@natlibfi/melinda-record-matching 5.0.4 → 5.0.5

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

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../src/candidate-search/query-list/component.js"],
-  "sourcesContent": ["import createDebugLogger from 'debug';\nimport {promisify} from 'util';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getSubfieldValues, testStringOrNumber, toMelindaIds} from '../../matching-utils.js';\n\nconst setTimeoutPromise = promisify(setTimeout); // eslint-disable-line\n\nexport function hostIdMelinda(record) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:hostIdMelinda');\n  const debugData = debug.extend('data'); // eslint-disable-line\n  debug(`Creating query for the Melinda Id host`);\n\n  const ids = hostId(record);\n  if (ids.length > 0) {\n    return toQueries(ids, 'melinda.partsofhost');\n  }\n\n  debug(`No valid Melinda Id host found`);\n  return [];\n\n  function hostId(record) {\n    // Multi 773 handling\n    const f773s = record.get(/^773$/u)\n      .filter(f773 => f773.subfields.some(sub => sub.code === 'w' && (/\\(FI-MELINDA\\).*/ui).test(sub.value)));\n\n    if (f773s.length === 0) {\n      return false;\n    }\n\n    // Multi $w handling\n    // $w (prefix)<id> handling\n    // $w <id> & $w (prefix)<id> Match\n    const melindaIds = f773s.map(f773 => toMelindaIds(f773, ['w'])).flat()\n      .filter(value => testStringOrNumber(value)) // drop invalid values\n      .filter((value, index, array) => array.indexOf(value) === index); // unique values;\n    return melindaIds;\n  }\n}\n\nexport async function hostIdOtherSource(record, client) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:hostIdOtherSource');\n  const debugData = debug.extend('data'); // eslint-disable-line\n  debug(`Creating query for the Other source Id host`);\n\n  const ids = getHostIds(record);\n  const otherSources = getOtherSources(record);\n  if (ids.length > 0 && otherSources.length > 0) {\n    const otherSourceIds = ids.map(id => otherSources.map(source => `${id}${source}`)).flat();\n    const melindaHostQueries = await handleSruCalls(otherSourceIds);\n    debug(JSON.stringify(melindaHostQueries));\n\n    return melindaHostQueries;\n  }\n\n  debug(`No valid other source hosts found`);\n  return [];\n\n\n  function getHostIds(record) {\n    const f773s = record.get(/^773$/u)\n      .filter(f773 => f773.subfields.some(sub => sub.code === 'w' && !(/\\(FI-MELINDA\\).*/ui).test(sub.value)));\n    if (f773s.length === 0) {\n      return false;\n    }\n\n    // Multi 773 handling\n    const subfieldWs = f773s\n      .map(f773 => {\n        debugData(`f773: ${JSON.stringify(f773)}`);\n        return getSubfieldValues(f773, 'w').flat();\n      }).flat();\n\n    // Multi $w handling\n    // $w (prefix)<id> handling\n    // $w <id> & $w (prefix)<id> Match\n    const ids = subfieldWs.map(value => `${value}`.replace(/\\(FI-.*\\)/ui, '')) // remove prefixes\n      .filter(value => testStringOrNumber(value)) // drop invalid values\n      .filter((value, index, array) => array.indexOf(value) === index); // unique values\n\n    return ids;\n  }\n\n  function getOtherSources(record) {\n    const fSids = record.get('SID');\n    return fSids.map(field => getSubfieldValues(field, 'b'));\n  }\n\n  async function handleSruCalls(otherSourceIds, ids = []) {\n    const [otherSourceId, ...rest] = otherSourceIds;\n\n    if (otherSourceId === undefined) {\n      debug(`host ids: ${ids}`);\n      const validIds = ids.filter(id => id);\n      return toQueries(validIds, 'melinda.partsofhost');\n    }\n\n    const otherSourceHostQuery = await toQueries([otherSourceId], 'melinda.sourceid');\n    const id = await new Promise((resolve, reject) => {\n      debug(`Searching for hosts with query: ${otherSourceHostQuery}`);\n      let recordId;\n\n      client.searchRetrieve(otherSourceHostQuery)\n        .on('error', err => {\n          debug(`SRU error for query: ${otherSourceHostQuery}: ${err}`);\n          reject(err);\n        })\n        .on('end', async () => {\n          try {\n            debug(`Searching for hosts: done`);\n            await setTimeoutPromise(10);\n            resolve(recordId);\n          } catch (err) {\n            debug(`Error caught on END`);\n            reject(err);\n          }\n        })\n        .on('record', record => {\n          const [field] = record.get(/^001$/u);\n          debug(field);\n          recordId = field.value ? field.value : '';\n        });\n    });\n\n    return handleSruCalls(rest, [...ids, id]);\n  }\n}\n\n"],
-  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,oBAAoB,oBAAmB;AAElE,MAAM,oBAAoB,UAAU,UAAU;AAEvC,gBAAS,cAAc,QAAQ;AACpC,QAAM,QAAQ,kBAAkB,wEAAwE;AACxG,QAAM,YAAY,MAAM,OAAO,MAAM;AACrC,QAAM,wCAAwC;AAE9C,QAAM,MAAM,OAAO,MAAM;AACzB,MAAI,IAAI,SAAS,GAAG;AAClB,WAAO,UAAU,KAAK,qBAAqB;AAAA,EAC7C;AAEA,QAAM,gCAAgC;AACtC,SAAO,CAAC;AAER,WAAS,OAAOA,SAAQ;AAEtB,UAAM,QAAQA,QAAO,IAAI,QAAQ,EAC9B,OAAO,UAAQ,KAAK,UAAU,KAAK,SAAO,IAAI,SAAS,OAAQ,qBAAsB,KAAK,IAAI,KAAK,CAAC,CAAC;AAExG,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO;AAAA,IACT;AAKA,UAAM,aAAa,MAAM,IAAI,UAAQ,aAAa,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAClE,OAAO,WAAS,mBAAmB,KAAK,CAAC,EACzC,OAAO,CAAC,OAAO,OAAO,UAAU,MAAM,QAAQ,KAAK,MAAM,KAAK;AACjE,WAAO;AAAA,EACT;AACF;AAEA,sBAAsB,kBAAkB,QAAQ,QAAQ;AACtD,QAAM,QAAQ,kBAAkB,4EAA4E;AAC5G,QAAM,YAAY,MAAM,OAAO,MAAM;AACrC,QAAM,6CAA6C;AAEnD,QAAM,MAAM,WAAW,MAAM;AAC7B,QAAM,eAAe,gBAAgB,MAAM;AAC3C,MAAI,IAAI,SAAS,KAAK,aAAa,SAAS,GAAG;AAC7C,UAAM,iBAAiB,IAAI,IAAI,QAAM,aAAa,IAAI,YAAU,GAAG,EAAE,GAAG,MAAM,EAAE,CAAC,EAAE,KAAK;AACxF,UAAM,qBAAqB,MAAM,eAAe,cAAc;AAC9D,UAAM,KAAK,UAAU,kBAAkB,CAAC;AAExC,WAAO;AAAA,EACT;AAEA,QAAM,mCAAmC;AACzC,SAAO,CAAC;AAGR,WAAS,WAAWA,SAAQ;AAC1B,UAAM,QAAQA,QAAO,IAAI,QAAQ,EAC9B,OAAO,UAAQ,KAAK,UAAU,KAAK,SAAO,IAAI,SAAS,OAAO,CAAE,qBAAsB,KAAK,IAAI,KAAK,CAAC,CAAC;AACzG,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO;AAAA,IACT;AAGA,UAAM,aAAa,MAChB,IAAI,UAAQ;AACX,gBAAU,SAAS,KAAK,UAAU,IAAI,CAAC,EAAE;AACzC,aAAO,kBAAkB,MAAM,GAAG,EAAE,KAAK;AAAA,IAC3C,CAAC,EAAE,KAAK;AAKV,UAAMC,OAAM,WAAW,IAAI,WAAS,GAAG,KAAK,GAAG,QAAQ,eAAe,EAAE,CAAC,EACtE,OAAO,WAAS,mBAAmB,KAAK,CAAC,EACzC,OAAO,CAAC,OAAO,OAAO,UAAU,MAAM,QAAQ,KAAK,MAAM,KAAK;AAEjE,WAAOA;AAAA,EACT;AAEA,WAAS,gBAAgBD,SAAQ;AAC/B,UAAM,QAAQA,QAAO,IAAI,KAAK;AAC9B,WAAO,MAAM,IAAI,WAAS,kBAAkB,OAAO,GAAG,CAAC;AAAA,EACzD;AAEA,iBAAe,eAAe,gBAAgBC,OAAM,CAAC,GAAG;AACtD,UAAM,CAAC,eAAe,GAAG,IAAI,IAAI;AAEjC,QAAI,kBAAkB,QAAW;AAC/B,YAAM,aAAaA,IAAG,EAAE;AACxB,YAAM,WAAWA,KAAI,OAAO,CAAAC,QAAMA,GAAE;AACpC,aAAO,UAAU,UAAU,qBAAqB;AAAA,IAClD;AAEA,UAAM,uBAAuB,MAAM,UAAU,CAAC,aAAa,GAAG,kBAAkB;AAChF,UAAM,KAAK,MAAM,IAAI,QAAQ,CAAC,SAAS,WAAW;AAChD,YAAM,mCAAmC,oBAAoB,EAAE;AAC/D,UAAI;AAEJ,aAAO,eAAe,oBAAoB,EACvC,GAAG,SAAS,SAAO;AAClB,cAAM,wBAAwB,oBAAoB,KAAK,GAAG,EAAE;AAC5D,eAAO,GAAG;AAAA,MACZ,CAAC,EACA,GAAG,OAAO,YAAY;AACrB,YAAI;AACF,gBAAM,2BAA2B;AACjC,gBAAM,kBAAkB,EAAE;AAC1B,kBAAQ,QAAQ;AAAA,QAClB,SAAS,KAAK;AACZ,gBAAM,qBAAqB;AAC3B,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF,CAAC,EACA,GAAG,UAAU,CAAAF,YAAU;AACtB,cAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,QAAQ;AACnC,cAAM,KAAK;AACX,mBAAW,MAAM,QAAQ,MAAM,QAAQ;AAAA,MACzC,CAAC;AAAA,IACL,CAAC;AAED,WAAO,eAAe,MAAM,CAAC,GAAGC,MAAK,EAAE,CAAC;AAAA,EAC1C;AACF;",
-  "names": ["record", "ids", "id"]
+  "sourcesContent": ["import createDebugLogger from 'debug';\nimport {promisify} from 'util';\nimport {toQueries} from '../candidate-search-utils.js';\nimport {getSubfieldValues, stringAfter, stringBefore, testStringOrNumber, toMelindaIds} from '../../matching-utils.js';\nimport {fieldToString, uniqArray} from '@natlibfi/marc-record-validators-melinda/dist/utils.js';\n\nconst setTimeoutPromise = promisify(setTimeout); // eslint-disable-line\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:component');\nconst debugData = debug.extend('data'); // eslint-disable-line\n\n\n\nexport function parse773g(field) {\n  const g = field?.subfields?.find(sf => sf.code === 'g'); // NB! returns the first $g\n\n  if (!g) {\n    return {};\n  }\n\n  const value = normalizeValue(g.value);\n\n  return {\n    // NB! We are not currently interested in volume (vuosikerta)\n    year: gToYear(value) || null,\n    number: gToNumber(value) || null,\n    pages: gToPages(value) || null // Might also be eg. Raita 5. However, returns only the number part\n  }\n\n  function normalizeValue(value) {\n    return value.replace(/(?:\\. -|\\.)$/u, '');\n  }\n\n  function gToNumber(value) {\n    // Allow up to 3-digit numbers\n    const extract1 = value.replace(/^.*: ([1-9][0-9]?[0-9]), (?:p\\.|page|pp\\.|s\\.|Seite|sida|sidor|sivu).*$/ui, '$1');\n    if (extract1 !== value) {\n      return extract1;\n    }\n    return undefined;\n  }\n\n  function gToPages(value) {\n    if (value.match(/^[0-9]+(?:-[0-9]+)?(?:, [0-9]+(?:-[0-9]+)?)*$/u)) {\n      return value;\n    }\n    const numberPartOnly = value.replace(/^.*(?:p\\.|raidat|raita|s\\.|Seite|sivut?|pages?) /ui, '');\n    if (numberPartOnly !== value) {\n      return gToPages(numberPartOnly);\n    }\n    return undefined;\n  }\n\n\n  function gToYear(value) {\n    debug(`EXTRACT YEAR FROM '${value}`);\n    // extract year from within parentheses: $g vsk (yyyy) AND $g (YYYY)\n    if (value.match(/^[1-9][0-9]?[0-9]? ?\\((?:20[012][0-9]|19[0-9][0-9])\\)/u) || value.match(/^\\((?:20[012][0-9]|19[0-9][0-9])\\)/u) ) {\n      return stringBefore(stringAfter(value, '('), ')');\n    }\n    // If volume is missing, the year often seems to come without them parentheses:\n    if (value.match(/^(?:20[012][0-9]|19[0-9][0-9]) :/u)) {\n      return stringBefore(value, ' ');\n    }\n    // Seen in SB $g (vsk)year joulukuu\n    if (value.match(/^\\([1-9][0-9]?\\) ?(?:20[012][0-9]|19[0-9][0-9])[^0-9]/)) {\n      return value.replace(/^\\([0-9]+\\) ?/u, '').replace(/[^0-9].*$/u, '');\n    }\n    // Some magazines use DD.MM.YYYY\n    if (value.match(/^[^0-9]*(?:[1-9]|[12][0-9]|30|31)\\.(?:1[012]|[1-9])\\. ?(?:20[012][0-9]|19[0-9][0-9])[^0-9]/u)) {\n      return value.replace(/^[^0-9]*[0-9]+\\.[0-9]+\\. ?/u, '').replace(/[^0-9].*$/u, '');\n    }\n\n    return undefined;\n  }\n}\n\nexport function extractPublicationYearFrom773(field) {\n  const data = parse773g(field);\n  debug(`Extracted year ${data.year} from f773$g ${fieldToString(field)}`);\n  return data.year;\n}\n\nfunction isMelindaSubfieldW(subfield) {\n  if (subfield.code !== 'w') {\n    return false;\n  }\n  return (/^\\((?:FI-MELINDA|FIN01)\\)0[0-9]{8}$/ui).test(subfield.value);\n}\n\nfunction removeMelindaPrefixFromValue(value) {\n  return value.replace(/^(?:\\(FI-MELINDA\\)|\\(FIN01\\))/, '');\n}\n\nexport function getHostMelindaFields(record) {\n  return record.get(/^773$/u).filter(f => f.subfields.some(sf => isMelindaSubfieldW(sf)));\n}\n\nexport function extractHostMelindaIdsFromField(field) {\n  return field.subfields.filter(sf => isMelindaSubfieldW(sf)).map(sf => removeMelindaPrefixFromValue(sf.value));\n}\n\nexport function getHostMelindaIds(record) {\n  return uniqArray(getHostMelindaFields(record).map(f => extractHostMelindaIdsFromField(f)).flat());\n}\n\nexport function hostIdMelinda(record) { // old function with replaced code\n  debug(`Creating query for the Melinda Id host`);\n  const ids = getHostMelindaIds(record);\n  if (ids.length === 0) {\n    debug(`No valid Melinda Id host found`);\n    return [];\n  }\n  return toQueries(getHostMelindaIds(record), 'melinda.partsofhost');\n}\n\nexport async function hostIdOtherSource(record, client) {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:hostIdOtherSource');\n  const debugData = debug.extend('data'); // eslint-disable-line\n  debug(`Creating query for the Other source Id host`);\n\n  const ids = getHostIds(record);\n  const otherSources = getOtherSources(record);\n  if (ids.length > 0 && otherSources.length > 0) {\n    const otherSourceIds = ids.map(id => otherSources.map(source => `${id}${source}`)).flat();\n    const melindaHostQueries = await handleSruCalls(otherSourceIds);\n    debug(JSON.stringify(melindaHostQueries));\n\n    return melindaHostQueries;\n  }\n\n  debug(`No valid other source hosts found`);\n  return [];\n\n\n  function getHostIds(record) {\n    const f773s = record.get(/^773$/u)\n      .filter(f773 => f773.subfields.some(sub => sub.code === 'w' && !(/\\(FI-MELINDA\\).*/ui).test(sub.value)));\n    if (f773s.length === 0) {\n      return false;\n    }\n\n    // Multi 773 handling\n    const subfieldWs = f773s\n      .map(f773 => {\n        debugData(`f773: ${JSON.stringify(f773)}`);\n        return getSubfieldValues(f773, 'w').flat();\n      }).flat();\n\n    // Multi $w handling\n    // $w (prefix)<id> handling\n    // $w <id> & $w (prefix)<id> Match\n    const ids = subfieldWs.map(value => `${value}`.replace(/\\(FI-.*\\)/ui, '')) // remove prefixes\n      .filter(value => testStringOrNumber(value)) // drop invalid values\n      .filter((value, index, array) => array.indexOf(value) === index); // unique values\n\n    return ids;\n  }\n\n  function getOtherSources(record) {\n    const fSids = record.get('SID');\n    return fSids.map(field => getSubfieldValues(field, 'b'));\n  }\n\n  async function handleSruCalls(otherSourceIds, ids = []) {\n    const [otherSourceId, ...rest] = otherSourceIds;\n\n    if (otherSourceId === undefined) {\n      debug(`host ids: ${ids}`);\n      const validIds = ids.filter(id => id);\n      return toQueries(validIds, 'melinda.partsofhost');\n    }\n\n    const otherSourceHostQuery = await toQueries([otherSourceId], 'melinda.sourceid');\n    const id = await new Promise((resolve, reject) => {\n      debug(`Searching for hosts with query: ${otherSourceHostQuery}`);\n      let recordId;\n\n      client.searchRetrieve(otherSourceHostQuery)\n        .on('error', err => {\n          debug(`SRU error for query: ${otherSourceHostQuery}: ${err}`);\n          reject(err);\n        })\n        .on('end', async () => {\n          try {\n            debug(`Searching for hosts: done`);\n            await setTimeoutPromise(10);\n            resolve(recordId);\n          } catch (err) {\n            debug(`Error caught on END`);\n            reject(err);\n          }\n        })\n        .on('record', record => {\n          const [field] = record.get(/^001$/u);\n          debug(field);\n          recordId = field.value ? field.value : '';\n        });\n    });\n\n    return handleSruCalls(rest, [...ids, id]);\n  }\n}\n\n"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAC9B,SAAQ,iBAAgB;AACxB,SAAQ,iBAAgB;AACxB,SAAQ,mBAAmB,aAAa,cAAc,oBAAoB,oBAAmB;AAC7F,SAAQ,eAAe,iBAAgB;AAEvC,MAAM,oBAAoB,UAAU,UAAU;AAE9C,MAAM,QAAQ,kBAAkB,oEAAoE;AACpG,MAAM,YAAY,MAAM,OAAO,MAAM;AAI9B,gBAAS,UAAU,OAAO;AAC/B,QAAM,IAAI,OAAO,WAAW,KAAK,QAAM,GAAG,SAAS,GAAG;AAEtD,MAAI,CAAC,GAAG;AACN,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,QAAQ,eAAe,EAAE,KAAK;AAEpC,SAAO;AAAA;AAAA,IAEL,MAAM,QAAQ,KAAK,KAAK;AAAA,IACxB,QAAQ,UAAU,KAAK,KAAK;AAAA,IAC5B,OAAO,SAAS,KAAK,KAAK;AAAA;AAAA,EAC5B;AAEA,WAAS,eAAeA,QAAO;AAC7B,WAAOA,OAAM,QAAQ,iBAAiB,EAAE;AAAA,EAC1C;AAEA,WAAS,UAAUA,QAAO;AAExB,UAAM,WAAWA,OAAM,QAAQ,6EAA6E,IAAI;AAChH,QAAI,aAAaA,QAAO;AACtB,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT;AAEA,WAAS,SAASA,QAAO;AACvB,QAAIA,OAAM,MAAM,gDAAgD,GAAG;AACjE,aAAOA;AAAA,IACT;AACA,UAAM,iBAAiBA,OAAM,QAAQ,sDAAsD,EAAE;AAC7F,QAAI,mBAAmBA,QAAO;AAC5B,aAAO,SAAS,cAAc;AAAA,IAChC;AACA,WAAO;AAAA,EACT;AAGA,WAAS,QAAQA,QAAO;AACtB,UAAM,sBAAsBA,MAAK,EAAE;AAEnC,QAAIA,OAAM,MAAM,wDAAwD,KAAKA,OAAM,MAAM,qCAAqC,GAAI;AAChI,aAAO,aAAa,YAAYA,QAAO,GAAG,GAAG,GAAG;AAAA,IAClD;AAEA,QAAIA,OAAM,MAAM,mCAAmC,GAAG;AACpD,aAAO,aAAaA,QAAO,GAAG;AAAA,IAChC;AAEA,QAAIA,OAAM,MAAM,uDAAuD,GAAG;AACxE,aAAOA,OAAM,QAAQ,kBAAkB,EAAE,EAAE,QAAQ,cAAc,EAAE;AAAA,IACrE;AAEA,QAAIA,OAAM,MAAM,6FAA6F,GAAG;AAC9G,aAAOA,OAAM,QAAQ,+BAA+B,EAAE,EAAE,QAAQ,cAAc,EAAE;AAAA,IAClF;AAEA,WAAO;AAAA,EACT;AACF;AAEO,gBAAS,8BAA8B,OAAO;AACnD,QAAM,OAAO,UAAU,KAAK;AAC5B,QAAM,kBAAkB,KAAK,IAAI,gBAAgB,cAAc,KAAK,CAAC,EAAE;AACvE,SAAO,KAAK;AACd;AAEA,SAAS,mBAAmB,UAAU;AACpC,MAAI,SAAS,SAAS,KAAK;AACzB,WAAO;AAAA,EACT;AACA,SAAQ,wCAAyC,KAAK,SAAS,KAAK;AACtE;AAEA,SAAS,6BAA6B,OAAO;AAC3C,SAAO,MAAM,QAAQ,iCAAiC,EAAE;AAC1D;AAEO,gBAAS,qBAAqB,QAAQ;AAC3C,SAAO,OAAO,IAAI,QAAQ,EAAE,OAAO,OAAK,EAAE,UAAU,KAAK,QAAM,mBAAmB,EAAE,CAAC,CAAC;AACxF;AAEO,gBAAS,+BAA+B,OAAO;AACpD,SAAO,MAAM,UAAU,OAAO,QAAM,mBAAmB,EAAE,CAAC,EAAE,IAAI,QAAM,6BAA6B,GAAG,KAAK,CAAC;AAC9G;AAEO,gBAAS,kBAAkB,QAAQ;AACxC,SAAO,UAAU,qBAAqB,MAAM,EAAE,IAAI,OAAK,+BAA+B,CAAC,CAAC,EAAE,KAAK,CAAC;AAClG;AAEO,gBAAS,cAAc,QAAQ;AACpC,QAAM,wCAAwC;AAC9C,QAAM,MAAM,kBAAkB,MAAM;AACpC,MAAI,IAAI,WAAW,GAAG;AACpB,UAAM,gCAAgC;AACtC,WAAO,CAAC;AAAA,EACV;AACA,SAAO,UAAU,kBAAkB,MAAM,GAAG,qBAAqB;AACnE;AAEA,sBAAsB,kBAAkB,QAAQ,QAAQ;AACtD,QAAMC,SAAQ,kBAAkB,4EAA4E;AAC5G,QAAMC,aAAYD,OAAM,OAAO,MAAM;AACrC,EAAAA,OAAM,6CAA6C;AAEnD,QAAM,MAAM,WAAW,MAAM;AAC7B,QAAM,eAAe,gBAAgB,MAAM;AAC3C,MAAI,IAAI,SAAS,KAAK,aAAa,SAAS,GAAG;AAC7C,UAAM,iBAAiB,IAAI,IAAI,QAAM,aAAa,IAAI,YAAU,GAAG,EAAE,GAAG,MAAM,EAAE,CAAC,EAAE,KAAK;AACxF,UAAM,qBAAqB,MAAM,eAAe,cAAc;AAC9D,IAAAA,OAAM,KAAK,UAAU,kBAAkB,CAAC;AAExC,WAAO;AAAA,EACT;AAEA,EAAAA,OAAM,mCAAmC;AACzC,SAAO,CAAC;AAGR,WAAS,WAAWE,SAAQ;AAC1B,UAAM,QAAQA,QAAO,IAAI,QAAQ,EAC9B,OAAO,UAAQ,KAAK,UAAU,KAAK,SAAO,IAAI,SAAS,OAAO,CAAE,qBAAsB,KAAK,IAAI,KAAK,CAAC,CAAC;AACzG,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO;AAAA,IACT;AAGA,UAAM,aAAa,MAChB,IAAI,UAAQ;AACX,MAAAD,WAAU,SAAS,KAAK,UAAU,IAAI,CAAC,EAAE;AACzC,aAAO,kBAAkB,MAAM,GAAG,EAAE,KAAK;AAAA,IAC3C,CAAC,EAAE,KAAK;AAKV,UAAME,OAAM,WAAW,IAAI,WAAS,GAAG,KAAK,GAAG,QAAQ,eAAe,EAAE,CAAC,EACtE,OAAO,WAAS,mBAAmB,KAAK,CAAC,EACzC,OAAO,CAAC,OAAO,OAAO,UAAU,MAAM,QAAQ,KAAK,MAAM,KAAK;AAEjE,WAAOA;AAAA,EACT;AAEA,WAAS,gBAAgBD,SAAQ;AAC/B,UAAM,QAAQA,QAAO,IAAI,KAAK;AAC9B,WAAO,MAAM,IAAI,WAAS,kBAAkB,OAAO,GAAG,CAAC;AAAA,EACzD;AAEA,iBAAe,eAAe,gBAAgBC,OAAM,CAAC,GAAG;AACtD,UAAM,CAAC,eAAe,GAAG,IAAI,IAAI;AAEjC,QAAI,kBAAkB,QAAW;AAC/B,MAAAH,OAAM,aAAaG,IAAG,EAAE;AACxB,YAAM,WAAWA,KAAI,OAAO,CAAAC,QAAMA,GAAE;AACpC,aAAO,UAAU,UAAU,qBAAqB;AAAA,IAClD;AAEA,UAAM,uBAAuB,MAAM,UAAU,CAAC,aAAa,GAAG,kBAAkB;AAChF,UAAM,KAAK,MAAM,IAAI,QAAQ,CAAC,SAAS,WAAW;AAChD,MAAAJ,OAAM,mCAAmC,oBAAoB,EAAE;AAC/D,UAAI;AAEJ,aAAO,eAAe,oBAAoB,EACvC,GAAG,SAAS,SAAO;AAClB,QAAAA,OAAM,wBAAwB,oBAAoB,KAAK,GAAG,EAAE;AAC5D,eAAO,GAAG;AAAA,MACZ,CAAC,EACA,GAAG,OAAO,YAAY;AACrB,YAAI;AACF,UAAAA,OAAM,2BAA2B;AACjC,gBAAM,kBAAkB,EAAE;AAC1B,kBAAQ,QAAQ;AAAA,QAClB,SAAS,KAAK;AACZ,UAAAA,OAAM,qBAAqB;AAC3B,iBAAO,GAAG;AAAA,QACZ;AAAA,MACF,CAAC,EACA,GAAG,UAAU,CAAAE,YAAU;AACtB,cAAM,CAAC,KAAK,IAAIA,QAAO,IAAI,QAAQ;AACnC,QAAAF,OAAM,KAAK;AACX,mBAAW,MAAM,QAAQ,MAAM,QAAQ;AAAA,MACzC,CAAC;AAAA,IACL,CAAC;AAED,WAAO,eAAe,MAAM,CAAC,GAAGG,MAAK,EAAE,CAAC;AAAA,EAC1C;AACF;",
+  "names": ["value", "debug", "debugData", "record", "ids", "id"]
 }

