@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.
- package/dist/candidate-search/index.js +40 -10
- package/dist/candidate-search/index.js.map +1 -1
- package/dist/candidate-search/index.spec.js +16 -2
- package/dist/candidate-search/index.spec.js.map +1 -1
- package/dist/candidate-search/query-list/bib.js +2 -2
- package/dist/candidate-search/query-list/bib.js.map +1 -1
- package/dist/index.js +139 -39
- package/dist/index.js.map +1 -1
- package/dist/index.spec.js +12 -6
- package/dist/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/authors.js +1 -1
- package/dist/match-detection/features/bib/authors.js.map +1 -1
- package/dist/match-detection/features/bib/standard-identifier-factory.js +3 -2
- package/dist/match-detection/features/bib/standard-identifier-factory.js.map +1 -1
- package/dist/match-detection/index.js +36 -30
- package/dist/match-detection/index.js.map +1 -1
- package/dist/match-detection/index.spec.js +20 -2
- package/dist/match-detection/index.spec.js.map +1 -1
- package/dist/matching-utils.js +1 -1
- package/dist/matching-utils.js.map +1 -1
- package/package.json +16 -15
- package/src/candidate-search/index.js +36 -14
- package/src/candidate-search/index.spec.js +13 -3
- package/src/candidate-search/query-list/bib.js +2 -2
- package/src/index.js +94 -35
- package/src/index.spec.js +10 -6
- package/src/match-detection/features/bib/authors.js +1 -1
- package/src/match-detection/features/bib/standard-identifier-factory.js +2 -1
- package/src/match-detection/index.js +29 -19
- package/src/match-detection/index.spec.js +17 -3
- package/src/matching-utils.js +1 -1
package/dist/index.js
CHANGED
|
@@ -48,6 +48,7 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
48
48
|
* for the JavaScript code in this file.
|
|
49
49
|
*
|
|
50
50
|
*/
|
|
51
|
+
//import inspect from 'util';
|
|
51
52
|
var _default = ({
|
|
52
53
|
detection: detectionOptions,
|
|
53
54
|
search: searchOptions,
|
|
@@ -55,7 +56,8 @@ var _default = ({
|
|
|
55
56
|
maxCandidates = 25,
|
|
56
57
|
returnStrategy = false,
|
|
57
58
|
returnQuery = false,
|
|
58
|
-
returnNonMatches = false
|
|
59
|
+
returnNonMatches = false,
|
|
60
|
+
returnFailures = false
|
|
59
61
|
}) => {
|
|
60
62
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:index');
|
|
61
63
|
const debugData = debug.extend('data');
|
|
@@ -66,6 +68,7 @@ var _default = ({
|
|
|
66
68
|
debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
|
|
67
69
|
debugData(`ReturnQuery: ${JSON.stringify(returnQuery)}`);
|
|
68
70
|
debugData(`ReturnNonMatches: ${JSON.stringify(returnNonMatches)}`);
|
|
71
|
+
debugData(`ReturnFailures: ${JSON.stringify(returnFailures)}`);
|
|
69
72
|
const detect = (0, matchDetection.default)(detectionOptions, returnStrategy);
|
|
70
73
|
return record => {
|
|
71
74
|
const search = (0, candidateSearch.default)({ ...searchOptions,
|
|
@@ -90,16 +93,22 @@ var _default = ({
|
|
|
90
93
|
candidateCount = 0,
|
|
91
94
|
nonMatches = [],
|
|
92
95
|
duplicateCount = 0,
|
|
93
|
-
nonMatchCount = 0
|
|
96
|
+
nonMatchCount = 0,
|
|
97
|
+
conversionFailures = [],
|
|
98
|
+
matchErrors = []
|
|
94
99
|
}) {
|
|
95
100
|
debugData(`Starting next matcher iteration.`);
|
|
96
101
|
const {
|
|
97
102
|
records,
|
|
103
|
+
failures,
|
|
98
104
|
...state
|
|
99
105
|
} = await search(initialState);
|
|
100
|
-
debugData(`Current state: ${JSON.stringify(state)}, matches: ${matches.length}, candidateCount: ${candidateCount}, nonMatches: ${nonMatches.length}`);
|
|
106
|
+
