,CAAC;AAAA,MACjC;AAAA,IACF;AAEA,aAAS,kBAAkBD,SAAQ;AACjC,YAAM,iBAAiB;AAEvB,YAAM,0BAA0B,+BAA+BA,SAAQ,cAAc;AACrF,YAAM,iBAAiB,KAAK,UAAU,uBAAuB,CAAC,EAAE;AAEhE,YAAM,4BAA4B,iCAAiCA,OAAM;AACzE,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,YAAM,4BAA4B,iCAAiCA,SAAQ,cAAc;AACzF,YAAM,yBAAyB,KAAK,UAAU,yBAAyB,CAAC,EAAE;AAE1E,aAAO,EAAC,aAAa,yBAAyB,gBAAgB,CAAC,GAAG,2BAA2B,GAAG,yBAAyB,EAAC;AAE1H,eAAS,+BAA+BA,SAAQE,iBAAgB;AAC9D,eAAOF,QAAO,IAAI,WAAW,EAC1B,OAAO,CAAC,UAAU,EAAE,MAAM,QAAQ,SAAS,MAAM,SAAS,IAAI,EAC9D,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAAS,CAACE,gBAAe,KAAK,KAAK,CAAC,EACxD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,OAAO,CAAC,UAAU,MAAM,SAAS,GAAG,EACpC,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,KAAK,EACzB,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQE,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,iCAAiCF,SAAQE,iBAAgB;AAChE,eAAOF,QAAO,IAAI,QAAQ,EACvB,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,QAAQ,SAAS,GAAG,EACvC,OAAO,CAAC,EAAC,MAAK,MAAM,SAASE,gBAAe,KAAK,KAAK,CAAC,EACvD,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK,EACtB,IAAI,CAAC,UAAU,MAAM,QAAQA,iBAAgB,EAAE,CAAC,EAChD,IAAI,CAAC,UAAU,sBAAsB,KAAK,CAAC;AAAA,MAChD;AAEA,eAAS,sBAAsB,OAAO;AACpC,cAAM,aAAa,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1C,cAAM,uBAAuB;AAC7B,eAAO,QAAQ,MAAM,QAAQ,sBAAsB,EAAE,IAAI;AAAA,MAC3D;AAAA,IACF;AAEA,aAAS,iBAAiBF,SAAQ;AAEhC,YAAM,eAAe;AACrB,YAAM,uBAAuBA,QAAO,IAAI,QAAQ,EAC7C,IAAI,CAAC,EAAC,UAAS,MAAM,SAAS,EAC9B,KAAK,EACL,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,GAAG,EAC/B,IAAI,CAAC,EAAC,MAAK,MAAM,KAAK;AAGzB,YAAM,gCAAgC,KAAK,UAAU,oBAAoB,CAAC,EAAE;AAE5E,YAAM,eAAe,qBAAqB,OAAO,CAAC,YAAY,WAAW,QAAQ,MAAM,YAAY,CAAC;AAEpG,YAAM,2CAA2C,KAAK,UAAU,YAAY,CAAC,EAAE;AAE/E,YAAMG,gBAAe,oBAAoB,YAAY;AAErD,aAAOA;AAAA,IACT;AAEA,aAAS,oBAAoB,UAAU;AACrC,YAAM,YAAY;AAClB,YAAM,QAAQ,SAAS,IAAI,aAAW,QAAQ,MAAM,SAAS,CAAC,EAC3D,KAAK;AACR,YAAM,GAAG,KAAK,UAAU,KAAK,CAAC,EAAE;AAChC,YAAM,YAAY,CAAC,GAAG,IAAI,IAAI,KAAK,CAAC,EAAE,KAAK;AAC3C,YAAM,GAAG,KAAK,UAAU,SAAS,CAAC,EAAE;AACpC,aAAO;AAAA,IACT;AAAA,EAEF;AAAA,EACA,SAAS,CAAC,GAAG,MAAM;AACjB,UAAM,QAAQ,kBAAkB,wGAAwG;AACxI,UAAM,aAAa,KAAK,UAAU,CAAC,CAAC,OAAO,KAAK,UAAU,CAAC,CAAC,EAAE;AAE9D,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AACjD,UAAM,CAAC,MAAM,IAAI,EAAE,cAAc,EAAE,cAAc;AAEjD,UAAM,aAAa,KAAK,UAAU,MAAM,CAAC,OAAO,KAAK,UAAU,MAAM,CAAC,EAAE;AAExE,QAAI,WAAW,QAAQ;AACrB,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,mBAAmB,MAAM,KAAK,CAAC,mBAAmB,MAAM,GAAG;AAC9D,aAAO;AAAA,IACT;AAEA,UAAM,eAAe,SAAS,QAAQ,EAAE;AACxC,UAAM,eAAe,SAAS,QAAQ,EAAE;AAIxC,QAAI,EAAE,MAAM,YAAY,KAAK,MAAM,YAAY,OAC5C,eAAe,MAAM,gBAAgB,eAAe,MAAM,eAAe;AAC1E,aAAO;AAAA,IACT;AAKA,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,EAAE,YAAY,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACvG,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAC3D,UAAM,mBAAmB,KAAK,UAAU,gBAAgB,CAAC,EAAE;AAE3D,QAAI,mBAAmB,KAAK,mBAAmB,GAAG;AAChD,aAAO;AAAA,IACT;AAGA,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,oBAAoB,EAAE,aAAa,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AACzG,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAC/D,UAAM,sBAAsB,KAAK,UAAU,iBAAiB,CAAC,EAAE;AAE/D,QAAI,oBAAoB,KAAK,oBAAoB,GAAG;AAClD,aAAO;AAAA,IACT;AAGA,UAAM,sBAAsB,EAAE,eAAe,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AAC7G,UAAM,sBAAqB,EAAE,eAAe,OAAO,YAAU,EAAE,YAAY,KAAK,YAAU,WAAW,MAAM,CAAC;AAC5G,UAAM,wBAAwB,KAAK,UAAU,mBAAmB,CAAC,EAAE;AACnE,UAAM,wBAAwB,KAAK,UAAU,mBAAmB,CAAC,EAAE;AAEnE,QAAI,sBAAsB,KAAK,sBAAsB,GAAG;AACtD,aAAO;AAAA,IACT;AAMA,WAAO;AAAA,EAET;AACF;",
   "names": ["record", "value", "copyrightRegex", "reprintYears"]
 }

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"]) {