package/dist/cli.js

@@ -49,7 +49,9 @@ async function cli() {
     if (["IDS"].includes(searchType2)) {
       return [
         matchDetection.features.bib.melindaId(),
-        matchDetection.features.bib.allSourceIds()
+        matchDetection.features.bib.allSourceIds(),
+        matchDetection.features.bib.f773()
+        // Host data
       ];
     }
     if (["STANDARD_IDS"].includes(searchType2)) {
@@ -70,7 +72,9 @@ async function cli() {
         //       normal publicationTime and a reprintPublication time
         matchDetection.features.bib.publicationTimeAllowConsYearsMulti(),
         matchDetection.features.bib.language(),
-        matchDetection.features.bib.bibliographicLevel()
+        matchDetection.features.bib.bibliographicLevel(),
+        matchDetection.features.bib.f773()
+        // Host data
       ];
     }
     if (["COMPONENT"].includes(searchType2)) {
@@ -81,7 +85,9 @@ async function cli() {
         matchDetection.features.bib.title(),
         matchDetection.features.bib.language(),
         matchDetection.features.bib.authors(),
-        matchDetection.features.bib.bibliographicLevel()
+        matchDetection.features.bib.bibliographicLevel(),
+        matchDetection.features.bib.f773()
+        // Host data
       ];
     }
     if (["CONTENT", "CONTENTALT"].includes(searchType2)) {
@@ -95,7 +101,9 @@ async function cli() {
         matchDetection.features.bib.recordType(),
         matchDetection.features.bib.publicationTime(),
         matchDetection.features.bib.language(),
-        matchDetection.features.bib.bibliographicLevel()
+        matchDetection.features.bib.bibliographicLevel(),
+        matchDetection.features.bib.f773()
+        // Host data
       ];
     }
     throw new Error("Unsupported match validation package");

