@natlibfi/melinda-record-matching 2.1.0 → 2.2.0-alpha.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (31) hide show
  1. package/dist/candidate-search/index.js +40 -10
  2. package/dist/candidate-search/index.js.map +1 -1
  3. package/dist/candidate-search/index.spec.js +16 -2
  4. package/dist/candidate-search/index.spec.js.map +1 -1
  5. package/dist/candidate-search/query-list/bib.js +2 -2
  6. package/dist/candidate-search/query-list/bib.js.map +1 -1
  7. package/dist/index.js +139 -39
  8. package/dist/index.js.map +1 -1
  9. package/dist/index.spec.js +12 -6
  10. package/dist/index.spec.js.map +1 -1
  11. package/dist/match-detection/features/bib/authors.js +1 -1
  12. package/dist/match-detection/features/bib/authors.js.map +1 -1
  13. package/dist/match-detection/features/bib/standard-identifier-factory.js +3 -2
  14. package/dist/match-detection/features/bib/standard-identifier-factory.js.map +1 -1
  15. package/dist/match-detection/index.js +36 -30
  16. package/dist/match-detection/index.js.map +1 -1
  17. package/dist/match-detection/index.spec.js +20 -2
  18. package/dist/match-detection/index.spec.js.map +1 -1
  19. package/dist/matching-utils.js +1 -1
  20. package/dist/matching-utils.js.map +1 -1
  21. package/package.json +16 -15
  22. package/src/candidate-search/index.js +36 -14
  23. package/src/candidate-search/index.spec.js +13 -3
  24. package/src/candidate-search/query-list/bib.js +2 -2
  25. package/src/index.js +94 -35
  26. package/src/index.spec.js +10 -6
  27. package/src/match-detection/features/bib/authors.js +1 -1
  28. package/src/match-detection/features/bib/standard-identifier-factory.js +2 -1
  29. package/src/match-detection/index.js +29 -19
  30. package/src/match-detection/index.spec.js +17 -3
  31. package/src/matching-utils.js +1 -1
@@ -21,6 +21,8 @@ var _marcRecordSerializers = require("@natlibfi/marc-record-serializers");
21
21
 
22
22
  var _queryList = _interopRequireWildcard(require("./query-list"));
23
23
 
24
+ var _melindaCommons = require("@natlibfi/melinda-commons");
25
+
24
26
  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); }
25
27
 
26
28
  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; }
@@ -90,7 +92,7 @@ var _default = ({
90
92
  retrieveAll: false
91
93
  });
92
94
  debug(`Searching matches for ${inputRecordId}`);
93
- debug(`Generated queryList ${JSON.stringify(queryList)}`);
95
+ debug(`Generated queryList ${JSON.stringify(queryList)}`); // if generateQueryList errored we should throw 422
94
96
 
95
97
  if (queryList.length === 0) {
96
98
  // eslint-disable-line functional/no-conditional-statement
@@ -98,11 +100,10 @@ var _default = ({
98
100
  } // state.totalRecords : amount of candidate records available to the current query (undefined, if there was no queries left)
99
101
  // state.query : current query (undefined if there was no queries left)
100
102
  // state.searchCounter : sequence for current search for current query (undefined, if there we no queries left)
101
- // 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)
103
+ // state.queryCandidateCounter: amount of candidates (records+failures) retrieved from SRU for matching for current query, including the current record+failure set (undefined if there were no queries left)
102
104
  // state.queriesLeft : amount of queries left
103
105
  // state.queryCounter : sequence for current query
104
106
  // state.maxedQueries : queries that resulted in more than serverMaxResults hits
105
- // eslint-disable-next-line max-statements
106
107
 
107
108
 
108
109
  return async ({
@@ -119,6 +120,7 @@ var _default = ({
119
120
  if (query) {
120
121
  const {
121
122
  records,
123
+ failures,
122
124
  nextOffset,
123
125
  total
124
126
