@natlibfi/melinda-record-matching 2.0.1-alpha.1 → 2.1.0

package/README.md

@@ -2,6 +2,6 @@
 
 ## License and copyright
 
-Copyright (c) 2020 **University Of Helsinki (The National Library Of Finland)**
+Copyright (c) 2020-2022 **University Of Helsinki (The National Library Of Finland)**
 
 This project's source code is licensed under the terms of **GNU Lesser General Public License Version 3** or any later version.

package/dist/candidate-search/index.js

@@ -33,7 +33,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -54,7 +54,9 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 * for the JavaScript code in this file.
 *
 */
-class CandidateSearchError extends Error {}
+class CandidateSearchError extends Error {} // serverMaxResults : maximum size of total search result available from the server, defaults to Aleph's 20000
+// eslint-disable-next-line max-statements
+
 
 exports.CandidateSearchError = CandidateSearchError;
 
@@ -62,18 +64,28 @@ var _default = ({
   record,
   searchSpec,
   url,
-  maxRecordsPerRequest = 50
+  maxCandidates,
+  maxRecordsPerRequest = 50,
+  serverMaxResult = 20000
 }) => {
   _marcRecord.MarcRecord.setValidationOptions({
     subfieldValues: false
   });
 
   const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:candidate-search');
+  const debugData = debug.extend('data');
+  debugData(`SearchSpec: ${JSON.stringify(searchSpec)}`);
+  debugData(`Url: ${url}`);
+  debugData(`MaxRecordsPerRequest ${maxRecordsPerRequest}`);
+  debugData(`ServerMaxResult: ${serverMaxResult}`);
+  debugData(`MaxCandidates: ${maxCandidates}`); // Do not retrieve more candidates than defined in maxCandidates
+
+  const adjustedMaxRecordsPerRequest = maxRecordsPerRequest >= maxCandidates ? maxCandidates : maxRecordsPerRequest;
   const inputRecordId = getRecordId(record);
   const queryList = (0, _queryList.default)(record, searchSpec);
   const client = (0, _sruClient.default)({
     url,
-    maxRecordsPerRequest,
+    maxRecordsPerRequest: adjustedMaxRecordsPerRequest,
     version: '2.0',
     retrieveAll: false
   });
@@ -83,47 +95,86 @@ var _default = ({
   if (queryList.length === 0) {
     // eslint-disable-line functional/no-conditional-statement
     throw new CandidateSearchError(`Generated query list contains no queries`);
-  } // eslint-disable-next-line max-statements
+  } // state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)
+  // state.query : current query (undefined if there was no queries left)
+  // state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)
+  // state.queryCandidateCounter: amount of candidate records retrieved from SRU for matching for current query, including the current record set (undefined if there were no queries left)
+  // state.queriesLeft : amount of queries left
+  // state.queryCounter : sequence for current query
+  // state.maxedQueries : queries that resulted in more than serverMaxResults hits
+  // eslint-disable-next-line max-statements
 
 
   return async ({
     queryOffset = 0,
-    resultSetOffset = 1
+    resultSetOffset = 1,
+    totalRecords = 0,
+    searchCounter = 0,
+    queryCandidateCounter = 0,
+    queryCounter = 0,
+    maxedQueries = []
   }) => {
     const query = queryList[queryOffset];
 
     if (query) {
       const {
         records,
-        nextOffset
-      } = await retrieveRecords();
+        nextOffset,
+        total
+      } = await retrieveRecords(); // If resultSetOffset === 1 this is the first search for the current query
+
+      debugData(`ResultSetOffset: ${resultSetOffset}`);
+      const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;
+      const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;
+      const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;
+      const newQueryCandidateCounter = resultSetOffset === 1 ? records.length : queryCandidateCounter + records.length;
+      const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;
+      const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;
 
       if (typeof nextOffset === 'number') {
-        debug(`Running next search for query ${queryOffset} ${query}`);
+        debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);
         return {
           records,
           queryOffset,
           resultSetOffset: nextOffset,
-          queriesLeft: queryList.length - (queryOffset + 1)
+          queriesLeft: queryList.length - (queryOffset + 1),
+          totalRecords: newTotalRecords,
+          query,
+          searchCounter: newSearchCounter,
+          queryCandidateCounter: newQueryCandidateCounter,
+          queryCounter: newQueryCounter,
+          maxedQueries: newMaxedQueries
         };
       }
 
-      debug(`Query ${queryOffset} ${query} done, moving to next query. (${queryList.length - (queryOffset + 1)} queries left)`);
+      debug(`Query ${queryOffset} ${query} done.`);
+      debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);
       return {
         records,
         queryOffset: queryOffset + 1,
-        queriesLeft: queryList.length - (queryOffset + 1)
+        queriesLeft: queryList.length - (queryOffset + 1),
+        totalRecords: newTotalRecords,
+        query,
+        searchCounter: newSearchCounter,
+        queryCandidateCounter: newQueryCandidateCounter,
+        queryCounter: newQueryCounter,
+        maxedQueries: newMaxedQueries
       };
     }
 
     debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);
     return {
-      records: []
+      records: [],
+      queriesLeft: 0,
+      queryCounter,
+      maxedQueries
     };
 
     function retrieveRecords() {
       return new Promise((resolve, reject) => {
-        const promises = [];
+        const promises = []; // eslint-disable-next-line functional/no-let
+
+        let totalRecords = 0;
         debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);
         client.searchRetrieve(query, {
           startRecord: resultSetOffset
@@ -136,6 +187,9 @@ var _default = ({
 
           debug(`SRU error for query: ${query}: ${err}`);
           reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));
+        }).on('total', total => {
+          debug(`Got total: ${total}`);
+          totalRecords += total;
         }).on('end', async nextOffset => {
           try {
             const records = await Promise.all(promises);
@@ -143,7 +197,8 @@ var _default = ({
             debug(`Found ${filtered.length} candidates`);
             resolve({
               nextOffset,
-              records: filtered
+              records: filtered,
+              total: totalRecords
             });
           } catch (err) {
             reject(err);
@@ -156,16 +211,7 @@ var _default = ({
               const foundRecordMarc = await _marcRecordSerializers.MARCXML.from(foundRecord, {
                 subfieldValues: false
               });
-              const foundRecordId = getRecordId(foundRecordMarc); // This does not work and might cause problems:
-              // Record *should* match itself AND in REST the input record is given id 000000001 always
-
-              debug(`Checking ${inputRecordId} vs ${foundRecordId}`);
-
-              if (inputRecordId === foundRecordId) {
-                debug(`Input and candidate are the same record per 001. Discarding candidate`);
-                return;
-              }
-
+              const foundRecordId = getRecordId(foundRecordMarc);
               return {
                 record: foundRecordMarc,
                 id: foundRecordId
@@ -179,6 +225,14 @@ var _default = ({
     }
   };
 
+  function checkMaxedQuery(query, total, serverMaxResult) {
+    // eslint-disable-next-line functional/no-conditional-statement
+    if (total >= serverMaxResult) {
+      debug(`WARNING: Query ${query} resulted in ${total} hits which meets the serverMaxResult (${serverMaxResult}) `);
+      return query;
+    }
+  }
+
   function getRecordId(record) {
     const [field] = record.get(/^001$/u);
     return field ? field.value : '';