debugData(`Current state: ${JSON.stringify(state)}, matches: ${matches.length}, candidateCount: ${candidateCount}, nonMatches: ${nonMatches.length}, nonMatchCount: ${nonMatchCount}, conversionFailures: ${conversionFailures}, matchErrors: ${matchErrors.length}`);
|
|
101
107
|
const recordSetSize = records.length;
|
|
102
|
-
const
|
|
108
|
+
const failureSetSize = failures.length;
|
|
109
|
+
const newCandidateCount = candidateCount + recordSetSize + failureSetSize;
|
|
110
|
+
const newConversionFailures = conversionFailures.concat(failures);
|
|
111
|
+
debugData(`Failures: ${failures.length}, ConversionFailures: ${conversionFailures.length}, NewConversionFailures: ${newConversionFailures.length}`);
|
|
103
112
|
|
|
104
113
|
if (recordSetSize > 0) {
|
|
105
114
|
return handleRecordSet();
|
|
@@ -112,7 +121,10 @@ var _default = ({
|
|
|
112
121
|
matches,
|
|
113
122
|
candidateCount: newCandidateCount,
|
|
114
123
|
nonMatches,
|
|
115
|
-
|
|
124
|
+
nonMatchCount,
|
|
125
|
+
duplicateCount,
|
|
126
|
+
conversionFailures: newConversionFailures,
|
|
127
|
+
matchErrors
|
|
116
128
|
});
|
|
117
129
|
}
|
|
118
130
|
|
|
@@ -122,8 +134,11 @@ var _default = ({
|
|
|
122
134
|
state,
|
|
123
135
|
stopReason: '',
|
|
124
136
|
nonMatches,
|
|
137
|
+
nonMatchCount,
|
|
125
138
|
candidateCount: newCandidateCount,
|
|
126
|
-
duplicateCount
|
|
139
|
+
duplicateCount,
|
|
140
|
+
conversionFailures: newConversionFailures,
|
|
141
|
+
matchErrors
|
|
127
142
|
});
|
|
128
143
|
|
|
129
144
|
function handleRecordSet() {
|
|
@@ -133,14 +148,16 @@ var _default = ({
|
|
|
133
148
|
recordSetSize,
|
|
134
149
|
maxMatches,
|
|
135
150
|
matches,
|
|
136
|
-
nonMatches
|
|
151
|
+
nonMatches,
|
|
152
|
+
nonMatchCount
|
|
137
153
|
});
|
|
138
154
|
const newDuplicateCount = duplicateCount + matchResult.duplicateCount;
|
|
139
155
|
const newNonMatchCount = nonMatchCount + matchResult.nonMatchCount;
|
|
140
156
|
const {
|
|
141
157
|
newMatches,
|
|
142
|
-
newNonMatches
|
|
143
|
-
|
|
158
|
+
newNonMatches,
|
|
159
|
+
newMatchErrors
|
|
160
|
+
} = handleMatchResult(matchResult, matches, nonMatches, matchErrors);
|
|
144
161
|
|
|
145
162
|
if (maxMatchesFound({
|
|
146
163
|
matches: newMatches,
|
|
@@ -153,7 +170,9 @@ var _default = ({
|
|
|
153
170
|
nonMatches: newNonMatches,
|
|
154
171
|
duplicateCount: newDuplicateCount,
|
|
155
172
|
candidateCount: newCandidateCount,
|
|
156
|
-
nonMatchCount: newNonMatchCount
|
|
173
|
+
nonMatchCount: newNonMatchCount,
|
|
174
|
+
conversionFailures: newConversionFailures,
|
|
175
|
+
matchErrors: newMatchErrors
|
|
157
176
|
});
|
|
158
177
|
}
|
|
159
178
|
|
|
@@ -165,7 +184,9 @@ var _default = ({
|
|
|
165
184
|
nonMatches: newNonMatches,
|
|
166
185
|
duplicateCount: newDuplicateCount,
|
|
167
186
|
candidateCount: newCandidateCount,
|
|
168
|
-
nonMatchCount: newNonMatchCount
|
|
187
|
+
nonMatchCount: newNonMatchCount,
|
|
188
|
+
conversionFailures: newConversionFailures,
|
|
189
|
+
matchErrors: newMatchErrors
|
|
169
190
|
});
|
|
170
191
|
}
|
|
171
192
|
|
|
@@ -175,11 +196,13 @@ var _default = ({
|
|
|
175
196
|
candidateCount: newCandidateCount,
|
|
176
197
|
nonMatches: newNonMatches,
|
|
177
198
|
duplicateCount: newDuplicateCount,
|
|
178
|
-
nonMatchCount: newNonMatchCount
|
|
199
|
+
nonMatchCount: newNonMatchCount,
|
|
200
|
+
conversionFailures: newConversionFailures,
|
|
201
|
+
matchErrors: newMatchErrors
|
|
179
202
|
});
|
|
180
203
|
}
|
|
181
204
|
|
|
182
|
-
function handleMatchResult(matchResult, matches, nonMatches) {
|
|
205
|
+
