@natlibfi/melinda-record-matching 2.2.0-alpha.4 → 2.2.1-alpha.2

package/.github/CODEOWNERS

@@ -0,0 +1,10 @@
+# With this line @NatLibFi/melinda-js-lead owns any files in the /.github/
+# directory at the root of the repository and any of its
+# subdirectories.
+/.github/ @NatLibFi/melinda-js-lead
+
+
+# With this line @NatLibFi/melinda-js-lead owns any files in the /src/
+# directory at the root of the repository and any of its
+# subdirectories.
+/src/ @NatLibFi/melinda-js-lead

package/.github/dependabot.yml

@@ -0,0 +1,40 @@
+# NatLibFi/Melinda maintenance strategy
+# https://docs.github.com/en/code-security/supply-chain-security/keeping-your-dependencies-updated-automatically/configuration-options-for-dependency-updates
+
+version: 2
+updates:
+  # Maintain dependencies for GitHub Actions
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "daily"
+      time: "06:30"
+      timezone: "Europe/Helsinki"
+
+  # Minor updates to npm production dependencies daily
+  - package-ecosystem: "npm"
+    directory: "/"
+    schedule:
+      interval: "daily"
+      time: "06:45"
+      timezone: "Europe/Helsinki"
+    versioning-strategy: lockfile-only
+    labels:
+      - "npm minor dependencies"
+    allow:
+      - dependency-type: "production"
+
+  # Major updates to npm dependencies weekly @tuesday
+  # Not possible yet https://github.com/dependabot/dependabot-core/issues/1778
+  # - package-ecosystem: "npm"
+  #   directory: "/"
+  #   schedule:
+  #     interval: "weekly"
+  #     day: "tuesday"
+  #     time: "07:00"
+  #     timezone: "Europe/Helsinki"
+  #   versioning-strategy: increase-if-necessary
+  #   labels:
+  #     - "npm major dependencies"
+  #   reviewers:
+  #     - "natlibfi/melinda-js-lead"

package/.github/workflows/melinda-node-tests.yml

@@ -0,0 +1,70 @@
+# Melinda node tests
+
+name: Melinda node tests
+
+on: push
+
+jobs:
+  build-node-versions:
+    name: Node version matrix
+    runs-on: ubuntu-latest
+
+    strategy:
+      matrix:
+        node-version: [14.x, 16.x, 18.x]
+        # See supported Node.js release schedule at https://nodejs.org/en/about/releases/
+
+    steps:
+    - name: Checkout the code
+      uses: actions/checkout@v2
+    - name: Use Node.js ${{ matrix.node-version }}
+      uses: actions/setup-node@v3
+      with:
+        node-version: ${{ matrix.node-version }}
+        cache: 'npm'
+      env:
+        NPM_CONFIG_IGNORE_SCRIPTS: true
+    - run: npm audit --package-lock-only --production --audit-level=high
+    - run: npm ci
+    - run: npm test
+    - run: npm run build --if-present
+
+  njsscan:
+    name: Njsscan check
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: Checkout the code
+      uses: actions/checkout@v3
+    - name: nodejsscan scan
+      id: njsscan
+      uses: ajinabraham/njsscan-action@master
+      with:
+        args: '.'
+
+  license-scan:
+    name: License compliance check
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+      - uses: mikaelvesavuori/license-compliance-action@v1.0.2
+        with:
+          exclude_pattern: /^@natlibfi/
+
+  publish:
+    runs-on: ubuntu-latest
+    needs: [build-node-versions, njsscan]
+    if: contains(github.ref, 'refs/tags/')
+
+    steps:
+      - uses: actions/checkout@v2
+      # Setup .npmrc file to publish to npm
+      - uses: actions/setup-node@v3
+        with:
+          node-version: '14.x'
+          registry-url: 'https://registry.npmjs.org'
+      - run: npm ci
+      - run: npm publish
+        env:
+          NODE_AUTH_TOKEN: ${{ secrets.MELINDA_RECORD_MATCHING_NPM_TOKEN }}

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

@@ -23,7 +23,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -104,7 +104,7 @@ function bibSourceIds(record) {
 
         function removeSourcePrefix(subfieldValue) {
           const sourcePrefixRegex = /^(?<sourcePrefix>\([A-Za-z0-9-]+\))(?<id>.+)$/u;
-          const normalizedValue = subfieldValue ? subfieldValue.replace(sourcePrefixRegex, '$<id>') : '';
+          const normalizedValue = (0, _matchingUtils.testStringOrNumber)(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';
           debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);
           return normalizedValue;
         }
@@ -112,7 +112,7 @@ function bibSourceIds(record) {
         function normalizeSidSubfieldValue(subfieldValue) {
           debugData(`Normalizing ${subfieldValue}`);
           const normalizeAwayRegex = /[- ]/u;
-          return subfieldValue ? subfieldValue.replace(normalizeAwayRegex, '') : '';
+          return (0, _matchingUtils.testStringOrNumber)(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';
         }
       }
     }
@@ -142,7 +142,7 @@ function bibMelindaIds(record) {
 
 function bibHostComponents(record) {
   const id = getHostId();
-  return id ? [`melinda.partsofhost=${id}`] : [];
+  return (0, _matchingUtils.testStringOrNumber)(id) ? [`melinda.partsofhost=${id}`] : [];
 
   function getHostId() {
     const [field] = record.get(/^773$/u);
@@ -154,12 +154,12 @@ function bibHostComponents(record) {
         code
       }) => code === 'w') || {};
 
-      if (value && /^\(FI-MELINDA\)/u.test(value)) {
-        return value.replace(/^\(FI-MELINDA\)/u, '');
+      if ((0, _matchingUtils.testStringOrNumber)(value) && /^\(FI-MELINDA\)/u.test(String(value))) {
+        return String(value).replace(/^\(FI-MELINDA\)/u, '');
       }
 
-      if (value && /^\(FIN01\)/u.test(value)) {
-        return value.replace(/^\(FIN01\)/u, '');
+      if ((0, _matchingUtils.testStringOrNumber)(value) && /^\(FIN01\)/u.test(String(value))) {
+        return String(value).replace(/^\(FIN01\)/u, '');
       }
 
       return false;
@@ -179,8 +179,8 @@ function bibTitle(record) {
   const title = getTitle();
   const booleanStartWords = ['and', 'or', 'nor', 'not'];
 
-  if (title) {
-    const formatted = title.replace(/[^\w\s\p{Alphabetic}]/gu, '') // Clean up concurrent spaces from fe. subfield changes
+  if ((0, _matchingUtils.testStringOrNumber)(title)) {
+    const formatted = String(title).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 titles starting with a boolean
 
     const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word)); // Prevent too many matches by having a minimum length requirement
@@ -198,7 +198,7 @@ function bibTitle(record) {
         code
       }) => ['a', 'b'].includes(code)).map(({
         value
-      }) => value) // In Melinda's index subfield separators are indexed as ' '
+      }) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').filter(value => value) // In Melinda's index subfield separators are indexed as ' '
       .join(' ');
     }
 
@@ -233,20 +233,20 @@ function bibStandardIdentifiers(record) {
     const otherIdReqExp = /^[A-Za-z0-9-:]+$/u;
 
     if (tag === '022') {
-      return subfields.filter(sub => ['a', 'z', 'y'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined).map(({
+      return subfields.filter(sub => ['a', 'z', 'y'].includes(sub.code) && (0, _matchingUtils.testStringOrNumber)(sub.value) && issnIsbnReqExp.test(String(sub.value))).map(({
         value
-      }) => value);
+      }) => String(value));
     }
 
     if (tag === '020') {
-      return subfields.filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined).map(({
+      return subfields.filter(sub => ['a', 'z'].includes(sub.code) && (0, _matchingUtils.testStringOrNumber)(sub.value) && issnIsbnReqExp.test(String(sub.value))).map(({
         value
-      }) => value);
+      }) => String(value));
     }
 
-    return subfields.filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined).map(({
+    return subfields.filter(sub => ['a', 'z'].includes(sub.code) && (0, _matchingUtils.testStringOrNumber)(sub.value) && otherIdReqExp.test(String(sub.value))).map(({
       value
-    }) => value);
+    }) => String(value));
   }
 }
 //# sourceMappingURL=bib.js.map