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

@@ -1 +1 @@
-{"version":3,"sources":["../../src/candidate-search/index.js"],"names":["CandidateSearchError","Error","record","searchSpec","url","maxRecordsPerRequest","MarcRecord","setValidationOptions","subfieldValues","debug","inputRecordId","getRecordId","queryList","client","version","retrieveAll","JSON","stringify","length","queryOffset","resultSetOffset","query","records","nextOffset","retrieveRecords","queriesLeft","Promise","resolve","reject","promises","searchRetrieve","startRecord","on","err","SruSearchError","all","filtered","filter","r","foundRecord","push","handleRecord","foundRecordMarc","MARCXML","from","foundRecordId","id","field","get","value"],"mappings":";;;;;;;;;;;;;AA4BA;;AACA;;AACA;;AACA;;AACA;;;;;;;;AAhCA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAUO,MAAMA,oBAAN,SAAmCC,KAAnC,CAAyC;;;;eAEjC,CAAC;AAACC,EAAAA,MAAD;AAASC,EAAAA,UAAT;AAAqBC,EAAAA,GAArB;AAA0BC,EAAAA,oBAAoB,GAAG;AAAjD,CAAD,KAA0D;AACvEC,yBAAWC,oBAAX,CAAgC;AAACC,IAAAA,cAAc,EAAE;AAAjB,GAAhC;;AAEA,QAAMC,KAAK,GAAG,oBAAkB,oDAAlB,CAAd;AACA,QAAMC,aAAa,GAAGC,WAAW,CAACT,MAAD,CAAjC;AACA,QAAMU,SAAS,GAAG,wBAAkBV,MAAlB,EAA0BC,UAA1B,CAAlB;AACA,QAAMU,MAAM,GAAG,wBAAa;AAC1BT,IAAAA,GAD0B;AACrBC,IAAAA,oBADqB;AAE1BS,IAAAA,OAAO,EAAE,KAFiB;AAG1BC,IAAAA,WAAW,EAAE;AAHa,GAAb,CAAf;AAMAN,EAAAA,KAAK,CAAE,yBAAwBC,aAAc,EAAxC,CAAL;AACAD,EAAAA,KAAK,CAAE,uBAAsBO,IAAI,CAACC,SAAL,CAAeL,SAAf,CAA0B,EAAlD,CAAL;;AACA,MAAIA,SAAS,CAACM,MAAV,KAAqB,CAAzB,EAA4B;AAAE;AAC5B,UAAM,IAAIlB,oBAAJ,CAA0B,0CAA1B,CAAN;AACD,GAhBsE,CAkBvE;;;AACA,SAAO,OAAO;AAACmB,IAAAA,WAAW,GAAG,CAAf;AAAkBC,IAAAA,eAAe,GAAG;AAApC,GAAP,KAAkD;AACvD,UAAMC,KAAK,GAAGT,SAAS,CAACO,WAAD,CAAvB;;AAEA,QAAIE,KAAJ,EAAW;AACT,YAAM;AAACC,QAAAA,OAAD;AAAUC,QAAAA;AAAV,UAAwB,MAAMC,eAAe,EAAnD;;AAEA,UAAI,OAAOD,UAAP,KAAsB,QAA1B,EAAoC;AAClCd,QAAAA,KAAK,CAAE,iCAAgCU,WAAY,IAAGE,KAAM,EAAvD,CAAL;AACA,eAAO;AAACC,UAAAA,OAAD;AAAUH,UAAAA,WAAV;AAAuBC,UAAAA,eAAe,EAAEG,UAAxC;AAAoDE,UAAAA,WAAW,EAAEb,SAAS,CAACM,MAAV,IAAoBC,WAAW,GAAG,CAAlC;AAAjE,SAAP;AACD;;AACDV,MAAAA,KAAK,CAAE,SAAQU,WAAY,IAAGE,KAAM,iCAAgCT,SAAS,CAACM,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAAqC,gBAApG,CAAL;AACA,aAAO;AAACG,QAAAA,OAAD;AAAUH,QAAAA,WAAW,EAAEA,WAAW,GAAG,CAArC;AAAwCM,QAAAA,WAAW,EAAEb,SAAS,CAACM,MAAV,IAAoBC,WAAW,GAAG,CAAlC;AAArD,OAAP;AACD;;AAEDV,IAAAA,KAAK,CAAE,OAAMG,SAAS,CAACM,MAAO,uCAAsCC,WAAY,EAA3E,CAAL;AACA,WAAO;AAACG,MAAAA,OAAO,EAAE;AAAV,KAAP;;AAEA,aAASE,eAAT,GAA2B;AACzB,aAAO,IAAIE,OAAJ,CAAY,CAACC,OAAD,EAAUC,MAAV,KAAqB;AACtC,cAAMC,QAAQ,GAAG,EAAjB;AAEApB,QAAAA,KAAK,CAAE,wCAAuCY,KAAM,YAAWD,eAAgB,GAA1E,CAAL;AAEAP,QAAAA,MAAM,CAACiB,cAAP,CAAsBT,KAAtB,EAA6B;AAACU,UAAAA,WAAW,EAAEX;AAAd,SAA7B,EACGY,EADH,CACM,OADN,EACeC,GAAG,IAAI;AAClB;AACA,cAAIA,GAAG,YAAYC,yBAAnB,EAAmC;AACjCzB,YAAAA,KAAK,CAAE,iCAAgCY,KAAM,KAAIY,GAAI,EAAhD,CAAL;AACAL,YAAAA,MAAM,CAAC,IAAI5B,oBAAJ,CAA0B,iCAAgCqB,KAAM,KAAIY,GAAI,EAAxE,CAAD,CAAN;AACD;;AACDxB,UAAAA,KAAK,CAAE,wBAAuBY,KAAM,KAAIY,GAAI,EAAvC,CAAL;AACAL,UAAAA,MAAM,CAAC,IAAI5B,oBAAJ,CAA0B,wBAAuBqB,KAAM,KAAIY,GAAI,EAA/D,CAAD,CAAN;AACD,SATH,EAUGD,EAVH,CAUM,KAVN,EAUa,MAAMT,UAAN,IAAoB;AAC7B,cAAI;AACF,kBAAMD,OAAO,GAAG,MAAMI,OAAO,CAACS,GAAR,CAAYN,QAAZ,CAAtB;AACA,kBAAMO,QAAQ,GAAGd,OAAO,CAACe,MAAR,CAAeC,CAAC,IAAIA,CAApB,CAAjB;AAEA7B,YAAAA,KAAK,CAAE,SAAQ2B,QAAQ,CAAClB,MAAO,aAA1B,CAAL;AAEAS,YAAAA,OAAO,CAAC;AAACJ,cAAAA,UAAD;AAAaD,cAAAA,OAAO,EAAEc;AAAtB,aAAD,CAAP;AACD,WAPD,CAOE,OAAOH,GAAP,EAAY;AACZL,YAAAA,MAAM,CAACK,GAAD,CAAN;AACD;AACF,SArBH,EAsBGD,EAtBH,CAsBM,QAtBN,EAsBgBO,WAAW,IAAI;AAC3BV,UAAAA,QAAQ,CAACW,IAAT,CAAcC,YAAY,EAA1B,EAD2B,CACI;;AAE/B,yBAAeA,YAAf,GAA8B;AAC5B,gBAAI;AACF,oBAAMC,eAAe,GAAG,MAAMC,+BAAQC,IAAR,CAAaL,WAAb,EAA0B;AAAC/B,gBAAAA,cAAc,EAAE;AAAjB,eAA1B,CAA9B;AACA,oBAAMqC,aAAa,GAAGlC,WAAW,CAAC+B,eAAD,CAAjC,CAFE,CAIF;AACA;;AACAjC,cAAAA,KAAK,CAAE,YAAWC,aAAc,OAAMmC,aAAc,EAA/C,CAAL;;AACA,kBAAInC,aAAa,KAAKmC,aAAtB,EAAqC;AACnCpC,gBAAAA,KAAK,CAAE,uEAAF,CAAL;AACA;AACD;;AAED,qBAAO;AAACP,gBAAAA,MAAM,EAAEwC,eAAT;AAA0BI,gBAAAA,EAAE,EAAED;AAA9B,eAAP;AACD,aAbD,CAaE,OAAOZ,GAAP,EAAY;AACZ,oBAAM,IAAIhC,KAAJ,CAAW,6BAA4BgC,GAAI,aAAYM,WAAY,EAAnE,CAAN;AACD;AACF;AACF,SA3CH;AA4CD,OAjDM,CAAP;AAkDD;AACF,GArED;;AAuEA,WAAS5B,WAAT,CAAqBT,MAArB,EAA6B;AAC3B,UAAM,CAAC6C,KAAD,IAAU7C,MAAM,CAAC8C,GAAP,CAAW,QAAX,CAAhB;AACA,WAAOD,KAAK,GAAGA,KAAK,CAACE,KAAT,GAAiB,EAA7B;AACD;AACF,C","sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport createClient, {SruSearchError} from '@natlibfi/sru-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport {MARCXML} from '@natlibfi/marc-record-serializers';\nimport generateQueryList from './query-list';\n\nexport {searchTypes} from './query-list';\n\nexport class CandidateSearchError extends Error {}\n\nexport default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {\n  MarcRecord.setValidationOptions({subfieldValues: false});\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search');\n  const inputRecordId = getRecordId(record);\n  const queryList = generateQueryList(record, searchSpec);\n  const client = createClient({\n    url, maxRecordsPerRequest,\n    version: '2.0',\n    retrieveAll: false\n  });\n\n  debug(`Searching matches for ${inputRecordId}`);\n  debug(`Generated queryList ${JSON.stringify(queryList)}`);\n  if (queryList.length === 0) { // eslint-disable-line functional/no-conditional-statement\n    throw new CandidateSearchError(`Generated query list contains no queries`);\n  }\n\n  // eslint-disable-next-line max-statements\n  return async ({queryOffset = 0, resultSetOffset = 1}) => {\n    const query = queryList[queryOffset];\n\n    if (query) {\n      const {records, nextOffset} = await retrieveRecords();\n\n      if (typeof nextOffset === 'number') {\n        debug(`Running next search for query ${queryOffset} ${query}`);\n        return {records, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1)};\n      }\n      debug(`Query ${queryOffset} ${query} done, moving to next query. (${queryList.length - (queryOffset + 1)} queries left)`);\n      return {records, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1)};\n    }\n\n    debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);\n    return {records: []};\n\n    function retrieveRecords() {\n      return new Promise((resolve, reject) => {\n        const promises = [];\n\n        debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);\n\n        client.searchRetrieve(query, {startRecord: resultSetOffset})\n          .on('error', err => {\n            // eslint-disable-next-line functional/no-conditional-statement\n            if (err instanceof SruSearchError) {\n              debug(`SRU SruSearchError for query: ${query}: ${err}`);\n              reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));\n            }\n            debug(`SRU error for query: ${query}: ${err}`);\n            reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));\n          })\n          .on('end', async nextOffset => {\n            try {\n              const records = await Promise.all(promises);\n              const filtered = records.filter(r => r);\n\n              debug(`Found ${filtered.length} candidates`);\n\n              resolve({nextOffset, records: filtered});\n            } catch (err) {\n              reject(err);\n            }\n          })\n          .on('record', foundRecord => {\n            promises.push(handleRecord()); // eslint-disable-line functional/immutable-data\n\n            async function handleRecord() {\n              try {\n                const foundRecordMarc = await MARCXML.from(foundRecord, {subfieldValues: false});\n                const foundRecordId = getRecordId(foundRecordMarc);\n\n                // This does not work and might cause problems:\n                // Record *should* match itself AND in REST the input record is given id 000000001 always\n                debug(`Checking ${inputRecordId} vs ${foundRecordId}`);\n                if (inputRecordId === foundRecordId) {\n                  debug(`Input and candidate are the same record per 001. Discarding candidate`);\n                  return;\n                }\n\n                return {record: foundRecordMarc, id: foundRecordId};\n              } catch (err) {\n                throw new Error(`Failed converting record: ${err}, record: ${foundRecord}`);\n              }\n            }\n          });\n      });\n    }\n  };\n\n  function getRecordId(record) {\n    const [field] = record.get(/^001$/u);\n    return field ? field.value : '';\n  }\n};\n"],"file":"index.js"}
+{"version":3,"sources":["../../src/candidate-search/index.js"],"names":["CandidateSearchError","Error","record","searchSpec","url","maxCandidates","maxRecordsPerRequest","serverMaxResult","MarcRecord","setValidationOptions","subfieldValues","debug","debugData","extend","JSON","stringify","adjustedMaxRecordsPerRequest","inputRecordId","getRecordId","queryList","client","version","retrieveAll","length","queryOffset","resultSetOffset","totalRecords","searchCounter","queryCandidateCounter","queryCounter","maxedQueries","query","records","nextOffset","total","retrieveRecords","newTotalRecords","newQueryCounter","newSearchCounter","newQueryCandidateCounter","maxedQuery","checkMaxedQuery","undefined","newMaxedQueries","concat","queriesLeft","Promise","resolve","reject","promises","searchRetrieve","startRecord","on","err","SruSearchError","all","filtered","filter","r","foundRecord","push","handleRecord","foundRecordMarc","MARCXML","from","foundRecordId","id","field","get","value"],"mappings":";;;;;;;;;;;;;AA4BA;;AACA;;AACA;;AACA;;AACA;;;;;;;;AAhCA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAUO,MAAMA,oBAAN,SAAmCC,KAAnC,CAAyC,E,CAEhD;AAEA;;;;;eACe,CAAC;AAACC,EAAAA,MAAD;AAASC,EAAAA,UAAT;AAAqBC,EAAAA,GAArB;AAA0BC,EAAAA,aAA1B;AAAyCC,EAAAA,oBAAoB,GAAG,EAAhE;AAAoEC,EAAAA,eAAe,GAAG;AAAtF,CAAD,KAAkG;AAC/GC,yBAAWC,oBAAX,CAAgC;AAACC,IAAAA,cAAc,EAAE;AAAjB,GAAhC;;AAEA,QAAMC,KAAK,GAAG,oBAAkB,oDAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AAEAD,EAAAA,SAAS,CAAE,eAAcE,IAAI,CAACC,SAAL,CAAeZ,UAAf,CAA2B,EAA3C,CAAT;AACAS,EAAAA,SAAS,CAAE,QAAOR,GAAI,EAAb,CAAT;AACAQ,EAAAA,SAAS,CAAE,wBAAuBN,oBAAqB,EAA9C,CAAT;AACAM,EAAAA,SAAS,CAAE,oBAAmBL,eAAgB,EAArC,CAAT;AACAK,EAAAA,SAAS,CAAE,kBAAiBP,aAAc,EAAjC,CAAT,CAV+G,CAY/G;;AACA,QAAMW,4BAA4B,GAAGV,oBAAoB,IAAID,aAAxB,GAAwCA,aAAxC,GAAwDC,oBAA7F;AAEA,QAAMW,aAAa,GAAGC,WAAW,CAAChB,MAAD,CAAjC;AACA,QAAMiB,SAAS,GAAG,wBAAkBjB,MAAlB,EAA0BC,UAA1B,CAAlB;AACA,QAAMiB,MAAM,GAAG,wBAAa;AAC1BhB,IAAAA,GAD0B;AAE1BE,IAAAA,oBAAoB,EAAEU,4BAFI;AAG1BK,IAAAA,OAAO,EAAE,KAHiB;AAI1BC,IAAAA,WAAW,EAAE;AAJa,GAAb,CAAf;AAOAX,EAAAA,KAAK,CAAE,yBAAwBM,aAAc,EAAxC,CAAL;AACAN,EAAAA,KAAK,CAAE,uBAAsBG,IAAI,CAACC,SAAL,CAAeI,SAAf,CAA0B,EAAlD,CAAL;;AACA,MAAIA,SAAS,CAACI,MAAV,KAAqB,CAAzB,EAA4B;AAAE;AAC5B,UAAM,IAAIvB,oBAAJ,CAA0B,0CAA1B,CAAN;AACD,GA5B8G,CA8B/G;AACA;AACA;AACA;AACA;AACA;AACA;AAGA;;;AACA,SAAO,OAAO;AAACwB,IAAAA,WAAW,GAAG,CAAf;AAAkBC,IAAAA,eAAe,GAAG,CAApC;AAAuCC,IAAAA,YAAY,GAAG,CAAtD;AAAyDC,IAAAA,aAAa,GAAG,CAAzE;AAA4EC,IAAAA,qBAAqB,GAAG,CAApG;AAAuGC,IAAAA,YAAY,GAAG,CAAtH;AAAyHC,IAAAA,YAAY,GAAG;AAAxI,GAAP,KAAuJ;AAC5J,UAAMC,KAAK,GAAGZ,SAAS,CAACK,WAAD,CAAvB;;AAEA,QAAIO,KAAJ,EAAW;AACT,YAAM;AAACC,QAAAA,OAAD;AAAUC,QAAAA,UAAV;AAAsBC,QAAAA;AAAtB,UAA+B,MAAMC,eAAe,EAA1D,CADS,CAGT;;AACAvB,MAAAA,SAAS,CAAE,oBAAmBa,eAAgB,EAArC,CAAT;AACA,YAAMW,eAAe,GAAGX,eAAe,KAAK,CAApB,GAAwBS,KAAxB,GAAgCR,YAAxD;AACA,YAAMW,eAAe,GAAGZ,eAAe,KAAK,CAApB,GAAwBI,YAAY,GAAG,CAAvC,GAA2CA,YAAnE;AACA,YAAMS,gBAAgB,GAAGb,eAAe,KAAK,CAApB,GAAwB,CAAxB,GAA4BE,aAAa,GAAG,CAArE;AACA,YAAMY,wBAAwB,GAAGd,eAAe,KAAK,CAApB,GAAwBO,OAAO,CAACT,MAAhC,GAAyCK,qBAAqB,GAAGI,OAAO,CAACT,MAA1G;AAEA,YAAMiB,UAAU,GAAGf,eAAe,KAAK,CAApB,GAAwBgB,eAAe,CAACV,KAAD,EAAQG,KAAR,EAAe3B,eAAf,CAAvC,GAAyEmC,SAA5F;AACA,YAAMC,eAAe,GAAGH,UAAU,GAAGV,YAAY,CAACc,MAAb,CAAoBJ,UAApB,CAAH,GAAqCV,YAAvE;;AAEA,UAAI,OAAOG,UAAP,KAAsB,QAA1B,EAAoC;AAClCtB,QAAAA,KAAK,CAAE,iCAAgCa,WAAY,IAAGO,KAAM,0BAAyBE,UAAW,EAA3F,CAAL;AACA,eAAO;AAACD,UAAAA,OAAD;AAAUR,UAAAA,WAAV;AAAuBC,UAAAA,eAAe,EAAEQ,UAAxC;AAAoDY,UAAAA,WAAW,EAAE1B,SAAS,CAACI,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAAjE;AAAuGE,UAAAA,YAAY,EAAEU,eAArH;AAAsIL,UAAAA,KAAtI;AAA6IJ,UAAAA,aAAa,EAAEW,gBAA5J;AAA8KV,UAAAA,qBAAqB,EAAEW,wBAArM;AAA+NV,UAAAA,YAAY,EAAEQ,eAA7O;AAA8PP,UAAAA,YAAY,EAAEa;AAA5Q,SAAP;AACD;;AACDhC,MAAAA,KAAK,CAAE,SAAQa,WAAY,IAAGO,KAAM,QAA/B,CAAL;AACApB,MAAAA,KAAK,CAAE,cAAaQ,SAAS,CAACI,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAAqC,gBAApD,CAAL;AACA,aAAO;AAACQ,QAAAA,OAAD;AAAUR,QAAAA,WAAW,EAAEA,WAAW,GAAG,CAArC;AAAwCqB,QAAAA,WAAW,EAAE1B,SAAS,CAACI,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAArD;AAA2FE,QAAAA,YAAY,EAAEU,eAAzG;AAA0HL,QAAAA,KAA1H;AAAiIJ,QAAAA,aAAa,EAAEW,gBAAhJ;AAAkKV,QAAAA,qBAAqB,EAAEW,wBAAzL;AAAmNV,QAAAA,YAAY,EAAEQ,eAAjO;AAAkPP,QAAAA,YAAY,EAAEa;AAAhQ,OAAP;AACD;;AAEDhC,IAAAA,KAAK,CAAE,OAAMQ,SAAS,CAACI,MAAO,uCAAsCC,WAAY,EAA3E,CAAL;AACA,WAAO;AAACQ,MAAAA,OAAO,EAAE,EAAV;AAAca,MAAAA,WAAW,EAAE,CAA3B;AAA8BhB,MAAAA,YAA9B;AAA4CC,MAAAA;AAA5C,KAAP;;AAEA,aAASK,eAAT,GAA2B;AACzB,aAAO,IAAIW,OAAJ,CAAY,CAACC,OAAD,EAAUC,MAAV,KAAqB;AACtC,cAAMC,QAAQ,GAAG,EAAjB,CADsC,CAEtC;;AACA,YAAIvB,YAAY,GAAG,CAAnB;AAEAf,QAAAA,KAAK,CAAE,wCAAuCoB,KAAM,YAAWN,eAAgB,GAA1E,CAAL;AAEAL,QAAAA,MAAM,CAAC8B,cAAP,CAAsBnB,KAAtB,EAA6B;AAACoB,UAAAA,WAAW,EAAE1B;AAAd,SAA7B,EACG2B,EADH,CACM,OADN,EACeC,GAAG,IAAI;AAClB;AACA,cAAIA,GAAG,YAAYC,yBAAnB,EAAmC;AACjC3C,YAAAA,KAAK,CAAE,iCAAgCoB,KAAM,KAAIsB,GAAI,EAAhD,CAAL;AACAL,YAAAA,MAAM,CAAC,IAAIhD,oBAAJ,CAA0B,iCAAgC+B,KAAM,KAAIsB,GAAI,EAAxE,CAAD,CAAN;AACD;;AACD1C,UAAAA,KAAK,CAAE,wBAAuBoB,KAAM,KAAIsB,GAAI,EAAvC,CAAL;AACAL,UAAAA,MAAM,CAAC,IAAIhD,oBAAJ,CAA0B,wBAAuB+B,KAAM,KAAIsB,GAAI,EAA/D,CAAD,CAAN;AACD,SATH,EAUGD,EAVH,CAUM,OAVN,EAUelB,KAAK,IAAI;AACpBvB,UAAAA,KAAK,CAAE,cAAauB,KAAM,EAArB,CAAL;AACAR,UAAAA,YAAY,IAAIQ,KAAhB;AACD,SAbH,EAcGkB,EAdH,CAcM,KAdN,EAca,MAAMnB,UAAN,IAAoB;AAC7B,cAAI;AACF,kBAAMD,OAAO,GAAG,MAAMc,OAAO,CAACS,GAAR,CAAYN,QAAZ,CAAtB;AACA,kBAAMO,QAAQ,GAAGxB,OAAO,CAACyB,MAAR,CAAeC,CAAC,IAAIA,CAApB,CAAjB;AAEA/C,YAAAA,KAAK,CAAE,SAAQ6C,QAAQ,CAACjC,MAAO,aAA1B,CAAL;AAEAwB,YAAAA,OAAO,CAAC;AAACd,cAAAA,UAAD;AAAaD,cAAAA,OAAO,EAAEwB,QAAtB;AAAgCtB,cAAAA,KAAK,EAAER;AAAvC,aAAD,CAAP;AACD,WAPD,CAOE,OAAO2B,GAAP,EAAY;AACZL,YAAAA,MAAM,CAACK,GAAD,CAAN;AACD;AACF,SAzBH,EA0BGD,EA1BH,CA0BM,QA1BN,EA0BgBO,WAAW,IAAI;AAC3BV,UAAAA,QAAQ,CAACW,IAAT,CAAcC,YAAY,EAA1B,EAD2B,CACI;;AAE/B,yBAAeA,YAAf,GAA8B;AAC5B,gBAAI;AACF,oBAAMC,eAAe,GAAG,MAAMC,+BAAQC,IAAR,CAAaL,WAAb,EAA0B;AAACjD,gBAAAA,cAAc,EAAE;AAAjB,eAA1B,CAA9B;AACA,oBAAMuD,aAAa,GAAG/C,WAAW,CAAC4C,eAAD,CAAjC;AAEA,qBAAO;AAAC5D,gBAAAA,MAAM,EAAE4D,eAAT;AAA0BI,gBAAAA,EAAE,EAAED;AAA9B,eAAP;AACD,aALD,CAKE,OAAOZ,GAAP,EAAY;AACZ,oBAAM,IAAIpD,KAAJ,CAAW,6BAA4BoD,GAAI,aAAYM,WAAY,EAAnE,CAAN;AACD;AACF;AACF,SAvCH;AAwCD,OA/CM,CAAP;AAgDD;AACF,GA9ED;;AAgFA,WAASlB,eAAT,CAAyBV,KAAzB,EAAgCG,KAAhC,EAAuC3B,eAAvC,EAAwD;AACtD;AACA,QAAI2B,KAAK,IAAI3B,eAAb,EAA8B;AAC5BI,MAAAA,KAAK,CAAE,kBAAiBoB,KAAM,gBAAeG,KAAM,0CAAyC3B,eAAgB,IAAvG,CAAL;AACA,aAAOwB,KAAP;AACD;AACF;;AAGD,WAASb,WAAT,CAAqBhB,MAArB,EAA6B;AAC3B,UAAM,CAACiE,KAAD,IAAUjE,MAAM,CAACkE,GAAP,CAAW,QAAX,CAAhB;AACA,WAAOD,KAAK,GAAGA,KAAK,CAACE,KAAT,GAAiB,EAA7B;AACD;AACF,C","sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport createClient, {SruSearchError} from '@natlibfi/sru-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport {MARCXML} from '@natlibfi/marc-record-serializers';\nimport generateQueryList from './query-list';\n\nexport {searchTypes} from './query-list';\n\nexport class CandidateSearchError extends Error {}\n\n// serverMaxResults : maximum size of total search result available from the server, defaults to Aleph's 20000\n\n// eslint-disable-next-line max-statements\nexport default ({record, searchSpec, url, maxCandidates, maxRecordsPerRequest = 50, serverMaxResult = 20000}) => {\n  MarcRecord.setValidationOptions({subfieldValues: false});\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search');\n  const debugData = debug.extend('data');\n\n  debugData(`SearchSpec: ${JSON.stringify(searchSpec)}`);\n  debugData(`Url: ${url}`);\n  debugData(`MaxRecordsPerRequest ${maxRecordsPerRequest}`);\n  debugData(`ServerMaxResult: ${serverMaxResult}`);\n  debugData(`MaxCandidates: ${maxCandidates}`);\n\n  // Do not retrieve more candidates than defined in maxCandidates\n  const adjustedMaxRecordsPerRequest = maxRecordsPerRequest >= maxCandidates ? maxCandidates : maxRecordsPerRequest;\n\n  const inputRecordId = getRecordId(record);\n  const queryList = generateQueryList(record, searchSpec);\n  const client = createClient({\n    url,\n    maxRecordsPerRequest: adjustedMaxRecordsPerRequest,\n    version: '2.0',\n    retrieveAll: false\n  });\n\n  debug(`Searching matches for ${inputRecordId}`);\n  debug(`Generated queryList ${JSON.stringify(queryList)}`);\n  if (queryList.length === 0) { // eslint-disable-line functional/no-conditional-statement\n    throw new CandidateSearchError(`Generated query list contains no queries`);\n  }\n\n  // state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)\n  // state.query : current query (undefined if there was no queries left)\n  // state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)\n  // state.queryCandidateCounter: amount of candidate records retrieved from SRU for matching for current query, including the current record set (undefined if there were no queries left)\n  // state.queriesLeft : amount of queries left\n  // state.queryCounter : sequence for current query\n  // state.maxedQueries : queries that resulted in more than serverMaxResults hits\n\n\n  // eslint-disable-next-line max-statements\n  return async ({queryOffset = 0, resultSetOffset = 1, totalRecords = 0, searchCounter = 0, queryCandidateCounter = 0, queryCounter = 0, maxedQueries = []}) => {\n    const query = queryList[queryOffset];\n\n    if (query) {\n      const {records, nextOffset, total} = await retrieveRecords();\n\n      // If resultSetOffset === 1 this is the first search for the current query\n      debugData(`ResultSetOffset: ${resultSetOffset}`);\n      const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;\n      const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;\n      const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;\n      const newQueryCandidateCounter = resultSetOffset === 1 ? records.length : queryCandidateCounter + records.length;\n\n      const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;\n      const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;\n\n      if (typeof nextOffset === 'number') {\n        debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);\n        return {records, queryOffset, resultSetOffset: nextOffset, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n      }\n      debug(`Query ${queryOffset} ${query} done.`);\n      debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);\n      return {records, queryOffset: queryOffset + 1, queriesLeft: queryList.length - (queryOffset + 1), totalRecords: newTotalRecords, query, searchCounter: newSearchCounter, queryCandidateCounter: newQueryCandidateCounter, queryCounter: newQueryCounter, maxedQueries: newMaxedQueries};\n    }\n\n    debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);\n    return {records: [], queriesLeft: 0, queryCounter, maxedQueries};\n\n    function retrieveRecords() {\n      return new Promise((resolve, reject) => {\n        const promises = [];\n        // eslint-disable-next-line functional/no-let\n        let totalRecords = 0;\n\n        debug(`Searching for candidates with query: ${query} (Offset ${resultSetOffset})`);\n\n        client.searchRetrieve(query, {startRecord: resultSetOffset})\n          .on('error', err => {\n            // eslint-disable-next-line functional/no-conditional-statement\n            if (err instanceof SruSearchError) {\n              debug(`SRU SruSearchError for query: ${query}: ${err}`);\n              reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));\n            }\n            debug(`SRU error for query: ${query}: ${err}`);\n            reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));\n          })\n          .on('total', total => {\n            debug(`Got total: ${total}`);\n            totalRecords += total;\n          })\n          .on('end', async nextOffset => {\n            try {\n              const records = await Promise.all(promises);\n              const filtered = records.filter(r => r);\n\n              debug(`Found ${filtered.length} candidates`);\n\n              resolve({nextOffset, records: filtered, total: totalRecords});\n            } catch (err) {\n              reject(err);\n            }\n          })\n          .on('record', foundRecord => {\n            promises.push(handleRecord()); // eslint-disable-line functional/immutable-data\n\n            async function handleRecord() {\n              try {\n                const foundRecordMarc = await MARCXML.from(foundRecord, {subfieldValues: false});\n                const foundRecordId = getRecordId(foundRecordMarc);\n\n                return {record: foundRecordMarc, id: foundRecordId};\n              } catch (err) {\n                throw new Error(`Failed converting record: ${err}, record: ${foundRecord}`);\n              }\n            }\n          });\n      });\n    }\n  };\n\n  function checkMaxedQuery(query, total, serverMaxResult) {\n    // eslint-disable-next-line functional/no-conditional-statement\n    if (total >= serverMaxResult) {\n      debug(`WARNING: Query ${query} resulted in ${total} hits which meets the serverMaxResult (${serverMaxResult}) `);\n      return query;\n    }\n  }\n\n\n  function getRecordId(record) {\n    const [field] = record.get(/^001$/u);\n    return field ? field.value : '';\n  }\n};\n"],"file":"index.js"}