package/dist/cli.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../src/cli.js"],
-  "sourcesContent": ["import fs from 'fs';\nimport yargs from 'yargs';\nimport createMatchOperator, {candidateSearch, matchDetection} from './index.js';\nimport createDebugLogger from 'debug';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\ncli();\n\nasync function cli() {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:cli');\n  const args = yargs(process.argv.slice(2))\n    .scriptName('melinda-record-matching-js')\n    .epilog('Copyright (C) 2022-2023 University Of Helsinki (The National Library Of Finland)')\n    .usage('$0 <file> [options] and env variable info in README')\n    .usage('Installed globally: $0 <file> [options] and env variable info in README')\n    .usage('Not installed: npx $0 <file> [options] and env variable info in README')\n    .usage('Build from source: node dist/index.js <file> [options] and env variable info in README')\n    .showHelpOnFail(true)\n    .example([['$ node /dist/cli.js record.json']])\n    .env('MELINDA_RECORD_MATCH')\n    .version()\n    .positional('file', {type: 'string', describe: 'Json file of records to match'})\n    .options({\n      t: {type: 'string', default: 'IDS', alias: 'searchType', describe: 'IDS, STANDARD_IDS, COMPONENT, CONTENT or CONTENTALT'},\n      m: {type: 'number', default: 1, alias: 'maxMatches', describe: ''},\n      c: {type: 'number', default: 1000, alias: 'maxCandidates', describe: ''},\n      s: {type: 'boolean', default: false, alias: 'returnStrategy', describe: ''},\n      q: {type: 'boolean', default: false, alias: 'returnQuery', describe: ''},\n      n: {type: 'boolean', default: false, alias: 'returnNonMatches', describe: ''}\n    })\n    .check((args) => {\n      const [file] = args._;\n      if (file === undefined) {\n        throw new Error('No file argument given');\n      }\n\n      if (!fs.existsSync(file)) {\n        throw new Error(`File ${file} does not exist`);\n      }\n\n      if (args.sruUrl === undefined) {\n        throw new Error('Setup sru url');\n      }\n\n      if (!['IDS', 'STANDARD_IDS', 'COMPONENT', 'CONTENT', 'CONTENTALT'].includes(args.searchType)) {\n        throw new Error('Invalid search type');\n      }\n\n      return true;\n    })\n    .parseSync();\n\n  const [file] = args._;\n  const {searchType} = args;\n  debug(JSON.stringify(args));\n\n  const detection = {\n    threshold: 0.8999,\n    strategy: generateStrategy(searchType)\n  };\n\n  const search = {\n    url: args.sruUrl, searchSpec: generateSearchSpec(searchType)\n  };\n\n  const matchOperator = await createMatchOperator({detection, search, ...args});\n\n  const fileRaw = fs.readFileSync(file, 'utf8');\n  const record = new MarcRecord(JSON.parse(fileRaw), {subfieldValues: false});\n\n  const result = await matchOperator({record});\n  debug(JSON.stringify(result));\n\n\n  function generateStrategy(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.melindaId(),\n        matchDetection.features.bib.allSourceIds()\n      ];\n    }\n\n    // We could have differing strategy for STANDARD_IDS\n    // Let's not run title in strategy when we found the candidates through standard_ids search\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        // Let's not use the same title matchDetection here\n        //matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        // We probably should have some leeway here for notated music as BK etc.\n        matchDetection.features.bib.recordType(),\n        // Use publicationTimeAllowConsYearsMulti to\n        //  - ignore one year differences in publicationTime\n        //  - extract publicationTimes from f008, f26x and reprint notes in f500\n        //  - do not substract points for mismatching (normal) publicationTime, if there's a match between\n        //       normal publicationTime and a reprintPublication time\n        matchDetection.features.bib.publicationTimeAllowConsYearsMulti(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel()\n      ];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.bibliographicLevel()\n      ];\n    }\n\n    if (['CONTENT', 'CONTENTALT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.publicationTime(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel()\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n\n\n  function generateSearchSpec(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.melindaId,\n        candidateSearch.searchTypes.bib.sourceIds\n      ];\n    }\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [candidateSearch.searchTypes.bib.standardIdentifiers];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        //candidateSearch.searchTypes.bib.sourceIds,\n        candidateSearch.searchTypes.component.hostIdMelinda,\n        candidateSearch.searchTypes.component.hostIdOtherSource\n      ];\n    }\n\n    if (['CONTENT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        //candidateSearch.searchTypes.bib.titleAuthor,\n        candidateSearch.searchTypes.bib.title\n      ];\n    }\n    if (['CONTENTALT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        // titleAuthorYearAlternates searches for matchCandidates\n        // with alternate queries, starting from more tight searches\n        candidateSearch.searchTypes.bib.titleAuthorYearAlternates\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n}\n"],
-  "mappings": "AAAA,OAAO,QAAQ;AACf,OAAO,WAAW;AAClB,OAAO,uBAAsB,iBAAiB,sBAAqB;AACnE,OAAO,uBAAuB;AAC9B,SAAQ,kBAAiB;AAEzB,IAAI;AAEJ,eAAe,MAAM;AACnB,QAAM,QAAQ,kBAAkB,uCAAuC;AACvE,QAAM,OAAO,MAAM,QAAQ,KAAK,MAAM,CAAC,CAAC,EACrC,WAAW,4BAA4B,EACvC,OAAO,kFAAkF,EACzF,MAAM,qDAAqD,EAC3D,MAAM,yEAAyE,EAC/E,MAAM,wEAAwE,EAC9E,MAAM,wFAAwF,EAC9F,eAAe,IAAI,EACnB,QAAQ,CAAC,CAAC,iCAAiC,CAAC,CAAC,EAC7C,IAAI,sBAAsB,EAC1B,QAAQ,EACR,WAAW,QAAQ,EAAC,MAAM,UAAU,UAAU,gCAA+B,CAAC,EAC9E,QAAQ;AAAA,IACP,GAAG,EAAC,MAAM,UAAU,SAAS,OAAO,OAAO,cAAc,UAAU,sDAAqD;AAAA,IACxH,GAAG,EAAC,MAAM,UAAU,SAAS,GAAG,OAAO,cAAc,UAAU,GAAE;AAAA,IACjE,GAAG,EAAC,MAAM,UAAU,SAAS,KAAM,OAAO,iBAAiB,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,kBAAkB,UAAU,GAAE;AAAA,IAC1E,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,eAAe,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,oBAAoB,UAAU,GAAE;AAAA,EAC9E,CAAC,EACA,MAAM,CAACA,UAAS;AACf,UAAM,CAACC,KAAI,IAAID,MAAK;AACpB,QAAIC,UAAS,QAAW;AACtB,YAAM,IAAI,MAAM,wBAAwB;AAAA,IAC1C;AAEA,QAAI,CAAC,GAAG,WAAWA,KAAI,GAAG;AACxB,YAAM,IAAI,MAAM,QAAQA,KAAI,iBAAiB;AAAA,IAC/C;AAEA,QAAID,MAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,eAAe;AAAA,IACjC;AAEA,QAAI,CAAC,CAAC,OAAO,gBAAgB,aAAa,WAAW,YAAY,EAAE,SAASA,MAAK,UAAU,GAAG;AAC5F,YAAM,IAAI,MAAM,qBAAqB;AAAA,IACvC;AAEA,WAAO;AAAA,EACT,CAAC,EACA,UAAU;AAEb,QAAM,CAAC,IAAI,IAAI,KAAK;AACpB,QAAM,EAAC,WAAU,IAAI;AACrB,QAAM,KAAK,UAAU,IAAI,CAAC;AAE1B,QAAM,YAAY;AAAA,IAChB,WAAW;AAAA,IACX,UAAU,iBAAiB,UAAU;AAAA,EACvC;AAEA,QAAM,SAAS;AAAA,IACb,KAAK,KAAK;AAAA,IAAQ,YAAY,mBAAmB,UAAU;AAAA,EAC7D;AAEA,QAAM,gBAAgB,MAAM,oBAAoB,EAAC,WAAW,QAAQ,GAAG,KAAI,CAAC;AAE5E,QAAM,UAAU,GAAG,aAAa,MAAM,MAAM;AAC5C,QAAM,SAAS,IAAI,WAAW,KAAK,MAAM,OAAO,GAAG,EAAC,gBAAgB,MAAK,CAAC;AAE1E,QAAM,SAAS,MAAM,cAAc,EAAC,OAAM,CAAC;AAC3C,QAAM,KAAK,UAAU,MAAM,CAAC;AAG5B,WAAS,iBAAiBE,aAAY;AACpC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,UAAU;AAAA,QACtC,eAAe,SAAS,IAAI,aAAa;AAAA,MAC3C;AAAA,IACF;AAKA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,wBAAwB;AAAA;AAAA;AAAA,QAGpD,eAAe,SAAS,IAAI,QAAQ;AAAA;AAAA,QAEpC,eAAe,SAAS,IAAI,WAAW;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,QAMvC,eAAe,SAAS,IAAI,mCAAmC;AAAA,QAC/D,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,MACjD;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,mBAAmB;AAAA,MACjD;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,YAAY,EAAE,SAASA,WAAU,GAAG;AAClD,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,gBAAgB;AAAA,QAC5C,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,MACjD;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AAGA,WAAS,mBAAmBA,aAAY;AACtC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA,QAChC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO,CAAC,gBAAgB,YAAY,IAAI,mBAAmB;AAAA,IAC7D;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA;AAAA,QAEL,gBAAgB,YAAY,UAAU;AAAA,QACtC,gBAAgB,YAAY,UAAU;AAAA,MACxC;AAAA,IACF;AAEA,QAAI,CAAC,SAAS,EAAE,SAASA,WAAU,GAAG;AACpC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA,QAEhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AACA,QAAI,CAAC,YAAY,EAAE,SAASA,WAAU,GAAG;AACvC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA;AAAA,QAGhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AACF;",
+  "sourcesContent": ["import fs from 'fs';\nimport yargs from 'yargs';\nimport createMatchOperator, {candidateSearch, matchDetection} from './index.js';\nimport createDebugLogger from 'debug';\nimport {MarcRecord} from '@natlibfi/marc-record';\n\ncli();\n\nasync function cli() {\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:cli');\n  const args = yargs(process.argv.slice(2))\n    .scriptName('melinda-record-matching-js')\n    .epilog('Copyright (C) 2022-2023 University Of Helsinki (The National Library Of Finland)')\n    .usage('$0 <file> [options] and env variable info in README')\n    .usage('Installed globally: $0 <file> [options] and env variable info in README')\n    .usage('Not installed: npx $0 <file> [options] and env variable info in README')\n    .usage('Build from source: node dist/index.js <file> [options] and env variable info in README')\n    .showHelpOnFail(true)\n    .example([['$ node /dist/cli.js record.json']])\n    .env('MELINDA_RECORD_MATCH')\n    .version()\n    .positional('file', {type: 'string', describe: 'Json file of records to match'})\n    .options({\n      t: {type: 'string', default: 'IDS', alias: 'searchType', describe: 'IDS, STANDARD_IDS, COMPONENT, CONTENT or CONTENTALT'},\n      m: {type: 'number', default: 1, alias: 'maxMatches', describe: ''},\n      c: {type: 'number', default: 1000, alias: 'maxCandidates', describe: ''},\n      s: {type: 'boolean', default: false, alias: 'returnStrategy', describe: ''},\n      q: {type: 'boolean', default: false, alias: 'returnQuery', describe: ''},\n      n: {type: 'boolean', default: false, alias: 'returnNonMatches', describe: ''}\n    })\n    .check((args) => {\n      const [file] = args._;\n      if (file === undefined) {\n        throw new Error('No file argument given');\n      }\n\n      if (!fs.existsSync(file)) {\n        throw new Error(`File ${file} does not exist`);\n      }\n\n      if (args.sruUrl === undefined) {\n        throw new Error('Setup sru url');\n      }\n\n      if (!['IDS', 'STANDARD_IDS', 'COMPONENT', 'CONTENT', 'CONTENTALT'].includes(args.searchType)) {\n        throw new Error('Invalid search type');\n      }\n\n      return true;\n    })\n    .parseSync();\n\n  const [file] = args._;\n  const {searchType} = args;\n  debug(JSON.stringify(args));\n\n  const detection = {\n    threshold: 0.8999,\n    strategy: generateStrategy(searchType)\n  };\n\n  const search = {\n    url: args.sruUrl, searchSpec: generateSearchSpec(searchType)\n  };\n\n  const matchOperator = await createMatchOperator({detection, search, ...args});\n\n  const fileRaw = fs.readFileSync(file, 'utf8');\n  const record = new MarcRecord(JSON.parse(fileRaw), {subfieldValues: false});\n\n  const result = await matchOperator({record});\n  debug(JSON.stringify(result));\n\n\n  function generateStrategy(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.melindaId(),\n        matchDetection.features.bib.allSourceIds(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    // We could have differing strategy for STANDARD_IDS\n    // Let's not run title in strategy when we found the candidates through standard_ids search\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        // Let's not use the same title matchDetection here\n        //matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        // We probably should have some leeway here for notated music as BK etc.\n        matchDetection.features.bib.recordType(),\n        // Use publicationTimeAllowConsYearsMulti to\n        //  - ignore one year differences in publicationTime\n        //  - extract publicationTimes from f008, f26x and reprint notes in f500\n        //  - do not substract points for mismatching (normal) publicationTime, if there's a match between\n        //       normal publicationTime and a reprintPublication time\n        matchDetection.features.bib.publicationTimeAllowConsYearsMulti(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    if (['CONTENT', 'CONTENTALT'].includes(searchType)) {\n      return [\n        matchDetection.features.bib.hostComponent(),\n        matchDetection.features.bib.isbn(),\n        matchDetection.features.bib.issn(),\n        matchDetection.features.bib.otherStandardIdentifier(),\n        matchDetection.features.bib.title(),\n        matchDetection.features.bib.authors(),\n        matchDetection.features.bib.recordType(),\n        matchDetection.features.bib.publicationTime(),\n        matchDetection.features.bib.language(),\n        matchDetection.features.bib.bibliographicLevel(),\n        matchDetection.features.bib.f773() // Host data\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n\n\n  function generateSearchSpec(searchType) {\n    if (['IDS'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.melindaId,\n        candidateSearch.searchTypes.bib.sourceIds\n      ];\n    }\n\n    if (['STANDARD_IDS'].includes(searchType)) {\n      return [candidateSearch.searchTypes.bib.standardIdentifiers];\n    }\n\n    if (['COMPONENT'].includes(searchType)) {\n      return [\n        //candidateSearch.searchTypes.bib.sourceIds,\n        candidateSearch.searchTypes.component.hostIdMelinda,\n        candidateSearch.searchTypes.component.hostIdOtherSource\n      ];\n    }\n\n    if (['CONTENT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        //candidateSearch.searchTypes.bib.titleAuthor,\n        candidateSearch.searchTypes.bib.title\n      ];\n    }\n    if (['CONTENTALT'].includes(searchType)) {\n      return [\n        candidateSearch.searchTypes.bib.hostComponents,\n        // titleAuthorYearAlternates searches for matchCandidates\n        // with alternate queries, starting from more tight searches\n        candidateSearch.searchTypes.bib.titleAuthorYearAlternates\n      ];\n    }\n\n    throw new Error('Unsupported match validation package');\n  }\n}\n"],
+  "mappings": "AAAA,OAAO,QAAQ;AACf,OAAO,WAAW;AAClB,OAAO,uBAAsB,iBAAiB,sBAAqB;AACnE,OAAO,uBAAuB;AAC9B,SAAQ,kBAAiB;AAEzB,IAAI;AAEJ,eAAe,MAAM;AACnB,QAAM,QAAQ,kBAAkB,uCAAuC;AACvE,QAAM,OAAO,MAAM,QAAQ,KAAK,MAAM,CAAC,CAAC,EACrC,WAAW,4BAA4B,EACvC,OAAO,kFAAkF,EACzF,MAAM,qDAAqD,EAC3D,MAAM,yEAAyE,EAC/E,MAAM,wEAAwE,EAC9E,MAAM,wFAAwF,EAC9F,eAAe,IAAI,EACnB,QAAQ,CAAC,CAAC,iCAAiC,CAAC,CAAC,EAC7C,IAAI,sBAAsB,EAC1B,QAAQ,EACR,WAAW,QAAQ,EAAC,MAAM,UAAU,UAAU,gCAA+B,CAAC,EAC9E,QAAQ;AAAA,IACP,GAAG,EAAC,MAAM,UAAU,SAAS,OAAO,OAAO,cAAc,UAAU,sDAAqD;AAAA,IACxH,GAAG,EAAC,MAAM,UAAU,SAAS,GAAG,OAAO,cAAc,UAAU,GAAE;AAAA,IACjE,GAAG,EAAC,MAAM,UAAU,SAAS,KAAM,OAAO,iBAAiB,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,kBAAkB,UAAU,GAAE;AAAA,IAC1E,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,eAAe,UAAU,GAAE;AAAA,IACvE,GAAG,EAAC,MAAM,WAAW,SAAS,OAAO,OAAO,oBAAoB,UAAU,GAAE;AAAA,EAC9E,CAAC,EACA,MAAM,CAACA,UAAS;AACf,UAAM,CAACC,KAAI,IAAID,MAAK;AACpB,QAAIC,UAAS,QAAW;AACtB,YAAM,IAAI,MAAM,wBAAwB;AAAA,IAC1C;AAEA,QAAI,CAAC,GAAG,WAAWA,KAAI,GAAG;AACxB,YAAM,IAAI,MAAM,QAAQA,KAAI,iBAAiB;AAAA,IAC/C;AAEA,QAAID,MAAK,WAAW,QAAW;AAC7B,YAAM,IAAI,MAAM,eAAe;AAAA,IACjC;AAEA,QAAI,CAAC,CAAC,OAAO,gBAAgB,aAAa,WAAW,YAAY,EAAE,SAASA,MAAK,UAAU,GAAG;AAC5F,YAAM,IAAI,MAAM,qBAAqB;AAAA,IACvC;AAEA,WAAO;AAAA,EACT,CAAC,EACA,UAAU;AAEb,QAAM,CAAC,IAAI,IAAI,KAAK;AACpB,QAAM,EAAC,WAAU,IAAI;AACrB,QAAM,KAAK,UAAU,IAAI,CAAC;AAE1B,QAAM,YAAY;AAAA,IAChB,WAAW;AAAA,IACX,UAAU,iBAAiB,UAAU;AAAA,EACvC;AAEA,QAAM,SAAS;AAAA,IACb,KAAK,KAAK;AAAA,IAAQ,YAAY,mBAAmB,UAAU;AAAA,EAC7D;AAEA,QAAM,gBAAgB,MAAM,oBAAoB,EAAC,WAAW,QAAQ,GAAG,KAAI,CAAC;AAE5E,QAAM,UAAU,GAAG,aAAa,MAAM,MAAM;AAC5C,QAAM,SAAS,IAAI,WAAW,KAAK,MAAM,OAAO,GAAG,EAAC,gBAAgB,MAAK,CAAC;AAE1E,QAAM,SAAS,MAAM,cAAc,EAAC,OAAM,CAAC;AAC3C,QAAM,KAAK,UAAU,MAAM,CAAC;AAG5B,WAAS,iBAAiBE,aAAY;AACpC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,UAAU;AAAA,QACtC,eAAe,SAAS,IAAI,aAAa;AAAA,QACzC,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAKA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,wBAAwB;AAAA;AAAA;AAAA,QAGpD,eAAe,SAAS,IAAI,QAAQ;AAAA;AAAA,QAEpC,eAAe,SAAS,IAAI,WAAW;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,QAMvC,eAAe,SAAS,IAAI,mCAAmC;AAAA,QAC/D,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,QAAI,CAAC,WAAW,YAAY,EAAE,SAASA,WAAU,GAAG;AAClD,aAAO;AAAA,QACL,eAAe,SAAS,IAAI,cAAc;AAAA,QAC1C,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,KAAK;AAAA,QACjC,eAAe,SAAS,IAAI,wBAAwB;AAAA,QACpD,eAAe,SAAS,IAAI,MAAM;AAAA,QAClC,eAAe,SAAS,IAAI,QAAQ;AAAA,QACpC,eAAe,SAAS,IAAI,WAAW;AAAA,QACvC,eAAe,SAAS,IAAI,gBAAgB;AAAA,QAC5C,eAAe,SAAS,IAAI,SAAS;AAAA,QACrC,eAAe,SAAS,IAAI,mBAAmB;AAAA,QAC/C,eAAe,SAAS,IAAI,KAAK;AAAA;AAAA,MACnC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AAGA,WAAS,mBAAmBA,aAAY;AACtC,QAAI,CAAC,KAAK,EAAE,SAASA,WAAU,GAAG;AAChC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA,QAChC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,QAAI,CAAC,cAAc,EAAE,SAASA,WAAU,GAAG;AACzC,aAAO,CAAC,gBAAgB,YAAY,IAAI,mBAAmB;AAAA,IAC7D;AAEA,QAAI,CAAC,WAAW,EAAE,SAASA,WAAU,GAAG;AACtC,aAAO;AAAA;AAAA,QAEL,gBAAgB,YAAY,UAAU;AAAA,QACtC,gBAAgB,YAAY,UAAU;AAAA,MACxC;AAAA,IACF;AAEA,QAAI,CAAC,SAAS,EAAE,SAASA,WAAU,GAAG;AACpC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA,QAEhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AACA,QAAI,CAAC,YAAY,EAAE,SAASA,WAAU,GAAG;AACvC,aAAO;AAAA,QACL,gBAAgB,YAAY,IAAI;AAAA;AAAA;AAAA,QAGhC,gBAAgB,YAAY,IAAI;AAAA,MAClC;AAAA,IACF;AAEA,UAAM,IAAI,MAAM,sCAAsC;AAAA,EACxD;AACF;",
   "names": ["args", "file", "searchType"]
 }