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","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","undefined"],"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 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} 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 = subfieldValue ? subfieldValue.replace(sourcePrefixRegex, '$<id>') : '';\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return subfieldValue ? subfieldValue.replace(normalizeAwayRegex, '') : '';\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${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 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const id = getHostId();\n  return id ? [`melinda.partsofhost=${id}`] : [];\n\n  function getHostId() {\n    const [field] = record.get(/^773$/u);\n\n    if (field) {\n      const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n      if (value && (/^\\(FI-MELINDA\\)/u).test(value)) {\n        return value.replace(/^\\(FI-MELINDA\\)/u, '');\n      }\n\n      if (value && (/^\\(FIN01\\)/u).test(value)) {\n        return value.replace(/^\\(FIN01\\)/u, '');\n      }\n\n      return false;\n    }\n    return false;\n  }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n  const title = getTitle();\n  const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n  if (title) {\n    const formatted = title\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for titles starting with a boolean\n    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n    // Prevent too many matches by having a minimum length requirement\n    return formatted.length >= 5 ? [`dc.title=\"${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}) => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n    }\n    return false;\n  }\n}\n\n// eslint-disable-next-line max-statements\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) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined)\n      .map(({value}) => value);\n  }\n}\n"],"mappings":";;;;;;;;;;;AA4BA;;AACA;;AACA;;;;AA7BA;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,YAAT,CAAsBC,MAAtB,EAA8B;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,qEAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB,CA3BmC,CA4BnC;;EAEAH,KAAK,CAAE,iCAAF,CAAL;EAEA,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAP,CAAW,KAAX,CAAd;EACAH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeJ,KAAf,CAAsB,EAAxD,CAAT;EAEA,OAAOA,KAAK,CAACE,MAAN,GAAe,CAAf,GAAmBG,YAAY,CAACL,KAAD,CAA/B,GAAyC,EAAhD;;EAEA,SAASK,YAAT,CAAsBL,KAAtB,EAA6B;IAC3BJ,KAAK,CAAE,wCAAF,CAAL;IAEA,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAD,CAAhC;;IAEA,IAAIM,UAAU,CAACJ,MAAX,GAAoB,CAAxB,EAA2B;MACzBN,KAAK,CAAE,2CAAF,CAAL;MACA,OAAO,EAAP;IACD;;IAED,MAAMY,UAAU,GAAG,IAAAC,+BAAA,EAAUH,UAAV,EAAsB,kBAAtB,CAAnB;IAEA,OAAOE,UAAP;;IAEA,SAASD,aAAT,CAAuBP,KAAvB,EAA8B;MAC5BJ,KAAK,CAAE,qCAAF,CAAL,CAD4B,CAG5B;;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAN,CAAUC,WAAV,EAAuBC,MAAvB,CAA8BC,WAAW,IAAIA,WAA7C,CAAnB;MACA,OAAOP,UAAP;;MAEA,SAASK,WAAT,CAAqBG,KAArB,EAA4B;QAC1BlB,KAAK,CAAE,6BAAF,CAAL;QAEA,OAAO,IAAAmB,6CAAA,EAA+BD,KAA/B,IAAwCE,eAAe,CAACF,KAAD,CAAvD,GAAiE,EAAxE;;QAEA,SAASE,eAAT,CAAyBF,KAAzB,EAAgC;UAC9BlB,KAAK,CAAE,8BAAF,CAAL;UACA,MAAM,CAACqB,GAAD,IAAQ,IAAAC,gCAAA,EAAkBJ,KAAlB,EAAyB,GAAzB,CAAd;UACA,MAAM,CAACK,GAAD,IAAQ,IAAAD,gCAAA,EAAkBJ,KAAlB,EAAyB,GAAzB,CAAd;UAEA,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAD,CAA1B,CAArC;UACA,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAD,CAA5C;UAEArB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAL,CAAea,GAAf,CAAoB,MAAKd,IAAI,CAACC,SAAL,CAAee,GAAf,CAAoB,EAAjD,CAAT;UACA,OAAOC,UAAU,CAACI,MAAX,CAAkBD,UAAlB,CAAP;QACD;;QAED,SAASF,kBAAT,CAA4BI,aAA5B,EAA2C;UACzC,MAAMC,iBAAiB,GAAI,gDAA3B;UACA,MAAMC,eAAe,GAAGF,aAAa,GAAGA,aAAa,CAACG,OAAd,CAAsBF,iBAAtB,EAAyC,OAAzC,CAAH,GAAuD,EAA5F;UACA5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAnD,CAAT;UACA,OAAOA,eAAP;QACD;;QAED,SAASL,yBAAT,CAAmCG,aAAnC,EAAkD;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAA9B,CAAT;UACA,MAAMI,kBAAkB,GAAI,OAA5B;UACA,OAAOJ,aAAa,GAAGA,aAAa,CAACG,OAAd,CAAsBC,kBAAtB,EAA0C,EAA1C,CAAH,GAAmD,EAAvE;QACD;MAEF;IACF;EACF;AACF;;AAEM,SAASC,aAAT,CAAuBnC,MAAvB,EAA+B;EACpC;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,wEAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EACAH,KAAK,CAAE,iCAAF,CAAL,CALoC,CAOpC;EACA;EACA;;EACA,MAAMmC,UAAU,GAAG,IAAAC,gCAAA,EAAkBrC,MAAlB,CAAnB;EAEAG,SAAS,CAAE,uBAAsBiC,UAAU,CAAC7B,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAe2B,UAAf,CAA2B,EAA1E,CAAT;;EAEA,IAAIA,UAAU,CAAC7B,MAAX,GAAoB,CAAxB,EAA2B;IACzBN,KAAK,CAAE,2CAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,OAAO,IAAAa,+BAAA,EAAUsB,UAAV,EAAsB,mBAAtB,CAAP;AACD,C,CAGD;AACA;;;AACO,SAASE,iBAAT,CAA2BtC,MAA3B,EAAmC;EACxC,MAAMuC,EAAE,GAAGC,SAAS,EAApB;EACA,OAAOD,EAAE,GAAG,CAAE,uBAAsBA,EAAG,EAA3B,CAAH,GAAmC,EAA5C;;EAEA,SAASC,SAAT,GAAqB;IACnB,MAAM,CAACrB,KAAD,IAAUnB,MAAM,CAACM,GAAP,CAAW,QAAX,CAAhB;;IAEA,IAAIa,KAAJ,EAAW;MACT,MAAM;QAACsB;MAAD,IAAUtB,KAAK,CAACuB,SAAN,CAAgBC,IAAhB,CAAqB,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA1C,KAAkD,EAAlE;;MAEA,IAAIH,KAAK,IAAK,kBAAD,CAAqBI,IAArB,CAA0BJ,KAA1B,CAAb,EAA+C;QAC7C,OAAOA,KAAK,CAACR,OAAN,CAAc,kBAAd,EAAkC,EAAlC,CAAP;MACD;;MAED,IAAIQ,KAAK,IAAK,aAAD,CAAgBI,IAAhB,CAAqBJ,KAArB,CAAb,EAA0C;QACxC,OAAOA,KAAK,CAACR,OAAN,CAAc,aAAd,EAA6B,EAA7B,CAAP;MACD;;MAED,OAAO,KAAP;IACD;;IACD,OAAO,KAAP;EACD;AACF,C,CAED;AACA;AACA;AAEA;AACA;AACA;;;AAEO,SAASa,QAAT,CAAkB9C,MAAlB,EAA0B;EAC/B,MAAM+C,KAAK,GAAGC,QAAQ,EAAtB;EACA,MAAMC,iBAAiB,GAAG,CAAC,KAAD,EAAQ,IAAR,EAAc,KAAd,EAAqB,KAArB,CAA1B;;EAEA,IAAIF,KAAJ,EAAW;IACT,MAAMG,SAAS,GAAGH,KAAK,CACpBd,OADe,CACP,yBADO,EACoB,EADpB,EAEhB;IAFgB,CAGfA,OAHe,CAGP,MAHO,EAGC,GAHD,EAIfkB,IAJe,GAKfC,KALe,CAKT,CALS,EAKN,EALM,EAMfD,IANe,EAAlB,CADS,CAST;;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAlB,CAAuBC,IAAI,IAAIL,SAAS,CAACM,WAAV,GAAwBC,UAAxB,CAAmCF,IAAnC,CAA/B,CAAtB,CAVS,CAWT;;IACA,OAAOL,SAAS,CAAC3C,MAAV,IAAoB,CAApB,GAAwB,CAAE,aAAY8C,aAAa,GAAG,EAAH,GAAQ,GAAI,GAAEH,SAAU,IAAnD,CAAxB,GAAkF,EAAzF;EACD;;EAED,OAAO,EAAP;;EAEA,SAASF,QAAT,GAAoB;IAClB,MAAM,CAAC7B,KAAD,IAAUnB,MAAM,CAACM,GAAP,CAAW,QAAX,CAAhB;;IAEA,IAAIa,KAAJ,EAAW;MACT,OAAOA,KAAK,CAACuB,SAAN,CACJzB,MADI,CACG,CAAC;QAAC2B;MAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWc,QAAX,CAAoBd,IAApB,CADf,EAEJ7B,GAFI,CAEA,CAAC;QAAC0B;MAAD,CAAD,KAAaA,KAFb,EAGL;MAHK,CAIJkB,IAJI,CAIC,GAJD,CAAP;IAKD;;IACD,OAAO,KAAP;EACD;AACF,C,CAED;;;AACO,SAASC,sBAAT,CAAgC5D,MAAhC,EAAwC;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,iFAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EACAH,KAAK,CAAE,2CAAF,CAAL;EAEA,MAAM4D,MAAM,GAAG7D,MAAM,CAACM,GAAP,CAAW,wBAAX,CAAf;EACA,MAAMwD,WAAW,GAAG,GAAGjC,MAAH,CAAU,GAAGgC,MAAM,CAAC9C,GAAP,CAAWgD,aAAX,CAAb,CAApB;EACA,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQH,WAAR,CAAJ,CAA1B;EAEA3D,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAL,CAAeoD,MAAf,CAAuB,EAAvD,CAAT;EACA1D,SAAS,CAAE,gBAAe2D,WAAW,CAACvD,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeqD,WAAf,CAA4B,EAArE,CAAT;EACA3D,SAAS,CAAE,uBAAsB6D,iBAAiB,CAACzD,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeuD,iBAAf,CAAkC,EAAxF,CAAT;;EAEA,IAAIA,iBAAiB,CAACzD,MAAlB,GAA2B,CAA/B,EAAkC;IAChCN,KAAK,CAAE,2CAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,OAAO,IAAAa,+BAAA,EAAUkD,iBAAV,EAA6B,eAA7B,CAAP;;EAEA,SAASD,aAAT,CAAuB;IAACG,GAAD;IAAMxB;EAAN,CAAvB,EAAyC;IACvC,MAAMyB,cAAc,GAAI,kBAAxB;IACA,MAAMC,aAAa,GAAI,mBAAvB;;IAEA,IAAIF,GAAG,KAAK,KAAZ,EAAmB;MACjB,OAAOxB,SAAS,CACbzB,MADI,CACGoD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAW,GAAX,EAAgBX,QAAhB,CAAyBW,GAAG,CAACzB,IAA7B,KAAsCuB,cAAc,CAACtB,IAAf,CAAoBwB,GAAG,CAAC5B,KAAxB,CAAtC,IAAwE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SADhG,EAEJvD,GAFI,CAEA,CAAC;QAAC0B;MAAD,CAAD,KAAaA,KAFb,CAAP;IAGD;;IAED,IAAIyB,GAAG,KAAK,KAAZ,EAAmB;MACjB,OAAOxB,SAAS,CACbzB,MADI,CACGoD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiCuB,cAAc,CAACtB,IAAf,CAAoBwB,GAAG,CAAC5B,KAAxB,CAAjC,IAAmE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SAD3F,EAEJvD,GAFI,CAEA,CAAC;QAAC0B;MAAD,CAAD,KAAaA,KAFb,CAAP;IAGD;;IAED,OAAOC,SAAS,CACbzB,MADI,CACGoD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiCwB,aAAa,CAACvB,IAAd,CAAmBwB,GAAG,CAAC5B,KAAvB,CAAjC,IAAkE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SAD1F,EAEJvD,GAFI,CAEA,CAAC;MAAC0B;IAAD,CAAD,KAAaA,KAFb,CAAP;EAGD;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","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\n// eslint-disable-next-line max-statements\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;;;;AA7BA;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,YAAT,CAAsBC,MAAtB,EAA8B;EAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;EAQE,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,qEAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB,CA3BmC,CA4BnC;;EAEAH,KAAK,CAAE,iCAAF,CAAL;EAEA,MAAMI,KAAK,GAAGL,MAAM,CAACM,GAAP,CAAW,KAAX,CAAd;EACAH,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeJ,KAAf,CAAsB,EAAxD,CAAT;EAEA,OAAOA,KAAK,CAACE,MAAN,GAAe,CAAf,GAAmBG,YAAY,CAACL,KAAD,CAA/B,GAAyC,EAAhD;;EAEA,SAASK,YAAT,CAAsBL,KAAtB,EAA6B;IAC3BJ,KAAK,CAAE,wCAAF,CAAL;IAEA,MAAMU,UAAU,GAAGC,aAAa,CAACP,KAAD,CAAhC;;IAEA,IAAIM,UAAU,CAACJ,MAAX,GAAoB,CAAxB,EAA2B;MACzBN,KAAK,CAAE,2CAAF,CAAL;MACA,OAAO,EAAP;IACD;;IAED,MAAMY,UAAU,GAAG,IAAAC,+BAAA,EAAUH,UAAV,EAAsB,kBAAtB,CAAnB;IAEA,OAAOE,UAAP;;IAEA,SAASD,aAAT,CAAuBP,KAAvB,EAA8B;MAC5BJ,KAAK,CAAE,qCAAF,CAAL,CAD4B,CAG5B;;MACA,MAAMU,UAAU,GAAGN,KAAK,CAACU,GAAN,CAAUC,WAAV,EAAuBC,MAAvB,CAA8BC,WAAW,IAAIA,WAA7C,CAAnB;MACA,OAAOP,UAAP;;MAEA,SAASK,WAAT,CAAqBG,KAArB,EAA4B;QAC1BlB,KAAK,CAAE,6BAAF,CAAL;QAEA,OAAO,IAAAmB,6CAAA,EAA+BD,KAA/B,IAAwCE,eAAe,CAACF,KAAD,CAAvD,GAAiE,EAAxE;;QAEA,SAASE,eAAT,CAAyBF,KAAzB,EAAgC;UAC9BlB,KAAK,CAAE,8BAAF,CAAL;UACA,MAAM,CAACqB,GAAD,IAAQ,IAAAC,gCAAA,EAAkBJ,KAAlB,EAAyB,GAAzB,CAAd;UACA,MAAM,CAACK,GAAD,IAAQ,IAAAD,gCAAA,EAAkBJ,KAAlB,EAAyB,GAAzB,CAAd;UAEA,MAAMM,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACL,GAAD,CAA1B,CAArC;UACA,MAAMM,UAAU,GAAGD,yBAAyB,CAACH,GAAD,CAA5C;UAEArB,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAL,CAAea,GAAf,CAAoB,MAAKd,IAAI,CAACC,SAAL,CAAee,GAAf,CAAoB,EAAjD,CAAT;UACA,OAAOC,UAAU,CAACI,MAAX,CAAkBD,UAAlB,CAAP;QACD;;QAED,SAASF,kBAAT,CAA4BI,aAA5B,EAA2C;UACzC,MAAMC,iBAAiB,GAAI,gDAA3B;UACA,MAAMC,eAAe,GAAG,IAAAC,iCAAA,EAAmBH,aAAnB,IAAoCI,MAAM,CAACJ,aAAD,CAAN,CAAsBK,OAAtB,CAA8BJ,iBAA9B,EAAiD,OAAjD,CAApC,GAAgG,EAAxH;UACA5B,SAAS,CAAE,cAAa2B,aAAc,OAAME,eAAgB,EAAnD,CAAT;UACA,OAAOA,eAAP;QACD;;QAED,SAASL,yBAAT,CAAmCG,aAAnC,EAAkD;UAChD3B,SAAS,CAAE,eAAc2B,aAAc,EAA9B,CAAT;UACA,MAAMM,kBAAkB,GAAI,OAA5B;UACA,OAAO,IAAAH,iCAAA,EAAmBH,aAAnB,IAAoCI,MAAM,CAACJ,aAAD,CAAN,CAAsBK,OAAtB,CAA8BC,kBAA9B,EAAkD,EAAlD,CAApC,GAA4F,EAAnG;QACD;MAEF;IACF;EACF;AACF;;AAEM,SAASC,aAAT,CAAuBrC,MAAvB,EAA+B;EACpC;EAEA,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,wEAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EACAH,KAAK,CAAE,iCAAF,CAAL,CALoC,CAOpC;EACA;EACA;;EACA,MAAMqC,UAAU,GAAG,IAAAC,gCAAA,EAAkBvC,MAAlB,CAAnB;EAEAG,SAAS,CAAE,uBAAsBmC,UAAU,CAAC/B,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAe6B,UAAf,CAA2B,EAA1E,CAAT;;EAEA,IAAIA,UAAU,CAAC/B,MAAX,GAAoB,CAAxB,EAA2B;IACzBN,KAAK,CAAE,2CAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,OAAO,IAAAa,+BAAA,EAAUwB,UAAV,EAAsB,mBAAtB,CAAP;AACD,C,CAGD;AACA;;;AACO,SAASE,iBAAT,CAA2BxC,MAA3B,EAAmC;EACxC,MAAMyC,EAAE,GAAGC,SAAS,EAApB;EACA,OAAO,IAAAT,iCAAA,EAAmBQ,EAAnB,IAAyB,CAAE,uBAAsBA,EAAG,EAA3B,CAAzB,GAAyD,EAAhE;;EAEA,SAASC,SAAT,GAAqB;IACnB,MAAM,CAACvB,KAAD,IAAUnB,MAAM,CAACM,GAAP,CAAW,QAAX,CAAhB;;IAEA,IAAIa,KAAJ,EAAW;MACT,MAAM;QAACwB;MAAD,IAAUxB,KAAK,CAACyB,SAAN,CAAgBC,IAAhB,CAAqB,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA1C,KAAkD,EAAlE;;MAEA,IAAI,IAAAb,iCAAA,EAAmBU,KAAnB,KAA8B,kBAAD,CAAqBI,IAArB,CAA0Bb,MAAM,CAACS,KAAD,CAAhC,CAAjC,EAA2E;QACzE,OAAOT,MAAM,CAACS,KAAD,CAAN,CAAcR,OAAd,CAAsB,kBAAtB,EAA0C,EAA1C,CAAP;MACD;;MAED,IAAI,IAAAF,iCAAA,EAAmBU,KAAnB,KAA8B,aAAD,CAAgBI,IAAhB,CAAqBb,MAAM,CAACS,KAAD,CAA3B,CAAjC,EAAsE;QACpE,OAAOT,MAAM,CAACS,KAAD,CAAN,CAAcR,OAAd,CAAsB,aAAtB,EAAqC,EAArC,CAAP;MACD;;MAED,OAAO,KAAP;IACD;;IACD,OAAO,KAAP;EACD;AACF,C,CAED;AACA;AACA;AAEA;AACA;AACA;;;AAEO,SAASa,QAAT,CAAkBhD,MAAlB,EAA0B;EAC/B,MAAMiD,KAAK,GAAGC,QAAQ,EAAtB;EACA,MAAMC,iBAAiB,GAAG,CAAC,KAAD,EAAQ,IAAR,EAAc,KAAd,EAAqB,KAArB,CAA1B;;EAEA,IAAI,IAAAlB,iCAAA,EAAmBgB,KAAnB,CAAJ,EAA+B;IAC7B,MAAMG,SAAS,GAAGlB,MAAM,CAACe,KAAD,CAAN,CACfd,OADe,CACP,yBADO,EACoB,EADpB,EAEhB;IAFgB,CAGfA,OAHe,CAGP,MAHO,EAGC,GAHD,EAIfkB,IAJe,GAKfC,KALe,CAKT,CALS,EAKN,EALM,EAMfD,IANe,EAAlB,CAD6B,CAS7B;;IACA,MAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAlB,CAAuBC,IAAI,IAAIL,SAAS,CAACM,WAAV,GAAwBC,UAAxB,CAAmCF,IAAnC,CAA/B,CAAtB,CAV6B,CAW7B;;IACA,OAAOL,SAAS,CAAC7C,MAAV,IAAoB,CAApB,GAAwB,CAAE,aAAYgD,aAAa,GAAG,EAAH,GAAQ,GAAI,GAAEH,SAAU,IAAnD,CAAxB,GAAkF,EAAzF;EACD;;EAED,OAAO,EAAP;;EAEA,SAASF,QAAT,GAAoB;IAClB,MAAM,CAAC/B,KAAD,IAAUnB,MAAM,CAACM,GAAP,CAAW,QAAX,CAAhB;;IAEA,IAAIa,KAAJ,EAAW;MACT,OAAOA,KAAK,CAACyB,SAAN,CACJ3B,MADI,CACG,CAAC;QAAC6B;MAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWc,QAAX,CAAoBd,IAApB,CADf,EAEJ/B,GAFI,CAEA,CAAC;QAAC4B;MAAD,CAAD,KAAa,IAAAV,iCAAA,EAAmBU,KAAnB,IAA4BT,MAAM,CAACS,KAAD,CAAlC,GAA4C,EAFzD,EAGJ1B,MAHI,CAGG0B,KAAK,IAAIA,KAHZ,EAIL;MAJK,CAKJkB,IALI,CAKC,GALD,CAAP;IAMD;;IACD,OAAO,KAAP;EACD;AACF,C,CAED;;;AACO,SAASC,sBAAT,CAAgC9D,MAAhC,EAAwC;EAE7C,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,iFAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EACAH,KAAK,CAAE,2CAAF,CAAL;EAEA,MAAM8D,MAAM,GAAG/D,MAAM,CAACM,GAAP,CAAW,wBAAX,CAAf;EACA,MAAM0D,WAAW,GAAG,GAAGnC,MAAH,CAAU,GAAGkC,MAAM,CAAChD,GAAP,CAAWkD,aAAX,CAAb,CAApB;EACA,MAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQH,WAAR,CAAJ,CAA1B;EAEA7D,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAL,CAAesD,MAAf,CAAuB,EAAvD,CAAT;EACA5D,SAAS,CAAE,gBAAe6D,WAAW,CAACzD,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeuD,WAAf,CAA4B,EAArE,CAAT;EACA7D,SAAS,CAAE,uBAAsB+D,iBAAiB,CAAC3D,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeyD,iBAAf,CAAkC,EAAxF,CAAT;;EAEA,IAAIA,iBAAiB,CAAC3D,MAAlB,GAA2B,CAA/B,EAAkC;IAChCN,KAAK,CAAE,2CAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,OAAO,IAAAa,+BAAA,EAAUoD,iBAAV,EAA6B,eAA7B,CAAP;;EAEA,SAASD,aAAT,CAAuB;IAACG,GAAD;IAAMxB;EAAN,CAAvB,EAAyC;IACvC,MAAMyB,cAAc,GAAI,kBAAxB;IACA,MAAMC,aAAa,GAAI,mBAAvB;;IAEA,IAAIF,GAAG,KAAK,KAAZ,EAAmB;MACjB,OAAOxB,SAAS,CACb3B,MADI,CACGsD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAW,GAAX,EAAgBX,QAAhB,CAAyBW,GAAG,CAACzB,IAA7B,KAAsC,IAAAb,iCAAA,EAAmBsC,GAAG,CAAC5B,KAAvB,CAAtC,IAAuE0B,cAAc,CAACtB,IAAf,CAAoBb,MAAM,CAACqC,GAAG,CAAC5B,KAAL,CAA1B,CADjF,EAEJ5B,GAFI,CAEA,CAAC;QAAC4B;MAAD,CAAD,KAAaT,MAAM,CAACS,KAAD,CAFnB,CAAP;IAGD;;IAED,IAAIyB,GAAG,KAAK,KAAZ,EAAmB;MACjB,OAAOxB,SAAS,CACb3B,MADI,CACGsD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiC,IAAAb,iCAAA,EAAmBsC,GAAG,CAAC5B,KAAvB,CAAjC,IAAkE0B,cAAc,CAACtB,IAAf,CAAoBb,MAAM,CAACqC,GAAG,CAAC5B,KAAL,CAA1B,CAD5E,EAEJ5B,GAFI,CAEA,CAAC;QAAC4B;MAAD,CAAD,KAAaT,MAAM,CAACS,KAAD,CAFnB,CAAP;IAGD;;IAED,OAAOC,SAAS,CACb3B,MADI,CACGsD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiC,IAAAb,iCAAA,EAAmBsC,GAAG,CAAC5B,KAAvB,CAAjC,IAAkE2B,aAAa,CAACvB,IAAd,CAAmBb,MAAM,CAACqC,GAAG,CAAC5B,KAAL,CAAzB,CAD5E,EAEJ5B,GAFI,CAEA,CAAC;MAAC4B;IAAD,CAAD,KAAaT,MAAM,CAACS,KAAD,CAFnB,CAAP;EAGD;AACF"}

