@natlibfi/melinda-record-matching 3.0.2-alpha.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/dist/match-detection/features/bib/isbn.js

@@ -5,8 +5,6 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.default = void 0;
 var _standardIdentifierFactory = _interopRequireDefault(require("./standard-identifier-factory"));
-var _isbn = require("isbn3");
-var _debug = _interopRequireDefault(require("debug"));
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
 /**
 *
@@ -35,42 +33,18 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 * for the JavaScript code in this file.
 *
 */
-
-const debug = (0, _debug.default)(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);
-const debugData = debug.extend('data');
 var _default = () => {
-  const IDENTIFIER_NAME = 'ISBN';
-  function validatorAndNormalizer(string) {
-    const isbnParseResult = (0, _isbn.parse)(string);
-    debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);
-    if (isbnParseResult === null) {
-      debug(`Not parseable ISBN, just removing hyphens`);
-      return {
-        valid: false,
-        value: string.replace(/-/ug, '')
-      };
-    }
-    debug(`Parseable ISBN, normalizing to ISBN-13`);
-    return {
-      valid: true,
-      value: isbnParseResult.isbn13
-    };
-  }
   const {
     extract,
     compare
   } = (0, _standardIdentifierFactory.default)({
-    identifier: IDENTIFIER_NAME,
     pattern: /^020$/u,
-    subfieldCodes: ['a', 'z'],
-    validIdentifierSubfieldCodes: ['a'],
-    invalidIdentifierSubfieldCodes: ['z'],
-    validatorAndNormalizer
+    subfieldCodes: ['a', 'z']
   });
   return {
     extract,
     compare,
-    name: IDENTIFIER_NAME
+    name: 'ISBN'
   };
 };
 exports.default = _default;

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

@@ -1 +1 @@
-{"version":3,"file":"isbn.js","names":["debug","createDebugLogger","debugData","extend","IDENTIFIER_NAME","validatorAndNormalizer","string","isbnParseResult","isbnParse","JSON","stringify","valid","value","replace","isbn13","extract","compare","createInterface","identifier","pattern","subfieldCodes","validIdentifierSubfieldCodes","invalidIdentifierSubfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/isbn.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createInterface from './standard-identifier-factory';\nimport {parse as isbnParse} from 'isbn3';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);\nconst debugData = debug.extend('data');\n\nexport default () => {\n  const IDENTIFIER_NAME = 'ISBN';\n\n  function validatorAndNormalizer(string) {\n    const isbnParseResult = isbnParse(string);\n    debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);\n    if (isbnParseResult === null) {\n      debug(`Not parseable ISBN, just removing hyphens`);\n      return {valid: false, value: string.replace(/-/ug, '')};\n    }\n    debug(`Parseable ISBN, normalizing to ISBN-13`);\n    return {valid: true, value: isbnParseResult.isbn13};\n  }\n\n  const {extract, compare} = createInterface({identifier: IDENTIFIER_NAME, pattern: /^020$/u, subfieldCodes: ['a', 'z'], validIdentifierSubfieldCodes: ['a'], invalidIdentifierSubfieldCodes: ['z'], validatorAndNormalizer});\n  return {extract, compare, name: IDENTIFIER_NAME};\n};\n\n"],"mappings":";;;;;;AA4BA;AACA;AACA;AAAsC;AA9BtC;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;;AAMA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAE,sFAAqF,CAAC;AACvH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,MAAM;EACnB,MAAMC,eAAe,GAAG,MAAM;EAE9B,SAASC,sBAAsB,CAACC,MAAM,EAAE;IACtC,MAAMC,eAAe,GAAG,IAAAC,WAAS,EAACF,MAAM,CAAC;IACzCJ,SAAS,CAAE,oBAAmBO,IAAI,CAACC,SAAS,CAACH,eAAe,CAAE,EAAC,CAAC;IAChE,IAAIA,eAAe,KAAK,IAAI,EAAE;MAC5BP,KAAK,CAAE,2CAA0C,CAAC;MAClD,OAAO;QAACW,KAAK,EAAE,KAAK;QAAEC,KAAK,EAAEN,MAAM,CAACO,OAAO,CAAC,KAAK,EAAE,EAAE;MAAC,CAAC;IACzD;IACAb,KAAK,CAAE,wCAAuC,CAAC;IAC/C,OAAO;MAACW,KAAK,EAAE,IAAI;MAAEC,KAAK,EAAEL,eAAe,CAACO;IAAM,CAAC;EACrD;EAEA,MAAM;IAACC,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,UAAU,EAAEd,eAAe;IAAEe,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IAAEC,4BAA4B,EAAE,CAAC,GAAG,CAAC;IAAEC,8BAA8B,EAAE,CAAC,GAAG,CAAC;IAAEjB;EAAsB,CAAC,CAAC;EAC3N,OAAO;IAACU,OAAO;IAAEC,OAAO;IAAEO,IAAI,EAAEnB;EAAe,CAAC;AAClD,CAAC;AAAA"}
+{"version":3,"file":"isbn.js","names":["extract","compare","createInterface","pattern","subfieldCodes","name"],"sources":["../../../../src/match-detection/features/bib/isbn.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createInterface from './standard-identifier-factory';\n\nexport default () => {\n  const {extract, compare} = createInterface({pattern: /^020$/u, subfieldCodes: ['a', 'z']});\n  return {extract, compare, name: 'ISBN'};\n};\n"],"mappings":";;;;;;AA4BA;AAA4D;AA5B5D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,eA8Be,MAAM;EACnB,MAAM;IAACA,OAAO;IAAEC;EAAO,CAAC,GAAG,IAAAC,kCAAe,EAAC;IAACC,OAAO,EAAE,QAAQ;IAAEC,aAAa,EAAE,CAAC,GAAG,EAAE,GAAG;EAAC,CAAC,CAAC;EAC1F,OAAO;IAACJ,OAAO;IAAEC,OAAO;IAAEI,IAAI,EAAE;EAAM,CAAC;AACzC,CAAC;AAAA"}