package/dist/match-detection/features/bib/f773.js

@@ -0,0 +1,82 @@
+import createDebugLogger from "debug";
+import { isComponentRecord } from "@natlibfi/melinda-commons";
+import { uniqArray } from "./issn.js";
+import { parse773g } from "../../../candidate-search/query-list/component.js";
+const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:features:issn");
+const debugData = debug.extend("data");
+const MAX_IDENTIFIER = 0.1;
+const MAX_G = 0.5;
+export default () => ({
+  name: "f773 ",
+  extract: ({
+    record
+    /*, recordExternal*/
+  }) => {
+    if (!isComponentRecord(record, false, [])) {
+      return [];
+    }
+    const f773s = record.get(/^773$/u);
+    const values = f773s.map((f) => f.subfields).flat().filter((sf) => ["w", "o", "x", "z"].includes(sf.code)).map((sf) => normalizeValue(sf.value));
+    const gData = parse773g(f773s[0]);
+    return [uniqArray(values), gData];
+    function normalizeValue(value) {
+      return value.replace(/(?:\. -|\.)$/u, "");
+    }
+  },
+  compare: (aa, bb) => {
+    const [aIdentifiers, ag] = aa;
+    const [bIdentifiers, bg] = bb;
+    const identifierScore = scoreIdentifiers();
+    const gScore = scoreG();
+    if (identifierScore === MAX_IDENTIFIER && gScore === MAX_G) {
+      return 1;
+    }
+    return identifierScore + gScore;
+    function scoreG() {
+      if (ag.number && ag.number === bg.number && ag.pages && ag.pages === bg.pages && ag.year && ag.year === bg.year) {
+        return MAX_G;
+      }
+      return scoreYear() + scoreNumber() + scorePages();
+    }
+    function scoreNumber() {
+      if (!ag.number || !bg.number) {
+        return 0;
+      }
+      if (ag.number === bg.number) {
+        return 0.05;
+      }
+      return -0.02;
+    }
+    function scorePages() {
+      if (!ag.pages || !bg.pages) {
+        return 0;
+      }
+      if (ag.pages === bg.pages) {
+        return 0.1;
+      }
+      return -0.05;
+    }
+    function scoreYear() {
+      if (!ag.year || !bg.year) {
+        return 0;
+      }
+      if (ag.year === bg.year) {
+        return 0.02;
+      }
+      return -0.02;
+    }
+    function scoreIdentifiers() {
+      debugData(`Comparing ISSN sets ${JSON.stringify(aIdentifiers)} and ${JSON.stringify(bIdentifiers)}`);
+      if (aIdentifiers.length === 0 || bIdentifiers.length === 0) {
+        return 0;
+      }
+      const firstSharedIdentifier = aIdentifiers.find((val) => bIdentifiers.includes(val));
+      if (firstSharedIdentifier) {
+        debug(`	 Shared identifier found: '${firstSharedIdentifier}'`);
+        return MAX_IDENTIFIER;
+      }
+      return -0.5;
+    }
+  }
+});
+//# sourceMappingURL=f773.js.map

package/dist/match-detection/features/bib/f773.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../../../src/match-detection/features/bib/f773.js"],
+  "sourcesContent": ["// NB! This checks identifiers ($w=ID, $x=ISSN, $z=ISBN and $o=others) to check whether the host is the same.\n// BB! $g is not checked *yet*. It could very strongly indicate that the records are the same... (I doubt $q could be useful here)\n// Rationale:\n// 773$w indicates sameness. However, non-Melinda records probably refer to non-melinda hosts.\n// Thus check other identifiers subfields ($x, $z and $o as well)\n//\n\n\nimport createDebugLogger from 'debug';\n\nimport {isComponentRecord} from '@natlibfi/melinda-commons';\nimport {uniqArray} from './issn.js';\nimport {parse773g} from '../../../candidate-search/query-list/component.js';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:issn');\nconst debugData = debug.extend('data');\n\nconst MAX_IDENTIFIER = 0.1; // This should be pretty low: it only says something about the host\nconst MAX_G = 0.5; // $g is about the comp itself, so score can be high here\n\nexport default () => ({\n  name: 'f773 ',\n  extract: ({record/*, recordExternal*/}) => {\n    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    if (!isComponentRecord(record, false, [])) {\n      return [];\n    }\n    const f773s = record.get(/^773$/u); // Identifier extract handles multiple 773...\n\n    // I think it's ok, if, say, $o and $x values match, so I'm not keeping the subfield codes\n    const values = f773s.map(f => f.subfields)\n      .flat()\n      .filter(sf => ['w', 'o', 'x', 'z'].includes(sf.code))\n      .map(sf =>  normalizeValue(sf.value));\n\n    // $g: only on $g is supported\n    const gData = parse773g(f773s[0]);\n\n    return [uniqArray(values), gData];\n\n    function normalizeValue(value) {\n      return value.replace(/(?:\\. -|\\.)$/u, '');\n    }\n\n\n  },\n\n  compare: (aa, bb) => {\n    const [aIdentifiers, ag] = aa;\n    const [bIdentifiers, bg] = bb;\n\n    const identifierScore = scoreIdentifiers();\n    const gScore = scoreG();\n\n    if (identifierScore === MAX_IDENTIFIER && gScore === MAX_G) { // Pretty impressive hit, even if title matches not\n      return 1.0;\n    }\n\n    return identifierScore + gScore;\n\n    function scoreG() {\n      // NB! $g contents are very noise, so wrong values may be extracted. Thus do not overpenalize.\n\n      // All exist match: things must be pretty good:\n      if (ag.number && ag.number === bg.number && ag.pages && ag.pages === bg.pages && ag.year && ag.year === bg.year) {\n        return MAX_G;\n      }\n      // Not comparing volume. It correlates with year.\n      return scoreYear() + scoreNumber() + scorePages();\n    }\n\n    function scoreNumber() {\n      if (!ag.number || !bg.number) {\n        return 0.0;\n      }\n      if (ag.number === bg.number) {\n        return 0.05;\n      }\n      return -0.02;\n    }\n\n    function scorePages() {\n      if (!ag.pages || !bg.pages) {\n        return 0.0;\n      }\n      if (ag.pages === bg.pages) { // If pages match, things must be pretty good\n        return 0.1;\n      }\n      return -0.05;\n    }\n\n    function scoreYear() { // publication-time.js also uses this, so don't score heavily here\n      if (!ag.year || !bg.year) {\n        return 0.0;\n      }\n      if (ag.year === bg.year) {\n        return 0.02;\n      }\n      return -0.02;\n    }\n\n    function scoreIdentifiers() {\n      debugData(`Comparing ISSN sets ${JSON.stringify(aIdentifiers)} and ${JSON.stringify(bIdentifiers)}`);\n      if (aIdentifiers.length === 0 || bIdentifiers.length === 0) {\n        // No data for decision\n        return 0;\n      }\n      const firstSharedIdentifier = aIdentifiers.find(val => bIdentifiers.includes(val));\n      if (firstSharedIdentifier) {\n        debug(`\\t Shared identifier found: '${firstSharedIdentifier}'`);\n        return MAX_IDENTIFIER;\n      }\n      return -0.5;\n    }\n  }\n});\n\n"],
+  "mappings": "AAQA,OAAO,uBAAuB;AAE9B,SAAQ,yBAAwB;AAChC,SAAQ,iBAAgB;AACxB,SAAQ,iBAAgB;AAExB,MAAM,QAAQ,kBAAkB,iEAAiE;AACjG,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,MAAM,iBAAiB;AACvB,MAAM,QAAQ;AAEd,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC;AAAA,IAAC;AAAA;AAAA,EAA0B,MAAM;AAGzC,QAAI,CAAC,kBAAkB,QAAQ,OAAO,CAAC,CAAC,GAAG;AACzC,aAAO,CAAC;AAAA,IACV;AACA,UAAM,QAAQ,OAAO,IAAI,QAAQ;AAGjC,UAAM,SAAS,MAAM,IAAI,OAAK,EAAE,SAAS,EACtC,KAAK,EACL,OAAO,QAAM,CAAC,KAAK,KAAK,KAAK,GAAG,EAAE,SAAS,GAAG,IAAI,CAAC,EACnD,IAAI,QAAO,eAAe,GAAG,KAAK,CAAC;AAGtC,UAAM,QAAQ,UAAU,MAAM,CAAC,CAAC;AAEhC,WAAO,CAAC,UAAU,MAAM,GAAG,KAAK;AAEhC,aAAS,eAAe,OAAO;AAC7B,aAAO,MAAM,QAAQ,iBAAiB,EAAE;AAAA,IAC1C;AAAA,EAGF;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,UAAM,CAAC,cAAc,EAAE,IAAI;AAC3B,UAAM,CAAC,cAAc,EAAE,IAAI;AAE3B,UAAM,kBAAkB,iBAAiB;AACzC,UAAM,SAAS,OAAO;AAEtB,QAAI,oBAAoB,kBAAkB,WAAW,OAAO;AAC1D,aAAO;AAAA,IACT;AAEA,WAAO,kBAAkB;AAEzB,aAAS,SAAS;AAIhB,UAAI,GAAG,UAAU,GAAG,WAAW,GAAG,UAAU,GAAG,SAAS,GAAG,UAAU,GAAG,SAAS,GAAG,QAAQ,GAAG,SAAS,GAAG,MAAM;AAC/G,eAAO;AAAA,MACT;AAEA,aAAO,UAAU,IAAI,YAAY,IAAI,WAAW;AAAA,IAClD;AAEA,aAAS,cAAc;AACrB,UAAI,CAAC,GAAG,UAAU,CAAC,GAAG,QAAQ;AAC5B,eAAO;AAAA,MACT;AACA,UAAI,GAAG,WAAW,GAAG,QAAQ;AAC3B,eAAO;AAAA,MACT;AACA,aAAO;AAAA,IACT;AAEA,aAAS,aAAa;AACpB,UAAI,CAAC,GAAG,SAAS,CAAC,GAAG,OAAO;AAC1B,eAAO;AAAA,MACT;AACA,UAAI,GAAG,UAAU,GAAG,OAAO;AACzB,eAAO;AAAA,MACT;AACA,aAAO;AAAA,IACT;AAEA,aAAS,YAAY;AACnB,UAAI,CAAC,GAAG,QAAQ,CAAC,GAAG,MAAM;AACxB,eAAO;AAAA,MACT;AACA,UAAI,GAAG,SAAS,GAAG,MAAM;AACvB,eAAO;AAAA,MACT;AACA,aAAO;AAAA,IACT;AAEA,aAAS,mBAAmB;AAC1B,gBAAU,uBAAuB,KAAK,UAAU,YAAY,CAAC,QAAQ,KAAK,UAAU,YAAY,CAAC,EAAE;AACnG,UAAI,aAAa,WAAW,KAAK,aAAa,WAAW,GAAG;AAE1D,eAAO;AAAA,MACT;AACA,YAAM,wBAAwB,aAAa,KAAK,SAAO,aAAa,SAAS,GAAG,CAAC;AACjF,UAAI,uBAAuB;AACzB,cAAM,+BAAgC,qBAAqB,GAAG;AAC9D,eAAO;AAAA,MACT;AACA,aAAO;AAAA,IACT;AAAA,EACF;AACF;",
+  "names": []
+}

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

@@ -3,7 +3,6 @@ export { default as isbn } from "./isbn.js";
 export { default as issn } from "./issn.js";
 export { default as otherStandardIdentifier } from "./other-standard-identifier.js";
 export { default as title } from "./title.js";
-export { default as titleVersionOriginal } from "./title-version-original.js";
 export { default as authors } from "./authors.js";
 export { default as recordType } from "./record-type.js";
 export { default as publicationTime } from "./publication-time.js";
@@ -14,4 +13,5 @@ export { default as bibliographicLevel } from "./bibliographic-level.js";
 export { default as melindaId } from "./melinda-id.js";
 export { default as allSourceIds } from "./all-source-ids.js";
 export { default as mediaType } from "./media-type.js";
+export { default as f773 } from "./f773.js";
 //# sourceMappingURL=index.js.map