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

@@ -17,7 +17,7 @@ function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj &&
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *

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

@@ -1 +1 @@
-{"version":3,"file":"index.js","names":["searchTypes","bib","standardIdentifiers","hostComponents","title","melindaId","sourceIds","record","searchSpec","extractors","map","generateQueryExtractor","cb","flat","type","Error"],"sources":["../../../src/candidate-search/query-list/index.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport * as bib from './bib';\n\nexport const searchTypes = {\n  bib: {\n    standardIdentifiers: 'bibStandardIdentifiers',\n    hostComponents: 'bibHostComponents',\n    title: 'bibTitle',\n    melindaId: 'bibMelindaIds',\n    sourceIds: 'bibSourceIds'\n  }\n};\n\nexport default (record, searchSpec) => {\n  const extractors = {...bib};\n\n  return searchSpec\n    .map(generateQueryExtractor)\n    .map(cb => cb(record))\n    .flat();\n\n  function generateQueryExtractor(type) {\n    if (extractors[type]) {\n      return extractors[type];\n    }\n\n    throw new Error(`Unknown search type: ${type}`);\n  }\n};\n"],"mappings":";;;;;;;AA4BA;;;;;;AA5BA;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;AAIO,MAAMA,WAAW,GAAG;EACzBC,GAAG,EAAE;IACHC,mBAAmB,EAAE,wBADlB;IAEHC,cAAc,EAAE,mBAFb;IAGHC,KAAK,EAAE,UAHJ;IAIHC,SAAS,EAAE,eAJR;IAKHC,SAAS,EAAE;EALR;AADoB,CAApB;;;eAUQ,CAACC,MAAD,EAASC,UAAT,KAAwB;EACrC,MAAMC,UAAU,GAAG,EAAC,GAAGR;EAAJ,CAAnB;EAEA,OAAOO,UAAU,CACdE,GADI,CACAC,sBADA,EAEJD,GAFI,CAEAE,EAAE,IAAIA,EAAE,CAACL,MAAD,CAFR,EAGJM,IAHI,EAAP;;EAKA,SAASF,sBAAT,CAAgCG,IAAhC,EAAsC;IACpC,IAAIL,UAAU,CAACK,IAAD,CAAd,EAAsB;MACpB,OAAOL,UAAU,CAACK,IAAD,CAAjB;IACD;;IAED,MAAM,IAAIC,KAAJ,CAAW,wBAAuBD,IAAK,EAAvC,CAAN;EACD;AACF,C"}
+{"version":3,"file":"index.js","names":["searchTypes","bib","standardIdentifiers","hostComponents","title","melindaId","sourceIds","record","searchSpec","extractors","map","generateQueryExtractor","cb","flat","type","Error"],"sources":["../../../src/candidate-search/query-list/index.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 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 * as bib from './bib';\n\nexport const searchTypes = {\n  bib: {\n    standardIdentifiers: 'bibStandardIdentifiers',\n    hostComponents: 'bibHostComponents',\n    title: 'bibTitle',\n    melindaId: 'bibMelindaIds',\n    sourceIds: 'bibSourceIds'\n  }\n};\n\nexport default (record, searchSpec) => {\n  const extractors = {...bib};\n\n  return searchSpec\n    .map(generateQueryExtractor)\n    .map(cb => cb(record))\n    .flat();\n\n  function generateQueryExtractor(type) {\n    if (extractors[type]) {\n      return extractors[type];\n    }\n\n    throw new Error(`Unknown search type: ${type}`);\n  }\n};\n"],"mappings":";;;;;;;AA4BA;;;;;;AA5BA;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;AAIO,MAAMA,WAAW,GAAG;EACzBC,GAAG,EAAE;IACHC,mBAAmB,EAAE,wBADlB;IAEHC,cAAc,EAAE,mBAFb;IAGHC,KAAK,EAAE,UAHJ;IAIHC,SAAS,EAAE,eAJR;IAKHC,SAAS,EAAE;EALR;AADoB,CAApB;;;eAUQ,CAACC,MAAD,EAASC,UAAT,KAAwB;EACrC,MAAMC,UAAU,GAAG,EAAC,GAAGR;EAAJ,CAAnB;EAEA,OAAOO,UAAU,CACdE,GADI,CACAC,sBADA,EAEJD,GAFI,CAEAE,EAAE,IAAIA,EAAE,CAACL,MAAD,CAFR,EAGJM,IAHI,EAAP;;EAKA,SAASF,sBAAT,CAAgCG,IAAhC,EAAsC;IACpC,IAAIL,UAAU,CAACK,IAAD,CAAd,EAAsB;MACpB,OAAOL,UAAU,CAACK,IAAD,CAAjB;IACD;;IAED,MAAM,IAAIC,KAAJ,CAAW,wBAAuBD,IAAK,EAAvC,CAAN;EACD;AACF,C"}

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

@@ -7,13 +7,15 @@ exports.default = void 0;
 
 var _natural = require("natural");
 
+var _matchingUtils = require("../../../matching-utils");
+
 /**
 *
 * @licstart  The following is the entire license notice for the JavaScript code in this file.
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -34,6 +36,7 @@ var _natural = require("natural");
 * for the JavaScript code in this file.
 *
 */
+// We should extract also organisational authors (110/710)
 var _default = ({
   nameTreshold = 10
 } = {}) => ({
@@ -56,7 +59,7 @@ var _default = ({
     }) {
       if (code === 'a') {
         return {
-          name: value ? value.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''
+          name: (0, _matchingUtils.testStringOrNumber)(value) ? String(value).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''
         };
       }
 

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

@@ -1 +1 @@
-{"version":3,"file":"authors.js","names":["nameTreshold","name","extract","record","get","map","subfields","filter","code","value","includes","toObj","reduce","acc","v","replace","toLowerCase","id","compare","a","b","maxAuthors","length","matchingIds","findMatchingIds","matchingNames","findMatchingNames","idPoints","namePoints","totalPoints","findMatches","distance","leven","maxLength","getMaxLength","percentage","key","cb","valuesA","o","valuesB","allValues","concat","occurrance","otherValue"],"sources":["../../../../src/match-detection/features/bib/authors.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 {LevenshteinDistance as leven} from 'natural';\n\nexport default ({nameTreshold = 10} = {}) => ({\n  name: 'Authors',\n  extract: record => record.get(/^(?<def>100|700)$/u)\n    .map(({subfields}) => {\n      return subfields\n        .filter(({code, value}) => code && value)\n        .filter(({code}) => ['a', '0'].includes(code))\n        .map(toObj)\n        .reduce((acc, v) => ({...acc, ...v}), {});\n\n      function toObj({code, value}) {\n        if (code === 'a') {\n          return {name: value ? value.replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase() : ''};\n        }\n\n        return {id: value};\n      }\n    }),\n  compare: (a, b) => {\n    const maxAuthors = a.length > b.length ? a.length : b.length;\n    const matchingIds = findMatchingIds();\n\n    if (maxAuthors >= 3 && matchingIds >= 3) {\n      return 0.3;\n    }\n\n    const matchingNames = findMatchingNames();\n    const idPoints = matchingIds / maxAuthors * 0.3;\n    const namePoints = matchingNames / maxAuthors * 0.2;\n    const totalPoints = idPoints + namePoints;\n\n    return totalPoints <= 0.2 ? totalPoints : 0.2;\n\n    function findMatchingIds() {\n      return findMatches('id', (a, b) => a === b);\n    }\n\n    function findMatchingNames() {\n      return findMatches('name', (a, b) => {\n        const distance = leven(a, b);\n\n        if (distance === 0) {\n          return true;\n        }\n\n        const maxLength = getMaxLength();\n        const percentage = distance / maxLength * 100;\n\n        return percentage <= nameTreshold;\n\n        function getMaxLength() {\n          return a.length > b.length ? a.length : b.length;\n        }\n      });\n    }\n\n    function findMatches(key, cb) {\n      const valuesA = a.filter(o => o[key]).map(o => o[key]);\n      const valuesB = b.filter(o => o[key]).map(o => o[key]);\n      const allValues = valuesA.concat(valuesB);\n\n      return allValues.reduce((acc, value) => {\n        const occurrance = allValues.filter(otherValue => cb(value, otherValue)).length;\n        return occurrance >= 2 ? acc + 1 : acc;\n      }, 0);\n    }\n  }\n});\n"],"mappings":";;;;;;;AA4BA;;AA5BA;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;eAIe,CAAC;EAACA,YAAY,GAAG;AAAhB,IAAsB,EAAvB,MAA+B;EAC5CC,IAAI,EAAE,SADsC;EAE5CC,OAAO,EAAEC,MAAM,IAAIA,MAAM,CAACC,GAAP,CAAW,oBAAX,EAChBC,GADgB,CACZ,CAAC;IAACC;EAAD,CAAD,KAAiB;IACpB,OAAOA,SAAS,CACbC,MADI,CACG,CAAC;MAACC,IAAD;MAAOC;IAAP,CAAD,KAAmBD,IAAI,IAAIC,KAD9B,EAEJF,MAFI,CAEG,CAAC;MAACC;IAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWE,QAAX,CAAoBF,IAApB,CAFf,EAGJH,GAHI,CAGAM,KAHA,EAIJC,MAJI,CAIG,CAACC,GAAD,EAAMC,CAAN,MAAa,EAAC,GAAGD,GAAJ;MAAS,GAAGC;IAAZ,CAAb,CAJH,EAIiC,EAJjC,CAAP;;IAMA,SAASH,KAAT,CAAe;MAACH,IAAD;MAAOC;IAAP,CAAf,EAA8B;MAC5B,IAAID,IAAI,KAAK,GAAb,EAAkB;QAChB,OAAO;UAACP,IAAI,EAAEQ,KAAK,GAAGA,KAAK,CAACM,OAAN,CAAc,2BAAd,EAA2C,EAA3C,EAA+CC,WAA/C,EAAH,GAAkE;QAA9E,CAAP;MACD;;MAED,OAAO;QAACC,EAAE,EAAER;MAAL,CAAP;IACD;EACF,CAfgB,CAFyB;EAkB5CS,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB,MAAMC,UAAU,GAAGF,CAAC,CAACG,MAAF,GAAWF,CAAC,CAACE,MAAb,GAAsBH,CAAC,CAACG,MAAxB,GAAiCF,CAAC,CAACE,MAAtD;IACA,MAAMC,WAAW,GAAGC,eAAe,EAAnC;;IAEA,IAAIH,UAAU,IAAI,CAAd,IAAmBE,WAAW,IAAI,CAAtC,EAAyC;MACvC,OAAO,GAAP;IACD;;IAED,MAAME,aAAa,GAAGC,iBAAiB,EAAvC;IACA,MAAMC,QAAQ,GAAGJ,WAAW,GAAGF,UAAd,GAA2B,GAA5C;IACA,MAAMO,UAAU,GAAGH,aAAa,GAAGJ,UAAhB,GAA6B,GAAhD;IACA,MAAMQ,WAAW,GAAGF,QAAQ,GAAGC,UAA/B;IAEA,OAAOC,WAAW,IAAI,GAAf,GAAqBA,WAArB,GAAmC,GAA1C;;IAEA,SAASL,eAAT,GAA2B;MACzB,OAAOM,WAAW,CAAC,IAAD,EAAO,CAACX,CAAD,EAAIC,CAAJ,KAAUD,CAAC,KAAKC,CAAvB,CAAlB;IACD;;IAED,SAASM,iBAAT,GAA6B;MAC3B,OAAOI,WAAW,CAAC,MAAD,EAAS,CAACX,CAAD,EAAIC,CAAJ,KAAU;QACnC,MAAMW,QAAQ,GAAG,IAAAC,4BAAA,EAAMb,CAAN,EAASC,CAAT,CAAjB;;QAEA,IAAIW,QAAQ,KAAK,CAAjB,EAAoB;UAClB,OAAO,IAAP;QACD;;QAED,MAAME,SAAS,GAAGC,YAAY,EAA9B;QACA,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAX,GAAuB,GAA1C;QAEA,OAAOE,UAAU,IAAInC,YAArB;;QAEA,SAASkC,YAAT,GAAwB;UACtB,OAAOf,CAAC,CAACG,MAAF,GAAWF,CAAC,CAACE,MAAb,GAAsBH,CAAC,CAACG,MAAxB,GAAiCF,CAAC,CAACE,MAA1C;QACD;MACF,CAfiB,CAAlB;IAgBD;;IAED,SAASQ,WAAT,CAAqBM,GAArB,EAA0BC,EAA1B,EAA8B;MAC5B,MAAMC,OAAO,GAAGnB,CAAC,CAACZ,MAAF,CAASgC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAf,EAAsB/B,GAAtB,CAA0BkC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAhC,CAAhB;MACA,MAAMI,OAAO,GAAGpB,CAAC,CAACb,MAAF,CAASgC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAf,EAAsB/B,GAAtB,CAA0BkC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAhC,CAAhB;MACA,MAAMK,SAAS,GAAGH,OAAO,CAACI,MAAR,CAAeF,OAAf,CAAlB;MAEA,OAAOC,SAAS,CAAC7B,MAAV,CAAiB,CAACC,GAAD,EAAMJ,KAAN,KAAgB;QACtC,MAAMkC,UAAU,GAAGF,SAAS,CAAClC,MAAV,CAAiBqC,UAAU,IAAIP,EAAE,CAAC5B,KAAD,EAAQmC,UAAR,CAAjC,EAAsDtB,MAAzE;QACA,OAAOqB,UAAU,IAAI,CAAd,GAAkB9B,GAAG,GAAG,CAAxB,GAA4BA,GAAnC;MACD,CAHM,EAGJ,CAHI,CAAP;IAID;EACF;AAlE2C,CAA/B,C"}
+{"version":3,"file":"authors.js","names":["nameTreshold","name","extract","record","get","map","subfields","filter","code","value","includes","toObj","reduce","acc","v","testStringOrNumber","String","replace","toLowerCase","id","compare","a","b","maxAuthors","length","matchingIds","findMatchingIds","matchingNames","findMatchingNames","idPoints","namePoints","totalPoints","findMatches","distance","leven","maxLength","getMaxLength","percentage","key","cb","valuesA","o","valuesB","allValues","concat","occurrance","otherValue"],"sources":["../../../../src/match-detection/features/bib/authors.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 {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\n\n// We should extract also organisational authors (110/710)\n\nexport default ({nameTreshold = 10} = {}) => ({\n  name: 'Authors',\n  extract: record => record.get(/^(?<def>100|700)$/u)\n    .map(({subfields}) => {\n      return subfields\n        .filter(({code, value}) => code && value)\n        .filter(({code}) => ['a', '0'].includes(code))\n        .map(toObj)\n        .reduce((acc, v) => ({...acc, ...v}), {});\n\n      function toObj({code, value}) {\n        if (code === 'a') {\n          return {name: testStringOrNumber(value) ? String(value).replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase() : ''};\n        }\n\n        return {id: value};\n      }\n    }),\n  compare: (a, b) => {\n    const maxAuthors = a.length > b.length ? a.length : b.length;\n    const matchingIds = findMatchingIds();\n\n    if (maxAuthors >= 3 && matchingIds >= 3) {\n      return 0.3;\n    }\n\n    const matchingNames = findMatchingNames();\n    const idPoints = matchingIds / maxAuthors * 0.3;\n    const namePoints = matchingNames / maxAuthors * 0.2;\n    const totalPoints = idPoints + namePoints;\n\n    return totalPoints <= 0.2 ? totalPoints : 0.2;\n\n    function findMatchingIds() {\n      return findMatches('id', (a, b) => a === b);\n    }\n\n    function findMatchingNames() {\n      return findMatches('name', (a, b) => {\n        const distance = leven(a, b);\n\n        if (distance === 0) {\n          return true;\n        }\n\n        const maxLength = getMaxLength();\n        const percentage = distance / maxLength * 100;\n\n        return percentage <= nameTreshold;\n\n        function getMaxLength() {\n          return a.length > b.length ? a.length : b.length;\n        }\n      });\n    }\n\n    function findMatches(key, cb) {\n      const valuesA = a.filter(o => o[key]).map(o => o[key]);\n      const valuesB = b.filter(o => o[key]).map(o => o[key]);\n      const allValues = valuesA.concat(valuesB);\n\n      return allValues.reduce((acc, value) => {\n        const occurrance = allValues.filter(otherValue => cb(value, otherValue)).length;\n        return occurrance >= 2 ? acc + 1 : acc;\n      }, 0);\n    }\n  }\n});\n"],"mappings":";;;;;;;AA4BA;;AACA;;AA7BA;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;eAEe,CAAC;EAACA,YAAY,GAAG;AAAhB,IAAsB,EAAvB,MAA+B;EAC5CC,IAAI,EAAE,SADsC;EAE5CC,OAAO,EAAEC,MAAM,IAAIA,MAAM,CAACC,GAAP,CAAW,oBAAX,EAChBC,GADgB,CACZ,CAAC;IAACC;EAAD,CAAD,KAAiB;IACpB,OAAOA,SAAS,CACbC,MADI,CACG,CAAC;MAACC,IAAD;MAAOC;IAAP,CAAD,KAAmBD,IAAI,IAAIC,KAD9B,EAEJF,MAFI,CAEG,CAAC;MAACC;IAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWE,QAAX,CAAoBF,IAApB,CAFf,EAGJH,GAHI,CAGAM,KAHA,EAIJC,MAJI,CAIG,CAACC,GAAD,EAAMC,CAAN,MAAa,EAAC,GAAGD,GAAJ;MAAS,GAAGC;IAAZ,CAAb,CAJH,EAIiC,EAJjC,CAAP;;IAMA,SAASH,KAAT,CAAe;MAACH,IAAD;MAAOC;IAAP,CAAf,EAA8B;MAC5B,IAAID,IAAI,KAAK,GAAb,EAAkB;QAChB,OAAO;UAACP,IAAI,EAAE,IAAAc,iCAAA,EAAmBN,KAAnB,IAA4BO,MAAM,CAACP,KAAD,CAAN,CAAcQ,OAAd,CAAsB,2BAAtB,EAAmD,EAAnD,EAAuDC,WAAvD,EAA5B,GAAmG;QAA1G,CAAP;MACD;;MAED,OAAO;QAACC,EAAE,EAAEV;MAAL,CAAP;IACD;EACF,CAfgB,CAFyB;EAkB5CW,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB,MAAMC,UAAU,GAAGF,CAAC,CAACG,MAAF,GAAWF,CAAC,CAACE,MAAb,GAAsBH,CAAC,CAACG,MAAxB,GAAiCF,CAAC,CAACE,MAAtD;IACA,MAAMC,WAAW,GAAGC,eAAe,EAAnC;;IAEA,IAAIH,UAAU,IAAI,CAAd,IAAmBE,WAAW,IAAI,CAAtC,EAAyC;MACvC,OAAO,GAAP;IACD;;IAED,MAAME,aAAa,GAAGC,iBAAiB,EAAvC;IACA,MAAMC,QAAQ,GAAGJ,WAAW,GAAGF,UAAd,GAA2B,GAA5C;IACA,MAAMO,UAAU,GAAGH,aAAa,GAAGJ,UAAhB,GAA6B,GAAhD;IACA,MAAMQ,WAAW,GAAGF,QAAQ,GAAGC,UAA/B;IAEA,OAAOC,WAAW,IAAI,GAAf,GAAqBA,WAArB,GAAmC,GAA1C;;IAEA,SAASL,eAAT,GAA2B;MACzB,OAAOM,WAAW,CAAC,IAAD,EAAO,CAACX,CAAD,EAAIC,CAAJ,KAAUD,CAAC,KAAKC,CAAvB,CAAlB;IACD;;IAED,SAASM,iBAAT,GAA6B;MAC3B,OAAOI,WAAW,CAAC,MAAD,EAAS,CAACX,CAAD,EAAIC,CAAJ,KAAU;QACnC,MAAMW,QAAQ,GAAG,IAAAC,4BAAA,EAAMb,CAAN,EAASC,CAAT,CAAjB;;QAEA,IAAIW,QAAQ,KAAK,CAAjB,EAAoB;UAClB,OAAO,IAAP;QACD;;QAED,MAAME,SAAS,GAAGC,YAAY,EAA9B;QACA,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAX,GAAuB,GAA1C;QAEA,OAAOE,UAAU,IAAIrC,YAArB;;QAEA,SAASoC,YAAT,GAAwB;UACtB,OAAOf,CAAC,CAACG,MAAF,GAAWF,CAAC,CAACE,MAAb,GAAsBH,CAAC,CAACG,MAAxB,GAAiCF,CAAC,CAACE,MAA1C;QACD;MACF,CAfiB,CAAlB;IAgBD;;IAED,SAASQ,WAAT,CAAqBM,GAArB,EAA0BC,EAA1B,EAA8B;MAC5B,MAAMC,OAAO,GAAGnB,CAAC,CAACd,MAAF,CAASkC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAf,EAAsBjC,GAAtB,CAA0BoC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAhC,CAAhB;MACA,MAAMI,OAAO,GAAGpB,CAAC,CAACf,MAAF,CAASkC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAf,EAAsBjC,GAAtB,CAA0BoC,CAAC,IAAIA,CAAC,CAACH,GAAD,CAAhC,CAAhB;MACA,MAAMK,SAAS,GAAGH,OAAO,CAACI,MAAR,CAAeF,OAAf,CAAlB;MAEA,OAAOC,SAAS,CAAC/B,MAAV,CAAiB,CAACC,GAAD,EAAMJ,KAAN,KAAgB;QACtC,MAAMoC,UAAU,GAAGF,SAAS,CAACpC,MAAV,CAAiBuC,UAAU,IAAIP,EAAE,CAAC9B,KAAD,EAAQqC,UAAR,CAAjC,EAAsDtB,MAAzE;QACA,OAAOqB,UAAU,IAAI,CAAd,GAAkBhC,GAAG,GAAG,CAAxB,GAA4BA,GAAnC;MACD,CAHM,EAGJ,CAHI,CAAP;IAID;EACF;AAlE2C,CAA/B,C"}

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

@@ -17,7 +17,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -99,7 +99,7 @@ var _default = () => ({
       return 0;
     }
 
-    debugData(`There area languages to compare`);
+    debugData(`There are languages to compare`);
 
     if (a[0] === b[0]) {
       return 0.1;