package/dist/match-detection/features/bib/standard-identifier-factory.js

@@ -7,7 +7,6 @@ exports.default = void 0;
 var _debug = _interopRequireDefault(require("debug"));
 var _matchingUtils = require("../../../matching-utils");
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
-/* eslint-disable max-statements */
 /**
 *
 * @licstart  The following is the entire license notice for the JavaScript code in this file.
@@ -35,172 +34,72 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 * for the JavaScript code in this file.
 *
 */
-// Note about validity of standardIdentifiers:
-// We have three types of invalid standardIdentifiers:
-// 1. Formally invalid standardIdentifiers (ie. typos either in the resource or the record)
-// 2. Formally valid standardIdentifiers that are used in a wrong resource
-// 3. Canceled standardIdentifiers
-// Matcher could and should check that a standardIdentifier found in a subfield for a valid identifier is formally valid, and if it's not formally valid, handle it as an invalid standardIdentifier
-// Formally valid standardIdentifiers found in subfield for invalid identifier cannot be handled as valid standardIdentifiers, because they can be a case of type 2) or 3) invalid standardIdentifiers
-// We could also do a separate handling for formally valid an formally invalid standardIdentifiers
+
+const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');
+const debugData = debug.extend('data');
 var _default = ({
   pattern,
-  subfieldCodes,
-  identifier,
-  validIdentifierSubfieldCodes = ['a'],
-  invalidIdentifierSubfieldCodes = ['z'],
-  validatorAndNormalizer = undefined
+  subfieldCodes
 }) => {
-  const debug = (0, _debug.default)(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:${identifier}`);
-  const debugData = debug.extend('data');
   return {
     extract,
     compare
   };
   function extract({
-    record,
-    recordExternal
+    record
   }) {
-    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
-    const fields = record.get(pattern);
-    debugData(`${label}: ${fields.length} ${identifier}-fields `);
-
-    // extractIdentifierSubfield normalizes hyphens away from the subfield values
-    const identifiersFromFields = fields.map(field => (0, _matchingUtils.extractSubfieldsFromField)(field, subfieldCodes));
-    debugData(`${label}: IDs from fields (${identifiersFromFields.length}): ${JSON.stringify(identifiersFromFields)}`);
-    const allIdentifiers = identifiersFromFields.flat();
-    debugData(`${label}: Flat IDs from fields (${allIdentifiers.length}): ${JSON.stringify(allIdentifiers)}`);
-    const validatedAndNormalizedIdentifiers = validateAndNormalizeIdentifiers({
-      identifierSubs: allIdentifiers,
-      validatorAndNormalizer,
-      validIdentifierSubfieldCodes,
-      invalidIdentifierSubfieldCodes
-    });
-    const identifiers = (0, _matchingUtils.uniqueSubfields)(validatedAndNormalizedIdentifiers);
-    debugData(`${label}: Unique IDs from fields (${identifiers.length}): ${JSON.stringify(identifiers)}`);
-    return identifiers;
-    function validateAndNormalizeIdentifiers({
-      identifierSubs,
-      validatorAndNormalizer,
-      validIdentifierSubfieldCodes,
-      invalidIdentifierSubfieldCodes
-    }) {
-      if (validatorAndNormalizer) {
-        return identifierSubs.map(idSub => validateAndNormalizeIdentifier({
-          idSub,
-          validatorAndNormalizer,
-          validIdentifierSubfieldCodes,
-          invalidIdentifierSubfieldCodes
-        }));
-      }
-      return identifierSubs.map(idSub => normalizeHyphens(idSub));
-    }
-    function validateAndNormalizeIdentifier({
-      idSub,
-      validatorAndNormalizer,
-      validIdentifierSubfieldCodes,
-      invalidIdentifierSubfieldCodes
-    }) {
-      const {
-        valid,
-        value
-      } = validatorAndNormalizer(idSub.value);
-      if (validIdentifierSubfieldCodes.includes(idSub.code) && valid === false) {
-        const [code] = invalidIdentifierSubfieldCodes;
-        return {
-          code,
-          value
-        };
-      }
-      return {
-        code: idSub.code,
+    const [field] = record.get(pattern);
+    if (field) {
+      return field.subfields.filter(({
+        code
+      }) => subfieldCodes.includes(code)).map(({
+        code,
         value
-      };
-    }
-    function normalizeHyphens(idSub) {
-      return {
-        code: idSub.code,
-        value: idSub.value.replace(/-/ug, '')
-      };
+      }) => ({
+        code,
+        value: (0, _matchingUtils.testStringOrNumber)(value) ? String(value).replace(/-/ug, '') : ''
+      })).filter(value => value);
     }
+    return [];
   }
   function compare(a, b) {
-    debug(`Comparing A and B`);
     if (a.length === 0 || b.length === 0) {
-      debugData(`No standardidentifiers (${identifier}) to compare`);
+      debugData(`No standardidentifiers to compare`);
       return 0;
     }
-    debugData(`A: ${JSON.stringify(a)}`);
-    debugData(`B: ${JSON.stringify(b)}`);
     if (bothHaveValidIdentifiers()) {
-      // Compare only valid identifiers, if both have valid idenfiers
       const {
         maxValues,
-        possibleMatchValues,
         matchingValues
       } = getValueCount(true);
-      if (matchingValues < 1) {
-        debug(`Both have valid standardidentifiers (${identifier}), but none of these match.`);
-        return -0.75;
-      }
-      debug(`Both have valid standardidentifiers (${identifier}), ${matchingValues}/${possibleMatchValues} valid identifiers match.`);
-      // ignore non-matches if there is mismatching amount of values
-      debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);
-      //we give some kind of penalty for mismatching amount of values instead of simple divide?
-      const penaltyForMissing = 0.1 * (maxValues - possibleMatchValues);
-      const penaltyForMisMatch = 0.2 * (possibleMatchValues - matchingValues);
-      debug(`\t points: penaltyForMissing: ${penaltyForMissing}`);
-      debug(`\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);
-      return 0.75 - penaltyForMisMatch - penaltyForMissing;
-      //return matchingValues / possibleMatchValues * 0.75;
+      return matchingValues / maxValues * 0.75;
     }
-    // If both do not have valid identifiers, compare all identifiers
     const {
       maxValues,
       matchingValues
     } = getValueCount();
-    debug(`Both do NOT have valid standardidentifiers (${identifier}), ${matchingValues}/${maxValues} valid/invalid identifiers match.`);
     return matchingValues / maxValues * 0.2;
     function bothHaveValidIdentifiers() {
       const aValues = a.filter(({
         code
-      }) => validIdentifierSubfieldCodes.includes(code));
+      }) => code === 'a');
       const bValues = a.filter(({
         code
-      }) => validIdentifierSubfieldCodes.includes(code));
-      debug(`A: ${aValues.length} valid ${identifier} identifiers`);
-      debug(`B: ${bValues.length} valid ${identifier} identifiers`);
+      }) => code === 'a');
       return aValues.length > 0 && bValues.length > 0;
     }
     function getValueCount(validOnly = false) {
-      const aValues = getIdentifiers(a, validOnly);
-      const bValues = getIdentifiers(b, validOnly);
-      const matchingValues = getMatchingValuesAmount(aValues, bValues);
+      const aValues = getIdentifiers(a);
+      const bValues = getIdentifiers(b);
       return {
         maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
-        // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
-        possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
-        matchingValues
+        matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length
       };
-      function getMatchingValuesAmount(aValues, bValues) {
-        if (bValues.length > aValues.length) {
-          return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
-        }
-        if (aValues.length > bValues.length) {
-          return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
-        }
-
-        // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases
-        // there would be duplicate values
-        const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
-        const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
-        return aToB < bToA ? aToB : bToA;
-      }
-      function getIdentifiers(values, validOnly) {
+      function getIdentifiers(values) {
         if (validOnly) {
           return values.filter(({
             code
-          }) => validIdentifierSubfieldCodes.includes(code)).map(({
+          }) => code === 'a').map(({
             value
           }) => value);
         }

