@natlibfi/melinda-record-matching 3.0.1 → 3.0.3-alpha.1

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

@@ -3,6 +3,7 @@
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
+exports.bibAuthors = bibAuthors;
 exports.bibHostComponents = bibHostComponents;
 exports.bibMelindaIds = bibMelindaIds;
 exports.bibSourceIds = bibSourceIds;
@@ -171,20 +172,78 @@ function bibTitle(record) {
 
     // use word search for titles starting with a boolean
     const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));
-    // Prevent too many matches by having a minimum length requirement
-    return formatted.length >= 5 ? [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`] : [];
+    // Prevent too many matches by having a minimum length
+    // Note that currently this fails matching if there are no matches from previous matchers
+    if (formatted.length >= 5) {
+      return [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`];
+    }
+    return addAuthorsToSearch(`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`);
   }
   return [];
+  function addAuthorsToSearch(titleQuery) {
+    const [authorQuery] = bibAuthors(record);
+    if (authorQuery !== undefined) {
+      return [`${authorQuery} AND ${titleQuery}`];
+    }
+    return [];
+  }
   function getTitle() {
     const [field] = record.get(/^245$/u);
     if (field) {
-      return field.subfields.filter(({
+      const titleString = field.subfields
+      //.filter(({code}) => ['a', 'b', 'n', 'p'].includes(code))
+      .filter(({
+        code
+      }) => ['a', 'b'].includes(code))
+      //.filter(({code}) => ['a'].includes(code))
+      .map(({
+        value
+      }) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').filter(value => value)
+      // In Melinda's index subfield separators are indexed as ' '
+      .join(' ');
+      return titleString;
+    }
+    return false;
+  }
+}
+function bibAuthors(record) {
+  const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');
+  const debugData = debug.extend('data');
+  debug(`Creating query for the first author`);
+  //debugData(record);
+
+  const author = getAuthor(record);
+  const booleanStartWords = ['and', 'or', 'nor', 'not'];
+  if ((0, _matchingUtils.testStringOrNumber)(author)) {
+    const formatted = String(author).replace(/[^\w\s\p{Alphabetic}]/gu, '')
+    // Clean up concurrent spaces from fe. subfield changes
+    .replace(/ +/gu, ' ').trim().slice(0, 30).trim();
+
+    // use word search for authors starting with a boolean
+    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));
+    // Prevent too many matches by having a minimum length
+    debugData(`Author string: ${formatted}`);
+    if (formatted.length >= 5) {
+      return [`dc.author="${useWordSearch ? '' : '^'}${formatted}*"`];
+    }
+    return [];
+  }
+  return [];
+  function getAuthor(record) {
+    //debugData(record);
+    // eslint-disable-next-line prefer-named-capture-group
+    const [field] = record.get(/^(100)|(110)|(111)|(700)|(710)|(711)$/u);
+    //debugData(field);
+
+    if (field) {
+      const authorString = field.subfields.filter(({
         code
-      }) => ['a', 'b'].includes(code)).map(({
+      }) => ['a'].includes(code)).map(({
         value
       }) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').filter(value => value)
       // In Melinda's index subfield separators are indexed as ' '
       .join(' ');
+      return authorString;
     }
     return false;
   }