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

@@ -1 +1 @@
-{"version":3,"file":"language.js","names":["debug","createDebugLogger","debugData","extend","name","extract","record","value008","get008Value","values041","get041Values","JSON","stringify","length","correspondingValue","find","v","value","get","undefined","code","slice","filter","ind2","map","subfields","flat","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/language.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 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:language');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'Language',\n  extract: record => {\n    const value008 = get008Value();\n    const values041 = get041Values();\n    debugData(`008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);\n\n    if (value008 && values041.length > 0) {\n      debugData(`There's both 008 and 041, searching for value in both`);\n      const correspondingValue = values041.find(v => v === value008);\n      debugData(`Corresponding value: ${correspondingValue}`);\n      return correspondingValue ? [correspondingValue] : [];\n    }\n\n    if (!value008 && values041.length < 1) {\n      debugData(`No actual values found`);\n      return [];\n    }\n\n    return value008 ? [value008] : [values041[0]];\n\n    function get008Value() {\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      debugData(`008 value: ${value}`);\n\n      if (!value) {\n        return undefined;\n      }\n\n      const code = value.slice(35, 38);\n      debugData(`008 code: ${code}`);\n      return code === '|||' || code === '   ' || code === '^^^' ? undefined : code;\n    }\n\n    // Main language for the resource: in the first f041 $a or f041 $d\n    // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes\n\n    function get041Values() {\n      return record.get(/^041$/u)\n        .filter(({ind2}) => ind2 === ' ')\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a' || code === 'd')\n        .map(({value}) => value);\n    }\n  },\n  compare: (a, b) => {\n    debugData(`Comparing ${JSON.stringify(a[0])} and ${JSON.stringify(b[0])}`);\n\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No language to compare`);\n      return 0;\n    }\n\n    debugData(`There area languages to compare`);\n\n    if (a[0] === b[0]) {\n      return 0.1;\n    }\n\n    return a[0] === 'und' || b[0] === 'und' ? 0.0 : -1.0;\n  }\n});\n"],"mappings":";;;;;;;AA6BA;;;;AA7BA;;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA,MAAMA,KAAK,GAAG,IAAAC,cAAA,EAAkB,qEAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;;eAEe,OAAO;EACpBC,IAAI,EAAE,UADc;EAEpBC,OAAO,EAAEC,MAAM,IAAI;IACjB,MAAMC,QAAQ,GAAGC,WAAW,EAA5B;IACA,MAAMC,SAAS,GAAGC,YAAY,EAA9B;IACAR,SAAS,CAAE,QAAOS,IAAI,CAACC,SAAL,CAAeL,QAAf,CAAyB,UAASI,IAAI,CAACC,SAAL,CAAeH,SAAf,CAA0B,EAArE,CAAT;;IAEA,IAAIF,QAAQ,IAAIE,SAAS,CAACI,MAAV,GAAmB,CAAnC,EAAsC;MACpCX,SAAS,CAAE,uDAAF,CAAT;MACA,MAAMY,kBAAkB,GAAGL,SAAS,CAACM,IAAV,CAAeC,CAAC,IAAIA,CAAC,KAAKT,QAA1B,CAA3B;MACAL,SAAS,CAAE,wBAAuBY,kBAAmB,EAA5C,CAAT;MACA,OAAOA,kBAAkB,GAAG,CAACA,kBAAD,CAAH,GAA0B,EAAnD;IACD;;IAED,IAAI,CAACP,QAAD,IAAaE,SAAS,CAACI,MAAV,GAAmB,CAApC,EAAuC;MACrCX,SAAS,CAAE,wBAAF,CAAT;MACA,OAAO,EAAP;IACD;;IAED,OAAOK,QAAQ,GAAG,CAACA,QAAD,CAAH,GAAgB,CAACE,SAAS,CAAC,CAAD,CAAV,CAA/B;;IAEA,SAASD,WAAT,GAAuB;MAAA;;MACrB,MAAMS,KAAK,GAAG,gBAAAX,MAAM,CAACY,GAAP,CAAW,QAAX,6EAAuB,CAAvB,+DAA2BD,KAA3B,KAAoCE,SAAlD;MACAjB,SAAS,CAAE,cAAae,KAAM,EAArB,CAAT;;MAEA,IAAI,CAACA,KAAL,EAAY;QACV,OAAOE,SAAP;MACD;;MAED,MAAMC,IAAI,GAAGH,KAAK,CAACI,KAAN,CAAY,EAAZ,EAAgB,EAAhB,CAAb;MACAnB,SAAS,CAAE,aAAYkB,IAAK,EAAnB,CAAT;MACA,OAAOA,IAAI,KAAK,KAAT,IAAkBA,IAAI,KAAK,KAA3B,IAAoCA,IAAI,KAAK,KAA7C,GAAqDD,SAArD,GAAiEC,IAAxE;IACD,CA9BgB,CAgCjB;IACA;;;IAEA,SAASV,YAAT,GAAwB;MACtB,OAAOJ,MAAM,CAACY,GAAP,CAAW,QAAX,EACJI,MADI,CACG,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GADxB,EAEJC,GAFI,CAEA,CAAC;QAACC;MAAD,CAAD,KAAiBA,SAFjB,EAGJC,IAHI,GAIJJ,MAJI,CAIG,CAAC;QAACF;MAAD,CAAD,KAAYA,IAAI,KAAK,GAAT,IAAgBA,IAAI,KAAK,GAJxC,EAKJI,GALI,CAKA,CAAC;QAACP;MAAD,CAAD,KAAaA,KALb,CAAP;IAMD;EACF,CA7CmB;EA8CpBU,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB3B,SAAS,CAAE,aAAYS,IAAI,CAACC,SAAL,CAAegB,CAAC,CAAC,CAAD,CAAhB,CAAqB,QAAOjB,IAAI,CAACC,SAAL,CAAeiB,CAAC,CAAC,CAAD,CAAhB,CAAqB,EAA/D,CAAT;;IAEA,IAAID,CAAC,CAACf,MAAF,KAAa,CAAb,IAAkBgB,CAAC,CAAChB,MAAF,KAAa,CAAnC,EAAsC;MACpCX,SAAS,CAAE,wBAAF,CAAT;MACA,OAAO,CAAP;IACD;;IAEDA,SAAS,CAAE,iCAAF,CAAT;;IAEA,IAAI0B,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAd,EAAmB;MACjB,OAAO,GAAP;IACD;;IAED,OAAOD,CAAC,CAAC,CAAD,CAAD,KAAS,KAAT,IAAkBC,CAAC,CAAC,CAAD,CAAD,KAAS,KAA3B,GAAmC,GAAnC,GAAyC,CAAC,GAAjD;EACD;AA7DmB,CAAP,C"}
+{"version":3,"file":"language.js","names":["debug","createDebugLogger","debugData","extend","name","extract","record","value008","get008Value","values041","get041Values","JSON","stringify","length","correspondingValue","find","v","value","get","undefined","code","slice","filter","ind2","map","subfields","flat","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/language.js"],"sourcesContent":["/* eslint-disable max-statements */\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:language');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'Language',\n  extract: record => {\n    const value008 = get008Value();\n    const values041 = get041Values();\n    debugData(`008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);\n\n    if (value008 && values041.length > 0) {\n      debugData(`There's both 008 and 041, searching for value in both`);\n      const correspondingValue = values041.find(v => v === value008);\n      debugData(`Corresponding value: ${correspondingValue}`);\n      return correspondingValue ? [correspondingValue] : [];\n    }\n\n    if (!value008 && values041.length < 1) {\n      debugData(`No actual values found`);\n      return [];\n    }\n\n    return value008 ? [value008] : [values041[0]];\n\n    function get008Value() {\n      const value = record.get(/^008$/u)?.[0]?.value || undefined;\n      debugData(`008 value: ${value}`);\n\n      if (!value) {\n        return undefined;\n      }\n\n      const code = value.slice(35, 38);\n      debugData(`008 code: ${code}`);\n      return code === '|||' || code === '   ' || code === '^^^' ? undefined : code;\n    }\n\n    // Main language for the resource: in the first f041 $a or f041 $d\n    // Uses only f041s that have 2nd ind ' ', which means that the codes used are MARC 21 language codes\n\n    function get041Values() {\n      return record.get(/^041$/u)\n        .filter(({ind2}) => ind2 === ' ')\n        .map(({subfields}) => subfields)\n        .flat()\n        .filter(({code}) => code === 'a' || code === 'd')\n        .map(({value}) => value);\n    }\n  },\n  compare: (a, b) => {\n    debugData(`Comparing ${JSON.stringify(a[0])} and ${JSON.stringify(b[0])}`);\n\n    if (a.length === 0 || b.length === 0) {\n      debugData(`No language to compare`);\n      return 0;\n    }\n\n    debugData(`There are languages to compare`);\n\n    if (a[0] === b[0]) {\n      return 0.1;\n    }\n\n    return a[0] === 'und' || b[0] === 'und' ? 0.0 : -1.0;\n  }\n});\n"],"mappings":";;;;;;;AA6BA;;;;AA7BA;;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA,MAAMA,KAAK,GAAG,IAAAC,cAAA,EAAkB,qEAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;;eAEe,OAAO;EACpBC,IAAI,EAAE,UADc;EAEpBC,OAAO,EAAEC,MAAM,IAAI;IACjB,MAAMC,QAAQ,GAAGC,WAAW,EAA5B;IACA,MAAMC,SAAS,GAAGC,YAAY,EAA9B;IACAR,SAAS,CAAE,QAAOS,IAAI,CAACC,SAAL,CAAeL,QAAf,CAAyB,UAASI,IAAI,CAACC,SAAL,CAAeH,SAAf,CAA0B,EAArE,CAAT;;IAEA,IAAIF,QAAQ,IAAIE,SAAS,CAACI,MAAV,GAAmB,CAAnC,EAAsC;MACpCX,SAAS,CAAE,uDAAF,CAAT;MACA,MAAMY,kBAAkB,GAAGL,SAAS,CAACM,IAAV,CAAeC,CAAC,IAAIA,CAAC,KAAKT,QAA1B,CAA3B;MACAL,SAAS,CAAE,wBAAuBY,kBAAmB,EAA5C,CAAT;MACA,OAAOA,kBAAkB,GAAG,CAACA,kBAAD,CAAH,GAA0B,EAAnD;IACD;;IAED,IAAI,CAACP,QAAD,IAAaE,SAAS,CAACI,MAAV,GAAmB,CAApC,EAAuC;MACrCX,SAAS,CAAE,wBAAF,CAAT;MACA,OAAO,EAAP;IACD;;IAED,OAAOK,QAAQ,GAAG,CAACA,QAAD,CAAH,GAAgB,CAACE,SAAS,CAAC,CAAD,CAAV,CAA/B;;IAEA,SAASD,WAAT,GAAuB;MAAA;;MACrB,MAAMS,KAAK,GAAG,gBAAAX,MAAM,CAACY,GAAP,CAAW,QAAX,6EAAuB,CAAvB,+DAA2BD,KAA3B,KAAoCE,SAAlD;MACAjB,SAAS,CAAE,cAAae,KAAM,EAArB,CAAT;;MAEA,IAAI,CAACA,KAAL,EAAY;QACV,OAAOE,SAAP;MACD;;MAED,MAAMC,IAAI,GAAGH,KAAK,CAACI,KAAN,CAAY,EAAZ,EAAgB,EAAhB,CAAb;MACAnB,SAAS,CAAE,aAAYkB,IAAK,EAAnB,CAAT;MACA,OAAOA,IAAI,KAAK,KAAT,IAAkBA,IAAI,KAAK,KAA3B,IAAoCA,IAAI,KAAK,KAA7C,GAAqDD,SAArD,GAAiEC,IAAxE;IACD,CA9BgB,CAgCjB;IACA;;;IAEA,SAASV,YAAT,GAAwB;MACtB,OAAOJ,MAAM,CAACY,GAAP,CAAW,QAAX,EACJI,MADI,CACG,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GADxB,EAEJC,GAFI,CAEA,CAAC;QAACC;MAAD,CAAD,KAAiBA,SAFjB,EAGJC,IAHI,GAIJJ,MAJI,CAIG,CAAC;QAACF;MAAD,CAAD,KAAYA,IAAI,KAAK,GAAT,IAAgBA,IAAI,KAAK,GAJxC,EAKJI,GALI,CAKA,CAAC;QAACP;MAAD,CAAD,KAAaA,KALb,CAAP;IAMD;EACF,CA7CmB;EA8CpBU,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB3B,SAAS,CAAE,aAAYS,IAAI,CAACC,SAAL,CAAegB,CAAC,CAAC,CAAD,CAAhB,CAAqB,QAAOjB,IAAI,CAACC,SAAL,CAAeiB,CAAC,CAAC,CAAD,CAAhB,CAAqB,EAA/D,CAAT;;IAEA,IAAID,CAAC,CAACf,MAAF,KAAa,CAAb,IAAkBgB,CAAC,CAAChB,MAAF,KAAa,CAAnC,EAAsC;MACpCX,SAAS,CAAE,wBAAF,CAAT;MACA,OAAO,CAAP;IACD;;IAEDA,SAAS,CAAE,gCAAF,CAAT;;IAEA,IAAI0B,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAd,EAAmB;MACjB,OAAO,GAAP;IACD;;IAED,OAAOD,CAAC,CAAC,CAAD,CAAD,KAAS,KAAT,IAAkBC,CAAC,CAAC,CAAD,CAAD,KAAS,KAA3B,GAAmC,GAAnC,GAAyC,CAAC,GAAjD;EACD;AA7DmB,CAAP,C"}

package/dist/match-detection/features/bib/publication-time.js

@@ -5,13 +5,15 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.default = void 0;
 
+var _matchingUtils = require("../../../matching-utils");
+
 /**
 *
 * @licstart  The following is the entire license notice for the JavaScript code in this file.
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -32,13 +34,15 @@ exports.default = void 0;
 * for the JavaScript code in this file.
 *
 */
+// We should also get copyright time and copyright/publication times from 26x
+// We could also enable matching of records which have consequent publication years
 var _default = () => ({
   name: 'Publication time',
   extract: record => {
     var _record$get, _record$get$;
 