package/dist/match-detection/features/bib/standard-identifier-factory.js.map

@@ -1 +1 @@
-{"version":3,"file":"standard-identifier-factory.js","names":["pattern","subfieldCodes","identifier","validIdentifierSubfieldCodes","invalidIdentifierSubfieldCodes","validatorAndNormalizer","undefined","debug","createDebugLogger","debugData","extend","extract","compare","record","recordExternal","label","fields","get","length","identifiersFromFields","map","field","extractSubfieldsFromField","JSON","stringify","allIdentifiers","flat","validatedAndNormalizedIdentifiers","validateAndNormalizeIdentifiers","identifierSubs","identifiers","uniqueSubfields","idSub","validateAndNormalizeIdentifier","normalizeHyphens","valid","value","includes","code","replace","a","b","bothHaveValidIdentifiers","maxValues","possibleMatchValues","matchingValues","getValueCount","penaltyForMissing","penaltyForMisMatch","aValues","filter","bValues","validOnly","getIdentifiers","getMatchingValuesAmount","aValue","some","bValue","aToB","bToA","values"],"sources":["../../../../src/match-detection/features/bib/standard-identifier-factory.js"],"sourcesContent":["/* eslint-disable max-statements */\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {extractSubfieldsFromField, uniqueSubfields} from '../../../matching-utils';\n\n// Note about validity of standardIdentifiers:\n// We have three types of invalid standardIdentifiers:\n// 1. Formally invalid standardIdentifiers (ie. typos either in the resource or the record)\n// 2. Formally valid standardIdentifiers that are used in a wrong resource\n// 3. Canceled standardIdentifiers\n\n// Matcher could and should check that a standardIdentifier found in a subfield for a valid identifier is formally valid, and if it's not formally valid, handle it as an invalid standardIdentifier\n// Formally valid standardIdentifiers found in subfield for invalid identifier cannot be handled as valid standardIdentifiers, because they can be a case of type 2) or 3) invalid standardIdentifiers\n// We could also do a separate handling for formally valid an formally invalid standardIdentifiers\n\nexport default ({pattern, subfieldCodes, identifier, validIdentifierSubfieldCodes = ['a'], invalidIdentifierSubfieldCodes = ['z'], validatorAndNormalizer = undefined}) => {\n  const debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:${identifier}`);\n  const debugData = debug.extend('data');\n\n  return {extract, compare};\n\n  function extract({record, recordExternal}) {\n    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n    const fields = record.get(pattern);\n    debugData(`${label}: ${fields.length} ${identifier}-fields `);\n\n    // extractIdentifierSubfield normalizes hyphens away from the subfield values\n    const identifiersFromFields = fields.map(field => extractSubfieldsFromField(field, subfieldCodes));\n    debugData(`${label}: IDs from fields (${identifiersFromFields.length}): ${JSON.stringify(identifiersFromFields)}`);\n    const allIdentifiers = identifiersFromFields.flat();\n    debugData(`${label}: Flat IDs from fields (${allIdentifiers.length}): ${JSON.stringify(allIdentifiers)}`);\n\n    const validatedAndNormalizedIdentifiers = validateAndNormalizeIdentifiers({identifierSubs: allIdentifiers, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes});\n\n    const identifiers = uniqueSubfields(validatedAndNormalizedIdentifiers);\n\n    debugData(`${label}: Unique IDs from fields (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n    return identifiers;\n\n    function validateAndNormalizeIdentifiers({identifierSubs, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}) {\n      if (validatorAndNormalizer) {\n        return identifierSubs.map((idSub) => validateAndNormalizeIdentifier({idSub, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}));\n      }\n      return identifierSubs.map((idSub) => normalizeHyphens(idSub));\n    }\n\n    function validateAndNormalizeIdentifier({idSub, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}) {\n      const {valid, value} = validatorAndNormalizer(idSub.value);\n      if (validIdentifierSubfieldCodes.includes(idSub.code) && valid === false) {\n        const [code] = invalidIdentifierSubfieldCodes;\n        return {code, value};\n      }\n      return {code: idSub.code, value};\n    }\n\n    function normalizeHyphens(idSub) {\n      return {code: idSub.code, value: idSub.value.replace(/-/ug, '')};\n    }\n\n\n  }\n\n  function compare(a, b) {\n    debug(`Comparing A and B`);\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No standardidentifiers (${identifier}) to compare`);\n      return 0;\n    }\n\n    debugData(`A: ${JSON.stringify(a)}`);\n    debugData(`B: ${JSON.stringify(b)}`);\n\n\n    if (bothHaveValidIdentifiers()) {\n      // Compare only valid identifiers, if both have valid idenfiers\n      const {maxValues, possibleMatchValues, matchingValues} = getValueCount(true);\n      if (matchingValues < 1) {\n        debug(`Both have valid standardidentifiers (${identifier}), but none of these match.`);\n        return -0.75;\n      }\n      debug(`Both have valid standardidentifiers (${identifier}), ${matchingValues}/${possibleMatchValues} valid identifiers match.`);\n      // ignore non-matches if there is mismatching amount of values\n      debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);\n      //we give some kind of penalty for mismatching amount of values instead of simple divide?\n      const penaltyForMissing = 0.1 * (maxValues - possibleMatchValues);\n      const penaltyForMisMatch = 0.2 * (possibleMatchValues - matchingValues);\n      debug(`\\t points: penaltyForMissing: ${penaltyForMissing}`);\n      debug(`\\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);\n\n      return 0.75 - penaltyForMisMatch - penaltyForMissing;\n      //return matchingValues / possibleMatchValues * 0.75;\n    }\n    // If both do not have valid identifiers, compare all identifiers\n    const {maxValues, matchingValues} = getValueCount();\n    debug(`Both do NOT have valid standardidentifiers (${identifier}), ${matchingValues}/${maxValues} valid/invalid identifiers match.