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

@@ -1 +1 @@
-{"version":3,"file":"bib.js","names":["bibSourceIds","record","debug","createDebugLogger","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","toQueries","map","toSidString","filter","nonEmptySid","field","validateSidFieldSubfieldCounts","createSidString","sfC","getSubfieldValues","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","testStringOrNumber","String","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","getMelindaIdsF035","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","booleanStartWords","formatted","trim","slice","useWordSearch","some","word","toLowerCase","startsWith","includes","join","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub"],"sources":["../../../src/candidate-search/query-list/bib.js"],"sourcesContent":["\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils';\n\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const id = getHostId();\n  return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];\n\n  function getHostId() {\n    const [field] = record.get(/^773$/u);\n\n    if (field) {\n      const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n      if (testStringOrNumber(value) && (/^\\(FI-MELINDA\\)/u).test(String(value))) {\n        return String(value).replace(/^\\(FI-MELINDA\\)/u, '');\n      }\n\n      if (testStringOrNumber(value) && (/^\\(FIN01\\)/u).test(String(value))) {\n        return String(value).replace(/^\\(FIN01\\)/u, '');\n      }\n\n      return false;\n    }\n    return false;\n  }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n  const title = getTitle();\n  const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n  if (testStringOrNumber(title)) {\n    const formatted = String(title)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for titles starting with a boolean\n    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n    // Prevent too many matches by having a minimum length requirement\n    return formatted.length >= 5 ? [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`] : [];\n  }\n\n  return [];\n\n  function getTitle() {\n    const [field] = record.get(/^245$/u);\n\n    if (field) {\n      return field.subfields\n        .filter(({code}) => ['a', '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    }\n    return false;\n  }\n}\n\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  const fields = record.get(/^(?<def>020|022|024)$/u);\n  const identifiers = [].concat(...fields.map(toIdentifiers));\n  const uniqueIdentifiers = [...new Set(identifiers)];\n\n  debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n  debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n  debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n  if (uniqueIdentifiers.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n  function toIdentifiers({tag, subfields}) {\n    const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n    const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n    if (tag === '022') {\n      return subfields\n        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))\n      .map(({value}) => String(value));\n  }\n}\n"],"mappings":";;;;;;;;;;AA4BA;AACA;AACA;AAA8H;AA7B9H;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAMO,SAASA,YAAY,CAACC,MAAM,EAAE;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtC;;EAEAH,KAAK,CAAE,iCAAgC,CAAC;EAExC,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAG,CAAC,KAAK,CAAC;EAC/BH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACJ,KAAK,CAAE,EAAC,CAAC;EAEnE,OAAOA,KAAK,CAACE,MAAM,GAAG,CAAC,GAAGG,YAAY,CAACL,KAAK,CAAC,GAAG,EAAE;EAElD,SAASK,YAAY,CAACL,KAAK,EAAE;IAC3BJ,KAAK,CAAE,wCAAuC,CAAC;IAE/C,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAK,CAAC;IAEvC,IAAIM,UAAU,CAACJ,MAAM,GAAG,CAAC,EAAE;MACzBN,