-    const value = ((_record$get = record.get(/^008$/u)) === null || _record$get === void 0 ? void 0 : (_record$get$ = _record$get[0]) === null || _record$get$ === void 0 ? void 0 : _record$get$.value) || '';
-    return value ? [value.slice(7, 11)] : [];
+    const value = ((_record$get = record.get(/^008$/u)) === null || _record$get === void 0 ? void 0 : (_record$get$ = _record$get[0]) === null || _record$get$ === void 0 ? void 0 : _record$get$.value) || undefined;
+    return (0, _matchingUtils.testStringOrNumber)(value) ? [String(value).slice(7, 11)] : [];
   },
   compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0
 });

package/dist/match-detection/features/bib/publication-time.js.map

@@ -1 +1 @@
-{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport default () => ({\n  name: 'Publication time',\n  extract: record => {\n    const value = record.get(/^008$/u)?.[0]?.value || '';\n    return value ? [value.slice(7, 11)] : [];\n  },\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;eAEe,OAAO;EACpBA,IAAI,EAAE,kBADc;EAEpBC,OAAO,EAAEC,MAAM,IAAI;IAAA;;IACjB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAP,CAAW,QAAX,6EAAuB,CAAvB,+DAA2BD,KAA3B,KAAoC,EAAlD;IACA,OAAOA,KAAK,GAAG,CAACA,KAAK,CAACE,KAAN,CAAY,CAAZ,EAAe,EAAf,CAAD,CAAH,GAA0B,EAAtC;EACD,CALmB;EAMpBC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAUD,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAV,GAAgB,GAAhB,GAAsB,CAAC;AANtB,CAAP,C"}
+{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.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 {testStringOrNumber} from '../../../matching-utils';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We could also enable matching of records which have consequent publication years\n\nexport default () => ({\n  name: 'Publication time',\n  extract: record => {\n    const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n  },\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;;AA4BA;;AA5BA;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;AACA;eAEe,OAAO;EACpBA,IAAI,EAAE,kBADc;EAEpBC,OAAO,EAAEC,MAAM,IAAI;IAAA;;IACjB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAP,CAAW,QAAX,6EAAuB,CAAvB,+DAA2BD,KAA3B,KAAoCE,SAAlD;IACA,OAAO,IAAAC,iCAAA,EAAmBH,KAAnB,IAA4B,CAACI,MAAM,CAACJ,KAAD,CAAN,CAAcK,KAAd,CAAoB,CAApB,EAAuB,EAAvB,CAAD,CAA5B,GAA2D,EAAlE;EACD,CALmB;EAMpBC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAUD,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAV,GAAgB,GAAhB,GAAsB,CAAC;AANtB,CAAP,C"}

package/dist/match-detection/features/bib/record-type.js

@@ -32,6 +32,7 @@ exports.default = void 0;
 * for the JavaScript code in this file.
 *
 */
+// we could handle the case of books/notes
 var _default = () => ({
   name: 'Record type',
   extract: r => r.leader[6] ? [r.leader[6]] : [],

package/dist/match-detection/features/bib/record-type.js.map

@@ -1 +1 @@
-{"version":3,"file":"record-type.js","names":["name","extract","r","leader","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/record-type.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nexport default () => ({\n  name: 'Record type',\n  extract: r => r.leader[6] ? [r.leader[6]] : [],\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5\n});\n"],"mappings":";;;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;eAEe,OAAO;EACpBA,IAAI,EAAE,aADc;EAEpBC,OAAO,EAAEC,CAAC,IAAIA,CAAC,CAACC,MAAF,CAAS,CAAT,IAAc,CAACD,CAAC,CAACC,MAAF,CAAS,CAAT,CAAD,CAAd,GAA8B,EAFxB;EAGpBC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAUD,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAV,GAAgB,GAAhB,GAAsB,CAAC;AAHtB,CAAP,C"}
+{"version":3,"file":"record-type.js","names":["name","extract","r","leader","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/record-type.js"],"sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\n// we could handle the case of books/notes\n\nexport default () => ({\n  name: 'Record type',\n  extract: r => r.leader[6] ? [r.leader[6]] : [],\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5\n});\n"],"mappings":";;;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;eAEe,OAAO;EACpBA,IAAI,EAAE,aADc;EAEpBC,OAAO,EAAEC,CAAC,IAAIA,CAAC,CAACC,MAAF,CAAS,CAAT,IAAc,CAACD,CAAC,CAACC,MAAF,CAAS,CAAT,CAAD,CAAd,GAA8B,EAFxB;EAGpBC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAUD,CAAC,CAAC,CAAD,CAAD,KAASC,CAAC,CAAC,CAAD,CAAV,GAAgB,GAAhB,GAAsB,CAAC;AAHtB,CAAP,C"}

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

@@ -7,6 +7,8 @@ exports.default = void 0;
 
 var _debug = _interopRequireDefault(require("debug"));
 
+var _matchingUtils = require("../../../matching-utils");
+
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
 
 /**
@@ -15,7 +17,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -54,14 +56,13 @@ var _default = ({
     if (field) {
       return field.subfields.filter(({
         code
-      }) => subfieldCodes.includes(code)) //  .map(({code, value}) => ({code, value: value.replace(/-/ug, '')}));
-      .map(({
+      }) => subfieldCodes.includes(code)).map(({
         code,
         value
       }) => ({
         code,
-        value: value ? value.replace(/-/ug, '') : ''
-      }));
+        value: (0, _matchingUtils.testStringOrNumber)(value) ? String(value).replace(/-/ug, '') : ''
+      })).filter(value => value);
     }
 
     return [];

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

@@ -1 +1 @@
-{"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","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 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features: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: value.replace(/-/ug, '')}));\n        .map(({code, value}) => ({code, value: value ? value.replace(/-/ug, '') : ''}));\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;;;;AA5BA;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,cAAA,EAAkB,iFAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;;eAEe,CAAC;EAACC,OAAD;EAAUC;AAAV,CAAD,KAA8B;EAC3C,OAAO;IAACC,OAAD;IAAUC;EAAV,CAAP;;EAEA,SAASD,OAAT,CAAiBE,MAAjB,EAAyB;IACvB,MAAM,CAACC,KAAD,IAAUD,MAAM,CAACE,GAAP,CAAWN,OAAX,CAAhB;;IAEA,IAAIK,KAAJ,EAAW;MACT,OAAOA,KAAK,CAACE,SAAN,CACJC,MADI,CACG,CAAC;QAACC;MAAD,CAAD,KAAYR,aAAa,CAACS,QAAd,CAAuBD,IAAvB,CADf,EAEP;MAFO,CAGJE,GAHI,CAGA,CAAC;QAACF,IAAD;QAAOG;MAAP,CAAD,MAAoB;QAACH,IAAD;QAAOG,KAAK,EAAEA,KAAK,GAAGA,KAAK,CAACC,OAAN,CAAc,KAAd,EAAqB,EAArB,CAAH,GAA8B;MAAjD,CAApB,CAHA,CAAP;IAID;;IAED,OAAO,EAAP;EACD;;EAED,SAASV,OAAT,CAAiBW,CAAjB,EAAoBC,CAApB,EAAuB;IACrB,IAAID,CAAC,CAACE,MAAF,KAAa,CAAb,IAAkBD,CAAC,CAACC,MAAF,KAAa,CAAnC,EAAsC;MACpClB,SAAS,CAAE,mCAAF,CAAT;MACA,OAAO,CAAP;IACD;;IAED,IAAImB,wBAAwB,EAA5B,EAAgC;MAC9B,MAAM;QAACC,SAAD;QAAYC;MAAZ,IAA8BC,aAAa,CAAC,IAAD,CAAjD;MACA,OAAOD,cAAc,GAAGD,SAAjB,GAA6B,IAApC;IACD;;IAED,MAAM;MAACA,SAAD;MAAYC;IAAZ,IAA8BC,aAAa,EAAjD;IACA,OAAOD,cAAc,GAAGD,SAAjB,GAA6B,GAApC;;IAEA,SAASD,wBAAT,GAAoC;MAClC,MAAMI,OAAO,GAAGP,CAAC,CAACN,MAAF,CAAS,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA9B,CAAhB;MACA,MAAMa,OAAO,GAAGR,CAAC,CAACN,MAAF,CAAS,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA9B,CAAhB;MACA,OAAOY,OAAO,CAACL,MAAR,GAAiB,CAAjB,IAAsBM,OAAO,CAACN,MAAR,GAAiB,CAA9C;IACD;;IAED,SAASI,aAAT,CAAuBG,SAAS,GAAG,KAAnC,EAA0C;MACxC,MAAMF,OAAO,GAAGG,cAAc,CAACV,CAAD,CAA9B;MACA,MAAMQ,OAAO,GAAGE,cAAc,CAACT,CAAD,CAA9B;MAEA,OAAO;QACLG,SAAS,EAAEG,OAAO,CAACL,MAAR,GAAiBM,OAAO,CAACN,MAAzB,GAAkCK,OAAO,CAACL,MAA1C,GAAmDM,OAAO,CAACN,MADjE;QAELG,cAAc,EAAEE,OAAO,CAACb,MAAR,CAAeiB,MAAM,IAAIH,OAAO,CAACI,IAAR,CAAaC,MAAM,IAAIF,MAAM,KAAKE,MAAlC,CAAzB,EAAoEX;MAF/E,CAAP;;MAKA,SAASQ,cAAT,CAAwBI,MAAxB,EAAgC;QAC9B,IAAIL,SAAJ,EAAe;UACb,OAAOK,MAAM,CACVpB,MADI,CACG,CAAC;YAACC;UAAD,CAAD,KAAYA,IAAI,KAAK,GADxB,EAEJE,GAFI,CAEA,CAAC;YAACC;UAAD,CAAD,KAAaA,KAFb,CAAP;QAGD;;QAED,OAAOgB,MAAM,CAACjB,GAAP,CAAW,CAAC;UAACC;QAAD,CAAD,KAAaA,KAAxB,CAAP;MACD;IACF;EACF;AACF,C"}
+{"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;;;;AA7BA;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,cAAA,EAAkB,iFAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;;eAEe,CAAC;EAACC,OAAD;EAAUC;AAAV,CAAD,KAA8B;EAC3C,OAAO;IAACC,OAAD;IAAUC;EAAV,CAAP;;EAEA,SAASD,OAAT,CAAiBE,MAAjB,EAAyB;IACvB,MAAM,CAACC,KAAD,IAAUD,MAAM,CAACE,GAAP,CAAWN,OAAX,CAAhB;;IAEA,IAAIK,KAAJ,EAAW;MACT,OAAOA,KAAK,CAACE,SAAN,CACJC,MADI,CACG,CAAC;QAACC;MAAD,CAAD,KAAYR,aAAa,CAACS,QAAd,CAAuBD,IAAvB,CADf,EAEJE,GAFI,CAEA,CAAC;QAACF,IAAD;QAAOG;MAAP,CAAD,MAAoB;QAACH,IAAD;QAAOG,KAAK,EAAE,IAAAC,iCAAA,EAAmBD,KAAnB,IAA4BE,MAAM,CAACF,KAAD,CAAN,CAAcG,OAAd,CAAsB,KAAtB,EAA6B,EAA7B,CAA5B,GAA+D;MAA7E,CAApB,CAFA,EAGJP,MAHI,CAGGI,KAAK,IAAIA,KAHZ,CAAP;IAID;;IAED,OAAO,EAAP;EACD;;EAED,SAAST,OAAT,CAAiBa,CAAjB,EAAoBC,CAApB,EAAuB;IACrB,IAAID,CAAC,CAACE,MAAF,KAAa,CAAb,IAAkBD,CAAC,CAACC,MAAF,KAAa,CAAnC,EAAsC;MACpCpB,SAAS,CAAE,mCAAF,CAAT;MACA,OAAO,CAAP;IACD;;IAED,IAAIqB,wBAAwB,EAA5B,EAAgC;MAC9B,MAAM;QAACC,SAAD;QAAYC;MAAZ,IAA8BC,aAAa,CAAC,IAAD,CAAjD;MACA,OAAOD,cAAc,GAAGD,SAAjB,GAA6B,IAApC;IACD;;IAED,MAAM;MAACA,SAAD;MAAYC;IAAZ,IAA8BC,aAAa,EAAjD;IACA,OAAOD,cAAc,GAAGD,SAAjB,GAA6B,GAApC;;IAEA,SAASD,wBAAT,GAAoC;MAClC,MAAMI,OAAO,GAAGP,CAAC,CAACR,MAAF,CAAS,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA9B,CAAhB;MACA,MAAMe,OAAO,GAAGR,CAAC,CAACR,MAAF,CAAS,CAAC;QAACC;MAAD,CAAD,KAAYA,IAAI,KAAK,GAA9B,CAAhB;MACA,OAAOc,OAAO,CAACL,MAAR,GAAiB,CAAjB,IAAsBM,OAAO,CAACN,MAAR,GAAiB,CAA9C;IACD;;IAED,SAASI,aAAT,CAAuBG,SAAS,GAAG,KAAnC,EAA0C;MACxC,MAAMF,OAAO,GAAGG,cAAc,CAACV,CAAD,CAA9B;MACA,MAAMQ,OAAO,GAAGE,cAAc,CAACT,CAAD,CAA9B;MAEA,OAAO;QACLG,SAAS,EAAEG,OAAO,CAACL,MAAR,GAAiBM,OAAO,CAACN,MAAzB,GAAkCK,OAAO,CAACL,MAA1C,GAAmDM,OAAO,CAACN,MADjE;QAELG,cAAc,EAAEE,OAAO,CAACf,MAAR,CAAemB,MAAM,IAAIH,OAAO,CAACI,IAAR,CAAaC,MAAM,IAAIF,MAAM,KAAKE,MAAlC,CAAzB,EAAoEX;MAF/E,CAAP;;MAKA,SAASQ,cAAT,CAAwBI,MAAxB,EAAgC;QAC9B,IAAIL,SAAJ,EAAe;UACb,OAAOK,MAAM,CACVtB,MADI,CACG,CAAC;YAACC;UAAD,CAAD,KAAYA,IAAI,KAAK,GADxB,EAEJE,GAFI,CAEA,CAAC;YAACC;UAAD,CAAD,KAAaA,KAFb,CAAP;QAGD;;QAED,OAAOkB,MAAM,CAACnB,GAAP,CAAW,CAAC;UAACC;QAAD,CAAD,KAAaA,KAAxB,CAAP;MACD;IACF;EACF;AACF,C"}