`);\n\n    return matchingValues / maxValues * 0.2;\n\n    function bothHaveValidIdentifiers() {\n      const aValues = a.filter(({code}) => validIdentifierSubfieldCodes.includes(code));\n      const bValues = a.filter(({code}) => validIdentifierSubfieldCodes.includes(code));\n      debug(`A: ${aValues.length} valid ${identifier} identifiers`);\n      debug(`B: ${bValues.length} valid ${identifier} identifiers`);\n      return aValues.length > 0 && bValues.length > 0;\n    }\n\n    function getValueCount(validOnly = false) {\n      const aValues = getIdentifiers(a, validOnly);\n      const bValues = getIdentifiers(b, validOnly);\n\n      const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n      return {\n        maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n        // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n        possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n        matchingValues\n      };\n\n      function getMatchingValuesAmount(aValues, bValues) {\n        if (bValues.length > aValues.length) {\n          return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n        }\n        if (aValues.length > bValues.length) {\n          return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n        }\n\n        // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n        // there would be duplicate values\n        const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n        const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n        return aToB < bToA ? aToB : bToA;\n      }\n\n      function getIdentifiers(values, validOnly) {\n        if (validOnly) {\n          return values\n            .filter(({code}) => validIdentifierSubfieldCodes.includes(code))\n            .map(({value}) => value);\n        }\n\n        return values.map(({value}) => value);\n      }\n    }\n  }\n};\n"],"mappings":";;;;;;AA6BA;AACA;AAAmF;AA9BnF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAKA;AACA;AACA;AACA;AACA;AAEA;AACA;AACA;AAAA,eAEe,CAAC;EAACA,OAAO;EAAEC,aAAa;EAAEC,UAAU;EAAEC,4BAA4B,GAAG,CAAC,GAAG,CAAC;EAAEC,8BAA8B,GAAG,CAAC,GAAG,CAAC;EAAEC,sBAAsB,GAAGC;AAAS,CAAC,KAAK;EACzK,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAE,mFAAkFN,UAAW,EAAC,CAAC;EAChI,MAAMO,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAAC;IAACE,MAAM;IAAEC;EAAc,CAAC,EAAE;IACzC,MAAMC,KAAK,GAAGD,cAAc,IAAIA,cAAc,CAACC,KAAK,GAAGD,cAAc,CAACC,KAAK,GAAG,QAAQ;IACtF,MAAMC,MAAM,GAAGH,MAAM,CAACI,GAAG,CAACjB,OAAO,CAAC;IAClCS,SAAS,CAAE,GAAEM,KAAM,KAAIC,MAAM,CAACE,MAAO,IAAGhB,UAAW,UAAS,CAAC;;IAE7D;IACA,MAAMiB,qBAAqB,GAAGH,MAAM,CAACI,GAAG,CAACC,KAAK,IAAI,IAAAC,wCAAyB,EAACD,KAAK,EAAEpB,aAAa,CAAC,CAAC;IAClGQ,SAAS,CAAE,GAAEM,KAAM,sBAAqBI,qBAAqB,CAACD,MAAO,MAAKK,IAAI,CAACC,SAAS,CAACL,qBAAqB,CAAE,EAAC,CAAC;IAClH,MAAMM,cAAc,GAAGN,qBAAqB,CAACO,IAAI,EAAE;IACnDjB,SAAS,CAAE,GAAEM,KAAM,2BAA0BU,cAAc,CAACP,MAAO,MAAKK,IAAI,CAACC,SAAS,CAACC,cAAc,CAAE,EAAC,CAAC;IAEzG,MAAME,iCAAiC,GAAGC,+BAA+B,CAAC;MAACC,cAAc,EAAEJ,cAAc;MAAEpB,sBAAsB;MAAEF,4BAA4B;MAAEC;IAA8B,CAAC,CAAC;IAEjM,MAAM0B,WAAW,GAAG,IAAAC,8BAAe,EAACJ,iCAAiC,CAAC;IAEtElB,SAAS,CAAE,GAAEM,KAAM,6BAA4Be,WAAW,CAACZ,MAAO,MAAKK,IAAI,CAACC,SAAS,CAACM,WAAW,CAAE,EAAC,CAAC;IACrG,OAAOA,WAAW;IAElB,SAASF,+BAA+B,CAAC;MAACC,cAAc;MAAExB,sBAAsB;MAAEF,4BAA4B;MAAEC;IAA8B,CAAC,EAAE;MAC/I,IAAIC,sBAAsB,EAAE;QAC1B,OAAOwB,cAAc,CAACT,GAAG,CAAEY,KAAK,IAAKC,8BAA8B,CAAC;UAACD,KAAK;UAAE3B,sBAAsB;UAAEF,4BAA4B;UAAEC;QAA8B,CAAC,CAAC,CAAC;MACrK;MACA,OAAOyB,cAAc,CAACT,GAAG,CAAEY,KAAK,IAAKE,gBAAgB,CAACF,KAAK,CAAC,CAAC;IAC/D;IAEA,SAASC,8BAA8B,CAAC;MAACD,KAAK;MAAE3B,sBAAsB;MAAEF,4BAA4B;MAAEC;IAA8B,CAAC,EAAE;MACrI,MAAM;QAAC+B,KAAK;QAAEC;MAAK,CAAC,GAAG/B,sBAAsB,CAAC2B,KAAK,CAACI,KAAK,CAAC;MAC1D,IAAIjC,4BAA4B,CAACkC,QAAQ,CAACL,KAAK,CAACM,IAAI,CAAC,IAAIH,KAAK,KAAK,KAAK,EAAE;QACxE,MAAM,CAACG,IAAI,CAAC,GAAGlC,8BAA8B;QAC7C,OAAO;UAACkC,IAAI;UAAEF;QAAK,CAAC;MACtB;MACA,OAAO;QAACE,IAAI,EAAEN,KAAK,CAACM,IAAI;QAAEF;MAAK,CAAC;IAClC;IAEA,SAASF,gBAAgB,CAACF,KAAK,EAAE;MAC/B,OAAO;QAACM,IAAI,EAAEN,KAAK,CAACM,IAAI;QAAEF,KAAK,EAAEJ,KAAK,CAACI,KAAK,CAACG,OAAO,CAAC,KAAK,EAAE,EAAE;MAAC,CAAC;IAClE;EAGF;EAEA,SAAS3B,OAAO,CAAC4B,CAAC,EAAEC,CAAC,EAAE;IACrBlC,KAAK,CAAE,mBAAkB,CAAC;IAC1B,IAAIiC,CAAC,CAACtB,MAAM,KAAK,CAAC,IAAIuB,CAAC,CAACvB,MAAM,KAAK,CAAC,EAAE;MACpCT,SAAS,CAAE,2BAA0BP,UAAW,cAAa,CAAC;MAC9D,OAAO,CAAC;IACV;IAEAO,SAAS,CAAE,MAAKc,IAAI,CAACC,SAAS,CAACgB,CAAC,CAAE,EAAC,CAAC;IACpC/B,SAAS,CAAE,MAAKc,IAAI,CAACC,SAAS,CAACiB,CAAC,CAAE,EAAC,CAAC;IAGpC,IAAIC,wBAAwB,EAAE,EAAE;MAC9B;MACA,MAAM;QAACC,SAAS;QAAEC,mBAAmB;QAAEC;MAAc,CAAC,GAAGC,aAAa,CAAC,IAAI,CAAC;MAC5E,IAAID,cAAc,GAAG,CAAC,EAAE;QACtBtC,KAAK,CAAE,wCAAuCL,UAAW,6BAA4B,CAAC;QACtF,OAAO,CAAC,IAAI;MACd;MACAK,KAAK,CAAE,wCAAuCL,UAAW,MAAK2C,cAAe,IAAGD,mBAAoB,2BAA0B,CAAC;MAC/H;MACArC,KAAK,CAAE,qBAAoBqC,mBAAoB,IAAGD,SAAU,EAAC,CAAC;MAC9D;MACA,MAAMI,iBAAiB,GAAG,GAAG,IAAIJ,SAAS,GAAGC,mBAAmB,CAAC;MACjE,MAAMI,kBAAkB,GAAG,GAAG,IAAIJ,mBAAmB,GAAGC,cAAc,CAAC;MACvEtC,KAAK,CAAE,iCAAgCwC,iBAAkB,EAAC,CAAC;MAC3DxC,KAAK,CAAE,kCAAiCyC,kBAAmB,EAAC,CAAC;MAE7D,OAAO,IAAI,GAAGA,kBAAkB,GAAGD,iBAAiB;MACpD;IACF;IACA;IACA,MAAM;MAACJ,SAAS;MAAEE;IAAc,CAAC,GAAGC,aAAa,EAAE;IACnDvC,KAAK,CAAE,+CAA8CL,UAAW,MAAK2C,cAAe,IAAGF,SAAU,mCAAkC,CAAC;IAEpI,OAAOE,cAAc,GAAGF,SAAS,GAAG,GAAG;IAEvC,SAASD,wBAAwB,GAAG;MAClC,MAAMO,OAAO,GAAGT,CAAC,CAACU,MAAM,CAAC,CAAC;QAACZ;MAAI,CAAC,KAAKnC,4BAA4B,CAACkC,QAAQ,CAACC,IAAI,CAAC,CAAC;MACjF,MAAMa,OAAO,GAAGX,CAAC,CAACU,MAAM,CAAC,CAAC;QAACZ;MAAI,CAAC,KAAKnC,4BAA4B,CAACkC,QAAQ,CAACC,IAAI,CAAC,CAAC;MACjF/B,KAAK,CAAE,MAAK0C,OAAO,CAAC/B,MAAO,UAAShB,UAAW,cAAa,CAAC;MAC7DK,KAAK,CAAE,MAAK4C,OAAO,CAACjC,MAAO,UAAShB,UAAW,cAAa,CAAC;MAC7D,OAAO+C,OAAO,CAAC/B,MAAM,GAAG,CAAC,IAAIiC,OAAO,CAACjC,MAAM,GAAG,CAAC;IACjD;IAEA,SAAS4B,aAAa,CAACM,SAAS,GAAG,KAAK,EAAE;MACxC,MAAMH,OAAO,GAAGI,cAAc,CAACb,CAAC,EAAEY,SAAS,CAAC;MAC5C,MAAMD,OAAO,GAAGE,cAAc,CAACZ,CAAC,EAAEW,SAAS,CAAC;MAE5C,MAAMP,cAAc,GAAGS,uBAAuB,CAACL,OAAO,EAAEE,OAAO,CAAC;MAEhE,OAAO;QACLR,SAAS,EAAEM,OAAO,CAAC/B,MAAM,GAAGiC,OAAO,CAACjC,MAAM,GAAG+B,OAAO,CAAC/B,MAAM,GAAGiC,OAAO,CAACjC,MAAM;QAC5E;QACA0B,mBAAmB,EAAEK,OAAO,CAAC/B,MAAM,GAAGiC,OAAO,CAACjC,MAAM,GAAGiC,OAAO,CAACjC,MAAM,GAAG+B,OAAO,CAAC/B,MAAM;QACtF2B;MACF,CAAC;MAED,SAASS,uBAAuB,CAACL,OAAO,EAAEE,OAAO,EAAE;QACjD,IAAIA,OAAO,CAACjC,MAAM,GAAG+B,OAAO,CAAC/B,MAAM,EAAE;UACnC,OAAO+B,OAAO,CAACC,MAAM,CAACK,MAAM,IAAIJ,OAAO,CAACK,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAACvC,MAAM;QACnF;QACA,IAAI+B,OAAO,CAAC/B,MAAM,GAAGiC,OAAO,CAACjC,MAAM,EAAE;UACnC,OAAOiC,OAAO,CAACD,MAAM,CAACO,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAACrC,MAAM;QACnF;;QAEA;QACA;QACA,MAAMwC,IAAI,GAAGT,OAAO,CAACC,MAAM,CAACK,MAAM,IAAIJ,OAAO,CAACK,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAACvC,MAAM;QACvF,MAAMyC,IAAI,GAAGR,OAAO,CAACD,MAAM,CAACO,MAAM,IAAIR,OAAO,CAACO,IAAI,CAACD,MAAM,IAAIE,MAAM,KAAKF,MAAM,CAAC,CAAC,CAACrC,MAAM;QAEvF,OAAOwC,IAAI,GAAGC,IAAI,GAAGD,IAAI,GAAGC,IAAI;MAClC;MAEA,SAASN,cAAc,CAACO,MAAM,EAAER,SAAS,EAAE;QACzC,IAAIA,SAAS,EAAE;UACb,OAAOQ,MAAM,CACVV,MAAM,CAAC,CAAC;YAACZ;UAAI,CAAC,KAAKnC,4BAA4B,CAACkC,QAAQ,CAACC,IAAI,CAAC,CAAC,CAC/DlB,GAAG,CAAC,CAAC;YAACgB;UAAK,CAAC,KAAKA,KAAK,CAAC;QAC5B;QAEA,OAAOwB,MAAM,CAACxC,GAAG,CAAC,CAAC;UAACgB;QAAK,CAAC,KAAKA,KAAK,CAAC;MACvC;IACF;EACF;AACF,CAAC;AAAA"}
+{"version":3,"file":"standard-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","pattern","subfieldCodes","extract","compare","record","field","get","subfields","filter","code","includes","map","value","testStringOrNumber","String","replace","a","b","length","bothHaveValidIdentifiers","maxValues","matchingValues","getValueCount","aValues","bValues","validOnly","getIdentifiers","aValue","some","bValue","values"],"sources":["../../../../src/match-detection/features/bib/standard-identifier-factory.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');\nconst debugData = debug.extend('data');\n\nexport default ({pattern, subfieldCodes}) => {\n  return {extract, compare};\n\n  function extract({record}) {\n    const [field] = record.get(pattern);\n\n    if (field) {\n      return field.subfields\n        .filter(({code}) => subfieldCodes.includes(code))\n        .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))\n        .filter(value => value);\n    }\n\n    return [];\n  }\n\n  function compare(a, b) {\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No standardidentifiers to compare`);\n      return 0;\n    }\n\n    if (bothHaveValidIdentifiers()) {\n      const {maxValues, matchingValues} = getValueCount(true);\n      return matchingValues / maxValues * 0.75;\n    }\n\n    const {maxValues, matchingValues} = getValueCount();\n    return matchingValues / maxValues * 0.2;\n\n    function bothHaveValidIdentifiers() {\n      const aValues = a.filter(({code}) => code === 'a');\n      const bValues = a.filter(({code}) => code === 'a');\n      return aValues.length > 0 && bValues.length > 0;\n    }\n\n    function getValueCount(validOnly = false) {\n      const aValues = getIdentifiers(a);\n      const bValues = getIdentifiers(b);\n\n      return {\n        maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n        matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length\n      };\n\n      function getIdentifiers(values) {\n        if (validOnly) {\n          return values\n            .filter(({code}) => code === 'a')\n            .map(({value}) => value);\n        }\n\n        return values.map(({value}) => value);\n      }\n    }\n  }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA2D;AA7B3D;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;;AAKA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;AAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,CAAC;EAACC,OAAO;EAAEC;AAAa,CAAC,KAAK;EAC3C,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAAC;IAACE;EAAM,CAAC,EAAE;IACzB,MAAM,CAACC,KAAK,CAAC,GAAGD,MAAM,CAACE,GAAG,CAACN,OAAO,CAAC;IAEnC,IAAIK,KAAK,EAAE;MACT,OAAOA,KAAK,CAACE,SAAS,CACnBC,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKR,aAAa,CAACS,QAAQ,CAACD,IAAI,CAAC,CAAC,CAChDE,GAAG,CAAC,CAAC;QAACF,IAAI;QAAEG;MAAK,CAAC,MAAM;QAACH,IAAI;QAAEG,KAAK,EAAE,IAAAC,iCAAkB,EAACD,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG;MAAE,CAAC,CAAC,CAAC,CAC1GP,MAAM,CAACI,KAAK,IAAIA,KAAK,CAAC;IAC3B;IAEA,OAAO,EAAE;EACX;EAEA,SAAST,OAAO,CAACa,CAAC,EAAEC,CAAC,EAAE;IACrB,IAAID,CAAC,CAACE,MAAM,KAAK,CAAC,IAAID,CAAC,CAACC,MAAM,KAAK,CAAC,EAAE;MACpCpB,SAAS,CAAE,mCAAkC,CAAC;MAC9C,OAAO,CAAC;IACV;IAEA,IAAIqB,wBAAwB,EAAE,EAAE;MAC9B,MAAM;QAACC,SAAS;QAAEC;MAAc,CAAC,GAAGC,aAAa,CAAC,IAAI,CAAC;MACvD,OAAOD,cAAc,GAAGD,SAAS,GAAG,IAAI;IAC1C;IAEA,MAAM;MAACA,SAAS;MAAEC;IAAc,CAAC,GAAGC,aAAa,EAAE;IACnD,OAAOD,cAAc,GAAGD,SAAS,GAAG,GAAG;IAEvC,SAASD,wBAAwB,GAAG;MAClC,MAAMI,OAAO,GAAGP,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,MAAMe,OAAO,GAAGR,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,OAAOc,OAAO,CAACL,MAAM,GAAG,CAAC,IAAIM,OAAO,CAACN,MAAM,GAAG,CAAC;IACjD;IAEA,SAASI,aAAa,CAACG,SAAS,GAAG,KAAK,EAAE;MACxC,MAAMF,OAAO,GAAGG,cAAc,CAACV,CAAC,CAAC;MACjC,MAAMQ,OAAO,GAAGE,cAAc,CAACT,CAAC,CAAC;MAEjC,OAAO;QACLG,SAAS,EAAEG,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM,GAAGK,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM;QAC5EG,cAAc,EAAEE,OAAO,CAACf,MAAM,CAACmB,MAAM,IAAIH,OAAO,CAACI,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAACX;MACtF,CAAC;MAED,SAASQ,cAAc,CAACI,MAAM,EAAE;QAC9B,IAAIL,SAAS,EAAE;UACb,OAAOK,MAAM,CACVtB,MAAM,CAAC,CAAC;YAACC;UAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCE,GAAG,CAAC,CAAC;YAACC;UAAK,CAAC,KAAKA,KAAK,CAAC;QAC5B;QAEA,OAAOkB,MAAM,CAACnB,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,CAAC;MACvC;IACF;EACF;AACF,CAAC;AAAA"}