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

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/index.js"],
-  "sourcesContent": ["\nexport {default as hostComponent} from './host-component.js';\nexport {default as isbn} from './isbn.js';\nexport {default as issn} from './issn.js';\nexport {default as otherStandardIdentifier} from './other-standard-identifier.js';\nexport {default as title} from './title.js';\nexport {default as titleVersionOriginal} from './title-version-original.js';\nexport {default as authors} from './authors.js';\nexport {default as recordType} from './record-type.js';\nexport {default as publicationTime} from './publication-time.js';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years.js';\nexport {default as publicationTimeAllowConsYearsMulti} from './publication-time-allow-cons-years-multi.js';\nexport {default as language} from './language.js';\nexport {default as bibliographicLevel} from './bibliographic-level.js';\nexport {default as melindaId} from './melinda-id.js';\nexport {default as allSourceIds} from './all-source-ids.js';\nexport {default as mediaType} from './media-type.js';\n"],
-  "mappings": "AACA,SAAQ,WAAW,qBAAoB;AACvC,SAAQ,WAAW,YAAW;AAC9B,SAAQ,WAAW,YAAW;AAC9B,SAAQ,WAAW,+BAA8B;AACjD,SAAQ,WAAW,aAAY;AAC/B,SAAQ,WAAW,4BAA2B;AAC9C,SAAQ,WAAW,eAAc;AACjC,SAAQ,WAAW,kBAAiB;AACpC,SAAQ,WAAW,uBAAsB;AACzC,SAAQ,WAAW,qCAAoC;AACvD,SAAQ,WAAW,0CAAyC;AAC5D,SAAQ,WAAW,gBAAe;AAClC,SAAQ,WAAW,0BAAyB;AAC5C,SAAQ,WAAW,iBAAgB;AACnC,SAAQ,WAAW,oBAAmB;AACtC,SAAQ,WAAW,iBAAgB;",
+  "sourcesContent": ["\nexport {default as hostComponent} from './host-component.js';\nexport {default as isbn} from './isbn.js';\nexport {default as issn} from './issn.js';\nexport {default as otherStandardIdentifier} from './other-standard-identifier.js';\nexport {default as title} from './title.js';\nexport {default as authors} from './authors.js';\nexport {default as recordType} from './record-type.js';\nexport {default as publicationTime} from './publication-time.js';\nexport {default as publicationTimeAllowConsYears} from './publication-time-allow-cons-years.js';\nexport {default as publicationTimeAllowConsYearsMulti} from './publication-time-allow-cons-years-multi.js';\nexport {default as language} from './language.js';\nexport {default as bibliographicLevel} from './bibliographic-level.js';\nexport {default as melindaId} from './melinda-id.js';\nexport {default as allSourceIds} from './all-source-ids.js';\nexport {default as mediaType} from './media-type.js';\nexport {default as f773} from './f773.js';\n"],
+  "mappings": "AACA,SAAQ,WAAW,qBAAoB;AACvC,SAAQ,WAAW,YAAW;AAC9B,SAAQ,WAAW,YAAW;AAC9B,SAAQ,WAAW,+BAA8B;AACjD,SAAQ,WAAW,aAAY;AAC/B,SAAQ,WAAW,eAAc;AACjC,SAAQ,WAAW,kBAAiB;AACpC,SAAQ,WAAW,uBAAsB;AACzC,SAAQ,WAAW,qCAAoC;AACvD,SAAQ,WAAW,0CAAyC;AAC5D,SAAQ,WAAW,gBAAe;AAClC,SAAQ,WAAW,0BAAyB;AAC5C,SAAQ,WAAW,iBAAgB;AACnC,SAAQ,WAAW,oBAAmB;AACtC,SAAQ,WAAW,iBAAgB;AACnC,SAAQ,WAAW,YAAW;",
   "names": []
 }

package/dist/match-detection/features/bib/index.test.js

@@ -30,7 +30,7 @@ describe("match-detection/features/bib/", () => {
       if (type === "compare") {
         const { featuresA, featuresB, expectedPoints } = expectations;
         const { compare } = features[feature](options);
-        assert.equal(compare(featuresA, featuresB), expectedPoints);
+        assert.equal(Math.round(compare(featuresA, featuresB) * 100) / 100, expectedPoints);
         return;
       }
       throw new Error(`Invalid type ${type}`);

package/dist/match-detection/features/bib/index.test.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/index.test.js"],
-  "sourcesContent": ["\nimport assert from 'node:assert';\nimport {describe} from 'node:test';\nimport createDebugLogger from 'debug';\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './index.js';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib:test');\nconst debugData = debug.extend('data');\n\n\ndescribe('match-detection/features/bib/', () => {\n  generateTests({\n    path: [import.meta.dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n    useMetadataFile: true,\n    fixura: {\n      reader: READERS.JSON\n    },\n\n    callback: ({enabled = true, feature, options, type, ...expectations}) => {\n\n      if (!enabled) {\n        return;\n      }\n\n      debug(`Testing: ${feature} ${type}`);\n\n      if (type === 'extract') {\n        const {expectedFeatures, inputRecord} = expectations;\n        const record = new MarcRecord(inputRecord, {subfieldValues: false});\n        debugData(`Record: ${record}`);\n        const {extract} = features[feature](options);\n\n        assert.deepStrictEqual(extract({record}), expectedFeatures);\n        return;\n      }\n\n      if (type === 'compare') {\n        const {featuresA, featuresB, expectedPoints} = expectations;\n        const {compare} = features[feature](options);\n\n        assert.equal(compare(featuresA, featuresB), expectedPoints);\n        return;\n      }\n\n      throw new Error(`Invalid type ${type}`);\n    }\n  });\n});\n"],
-  "mappings": "AACA,OAAO,YAAY;AACnB,SAAQ,gBAAe;AACvB,OAAO,uBAAuB;AAC9B,OAAO,mBAAmB;AAC1B,SAAQ,eAAc;AACtB,SAAQ,kBAAiB;AACzB,YAAY,cAAc;AAG1B,MAAM,QAAQ,kBAAkB,qEAAqE;AACrG,MAAM,YAAY,MAAM,OAAO,MAAM;AAGrC,SAAS,iCAAiC,MAAM;AAC9C,gBAAc;AAAA,IACZ,MAAM,CAAC,YAAY,SAAS,MAAM,MAAM,MAAM,MAAM,iBAAiB,mBAAmB,YAAY,KAAK;AAAA,IACzG,iBAAiB;AAAA,IACjB,QAAQ;AAAA,MACN,QAAQ,QAAQ;AAAA,IAClB;AAAA,IAEA,UAAU,CAAC,EAAC,UAAU,MAAM,SAAS,SAAS,MAAM,GAAG,aAAY,MAAM;AAEvE,UAAI,CAAC,SAAS;AACZ;AAAA,MACF;AAEA,YAAM,YAAY,OAAO,IAAI,IAAI,EAAE;AAEnC,UAAI,SAAS,WAAW;AACtB,cAAM,EAAC,kBAAkB,YAAW,IAAI;AACxC,cAAM,SAAS,IAAI,WAAW,aAAa,EAAC,gBAAgB,MAAK,CAAC;AAClE,kBAAU,WAAW,MAAM,EAAE;AAC7B,cAAM,EAAC,QAAO,IAAI,SAAS,OAAO,EAAE,OAAO;AAE3C,eAAO,gBAAgB,QAAQ,EAAC,OAAM,CAAC,GAAG,gBAAgB;AAC1D;AAAA,MACF;AAEA,UAAI,SAAS,WAAW;AACtB,cAAM,EAAC,WAAW,WAAW,eAAc,IAAI;AAC/C,cAAM,EAAC,QAAO,IAAI,SAAS,OAAO,EAAE,OAAO;AAE3C,eAAO,MAAM,QAAQ,WAAW,SAAS,GAAG,cAAc;AAC1D;AAAA,MACF;AAEA,YAAM,IAAI,MAAM,gBAAgB,IAAI,EAAE;AAAA,IACxC;AAAA,EACF,CAAC;AACH,CAAC;",
+  "sourcesContent": ["\nimport assert from 'node:assert';\nimport {describe} from 'node:test';\nimport createDebugLogger from 'debug';\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './index.js';\n\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib:test');\nconst debugData = debug.extend('data');\n\n\ndescribe('match-detection/features/bib/', () => {\n  generateTests({\n    path: [import.meta.dirname, '..', '..', '..', '..', 'test-fixtures', 'match-detection', 'features', 'bib'],\n    useMetadataFile: true,\n    fixura: {\n      reader: READERS.JSON\n    },\n\n    callback: ({enabled = true, feature, options, type, ...expectations}) => {\n\n      if (!enabled) {\n        return;\n      }\n\n      debug(`Testing: ${feature} ${type}`);\n\n      if (type === 'extract') {\n        const {expectedFeatures, inputRecord} = expectations;\n        const record = new MarcRecord(inputRecord, {subfieldValues: false});\n        debugData(`Record: ${record}`);\n        const {extract} = features[feature](options);\n\n        assert.deepStrictEqual(extract({record}), expectedFeatures);\n        return;\n      }\n\n      if (type === 'compare') {\n        const {featuresA, featuresB, expectedPoints} = expectations;\n        const {compare} = features[feature](options);\n\n        assert.equal(Math.round(compare(featuresA, featuresB) *100)/100, expectedPoints);\n        return;\n      }\n\n      throw new Error(`Invalid type ${type}`);\n    }\n  });\n});\n"],
+  "mappings": "AACA,OAAO,YAAY;AACnB,SAAQ,gBAAe;AACvB,OAAO,uBAAuB;AAC9B,OAAO,mBAAmB;AAC1B,SAAQ,eAAc;AACtB,SAAQ,kBAAiB;AACzB,YAAY,cAAc;AAG1B,MAAM,QAAQ,kBAAkB,qEAAqE;AACrG,MAAM,YAAY,MAAM,OAAO,MAAM;AAGrC,SAAS,iCAAiC,MAAM;AAC9C,gBAAc;AAAA,IACZ,MAAM,CAAC,YAAY,SAAS,MAAM,MAAM,MAAM,MAAM,iBAAiB,mBAAmB,YAAY,KAAK;AAAA,IACzG,iBAAiB;AAAA,IACjB,QAAQ;AAAA,MACN,QAAQ,QAAQ;AAAA,IAClB;AAAA,IAEA,UAAU,CAAC,EAAC,UAAU,MAAM,SAAS,SAAS,MAAM,GAAG,aAAY,MAAM;AAEvE,UAAI,CAAC,SAAS;AACZ;AAAA,MACF;AAEA,YAAM,YAAY,OAAO,IAAI,IAAI,EAAE;AAEnC,UAAI,SAAS,WAAW;AACtB,cAAM,EAAC,kBAAkB,YAAW,IAAI;AACxC,cAAM,SAAS,IAAI,WAAW,aAAa,EAAC,gBAAgB,MAAK,CAAC;AAClE,kBAAU,WAAW,MAAM,EAAE;AAC7B,cAAM,EAAC,QAAO,IAAI,SAAS,OAAO,EAAE,OAAO;AAE3C,eAAO,gBAAgB,QAAQ,EAAC,OAAM,CAAC,GAAG,gBAAgB;AAC1D;AAAA,MACF;AAEA,UAAI,SAAS,WAAW;AACtB,cAAM,EAAC,WAAW,WAAW,eAAc,IAAI;AAC/C,cAAM,EAAC,QAAO,IAAI,SAAS,OAAO,EAAE,OAAO;AAE3C,eAAO,MAAM,KAAK,MAAM,QAAQ,WAAW,SAAS,IAAG,GAAG,IAAE,KAAK,cAAc;AAC/E;AAAA,MACF;AAEA,YAAM,IAAI,MAAM,gBAAgB,IAAI,EAAE;AAAA,IACxC;AAAA,EACF,CAAC;AACH,CAAC;",
   "names": []
 }

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

@@ -1,7 +1,6 @@
 import createDebugLogger from "debug";
 import { parse as isbnParse } from "isbn3";
 import { uniqArray } from "./issn.js";
-import { isComponentRecord } from "@natlibfi/melinda-commons";
 const debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);
 const debugData = debug.extend("data");
 const MAX_SCORE = 0.75;
@@ -11,11 +10,7 @@ export default () => ({
     record
     /*, recordExternal*/
   }) => {
-    const host = !isComponentRecord(record, false, []);
-    if (host) {
-      return record.get("020").filter((f) => f.subfields?.some((sf) => ["a", "z"].includes(sf.code) && sf.value));
-    }
-    return record.get("773").filter((f) => f.subfields?.some((sf) => ["z"].includes(sf.code) && sf.value));
+    return record.get("020").filter((f) => f.subfields?.some((sf) => ["a", "z"].includes(sf.code) && sf.value));
   },
   // eslint-disable-next-line max-statements
   compare: (aa, bb) => {
@@ -23,7 +18,8 @@ export default () => ({
     if (aa.length === 0 || bb.length === 0) {
       return 0;
     }
-    const [subfieldCodeForGoodValues, subfieldCodeForBadValues] = getSubfieldCodes(aa[0].tag);
+    const subfieldCodeForGoodValues = "a";
+    const subfieldCodeForBadValues = "z";
     const [aValidValuesA, aInvalidValuesA, zValidValuesA, zInvalidValuesA] = getValuesWrapper(aa, "AA");
     const [aValidValuesB, aInvalidValuesB, zValidValuesB, zInvalidValuesB] = getValuesWrapper(bb, "BB");
     function getValuesWrapper(data, prefix) {
@@ -81,12 +77,6 @@ export default () => ({
       return 0;
     }
     return -0.75;
-    function getSubfieldCodes(tag) {
-      if (tag === "773") {
-        return ["z", void 0];
-      }
-      return ["a", "z"];
-    }
     function getValues(fields) {
       const goodValues = fields.flatMap((f) => f.subfields.filter((sf) => sf.code === subfieldCodeForGoodValues)).map((sf) => validatorAndNormalizer(sf.value));
       const trueGoodValues = goodValues.filter((val) => val.valid).map((val) => val.value);