  } = await retrieveRecords(); // If resultSetOffset === 1 this is the first search for the current query
@@ -127,7 +129,7 @@ var _default = ({
127
129
  const newTotalRecords = resultSetOffset === 1 ? total : totalRecords;
128
130
  const newQueryCounter = resultSetOffset === 1 ? queryCounter + 1 : queryCounter;
129
131
  const newSearchCounter = resultSetOffset === 1 ? 1 : searchCounter + 1;
130
- const newQueryCandidateCounter = resultSetOffset === 1 ? records.length : queryCandidateCounter + records.length;
132
+ const newQueryCandidateCounter = resultSetOffset === 1 ? records.length + failures.length : queryCandidateCounter + records.length + failures.length;
131
133
  const maxedQuery = resultSetOffset === 1 ? checkMaxedQuery(query, total, serverMaxResult) : undefined;
132
134
  const newMaxedQueries = maxedQuery ? maxedQueries.concat(maxedQuery) : maxedQueries;
133
135
 
@@ -135,6 +137,7 @@ var _default = ({
135
137
  debug(`Next search will be for query ${queryOffset} ${query}, starting from record ${nextOffset}`);
136
138
  return {
137
139
  records,
140
+ failures,
138
141
  queryOffset,
139
142
  resultSetOffset: nextOffset,
140
143
  queriesLeft: queryList.length - (queryOffset + 1),
@@ -151,6 +154,7 @@ var _default = ({
151
154
  debug(`There are (${queryList.length - (queryOffset + 1)} queries left)`);
152
155
  return {
153
156
  records,
157
+ failures,
154
158
  queryOffset: queryOffset + 1,
155
159
  queriesLeft: queryList.length - (queryOffset + 1),
156
160
  totalRecords: newTotalRecords,
@@ -165,6 +169,7 @@ var _default = ({
165
169
  debug(`All ${queryList.length} queries done, there's no query for ${queryOffset}`);
166
170
  return {
167
171
  records: [],
172
+ failures: [],
168
173
  queriesLeft: 0,
169
174
  queryCounter,
170
175
  maxedQueries
@@ -192,23 +197,34 @@ var _default = ({
192
197
  totalRecords += total;
193
198
  }).on('end', async nextOffset => {
194
199
  try {
195
- const records = await Promise.all(promises);
196
- const filtered = records.filter(r => r);
197
- debug(`Found ${filtered.length} candidates`);
200
+ const recordPromises = await Promise.allSettled(promises);
201
+ debug(`All recordPromises: ${JSON.stringify(recordPromises)}`);
202
+ const filtered = recordPromises.filter(r => r.status === 'fulfilled').map(r => r.value);
203
+ const failures = recordPromises.filter(r => r.status === 'rejected').map(r => ({
204
+ status: r.reason.status,
205
+ payload: r.reason.payload
206
+ }));
207
+ debug(`Found ${JSON.stringify(recordPromises)} records`);
208
+ debug(`Found ${filtered.length} convertable candidates`);
209
+ debug(`Found ${failures.length} NON-convertable candidates`);
210
+ debug(`Converted: ${JSON.stringify(filtered)}.`);
211
+ debug(`Not converted: ${JSON.stringify(failures)}.`);
198
212
  resolve({
199
213
  nextOffset,
200
214
  records: filtered,
215
+ failures,
201
216
  total: totalRecords
202
217
  });
203
218
  } catch (err) {
219
+ debug(`Error caught on END`);
204
220
  reject(err);
205
221
  }
206
- }).on('record', foundRecord => {
222
+ }).on('record', foundRecordXML => {
207
223
  promises.push(handleRecord()); // eslint-disable-line functional/immutable-data
208
224
 
209
225
  async function handleRecord() {
210
226
  try {
211
- const foundRecordMarc = await _marcRecordSerializers.MARCXML.from(foundRecord, {
227
+ const foundRecordMarc = await _marcRecordSerializers.MARCXML.from(foundRecordXML, {
212
228
  subfieldValues: false
213
229
  });
214
230
  const foundRecordId = getRecordId(foundRecordMarc);
@@ -217,7 +233,15 @@ var _default = ({
217
233
  id: foundRecordId
218
234
  };
219
235
  } catch (err) {
220
- throw new Error(`Failed converting record: ${err}, record: ${foundRecord}`);
236
+ // What should this do?