package/dist/matching-utils.js

@@ -3,11 +3,9 @@
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
-exports.extractSubfieldsFromField = extractSubfieldsFromField;
 exports.getMelindaIdsF035 = getMelindaIdsF035;
 exports.getSubfieldValues = getSubfieldValues;
 exports.testStringOrNumber = testStringOrNumber;
-exports.uniqueSubfields = uniqueSubfields;
 exports.validateSidFieldSubfieldCounts = validateSidFieldSubfieldCounts;
 var _debug = _interopRequireDefault(require("debug"));
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
@@ -92,28 +90,4 @@ function testStringOrNumber(value) {
   }
   return false;
 }
-function extractSubfieldsFromField(field, subfieldCodes) {
-  if (field === undefined || field.subfields === undefined) {
-    return [];
-  }
-  const resultSubfields = field.subfields.filter(({
-    code
-  }) => subfieldCodes.includes(code)).map(({
-    code,
-    value
-  }) => ({
-    code,
-    value: testStringOrNumber(value) ? String(value) : ''
-  })).filter(value => value);
-  return resultSubfields;
-}
-function uniqueSubfields(subfields) {
-  return subfields.reduce((arr, e) => {
-    if (!arr.find(item => item.code === e.code && item.value === e.value)) {
-      const newArr = arr.concat(e);
-      return newArr;
-    }
-    return arr;
-  }, []);
-}
 //# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -1 +1 @@
-{"version":3,"file":"matching-utils.js","names":["debug","createDebugLogger","debugData","extend","getMelindaIdsF035","record","f035s","getFields","length","allF035MelindaIds","concat","map","toMelindaIds","f035MelindaIds","Set","JSON","stringify","subfields","melindaIdRegExp","filter","sub","includes","code","testStringOrNumber","value","test","String","replace","validateSidFieldSubfieldCounts","field","countC","countSubfields","countB","subfieldCode","getSubfieldValues","extractSubfieldsFromField","subfieldCodes","undefined","resultSubfields","uniqueSubfields","reduce","arr","e","find","item","newArr"],"sources":["../src/matching-utils.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n\n  function toMelindaIds({subfields}) {\n    const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code))\n      .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n      .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n\n  }\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n"],"mappings":";;;;;;;;;;;AA4BA;AAAsC;AA5BtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAIA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiB,CAACC,MAAM,EAAE;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAiB,EAAC,8CAA8C,CAAC;EAC/E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAS,CAAC,KAAK,CAAC;EAErC,IAAID,KAAK,CAACE,MAAM,GAAG,CAAC,EAAE;IACpBR,KAAK,CAAE,iBAAgB,CAAC;IACxB,OAAO,EAAE;EACX;EAEA,MAAMS,iBAAiB,GAAG,EAAE,CAACC,MAAM,CAAC,GAAGJ,KAAK,CAACK,GAAG,CAACC,YAAY,CAAC,CAAC;EAC/D,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACL,iBAAiB,CAAC,CAAC;EAEtDP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACV,KAAK,CAAE,EAAC,CAAC;EAC/DJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACP,iBAAiB,CAAE,EAAC,CAAC;EACpFP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACH,cAAc,CAAE,EAAC,CAAC;EAErF,OAAOA,cAAc;EAErB,SAASD,YAAY,CAAC;IAACK;EAAS,CAAC,EAAE;IACjC,MAAMC,eAAe,GAAG,8CAA8C;IAEtE,OAAOD,SAAS,CACbE,MAAM,CAACC,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,GAAG,CAACE,IAAI,CAAC,CAAC,CAC5CH,MAAM,CAACC,GAAG,IAAIG,kBAAkB,CAACH,GAAG,CAACI,KAAK,CAAC,IAAIN,eAAe,CAACO,IAAI,CAACC,MAAM,CAACN,GAAG,CAACI,KAAK,CAAC,CAAC,CAAC,CACvFb,GAAG,CAAC,CAAC;MAACa;IAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAACT,eAAe,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC;EAEvG;AACF;AAEO,SAASU,8BAA8B,CAACC,KAAK,EAAE;EACpD;EACA3B,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC1D,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC7B,KAAK,CAAE,SAAQ8B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAE,EAAC,CAAC;EAE9F,OAAOF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAC;AACrC;AAEA,SAASD,cAAc,CAACF,KAAK,EAAEI,YAAY,EAAE;EAC3C;EACA,OAAOJ,KAAK,CAACZ,SAAS,CAACE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CAACzB,MAAM;AACzE;AAEO,SAAS0B,iBAAiB,CAACL,KAAK,EAAEI,YAAY,EAAE;EACrD/B,SAAS,CAAE,oBAAmB+B,YAAa,SAAQlB,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC3E,OAAOA,KAAK,CAACZ,SAAS,CACnBE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CACzCtB,GAAG,CAAC,CAAC;IAACa;EAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;AAC3B;AAEO,SAASD,kBAAkB,CAACC,KAAK,EAAE;EACxC,IAAI,OAAOA,KAAK,KAAK,QAAQ,IAAI,OAAOA,KAAK,KAAK,QAAQ,EAAE;IAC1D,OAAO,IAAI;EACb;EACA,OAAO,KAAK;AACd;AAEO,SAASW,yBAAyB,CAACN,KAAK,EAAEO,aAAa,EAAE;EAC9D,IAAIP,KAAK,KAAKQ,SAAS,IAAIR,KAAK,CAACZ,SAAS,KAAKoB,SAAS,EAAE;IACxD,OAAO,EAAE;EACX;EACA,MAAMC,eAAe,GAAGT,KAAK,CAACZ,SAAS,CACpCE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKc,aAAa,CAACf,QAAQ,CAACC,IAAI,CAAC,CAAC,CAChDX,GAAG,CAAC,CAAC;IAACW,IAAI;IAAEE;EAAK,CAAC,MAAM;IAACF,IAAI;IAAEE,KAAK,EAAED,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG;EAAE,CAAC,CAAC,CAAC,CACvFL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;EACzB,OAAOc,eAAe;AACxB;AAEO,SAASC,eAAe,CAACtB,SAAS,EAAE;EACzC,OAAOA,SAAS,CAACuB,MAAM,CAAC,CAACC,GAAG,EAAEC,CAAC,KAAK;IAClC,IAAI,CAACD,GAAG,CAACE,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACtB,IAAI,KAAKoB,CAAC,CAACpB,IAAI,IAAIsB,IAAI,CAACpB,KAAK,KAAKkB,CAAC,CAAClB,KAAK,CAAC,EAAE;MACrE,MAAMqB,MAAM,GAAGJ,GAAG,CAAC/B,MAAM,CAACgC,CAAC,CAAC;MAC5B,OAAOG,MAAM;IACf;IACA,OAAOJ,GAAG;EACZ,CAAC,EAAE,EAAE,CAAC;AACR"}
+{"version":3,"file":"matching-utils.js","names":["debug","createDebugLogger","debugData","extend","getMelindaIdsF035","record","f035s","getFields","length","allF035MelindaIds","concat","map","toMelindaIds","f035MelindaIds","Set","JSON","stringify","subfields","melindaIdRegExp","filter","sub","includes","code","testStringOrNumber","value","test","String","replace","validateSidFieldSubfieldCounts","field","countC","countSubfields","countB","subfieldCode","getSubfieldValues"],"sources":["../src/matching-utils.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(toMelindaIds));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n\n  function toMelindaIds({subfields}) {\n    const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code))\n      .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n      .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n\n  }\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n"],"mappings":";;;;;;;;;AA4BA;AAAsC;AA5BtC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAIA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,yCAAyC,CAAC;AAC1E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAE/B,SAASC,iBAAiB,CAACC,MAAM,EAAE;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAiB,EAAC,8CAA8C,CAAC;EAC/E,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAS,CAAC,KAAK,CAAC;EAErC,IAAID,KAAK,CAACE,MAAM,GAAG,CAAC,EAAE;IACpBR,KAAK,CAAE,iBAAgB,CAAC;IACxB,OAAO,EAAE;EACX;EAEA,MAAMS,iBAAiB,GAAG,EAAE,CAACC,MAAM,CAAC,GAAGJ,KAAK,CAACK,GAAG,CAACC,YAAY,CAAC,CAAC;EAC/D,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAG,CAACL,iBAAiB,CAAC,CAAC;EAEtDP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACV,KAAK,CAAE,EAAC,CAAC;EAC/DJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACP,iBAAiB,CAAE,EAAC,CAAC;EACpFP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAS,CAACH,cAAc,CAAE,EAAC,CAAC;EAErF,OAAOA,cAAc;EAErB,SAASD,YAAY,CAAC;IAACK;EAAS,CAAC,EAAE;IACjC,MAAMC,eAAe,GAAG,8CAA8C;IAEtE,OAAOD,SAAS,CACbE,MAAM,CAACC,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,GAAG,CAACE,IAAI,CAAC,CAAC,CAC5CH,MAAM,CAACC,GAAG,IAAIG,kBAAkB,CAACH,GAAG,CAACI,KAAK,CAAC,IAAIN,eAAe,CAACO,IAAI,CAACC,MAAM,CAACN,GAAG,CAACI,KAAK,CAAC,CAAC,CAAC,CACvFb,GAAG,CAAC,CAAC;MAACa;IAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAACT,eAAe,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC;EAEvG;AACF;AAEO,SAASU,8BAA8B,CAACC,KAAK,EAAE;EACpD;EACA3B,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC1D,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAK,EAAE,GAAG,CAAC;EACzC7B,KAAK,CAAE,SAAQ8B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAE,EAAC,CAAC;EAE9F,OAAOF,MAAM,KAAK,CAAC,IAAIE,MAAM,KAAK,CAAC;AACrC;AAEA,SAASD,cAAc,CAACF,KAAK,EAAEI,YAAY,EAAE;EAC3C;EACA,OAAOJ,KAAK,CAACZ,SAAS,CAACE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CAACzB,MAAM;AACzE;AAEO,SAAS0B,iBAAiB,CAACL,KAAK,EAAEI,YAAY,EAAE;EACrD/B,SAAS,CAAE,oBAAmB+B,YAAa,SAAQlB,IAAI,CAACC,SAAS,CAACa,KAAK,CAAE,EAAC,CAAC;EAC3E,OAAOA,KAAK,CAACZ,SAAS,CACnBE,MAAM,CAAC,CAAC;IAACG;EAAI,CAAC,KAAKA,IAAI,KAAKW,YAAY,CAAC,CACzCtB,GAAG,CAAC,CAAC;IAACa;EAAK,CAAC,KAAKD,kBAAkB,CAACC,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEL,MAAM,CAACK,KAAK,IAAIA,KAAK,CAAC;AAC3B;AAEO,SAASD,kBAAkB,CAACC,KAAK,EAAE;EACxC,IAAI,OAAOA,KAAK,KAAK,QAAQ,IAAI,OAAOA,KAAK,KAAK,QAAQ,EAAE;IAC1D,OAAO,IAAI;EACb;EACA,OAAO,KAAK;AACd"}