package/dist/candidate-search/index.spec.js

@@ -10,6 +10,8 @@ var _marcRecord = require("@natlibfi/marc-record");
 
 var _ = _interopRequireWildcard(require("."));
 
+var _debug = _interopRequireDefault(require("debug"));
+
 function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
 
 function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || typeof obj !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
@@ -22,7 +24,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 *
 * Melinda record matching modules for Javascript
 *
-* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)
+* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)
 *
 * This file is part of melinda-record-matching-js
 *
@@ -43,6 +45,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
 * for the JavaScript code in this file.
 *
 */
+const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:candidate-search:test');
 describe('candidate-search', () => {
   (0, _fixugenHttpClient.default)({
     callback,
@@ -57,8 +60,8 @@ describe('candidate-search', () => {
     getFixture,
     factoryOptions,
     searchOptions,
-    expectedFactoryError,
-    expectedSearchError,
+    expectedFactoryError = false,
+    expectedSearchError = false,
     enabled = true
   }) {
     const url = 'http://foo.bar';
@@ -85,12 +88,15 @@ describe('candidate-search', () => {
       url
     });
     await iterate({
-      searchOptions
+      searchOptions,
+      expectedSearchError
     });
 
     function formatFactoryOptions() {
+      debug(`Using factoryOptions: ${JSON.stringify(factoryOptions)}`);
       return { ...factoryOptions,
-        maxRecordsPerRequest: 1,
+        maxRecordsPerRequest: factoryOptions.maxRecordsPerRequest || 1,
+        maxServerResults: factoryOptions.maxServerResults || undefined,
         record: new _marcRecord.MarcRecord(factoryOptions.record, {
           subfieldValues: false
         })
@@ -100,6 +106,7 @@ describe('candidate-search', () => {
 
     async function iterate({
       searchOptions,
+      expectedSearchError,
       count = 1
     }) {
       const expectedResults = getFixture(`expectedResults${count}.json`);
@@ -114,16 +121,12 @@ describe('candidate-search', () => {
           (0, _chai.expect)(err.message).to.match(new RegExp(expectedSearchError, 'u'));
           return;
         }
-      }
+      } // eslint-disable-next-line functional/no-conditional-statement
 
-      const results = await search(searchOptions);
-      (0, _chai.expect)(formatResults(results)).to.eql(expectedResults);
 
-      if (results.records.length > 0) {
-        return iterate({
-          searchOptions: resultsToOptions(results),
-          count: count + 1
-        });
+      if (!expectedSearchError) {
+        const results = await search(searchOptions);
+        (0, _chai.expect)(formatResults(results)).to.eql(expectedResults);
       }
 
       function formatResults(results) {
@@ -138,13 +141,6 @@ describe('candidate-search', () => {
           }))
         };
       }
-
-      function resultsToOptions(results) {
-        return Object.entries(results).filter(([k]) => k === 'records' === false) // If key is 'records' return false
-        .reduce((acc, [k, v]) => ({ ...acc,
-          [k]: v
-        }), {});
-      }
     }
   }
 });