function handleMatchResult(matchResult, matches, nonMatches, matchErrors) {
|
|
183
206
|
debugData(`- Amount of new matches from record set: ${matchResult.matches.length}`); // eslint-disable-next-line functional/no-conditional-statement
|
|
184
207
|
|
|
185
208
|
if (returnNonMatches) {
|
|
@@ -188,15 +211,20 @@ var _default = ({
|
|
|
188
211
|
|
|
189
212
|
const newMatches = matches.concat(returnQuery ? addQuery(matchResult.matches) : matchResult.matches);
|
|
190
213
|
const newNonMatches = returnNonMatches ? nonMatches.concat(returnQuery ? addQuery(matchResult.nonMatches) : matchResult.nonMatches) : [];
|
|
214
|
+
const newMatchErrors = matchErrors.concat(matchResult.matchErrors);
|
|
191
215
|
debugData(`- Total amount of matches: ${newMatches.length}`); // eslint-disable-next-line functional/no-conditional-statement
|
|
192
216
|
|
|
193
217
|
if (returnNonMatches) {
|
|
194
218
|
debugData(`- Total amount of nonMatches: ${newNonMatches.length}`);
|
|
195
219
|
}
|
|
196
220
|
|
|
221
|
+
debugData(`MatchResult: ${JSON.stringify(matchResult)}`);
|
|
222
|
+
debugData(`Old matchErrors: ${JSON.stringify(matchErrors)}, matchErrors from matchResult: ${JSON.stringify(matchResult.matchErrors)}, New matchErrors: ${JSON.stringify(newMatchErrors)}`);
|
|
223
|
+
debugData(`- Total amount of matchErrors: ${newMatchErrors.length}`);
|
|
197
224
|
return {
|
|
198
225
|
newMatches,
|
|
199
|
-
newNonMatches
|
|
226
|
+
newNonMatches,
|
|
227
|
+
newMatchErrors
|
|
200
228
|
};
|
|
201
229
|
}
|
|
202
230
|
|
|
@@ -223,9 +251,10 @@ var _default = ({
|
|
|
223
251
|
// - strategy (if returnStrategy option is true)
|
|
224
252
|
// - treshold (if returnStrategy option is true)
|
|
225
253
|
// - matchQuery (if returnQuery option is true)
|
|
254
|
+
// failures: array of conversionFailures from candidate-search and matchErrors from matchDetection in error format {status, payload: {message, id}} if returnFailures is true
|
|
226
255
|
// we could have here also returnRecords/returnMatchRecords/returnNonMatchRecord options that could be turned false for not to return actual record data
|
|
227
256
|
// matchStatus.status: boolean, true if matcher retrieved and handled all found candidate records, false if it did not
|
|
228
|
-
// matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries',empty string/undefined), reason for stopping retrieving or handling the candidate records
|
|
257
|
+
// matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries','conversionFailures', empty string/undefined), reason for stopping retrieving or handling the candidate records
|
|
229
258
|
// - only one stopReason is returned (if there would be several possible stopReasons, stopReason is picked in the above order)
|
|
230
259
|
// - currently stopReason can be non-empty also in cases where status is true, if matcher hit the stop reason when handling the last available candidate record
|
|
231
260
|
|
|
@@ -237,18 +266,24 @@ var _default = ({
|
|
|
237
266
|
nonMatches,
|
|
238
267
|
duplicateCount,
|
|
239
268
|
candidateCount,
|
|
240
|
-
nonMatchCount
|
|
269
|
+
nonMatchCount,
|
|
270
|
+
conversionFailures,
|
|
271
|
+
matchErrors
|
|
241
272
|
}) {
|
|
273
|
+
const conversionFailureCount = conversionFailures.length;
|
|
274
|
+
const matchErrorCount = matchErrors.length;
|
|
242
275
|
checkCounts({
|
|
243
276
|
matches,
|
|
244
277
|
nonMatches,
|
|
245
278
|
candidateCount,
|
|
246
279
|
duplicateCount,
|
|
247
|
-
nonMatchCount
|
|
280
|
+
nonMatchCount,
|
|
281
|
+
conversionFailureCount,
|
|