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

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/isbn.js"],
-  "sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {parse as isbnParse} from 'isbn3';\n\nimport {uniqArray} from './issn.js';\nimport {isComponentRecord} from '@natlibfi/melinda-commons';\n\n\nconst debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);\nconst debugData = debug.extend('data');\n\nconst MAX_SCORE = 0.75;\n\nexport default () => ({\n  name: 'ISBN',\n  extract: ({record/*, recordExternal*/}) => {\n    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const host = !isComponentRecord(record, false, []);\n\n    if (host) {\n      return record.get('020').filter(f => f.subfields?.some(sf => ['a', 'z'].includes(sf.code) && sf.value));\n    }\n    return record.get('773').filter(f => f.subfields?.some(sf => ['z'].includes(sf.code) && sf.value));\n  },\n  // eslint-disable-next-line max-statements\n  compare: (aa, bb) => {\n    debugData(`Comparing ISBN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);\n    if (aa.length === 0 || bb.length === 0) {\n      // No data for decision\n      return 0;\n    }\n\n    const [subfieldCodeForGoodValues, subfieldCodeForBadValues] = getSubfieldCodes(aa[0].tag);\n\n    const [aValidValuesA, aInvalidValuesA, zValidValuesA, zInvalidValuesA] = getValuesWrapper(aa, 'AA'); // initial 'a' and 'z' refer to 020 subfields codes\n    const [aValidValuesB, aInvalidValuesB, zValidValuesB, zInvalidValuesB] = getValuesWrapper(bb, 'BB');\n\n    function getValuesWrapper(data, prefix) {\n      const [aValidValues, aInvalidValues, zValidValues, zInvalidValues] = getValues(data);\n      if (aValidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForGoodValues} VALID: ${aValidValues.join(', ')}`);\n      }\n      if (aInvalidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForGoodValues} INVALID: ${aInvalidValues.join(', ')}`);\n      }\n      if (zValidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForBadValues} VALID: ${zValidValues.join(', ')}`);\n      }\n      if (zInvalidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForBadValues} INVALID: ${zInvalidValues.join(', ')}`);\n      }\n      return [aValidValues, aInvalidValues, zValidValues, zInvalidValues];\n    }\n\n    const [sharedGoodValues, goodValuesAOnly, goodValuesBOnly] = getUnionData(aValidValuesA, aValidValuesB);\n\n    debug(`GOOD\\tBOTH: ${sharedGoodValues.length}, A only: ${goodValuesAOnly.length}, B only: ${goodValuesBOnly.length}`);\n\n    if (sharedGoodValues.length > 0) {\n      // Third argument (aka 'N times') is >= 0\n      return scoreData(MAX_SCORE, 0.8, goodValuesAOnly.length + goodValuesBOnly.length);\n    }\n\n    const hitScore = scoreSuboptimalHit();\n\n    function scoreSuboptimalHit() {\n      // One record consider ISBN good and the other record considered it's canceled:\n      if (aValidValuesA.some(valA => zValidValuesB.includes(valA)) || aValidValuesB.some(valB => zValidValuesA.includes(valB))) {\n        return MAX_SCORE;\n      }\n\n      // Subfield is for cancelled/whatever values, but the value is syntactically valid:\n      // Could happen for two canceled ISBNs for example. I'll give this two thirds of the full score\n      const zzValid = zValidValuesA.find(valA => zValidValuesB.includes(valA));\n      if (zzValid) {\n        debug(`Both contain a valid value in 020$z: ${zzValid}`);\n        return MAX_SCORE * 2 / 3;\n      }\n      // Shared invalid identifiers:\n      const aaInvalid = aInvalidValuesA.find(valA => aInvalidValuesB.includes(valA) || zInvalidValuesB.includes(valA)) || aInvalidValuesB.find(valB => zInvalidValuesA.includes(valB));\n      if (aaInvalid) {\n        debug(`Shared invalid value in 020$a and 020$a-or-$z subfields: ${aaInvalid}`);\n        return MAX_SCORE * 2 / 3;\n      }\n\n      /* // Currently I think that paired invalid idenfiers in 020$z are meaningless...\n      const zzInvalid = zInvalidValuesB.find(valB => zInvalidValuesA.includes(valB)) || zInvalidValuesA.find(valA => zInvalidValuesB.includes(valA));\n      if (zzInvalid) {\n        debug(`Shared invalid value in 020$z subfields: ${zzInvalid}`);\n        return MAX_SCORE / 3;\n      }\n      */\n\n      return 0;\n    }\n\n\n\n    if (hitScore === MAX_SCORE) {\n      // -1 is needed to make the third argument >= 0 (otherwise min val would be 0)\n      return scoreData(hitScore, 0.8, goodValuesAOnly.length + goodValuesBOnly.length - 1);\n    }\n\n    if (hitScore > 0) { // Canceled/invalid ISBNs match\n      // Note that this is not (currently) penalized for non-matching canceled/invalid ISBNs. Maybe it should be.\n      return scoreData(hitScore, 0.8, goodValuesAOnly.length + goodValuesBOnly.length);\n    }\n\n    // No match:\n\n    if (aValidValuesA.length + aInvalidValuesA.length === 0 || aValidValuesB.length + aInvalidValuesB.length === 0) { // At least one record did not have any good ISBNs, so not penalizing here! (Invalid 020$as are counted a bad.)\n      return 0.0;\n    }\n    // We have same matching invalid identifiers. Don't penalize:\n    const allA = [...aValidValuesA, ...aInvalidValuesA, ...zValidValuesA, ...zInvalidValuesA];\n    const allB = [...aValidValuesB, ...aInvalidValuesB, ...zValidValuesB, ...zInvalidValuesB];\n    const [sharedValues, tmp1, tmp2] = getUnionData(allA, allB);\n    debug(`WHATEVER\\tBOTH: ${sharedGoodValues.length}, A only: ${tmp1.length}, B only: ${tmp2.length}`);\n\n    if (sharedValues.length > 0) {\n      return 0;\n    }\n    // We have values but they disagree:\n    return -0.75; // Has good ISBNs on both records, but they did not match\n\n\n    function getSubfieldCodes(tag) {\n      if (tag === '773') {\n        return ['z', undefined];\n      }\n      return ['a', 'z'];\n    }\n\n    function getValues(fields) {\n      // Valid values are normalized to their isbn-13 form. Invalid values get their '-'s removed.\n      const goodValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForGoodValues)).map(sf => validatorAndNormalizer(sf.value));\n      const trueGoodValues = goodValues.filter(val => val.valid).map(val => val.value);\n      const wannabeGoodValues = goodValues.filter(val => !val.valid).map(val => val.value);\n      if (!subfieldCodeForBadValues) { // 773\n        return [trueGoodValues, wannabeGoodValues, [], []];\n      }\n      const badValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForBadValues)).map(sf => validatorAndNormalizer(sf.value));\n      const validBadValues = badValues.filter(val => val.valid).map(val => val.value);\n      const invalidBadValues = badValues.filter(val => !val.valid).map(val => val.value);\n      //const badValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForBadValues)).map(sf => validatorAndNormalizer(sf.value).value);\n      return [uniqArray(trueGoodValues), uniqArray(wannabeGoodValues), uniqArray(validBadValues), uniqArray(invalidBadValues)];\n    }\n\n    function validatorAndNormalizer(string) {\n      const string2 = string.replace(/\\. -$/u, ''); // Remove punctuation (773$z)\n\n      // Hack: Historically we 020$a \"1234567890 sidottu\" etc. Try the LHS alone:\n      const string3 = string2.replace(/ .*$/u, '');\n      if (string2 !== string3) {\n        const altResult = validatorAndNormalizer(string3);\n        if (altResult.valid) {\n          return altResult;\n        }\n      }\n\n      const isbnParseResult = isbnParse(string2, '') || '';\n      debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);\n      if (!isbnParseResult.isValid) {\n        debug(`Not parseable ISBN '${string2}', just removing hyphens`);\n        return {valid: false, value: string2.replace(/-/ug, '')};\n      }\n\n      debug(`Parseable ISBN '${string2}', normalizing to ISBN-13 '${isbnParseResult.isbn13}'`);\n      return {valid: true, value: isbnParseResult.isbn13};\n\n\n    }\n  }\n});\n\n// These are outside the default function as I'll probably want to export these later on\n\nfunction getUnionData(set1, set2) {\n  const shared = set1.filter(val => set2.includes(val));\n  const onlyInSet1 = set1.filter(val => !shared.includes(val));\n  const onlyInSet2 = set2.filter(val => !shared.includes(val));\n  return [shared, onlyInSet1, onlyInSet2];\n}\n\nfunction scoreData(score, factor, n) {\n  return innerScoreData(score, n);\n  function innerScoreData(currScore, remaining) {\n    if (remaining > 0) {\n      return innerScoreData(currScore * factor, remaining-1);\n    }\n    return Math.round(currScore * 100)/100; // 0.600000000001 => 0.6\n  }\n}"],
-  "mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,SAAS,iBAAgB;AAEjC,SAAQ,iBAAgB;AACxB,SAAQ,yBAAwB;AAGhC,MAAM,QAAQ,kBAAkB,sFAAsF;AACtH,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,MAAM,YAAY;AAElB,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC;AAAA,IAAC;AAAA;AAAA,EAA0B,MAAM;AAGzC,UAAM,OAAO,CAAC,kBAAkB,QAAQ,OAAO,CAAC,CAAC;AAEjD,QAAI,MAAM;AACR,aAAO,OAAO,IAAI,KAAK,EAAE,OAAO,OAAK,EAAE,WAAW,KAAK,QAAM,CAAC,KAAK,GAAG,EAAE,SAAS,GAAG,IAAI,KAAK,GAAG,KAAK,CAAC;AAAA,IACxG;AACA,WAAO,OAAO,IAAI,KAAK,EAAE,OAAO,OAAK,EAAE,WAAW,KAAK,QAAM,CAAC,GAAG,EAAE,SAAS,GAAG,IAAI,KAAK,GAAG,KAAK,CAAC;AAAA,EACnG;AAAA;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,cAAU,uBAAuB,KAAK,UAAU,EAAE,CAAC,QAAQ,KAAK,UAAU,EAAE,CAAC,EAAE;AAC/E,QAAI,GAAG,WAAW,KAAK,GAAG,WAAW,GAAG;AAEtC,aAAO;AAAA,IACT;AAEA,UAAM,CAAC,2BAA2B,wBAAwB,IAAI,iBAAiB,GAAG,CAAC,EAAE,GAAG;AAExF,UAAM,CAAC,eAAe,iBAAiB,eAAe,eAAe,IAAI,iBAAiB,IAAI,IAAI;AAClG,UAAM,CAAC,eAAe,iBAAiB,eAAe,eAAe,IAAI,iBAAiB,IAAI,IAAI;AAElG,aAAS,iBAAiB,MAAM,QAAQ;AACtC,YAAM,CAAC,cAAc,gBAAgB,cAAc,cAAc,IAAI,UAAU,IAAI;AACnF,UAAI,aAAa,QAAQ;AACvB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,yBAAyB,WAAW,aAAa,KAAK,IAAI,CAAC,EAAE;AAAA,MAChG;AACA,UAAI,eAAe,QAAQ;AACzB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,yBAAyB,aAAa,eAAe,KAAK,IAAI,CAAC,EAAE;AAAA,MACpG;AACA,UAAI,aAAa,QAAQ;AACvB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,wBAAwB,WAAW,aAAa,KAAK,IAAI,CAAC,EAAE;AAAA,MAC/F;AACA,UAAI,eAAe,QAAQ;AACzB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,wBAAwB,aAAa,eAAe,KAAK,IAAI,CAAC,EAAE;AAAA,MACnG;AACA,aAAO,CAAC,cAAc,gBAAgB,cAAc,cAAc;AAAA,IACpE;AAEA,UAAM,CAAC,kBAAkB,iBAAiB,eAAe,IAAI,aAAa,eAAe,aAAa;AAEtG,UAAM,cAAe,iBAAiB,MAAM,aAAa,gBAAgB,MAAM,aAAa,gBAAgB,MAAM,EAAE;AAEpH,QAAI,iBAAiB,SAAS,GAAG;AAE/B,aAAO,UAAU,WAAW,KAAK,gBAAgB,SAAS,gBAAgB,MAAM;AAAA,IAClF;AAEA,UAAM,WAAW,mBAAmB;AAEpC,aAAS,qBAAqB;AAE5B,UAAI,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC,KAAK,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC,GAAG;AACxH,eAAO;AAAA,MACT;AAIA,YAAM,UAAU,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC;AACvE,UAAI,SAAS;AACX,cAAM,wCAAwC,OAAO,EAAE;AACvD,eAAO,YAAY,IAAI;AAAA,MACzB;AAEA,YAAM,YAAY,gBAAgB,KAAK,UAAQ,gBAAgB,SAAS,IAAI,KAAK,gBAAgB,SAAS,IAAI,CAAC,KAAK,gBAAgB,KAAK,UAAQ,gBAAgB,SAAS,IAAI,CAAC;AAC/K,UAAI,WAAW;AACb,cAAM,4DAA4D,SAAS,EAAE;AAC7E,eAAO,YAAY,IAAI;AAAA,MACzB;AAUA,aAAO;AAAA,IACT;AAIA,QAAI,aAAa,WAAW;AAE1B,aAAO,UAAU,UAAU,KAAK,gBAAgB,SAAS,gBAAgB,SAAS,CAAC;AAAA,IACrF;AAEA,QAAI,WAAW,GAAG;AAEhB,aAAO,UAAU,UAAU,KAAK,gBAAgB,SAAS,gBAAgB,MAAM;AAAA,IACjF;AAIA,QAAI,cAAc,SAAS,gBAAgB,WAAW,KAAK,cAAc,SAAS,gBAAgB,WAAW,GAAG;AAC9G,aAAO;AAAA,IACT;AAEA,UAAM,OAAO,CAAC,GAAG,eAAe,GAAG,iBAAiB,GAAG,eAAe,GAAG,eAAe;AACxF,UAAM,OAAO,CAAC,GAAG,eAAe,GAAG,iBAAiB,GAAG,eAAe,GAAG,eAAe;AACxF,UAAM,CAAC,cAAc,MAAM,IAAI,IAAI,aAAa,MAAM,IAAI;AAC1D,UAAM,kBAAmB,iBAAiB,MAAM,aAAa,KAAK,MAAM,aAAa,KAAK,MAAM,EAAE;AAElG,QAAI,aAAa,SAAS,GAAG;AAC3B,aAAO;AAAA,IACT;AAEA,WAAO;AAGP,aAAS,iBAAiB,KAAK;AAC7B,UAAI,QAAQ,OAAO;AACjB,eAAO,CAAC,KAAK,MAAS;AAAA,MACxB;AACA,aAAO,CAAC,KAAK,GAAG;AAAA,IAClB;AAEA,aAAS,UAAU,QAAQ;AAEzB,YAAM,aAAa,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,GAAG,SAAS,yBAAyB,CAAC,EAAE,IAAI,QAAM,uBAAuB,GAAG,KAAK,CAAC;AAClJ,YAAM,iBAAiB,WAAW,OAAO,SAAO,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAC/E,YAAM,oBAAoB,WAAW,OAAO,SAAO,CAAC,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AACnF,UAAI,CAAC,0BAA0B;AAC7B,eAAO,CAAC,gBAAgB,mBAAmB,CAAC,GAAG,CAAC,CAAC;AAAA,MACnD;AACA,YAAM,YAAY,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,GAAG,SAAS,wBAAwB,CAAC,EAAE,IAAI,QAAM,uBAAuB,GAAG,KAAK,CAAC;AAChJ,YAAM,iBAAiB,UAAU,OAAO,SAAO,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAC9E,YAAM,mBAAmB,UAAU,OAAO,SAAO,CAAC,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAEjF,aAAO,CAAC,UAAU,cAAc,GAAG,UAAU,iBAAiB,GAAG,UAAU,cAAc,GAAG,UAAU,gBAAgB,CAAC;AAAA,IACzH;AAEA,aAAS,uBAAuB,QAAQ;AACtC,YAAM,UAAU,OAAO,QAAQ,UAAU,EAAE;AAG3C,YAAM,UAAU,QAAQ,QAAQ,SAAS,EAAE;AAC3C,UAAI,YAAY,SAAS;AACvB,cAAM,YAAY,uBAAuB,OAAO;AAChD,YAAI,UAAU,OAAO;AACnB,iBAAO;AAAA,QACT;AAAA,MACF;AAEA,YAAM,kBAAkB,UAAU,SAAS,EAAE,KAAK;AAClD,gBAAU,oBAAoB,KAAK,UAAU,eAAe,CAAC,EAAE;AAC/D,UAAI,CAAC,gBAAgB,SAAS;AAC5B,cAAM,uBAAuB,OAAO,0BAA0B;AAC9D,eAAO,EAAC,OAAO,OAAO,OAAO,QAAQ,QAAQ,OAAO,EAAE,EAAC;AAAA,MACzD;AAEA,YAAM,mBAAmB,OAAO,8BAA8B,gBAAgB,MAAM,GAAG;AACvF,aAAO,EAAC,OAAO,MAAM,OAAO,gBAAgB,OAAM;AAAA,IAGpD;AAAA,EACF;AACF;AAIA,SAAS,aAAa,MAAM,MAAM;AAChC,QAAM,SAAS,KAAK,OAAO,SAAO,KAAK,SAAS,GAAG,CAAC;AACpD,QAAM,aAAa,KAAK,OAAO,SAAO,CAAC,OAAO,SAAS,GAAG,CAAC;AAC3D,QAAM,aAAa,KAAK,OAAO,SAAO,CAAC,OAAO,SAAS,GAAG,CAAC;AAC3D,SAAO,CAAC,QAAQ,YAAY,UAAU;AACxC;AAEA,SAAS,UAAU,OAAO,QAAQ,GAAG;AACnC,SAAO,eAAe,OAAO,CAAC;AAC9B,WAAS,eAAe,WAAW,WAAW;AAC5C,QAAI,YAAY,GAAG;AACjB,aAAO,eAAe,YAAY,QAAQ,YAAU,CAAC;AAAA,IACvD;AACA,WAAO,KAAK,MAAM,YAAY,GAAG,IAAE;AAAA,EACrC;AACF;",
+  "sourcesContent": ["\nimport createDebugLogger from 'debug';\nimport {parse as isbnParse} from 'isbn3';\n\nimport {uniqArray} from './issn.js';\n\nconst debug = createDebugLogger(`@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers:ISBN`);\nconst debugData = debug.extend('data');\n\nconst MAX_SCORE = 0.75;\n\nexport default () => ({\n  name: 'ISBN',\n  extract: ({record/*, recordExternal*/}) => {\n    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    return record.get('020').filter(f => f.subfields?.some(sf => ['a', 'z'].includes(sf.code) && sf.value));\n  },\n  // eslint-disable-next-line max-statements\n  compare: (aa, bb) => {\n    debugData(`Comparing ISBN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);\n    if (aa.length === 0 || bb.length === 0) {\n      // No data for decision\n      return 0;\n    }\n\n    const subfieldCodeForGoodValues = 'a';\n    const subfieldCodeForBadValues = 'z';\n\n    const [aValidValuesA, aInvalidValuesA, zValidValuesA, zInvalidValuesA] = getValuesWrapper(aa, 'AA'); // initial 'a' and 'z' refer to 020 subfields codes\n    const [aValidValuesB, aInvalidValuesB, zValidValuesB, zInvalidValuesB] = getValuesWrapper(bb, 'BB');\n\n    function getValuesWrapper(data, prefix) {\n      const [aValidValues, aInvalidValues, zValidValues, zInvalidValues] = getValues(data);\n      if (aValidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForGoodValues} VALID: ${aValidValues.join(', ')}`);\n      }\n      if (aInvalidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForGoodValues} INVALID: ${aInvalidValues.join(', ')}`);\n      }\n      if (zValidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForBadValues} VALID: ${zValidValues.join(', ')}`);\n      }\n      if (zInvalidValues.length) {\n        debug(`${prefix}: ${aa[0].tag}$${subfieldCodeForBadValues} INVALID: ${zInvalidValues.join(', ')}`);\n      }\n      return [aValidValues, aInvalidValues, zValidValues, zInvalidValues];\n    }\n\n    const [sharedGoodValues, goodValuesAOnly, goodValuesBOnly] = getUnionData(aValidValuesA, aValidValuesB);\n\n    debug(`GOOD\\tBOTH: ${sharedGoodValues.length}, A only: ${goodValuesAOnly.length}, B only: ${goodValuesBOnly.length}`);\n\n    if (sharedGoodValues.length > 0) {\n      // Third argument (aka 'N times') is >= 0\n      return scoreData(MAX_SCORE, 0.8, goodValuesAOnly.length + goodValuesBOnly.length);\n    }\n\n    const hitScore = scoreSuboptimalHit();\n\n    function scoreSuboptimalHit() {\n      // One record consider ISBN good and the other record considered it's canceled:\n      if (aValidValuesA.some(valA => zValidValuesB.includes(valA)) || aValidValuesB.some(valB => zValidValuesA.includes(valB))) {\n        return MAX_SCORE;\n      }\n\n      // Subfield is for cancelled/whatever values, but the value is syntactically valid:\n      // Could happen for two canceled ISBNs for example. I'll give this two thirds of the full score\n      const zzValid = zValidValuesA.find(valA => zValidValuesB.includes(valA));\n      if (zzValid) {\n        debug(`Both contain a valid value in 020$z: ${zzValid}`);\n        return MAX_SCORE * 2 / 3;\n      }\n      // Shared invalid identifiers:\n      const aaInvalid = aInvalidValuesA.find(valA => aInvalidValuesB.includes(valA) || zInvalidValuesB.includes(valA)) || aInvalidValuesB.find(valB => zInvalidValuesA.includes(valB));\n      if (aaInvalid) {\n        debug(`Shared invalid value in 020$a and 020$a-or-$z subfields: ${aaInvalid}`);\n        return MAX_SCORE * 2 / 3;\n      }\n\n      /* // Currently I think that paired invalid idenfiers in 020$z are meaningless...\n      const zzInvalid = zInvalidValuesB.find(valB => zInvalidValuesA.includes(valB)) || zInvalidValuesA.find(valA => zInvalidValuesB.includes(valA));\n      if (zzInvalid) {\n        debug(`Shared invalid value in 020$z subfields: ${zzInvalid}`);\n        return MAX_SCORE / 3;\n      }\n      */\n\n      return 0;\n    }\n\n\n\n    if (hitScore === MAX_SCORE) {\n      // -1 is needed to make the third argument >= 0 (otherwise min val would be 0)\n      return scoreData(hitScore, 0.8, goodValuesAOnly.length + goodValuesBOnly.length - 1);\n    }\n\n    if (hitScore > 0) { // Canceled/invalid ISBNs match\n      // Note that this is not (currently) penalized for non-matching canceled/invalid ISBNs. Maybe it should be.\n      return scoreData(hitScore, 0.8, goodValuesAOnly.length + goodValuesBOnly.length);\n    }\n\n    // No match:\n\n    if (aValidValuesA.length + aInvalidValuesA.length === 0 || aValidValuesB.length + aInvalidValuesB.length === 0) { // At least one record did not have any good ISBNs, so not penalizing here! (Invalid 020$as are counted a bad.)\n      return 0.0;\n    }\n    // We have same matching invalid identifiers. Don't penalize:\n    const allA = [...aValidValuesA, ...aInvalidValuesA, ...zValidValuesA, ...zInvalidValuesA];\n    const allB = [...aValidValuesB, ...aInvalidValuesB, ...zValidValuesB, ...zInvalidValuesB];\n    const [sharedValues, tmp1, tmp2] = getUnionData(allA, allB);\n    debug(`WHATEVER\\tBOTH: ${sharedGoodValues.length}, A only: ${tmp1.length}, B only: ${tmp2.length}`);\n\n    if (sharedValues.length > 0) {\n      return 0;\n    }\n    // We have values but they disagree:\n    return -0.75; // Has good ISBNs on both records, but they did not match\n\n    function getValues(fields) {\n      // Valid values are normalized to their isbn-13 form. Invalid values get their '-'s removed.\n      const goodValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForGoodValues)).map(sf => validatorAndNormalizer(sf.value));\n      const trueGoodValues = goodValues.filter(val => val.valid).map(val => val.value);\n      const wannabeGoodValues = goodValues.filter(val => !val.valid).map(val => val.value);\n      if (!subfieldCodeForBadValues) { // 773\n        return [trueGoodValues, wannabeGoodValues, [], []];\n      }\n      const badValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForBadValues)).map(sf => validatorAndNormalizer(sf.value));\n      const validBadValues = badValues.filter(val => val.valid).map(val => val.value);\n      const invalidBadValues = badValues.filter(val => !val.valid).map(val => val.value);\n      //const badValues = fields.flatMap(f => f.subfields.filter(sf => sf.code === subfieldCodeForBadValues)).map(sf => validatorAndNormalizer(sf.value).value);\n      return [uniqArray(trueGoodValues), uniqArray(wannabeGoodValues), uniqArray(validBadValues), uniqArray(invalidBadValues)];\n    }\n\n    function validatorAndNormalizer(string) {\n      const string2 = string.replace(/\\. -$/u, ''); // Remove punctuation (773$z)\n\n      // Hack: Historically we 020$a \"1234567890 sidottu\" etc. Try the LHS alone:\n      const string3 = string2.replace(/ .*$/u, '');\n      if (string2 !== string3) {\n        const altResult = validatorAndNormalizer(string3);\n        if (altResult.valid) {\n          return altResult;\n        }\n      }\n\n      const isbnParseResult = isbnParse(string2, '') || '';\n      debugData(`isbnParseResult: ${JSON.stringify(isbnParseResult)}`);\n      if (!isbnParseResult.isValid) {\n        debug(`Not parseable ISBN '${string2}', just removing hyphens`);\n        return {valid: false, value: string2.replace(/-/ug, '')};\n      }\n\n      debug(`Parseable ISBN '${string2}', normalizing to ISBN-13 '${isbnParseResult.isbn13}'`);\n      return {valid: true, value: isbnParseResult.isbn13};\n\n\n    }\n  }\n});\n\n// These are outside the default function as I'll probably want to export these later on\n\nfunction getUnionData(set1, set2) {\n  const shared = set1.filter(val => set2.includes(val));\n  const onlyInSet1 = set1.filter(val => !shared.includes(val));\n  const onlyInSet2 = set2.filter(val => !shared.includes(val));\n  return [shared, onlyInSet1, onlyInSet2];\n}\n\nfunction scoreData(score, factor, n) {\n  return innerScoreData(score, n);\n  function innerScoreData(currScore, remaining) {\n    if (remaining > 0) {\n      return innerScoreData(currScore * factor, remaining-1);\n    }\n    return Math.round(currScore * 100)/100; // 0.600000000001 => 0.6\n  }\n}"],
+  "mappings": "AACA,OAAO,uBAAuB;AAC9B,SAAQ,SAAS,iBAAgB;AAEjC,SAAQ,iBAAgB;AAExB,MAAM,QAAQ,kBAAkB,sFAAsF;AACtH,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,MAAM,YAAY;AAElB,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC;AAAA,IAAC;AAAA;AAAA,EAA0B,MAAM;AAGzC,WAAO,OAAO,IAAI,KAAK,EAAE,OAAO,OAAK,EAAE,WAAW,KAAK,QAAM,CAAC,KAAK,GAAG,EAAE,SAAS,GAAG,IAAI,KAAK,GAAG,KAAK,CAAC;AAAA,EACxG;AAAA;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,cAAU,uBAAuB,KAAK,UAAU,EAAE,CAAC,QAAQ,KAAK,UAAU,EAAE,CAAC,EAAE;AAC/E,QAAI,GAAG,WAAW,KAAK,GAAG,WAAW,GAAG;AAEtC,aAAO;AAAA,IACT;AAEA,UAAM,4BAA4B;AAClC,UAAM,2BAA2B;AAEjC,UAAM,CAAC,eAAe,iBAAiB,eAAe,eAAe,IAAI,iBAAiB,IAAI,IAAI;AAClG,UAAM,CAAC,eAAe,iBAAiB,eAAe,eAAe,IAAI,iBAAiB,IAAI,IAAI;AAElG,aAAS,iBAAiB,MAAM,QAAQ;AACtC,YAAM,CAAC,cAAc,gBAAgB,cAAc,cAAc,IAAI,UAAU,IAAI;AACnF,UAAI,aAAa,QAAQ;AACvB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,yBAAyB,WAAW,aAAa,KAAK,IAAI,CAAC,EAAE;AAAA,MAChG;AACA,UAAI,eAAe,QAAQ;AACzB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,yBAAyB,aAAa,eAAe,KAAK,IAAI,CAAC,EAAE;AAAA,MACpG;AACA,UAAI,aAAa,QAAQ;AACvB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,wBAAwB,WAAW,aAAa,KAAK,IAAI,CAAC,EAAE;AAAA,MAC/F;AACA,UAAI,eAAe,QAAQ;AACzB,cAAM,GAAG,MAAM,KAAK,GAAG,CAAC,EAAE,GAAG,IAAI,wBAAwB,aAAa,eAAe,KAAK,IAAI,CAAC,EAAE;AAAA,MACnG;AACA,aAAO,CAAC,cAAc,gBAAgB,cAAc,cAAc;AAAA,IACpE;AAEA,UAAM,CAAC,kBAAkB,iBAAiB,eAAe,IAAI,aAAa,eAAe,aAAa;AAEtG,UAAM,cAAe,iBAAiB,MAAM,aAAa,gBAAgB,MAAM,aAAa,gBAAgB,MAAM,EAAE;AAEpH,QAAI,iBAAiB,SAAS,GAAG;AAE/B,aAAO,UAAU,WAAW,KAAK,gBAAgB,SAAS,gBAAgB,MAAM;AAAA,IAClF;AAEA,UAAM,WAAW,mBAAmB;AAEpC,aAAS,qBAAqB;AAE5B,UAAI,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC,KAAK,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC,GAAG;AACxH,eAAO;AAAA,MACT;AAIA,YAAM,UAAU,cAAc,KAAK,UAAQ,cAAc,SAAS,IAAI,CAAC;AACvE,UAAI,SAAS;AACX,cAAM,wCAAwC,OAAO,EAAE;AACvD,eAAO,YAAY,IAAI;AAAA,MACzB;AAEA,YAAM,YAAY,gBAAgB,KAAK,UAAQ,gBAAgB,SAAS,IAAI,KAAK,gBAAgB,SAAS,IAAI,CAAC,KAAK,gBAAgB,KAAK,UAAQ,gBAAgB,SAAS,IAAI,CAAC;AAC/K,UAAI,WAAW;AACb,cAAM,4DAA4D,SAAS,EAAE;AAC7E,eAAO,YAAY,IAAI;AAAA,MACzB;AAUA,aAAO;AAAA,IACT;AAIA,QAAI,aAAa,WAAW;AAE1B,aAAO,UAAU,UAAU,KAAK,gBAAgB,SAAS,gBAAgB,SAAS,CAAC;AAAA,IACrF;AAEA,QAAI,WAAW,GAAG;AAEhB,aAAO,UAAU,UAAU,KAAK,gBAAgB,SAAS,gBAAgB,MAAM;AAAA,IACjF;AAIA,QAAI,cAAc,SAAS,gBAAgB,WAAW,KAAK,cAAc,SAAS,gBAAgB,WAAW,GAAG;AAC9G,aAAO;AAAA,IACT;AAEA,UAAM,OAAO,CAAC,GAAG,eAAe,GAAG,iBAAiB,GAAG,eAAe,GAAG,eAAe;AACxF,UAAM,OAAO,CAAC,GAAG,eAAe,GAAG,iBAAiB,GAAG,eAAe,GAAG,eAAe;AACxF,UAAM,CAAC,cAAc,MAAM,IAAI,IAAI,aAAa,MAAM,IAAI;AAC1D,UAAM,kBAAmB,iBAAiB,MAAM,aAAa,KAAK,MAAM,aAAa,KAAK,MAAM,EAAE;AAElG,QAAI,aAAa,SAAS,GAAG;AAC3B,aAAO;AAAA,IACT;AAEA,WAAO;AAEP,aAAS,UAAU,QAAQ;AAEzB,YAAM,aAAa,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,GAAG,SAAS,yBAAyB,CAAC,EAAE,IAAI,QAAM,uBAAuB,GAAG,KAAK,CAAC;AAClJ,YAAM,iBAAiB,WAAW,OAAO,SAAO,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAC/E,YAAM,oBAAoB,WAAW,OAAO,SAAO,CAAC,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AACnF,UAAI,CAAC,0BAA0B;AAC7B,eAAO,CAAC,gBAAgB,mBAAmB,CAAC,GAAG,CAAC,CAAC;AAAA,MACnD;AACA,YAAM,YAAY,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,GAAG,SAAS,wBAAwB,CAAC,EAAE,IAAI,QAAM,uBAAuB,GAAG,KAAK,CAAC;AAChJ,YAAM,iBAAiB,UAAU,OAAO,SAAO,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAC9E,YAAM,mBAAmB,UAAU,OAAO,SAAO,CAAC,IAAI,KAAK,EAAE,IAAI,SAAO,IAAI,KAAK;AAEjF,aAAO,CAAC,UAAU,cAAc,GAAG,UAAU,iBAAiB,GAAG,UAAU,cAAc,GAAG,UAAU,gBAAgB,CAAC;AAAA,IACzH;AAEA,aAAS,uBAAuB,QAAQ;AACtC,YAAM,UAAU,OAAO,QAAQ,UAAU,EAAE;AAG3C,YAAM,UAAU,QAAQ,QAAQ,SAAS,EAAE;AAC3C,UAAI,YAAY,SAAS;AACvB,cAAM,YAAY,uBAAuB,OAAO;AAChD,YAAI,UAAU,OAAO;AACnB,iBAAO;AAAA,QACT;AAAA,MACF;AAEA,YAAM,kBAAkB,UAAU,SAAS,EAAE,KAAK;AAClD,gBAAU,oBAAoB,KAAK,UAAU,eAAe,CAAC,EAAE;AAC/D,UAAI,CAAC,gBAAgB,SAAS;AAC5B,cAAM,uBAAuB,OAAO,0BAA0B;AAC9D,eAAO,EAAC,OAAO,OAAO,OAAO,QAAQ,QAAQ,OAAO,EAAE,EAAC;AAAA,MACzD;AAEA,YAAM,mBAAmB,OAAO,8BAA8B,gBAAgB,MAAM,GAAG;AACvF,aAAO,EAAC,OAAO,MAAM,OAAO,gBAAgB,OAAM;AAAA,IAGpD;AAAA,EACF;AACF;AAIA,SAAS,aAAa,MAAM,MAAM;AAChC,QAAM,SAAS,KAAK,OAAO,SAAO,KAAK,SAAS,GAAG,CAAC;AACpD,QAAM,aAAa,KAAK,OAAO,SAAO,CAAC,OAAO,SAAS,GAAG,CAAC;AAC3D,QAAM,aAAa,KAAK,OAAO,SAAO,CAAC,OAAO,SAAS,GAAG,CAAC;AAC3D,SAAO,CAAC,QAAQ,YAAY,UAAU;AACxC;AAEA,SAAS,UAAU,OAAO,QAAQ,GAAG;AACnC,SAAO,eAAe,OAAO,CAAC;AAC9B,WAAS,eAAe,WAAW,WAAW;AAC5C,QAAI,YAAY,GAAG;AACjB,aAAO,eAAe,YAAY,QAAQ,YAAU,CAAC;AAAA,IACvD;AACA,WAAO,KAAK,MAAM,YAAY,GAAG,IAAE;AAAA,EACrC;AACF;",
   "names": []
 }