237
+ const idFromXML = getRecordIdFromXML(foundRecordXML);
238
+ debugData(`Failed converting record: ${err.message}, id: ${idFromXML}, data: ${foundRecordXML}`); //return {message: `Failed converting record: ${err.message}`, id: idFromXML, data: foundRecordXML};
239
+
240
+ throw new _melindaCommons.Error(422, {
241
+ message: `Failed converting record: ${err.message}`,
242
+ id: idFromXML || '000000000',
243
+ data: foundRecordXML
244
+ });
221
245
  }
222
246
  }
223
247
  });
@@ -237,6 +261,12 @@ var _default = ({
237
261
  const [field] = record.get(/^001$/u);
238
262
  return field ? field.value : '';
239
263
  }
264
+
265
+ function getRecordIdFromXML(recordXML) {
266
+ //<controlfield tag=\"001\">015376846</controlfield
267
+ debug(`Cannot yet find possible database record id from recordXML (length ${recordXML.length})`);
268
+ return undefined;
269
+ }
240
270
  };
241
271
 
242
272
  exports.default = _default;
@@ -1 +1 @@
1
- {"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"}
1
+ {"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","failures","nextOffset","total","retrieveRecords","newTotalRecords","newQueryCounter","newSearchCounter","newQueryCandidateCounter","maxedQuery","checkMaxedQuery","undefined","newMaxedQueries","concat","queriesLeft","Promise","resolve","reject","promises","searchRetrieve","startRecord","on","err","SruSearchError","recordPromises","allSettled","filtered","filter","r","status","map","value","reason","payload","foundRecordXML","push","handleRecord","foundRecordMarc","MARCXML","from","foundRecordId","id","idFromXML","getRecordIdFromXML","message","MatchingError","data","field","get","recordXML"],"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;AAWO,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,CAzB+G,CA2B/G;;AAEA,MAAIA,SAAS,CAACI,MAAV,KAAqB,CAAzB,EAA4B;AAAE;AAC5B,UAAM,IAAIvB,oBAAJ,CAA0B,0CAA1B,CAAN;AACD,GA/B8G,CAiC/G;AACA;AACA;AACA;AACA;AACA;AACA;;;AAGA,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,QAAV;AAAoBC,QAAAA,UAApB;AAAgCC,QAAAA;AAAhC,UAAyC,MAAMC,eAAe,EAApE,CADS,CAGT;;AACAxB,MAAAA,SAAS,CAAE,oBAAmBa,eAAgB,EAArC,CAAT;AACA,YAAMY,eAAe,GAAGZ,eAAe,KAAK,CAApB,GAAwBU,KAAxB,GAAgCT,YAAxD;AACA,YAAMY,eAAe,GAAGb,eAAe,KAAK,CAApB,GAAwBI,YAAY,GAAG,CAAvC,GAA2CA,YAAnE;AACA,YAAMU,gBAAgB,GAAGd,eAAe,KAAK,CAApB,GAAwB,CAAxB,GAA4BE,aAAa,GAAG,CAArE;AACA,YAAMa,wBAAwB,GAAGf,eAAe,KAAK,CAApB,GAAwBO,OAAO,CAACT,MAAR,GAAiBU,QAAQ,CAACV,MAAlD,GAA2DK,qBAAqB,GAAGI,OAAO,CAACT,MAAhC,GAAyCU,QAAQ,CAACV,MAA9I;AAEA,YAAMkB,UAAU,GAAGhB,eAAe,KAAK,CAApB,GAAwBiB,eAAe,CAACX,KAAD,EAAQI,KAAR,EAAe5B,eAAf,CAAvC,GAAyEoC,SAA5F;AACA,YAAMC,eAAe,GAAGH,UAAU,GAAGX,YAAY,CAACe,MAAb,CAAoBJ,UAApB,CAAH,GAAqCX,YAAvE;;AAEA,UAAI,OAAOI,UAAP,KAAsB,QAA1B,EAAoC;AAClCvB,QAAAA,KAAK,CAAE,iCAAgCa,WAAY,IAAGO,KAAM,0BAAyBG,UAAW,EAA3F,CAAL;AACA,eAAO;AAACF,UAAAA,OAAD;AAAUC,UAAAA,QAAV;AAAoBT,UAAAA,WAApB;AAAiCC,UAAAA,eAAe,EAAES,UAAlD;AAA8DY,UAAAA,WAAW,EAAE3B,SAAS,CAACI,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAA3E;AAAiHE,UAAAA,YAAY,EAAEW,eAA/H;AAAgJN,UAAAA,KAAhJ;AAAuJJ,UAAAA,aAAa,EAAEY,gBAAtK;AAAwLX,UAAAA,qBAAqB,EAAEY,wBAA/M;AAAyOX,UAAAA,YAAY,EAAES,eAAvP;AAAwQR,UAAAA,YAAY,EAAEc;AAAtR,SAAP;AACD;;AACDjC,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;AAAUC,QAAAA,QAAV;AAAoBT,QAAAA,WAAW,EAAEA,WAAW,GAAG,CAA/C;AAAkDsB,QAAAA,WAAW,EAAE3B,SAAS,CAACI,MAAV,IAAoBC,WAAW,GAAG,CAAlC,CAA/D;AAAqGE,QAAAA,YAAY,EAAEW,eAAnH;AAAoIN,QAAAA,KAApI;AAA2IJ,QAAAA,aAAa,EAAEY,gBAA1J;AAA4KX,QAAAA,qBAAqB,EAAEY,wBAAnM;AAA6NX,QAAAA,YAAY,EAAES,eAA3O;AAA4PR,QAAAA,YAAY,EAAEc;AAA1Q,OAAP;AACD;;AAEDjC,IAAAA,KAAK,CAAE,OAAMQ,SAAS,CAACI,MAAO,uCAAsCC,WAAY,EAA3E,CAAL;AACA,WAAO;AAACQ,MAAAA,OAAO,EAAE,EAAV;AAAcC,MAAAA,QAAQ,EAAE,EAAxB;AAA4Ba,MAAAA,WAAW,EAAE,CAAzC;AAA4CjB,MAAAA,YAA5C;AAA0DC,MAAAA;AAA1D,KAAP;;AAEA,aAASM,eAAT,GAA2B;AACzB,aAAO,IAAIW,OAAJ,CAAY,CAACC,OAAD,EAAUC,MAAV,KAAqB;AACtC,cAAMC,QAAQ,GAAG,EAAjB,CADsC,CAEtC;;AACA,YAAIxB,YAAY,GAAG,CAAnB;AAEAf,QAAAA,KAAK,CAAE,wCAAuCoB,KAAM,YAAWN,eAAgB,GAA1E,CAAL;AAEAL,QAAAA,MAAM,CAAC+B,cAAP,CAAsBpB,KAAtB,EAA6B;AAACqB,UAAAA,WAAW,EAAE3B;AAAd,SAA7B,EACG4B,EADH,CACM,OADN,EACeC,GAAG,IAAI;AAClB;AACA,cAAIA,GAAG,YAAYC,yBAAnB,EAAmC;AACjC5C,YAAAA,KAAK,CAAE,iCAAgCoB,KAAM,KAAIuB,GAAI,EAAhD,CAAL;AACAL,YAAAA,MAAM,CAAC,IAAIjD,oBAAJ,CAA0B,iCAAgC+B,KAAM,KAAIuB,GAAI,EAAxE,CAAD,CAAN;AACD;;AACD3C,UAAAA,KAAK,CAAE,wBAAuBoB,KAAM,KAAIuB,GAAI,EAAvC,CAAL;AACAL,UAAAA,MAAM,CAAC,IAAIjD,oBAAJ,CAA0B,wBAAuB+B,KAAM,KAAIuB,GAAI,EAA/D,CAAD,CAAN;AACD,SATH,EAUGD,EAVH,CAUM,OAVN,EAUelB,KAAK,IAAI;AACpBxB,UAAAA,KAAK,CAAE,cAAawB,KAAM,EAArB,CAAL;AACAT,UAAAA,YAAY,IAAIS,KAAhB;AACD,SAbH,EAcGkB,EAdH,CAcM,KAdN,EAca,MAAMnB,UAAN,IAAoB;AAC7B,cAAI;AACF,kBAAMsB,cAAc,GAAG,MAAMT,OAAO,CAACU,UAAR,CAAmBP,QAAnB,CAA7B;AACAvC,YAAAA,KAAK,CAAE,uBAAsBG,IAAI,CAACC,SAAL,CAAeyC,cAAf,CAA+B,EAAvD,CAAL;AACA,kBAAME,QAAQ,GAAGF,cAAc,CAACG,MAAf,CAAsBC,CAAC,IAAIA,CAAC,CAACC,MAAF,KAAa,WAAxC,EAAqDC,GAArD,CAAyDF,CAAC,IAAIA,CAAC,CAACG,KAAhE,CAAjB;AACA,kBAAM9B,QAAQ,GAAGuB,cAAc,CAACG,MAAf,CAAsBC,CAAC,IAAIA,CAAC,CAACC,MAAF,KAAa,UAAxC,EAAoDC,GAApD,CAAwDF,CAAC,KAAK;AAACC,cAAAA,MAAM,EAAED,CAAC,CAACI,MAAF,CAASH,MAAlB;AAA0BI,cAAAA,OAAO,EAAEL,CAAC,CAACI,MAAF,CAASC;AAA5C,aAAL,CAAzD,CAAjB;AAEAtD,YAAAA,KAAK,CAAE,SAAQG,IAAI,CAACC,SAAL,CAAeyC,cAAf,CAA+B,UAAzC,CAAL;AACA7C,YAAAA,KAAK,CAAE,SAAQ+C,QAAQ,CAACnC,MAAO,yBAA1B,CAAL;AACAZ,YAAAA,KAAK,CAAE,SAAQsB,QAAQ,CAACV,MAAO,6BAA1B,CAAL;AACAZ,YAAAA,KAAK,CAAE,cAAaG,IAAI,CAACC,SAAL,CAAe2C,QAAf,CAAyB,GAAxC,CAAL;AACA/C,YAAAA,KAAK,CAAE,kBAAiBG,IAAI,CAACC,SAAL,CAAekB,QAAf,CAAyB,GAA5C,CAAL;AAGAe,YAAAA,OAAO,CAAC;AAACd,cAAAA,UAAD;AAAaF,cAAAA,OAAO,EAAE0B,QAAtB;AAAgCzB,cAAAA,QAAhC;AAA0CE,cAAAA,KAAK,EAAET;AAAjD,aAAD,CAAP;AACD,WAdD,CAcE,OAAO4B,GAAP,EAAY;AACZ3C,YAAAA,KAAK,CAAE,qBAAF,CAAL;AACAsC,YAAAA,MAAM,CAACK,GAAD,CAAN;AACD;AACF,SAjCH,EAkCGD,EAlCH,CAkCM,QAlCN,EAkCgBa,cAAc,IAAI;AAC9BhB,UAAAA,QAAQ,CAACiB,IAAT,CAAcC,YAAY,EAA1B,EAD8B,CACC;;AAE/B,yBAAeA,YAAf,GAA8B;AAC5B,gBAAI;AACF,oBAAMC,eAAe,GAAG,MAAMC,+BAAQC,IAAR,CAAaL,cAAb,EAA6B;AAACxD,gBAAAA,cAAc,EAAE;AAAjB,eAA7B,CAA9B;AACA,oBAAM8D,aAAa,GAAGtD,WAAW,CAACmD,eAAD,CAAjC;AAEA,qBAAO;AAACnE,gBAAAA,MAAM,EAAEmE,eAAT;AAA0BI,gBAAAA,EAAE,EAAED;AAA9B,eAAP;AACD,aALD,CAKE,OAAOlB,GAAP,EAAY;AACZ;AACA,oBAAMoB,SAAS,GAAGC,kBAAkB,CAACT,cAAD,CAApC;AACAtD,cAAAA,SAAS,CAAE,6BAA4B0C,GAAG,CAACsB,OAAQ,SAAQF,SAAU,WAAUR,cAAe,EAArF,CAAT,CAHY,CAIZ;;AACA,oBAAM,IAAIW,qBAAJ,CAAkB,GAAlB,EAAuB;AAACD,gBAAAA,OAAO,EAAG,6BAA4BtB,GAAG,CAACsB,OAAQ,EAAnD;AAAsDH,gBAAAA,EAAE,EAAEC,SAAS,IAAI,WAAvE;AAAoFI,gBAAAA,IAAI,EAAEZ;AAA1F,eAAvB,CAAN;AACD;AACF;AACF,SAnDH;AAoDD,OA3DM,CAAP;AA4DD;AACF,GA1FD;;AA4FA,WAASxB,eAAT,CAAyBX,KAAzB,EAAgCI,KAAhC,EAAuC5B,eAAvC,EAAwD;AACtD;AACA,QAAI4B,KAAK,IAAI5B,eAAb,EAA8B;AAC5BI,MAAAA,KAAK,CAAE,kBAAiBoB,KAAM,gBAAeI,KAAM,0CAAyC5B,eAAgB,IAAvG,CAAL;AACA,aAAOwB,KAAP;AACD;AACF;;AAGD,WAASb,WAAT,CAAqBhB,MAArB,EAA6B;AAC3B,UAAM,CAAC6E,KAAD,IAAU7E,MAAM,CAAC8E,GAAP,CAAW,QAAX,CAAhB;AACA,WAAOD,KAAK,GAAGA,KAAK,CAAChB,KAAT,GAAiB,EAA7B;AACD;;AAED,WAASY,kBAAT,CAA4BM,SAA5B,EAAuC;AACrC;AACAtE,IAAAA,KAAK,CAAE,sEAAqEsE,SAAS,CAAC1D,MAAO,GAAxF,CAAL;AACA,WAAOoB,SAAP;AACD;AAEF,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';\nimport {Error as MatchingError} from '@natlibfi/melinda-commons';\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\n // if generateQueryList errored we should throw 422\n\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 candidates (records+failures) retrieved from SRU for matching for current query, including the current record+failure 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 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, failures, 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 + failures.length : queryCandidateCounter + records.length + failures.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, failures, 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, failures, 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: [], failures: [], 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 recordPromises = await Promise.allSettled(promises);\n debug(`All recordPromises: ${JSON.stringify(recordPromises)}`);\n const filtered = recordPromises.filter(r => r.status === 'fulfilled').map(r => r.value);\n const failures = recordPromises.filter(r => r.status === 'rejected').map(r => ({status: r.reason.status, payload: r.reason.payload}));\n\n debug(`Found ${JSON.stringify(recordPromises)} records`);\n debug(`Found ${filtered.length} convertable candidates`);\n debug(`Found ${failures.length} NON-convertable candidates`);\n debug(`Converted: ${JSON.stringify(filtered)}.`);\n debug(`Not converted: ${JSON.stringify(failures)}.`);\n\n\n resolve({nextOffset, records: filtered, failures, total: totalRecords});\n } catch (err) {\n debug(`Error caught on END`);\n reject(err);\n }\n })\n .on('record', foundRecordXML => {\n promises.push(handleRecord()); // eslint-disable-line functional/immutable-data\n\n async function handleRecord() {\n try {\n const foundRecordMarc = await MARCXML.from(foundRecordXML, {subfieldValues: false});\n const foundRecordId = getRecordId(foundRecordMarc);\n\n return {record: foundRecordMarc, id: foundRecordId};\n } catch (err) {\n // What should this do?\n const idFromXML = getRecordIdFromXML(foundRecordXML);\n debugData(`Failed converting record: ${err.message}, id: ${idFromXML}, data: ${foundRecordXML}`);\n //return {message: `Failed converting record: ${err.message}`, id: idFromXML, data: foundRecordXML};\n throw new MatchingError(422, {message: `Failed converting record: ${err.message}`, id: idFromXML || '000000000', data: foundRecordXML});\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 function getRecordIdFromXML(recordXML) {\n //<controlfield tag=\\\"001\\\">015376846</controlfield\n debug(`Cannot yet find possible database record id from recordXML (length ${recordXML.length})`);\n return undefined;\n }\n\n};\n"],"file":"index.js"}
@@ -8,6 +8,8 @@ var _fixugenHttpClient = _interopRequireDefault(require("@natlibfi/fixugen-http-
8
8
 
9
9
  var _marcRecord = require("@natlibfi/marc-record");
10
10
 
11
+ var _melindaCommons = require("@natlibfi/melinda-commons");
12
+
11
13
  var _ = _interopRequireWildcard(require("."));
12
14
 
13
15
  var _debug = _interopRequireDefault(require("debug"));
@@ -107,6 +109,7 @@ describe('candidate-search', () => {
107
109
  async function iterate({
108
110
  searchOptions,
109
111
  expectedSearchError,
112
+ expectedErrorStatus,
110
113
  count = 1
111
114
  }) {
112
115
  const expectedResults = getFixture(`expectedResults${count}.json`);
@@ -117,8 +120,18 @@ describe('candidate-search', () => {
117
120
  await search(searchOptions);
118
121
  throw new Error('Expected an error');
119
122
  } catch (err) {