282
|
+
matchErrorCount
|
|
248
283
|
});
|
|
249
|
-
const matchStatus = getMatchState(state, stopReason); // add nonMatches to result only if returnNonMatches is 'true', otherwise nonMatches have not been gathered
|
|
284
|
+
const matchStatus = getMatchState(state, stopReason, conversionFailureCount, matchErrorCount); // add nonMatches to result only if returnNonMatches is 'true', otherwise nonMatches have not been gathered
|
|
250
285
|
|
|
251
|
-
const
|
|
286
|
+
const matchesResult = returnNonMatches ? {
|
|
252
287
|
matches,
|
|
253
288
|
matchStatus,
|
|
254
289
|
nonMatches
|
|
@@ -256,6 +291,11 @@ var _default = ({
|
|
|
256
291
|
matches,
|
|
257
292
|
matchStatus
|
|
258
293
|
};
|
|
294
|
+
const failures = [...conversionFailures, ...matchErrors];
|
|
295
|
+
const result = returnFailures ? { ...matchesResult,
|
|
296
|
+
conversionFailures: failures
|
|
297
|
+
} : matchesResult;
|
|
298
|
+
debugData(`ReturnFailures ${returnFailures}`);
|
|
259
299
|
debugData(`${JSON.stringify(result)}`);
|
|
260
300
|
return result; // note that in cases where the matching has been stopped because of maxMatches checkCounts won't (in most cases) match
|
|
261
301
|
|
|
@@ -264,13 +304,15 @@ var _default = ({
|
|
|
264
304
|
nonMatches,
|
|
265
305
|
candidateCount,
|
|
266
306
|
duplicateCount,
|
|
267
|
-
nonMatchCount
|
|
307
|
+
nonMatchCount,
|
|
308
|
+
conversionFailureCount,
|
|
309
|
+
matchErrorCount
|
|
268
310
|
}) {
|
|
269
311
|
const matchCount = matches.length;
|
|
270
312
|
debugData(`Return nonMatches: ${returnNonMatches}`);
|
|
271
313
|
const chosenNonMatchCount = returnNonMatches ? nonMatches.length : nonMatchCount;
|
|
272
|
-
const totalHandled = matchCount +
|
|
273
|
-
debug(`candidateCount: ${candidateCount}, matches: ${matchCount}, nonMatches: ${chosenNonMatchCount}, duplicateCount: ${duplicateCount}`);
|
|
314
|
+
const totalHandled = matchCount + chosenNonMatchCount + duplicateCount;
|
|
315
|
+
debug(`candidateCount: ${candidateCount}, matches: ${matchCount}, nonMatches: ${chosenNonMatchCount}, duplicateCount: ${duplicateCount}, conversionFailureCount: ${conversionFailureCount}, matchErrorCount: ${matchErrorCount}`);
|
|
274
316
|
debug(`We got result for ${totalHandled} / ${candidateCount} retrieved candidates`);
|
|
275
317
|
|
|
276
318
|
if (totalHandled !== candidateCount) {
|
|
@@ -279,20 +321,28 @@ var _default = ({
|
|
|
279
321
|
}
|
|
280
322
|
|
|
281
323
|
return;
|
|
282
|
-
}
|
|
324
|
+
} // eslint-disable-next-line max-statements
|
|
283
325
|
|
|
284
|
-
|
|
326
|
+
|
|
327
|
+
function getMatchState(state, stopReason, conversionFailuresCount, matchErrorCount) {
|
|
285
328
|
debugData(`${JSON.stringify(state)}`);
|
|
329
|
+
debug(`We had ${conversionFailuresCount} retrieved candidates that could not be converted.`);
|
|
330
|
+
debug(`We had ${matchErrorCount} retrieved candidates that errored in matchDetection.`);
|
|
286
331
|
debug(`Queries left ${state.queriesLeft}, Searches for current query left: ${state.resultSetOffset && state.resultSetOffset <= state.totalRecords}, non-retrieved records: ${state.totalRecords - state.queryCandidateCounter}, maxedQueries (${state.maxedQueries.length}): ${state.maxedQueries}`);
|
|
287
332
|
debugData(`StopReason: <${stopReason}>`);
|
|
288
333
|
const searchesLeft = state.resultSetOffset && state.resultSetOffset <= state.totalRecords;
|
|
289
334
|
const nonRetrieved = searchesLeft ? state.totalRecords - state.queryCandidateCounter : 0;
|
|
290
|
-
debugData(`nonRetrieved: ${nonRetrieved}`);