package/dist/candidate-search/index.spec.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/candidate-search/index.spec.js"],"names":["describe","callback","path","__dirname","recurse","fixura","reader","READERS","JSON","getFixture","factoryOptions","searchOptions","expectedFactoryError","expectedSearchError","enabled","url","isCandidateSearchError","formatFactoryOptions","to","throw","CandidateSearchError","RegExp","search","iterate","maxRecordsPerRequest","record","MarcRecord","subfieldValues","count","expectedResults","Error","err","be","an","message","match","results","formatResults","eql","records","length","resultsToOptions","map","id","toObject","Object","entries","filter","k","reduce","acc","v"],"mappings":";;AA4BA;;AACA;;AACA;;AACA;;AACA;;;;;;;;AAhCA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAQAA,QAAQ,CAAC,kBAAD,EAAqB,MAAM;AACjC,kCAAc;AACZC,IAAAA,QADY;AAEZC,IAAAA,IAAI,EAAE,CAACC,SAAD,EAAY,IAAZ,EAAkB,IAAlB,EAAwB,eAAxB,EAAyC,kBAAzC,EAA6D,OAA7D,CAFM;AAGZC,IAAAA,OAAO,EAAE,KAHG;AAIZC,IAAAA,MAAM,EAAE;AACNC,MAAAA,MAAM,EAAEC,gBAAQC;AADV;AAJI,GAAd,EADiC,CAUjC;;AACA,iBAAeP,QAAf,CAAwB;AAACQ,IAAAA,UAAD;AAAaC,IAAAA,cAAb;AAA6BC,IAAAA,aAA7B;AAA4CC,IAAAA,oBAA5C;AAAkEC,IAAAA,mBAAlE;AAAuFC,IAAAA,OAAO,GAAG;AAAjG,GAAxB,EAAgI;AAC9H,UAAMC,GAAG,GAAG,gBAAZ;;AAEA,QAAI,CAACD,OAAL,EAAc;AACZ;AACD;;AAED,QAAIF,oBAAJ,EAA0B;AACxB,UAAIA,oBAAoB,CAACI,sBAAzB,EAAiD;AAC/C,0BAAO,MAAM,eAAsB,EAAC,GAAGC,oBAAoB,EAAxB;AAA4BF,UAAAA;AAA5B,SAAtB,CAAb,EAAsEG,EAAtE,CAAyEC,KAAzE,CAA+EC,sBAA/E,EAAqG,IAAIC,MAAJ,CAAWT,oBAAX,EAAiC,GAAjC,CAArG;AACA;AACD;;AAED,wBAAO,MAAM,eAAsB,EAAC,GAAGK,oBAAoB,EAAxB;AAA4BF,QAAAA;AAA5B,OAAtB,CAAb,EAAsEG,EAAtE,CAAyEC,KAAzE,CAA+E,IAAIE,MAAJ,CAAWT,oBAAX,EAAiC,GAAjC,CAA/E;AACA;AACD;;AAED,UAAMU,MAAM,GAAG,eAAsB,EAAC,GAAGL,oBAAoB,EAAxB;AAA4BF,MAAAA;AAA5B,KAAtB,CAAf;AACA,UAAMQ,OAAO,CAAC;AAACZ,MAAAA;AAAD,KAAD,CAAb;;AAEA,aAASM,oBAAT,GAAgC;AAC9B,aAAO,EACL,GAAGP,cADE;AAELc,QAAAA,oBAAoB,EAAE,CAFjB;AAGLC,QAAAA,MAAM,EAAE,IAAIC,sBAAJ,CAAehB,cAAc,CAACe,MAA9B,EAAsC;AAACE,UAAAA,cAAc,EAAE;AAAjB,SAAtC;AAHH,OAAP;AAKD,KA1B6H,CA4B9H;;;AACA,mBAAeJ,OAAf,CAAuB;AAACZ,MAAAA,aAAD;AAAgBiB,MAAAA,KAAK,GAAG;AAAxB,KAAvB,EAAmD;AACjD,YAAMC,eAAe,GAAGpB,UAAU,CAAE,kBAAiBmB,KAAM,OAAzB,CAAlC;;AAEA,UAAIf,mBAAJ,EAAyB;AAAE;AACzB,YAAI;AACF,gBAAMS,MAAM,CAACX,aAAD,CAAZ;AACA,gBAAM,IAAImB,KAAJ,CAAU,mBAAV,CAAN;AACD,SAHD,CAGE,OAAOC,GAAP,EAAY;AACZ,4BAAOA,GAAP,EAAYb,EAAZ,CAAec,EAAf,CAAkBC,EAAlB,CAAqB,OAArB;AACA,4BAAOF,GAAG,CAACG,OAAX,EAAoBhB,EAApB,CAAuBiB,KAAvB,CAA6B,IAAId,MAAJ,CAAWR,mBAAX,EAAgC,GAAhC,CAA7B;AACA;AACD;AACF;;AAED,YAAMuB,OAAO,GAAG,MAAMd,MAAM,CAACX,aAAD,CAA5B;AAEA,wBAAO0B,aAAa,CAACD,OAAD,CAApB,EAA+BlB,EAA/B,CAAkCoB,GAAlC,CAAsCT,eAAtC;;AAEA,UAAIO,OAAO,CAACG,OAAR,CAAgBC,MAAhB,GAAyB,CAA7B,EAAgC;AAC9B,eAAOjB,OAAO,CAAC;AACbZ,UAAAA,aAAa,EAAE8B,gBAAgB,CAACL,OAAD,CADlB;AAEbR,UAAAA,KAAK,EAAEA,KAAK,GAAG;AAFF,SAAD,CAAd;AAID;;AAED,eAASS,aAAT,CAAuBD,OAAvB,EAAgC;AAC9B;AACA,eAAO,EACL,GAAGA,OADE;AAELG,UAAAA,OAAO,EAAEH,OAAO,CAACG,OAAR,CAAgBG,GAAhB,CAAoB,CAAC;AAACjB,YAAAA,MAAD;AAASkB,YAAAA;AAAT,WAAD,MAAmB;AAACA,YAAAA,EAAD;AAAKlB,YAAAA,MAAM,EAAEA,MAAM,CAACmB,QAAP;AAAb,WAAnB,CAApB;AAFJ,SAAP;AAID;;AAED,eAASH,gBAAT,CAA0BL,OAA1B,EAAmC;AACjC,eAAOS,MAAM,CAACC,OAAP,CAAeV,OAAf,EACJW,MADI,CACG,CAAC,CAACC,CAAD,CAAD,KAASA,CAAC,KAAK,SAAN,KAAoB,KADhC,EACuC;AADvC,SAEJC,MAFI,CAEG,CAACC,GAAD,EAAM,CAACF,CAAD,EAAIG,CAAJ,CAAN,MAAkB,EAAC,GAAGD,GAAJ;AAAS,WAACF,CAAD,GAAKG;AAAd,SAAlB,CAFH,EAEwC,EAFxC,CAAP;AAGD;AACF;AACF;AACF,CAhFO,CAAR","sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport {expect} from 'chai';\nimport {READERS} from '@natlibfi/fixura';\nimport generateTests from '@natlibfi/fixugen-http-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport createSearchInterface, {CandidateSearchError} from '.';\n\ndescribe('candidate-search', () => {\n  generateTests({\n    callback,\n    path: [__dirname, '..', '..', 'test-fixtures', 'candidate-search', 'index'],\n    recurse: false,\n    fixura: {\n      reader: READERS.JSON\n    }\n  });\n\n  // eslint-disable-next-line max-statements\n  async function callback({getFixture, factoryOptions, searchOptions, expectedFactoryError, expectedSearchError, enabled = true}) {\n    const url = 'http://foo.bar';\n\n    if (!enabled) {\n      return;\n    }\n\n    if (expectedFactoryError) {\n      if (expectedFactoryError.isCandidateSearchError) {\n        expect(() => createSearchInterface({...formatFactoryOptions(), url})).to.throw(CandidateSearchError, new RegExp(expectedFactoryError, 'u'));\n        return;\n      }\n\n      expect(() => createSearchInterface({...formatFactoryOptions(), url})).to.throw(new RegExp(expectedFactoryError, 'u'));\n      return;\n    }\n\n    const search = createSearchInterface({...formatFactoryOptions(), url});\n    await iterate({searchOptions});\n\n    function formatFactoryOptions() {\n      return {\n        ...factoryOptions,\n        maxRecordsPerRequest: 1,\n        record: new MarcRecord(factoryOptions.record, {subfieldValues: false})\n      };\n    }\n\n    // eslint-disable-next-line max-statements\n    async function iterate({searchOptions, count = 1}) {\n      const expectedResults = getFixture(`expectedResults${count}.json`);\n\n      if (expectedSearchError) { // eslint-disable-line functional/no-conditional-statement\n        try {\n          await search(searchOptions);\n          throw new Error('Expected an error');\n        } catch (err) {\n          expect(err).to.be.an('error');\n          expect(err.message).to.match(new RegExp(expectedSearchError, 'u'));\n          return;\n        }\n      }\n\n      const results = await search(searchOptions);\n\n      expect(formatResults(results)).to.eql(expectedResults);\n\n      if (results.records.length > 0) {\n        return iterate({\n          searchOptions: resultsToOptions(results),\n          count: count + 1\n        });\n      }\n\n      function formatResults(results) {\n        // console.log(results); //eslint-disable-line\n        return {\n          ...results,\n          records: results.records.map(({record, id}) => ({id, record: record.toObject()}))\n        };\n      }\n\n      function resultsToOptions(results) {\n        return Object.entries(results)\n          .filter(([k]) => k === 'records' === false) // If key is 'records' return false\n          .reduce((acc, [k, v]) => ({...acc, [k]: v}), {});\n      }\n    }\n  }\n});\n"],"file":"index.spec.js"}
+{"version":3,"sources":["../../src/candidate-search/index.spec.js"],"names":["debug","describe","callback","path","__dirname","recurse","fixura","reader","READERS","JSON","getFixture","factoryOptions","searchOptions","expectedFactoryError","expectedSearchError","enabled","url","isCandidateSearchError","formatFactoryOptions","to","throw","CandidateSearchError","RegExp","search","iterate","stringify","maxRecordsPerRequest","maxServerResults","undefined","record","MarcRecord","subfieldValues","count","expectedResults","Error","err","be","an","message","match","results","formatResults","eql","records","map","id","toObject"],"mappings":";;AA4BA;;AACA;;AACA;;AACA;;AACA;;AACA;;;;;;;;AAjCA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AASA,MAAMA,KAAK,GAAG,oBAAkB,yDAAlB,CAAd;AAEAC,QAAQ,CAAC,kBAAD,EAAqB,MAAM;AACjC,kCAAc;AACZC,IAAAA,QADY;AAEZC,IAAAA,IAAI,EAAE,CAACC,SAAD,EAAY,IAAZ,EAAkB,IAAlB,EAAwB,eAAxB,EAAyC,kBAAzC,EAA6D,OAA7D,CAFM;AAGZC,IAAAA,OAAO,EAAE,KAHG;AAIZC,IAAAA,MAAM,EAAE;AACNC,MAAAA,MAAM,EAAEC,gBAAQC;AADV;AAJI,GAAd,EADiC,CAUjC;;AACA,iBAAeP,QAAf,CAAwB;AAACQ,IAAAA,UAAD;AAAaC,IAAAA,cAAb;AAA6BC,IAAAA,aAA7B;AAA4CC,IAAAA,oBAAoB,GAAG,KAAnE;AAA0EC,IAAAA,mBAAmB,GAAG,KAAhG;AAAuGC,IAAAA,OAAO,GAAG;AAAjH,GAAxB,EAAgJ;AAC9I,UAAMC,GAAG,GAAG,gBAAZ;;AAEA,QAAI,CAACD,OAAL,EAAc;AACZ;AACD;;AAED,QAAIF,oBAAJ,EAA0B;AACxB,UAAIA,oBAAoB,CAACI,sBAAzB,EAAiD;AAC/C,0BAAO,MAAM,eAAsB,EAAC,GAAGC,oBAAoB,EAAxB;AAA4BF,UAAAA;AAA5B,SAAtB,CAAb,EAAsEG,EAAtE,CAAyEC,KAAzE,CAA+EC,sBAA/E,EAAqG,IAAIC,MAAJ,CAAWT,oBAAX,EAAiC,GAAjC,CAArG;AACA;AACD;;AAED,wBAAO,MAAM,eAAsB,EAAC,GAAGK,oBAAoB,EAAxB;AAA4BF,QAAAA;AAA5B,OAAtB,CAAb,EAAsEG,EAAtE,CAAyEC,KAAzE,CAA+E,IAAIE,MAAJ,CAAWT,oBAAX,EAAiC,GAAjC,CAA/E;AACA;AACD;;AAED,UAAMU,MAAM,GAAG,eAAsB,EAAC,GAAGL,oBAAoB,EAAxB;AAA4BF,MAAAA;AAA5B,KAAtB,CAAf;AACA,UAAMQ,OAAO,CAAC;AAACZ,MAAAA,aAAD;AAAgBE,MAAAA;AAAhB,KAAD,CAAb;;AAEA,aAASI,oBAAT,GAAgC;AAC9BlB,MAAAA,KAAK,CAAE,yBAAwBS,IAAI,CAACgB,SAAL,CAAed,cAAf,CAA+B,EAAzD,CAAL;AACA,aAAO,EACL,GAAGA,cADE;AAELe,QAAAA,oBAAoB,EAAEf,cAAc,CAACe,oBAAf,IAAuC,CAFxD;AAGLC,QAAAA,gBAAgB,EAAEhB,cAAc,CAACgB,gBAAf,IAAmCC,SAHhD;AAILC,QAAAA,MAAM,EAAE,IAAIC,sBAAJ,CAAenB,cAAc,CAACkB,MAA9B,EAAsC;AAACE,UAAAA,cAAc,EAAE;AAAjB,SAAtC;AAJH,OAAP;AAMD,KA5B6I,CA8B9I;;;AACA,mBAAeP,OAAf,CAAuB;AAACZ,MAAAA,aAAD;AAAgBE,MAAAA,mBAAhB;AAAqCkB,MAAAA,KAAK,GAAG;AAA7C,KAAvB,EAAwE;AACtE,YAAMC,eAAe,GAAGvB,UAAU,CAAE,kBAAiBsB,KAAM,OAAzB,CAAlC;;AAEA,UAAIlB,mBAAJ,EAAyB;AAAE;AACzB,YAAI;AACF,gBAAMS,MAAM,CAACX,aAAD,CAAZ;AACA,gBAAM,IAAIsB,KAAJ,CAAU,mBAAV,CAAN;AACD,SAHD,CAGE,OAAOC,GAAP,EAAY;AACZ,4BAAOA,GAAP,EAAYhB,EAAZ,CAAeiB,EAAf,CAAkBC,EAAlB,CAAqB,OAArB;AACA,4BAAOF,GAAG,CAACG,OAAX,EAAoBnB,EAApB,CAAuBoB,KAAvB,CAA6B,IAAIjB,MAAJ,CAAWR,mBAAX,EAAgC,GAAhC,CAA7B;AACA;AACD;AACF,OAZqE,CActE;;;AACA,UAAI,CAACA,mBAAL,EAA0B;AACxB,cAAM0B,OAAO,GAAG,MAAMjB,MAAM,CAACX,aAAD,CAA5B;AACA,0BAAO6B,aAAa,CAACD,OAAD,CAApB,EAA+BrB,EAA/B,CAAkCuB,GAAlC,CAAsCT,eAAtC;AACD;;AAED,eAASQ,aAAT,CAAuBD,OAAvB,EAAgC;AAC9B;AACA,eAAO,EACL,GAAGA,OADE;AAELG,UAAAA,OAAO,EAAEH,OAAO,CAACG,OAAR,CAAgBC,GAAhB,CAAoB,CAAC;AAACf,YAAAA,MAAD;AAASgB,YAAAA;AAAT,WAAD,MAAmB;AAACA,YAAAA,EAAD;AAAKhB,YAAAA,MAAM,EAAEA,MAAM,CAACiB,QAAP;AAAb,WAAnB,CAApB;AAFJ,SAAP;AAID;AAEF;AACF;AACF,CAxEO,CAAR","sourcesContent":["/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport {expect} from 'chai';\nimport {READERS} from '@natlibfi/fixura';\nimport generateTests from '@natlibfi/fixugen-http-client';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport createSearchInterface, {CandidateSearchError} from '.';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:test');\n\ndescribe('candidate-search', () => {\n  generateTests({\n    callback,\n    path: [__dirname, '..', '..', 'test-fixtures', 'candidate-search', 'index'],\n    recurse: false,\n    fixura: {\n      reader: READERS.JSON\n    }\n  });\n\n  // eslint-disable-next-line max-statements\n  async function callback({getFixture, factoryOptions, searchOptions, expectedFactoryError = false, expectedSearchError = false, enabled = true}) {\n    const url = 'http://foo.bar';\n\n    if (!enabled) {\n      return;\n    }\n\n    if (expectedFactoryError) {\n      if (expectedFactoryError.isCandidateSearchError) {\n        expect(() => createSearchInterface({...formatFactoryOptions(), url})).to.throw(CandidateSearchError, new RegExp(expectedFactoryError, 'u'));\n        return;\n      }\n\n      expect(() => createSearchInterface({...formatFactoryOptions(), url})).to.throw(new RegExp(expectedFactoryError, 'u'));\n      return;\n    }\n\n    const search = createSearchInterface({...formatFactoryOptions(), url});\n    await iterate({searchOptions, expectedSearchError});\n\n    function formatFactoryOptions() {\n      debug(`Using factoryOptions: ${JSON.stringify(factoryOptions)}`);\n      return {\n        ...factoryOptions,\n        maxRecordsPerRequest: factoryOptions.maxRecordsPerRequest || 1,\n        maxServerResults: factoryOptions.maxServerResults || undefined,\n        record: new MarcRecord(factoryOptions.record, {subfieldValues: false})\n      };\n    }\n\n    // eslint-disable-next-line max-statements\n    async function iterate({searchOptions, expectedSearchError, count = 1}) {\n      const expectedResults = getFixture(`expectedResults${count}.json`);\n\n      if (expectedSearchError) { // eslint-disable-line functional/no-conditional-statement\n        try {\n          await search(searchOptions);\n          throw new Error('Expected an error');\n        } catch (err) {\n          expect(err).to.be.an('error');\n          expect(err.message).to.match(new RegExp(expectedSearchError, 'u'));\n          return;\n        }\n      }\n\n      // eslint-disable-next-line functional/no-conditional-statement\n      if (!expectedSearchError) {\n        const results = await search(searchOptions);\n        expect(formatResults(results)).to.eql(expectedResults);\n      }\n\n      function formatResults(results) {\n        // console.log(results); //eslint-disable-line\n        return {\n          ...results,\n          records: results.records.map(({record, id}) => ({id, record: record.toObject()}))\n        };\n      }\n\n    }\n  }\n});\n"],"file":"index.spec.js"}