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

@@ -1,5 +1,4 @@
 import createDebugLogger from "debug";
-import { isComponentRecord } from "@natlibfi/melinda-commons";
 const debug = createDebugLogger("@natlibfi/melinda-record-matching:match-detection:features:issn");
 const debugData = debug.extend("data");
 export default () => ({
@@ -8,15 +7,14 @@ export default () => ({
     record
     /*, recordExternal*/
   }) => {
-    const isHost = !isComponentRecord(record, false, []);
     return getIssns();
     function getIssns() {
-      const fields = getRelevantFields();
+      const fields = record.get(/^022$/u);
       if (fields.length === 0) {
         return [];
       }
       debug(`	 ${fields.length} potential ISSN fields (${fields[0].tag})`);
-      const subfieldCodes = getRelevantSubfieldCodes();
+      const subfieldCodes = ["a", "y", "z"];
       const subfieldValues = fields.flatMap((f) => f.subfields.filter((sf) => subfieldCodes.includes(sf.code)).map((sf) => sf.value));
       const validSubfieldValues = subfieldValues?.map((val) => normalizeSubfieldValue(val)).filter((val) => isValidIssn(val));
       if (!validSubfieldValues) {
@@ -24,24 +22,12 @@ export default () => ({
       }
       return uniqArray(validSubfieldValues);
     }
-    function getRelevantFields() {
-      if (isHost) {
-        return record.get(/^022$/u);
-      }
-      return record.get(/^[79]73$/u);
-    }
     function normalizeSubfieldValue(val) {
       return val.replace(/[., :;].*$/u, "");
     }
     function isValidIssn(val) {
       return /^[0-9][0-9][0-9][0-9]-[0-9][0-9][0-9][0-9X]$/u.test(val);
     }
-    function getRelevantSubfieldCodes() {
-      if (isHost) {
-        return ["a", "y", "z"];
-      }
-      return ["x"];
-    }
   },
   // eslint-disable-next-line max-statements
   compare: (aa, bb) => {