|
|
291
|
-
|
|
292
|
-
|
|
293
|
-
|
|
294
|
-
const
|
|
335
|
+
debugData(`nonRetrieved: ${nonRetrieved}`); // matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries','conversionFailures', empty string/undefined), reason for stopping retrieving or handling the candidate records
|
|
336
|
+
// 'maxMatches' and 'maxCandidates' are in stopReason, 'maxedQueries', 'conversionFailures' and 'matchErrors' are created here
|
|
337
|
+
|
|
338
|
+
if (state.queriesLeft > 0 || nonRetrieved > 0 || state.maxedQueries.length > 0 || conversionFailureCount > 0 || matchErrorCount > 0) {
|
|
339
|
+
const maxedQueriesStopReason = state.maxedQueries.length > 0 ? 'maxedQueries' : undefined;
|
|
340
|
+
const conversionFailuresStopReason = conversionFailureCount > 0 ? 'conversionFailures' : undefined;
|
|
341
|
+
const matchErrorsStopReason = matchErrorCount > 0 ? 'matchErrors' : undefined;
|
|
342
|
+
const newStopReason = stopReason === '' || stopReason === undefined ? maxedQueriesStopReason || conversionFailuresStopReason || matchErrorsStopReason : stopReason;
|
|
295
343
|
debugData(`MaxedQueriesStopReason: <${maxedQueriesStopReason}>`);
|
|
344
|
+
debugData(`ConversionFailureStopReason <${conversionFailuresStopReason}>`);
|
|
345
|
+
debugData(`MatchErrorsStopReason <${matchErrorsStopReason}>`);
|
|
296
346
|
debugData(`NewStopReason: <${newStopReason}>`);
|
|
297
347
|
debug(`Match status: false`);
|
|
298
348
|
return {
|
|
@@ -307,7 +357,12 @@ var _default = ({
|
|
|
307
357
|
stopReason
|
|
308
358
|
};
|
|
309
359
|
}
|
|
310
|
-
}
|
|
360
|
+
} // NOTES:
|
|
361
|
+
// - we could optimize by creating the featureSet for the incoming record once and using it for all database/candidateRecords
|
|
362
|
+
// - if creating the featureSet for the incoming record fails we have an unprocessable entity
|
|
363
|
+
// - if creating the featureSet for a candidate record fails we could skip that candidate - but list the case as a detectionFailure, same as conversionFailures
|
|
364
|
+
// eslint-disable-next-line max-statements
|
|
365
|
+
|
|
311
366
|
|
|
312
367
|
function iterateRecords({
|
|
313
368
|
records,
|
|
@@ -319,7 +374,8 @@ var _default = ({
|
|
|
319
374
|
recordNonMatches = [],
|
|
320
375
|
recordCount = 0,
|
|
321
376
|
recordDuplicateCount = 0,
|
|
322
|
-
recordNonMatchCount = 0
|
|
377
|
+
recordNonMatchCount = 0,
|
|
378
|
+
recordMatchErrors = []
|
|
323
379
|
}) {
|
|
324
380
|
// recordSetSize : total amount of records in the current record set
|
|
325
381
|
// recordCount : amount of records from the current record set that have been handled
|
|
@@ -330,21 +386,57 @@ var _default = ({
|
|
|
330
386
|
// matches : found matches in the current matcher job
|
|
331
387
|
// recordMatches : found matches in the current record set
|
|
332
388
|
// recordNonMatches : found nonMatches in the current record set (only if returnNonMatches setting is true)
|
|
389
|
+
// recordMatchErrors: errored matchDetection in the current record set
|
|
333
390
|
const [candidate] = records;
|
|
334
391
|
const newRecordCount = candidate ? recordCount + 1 : recordCount; // The matcher uses same matchDetection strategy for candidates from all candidate-searches -> matchDetection result for the same candidate is always same
|
|
335
392
|
// Exceptions would happen if the candidate would have been updated in the database between candidate searches
|
|
336
393
|
// Note that if returnNonMatches is false, matcher won't remember candidates that didn't match, so they will be matched again everytime they are retrieved by