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

@@ -169,6 +169,7 @@ function bibHostComponents(record) {
   }
 } // SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to
 // (probably) to *headings* index TIT
+// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search
 // Headings index TIT drops articles etc. from the start of the title according to the filing indicator
 // Currently filing indicator is not implemented - if the title starts with an article and the Melinda
 // record is correctly catalogued using a filing indicator -> dc.title search won't match
@@ -176,12 +177,15 @@ function bibHostComponents(record) {
 
 function bibTitle(record) {
   const title = getTitle();
+  const booleanStartWords = ['and', 'or', 'nor', 'not'];
 
   if (title) {
     const formatted = title.replace(/[^\w\s\p{Alphabetic}]/gu, '') // Clean up concurrent spaces from fe. subfield changes
-    .replace(/ +/gu, ' ').trim().slice(0, 30).trim(); // Prevent too many matches by having a minimum length requirement
+    .replace(/ +/gu, ' ').trim().slice(0, 30).trim(); // use word search for titles starting with a boolean
 
-    return formatted.length >= 5 ? [`dc.title="^${formatted}*"`] : [];
+    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word)); // Prevent too many matches by having a minimum length requirement
+
+    return formatted.length >= 5 ? [`dc.title="${useWordSearch ? '' : '^'}${formatted}*"`] : [];
   }
 
   return [];