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

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/issn.js"],
-  "sourcesContent": ["/*\n//import createInterface from './standard-identifier-factory.js';\n\n\nexport default () => {\n  const {extract, compare} = createInterface({pattern: /^022$/u, subfieldCodes: ['a', 'z', 'y']});\n  return {extract, compare, name: 'ISSN'};\n};\n*/\n\n\nimport createDebugLogger from 'debug';\n\nimport {isComponentRecord} from '@natlibfi/melinda-commons';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:issn');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'ISSN',\n  extract: ({record/*, recordExternal*/}) => {\n    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    const isHost = !isComponentRecord(record, false, []);\n\n    // debug(`\\t Is HOST: '${isHost}'`);\n    return getIssns();\n\n    function getIssns() {\n      const fields = getRelevantFields();\n      if (fields.length === 0) {\n        return [];\n      }\n      debug(`\\t ${fields.length} potential ISSN fields (${fields[0].tag})`);\n      const subfieldCodes = getRelevantSubfieldCodes();\n      //debug(`\\t subfield codes: '${subfieldCodes.join(\"', '\")}'`);\n      const subfieldValues = fields.flatMap(f => f.subfields.filter(sf => subfieldCodes.includes(sf.code)).map(sf => sf.value));\n      //debug(`\\t cand values: '${subfieldValues.join(\"', '\")}'`);\n      // Stripping punctuaction with substring here is pretty quick and dirty approach...\n      const validSubfieldValues = subfieldValues?.map(val => normalizeSubfieldValue(val)).filter(val => isValidIssn(val));\n      if (!validSubfieldValues) {\n        return [];\n      }\n      return uniqArray(validSubfieldValues);\n    }\n\n    function getRelevantFields() {\n      if (isHost) {\n        return record.get(/^022$/u);\n      }\n      return record.get(/^[79]73$/u);\n    }\n\n    function normalizeSubfieldValue(val) {\n      return val.replace(/[., :;].*$/u, '');\n    }\n\n    function isValidIssn(val) {\n      return /^[0-9][0-9][0-9][0-9]-[0-9][0-9][0-9][0-9X]$/u.test(val);\n    }\n\n    function getRelevantSubfieldCodes() {\n      if (isHost) {\n        return ['a', 'y', 'z'];\n      }\n      return ['x'];\n    }\n  },\n  // eslint-disable-next-line max-statements\n  compare: (aa, bb) => {\n    debugData(`Comparing ISSN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);\n    if (aa.length === 0 || bb.length === 0) {\n      // No data for decision\n      return 0;\n    }\n    const firstSharedIssn = aa.find(val => bb.includes(val));\n    if (firstSharedIssn) {\n      debug(`\\t Shared ISSN found: '${firstSharedIssn}'`);\n      // Maybe less for comps?\n      return 0.2;\n    }\n    return -0.2;\n\n  }\n});\n\nexport function uniqArray(arr) {\n  return arr.filter((val, i) => arr.indexOf(val) === i);\n}"],
-  "mappings": "AAWA,OAAO,uBAAuB;AAE9B,SAAQ,yBAAwB;AAEhC,MAAM,QAAQ,kBAAkB,iEAAiE;AACjG,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC;AAAA,IAAC;AAAA;AAAA,EAA0B,MAAM;AAGzC,UAAM,SAAS,CAAC,kBAAkB,QAAQ,OAAO,CAAC,CAAC;AAGnD,WAAO,SAAS;AAEhB,aAAS,WAAW;AAClB,YAAM,SAAS,kBAAkB;AACjC,UAAI,OAAO,WAAW,GAAG;AACvB,eAAO,CAAC;AAAA,MACV;AACA,YAAM,KAAM,OAAO,MAAM,2BAA2B,OAAO,CAAC,EAAE,GAAG,GAAG;AACpE,YAAM,gBAAgB,yBAAyB;AAE/C,YAAM,iBAAiB,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,cAAc,SAAS,GAAG,IAAI,CAAC,EAAE,IAAI,QAAM,GAAG,KAAK,CAAC;AAGxH,YAAM,sBAAsB,gBAAgB,IAAI,SAAO,uBAAuB,GAAG,CAAC,EAAE,OAAO,SAAO,YAAY,GAAG,CAAC;AAClH,UAAI,CAAC,qBAAqB;AACxB,eAAO,CAAC;AAAA,MACV;AACA,aAAO,UAAU,mBAAmB;AAAA,IACtC;AAEA,aAAS,oBAAoB;AAC3B,UAAI,QAAQ;AACV,eAAO,OAAO,IAAI,QAAQ;AAAA,MAC5B;AACA,aAAO,OAAO,IAAI,WAAW;AAAA,IAC/B;AAEA,aAAS,uBAAuB,KAAK;AACnC,aAAO,IAAI,QAAQ,eAAe,EAAE;AAAA,IACtC;AAEA,aAAS,YAAY,KAAK;AACxB,aAAO,gDAAgD,KAAK,GAAG;AAAA,IACjE;AAEA,aAAS,2BAA2B;AAClC,UAAI,QAAQ;AACV,eAAO,CAAC,KAAK,KAAK,GAAG;AAAA,MACvB;AACA,aAAO,CAAC,GAAG;AAAA,IACb;AAAA,EACF;AAAA;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,cAAU,uBAAuB,KAAK,UAAU,EAAE,CAAC,QAAQ,KAAK,UAAU,EAAE,CAAC,EAAE;AAC/E,QAAI,GAAG,WAAW,KAAK,GAAG,WAAW,GAAG;AAEtC,aAAO;AAAA,IACT;AACA,UAAM,kBAAkB,GAAG,KAAK,SAAO,GAAG,SAAS,GAAG,CAAC;AACvD,QAAI,iBAAiB;AACnB,YAAM,yBAA0B,eAAe,GAAG;AAElD,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EAET;AACF;AAEO,gBAAS,UAAU,KAAK;AAC7B,SAAO,IAAI,OAAO,CAAC,KAAK,MAAM,IAAI,QAAQ,GAAG,MAAM,CAAC;AACtD;",
+  "sourcesContent": ["import createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:issn');\nconst debugData = debug.extend('data');\n\nexport default () => ({\n  name: 'ISSN',\n  extract: ({record/*, recordExternal*/}) => {\n    //const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';\n\n    return getIssns();\n\n    function getIssns() {\n      const fields = record.get(/^022$/u);\n      if (fields.length === 0) {\n        return [];\n      }\n\n      debug(`\\t ${fields.length} potential ISSN fields (${fields[0].tag})`);\n      const subfieldCodes = ['a', 'y', 'z']\n      //debug(`\\t subfield codes: '${subfieldCodes.join(\"', '\")}'`);\n      const subfieldValues = fields.flatMap(f => f.subfields.filter(sf => subfieldCodes.includes(sf.code)).map(sf => sf.value));\n      //debug(`\\t cand values: '${subfieldValues.join(\"', '\")}'`);\n      // Stripping punctuaction with substring here is pretty quick and dirty approach...\n      const validSubfieldValues = subfieldValues?.map(val => normalizeSubfieldValue(val)).filter(val => isValidIssn(val));\n      if (!validSubfieldValues) {\n        return [];\n      }\n      return uniqArray(validSubfieldValues);\n    }\n\n    function normalizeSubfieldValue(val) {\n      return val.replace(/[., :;].*$/u, '');\n    }\n\n    function isValidIssn(val) {\n      return /^[0-9][0-9][0-9][0-9]-[0-9][0-9][0-9][0-9X]$/u.test(val);\n    }\n  },\n  // eslint-disable-next-line max-statements\n  compare: (aa, bb) => {\n    debugData(`Comparing ISSN sets ${JSON.stringify(aa)} and ${JSON.stringify(bb)}`);\n    if (aa.length === 0 || bb.length === 0) {\n      // No data for decision\n      return 0;\n    }\n    const firstSharedIssn = aa.find(val => bb.includes(val));\n    if (firstSharedIssn) {\n      debug(`\\t Shared ISSN found: '${firstSharedIssn}'`);\n      // Maybe less for comps?\n      return 0.2;\n    }\n    return -0.2;\n\n  }\n});\n\nexport function uniqArray(arr) {\n  return arr.filter((val, i) => arr.indexOf(val) === i);\n}"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAE9B,MAAM,QAAQ,kBAAkB,iEAAiE;AACjG,MAAM,YAAY,MAAM,OAAO,MAAM;AAErC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC;AAAA,IAAC;AAAA;AAAA,EAA0B,MAAM;AAGzC,WAAO,SAAS;AAEhB,aAAS,WAAW;AAClB,YAAM,SAAS,OAAO,IAAI,QAAQ;AAClC,UAAI,OAAO,WAAW,GAAG;AACvB,eAAO,CAAC;AAAA,MACV;AAEA,YAAM,KAAM,OAAO,MAAM,2BAA2B,OAAO,CAAC,EAAE,GAAG,GAAG;AACpE,YAAM,gBAAgB,CAAC,KAAK,KAAK,GAAG;AAEpC,YAAM,iBAAiB,OAAO,QAAQ,OAAK,EAAE,UAAU,OAAO,QAAM,cAAc,SAAS,GAAG,IAAI,CAAC,EAAE,IAAI,QAAM,GAAG,KAAK,CAAC;AAGxH,YAAM,sBAAsB,gBAAgB,IAAI,SAAO,uBAAuB,GAAG,CAAC,EAAE,OAAO,SAAO,YAAY,GAAG,CAAC;AAClH,UAAI,CAAC,qBAAqB;AACxB,eAAO,CAAC;AAAA,MACV;AACA,aAAO,UAAU,mBAAmB;AAAA,IACtC;AAEA,aAAS,uBAAuB,KAAK;AACnC,aAAO,IAAI,QAAQ,eAAe,EAAE;AAAA,IACtC;AAEA,aAAS,YAAY,KAAK;AACxB,aAAO,gDAAgD,KAAK,GAAG;AAAA,IACjE;AAAA,EACF;AAAA;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,cAAU,uBAAuB,KAAK,UAAU,EAAE,CAAC,QAAQ,KAAK,UAAU,EAAE,CAAC,EAAE;AAC/E,QAAI,GAAG,WAAW,KAAK,GAAG,WAAW,GAAG;AAEtC,aAAO;AAAA,IACT;AACA,UAAM,kBAAkB,GAAG,KAAK,SAAO,GAAG,SAAS,GAAG,CAAC;AACvD,QAAI,iBAAiB;AACnB,YAAM,yBAA0B,eAAe,GAAG;AAElD,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EAET;AACF;AAEO,gBAAS,UAAU,KAAK;AAC7B,SAAO,IAAI,OAAO,CAAC,KAAK,MAAM,IAAI,QAAQ,GAAG,MAAM,CAAC;AACtD;",
   "names": []
 }

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

@@ -11,7 +11,8 @@ export default () => ({
     const clonedRecord = new MarcRecord(record, record._validationOptions);
     FixSami041().fix(clonedRecord);
     Remove041zxx().fix(clonedRecord);
-    return [{ leader: clonedRecord.leader, fields: clonedRecord.fields }, label];
+    clonedRecord.fields = clonedRecord.fields.filter((f) => ["008", "041"].includes(f.tag));
+    return [clonedRecord.fields, label];
   },
   // eslint-disable-next-line max-statements
   compare: (aa, bb) => {
@@ -25,9 +26,6 @@ export default () => ({
       if (score > 0.1) {
         return 0.1;
       }
-      if (score < -1) {
-        return -1;
-      }
       return score;
     }
     function compare008() {
@@ -60,11 +58,11 @@ export default () => ({
         return mainPenalty + sourcePenalty + scoreInd1;
       }
       return compareLanguageCodeLists(getLanguageCodesFrom041Fields(a), getLanguageCodesFrom041Fields(b));
-      function getFields041(record) {
-        if (!record.fields) {
+      function getFields041(fields) {
+        if (!fields) {
           return [];
         }
-        return record.fields.filter((f) => f.tag === "041");
+        return fields.filter((f) => f.tag === "041");
       }
     }
     function getSourceOfLanguageCode(field) {
@@ -91,7 +89,7 @@ export default () => ({
       return ["   ", "|||", "und"].includes(languageCode);
     }
     function compareLanguageCodeLists(a2, b2) {
-      if (a2.length === 0 || b2.length === 0) {
+      if (a2.length === 0 || b2.length === 0 || a2.every((val) => containsNoData(val)) || b2.every((val) => containsNoData(val))) {
         debugData(`No language to compare`);
         return 0;
       }
@@ -114,16 +112,8 @@ export default () => ({
       if (b2.length === 1 && b2[0] === "mul" && a2.length > 1) {
         return 0;
       }
-      const sharedValues = getSharedValues(a2, b2);
-      const aOnly = a2.filter((val) => !sharedValues.includes(val));
-      const bOnly = b2.filter((val) => !sharedValues.includes(val));
-      const hasUnd = [...aOnly, ...bOnly].includes("und");
+      const sharedValues = getSharedValues(a2, b2).filter((val) => !containsNoData(val));
       if (sharedValues.length < 1) {
-        console.info(`NV: ${sharedValues.join(", ")}`);
-        if (aOnly.length === 1 && bOnly.length === 1 && hasUnd) {
-          debug(`Both have languages, but none of these match. However, the benefit of doubt is given: '${aOnly[0]}' and '${bOnly[0]}' might mean the same}`);
-          return 0;
-        }
         debug(`Both have languages, but none of these match.`);
         return -1;
       }
@@ -144,11 +134,11 @@ export default () => ({
     }
   }
 });
-function get008Value(record, label = "record") {
-  if (!record || !record.fields) {
+function get008Value(fields, label = "record") {
+  if (!fields) {
     return;
   }
-  const f008 = record.fields.find((f) => f.tag === "008");
+  const f008 = fields.find((f) => f.tag === "008");
   if (!f008) {
     return void 0;
   }
@@ -182,8 +172,8 @@ function isLangCodeForALanguage(code, label = "record", encoding = void 0) {
   }
   return true;
 }
-function getLanguageCodesFrom041Fields(record) {
-  const values = record.fields.filter((f) => f.tag === "041").flatMap((f) => getFieldsLanguageCodes(f));
+function getLanguageCodesFrom041Fields(fields) {
+  const values = fields.filter((f) => f.tag === "041").flatMap((f) => getFieldsLanguageCodes(f));
   return [...new Set(values)].sort();
 }
 function getFieldsLanguageCodes(field) {