|
|
337
394
|
// different candidate search queries. Same candidate search query won't have duplicate records.
|
|
338
395
|
|
|
396
|
+
/* We could optimize and detect all retrieved candidates at once
|
|
397
|
+
const candidateRecords = records.map(record => record.record);
|
|
398
|
+
const recordsIsArray = Array.isArray(candidateRecords);
|
|
399
|
+
debug(`records is an array: ${recordsIsArray}`);
|
|
400
|
+
const result = detect(record, candidateRecords);
|
|
401
|
+
debug(`${JSON.stringify(result)}`);
|
|
402
|
+
*/
|
|
403
|
+
|
|
339
404
|
if (candidate) {
|
|
405
|
+
// eslint-disable-next-line functional/no-conditional-statement
|
|
340
406
|
if (candidateNotInMatches(matches.concat(nonMatches), candidate)) {
|
|
341
407
|
const {
|
|
342
408
|
record: candidateRecord,
|
|
343
409
|
id: candidateId
|
|
344
410
|
} = candidate;
|
|
345
|
-
|
|
346
|
-
|
|
347
|
-
|
|
411
|
+
|
|
412
|
+
try {
|
|
413
|
+
debug(`Running matchDetection for record ${candidateId} (${newRecordCount}/${recordSetSize})`); // we should handle errors from detection somehow - ie. cases where either record or candidateRecord errors
|
|
414
|
+
|
|
415
|
+
const detectionResult = detect(record, candidateRecord);
|
|
416
|
+
return handleDetectionResult(detectionResult, candidateId, candidateRecord);
|
|
417
|
+
} catch (error) {
|
|
418
|
+
debug(`MatchDetection errored: database record ${candidateId}: ${error}`);
|
|
419
|
+
const matchError = {
|
|
420
|
+
status: 422,
|
|
421
|
+
payload: {
|
|
422
|
+
message: `Matching errored for database record ${candidateId}. ${error.message}.`,
|
|
423
|
+
id: candidateId
|
|
424
|
+
}
|
|
425
|
+
};
|
|
426
|
+
const newRecordMatchErrors = recordMatchErrors.concat(matchError);
|
|
427
|
+
return iterateRecords({
|
|
428
|
+
records: records.slice(1),
|
|
429
|
+
recordSetSize,
|
|
430
|
+
maxMatches,
|
|
431
|
+
matches,
|
|
432
|
+
recordMatches,
|
|
433
|
+
recordCount: newRecordCount,
|
|
434
|
+
recordNonMatches,
|
|
435
|
+
recordDuplicateCount,
|
|
436
|
+
recordNonMatchCount,
|
|
437
|
+
recordMatchErrors: newRecordMatchErrors
|
|
438
|
+
});
|
|
439
|
+
}
|
|
348
440
|
}
|
|
349
441
|
|
|
350
442
|
return iterateRecords({
|
|
@@ -356,7 +448,8 @@ var _default = ({
|
|
|
356
448
|
recordCount: newRecordCount,
|
|
357
449
|
recordNonMatches,
|
|
358
450
|
recordDuplicateCount: recordDuplicateCount + 1,
|
|
359
|
-
recordNonMatchCount
|
|
451
|
+
recordNonMatchCount,
|
|
452
|
+
recordMatchErrors
|
|
360
453
|
});
|
|
361
454
|
}
|
|
362
455
|
|
|
@@ -365,7 +458,8 @@ var _default = ({
|
|
|
365
458
|
matches: recordMatches,
|
|
366
459
|
nonMatches: returnNonMatches ? recordNonMatches : [],
|
|
367
460
|
duplicateCount: recordDuplicateCount,
|
|
368
|
-
nonMatchCount: recordNonMatchCount
|
|
461
|
+
nonMatchCount: recordNonMatchCount,
|
|
462
|
+
matchErrors: recordMatchErrors
|
|
369
463
|
};
|
|
370
464
|
|
|
371
465
|
function handleDetectionResult(detectionResult, candidateId, candidateRecord) {
|
|
@@ -404,19 +498,23 @@ var _default = ({
|
|
|
404
498
|
recordCount: newRecordCount,
|
|
405
499
|
recordNonMatches,
|
|
406
500
|
recordDuplicateCount,
|
|
407
|
-
recordNonMatchCount: newRecordNonMatchCount
|
|
501
|
+
recordNonMatchCount: newRecordNonMatchCount,
|
|
502
|
+
recordMatchErrors
|
|
408
503
|
});
|
|
409
504
|
}
|
|
410
505
|
|
|
411
506
|
function handleRecordMatch(isMatch, newMatch) {
|
|
412
507
|
const newRecordMatches = isMatch ? recordMatches.concat(newMatch) : recordMatches;
|
|
413
508
|