123
+ debug(`Got an error: ${err}`);
120
124
  (0, _chai.expect)(err).to.be.an('error');
121
- (0, _chai.expect)(err.message).to.match(new RegExp(expectedSearchError, 'u'));
125
+ const errorMessage = err instanceof _melindaCommons.Error ? err.payload.message : err.message;
126
+ const errorStatus = err instanceof _melindaCommons.Error ? err.status : undefined;
127
+ debug(`errorMessage: ${errorMessage}, errorStatus: ${errorStatus}`);
128
+ (0, _chai.expect)(errorMessage).to.match(new RegExp(expectedSearchError, 'u'));
129
+
130
+ if (expectedErrorStatus) {
131
+ (0, _chai.expect)(errorStatus).to.be(expectedErrorStatus);
132
+ return;
133
+ }
134
+
122
135
  return;
123
136
  }
124
137
  } // eslint-disable-next-line functional/no-conditional-statement
@@ -130,7 +143,8 @@ describe('candidate-search', () => {
130
143
  }
131
144
 
132
145
  function formatResults(results) {
133
- // console.log(results); //eslint-disable-line
146
+ debug(results); //eslint-disable-line
147
+
134
148
  return { ...results,
135
149
  records: results.records.map(({
136
150
  record,
@@ -1 +1 @@
1
- {"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"}
1
+ {"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","expectedErrorStatus","count","expectedResults","Error","err","be","an","errorMessage","MatchingError","payload","message","errorStatus","status","match","results","formatResults","eql","records","map","id","toObject"],"mappings":";;AA4BA;;AACA;;AACA;;AACA;;AACA;;AACA;;AACA;;;;;;;;AAlCA;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;AAUA,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,mBAArC;AAA0DC,MAAAA,KAAK,GAAG;AAAlE,KAAvB,EAA6F;AAC3F,YAAMC,eAAe,GAAGxB,UAAU,CAAE,kBAAiBuB,KAAM,OAAzB,CAAlC;;AAEA,UAAInB,mBAAJ,EAAyB;AAAE;AACzB,YAAI;AACF,gBAAMS,MAAM,CAACX,aAAD,CAAZ;AACA,gBAAM,IAAIuB,KAAJ,CAAU,mBAAV,CAAN;AACD,SAHD,CAGE,OAAOC,GAAP,EAAY;AACZpC,UAAAA,KAAK,CAAE,iBAAgBoC,GAAI,EAAtB,CAAL;AACA,4BAAOA,GAAP,EAAYjB,EAAZ,CAAekB,EAAf,CAAkBC,EAAlB,CAAqB,OAArB;AACA,gBAAMC,YAAY,GAAGH,GAAG,YAAYI,qBAAf,GAA+BJ,GAAG,CAACK,OAAJ,CAAYC,OAA3C,GAAqDN,GAAG,CAACM,OAA9E;AACA,gBAAMC,WAAW,GAAGP,GAAG,YAAYI,qBAAf,GAA+BJ,GAAG,CAACQ,MAAnC,GAA4ChB,SAAhE;AACA5B,UAAAA,KAAK,CAAE,iBAAgBuC,YAAa,kBAAiBI,WAAY,EAA5D,CAAL;AACA,4BAAOJ,YAAP,EAAqBpB,EAArB,CAAwB0B,KAAxB,CAA8B,IAAIvB,MAAJ,CAAWR,mBAAX,EAAgC,GAAhC,CAA9B;;AAEA,cAAIkB,mBAAJ,EAAyB;AACvB,8BAAOW,WAAP,EAAoBxB,EAApB,CAAuBkB,EAAvB,CAA0BL,mBAA1B;AACA;AACD;;AACD;AACD;AACF,OArB0F,CAuB3F;;;AACA,UAAI,CAAClB,mBAAL,EAA0B;AACxB,cAAMgC,OAAO,GAAG,MAAMvB,MAAM,CAACX,aAAD,CAA5B;AACA,0BAAOmC,aAAa,CAACD,OAAD,CAApB,EAA+B3B,EAA/B,CAAkC6B,GAAlC,CAAsCd,eAAtC;AACD;;AAED,eAASa,aAAT,CAAuBD,OAAvB,EAAgC;AAC9B9C,QAAAA,KAAK,CAAC8C,OAAD,CAAL,CAD8B,CACd;;AAChB,eAAO,EACL,GAAGA,OADE;AAELG,UAAAA,OAAO,EAAEH,OAAO,CAACG,OAAR,CAAgBC,GAAhB,CAAoB,CAAC;AAACrB,YAAAA,MAAD;AAASsB,YAAAA;AAAT,WAAD,MAAmB;AAACA,YAAAA,EAAD;AAAKtB,YAAAA,MAAM,EAAEA,MAAM,CAACuB,QAAP;AAAb,WAAnB,CAApB;AAFJ,SAAP;AAID;AAEF;AACF;AACF,CAjFO,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 {Error as MatchingError} from '@natlibfi/melinda-commons';\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: ${JSON.stringify(factoryOptions)}`);\n return {\n ...factoryOptions,\n maxRecordsPerRequest: factoryOptions.maxRecordsPerRequest || 1,\n maxServerResults: factoryOptions.maxServerResults || undefined,\n record: new MarcRecord(factoryOptions.record, {subfieldValues: false})\n };\n }\n\n // eslint-disable-next-line max-statements\n async function iterate({searchOptions, expectedSearchError, expectedErrorStatus, count = 1}) {\n const expectedResults = getFixture(`expectedResults${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 debug(`Got an error: ${err}`);\n expect(err).to.be.an('error');\n const errorMessage = err instanceof MatchingError ? err.payload.message : err.message;\n const errorStatus = err instanceof MatchingError ? err.status : undefined;\n debug(`errorMessage: ${errorMessage}, errorStatus: ${errorStatus}`);\n expect(errorMessage).to.match(new RegExp(expectedSearchError, 'u'));\n\n if (expectedErrorStatus) {\n expect(errorStatus).to.be(expectedErrorStatus);\n return;\n }\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 debug(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"}
@@ -104,7 +104,7 @@ function bibSourceIds(record) {
104
104
 
105
105
  function removeSourcePrefix(subfieldValue) {
106
106
  const sourcePrefixRegex = /^(?<sourcePrefix>\([A-Za-z0-9-]+\))(?<id>.+)$/u;
107
- const normalizedValue = subfieldValue.replace(sourcePrefixRegex, '$<id>');
107
+ const normalizedValue = subfieldValue ? subfieldValue.replace(sourcePrefixRegex, '$<id>') : '';
108
108
  debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);
109
109
  return normalizedValue;
110
110
  }
@@ -112,7 +112,7 @@ function bibSourceIds(record) {
112
112
  function normalizeSidSubfieldValue(subfieldValue) {
113
113
  debugData(`Normalizing ${subfieldValue}`);
114
114
  const normalizeAwayRegex = /[- ]/u;
115
- return subfieldValue.replace(normalizeAwayRegex, '');
115
+ return subfieldValue ? subfieldValue.replace(normalizeAwayRegex, '') : '';
116
116
  }
117
117
  }
118
118
  }
@@ -1 +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","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"}
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","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,GAAGA,aAAa,CAACG,OAAd,CAAsBF,iBAAtB,EAAyC,OAAzC,CAAH,GAAuD,EAA5F;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,GAAGA,aAAa,CAACG,OAAd,CAAsBC,kBAAtB,EAA0C,EAA1C,CAAH,GAAmD,EAAvE;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 ? subfieldValue.replace(sourcePrefixRegex, '$<id>') : '';\n debugData(`Normalized ${subfieldValue} to ${normalizedValue}`);\n return normalizedValue;\n }\n\n function normalizeSidSubfieldValue(subfieldValue) {\n debugData(`Normalizing ${subfieldValue}`);\n const normalizeAwayRegex = (/[- ]/u);\n return subfieldValue ? subfieldValue.replace(normalizeAwayRegex, '') : '';\n }\n\n }\n }\n }\n}\n\nexport function bibMelindaIds(record) {\n // Melinda's SRU-index melinda.melindaid includes f001 controlnumbers and old Melinda-IDs from f035z's for all non-deleted Melinda-records\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibMelindaIds');\n const debugData = debug.extend('data');\n debug(`Creating queries for MelindaIds`);\n\n // Note: Melinda-ID's for search queries are created just from records f035a's and f035z's\n // Both (FI-MELINDA)- and FCC-prefixed forms are found\n // f001 controlnumber is not currently included, even if record's f003 is FI-MELINDA\n const melindaIds = getMelindaIdsF035(record);\n\n debugData(`Unique identifiers (${melindaIds.length}): ${JSON.stringify(melindaIds)}`);\n\n if (melindaIds.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(melindaIds, 'melinda.melindaid');\n}\n\n\n// bibHostComponents returns host id from the first subfield $w of first field f773, see test-fixtures 04 and 05\n// bibHostComponents should search all 773 $ws for possible host id, but what should it do in case of multiple host ids?