package/package.json

@@ -14,7 +14,7 @@
 		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
 	},
 	"license": "LGPL-3.0+",
-	"version": "3.0.2-alpha.1",
+	"version": "3.0.3-alpha.1",
 	"main": "./dist/index.js",
 	"engines": {
 		"node": ">=14"
@@ -41,7 +41,6 @@
 		"@natlibfi/melinda-commons": "^12.0.6",
 		"@natlibfi/sru-client": "^5.0.3",
 		"debug": "^4.3.4",
-		"isbn3": "^1.1.28",
 		"moment": "^2.29.4",
 		"natural": "^5.2.3",
 		"uuid": "^9.0.0",

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');

package/src/match-detection/features/bib/isbn.js

@@ -27,27 +27,8 @@
 */
 
 import createInterface from './standard-identifier-factory';
-import {parse as isbnParse} from 'isbn3';
-import createDebugLogger from 'debug';
-
-const debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);
-const debugData = debug.extend('data');
 
 export default () => {
-  const IDENTIFIER_NAME = 'ISBN';
-
-  function validatorAndNormalizer(string) {
-    const isbnParseResult = isbnParse(string);
-    debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);
-    if (isbnParseResult === null) {
-      debug(`Not parseable ISBN, just removing hyphens`);
-      return {valid: false, value: string.replace(/-/ug, '')};
-    }
-    debug(`Parseable ISBN, normalizing to ISBN-13`);
-    return {valid: true, value: isbnParseResult.isbn13};
-  }
-
-  const {extract, compare} = createInterface({identifier: IDENTIFIER_NAME, pattern: /^020$/u, subfieldCodes: ['a', 'z'], validIdentifierSubfieldCodes: ['a'], invalidIdentifierSubfieldCodes: ['z'], validatorAndNormalizer});
-  return {extract, compare, name: IDENTIFIER_NAME};
+  const {extract, compare} = createInterface({pattern: /^020$/u, subfieldCodes: ['a', 'z']});
+  return {extract, compare, name: 'ISBN'};
 };
-

package/src/match-detection/features/bib/standard-identifier-factory.js

@@ -1,4 +1,3 @@
-/* eslint-disable max-statements */
 /**
 *
 * @licstart  The following is the entire license notice for the JavaScript code in this file.
@@ -28,142 +27,60 @@
 */
 
 import createDebugLogger from 'debug';
-import {extractSubfieldsFromField, uniqueSubfields} from '../../../matching-utils';
+import {testStringOrNumber} from '../../../matching-utils';
 