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

@@ -1 +1 @@
-{"version":3,"sources":["../../../src/candidate-search/query-list/bib.js"],"names":["bibSourceIds","record","debug","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","map","toSidString","filter","nonEmptySid","field","createSidString","sfC","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","formatted","trim","slice","includes","join","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub","undefined"],"mappings":";;;;;;;;;;;AA4BA;;AACA;;AACA;;;;AA7BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAMO,SAASA,YAAT,CAAsBC,MAAtB,EAA8B;AAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAQE,QAAMC,KAAK,GAAG,oBAAkB,qEAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB,CA3BmC,CA4BnC;;AAEAF,EAAAA,KAAK,CAAE,iCAAF,CAAL;AAEA,QAAMG,KAAK,GAAGJ,MAAM,CAACK,GAAP,CAAW,KAAX,CAAd;AACAH,EAAAA,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeJ,KAAf,CAAsB,EAAxD,CAAT;AAEA,SAAOA,KAAK,CAACE,MAAN,GAAe,CAAf,GAAmBG,YAAY,CAACL,KAAD,CAA/B,GAAyC,EAAhD;;AAEA,WAASK,YAAT,CAAsBL,KAAtB,EAA6B;AAC3BH,IAAAA,KAAK,CAAE,wCAAF,CAAL;AAEA,UAAMS,UAAU,GAAGC,aAAa,CAACP,KAAD,CAAhC;;AAEA,QAAIM,UAAU,CAACJ,MAAX,GAAoB,CAAxB,EAA2B;AACzBL,MAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,aAAO,EAAP;AACD;;AAED,UAAMW,UAAU,GAAG,qCAAUF,UAAV,EAAsB,kBAAtB,CAAnB;AAEA,WAAOE,UAAP;;AAEA,aAASD,aAAT,CAAuBP,KAAvB,EAA8B;AAC5BH,MAAAA,KAAK,CAAE,qCAAF,CAAL,CAD4B,CAG5B;;AACA,YAAMS,UAAU,GAAGN,KAAK,CAACS,GAAN,CAAUC,WAAV,EAAuBC,MAAvB,CAA8BC,WAAW,IAAIA,WAA7C,CAAnB;AACA,aAAON,UAAP;;AAEA,eAASI,WAAT,CAAqBG,KAArB,EAA4B;AAC1BhB,QAAAA,KAAK,CAAE,6BAAF,CAAL;AAEA,eAAO,mDAA+BgB,KAA/B,IAAwCC,eAAe,CAACD,KAAD,CAAvD,GAAiE,EAAxE;;AAEA,iBAASC,eAAT,CAAyBD,KAAzB,EAAgC;AAC9BhB,UAAAA,KAAK,CAAE,8BAAF,CAAL;AACA,gBAAM,CAACkB,GAAD,IAAQ,sCAAkBF,KAAlB,EAAyB,GAAzB,CAAd;AACA,gBAAM,CAACG,GAAD,IAAQ,sCAAkBH,KAAlB,EAAyB,GAAzB,CAAd;AAEA,gBAAMI,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACJ,GAAD,CAA1B,CAArC;AACA,gBAAMK,UAAU,GAAGD,yBAAyB,CAACH,GAAD,CAA5C;AAEAlB,UAAAA,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAL,CAAeW,GAAf,CAAoB,MAAKZ,IAAI,CAACC,SAAL,CAAeY,GAAf,CAAoB,EAAjD,CAAT;AACA,iBAAOC,UAAU,CAACI,MAAX,CAAkBD,UAAlB,CAAP;AACD;;AAED,iBAASF,kBAAT,CAA4BI,aAA5B,EAA2C;AACzC,gBAAMC,iBAAiB,GAAI,gDAA3B;AACA,gBAAMC,eAAe,GAAGF,aAAa,CAACG,OAAd,CAAsBF,iBAAtB,EAAyC,OAAzC,CAAxB;AACAzB,UAAAA,SAAS,CAAE,cAAawB,aAAc,OAAME,eAAgB,EAAnD,CAAT;AACA,iBAAOA,eAAP;AACD;;AAED,iBAASL,yBAAT,CAAmCG,aAAnC,EAAkD;AAChDxB,UAAAA,SAAS,CAAE,eAAcwB,aAAc,EAA9B,CAAT;AACA,gBAAMI,kBAAkB,GAAI,OAA5B;AACA,iBAAOJ,aAAa,CAACG,OAAd,CAAsBC,kBAAtB,EAA0C,EAA1C,CAAP;AACD;AAEF;AACF;AACF;AACF;;AAEM,SAASC,aAAT,CAAuB/B,MAAvB,EAA+B;AACpC;AAEA,QAAMC,KAAK,GAAG,oBAAkB,wEAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AACAF,EAAAA,KAAK,CAAE,iCAAF,CAAL,CALoC,CAOpC;AACA;AACA;;AACA,QAAM+B,UAAU,GAAG,sCAAkBhC,MAAlB,CAAnB;AAEAE,EAAAA,SAAS,CAAE,uBAAsB8B,UAAU,CAAC1B,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAewB,UAAf,CAA2B,EAA1E,CAAT;;AAEA,MAAIA,UAAU,CAAC1B,MAAX,GAAoB,CAAxB,EAA2B;AACzBL,IAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,WAAO,EAAP;AACD;;AAED,SAAO,qCAAU+B,UAAV,EAAsB,mBAAtB,CAAP;AACD,C,CAGD;AACA;;;AACO,SAASC,iBAAT,CAA2BjC,MAA3B,EAAmC;AACxC,QAAMkC,EAAE,GAAGC,SAAS,EAApB;AACA,SAAOD,EAAE,GAAG,CAAE,uBAAsBA,EAAG,EAA3B,CAAH,GAAmC,EAA5C;;AAEA,WAASC,SAAT,GAAqB;AACnB,UAAM,CAAClB,KAAD,IAAUjB,MAAM,CAACK,GAAP,CAAW,QAAX,CAAhB;;AAEA,QAAIY,KAAJ,EAAW;AACT,YAAM;AAACmB,QAAAA;AAAD,UAAUnB,KAAK,CAACoB,SAAN,CAAgBC,IAAhB,CAAqB,CAAC;AAACC,QAAAA;AAAD,OAAD,KAAYA,IAAI,KAAK,GAA1C,KAAkD,EAAlE;;AAEA,UAAIH,KAAK,IAAK,kBAAD,CAAqBI,IAArB,CAA0BJ,KAA1B,CAAb,EAA+C;AAC7C,eAAOA,KAAK,CAACP,OAAN,CAAc,kBAAd,EAAkC,EAAlC,CAAP;AACD;;AAED,UAAIO,KAAK,IAAK,aAAD,CAAgBI,IAAhB,CAAqBJ,KAArB,CAAb,EAA0C;AACxC,eAAOA,KAAK,CAACP,OAAN,CAAc,aAAd,EAA6B,EAA7B,CAAP;AACD;;AAED,aAAO,KAAP;AACD;;AACD,WAAO,KAAP;AACD;AACF,C,CAED;AACA;AACA;AACA;AACA;;;AAEO,SAASY,QAAT,CAAkBzC,MAAlB,EAA0B;AAC/B,QAAM0C,KAAK,GAAGC,QAAQ,EAAtB;;AAEA,MAAID,KAAJ,EAAW;AACT,UAAME,SAAS,GAAGF,KAAK,CACpBb,OADe,CACP,yBADO,EACoB,EADpB,EAEhB;AAFgB,KAGfA,OAHe,CAGP,MAHO,EAGC,GAHD,EAIfgB,IAJe,GAKfC,KALe,CAKT,CALS,EAKN,EALM,EAMfD,IANe,EAAlB,CADS,CAST;;AACA,WAAOD,SAAS,CAACtC,MAAV,IAAoB,CAApB,GAAwB,CAAE,cAAasC,SAAU,IAAzB,CAAxB,GAAwD,EAA/D;AACD;;AAED,SAAO,EAAP;;AAEA,WAASD,QAAT,GAAoB;AAClB,UAAM,CAAC1B,KAAD,IAAUjB,MAAM,CAACK,GAAP,CAAW,QAAX,CAAhB;;AAEA,QAAIY,KAAJ,EAAW;AACT,aAAOA,KAAK,CAACoB,SAAN,CACJtB,MADI,CACG,CAAC;AAACwB,QAAAA;AAAD,OAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWQ,QAAX,CAAoBR,IAApB,CADf,EAEJ1B,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,EAGL;AAHK,OAIJY,IAJI,CAIC,GAJD,CAAP;AAKD;;AACD,WAAO,KAAP;AACD;AACF,C,CAED;;;AACO,SAASC,sBAAT,CAAgCjD,MAAhC,EAAwC;AAE7C,QAAMC,KAAK,GAAG,oBAAkB,iFAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AACAF,EAAAA,KAAK,CAAE,2CAAF,CAAL;AAEA,QAAMiD,MAAM,GAAGlD,MAAM,CAACK,GAAP,CAAW,wBAAX,CAAf;AACA,QAAM8C,WAAW,GAAG,GAAG1B,MAAH,CAAU,GAAGyB,MAAM,CAACrC,GAAP,CAAWuC,aAAX,CAAb,CAApB;AACA,QAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQH,WAAR,CAAJ,CAA1B;AAEAjD,EAAAA,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAL,CAAe0C,MAAf,CAAuB,EAAvD,CAAT;AACAhD,EAAAA,SAAS,CAAE,gBAAeiD,WAAW,CAAC7C,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAe2C,WAAf,CAA4B,EAArE,CAAT;AACAjD,EAAAA,SAAS,CAAE,uBAAsBmD,iBAAiB,CAAC/C,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAe6C,iBAAf,CAAkC,EAAxF,CAAT;;AAEA,MAAIA,iBAAiB,CAAC/C,MAAlB,GAA2B,CAA/B,EAAkC;AAChCL,IAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,WAAO,EAAP;AACD;;AAED,SAAO,qCAAUoD,iBAAV,EAA6B,eAA7B,CAAP;;AAEA,WAASD,aAAT,CAAuB;AAACG,IAAAA,GAAD;AAAMlB,IAAAA;AAAN,GAAvB,EAAyC;AACvC,UAAMmB,cAAc,GAAI,kBAAxB;AACA,UAAMC,aAAa,GAAI,mBAAvB;;AAEA,QAAIF,GAAG,KAAK,KAAZ,EAAmB;AACjB,aAAOlB,SAAS,CACbtB,MADI,CACG2C,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAW,GAAX,EAAgBX,QAAhB,CAAyBW,GAAG,CAACnB,IAA7B,KAAsCiB,cAAc,CAAChB,IAAf,CAAoBkB,GAAG,CAACtB,KAAxB,CAAtC,IAAwEsB,GAAG,CAACtB,KAAJ,KAAcuB,SADhG,EAEJ9C,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,CAAP;AAGD;;AAED,QAAImB,GAAG,KAAK,KAAZ,EAAmB;AACjB,aAAOlB,SAAS,CACbtB,MADI,CACG2C,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACnB,IAAxB,KAAiCiB,cAAc,CAAChB,IAAf,CAAoBkB,GAAG,CAACtB,KAAxB,CAAjC,IAAmEsB,GAAG,CAACtB,KAAJ,KAAcuB,SAD3F,EAEJ9C,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,CAAP;AAGD;;AAED,WAAOC,SAAS,CACbtB,MADI,CACG2C,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACnB,IAAxB,KAAiCkB,aAAa,CAACjB,IAAd,CAAmBkB,GAAG,CAACtB,KAAvB,CAAjC,IAAkEsB,GAAG,CAACtB,KAAJ,KAAcuB,SAD1F,EAEJ9C,GAFI,CAEA,CAAC;AAACuB,MAAAA;AAAD,KAAD,KAAaA,KAFb,CAAP;AAGD;AACF","sourcesContent":["\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues} from '../../matching-utils';\n\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = subfieldValue.replace(sourcePrefixRegex, '$<id>');\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return subfieldValue.replace(normalizeAwayRegex, '');\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const id = getHostId();\n  return id ? [`melinda.partsofhost=${id}`] : [];\n\n  function getHostId() {\n    const [field] = record.get(/^773$/u);\n\n    if (field) {\n      const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n      if (value && (/^\\(FI-MELINDA\\)/u).test(value)) {\n        return value.replace(/^\\(FI-MELINDA\\)/u, '');\n      }\n\n      if (value && (/^\\(FIN01\\)/u).test(value)) {\n        return value.replace(/^\\(FIN01\\)/u, '');\n      }\n\n      return false;\n    }\n    return false;\n  }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n  const title = getTitle();\n\n  if (title) {\n    const formatted = title\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // Prevent too many matches by having a minimum length requirement\n    return formatted.length >= 5 ? [`dc.title=\"^${formatted}*\"`] : [];\n  }\n\n  return [];\n\n  function getTitle() {\n    const [field] = record.get(/^245$/u);\n\n    if (field) {\n      return field.subfields\n        .filter(({code}) => ['a', 'b'].includes(code))\n        .map(({value}) => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n    }\n    return false;\n  }\n}\n\n// eslint-disable-next-line max-statements\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  const fields = record.get(/^(?<def>020|022|024)$/u);\n  const identifiers = [].concat(...fields.map(toIdentifiers));\n  const uniqueIdentifiers = [...new Set(identifiers)];\n\n  debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n  debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n  debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n  if (uniqueIdentifiers.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n  function toIdentifiers({tag, subfields}) {\n    const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n    const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n    if (tag === '022') {\n      return subfields\n        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined)\n      .map(({value}) => value);\n  }\n}\n"],"file":"bib.js"}
+{"version":3,"sources":["../../../src/candidate-search/query-list/bib.js"],"names":["bibSourceIds","record","debug","debugData","extend","fSids","get","length","JSON","stringify","toSidQueries","sidStrings","getSidStrings","sidQueries","map","toSidString","filter","nonEmptySid","field","createSidString","sfC","sfB","cleanedSfC","removeSourcePrefix","normalizeSidSubfieldValue","cleanedSfB","concat","subfieldValue","sourcePrefixRegex","normalizedValue","replace","normalizeAwayRegex","bibMelindaIds","melindaIds","bibHostComponents","id","getHostId","value","subfields","find","code","test","bibTitle","title","getTitle","booleanStartWords","formatted","trim","slice","useWordSearch","some","word","toLowerCase","startsWith","includes","join","bibStandardIdentifiers","fields","identifiers","toIdentifiers","uniqueIdentifiers","Set","tag","issnIsbnReqExp","otherIdReqExp","sub","undefined"],"mappings":";;;;;;;;;;;AA4BA;;AACA;;AACA;;;;AA7BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAMO,SAASA,YAAT,CAAsBC,MAAtB,EAA8B;AAEnC;AACF;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAQE,QAAMC,KAAK,GAAG,oBAAkB,qEAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB,CA3BmC,CA4BnC;;AAEAF,EAAAA,KAAK,CAAE,iCAAF,CAAL;AAEA,QAAMG,KAAK,GAAGJ,MAAM,CAACK,GAAP,CAAW,KAAX,CAAd;AACAH,EAAAA,SAAS,CAAE,eAAcE,KAAK,CAACE,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeJ,KAAf,CAAsB,EAAxD,CAAT;AAEA,SAAOA,KAAK,CAACE,MAAN,GAAe,CAAf,GAAmBG,YAAY,CAACL,KAAD,CAA/B,GAAyC,EAAhD;;AAEA,WAASK,YAAT,CAAsBL,KAAtB,EAA6B;AAC3BH,IAAAA,KAAK,CAAE,wCAAF,CAAL;AAEA,UAAMS,UAAU,GAAGC,aAAa,CAACP,KAAD,CAAhC;;AAEA,QAAIM,UAAU,CAACJ,MAAX,GAAoB,CAAxB,EAA2B;AACzBL,MAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,aAAO,EAAP;AACD;;AAED,UAAMW,UAAU,GAAG,qCAAUF,UAAV,EAAsB,kBAAtB,CAAnB;AAEA,WAAOE,UAAP;;AAEA,aAASD,aAAT,CAAuBP,KAAvB,EAA8B;AAC5BH,MAAAA,KAAK,CAAE,qCAAF,CAAL,CAD4B,CAG5B;;AACA,YAAMS,UAAU,GAAGN,KAAK,CAACS,GAAN,CAAUC,WAAV,EAAuBC,MAAvB,CAA8BC,WAAW,IAAIA,WAA7C,CAAnB;AACA,aAAON,UAAP;;AAEA,eAASI,WAAT,CAAqBG,KAArB,EAA4B;AAC1BhB,QAAAA,KAAK,CAAE,6BAAF,CAAL;AAEA,eAAO,mDAA+BgB,KAA/B,IAAwCC,eAAe,CAACD,KAAD,CAAvD,GAAiE,EAAxE;;AAEA,iBAASC,eAAT,CAAyBD,KAAzB,EAAgC;AAC9BhB,UAAAA,KAAK,CAAE,8BAAF,CAAL;AACA,gBAAM,CAACkB,GAAD,IAAQ,sCAAkBF,KAAlB,EAAyB,GAAzB,CAAd;AACA,gBAAM,CAACG,GAAD,IAAQ,sCAAkBH,KAAlB,EAAyB,GAAzB,CAAd;AAEA,gBAAMI,UAAU,GAAGC,kBAAkB,CAACC,yBAAyB,CAACJ,GAAD,CAA1B,CAArC;AACA,gBAAMK,UAAU,GAAGD,yBAAyB,CAACH,GAAD,CAA5C;AAEAlB,UAAAA,SAAS,CAAE,GAAEK,IAAI,CAACC,SAAL,CAAeW,GAAf,CAAoB,MAAKZ,IAAI,CAACC,SAAL,CAAeY,GAAf,CAAoB,EAAjD,CAAT;AACA,iBAAOC,UAAU,CAACI,MAAX,CAAkBD,UAAlB,CAAP;AACD;;AAED,iBAASF,kBAAT,CAA4BI,aAA5B,EAA2C;AACzC,gBAAMC,iBAAiB,GAAI,gDAA3B;AACA,gBAAMC,eAAe,GAAGF,aAAa,CAACG,OAAd,CAAsBF,iBAAtB,EAAyC,OAAzC,CAAxB;AACAzB,UAAAA,SAAS,CAAE,cAAawB,aAAc,OAAME,eAAgB,EAAnD,CAAT;AACA,iBAAOA,eAAP;AACD;;AAED,iBAASL,yBAAT,CAAmCG,aAAnC,EAAkD;AAChDxB,UAAAA,SAAS,CAAE,eAAcwB,aAAc,EAA9B,CAAT;AACA,gBAAMI,kBAAkB,GAAI,OAA5B;AACA,iBAAOJ,aAAa,CAACG,OAAd,CAAsBC,kBAAtB,EAA0C,EAA1C,CAAP;AACD;AAEF;AACF;AACF;AACF;;AAEM,SAASC,aAAT,CAAuB/B,MAAvB,EAA+B;AACpC;AAEA,QAAMC,KAAK,GAAG,oBAAkB,wEAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AACAF,EAAAA,KAAK,CAAE,iCAAF,CAAL,CALoC,CAOpC;AACA;AACA;;AACA,QAAM+B,UAAU,GAAG,sCAAkBhC,MAAlB,CAAnB;AAEAE,EAAAA,SAAS,CAAE,uBAAsB8B,UAAU,CAAC1B,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAewB,UAAf,CAA2B,EAA1E,CAAT;;AAEA,MAAIA,UAAU,CAAC1B,MAAX,GAAoB,CAAxB,EAA2B;AACzBL,IAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,WAAO,EAAP;AACD;;AAED,SAAO,qCAAU+B,UAAV,EAAsB,mBAAtB,CAAP;AACD,C,CAGD;AACA;;;AACO,SAASC,iBAAT,CAA2BjC,MAA3B,EAAmC;AACxC,QAAMkC,EAAE,GAAGC,SAAS,EAApB;AACA,SAAOD,EAAE,GAAG,CAAE,uBAAsBA,EAAG,EAA3B,CAAH,GAAmC,EAA5C;;AAEA,WAASC,SAAT,GAAqB;AACnB,UAAM,CAAClB,KAAD,IAAUjB,MAAM,CAACK,GAAP,CAAW,QAAX,CAAhB;;AAEA,QAAIY,KAAJ,EAAW;AACT,YAAM;AAACmB,QAAAA;AAAD,UAAUnB,KAAK,CAACoB,SAAN,CAAgBC,IAAhB,CAAqB,CAAC;AAACC,QAAAA;AAAD,OAAD,KAAYA,IAAI,KAAK,GAA1C,KAAkD,EAAlE;;AAEA,UAAIH,KAAK,IAAK,kBAAD,CAAqBI,IAArB,CAA0BJ,KAA1B,CAAb,EAA+C;AAC7C,eAAOA,KAAK,CAACP,OAAN,CAAc,kBAAd,EAAkC,EAAlC,CAAP;AACD;;AAED,UAAIO,KAAK,IAAK,aAAD,CAAgBI,IAAhB,CAAqBJ,KAArB,CAAb,EAA0C;AACxC,eAAOA,KAAK,CAACP,OAAN,CAAc,aAAd,EAA6B,EAA7B,CAAP;AACD;;AAED,aAAO,KAAP;AACD;;AACD,WAAO,KAAP;AACD;AACF,C,CAED;AACA;AACA;AAEA;AACA;AACA;;;AAEO,SAASY,QAAT,CAAkBzC,MAAlB,EAA0B;AAC/B,QAAM0C,KAAK,GAAGC,QAAQ,EAAtB;AACA,QAAMC,iBAAiB,GAAG,CAAC,KAAD,EAAQ,IAAR,EAAc,KAAd,EAAqB,KAArB,CAA1B;;AAEA,MAAIF,KAAJ,EAAW;AACT,UAAMG,SAAS,GAAGH,KAAK,CACpBb,OADe,CACP,yBADO,EACoB,EADpB,EAEhB;AAFgB,KAGfA,OAHe,CAGP,MAHO,EAGC,GAHD,EAIfiB,IAJe,GAKfC,KALe,CAKT,CALS,EAKN,EALM,EAMfD,IANe,EAAlB,CADS,CAST;;AACA,UAAME,aAAa,GAAGJ,iBAAiB,CAACK,IAAlB,CAAuBC,IAAI,IAAIL,SAAS,CAACM,WAAV,GAAwBC,UAAxB,CAAmCF,IAAnC,CAA/B,CAAtB,CAVS,CAWT;;AACA,WAAOL,SAAS,CAACvC,MAAV,IAAoB,CAApB,GAAwB,CAAE,aAAY0C,aAAa,GAAG,EAAH,GAAQ,GAAI,GAAEH,SAAU,IAAnD,CAAxB,GAAkF,EAAzF;AACD;;AAED,SAAO,EAAP;;AAEA,WAASF,QAAT,GAAoB;AAClB,UAAM,CAAC1B,KAAD,IAAUjB,MAAM,CAACK,GAAP,CAAW,QAAX,CAAhB;;AAEA,QAAIY,KAAJ,EAAW;AACT,aAAOA,KAAK,CAACoB,SAAN,CACJtB,MADI,CACG,CAAC;AAACwB,QAAAA;AAAD,OAAD,KAAY,CAAC,GAAD,EAAM,GAAN,EAAWc,QAAX,CAAoBd,IAApB,CADf,EAEJ1B,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,EAGL;AAHK,OAIJkB,IAJI,CAIC,GAJD,CAAP;AAKD;;AACD,WAAO,KAAP;AACD;AACF,C,CAED;;;AACO,SAASC,sBAAT,CAAgCvD,MAAhC,EAAwC;AAE7C,QAAMC,KAAK,GAAG,oBAAkB,iFAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AACAF,EAAAA,KAAK,CAAE,2CAAF,CAAL;AAEA,QAAMuD,MAAM,GAAGxD,MAAM,CAACK,GAAP,CAAW,wBAAX,CAAf;AACA,QAAMoD,WAAW,GAAG,GAAGhC,MAAH,CAAU,GAAG+B,MAAM,CAAC3C,GAAP,CAAW6C,aAAX,CAAb,CAApB;AACA,QAAMC,iBAAiB,GAAG,CAAC,GAAG,IAAIC,GAAJ,CAAQH,WAAR,CAAJ,CAA1B;AAEAvD,EAAAA,SAAS,CAAE,+BAA8BK,IAAI,CAACC,SAAL,CAAegD,MAAf,CAAuB,EAAvD,CAAT;AACAtD,EAAAA,SAAS,CAAE,gBAAeuD,WAAW,CAACnD,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAeiD,WAAf,CAA4B,EAArE,CAAT;AACAvD,EAAAA,SAAS,CAAE,uBAAsByD,iBAAiB,CAACrD,MAAO,MAAKC,IAAI,CAACC,SAAL,CAAemD,iBAAf,CAAkC,EAAxF,CAAT;;AAEA,MAAIA,iBAAiB,CAACrD,MAAlB,GAA2B,CAA/B,EAAkC;AAChCL,IAAAA,KAAK,CAAE,2CAAF,CAAL;AACA,WAAO,EAAP;AACD;;AAED,SAAO,qCAAU0D,iBAAV,EAA6B,eAA7B,CAAP;;AAEA,WAASD,aAAT,CAAuB;AAACG,IAAAA,GAAD;AAAMxB,IAAAA;AAAN,GAAvB,EAAyC;AACvC,UAAMyB,cAAc,GAAI,kBAAxB;AACA,UAAMC,aAAa,GAAI,mBAAvB;;AAEA,QAAIF,GAAG,KAAK,KAAZ,EAAmB;AACjB,aAAOxB,SAAS,CACbtB,MADI,CACGiD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAW,GAAX,EAAgBX,QAAhB,CAAyBW,GAAG,CAACzB,IAA7B,KAAsCuB,cAAc,CAACtB,IAAf,CAAoBwB,GAAG,CAAC5B,KAAxB,CAAtC,IAAwE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SADhG,EAEJpD,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,CAAP;AAGD;;AAED,QAAIyB,GAAG,KAAK,KAAZ,EAAmB;AACjB,aAAOxB,SAAS,CACbtB,MADI,CACGiD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiCuB,cAAc,CAACtB,IAAf,CAAoBwB,GAAG,CAAC5B,KAAxB,CAAjC,IAAmE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SAD3F,EAEJpD,GAFI,CAEA,CAAC;AAACuB,QAAAA;AAAD,OAAD,KAAaA,KAFb,CAAP;AAGD;;AAED,WAAOC,SAAS,CACbtB,MADI,CACGiD,GAAG,IAAI,CAAC,GAAD,EAAM,GAAN,EAAWX,QAAX,CAAoBW,GAAG,CAACzB,IAAxB,KAAiCwB,aAAa,CAACvB,IAAd,CAAmBwB,GAAG,CAAC5B,KAAvB,CAAjC,IAAkE4B,GAAG,CAAC5B,KAAJ,KAAc6B,SAD1F,EAEJpD,GAFI,CAEA,CAAC;AAACuB,MAAAA;AAAD,KAAD,KAAaA,KAFb,CAAP;AAGD;AACF","sourcesContent":["\n/**\n*\n* @licstart  The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program.  If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend  The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\nimport createDebugLogger from 'debug';\nimport {toQueries} from '../candidate-search-utils';\nimport {getMelindaIdsF035, validateSidFieldSubfieldCounts, getSubfieldValues} from '../../matching-utils';\n\n\nexport function bibSourceIds(record) {\n\n  /* Melinda's SRU-index melinda.sourceid includes source IDs from SID fields in Melinda records\n     SID-fields in Melinda have sf $c with local id and sf $b with a code for the local db:\n\n     SID__ $c 123457 $b helka\n     SID__ $c (ANDL100020)1077305 $b sata\n     SID__ $c VER999999 $ FI-KV\n     SID__ $c /10024/508126 $ REPO_THESEUS\n\n    In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:\n    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted\n\n    1234567helka\n    1077305sata\n    VER999999FIKV\n    /10024/508126REPO_THESEUS\n\n    Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,\n          existence of a SID field depends on how the record has been added to Melinda and how it has been handled\n          afterwards. SIDs are also not reliably maintained. A record might or might not have a SID for a local db\n          after the matching record is removed from the local db.\n\n   */\n\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');\n  const debugData = debug.extend('data');\n  //const debugInfo = debug.extend('info');\n\n  debug(`Creating queries for sourceid's`);\n\n  const fSids = record.get('SID');\n  debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);\n\n  return fSids.length > 0 ? toSidQueries(fSids) : [];\n\n  function toSidQueries(fSids) {\n    debug(`Creating actual queries for sourceid's`);\n\n    const sidStrings = getSidStrings(fSids);\n\n    if (sidStrings.length < 1) {\n      debug(`No identifiers found, no queries created.`);\n      return [];\n    }\n\n    const sidQueries = toQueries(sidStrings, 'melinda.sourceid');\n\n    return sidQueries;\n\n    function getSidStrings(fSids) {\n      debug(`Getting Sid strings from SID fields`);\n\n      // Map SID fields to valid sidStrings, filter out empty strings\n      const sidStrings = fSids.map(toSidString).filter(nonEmptySid => nonEmptySid);\n      return sidStrings;\n\n      function toSidString(field) {\n        debug(`Getting string from a field`);\n\n        return validateSidFieldSubfieldCounts(field) ? createSidString(field) : '';\n\n        function createSidString(field) {\n          debug(`Creating string from a field`);\n          const [sfC] = getSubfieldValues(field, 'c');\n          const [sfB] = getSubfieldValues(field, 'b');\n\n          const cleanedSfC = removeSourcePrefix(normalizeSidSubfieldValue(sfC));\n          const cleanedSfB = normalizeSidSubfieldValue(sfB);\n\n          debugData(`${JSON.stringify(sfC)} + ${JSON.stringify(sfB)}`);\n          return cleanedSfC.concat(cleanedSfB);\n        }\n\n        function removeSourcePrefix(subfieldValue) {\n          const sourcePrefixRegex = (/^(?<sourcePrefix>\\([A-Za-z0-9-]+\\))(?<id>.+)$/u);\n          const normalizedValue = subfieldValue.replace(sourcePrefixRegex, '$<id>');\n          debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n          return normalizedValue;\n        }\n\n        function normalizeSidSubfieldValue(subfieldValue) {\n          debugData(`Normalizing ${subfieldValue}`);\n          const normalizeAwayRegex = (/[- ]/u);\n          return subfieldValue.replace(normalizeAwayRegex, '');\n        }\n\n      }\n    }\n  }\n}\n\nexport function bibMelindaIds(record) {\n  // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for MelindaIds`);\n\n  // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n  // Both (FI-MELINDA)- and FCC-prefixed forms are found\n  // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n  const melindaIds = getMelindaIdsF035(record);\n\n  debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n  if (melindaIds.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n  const id = getHostId();\n  return id ? [`melinda.partsofhost=${id}`] : [];\n\n  function getHostId() {\n    const [field] = record.get(/^773$/u);\n\n    if (field) {\n      const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n      if (value && (/^\\(FI-MELINDA\\)/u).test(value)) {\n        return value.replace(/^\\(FI-MELINDA\\)/u, '');\n      }\n\n      if (value && (/^\\(FIN01\\)/u).test(value)) {\n        return value.replace(/^\\(FIN01\\)/u, '');\n      }\n\n      return false;\n    }\n    return false;\n  }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n  const title = getTitle();\n  const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n  if (title) {\n    const formatted = title\n      .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n      // Clean up concurrent spaces from fe. subfield changes\n      .replace(/ +/gu, ' ')\n      .trim()\n      .slice(0, 30)\n      .trim();\n\n    // use word search for titles starting with a boolean\n    const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n    // Prevent too many matches by having a minimum length requirement\n    return formatted.length >= 5 ? [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`] : [];\n  }\n\n  return [];\n\n  function getTitle() {\n    const [field] = record.get(/^245$/u);\n\n    if (field) {\n      return field.subfields\n        .filter(({code}) => ['a', 'b'].includes(code))\n        .map(({value}) => value)\n        // In Melinda's index subfield separators are indexed as ' '\n        .join(' ');\n    }\n    return false;\n  }\n}\n\n// eslint-disable-next-line max-statements\nexport function bibStandardIdentifiers(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n  const debugData = debug.extend('data');\n  debug(`Creating queries for standard identifiers`);\n\n  const fields = record.get(/^(?<def>020|022|024)$/u);\n  const identifiers = [].concat(...fields.map(toIdentifiers));\n  const uniqueIdentifiers = [...new Set(identifiers)];\n\n  debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n  debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n  debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n  if (uniqueIdentifiers.length < 1) {\n    debug(`No identifiers found, no queries created.`);\n    return [];\n  }\n\n  return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n  function toIdentifiers({tag, subfields}) {\n    const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n    const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n    if (tag === '022') {\n      return subfields\n        .filter(sub => ['a', 'z', 'y'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    if (tag === '020') {\n      return subfields\n        .filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n        .map(({value}) => value);\n    }\n\n    return subfields\n      .filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined)\n      .map(({value}) => value);\n  }\n}\n"],"file":"bib.js"}