\nexport function bibHostComponents(record) {\n const id = getHostId();\n return id ? [`melinda.partsofhost=${id}`] : [];\n\n function getHostId() {\n const [field] = record.get(/^773$/u);\n\n if (field) {\n const {value} = field.subfields.find(({code}) => code === 'w') || {};\n\n if (value && (/^\\(FI-MELINDA\\)/u).test(value)) {\n return value.replace(/^\\(FI-MELINDA\\)/u, '');\n }\n\n if (value && (/^\\(FIN01\\)/u).test(value)) {\n return value.replace(/^\\(FIN01\\)/u, '');\n }\n\n return false;\n }\n return false;\n }\n}\n\n// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to\n// (probably) to *headings* index TIT\n// - Aleph cannot currently handle headings searches starting with a boolean - in these cases use word search\n\n// Headings index TIT drops articles etc. from the start of the title according to the filing indicator\n// Currently filing indicator is not implemented - if the title starts with an article and the Melinda\n// record is correctly catalogued using a filing indicator -> dc.title search won't match\n\nexport function bibTitle(record) {\n const title = getTitle();\n const booleanStartWords = ['and', 'or', 'nor', 'not'];\n\n if (title) {\n const formatted = title\n .replace(/[^\\w\\s\\p{Alphabetic}]/gu, '')\n // Clean up concurrent spaces from fe. subfield changes\n .replace(/ +/gu, ' ')\n .trim()\n .slice(0, 30)\n .trim();\n\n // use word search for titles starting with a boolean\n const useWordSearch = booleanStartWords.some(word => formatted.toLowerCase().startsWith(word));\n // Prevent too many matches by having a minimum length requirement\n return formatted.length >= 5 ? [`dc.title=\"${useWordSearch ? '' : '^'}${formatted}*\"`] : [];\n }\n\n return [];\n\n function getTitle() {\n const [field] = record.get(/^245$/u);\n\n if (field) {\n return field.subfields\n .filter(({code}) => ['a', 'b'].includes(code))\n .map(({value}) => value)\n // In Melinda's index subfield separators are indexed as ' '\n .join(' ');\n }\n return false;\n }\n}\n\n// eslint-disable-next-line max-statements\nexport function bibStandardIdentifiers(record) {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:bibStandardIdentifiers');\n const debugData = debug.extend('data');\n debug(`Creating queries for standard identifiers`);\n\n const fields = record.get(/^(?<def>020|022|024)$/u);\n const identifiers = [].concat(...fields.map(toIdentifiers));\n const uniqueIdentifiers = [...new Set(identifiers)];\n\n debugData(`Standard identifier fields: ${JSON.stringify(fields)}`);\n debugData(`Identifiers (${identifiers.length}): ${JSON.stringify(identifiers)}`);\n debugData(`Unique identifiers (${uniqueIdentifiers.length}): ${JSON.stringify(uniqueIdentifiers)}`);\n\n if (uniqueIdentifiers.length < 1) {\n debug(`No identifiers found, no queries created.`);\n return [];\n }\n\n return toQueries(uniqueIdentifiers, 'dc.identifier');\n\n function toIdentifiers({tag, subfields}) {\n const issnIsbnReqExp = (/^[A-Za-z0-9-]+$/u);\n const otherIdReqExp = (/^[A-Za-z0-9-:]+$/u);\n\n if (tag === '022') {\n return subfields\n .filter(sub => ['a', 'z', 'y'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n .map(({value}) => value);\n }\n\n if (tag === '020') {\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && issnIsbnReqExp.test(sub.value) && sub.value !== undefined)\n .map(({value}) => value);\n }\n\n return subfields\n .filter(sub => ['a', 'z'].includes(sub.code) && otherIdReqExp.test(sub.value) && sub.value !== undefined)\n .map(({value}) => value);\n }\n}\n"],"file":"bib.js"}