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

@@ -9,6 +9,8 @@ var _debug = _interopRequireDefault(require("debug"));
 
 var _natural = require("natural");
 
+var _matchingUtils = require("../../../matching-utils");
+
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
 
 /**
@@ -17,7 +19,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -45,8 +47,8 @@ var _default = ({
   extract: record => {
     const title = getTitle();
 
-    if (title) {
-      return [title.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
+    if ((0, _matchingUtils.testStringOrNumber)(title)) {
+      return [String(title).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
     }
 
     return [];
@@ -59,8 +61,10 @@ var _default = ({
           code
         }) => ['a', 'b'].includes(code)).map(({
           value
-        }) => value).join('');
+        }) => (0, _matchingUtils.testStringOrNumber)(value) ? String(value) : '').join('');
       }
+
+      return false;
     }
   },
   compare: (a, b) => {

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

@@ -1 +1 @@
-{"version":3,"file":"title.js","names":["treshold","name","extract","record","title","getTitle","replace","toLowerCase","field","get","subfields","filter","code","includes","map","value","join","compare","a","b","debug","createDebugLogger","distance","leven","maxLength","getMaxLength","percentage","length"],"sources":["../../../../src/match-detection/features/bib/title.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 createDebugLogger from 'debug';\nimport {LevenshteinDistance as leven} from 'natural';\n\nexport default ({treshold = 10} = {}) => ({\n  name: 'Title',\n  extract: record => {\n    const title = getTitle();\n\n    if (title) {\n      return [title.replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase()];\n    }\n\n    return [];\n\n    function getTitle() {\n      const [field] = record.get(/^245$/u);\n\n      if (field) {\n        return field.subfields\n          .filter(({code}) => ['a', 'b'].includes(code))\n          .map(({value}) => value)\n          .join('');\n      }\n    }\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n    const distance = leven(a[0], b[0]);\n\n    if (distance === 0) {\n      return 0.5;\n    }\n\n    const maxLength = getMaxLength();\n    const percentage = distance / maxLength * 100;\n\n    debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);\n\n    if (percentage <= treshold) {\n      return 0.3;\n    }\n\n    return -0.5;\n\n    function getMaxLength() {\n      return a[0].length > b[0].length ? a[0].length : b[0].length;\n    }\n\n  }\n});\n"],"mappings":";;;;;;;AA4BA;;AACA;;;;AA7BA;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;eAKe,CAAC;EAACA,QAAQ,GAAG;AAAZ,IAAkB,EAAnB,MAA2B;EACxCC,IAAI,EAAE,OADkC;EAExCC,OAAO,EAAEC,MAAM,IAAI;IACjB,MAAMC,KAAK,GAAGC,QAAQ,EAAtB;;IAEA,IAAID,KAAJ,EAAW;MACT,OAAO,CAACA,KAAK,CAACE,OAAN,CAAc,2BAAd,EAA2C,EAA3C,EAA+CC,WAA/C,EAAD,CAAP;IACD;;IAED,OAAO,EAAP;;IAEA,SAASF,QAAT,GAAoB;MAClB,MAAM,CAACG,KAAD,IAAUL,MAAM,CAACM,GAAP,CAAW,QAAX,CAAhB;;MAEA,IAAID,KAAJ,EAAW;QACT,OAAOA,KAAK,CAACE,SAAN,CACJC,MADI,CACG,CAAC;UAACC;QAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWC,QAAX,CAAoBD,IAApB,CADf,EAEJE,GAFI,CAEA,CAAC;UAACC;QAAD,CAAD,KAAaA,KAFb,EAGJC,IAHI,CAGC,EAHD,CAAP;MAID;IACF;EACF,CArBuC;EAsBxCC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,sEAAlB,CAAd;IACA,MAAMC,QAAQ,GAAG,IAAAC,4BAAA,EAAML,CAAC,CAAC,CAAD,CAAP,EAAYC,CAAC,CAAC,CAAD,CAAb,CAAjB;;IAEA,IAAIG,QAAQ,KAAK,CAAjB,EAAoB;MAClB,OAAO,GAAP;IACD;;IAED,MAAME,SAAS,GAAGC,YAAY,EAA9B;IACA,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAX,GAAuB,GAA1C;IAEAJ,KAAK,CAAE,IAAGF,CAAE,SAAQC,CAAE,mBAAkBK,SAAU,gBAAeF,QAAS,kBAAiBI,UAAW,EAAjG,CAAL;;IAEA,IAAIA,UAAU,IAAI1B,QAAlB,EAA4B;MAC1B,OAAO,GAAP;IACD;;IAED,OAAO,CAAC,GAAR;;IAEA,SAASyB,YAAT,GAAwB;MACtB,OAAOP,CAAC,CAAC,CAAD,CAAD,CAAKS,MAAL,GAAcR,CAAC,CAAC,CAAD,CAAD,CAAKQ,MAAnB,GAA4BT,CAAC,CAAC,CAAD,CAAD,CAAKS,MAAjC,GAA0CR,CAAC,CAAC,CAAD,CAAD,CAAKQ,MAAtD;IACD;EAEF;AA7CuC,CAA3B,C"}
+{"version":3,"file":"title.js","names":["treshold","name","extract","record","title","getTitle","testStringOrNumber","String","replace","toLowerCase","field","get","subfields","filter","code","includes","map","value","join","compare","a","b","debug","createDebugLogger","distance","leven","maxLength","getMaxLength","percentage","length"],"sources":["../../../../src/match-detection/features/bib/title.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 {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nexport default ({treshold = 10} = {}) => ({\n  name: 'Title',\n  extract: record => {\n    const title = getTitle();\n\n    if (testStringOrNumber(title)) {\n      return [String(title).replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase()];\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          .join('');\n      }\n      return false;\n    }\n  },\n  compare: (a, b) => {\n    const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n    const distance = leven(a[0], b[0]);\n\n    if (distance === 0) {\n      return 0.5;\n    }\n\n    const maxLength = getMaxLength();\n    const percentage = distance / maxLength * 100;\n\n    debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);\n\n    if (percentage <= treshold) {\n      return 0.3;\n    }\n\n    return -0.5;\n\n    function getMaxLength() {\n      return a[0].length > b[0].length ? a[0].length : b[0].length;\n    }\n\n  }\n});\n"],"mappings":";;;;;;;AA4BA;;AACA;;AACA;;;;AA9BA;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;eAMe,CAAC;EAACA,QAAQ,GAAG;AAAZ,IAAkB,EAAnB,MAA2B;EACxCC,IAAI,EAAE,OADkC;EAExCC,OAAO,EAAEC,MAAM,IAAI;IACjB,MAAMC,KAAK,GAAGC,QAAQ,EAAtB;;IAEA,IAAI,IAAAC,iCAAA,EAAmBF,KAAnB,CAAJ,EAA+B;MAC7B,OAAO,CAACG,MAAM,CAACH,KAAD,CAAN,CAAcI,OAAd,CAAsB,2BAAtB,EAAmD,EAAnD,EAAuDC,WAAvD,EAAD,CAAP;IACD;;IAED,OAAO,EAAP;;IAEA,SAASJ,QAAT,GAAoB;MAClB,MAAM,CAACK,KAAD,IAAUP,MAAM,CAACQ,GAAP,CAAW,QAAX,CAAhB;;MAEA,IAAID,KAAJ,EAAW;QACT,OAAOA,KAAK,CAACE,SAAN,CACJC,MADI,CACG,CAAC;UAACC;QAAD,CAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWC,QAAX,CAAoBD,IAApB,CADf,EAEJE,GAFI,CAEA,CAAC;UAACC;QAAD,CAAD,KAAa,IAAAX,iCAAA,EAAmBW,KAAnB,IAA4BV,MAAM,CAACU,KAAD,CAAlC,GAA4C,EAFzD,EAGJC,IAHI,CAGC,EAHD,CAAP;MAID;;MACD,OAAO,KAAP;IACD;EACF,CAtBuC;EAuBxCC,OAAO,EAAE,CAACC,CAAD,EAAIC,CAAJ,KAAU;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAA,EAAkB,sEAAlB,CAAd;IACA,MAAMC,QAAQ,GAAG,IAAAC,4BAAA,EAAML,CAAC,CAAC,CAAD,CAAP,EAAYC,CAAC,CAAC,CAAD,CAAb,CAAjB;;IAEA,IAAIG,QAAQ,KAAK,CAAjB,EAAoB;MAClB,OAAO,GAAP;IACD;;IAED,MAAME,SAAS,GAAGC,YAAY,EAA9B;IACA,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAX,GAAuB,GAA1C;IAEAJ,KAAK,CAAE,IAAGF,CAAE,SAAQC,CAAE,mBAAkBK,SAAU,gBAAeF,QAAS,kBAAiBI,UAAW,EAAjG,CAAL;;IAEA,IAAIA,UAAU,IAAI5B,QAAlB,EAA4B;MAC1B,OAAO,GAAP;IACD;;IAED,OAAO,CAAC,GAAR;;IAEA,SAAS2B,YAAT,GAAwB;MACtB,OAAOP,CAAC,CAAC,CAAD,CAAD,CAAKS,MAAL,GAAcR,CAAC,CAAC,CAAD,CAAD,CAAKQ,MAAnB,GAA4BT,CAAC,CAAC,CAAD,CAAD,CAAKS,MAAjC,GAA0CR,CAAC,CAAC,CAAD,CAAD,CAAKQ,MAAtD;IACD;EAEF;AA9CuC,CAA3B,C"}

package/dist/matching-utils.js

@@ -5,6 +5,7 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.getMelindaIdsF035 = getMelindaIdsF035;
 exports.getSubfieldValues = getSubfieldValues;
+exports.testStringOrNumber = testStringOrNumber;
 exports.validateSidFieldSubfieldCounts = validateSidFieldSubfieldCounts;
 
 var _debug = _interopRequireDefault(require("debug"));
@@ -17,7 +18,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -62,9 +63,9 @@ function getMelindaIdsF035(record) {
     subfields
   }) {
     const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
-    return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => melindaIdRegExp.test(sub.value)).map(({
+    return subfields.filter(sub => ['a', 'z'].includes(sub.code)).filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({
       value
-    }) => value ? value.replace(melindaIdRegExp, '$<id>') : '');
+    }) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');
   }
 }
 