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO,EAAE;IACX;IAEA,MAAMY,UAAU,GAAG,IAAAC,+BAAS,EAACH,UAAU,EAAE,kBAAkB,CAAC;IAE5D,OAAOE,UAAU;IAEjB,SAASD,aAAa,CAACP,KAAK,EAAE;MAC5BJ,KAAK,CAAE,qCAAoC,CAAC;;MAE5C;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAG,CAACC,WAAW,CAAC,CAACC,MAAM,CAACC,WAAW,IAAIA,WAAW,CAAC;MAC5E,OAAOP,UAAU;MAEjB,SAASK,WAAW,CAACG,KAAK,EAAE;QAC1BlB,KAAK,CAAE,6BAA4B,CAAC;QAEpC,OAAO,IAAAmB,6CAA8B,EAACD,KAAK,CAAC,GAAGE,eAAe,CAACF,KAAK,CAAC,GAAG,EAAE;QAE1E,SAASE,eAAe,CAACF,KAAK,EAAE;UAC9BlB,KAAK,CAAE,8BAA6B,CAAC;UACrC,MAAM,CAACqB,GAAG,CAAC,GAAG,IAAAC,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAC3C,MAAM,CAACK,GAAG,CAAC,GAAG,IAAAD,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAE3C,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAG,CAAC,CAAC;UACrE,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAG,CAAC;UAEjDrB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACa,GAAG,CAAE,MAAKd,IAAI,CAACC,SAAS,CAACe,GAAG,CAAE,EAAC,CAAC;UAC5D,OAAOC,UAAU,CAACI,MAAM,CAACD,UAAU,CAAC;QACtC;QAEA,SAASF,kBAAkB,CAACI,aAAa,EAAE;UACzC,MAAMC,iBAAiB,GAAI,gDAAiD;UAC5E,MAAMC,eAAe,GAAG,IAAAC,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACJ,iBAAiB,EAAE,OAAO,CAAC,GAAG,EAAE;UAC1H5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAC,CAAC;UAC9D,OAAOA,eAAe;QACxB;QAEA,SAASL,yBAAyB,CAACG,aAAa,EAAE;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAAC,CAAC;UACzC,MAAMM,kBAAkB,GAAI,OAAQ;UACpC,OAAO,IAAAH,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACC,kBAAkB,EAAE,EAAE,CAAC,GAAG,EAAE;QACvG;MAEF;IACF;EACF;AACF;AAEO,SAASC,aAAa,CAACrC,MAAM,EAAE;EACpC;;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wEAAwE,CAAC;EACzG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,iCAAgC,CAAC;;EAExC;EACA;EACA;EACA,MAAMqC,UAAU,GAAG,IAAAC,gCAAiB,EAACvC,MAAM,CAAC;EAE5CG,SAAS,CAAE,uBAAsBmC,UAAU,CAAC/B,MAAO,MAAKC,IAAI,CAACC,SAAS,CAAC6B,UAAU,CAAE,EAAC,CAAC;EAErF,IAAIA,UAAU,CAAC/B,MAAM,GAAG,CAAC,EAAE;IACzBN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACwB,UAAU,EAAE,mBAAmB,CAAC;AACnD;;AAGA;AACA;AACO,SAASE,iBAAiB,CAACxC,MAAM,EAAE;EACxC,MAAMyC,EAAE,GAAGC,SAAS,EAAE;EACtB,OAAO,IAAAT,iCAAkB,EAACQ,EAAE,CAAC,GAAG,CAAE,uBAAsBA,EAAG,EAAC,CAAC,GAAG,EAAE;EAElE,SAASC,SAAS,GAAG;IACnB,MAAM,CAACvB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM;QAACwB;MAAK,CAAC,GAAGxB,KAAK,CAACyB,SAAS,CAACC,IAAI,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC;MAEpE,IAAI,IAAAb,iCAAkB,EAACU,KAAK,CAAC,IAAK,kBAAkB,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACzE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,kBAAkB,EAAE,EAAE,CAAC;MACtD;MAEA,IAAI,IAAAF,iCAAkB,EAACU,KAAK,CAAC,IAAK,aAAa,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACpE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC;MACjD;MAEA,OAAO,KAAK;IACd;IACA,OAAO,KAAK;EACd;AACF;;AAEA;AACA;AACA;;AAEA;AACA;AACA;;AAEO,SAASa,QAAQ,CAAChD,MAAM,EAAE;EAC/B,MAAMiD,KAAK,GAAGC,QAAQ,EAAE;EACxB,MAAMC,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAErD,IAAI,IAAAlB,iCAAkB,EAACgB,KAAK,CAAC,EAAE;IAC7B,MAAMG,SAAS,GAAGlB,MAAM,CAACe,KAAK,CAAC,CAC5Bd,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,EAAE,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,EAAE;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,EAAE,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACA,OAAOL,SAAS,CAAC7C,MAAM,IAAI,CAAC,GAAG,CAAE,aAAYgD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC,GAAG,EAAE;EAC7F;EAEA,OAAO,EAAE;EAET,SAASF,QAAQ,GAAG;IAClB,MAAM,CAAC/B,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,OAAOA,KAAK,CAACyB,SAAS,CACnB3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACc,QAAQ,CAACd,IAAI,CAAC,CAAC,CAC7C/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACCkB,IAAI,CAAC,GAAG,CAAC;IACd;IACA,OAAO,KAAK;EACd;AACF;AAEO,SAASC,sBAAsB,CAAC9D,MAAM,EAAE;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;EAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,2CAA0C,CAAC;EAElD,MAAM8D,MAAM,GAAG/D,MAAM,CAACM,GAAG,CAAC,wBAAwB,CAAC;EACnD,MAAM0D,WAAW,GAAG,EAAE,CAACnC,MAAM,CAAC,GAAGkC,MAAM,CAAChD,GAAG,CAACkD,aAAa,CAAC,CAAC;EAC3D,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACH,WAAW,CAAC,CAAC;EAEnD7D,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAS,CAACsD,MAAM,CAAE,EAAC,CAAC;EAClE5D,SAAS,CAAE,gBAAe6D,WAAW,CAACzD,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACuD,WAAW,CAAE,EAAC,CAAC;EAChF7D,SAAS,CAAE,uBAAsB+D,iBAAiB,CAAC3D,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACyD,iBAAiB,CAAE,EAAC,CAAC;EAEnG,IAAIA,iBAAiB,CAAC3D,MAAM,GAAG,CAAC,EAAE;IAChCN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACoD,iBAAiB,EAAE,eAAe,CAAC;EAEpD,SAASD,aAAa,CAAC;IAACG,GAAG;IAAExB;EAAS,CAAC,EAAE;IACvC,MAAMyB,cAAc,GAAI,kBAAmB;IAC3C,MAAMC,aAAa,GAAI,mBAAoB;IAE3C,IAAIF,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOxB,SAAS,CACb3B,MAAM,CAACsD,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAACX,QAAQ,CAACW,GAAG,CAACzB,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsC,GAAG,CAAC5B,KAAK,CAAC,IAAI0B,cAAc,CAACtB,IAAI,CAACb,MAAM,CAACqC,GAAG,CAAC5B,KAAK,CAAC,CAAC,CAAC,CAC5H5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,IAAIyB,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOxB,SAAS,CACb3B,MAAM,CAACsD,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACX,QAAQ,CAACW,GAAG,CAACzB,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsC,GAAG,CAAC5B,KAAK,CAAC,IAAI0B,cAAc,CAACtB,IAAI,CAACb,MAAM,CAACqC,GAAG,CAAC5B,KAAK,CAAC,CAAC,CAAC,CACvH5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,OAAOC,SAAS,CACb3B,MAAM,CAACsD,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACX,QAAQ,CAACW,GAAG,CAACzB,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAACsC,GAAG,CAAC5B,KAAK,CAAC,IAAI2B,aAAa,CAACvB,IAAI,CAACb,MAAM,CAACqC,GAAG,CAAC5B,KAAK,CAAC,CAAC,CAAC,CACtH5B,GAAG,CAAC,CAAC;MAAC4B;IAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;EACpC;AACF"}
+{"version":3,"file":"bib.js","names":["bibSourceIds","record","debug","createDebugLogger","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","toQueries","map","toSidString","filter","nonEmptySid","field","validateSidFieldSubfieldCounts","createSidString","sfC","getSubfieldValues","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","testStringOrNumber","String","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","getMelindaIdsF035","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","booleanStartWords","formatted","trim","slice","useWordSearch","some","word","toLowerCase","startsWith","addAuthorsToSearch","titleQuery","authorQuery","bibAuthors","undefined","titleString","includes","join","author","getAuthor","authorString","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub"],"sources":["../../../src/candidate-search/query-list/bib.js"],"sourcesContent":["\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils';\n\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const id = getHostId();\n  return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];\n\n  function getHostId() {\n    const [field] = record.get(/^773$/u);\n\n    if (field) {\n      const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n      if (testStringOrNumber(value) && (/^\\(FI-MELINDA\\)/u).test(String(value))) {\n        return String(value).replace(/^\\(FI-MELINDA\\)/u, '');\n      }\n\n      if (testStringOrNumber(value) && (/^\\(FIN01\\)/u).test(String(value))) {\n        return String(value).replace(/^\\(FIN01\\)/u, '');\n      }\n\n      return false;\n    }\n    return false;\n  }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n  const title = getTitle();\n  const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n  if (testStringOrNumber(title)) {\n    const formatted = String(title)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for titles starting with a boolean\n    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n    // Prevent too many matches by having a minimum length\n    // Note that currently this fails matching if there are no matches from previous matchers\n    if (formatted.length >= 5) {\n      return [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n    }\n    return addAuthorsToSearch(`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`);\n  }\n\n  return [];\n\n  function addAuthorsToSearch(titleQuery) {\n    const [authorQuery] = bibAuthors(record);\n    if (authorQuery !== undefined) {\n      return [`${authorQuery} AND ${titleQuery}`];\n    }\n    return [];\n  }\n\n  function getTitle() {\n    const [field] = record.get(/^245$/u);\n\n    if (field) {\n      const titleString = field.subfields\n        //.filter(({code}) => ['a', 'b', 'n', 'p'].includes(code))\n        .filter(({code}) => ['a', 'b'].includes(code))\n        //.filter(({code}) => ['a'].includes(code))\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n      return titleString;\n    }\n    return false;\n  }\n}\n\nexport function bibAuthors(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');\n  const debugData = debug.extend('data');\n  debug(`Creating query for the first author`);\n  //debugData(record);\n\n  const author = getAuthor(record);\n  const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n\n  if (testStringOrNumber(author)) {\n    const formatted = String(author)\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for authors starting with a boolean\n    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n    // Prevent too many matches by having a minimum length\n    debugData(`Author string: ${formatted}`);\n    if (formatted.length >= 5) {\n      return [`dc.author=\"${useWordSearch ? '' : '^'}${formatted}*\"`];\n    }\n    return [];\n  }\n\n  return [];\n\n  function getAuthor(record) {\n    //debugData(record);\n    // eslint-disable-next-line prefer-named-capture-group\n    const [field] = record.get(/^(100)|(110)|(111)|(700)|(710)|(711)$/u);\n    //debugData(field);\n\n    if (field) {\n      const authorString = field.subfields\n        .filter(({code}) => ['a'].includes(code))\n        .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n        .filter(value => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n      return authorString;\n    }\n    return false;\n  }\n}\n\n\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  const fields = record.get(/^(?<def>020|022|024)$/u);\n  const identifiers = [].concat(...fields.map(toIdentifiers));\n  const uniqueIdentifiers = [...new Set(identifiers)];\n\n  debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n  debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n  debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n  if (uniqueIdentifiers.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n  function toIdentifiers({tag, subfields}) {\n    const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n    const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n    if (tag === '022') {\n      return subfields\n        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))\n        .map(({value}) => String(value));\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))\n      .map(({value}) => String(value));\n  }\n}\n"],"mappings":";;;;;;;;;;;AA4BA;AACA;AACA;AAA8H;AA7B9H;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAMO,SAASA,YAAY,CAACC,MAAM,EAAE;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtC;;EAEAH,KAAK,CAAE,iCAAgC,CAAC;EAExC,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAG,CAAC,KAAK,CAAC;EAC/BH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACJ,KAAK,CAAE,EAAC,CAAC;EAEnE,OAAOA,KAAK,CAACE,MAAM,GAAG,CAAC,GAAGG,YAAY,CAACL,KAAK,CAAC,GAAG,EAAE;EAElD,SAASK,YAAY,CAACL,KAAK,EAAE;IAC3BJ,KAAK,CAAE,wCAAuC,CAAC;IAE/C,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAK,CAAC;IAEvC,IAAIM,UAAU,CAACJ,MAAM,GAAG,CAAC,EAAE;MACzBN,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO,EAAE;IACX;IAEA,MAAMY,UAAU,GAAG,IAAAC,+BAAS,EAACH,UAAU,EAAE,kBAAkB,CAAC;IAE5D,OAAOE,UAAU;IAEjB,SAASD,aAAa,CAACP,KAAK,EAAE;MAC5BJ,KAAK,CAAE,qCAAoC,CAAC;;MAE5C;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAG,CAACC,WAAW,CAAC,CAACC,MAAM,CAACC,WAAW,IAAIA,WAAW,CAAC;MAC5E,OAAOP,UAAU;MAEjB,SAASK,WAAW,CAACG,KAAK,EAAE;QAC1BlB,KAAK,CAAE,6BAA4B,CAAC;QAEpC,OAAO,IAAAmB,6CAA8B,EAACD,KAAK,CAAC,GAAGE,eAAe,CAACF,KAAK,CAAC,GAAG,EAAE;QAE1E,SAASE,eAAe,CAACF,KAAK,EAAE;UAC9BlB,KAAK,CAAE,8BAA6B,CAAC;UACrC,MAAM,CAACqB,GAAG,CAAC,GAAG,IAAAC,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAC3C,MAAM,CAACK,GAAG,CAAC,GAAG,IAAAD,gCAAiB,EAACJ,KAAK,EAAE,GAAG,CAAC;UAE3C,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAG,CAAC,CAAC;UACrE,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAG,CAAC;UAEjDrB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAS,CAACa,GAAG,CAAE,MAAKd,IAAI,CAACC,SAAS,CAACe,GAAG,CAAE,EAAC,CAAC;UAC5D,OAAOC,UAAU,CAACI,MAAM,CAACD,UAAU,CAAC;QACtC;QAEA,SAASF,kBAAkB,CAACI,aAAa,EAAE;UACzC,MAAMC,iBAAiB,GAAI,gDAAiD;UAC5E,MAAMC,eAAe,GAAG,IAAAC,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACJ,iBAAiB,EAAE,OAAO,CAAC,GAAG,EAAE;UAC1H5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAC,CAAC;UAC9D,OAAOA,eAAe;QACxB;QAEA,SAASL,yBAAyB,CAACG,aAAa,EAAE;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAAC,CAAC;UACzC,MAAMM,kBAAkB,GAAI,OAAQ;UACpC,OAAO,IAAAH,iCAAkB,EAACH,aAAa,CAAC,GAAGI,MAAM,CAACJ,aAAa,CAAC,CAACK,OAAO,CAACC,kBAAkB,EAAE,EAAE,CAAC,GAAG,EAAE;QACvG;MAEF;IACF;EACF;AACF;AAEO,SAASC,aAAa,CAACrC,MAAM,EAAE;EACpC;;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,wEAAwE,CAAC;EACzG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,iCAAgC,CAAC;;EAExC;EACA;EACA;EACA,MAAMqC,UAAU,GAAG,IAAAC,gCAAiB,EAACvC,MAAM,CAAC;EAE5CG,SAAS,CAAE,uBAAsBmC,UAAU,CAAC/B,MAAO,MAAKC,IAAI,CAACC,SAAS,CAAC6B,UAAU,CAAE,EAAC,CAAC;EAErF,IAAIA,UAAU,CAAC/B,MAAM,GAAG,CAAC,EAAE;IACzBN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAACwB,UAAU,EAAE,mBAAmB,CAAC;AACnD;;AAGA;AACA;AACO,SAASE,iBAAiB,CAACxC,MAAM,EAAE;EACxC,MAAMyC,EAAE,GAAGC,SAAS,EAAE;EACtB,OAAO,IAAAT,iCAAkB,EAACQ,EAAE,CAAC,GAAG,CAAE,uBAAsBA,EAAG,EAAC,CAAC,GAAG,EAAE;EAElE,SAASC,SAAS,GAAG;IACnB,MAAM,CAACvB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM;QAACwB;MAAK,CAAC,GAAGxB,KAAK,CAACyB,SAAS,CAACC,IAAI,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,CAAC;MAEpE,IAAI,IAAAb,iCAAkB,EAACU,KAAK,CAAC,IAAK,kBAAkB,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACzE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,kBAAkB,EAAE,EAAE,CAAC;MACtD;MAEA,IAAI,IAAAF,iCAAkB,EAACU,KAAK,CAAC,IAAK,aAAa,CAAEI,IAAI,CAACb,MAAM,CAACS,KAAK,CAAC,CAAC,EAAE;QACpE,OAAOT,MAAM,CAACS,KAAK,CAAC,CAACR,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC;MACjD;MAEA,OAAO,KAAK;IACd;IACA,OAAO,KAAK;EACd;AACF;;AAEA;AACA;AACA;;AAEA;AACA;AACA;;AAEO,SAASa,QAAQ,CAAChD,MAAM,EAAE;EAC/B,MAAMiD,KAAK,GAAGC,QAAQ,EAAE;EACxB,MAAMC,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAErD,IAAI,IAAAlB,iCAAkB,EAACgB,KAAK,CAAC,EAAE;IAC7B,MAAMG,SAAS,GAAGlB,MAAM,CAACe,KAAK,CAAC,CAC5Bd,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,EAAE,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,EAAE;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,EAAE,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACA;IACA,IAAIL,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,aAAYgD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IAChE;IACA,OAAOQ,kBAAkB,CAAE,aAAYL,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;EAClF;EAEA,OAAO,EAAE;EAET,SAASQ,kBAAkB,CAACC,UAAU,EAAE;IACtC,MAAM,CAACC,WAAW,CAAC,GAAGC,UAAU,CAAC/D,MAAM,CAAC;IACxC,IAAI8D,WAAW,KAAKE,SAAS,EAAE;MAC7B,OAAO,CAAE,GAAEF,WAAY,QAAOD,UAAW,EAAC,CAAC;IAC7C;IACA,OAAO,EAAE;EACX;EAEA,SAASX,QAAQ,GAAG;IAClB,MAAM,CAAC/B,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,QAAQ,CAAC;IAEpC,IAAIa,KAAK,EAAE;MACT,MAAM8C,WAAW,GAAG9C,KAAK,CAACyB;MACxB;MAAA,CACC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACoB,QAAQ,CAACpB,IAAI,CAAC;MAC7C;MAAA,CACC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACCwB,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOF,WAAW;IACpB;IACA,OAAO,KAAK;EACd;AACF;AAEO,SAASF,UAAU,CAAC/D,MAAM,EAAE;EACjC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,qEAAqE,CAAC;EACtG,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,qCAAoC,CAAC;EAC5C;;EAEA,MAAMmE,MAAM,GAAGC,SAAS,CAACrE,MAAM,CAAC;EAChC,MAAMmD,iBAAiB,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC;EAGrD,IAAI,IAAAlB,iCAAkB,EAACmC,MAAM,CAAC,EAAE;IAC9B,MAAMhB,SAAS,GAAGlB,MAAM,CAACkC,MAAM,CAAC,CAC7BjC,OAAO,CAAC,yBAAyB,EAAE,EAAE;IACtC;IAAA,CACCA,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CACpBkB,IAAI,EAAE,CACNC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CACZD,IAAI,EAAE;;IAET;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAI,CAACC,IAAI,IAAIL,SAAS,CAACM,WAAW,EAAE,CAACC,UAAU,CAACF,IAAI,CAAC,CAAC;IAC9F;IACAtD,SAAS,CAAE,kBAAiBiD,SAAU,EAAC,CAAC;IACxC,IAAIA,SAAS,CAAC7C,MAAM,IAAI,CAAC,EAAE;MACzB,OAAO,CAAE,cAAagD,aAAa,GAAG,EAAE,GAAG,GAAI,GAAEH,SAAU,IAAG,CAAC;IACjE;IACA,OAAO,EAAE;EACX;EAEA,OAAO,EAAE;EAET,SAASiB,SAAS,CAACrE,MAAM,EAAE;IACzB;IACA;IACA,MAAM,CAACmB,KAAK,CAAC,GAAGnB,MAAM,CAACM,GAAG,CAAC,wCAAwC,CAAC;IACpE;;IAEA,IAAIa,KAAK,EAAE;MACT,MAAMmD,YAAY,GAAGnD,KAAK,CAACyB,SAAS,CACjC3B,MAAM,CAAC,CAAC;QAAC6B;MAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAACoB,QAAQ,CAACpB,IAAI,CAAC,CAAC,CACxC/B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAK,IAAAV,iCAAkB,EAACU,KAAK,CAAC,GAAGT,MAAM,CAACS,KAAK,CAAC,GAAG,EAAE,CAAC,CAChE1B,MAAM,CAAC0B,KAAK,IAAIA,KAAK;MACtB;MAAA,CACCwB,IAAI,CAAC,GAAG,CAAC;MACZ,OAAOG,YAAY;IACrB;IACA,OAAO,KAAK;EACd;AACF;AAGO,SAASC,sBAAsB,CAACvE,MAAM,EAAE;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;EAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EACtCH,KAAK,CAAE,2CAA0C,CAAC;EAElD,MAAMuE,MAAM,GAAGxE,MAAM,CAACM,GAAG,CAAC,wBAAwB,CAAC;EACnD,MAAMmE,WAAW,GAAG,EAAE,CAAC5C,MAAM,CAAC,GAAG2C,MAAM,CAACzD,GAAG,CAAC2D,aAAa,CAAC,CAAC;EAC3D,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACH,WAAW,CAAC,CAAC;EAEnDtE,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAS,CAAC+D,MAAM,CAAE,EAAC,CAAC;EAClErE,SAAS,CAAE,gBAAesE,WAAW,CAAClE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACgE,WAAW,CAAE,EAAC,CAAC;EAChFtE,SAAS,CAAE,uBAAsBwE,iBAAiB,CAACpE,MAAO,MAAKC,IAAI,CAACC,SAAS,CAACkE,iBAAiB,CAAE,EAAC,CAAC;EAEnG,IAAIA,iBAAiB,CAACpE,MAAM,GAAG,CAAC,EAAE;IAChCN,KAAK,CAAE,2CAA0C,CAAC;IAClD,OAAO,EAAE;EACX;EAEA,OAAO,IAAAa,+BAAS,EAAC6D,iBAAiB,EAAE,eAAe,CAAC;EAEpD,SAASD,aAAa,CAAC;IAACG,GAAG;IAAEjC;EAAS,CAAC,EAAE;IACvC,MAAMkC,cAAc,GAAI,kBAAmB;IAC3C,MAAMC,aAAa,GAAI,mBAAoB;IAE3C,IAAIF,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOjC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAImC,cAAc,CAAC/B,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CAC5H5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,IAAIkC,GAAG,KAAK,KAAK,EAAE;MACjB,OAAOjC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAImC,cAAc,CAAC/B,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CACvH5B,GAAG,CAAC,CAAC;QAAC4B;MAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;IACpC;IAEA,OAAOC,SAAS,CACb3B,MAAM,CAAC+D,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACd,QAAQ,CAACc,GAAG,CAAClC,IAAI,CAAC,IAAI,IAAAb,iCAAkB,EAAC+C,GAAG,CAACrC,KAAK,CAAC,IAAIoC,aAAa,CAAChC,IAAI,CAACb,MAAM,CAAC8C,GAAG,CAACrC,KAAK,CAAC,CAAC,CAAC,CACtH5B,GAAG,CAAC,CAAC;MAAC4B;IAAK,CAAC,KAAKT,MAAM,CAACS,KAAK,CAAC,CAAC;EACpC;AACF"}