-// Note about validity of standardIdentifiers:
-// We have three types of invalid standardIdentifiers:
-// 1. Formally invalid standardIdentifiers (ie. typos either in the resource or the record)
-// 2. Formally valid standardIdentifiers that are used in a wrong resource
-// 3. Canceled standardIdentifiers
-
-// Matcher could and should check that a standardIdentifier found in a subfield for a valid identifier is formally valid, and if it's not formally valid, handle it as an invalid standardIdentifier
-// Formally valid standardIdentifiers found in subfield for invalid identifier cannot be handled as valid standardIdentifiers, because they can be a case of type 2) or 3) invalid standardIdentifiers
-// We could also do a separate handling for formally valid an formally invalid standardIdentifiers
-
-export default ({pattern, subfieldCodes, identifier, validIdentifierSubfieldCodes = ['a'], invalidIdentifierSubfieldCodes = ['z'], validatorAndNormalizer = undefined}) => {
-  const debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:${identifier}`);
-  const debugData = debug.extend('data');
+const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');
+const debugData = debug.extend('data');
 
+export default ({pattern, subfieldCodes}) => {
   return {extract, compare};
 
-  function extract({record, recordExternal}) {
-    const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
-    const fields = record.get(pattern);
-    debugData(`${label}: ${fields.length} ${identifier}-fields `);
-
-    // extractIdentifierSubfield normalizes hyphens away from the subfield values
-    const identifiersFromFields = fields.map(field => extractSubfieldsFromField(field, subfieldCodes));
-    debugData(`${label}: IDs from fields (${identifiersFromFields.length}): ${JSON.stringify(identifiersFromFields)}`);
-    const allIdentifiers = identifiersFromFields.flat();
-    debugData(`${label}: Flat IDs from fields (${allIdentifiers.length}): ${JSON.stringify(allIdentifiers)}`);
-
-    const validatedAndNormalizedIdentifiers = validateAndNormalizeIdentifiers({identifierSubs: allIdentifiers, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes});
+  function extract({record}) {
+    const [field] = record.get(pattern);
 
-    const identifiers = uniqueSubfields(validatedAndNormalizedIdentifiers);
-
-    debugData(`${label}: Unique IDs from fields (${identifiers.length}): ${JSON.stringify(identifiers)}`);
-    return identifiers;
-
-    function validateAndNormalizeIdentifiers({identifierSubs, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}) {
-      if (validatorAndNormalizer) {
-        return identifierSubs.map((idSub) => validateAndNormalizeIdentifier({idSub, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}));
-      }
-      return identifierSubs.map((idSub) => normalizeHyphens(idSub));
+    if (field) {
+      return field.subfields
+        .filter(({code}) => subfieldCodes.includes(code))
+        .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))
+        .filter(value => value);
     }
 
-    function validateAndNormalizeIdentifier({idSub, validatorAndNormalizer, validIdentifierSubfieldCodes, invalidIdentifierSubfieldCodes}) {
-      const {valid, value} = validatorAndNormalizer(idSub.value);
-      if (validIdentifierSubfieldCodes.includes(idSub.code) && valid === false) {
-        const [code] = invalidIdentifierSubfieldCodes;
-        return {code, value};
-      }
-      return {code: idSub.code, value};
-    }
-
-    function normalizeHyphens(idSub) {
-      return {code: idSub.code, value: idSub.value.replace(/-/ug, '')};
-    }
-
-
+    return [];
   }
 
   function compare(a, b) {
-    debug(`Comparing A and B`);
     if (a.length === 0 || b.length === 0) {
-      debugData(`No standardidentifiers (${identifier}) to compare`);
+      debugData(`No standardidentifiers to compare`);
       return 0;
     }
 
-    debugData(`A: ${JSON.stringify(a)}`);
-    debugData(`B: ${JSON.stringify(b)}`);
-
-
     if (bothHaveValidIdentifiers()) {
-      // Compare only valid identifiers, if both have valid idenfiers
-      const {maxValues, possibleMatchValues, matchingValues} = getValueCount(true);
-      if (matchingValues < 1) {
-        debug(`Both have valid standardidentifiers (${identifier}), but none of these match.`);
-        return -0.75;
-      }
-      debug(`Both have valid standardidentifiers (${identifier}), ${matchingValues}/${possibleMatchValues} valid identifiers match.`);
-      // ignore non-matches if there is mismatching amount of values
-      debug(`Possible matches: ${possibleMatchValues}/${maxValues}`);
-      //we give some kind of penalty for mismatching amount of values instead of simple divide?
-      const penaltyForMissing = 0.1 * (maxValues - possibleMatchValues);
-      const penaltyForMisMatch = 0.2 * (possibleMatchValues - matchingValues);
-      debug(`\t points: penaltyForMissing: ${penaltyForMissing}`);
-      debug(`\t points: penaltyForMisMatch: ${penaltyForMisMatch}`);
-
-      return 0.75 - penaltyForMisMatch - penaltyForMissing;
-      //return matchingValues / possibleMatchValues * 0.75;
+      const {maxValues, matchingValues} = getValueCount(true);
+      return matchingValues / maxValues * 0.75;
     }
-    // If both do not have valid identifiers, compare all identifiers
-    const {maxValues, matchingValues} = getValueCount();
-    debug(`Both do NOT have valid standardidentifiers (${identifier}), ${matchingValues}/${maxValues} valid/invalid identifiers match.`);
 
+    const {maxValues, matchingValues} = getValueCount();
     return matchingValues / maxValues * 0.2;
 
     function bothHaveValidIdentifiers() {
-      const aValues = a.filter(({code}) => validIdentifierSubfieldCodes.includes(code));
-      const bValues = a.filter(({code}) => validIdentifierSubfieldCodes.includes(code));
-      debug(`A: ${aValues.length} valid ${identifier} identifiers`);
-      debug(`B: ${bValues.length} valid ${identifier} identifiers`);
+      const aValues = a.filter(({code}) => code === 'a');
+      const bValues = a.filter(({code}) => code === 'a');
       return aValues.length > 0 && bValues.length > 0;
     }
 
     function getValueCount(validOnly = false) {
-      const aValues = getIdentifiers(a, validOnly);
-      const bValues = getIdentifiers(b, validOnly);
-
-      const matchingValues = getMatchingValuesAmount(aValues, bValues);
+      const aValues = getIdentifiers(a);
+      const bValues = getIdentifiers(b);
 
       return {
         maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
-        // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
-        possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
-        matchingValues
+        matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length
       };
 
-      function getMatchingValuesAmount(aValues, bValues) {
-        if (bValues.length > aValues.length) {
-          return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
-        }
-        if (aValues.length > bValues.length) {
-          return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
-        }
-
-        // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases
-        // there would be duplicate values
-        const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;
-        const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;
-
-        return aToB < bToA ? aToB : bToA;
-      }
-
-      function getIdentifiers(values, validOnly) {
+      function getIdentifiers(values) {
         if (validOnly) {
           return values
-            .filter(({code}) => validIdentifierSubfieldCodes.includes(code))
+            .filter(({code}) => code === 'a')
             .map(({value}) => value);
         }
 

package/src/matching-utils.js

@@ -92,24 +92,3 @@ export function testStringOrNumber(value) {
   }
   return false;
 }
-
-export function extractSubfieldsFromField(field, subfieldCodes) {
-  if (field === undefined || field.subfields === undefined) {
-    return [];
-  }
-  const resultSubfields = field.subfields
-    .filter(({code}) => subfieldCodes.includes(code))
-    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))
-    .filter(value => value);
-  return resultSubfields;
-}
-
-export function uniqueSubfields(subfields) {
-  return subfields.reduce((arr, e) => {
-    if (!arr.find(item => item.code === e.code && item.value === e.value)) {
-      const newArr = arr.concat(e);
-      return newArr;
-    }
-    return arr;
-  }, []);
-}