@@ -90,6 +91,14 @@ function getSubfieldValues(field, subfieldCode) {
     code
   }) => code === subfieldCode).map(({
     value
-  }) => value);
+  }) => testStringOrNumber(value) ? String(value) : '').filter(value => value);
+}
+
+function testStringOrNumber(value) {
+  if (typeof value === 'string' || typeof value === 'number') {
+    return true;
+  }
+
+  return false;
 }
 //# 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","test","value","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 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\n// eslint-disable-next-line max-statements\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 => melindaIdRegExp.test(sub.value))\n      .map(({value}) => value ? 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.filter(({code}) => code === subfieldCode).map(({value}) => value);\n}\n\n"],"mappings":";;;;;;;;;AA4BA;;;;AA5BA;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,cAAA,EAAkB,yCAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB,C,CAEA;;AACO,SAASC,iBAAT,CAA2BC,MAA3B,EAAmC;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAA,EAAkB,8CAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EAEA,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAP,CAAiB,KAAjB,CAAd;;EAEA,IAAID,KAAK,CAACE,MAAN,GAAe,CAAnB,EAAsB;IACpBR,KAAK,CAAE,iBAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,MAAMS,iBAAiB,GAAG,GAAGC,MAAH,CAAU,GAAGJ,KAAK,CAACK,GAAN,CAAUC,YAAV,CAAb,CAA1B;EACA,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQL,iBAAR,CAAJ,CAAvB;EAEAP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeV,KAAf,CAAsB,EAApD,CAAT;EACAJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeP,iBAAf,CAAkC,EAAzE,CAAT;EACAP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeH,cAAf,CAA+B,EAA1E,CAAT;EAEA,OAAOA,cAAP;;EAEA,SAASD,YAAT,CAAsB;IAACK;EAAD,CAAtB,EAAmC;IACjC,MAAMC,eAAe,GAAG,8CAAxB;IAEA,OAAOD,SAAS,CACbE,MADI,CACGC,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWC,QAAX,CAAoBD,GAAG,CAACE,IAAxB,CADV,EAEJH,MAFI,CAEGC,GAAG,IAAIF,eAAe,CAACK,IAAhB,CAAqBH,GAAG,CAACI,KAAzB,CAFV,EAGJb,GAHI,CAGA,CAAC;MAACa;IAAD,CAAD,KAAaA,KAAK,GAAGA,KAAK,CAACC,OAAN,CAAcP,eAAd,EAA+B,OAA/B,CAAH,GAA6C,EAH/D,CAAP;EAKD;AACF;;AAEM,SAASQ,8BAAT,CAAwCC,KAAxC,EAA+C;EACpD;EACAzB,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,EAA/C,CAAT;EACA,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;EACA,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;EACA3B,KAAK,CAAE,SAAQ4B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAE,EAAvF,CAAL;EAEA,OAAOF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAlC;AACD;;AAED,SAASD,cAAT,CAAwBF,KAAxB,EAA+BI,YAA/B,EAA6C;EAC3C;EACA,OAAOJ,KAAK,CAACV,SAAN,CAAgBE,MAAhB,CAAuB,CAAC;IAACG;EAAD,CAAD,KAAYA,IAAI,KAAKS,YAA5C,EAA0DvB,MAAjE;AACD;;AAEM,SAASwB,iBAAT,CAA2BL,KAA3B,EAAkCI,YAAlC,EAAgD;EACrD7B,SAAS,CAAE,oBAAmB6B,YAAa,SAAQhB,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,EAAhE,CAAT;EACA,OAAOA,KAAK,CAACV,SAAN,CAAgBE,MAAhB,CAAuB,CAAC;IAACG;EAAD,CAAD,KAAYA,IAAI,KAAKS,YAA5C,EAA0DpB,GAA1D,CAA8D,CAAC;IAACa;EAAD,CAAD,KAAaA,KAA3E,CAAP;AACD"}
+{"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\n// eslint-disable-next-line max-statements\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;;;;AA5BA;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,cAAA,EAAkB,yCAAlB,CAAd;AACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB,C,CAEA;;AACO,SAASC,iBAAT,CAA2BC,MAA3B,EAAmC;EAExC,MAAML,KAAK,GAAG,IAAAC,cAAA,EAAkB,8CAAlB,CAAd;EACA,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAN,CAAa,MAAb,CAAlB;EAEA,MAAMG,KAAK,GAAGD,MAAM,CAACE,SAAP,CAAiB,KAAjB,CAAd;;EAEA,IAAID,KAAK,CAACE,MAAN,GAAe,CAAnB,EAAsB;IACpBR,KAAK,CAAE,iBAAF,CAAL;IACA,OAAO,EAAP;EACD;;EAED,MAAMS,iBAAiB,GAAG,GAAGC,MAAH,CAAU,GAAGJ,KAAK,CAACK,GAAN,CAAUC,YAAV,CAAb,CAA1B;EACA,MAAMC,cAAc,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQL,iBAAR,CAAJ,CAAvB;EAEAP,SAAS,CAAE,WAAUI,KAAK,CAACE,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeV,KAAf,CAAsB,EAApD,CAAT;EACAJ,SAAS,CAAE,QAAOO,iBAAiB,CAACD,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeP,iBAAf,CAAkC,EAAzE,CAAT;EACAP,SAAS,CAAE,eAAcW,cAAc,CAACL,MAAO,MAAKO,IAAI,CAACC,SAAL,CAAeH,cAAf,CAA+B,EAA1E,CAAT;EAEA,OAAOA,cAAP;;EAEA,SAASD,YAAT,CAAsB;IAACK;EAAD,CAAtB,EAAmC;IACjC,MAAMC,eAAe,GAAG,8CAAxB;IAEA,OAAOD,SAAS,CACbE,MADI,CACGC,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWC,QAAX,CAAoBD,GAAG,CAACE,IAAxB,CADV,EAEJH,MAFI,CAEGC,GAAG,IAAIG,kBAAkB,CAACH,GAAG,CAACI,KAAL,CAAlB,IAAiCN,eAAe,CAACO,IAAhB,CAAqBC,MAAM,CAACN,GAAG,CAACI,KAAL,CAA3B,CAF3C,EAGJb,GAHI,CAGA,CAAC;MAACa;IAAD,CAAD,KAAaD,kBAAkB,CAACC,KAAD,CAAlB,GAA4BE,MAAM,CAACF,KAAD,CAAN,CAAcG,OAAd,CAAsBT,eAAtB,EAAuC,OAAvC,CAA5B,GAA8E,EAH3F,CAAP;EAKD;AACF;;AAEM,SAASU,8BAAT,CAAwCC,KAAxC,EAA+C;EACpD;EACA3B,SAAS,CAAE,wBAAuBa,IAAI,CAACC,SAAL,CAAea,KAAf,CAAsB,EAA/C,CAAT;EACA,MAAMC,MAAM,GAAGC,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;EACA,MAAMG,MAAM,GAAGD,cAAc,CAACF,KAAD,EAAQ,GAAR,CAA7B;EACA7B,KAAK,CAAE,SAAQ8B,MAAO,eAAcE,MAAO,qBAAoBF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAE,EAAvF,CAAL;EAEA,OAAOF,MAAM,KAAK,CAAX,IAAgBE,MAAM,KAAK,CAAlC;AACD;;AAED,SAASD,cAAT,CAAwBF,KAAxB,EAA+BI,YAA/B,EAA6C;EAC3C;EACA,OAAOJ,KAAK,CAACZ,SAAN,CAAgBE,MAAhB,CAAuB,CAAC;IAACG;EAAD,CAAD,KAAYA,IAAI,KAAKW,YAA5C,EAA0DzB,MAAjE;AACD;;AAEM,SAAS0B,iBAAT,CAA2BL,KAA3B,EAAkCI,YAAlC,EAAgD;EACrD/B,SAAS,CAAE,oBAAmB+B,YAAa,SAAQlB,IAAI,CAACC,SAAL,CAAea,KAAf,CAAsB,EAAhE,CAAT;EACA,OAAOA,KAAK,CAACZ,SAAN,CACJE,MADI,CACG,CAAC;IAACG;EAAD,CAAD,KAAYA,IAAI,KAAKW,YADxB,EAEJtB,GAFI,CAEA,CAAC;IAACa;EAAD,CAAD,KAAaD,kBAAkB,CAACC,KAAD,CAAlB,GAA4BE,MAAM,CAACF,KAAD,CAAlC,GAA4C,EAFzD,EAGJL,MAHI,CAGGK,KAAK,IAAIA,KAHZ,CAAP;AAID;;AAEM,SAASD,kBAAT,CAA4BC,KAA5B,EAAmC;EACxC,IAAI,OAAOA,KAAP,KAAiB,QAAjB,IAA6B,OAAOA,KAAP,KAAiB,QAAlD,EAA4D;IAC1D,OAAO,IAAP;EACD;;EACD,OAAO,KAAP;AACD"}

package/package.json

@@ -14,7 +14,7 @@
 		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
 	},
 	"license": "LGPL-3.0+",
-	"version": "2.2.0-alpha.4",
+	"version": "2.2.1-alpha.2",
 	"main": "./dist/index.js",
 	"engines": {
 		"node": ">=14"
@@ -23,6 +23,7 @@
 		"access": "public"
 	},
 	"scripts": {
+		"prepare": "npm run build",
 		"start": "node dist/index.js",
 		"cli": "node dist/cli.js",
 		"lint": "eslint src",

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

@@ -5,7 +5,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -28,7 +28,7 @@
 */
 import createDebugLogger from 'debug';
 import {toQueries} from '../candidate-search-utils';
-import {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues} from '../../matching-utils';
+import {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues, testStringOrNumber} from '../../matching-utils';
 
 
 export function bibSourceIds(record) {
@@ -108,7 +108,7 @@ export function bibSourceIds(record) {
 
         function removeSourcePrefix(subfieldValue) {
           const sourcePrefixRegex = (/^(?<sourcePrefix>\([A-Za-z0-9-]+\))(?<id>.+)$/u);
-          const normalizedValue = subfieldValue ? subfieldValue.replace(sourcePrefixRegex, '$<id>') : '';
+          const normalizedValue = testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(sourcePrefixRegex, '$<id>') : '';
           debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);
           return normalizedValue;
         }
@@ -116,7 +116,7 @@ export function bibSourceIds(record) {
         function normalizeSidSubfieldValue(subfieldValue) {
           debugData(`Normalizing ${subfieldValue}`);
           const normalizeAwayRegex = (/[- ]/u);
-          return subfieldValue ? subfieldValue.replace(normalizeAwayRegex, '') : '';
+          return testStringOrNumber(subfieldValue) ? String(subfieldValue).replace(normalizeAwayRegex, '') : '';
         }
 
       }
@@ -151,7 +151,7 @@ export function bibMelindaIds(record) {
 // bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?
 export function bibHostComponents(record) {
   const id = getHostId();
-  return id ? [`melinda.partsofhost=${id}`] : [];
+  return testStringOrNumber(id) ? [`melinda.partsofhost=${id}`] : [];
 
   function getHostId() {
     const [field] = record.get(/^773$/u);
@@ -159,12 +159,12 @@ export function bibHostComponents(record) {
     if (field) {
       const {value} = field.subfields.find(({code}) => code === 'w') || {};
 
-      if (value && (/^\(FI-MELINDA\)/u).test(value)) {
-        return value.replace(/^\(FI-MELINDA\)/u, '');
+      if (testStringOrNumber(value) && (/^\(FI-MELINDA\)/u).test(String(value))) {
+        return String(value).replace(/^\(FI-MELINDA\)/u, '');
       }
 
-      if (value && (/^\(FIN01\)/u).test(value)) {
-        return value.replace(/^\(FIN01\)/u, '');
+      if (testStringOrNumber(value) && (/^\(FIN01\)/u).test(String(value))) {
+        return String(value).replace(/^\(FIN01\)/u, '');
       }
 
       return false;
@@ -185,8 +185,8 @@ export function bibTitle(record) {
   const title = getTitle();
   const booleanStartWords = ['and', 'or', 'nor', 'not'];
 
-  if (title) {
-    const formatted = title
+  if (testStringOrNumber(title)) {
+    const formatted = String(title)
       .replace(/[^\w\s\p{Alphabetic}]/gu, '')
       // Clean up concurrent spaces from fe. subfield changes
       .replace(/ +/gu, ' ')
@@ -208,7 +208,8 @@ export function bibTitle(record) {
     if (field) {
       return field.subfields
         .filter(({code}) => ['a', 'b'].includes(code))
-        .map(({value}) => value)
+        .map(({value}) => testStringOrNumber(value) ? String(value) : '')
+        .filter(value => value)
         // In Melinda's index subfield separators are indexed as ' '
         .join(' ');
     }
@@ -244,18 +245,18 @@ export function bibStandardIdentifiers(record) {
 
     if (tag === '022') {
       return subfields
-        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)
-        .map(({value}) => value);
+        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))
+        .map(({value}) => String(value));
     }
 
     if (tag === '020') {
       return subfields
-        .filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)
-        .map(({value}) => value);
+        .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && issnIsbnReqExp.test(String(sub.value)))
+        .map(({value}) => String(value));
     }
 
     return subfields
-      .filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined)
-      .map(({value}) => value);
+      .filter(sub => ['a', 'z'].includes(sub.code) && testStringOrNumber(sub.value) && otherIdReqExp.test(String(sub.value)))
+      .map(({value}) => String(value));
   }
 }

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

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *

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

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -27,6 +27,9 @@
 */
 
 import {LevenshteinDistance as leven} from 'natural';
+import {testStringOrNumber} from '../../../matching-utils';
+
+// We should extract also organisational authors (110/710)
 
 export default ({nameTreshold = 10} = {}) => ({
   name: 'Authors',
@@ -40,7 +43,7 @@ export default ({nameTreshold = 10} = {}) => ({
 
       function toObj({code, value}) {
         if (code === 'a') {
-          return {name: value ? value.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''};
+          return {name: testStringOrNumber(value) ? String(value).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''};
         }
 
         return {id: value};

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

@@ -5,7 +5,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -86,7 +86,7 @@ export default () => ({
       return 0;
     }
 
-    debugData(`There area languages to compare`);
+    debugData(`There are languages to compare`);
 
     if (a[0] === b[0]) {
       return 0.1;

package/src/match-detection/features/bib/publication-time.js

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -26,11 +26,16 @@
 *
 */
 
+import {testStringOrNumber} from '../../../matching-utils';
+
+// We should also get copyright time and copyright/publication times from 26x
+// We could also enable matching of records which have consequent publication years
+
 export default () => ({
   name: 'Publication time',
   extract: record => {
-    const value = record.get(/^008$/u)?.[0]?.value || '';
-    return value ? [value.slice(7, 11)] : [];
+    const value = record.get(/^008$/u)?.[0]?.value || undefined;
+    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
   },
   compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0
 });

package/src/match-detection/features/bib/record-type.js

@@ -26,6 +26,8 @@
 *
 */
 
+// we could handle the case of books/notes
+
 export default () => ({
   name: 'Record type',
   extract: r => r.leader[6] ? [r.leader[6]] : [],

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

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -27,6 +27,7 @@
 */
 
 import createDebugLogger from 'debug';
+import {testStringOrNumber} from '../../../matching-utils';
 
 const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');
 const debugData = debug.extend('data');
@@ -40,8 +41,8 @@ export default ({pattern, subfieldCodes}) => {
     if (field) {
       return field.subfields
         .filter(({code}) => subfieldCodes.includes(code))
-      //  .map(({code, value}) => ({code, value: value.replace(/-/ug, '')}));
-        .map(({code, value}) => ({code, value: value ? value.replace(/-/ug, '') : ''}));
+        .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))
+        .filter(value => value);
     }
 
     return [];

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

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -28,14 +28,15 @@
 
 import createDebugLogger from 'debug';
 import {LevenshteinDistance as leven} from 'natural';
+import {testStringOrNumber} from '../../../matching-utils';
 
 export default ({treshold = 10} = {}) => ({
   name: 'Title',
   extract: record => {
     const title = getTitle();
 
-    if (title) {
-      return [title.replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
+    if (testStringOrNumber(title)) {
+      return [String(title).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
     }
 
     return [];
@@ -46,9 +47,10 @@ export default ({treshold = 10} = {}) => ({
       if (field) {
         return field.subfields
           .filter(({code}) => ['a', 'b'].includes(code))
-          .map(({value}) => value)
+          .map(({value}) => testStringOrNumber(value) ? String(value) : '')
           .join('');
       }
+      return false;
     }
   },
   compare: (a, b) => {

package/src/matching-utils.js

@@ -4,7 +4,7 @@
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -58,8 +58,8 @@ export function getMelindaIdsF035(record) {
 
     return subfields
       .filter(sub => ['a', 'z'].includes(sub.code))
-      .filter(sub => melindaIdRegExp.test(sub.value))
-      .map(({value}) => value ? value.replace(melindaIdRegExp, '$<id>') : '');
+      .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))
+      .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');
 
   }
 }
@@ -81,6 +81,15 @@ function countSubfields(field, subfieldCode) {
 
 export function getSubfieldValues(field, subfieldCode) {
   debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);
-  return field.subfields.filter(({code}) => code === subfieldCode).map(({value}) => value);
+  return field.subfields
+    .filter(({code}) => code === subfieldCode)
+    .map(({value}) => testStringOrNumber(value) ? String(value) : '')
+    .filter(value => value);
 }
 
+export function testStringOrNumber(value) {
+  if (typeof value === 'string' || typeof value === 'number') {
+    return true;
+  }
+  return false;
+}