package/package.json

@@ -14,7 +14,7 @@
 		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
 	},
 	"license": "LGPL-3.0+",
-	"version": "3.0.1",
+	"version": "3.0.3-alpha.1",
 	"main": "./dist/index.js",
 	"engines": {
 		"node": ">=14"

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

@@ -196,27 +196,93 @@ export function bibTitle(record) {
 
     // use word search for titles starting with a boolean
     const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));
-    // Prevent too many matches by having a minimum length requirement
-    return formatted.length >= 5 ? [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`] : [];
+    // Prevent too many matches by having a minimum length
+    // Note that currently this fails matching if there are no matches from previous matchers
+    if (formatted.length >= 5) {
+      return [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`];
+    }
+    return addAuthorsToSearch(`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`);
   }
 
   return [];
 
+  function addAuthorsToSearch(titleQuery) {
+    const [authorQuery] = bibAuthors(record);
+    if (authorQuery !== undefined) {
+      return [`${authorQuery} AND ${titleQuery}`];
+    }
+    return [];
+  }
+
   function getTitle() {
     const [field] = record.get(/^245$/u);
 
     if (field) {
-      return field.subfields
+      const titleString = field.subfields
+        //.filter(({code}) => ['a', 'b', 'n', 'p'].includes(code))
         .filter(({code}) => ['a', 'b'].includes(code))
+        //.filter(({code}) => ['a'].includes(code))
+        .map(({value}) => testStringOrNumber(value) ? String(value) : '')
+        .filter(value => value)
+        // In Melinda's index subfield separators are indexed as ' '
+        .join(' ');
+      return titleString;
+    }
+    return false;
+  }
+}
+
+export function bibAuthors(record) {
+  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibAuthors');
+  const debugData = debug.extend('data');
+  debug(`Creating query for the first author`);
+  //debugData(record);
+
+  const author = getAuthor(record);
+  const booleanStartWords = ['and', 'or', 'nor', 'not'];
+
+
+  if (testStringOrNumber(author)) {
+    const formatted = String(author)
+      .replace(/[^\w\s\p{Alphabetic}]/gu, '')
+      // Clean up concurrent spaces from fe. subfield changes
+      .replace(/ +/gu, ' ')
+      .trim()
+      .slice(0, 30)
+      .trim();
+
+    // use word search for authors starting with a boolean
+    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));
+    // Prevent too many matches by having a minimum length
+    debugData(`Author string: ${formatted}`);
+    if (formatted.length >= 5) {
+      return [`dc.author="${useWordSearch ? '' : '^'}${formatted}*"`];
+    }
+    return [];
+  }
+
+  return [];
+
+  function getAuthor(record) {
+    //debugData(record);
+    // eslint-disable-next-line prefer-named-capture-group
+    const [field] = record.get(/^(100)|(110)|(111)|(700)|(710)|(711)$/u);
+    //debugData(field);
+
+    if (field) {
+      const authorString = field.subfields
+        .filter(({code}) => ['a'].includes(code))
         .map(({value}) => testStringOrNumber(value) ? String(value) : '')
         .filter(value => value)
         // In Melinda's index subfield separators are indexed as ' '
         .join(' ');
+      return authorString;
     }
     return false;
   }
 }
 
+
 export function bibStandardIdentifiers(record) {
 
   const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');