const newRecordNonMatches = isMatch ? recordNonMatches : recordNonMatches.concat(newMatch);
|
|
509
|
+
const newRecordNonMatchCount = isMatch ? recordNonMatchCount : recordNonMatchCount + 1;
|
|
414
510
|
debugData(`- Total matches after this detection: ${matches.concat(newRecordMatches).length} (max: ${maxMatches})`); // eslint-disable-next-line functional/no-conditional-statement
|
|
415
511
|
|
|
416
512
|
if (returnNonMatches) {
|
|
417
513
|
debugData(`- Total nonMatches after this detection: ${nonMatches.concat(newRecordNonMatches).length}`);
|
|
418
514
|
}
|
|
419
515
|
|
|
516
|
+
debugData(`- Total nonMatchCount after this detection: ${recordNonMatchCount}`);
|
|
517
|
+
|
|
420
518
|
if (maxMatchesFound({
|
|
421
519
|
matches: matches.concat(newRecordMatches),
|
|
422
520
|
maxMatches
|
|
@@ -426,7 +524,8 @@ var _default = ({
|
|
|
426
524
|
matches: newRecordMatches,
|
|
427
525
|
nonMatches: returnNonMatches ? newRecordNonMatches : [],
|
|
428
526
|
duplicateCount: recordDuplicateCount,
|
|
429
|
-
nonMatchCount:
|
|
527
|
+
nonMatchCount: newRecordNonMatchCount,
|
|
528
|
+
matchErrors: recordMatchErrors
|
|
430
529
|
};
|
|
431
530
|
}
|
|
432
531
|
|
|
@@ -439,7 +538,8 @@ var _default = ({
|
|
|
439
538
|
recordCount: newRecordCount,
|
|
440
539
|
recordNonMatches: returnNonMatches ? newRecordNonMatches : [],
|
|
441
540
|
duplicateCount: recordDuplicateCount,
|
|
442
|
-
recordNonMatchCount
|
|
541
|
+
recordNonMatchCount: newRecordNonMatchCount,
|
|
542
|
+
matchErrors: recordMatchErrors
|
|
443
543
|
});
|
|
444
544
|
}
|
|
445
545
|
|
package/dist/index.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/index.js"],"names":["detection","detectionOptions","search","searchOptions","maxMatches","maxCandidates","returnStrategy","returnQuery","returnNonMatches","debug","debugData","extend","JSON","stringify","detect","record","iterate","initialState","matches","candidateCount","nonMatches","duplicateCount","nonMatchCount","records","state","length","recordSetSize","newCandidateCount","handleRecordSet","queriesLeft","searchCounter","query","returnResult","stopReason","matchResult","iterateRecords","newDuplicateCount","newNonMatchCount","newMatches","newNonMatches","handleMatchResult","maxMatchesFound","maxCandidatesRetrieved","concat","addQuery","map","match","matchQuery","checkCounts","matchStatus","getMatchState","result","matchCount","chosenNonMatchCount","totalHandled","resultSetOffset","totalRecords","queryCandidateCounter","maxedQueries","searchesLeft","nonRetrieved","maxedQueriesStopReason","newStopReason","status","recordMatches","recordNonMatches","recordCount","recordDuplicateCount","recordNonMatchCount","candidate","newRecordCount","candidateNotInMatches","candidateRecord","id","candidateId","detectionResult","handleDetectionResult","slice","strategy","treshold","probability","strategyResult","newMatch","handleRecordMatch","newRecordNonMatchCount","isMatch","newRecordMatches","newRecordNonMatches","newCandidateId","find"],"mappings":";;;;;;;AA4BA;;AACA;;;;AACA;;;;;;;;;;AA9BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;eAQe,CAAC;AAACA,EAAAA,SAAS,EAAEC,gBAAZ;AAA8BC,EAAAA,MAAM,EAAEC,aAAtC;AAAqDC,EAAAA,UAAU,GAAG,CAAlE;AAAqEC,EAAAA,aAAa,GAAG,EAArF;AAAyFC,EAAAA,cAAc,GAAG,KAA1G;AAAiHC,EAAAA,WAAW,GAAG,KAA/H;AAAsIC,EAAAA,gBAAgB,GAAG;AAAzJ,CAAD,KAAqK;AAClL,QAAMC,KAAK,GAAG,oBAAkB,yCAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AAEAD,EAAAA,SAAS,CAAE,qBAAoBE,IAAI,CAACC,SAAL,CAAeZ,gBAAf,CAAiC,EAAvD,CAAT;AACAS,EAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAeV,aAAf,CAA8B,EAAjD,CAAT;AACAO,EAAAA,SAAS,CAAE,eAAcE,IAAI,CAACC,SAAL,CAAeT,UAAf,CAA2B,EAA3C,CAAT;AACAM,EAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAeR,aAAf,CAA8B,EAAjD,CAAT;AACAK,EAAAA,SAAS,CAAE,mBAAkBE,IAAI,CAACC,SAAL,CAAeP,cAAf,CAA+B,EAAnD,CAAT;AACAI,EAAAA,SAAS,CAAE,gBAAeE,IAAI,CAACC,SAAL,CAAeN,WAAf,CAA4B,EAA7C,CAAT;AACAG,EAAAA,SAAS,CAAE,qBAAoBE,IAAI,CAACC,SAAL,CAAeL,gBAAf,CAAiC,EAAvD,CAAT;AAEA,QAAMM,MAAM,GAAG,4BAAyBb,gBAAzB,EAA2CK,cAA3C,CAAf;AAEA,SAAOS,MAAM,IAAI;AACf,UAAMb,MAAM,GAAG,6BAAsB,EAAC,GAAGC,aAAJ;AAAmBY,MAAAA,MAAnB;AAA2BV,MAAAA;AAA3B,KAAtB,CAAf;AACA,WAAOW,OAAO,CAAC,EAAD,CAAd,CAFe,CAIf;AACA;AACA;AACA;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;;AAEA,mBAAeA,OAAf,CAAuB;AAACC,MAAAA,YAAY,GAAG,EAAhB;AAAoBC,MAAAA,OAAO,GAAG,EAA9B;AAAkCC,MAAAA,cAAc,GAAG,CAAnD;AAAsDC,MAAAA,UAAU,GAAG,EAAnE;AAAuEC,MAAAA,cAAc,GAAG,CAAxF;AAA2FC,MAAAA,aAAa,GAAG;AAA3G,KAAvB,EAAsI;AACpIZ,MAAAA,SAAS,CAAE,kCAAF,CAAT;AACA,YAAM;AAACa,QAAAA,OAAD;AAAU,WAAGC;AAAb,UAAsB,MAAMtB,MAAM,CAACe,YAAD,CAAxC;AAEAP,MAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,cAAaN,OAAO,CAACO,MAAO,qBAAoBN,cAAe,iBAAgBC,UAAU,CAACK,MAAO,EAA1I,CAAT;AACA,YAAMC,aAAa,GAAGH,OAAO,CAACE,MAA9B;AACA,YAAME,iBAAiB,GAAGR,cAAc,GAAGO,aAA3C;;AAEA,UAAIA,aAAa,GAAG,CAApB,EAAuB;AACrB,eAAOE,eAAe,EAAtB;AACD;;AAED,UAAIJ,KAAK,CAACK,WAAN,GAAoB,CAAxB,EAA2B;AACzBpB,QAAAA,KAAK,CAAE,oBAAmBe,KAAK,CAACM,aAAc,QAAON,KAAK,CAACO,KAAM,mBAAkBP,KAAK,CAACK,WAAY,eAAhG,CAAL;AACA,eAAOb,OAAO,CAAC;AAACC,UAAAA,YAAY,EAAEO,KAAf;AAAsBN,UAAAA,OAAtB;AAA+BC,UAAAA,cAAc,EAAEQ,iBAA/C;AAAkEP,UAAAA,UAAlE;AAA8EC,UAAAA;AAA9E,SAAD,CAAd;AACD;;AAEDZ,MAAAA,KAAK,CAAE,wEAAuES,OAAO,CAACO,MAAO,EAAxF,CAAL;AACA,aAAOO,YAAY,CAAC;AAACd,QAAAA,OAAD;AAAUM,QAAAA,KAAV;AAAiBS,QAAAA,UAAU,EAAE,EAA7B;AAAiCb,QAAAA,UAAjC;AAA6CD,QAAAA,cAAc,EAAEQ,iBAA7D;AAAgFN,QAAAA;AAAhF,OAAD,CAAnB;;AAEA,eAASO,eAAT,GAA2B;AACzBnB,QAAAA,KAAK,CAAE,0BAAyBiB,aAAc,qDAAoDF,KAAK,CAACM,aAAc,eAAcN,KAAK,CAACO,KAAM,EAA3I,CAAL;AAEA,cAAMG,WAAW,GAAGC,cAAc,CAAC;AAACZ,UAAAA,OAAD;AAAUG,UAAAA,aAAV;AAAyBtB,UAAAA,UAAzB;AAAqCc,UAAAA,OAArC;AAA8CE,UAAAA;AAA9C,SAAD,CAAlC;AACA,cAAMgB,iBAAiB,GAAGf,cAAc,GAAGa,WAAW,CAACb,cAAvD;AACA,cAAMgB,gBAAgB,GAAGf,aAAa,GAAGY,WAAW,CAACZ,aAArD;AACA,cAAM;AAACgB,UAAAA,UAAD;AAAaC,UAAAA;AAAb,YAA8BC,iBAAiB,CAACN,WAAD,EAAchB,OAAd,EAAuBE,UAAvB,CAArD;;AAEA,YAAIqB,eAAe,CAAC;AAACvB,UAAAA,OAAO,EAAEoB,UAAV;AAAsBlC,UAAAA;AAAtB,SAAD,CAAnB,EAAwD;AACtD,iBAAO4B,YAAY,CAAC;AAACd,YAAAA,OAAO,EAAEoB,UAAV;AAAsBd,YAAAA,KAAtB;AAA6BS,YAAAA,UAAU,EAAE,YAAzC;AAAuDb,YAAAA,UAAU,EAAEmB,aAAnE;AAAkFlB,YAAAA,cAAc,EAAEe,iBAAlG;AAAqHjB,YAAAA,cAAc,EAAEQ,iBAArI;AAAwJL,YAAAA,aAAa,EAAEe;AAAvK,WAAD,CAAnB;AACD;;AAED,YAAIK,sBAAsB,CAACf,iBAAD,EAAoBtB,aAApB,CAA1B,EAA8D;AAC5D,iBAAO2B,YAAY,CAAC;AAACd,YAAAA,OAAO,EAAEoB,UAAV;AAAsBd,YAAAA,KAAtB;AAA6BS,YAAAA,UAAU,EAAE,eAAzC;AAA0Db,YAAAA,UAAU,EAAEmB,aAAtE;AAAqFlB,YAAAA,cAAc,EAAEe,iBAArG;AAAwHjB,YAAAA,cAAc,EAAEQ,iBAAxI;AAA2JL,YAAAA,aAAa,EAAEe;AAA1K,WAAD,CAAnB;AACD;;AAED,eAAOrB,OAAO,CAAC;AAACC,UAAAA,YAAY,EAAEO,KAAf;AAAsBN,UAAAA,OAAO,EAAEoB,UAA/B;AAA2CnB,UAAAA,cAAc,EAAEQ,iBAA3D;AAA8EP,UAAAA,UAAU,EAAEmB,aAA1F;AAAyGlB,UAAAA,cAAc,EAAEe,iBAAzH;AAA4Id,UAAAA,aAAa,EAAEe;AAA3J,SAAD,CAAd;AACD;;AAED,eAASG,iBAAT,CAA2BN,WAA3B,EAAwChB,OAAxC,EAAiDE,UAAjD,EAA6D;AAC3DV,QAAAA,SAAS,CAAE,4CAA2CwB,WAAW,CAAChB,OAAZ,CAAoBO,MAAO,EAAxE,CAAT,CAD2D,CAE3D;;AACA,YAAIjB,gBAAJ,EAAsB;AACpBE,UAAAA,SAAS,CAAE,+CAA8CwB,WAAW,CAACd,UAAZ,CAAuBK,MAAO,EAA9E,CAAT;AACD;;AAED,cAAMa,UAAU,GAAGpB,OAAO,CAACyB,MAAR,CAAepC,WAAW,GAAGqC,QAAQ,CAACV,WAAW,CAAChB,OAAb,CAAX,GAAmCgB,WAAW,CAAChB,OAAzE,CAAnB;AACA,cAAMqB,aAAa,GAAG/B,gBAAgB,GAAGY,UAAU,CAACuB,MAAX,CAAkBpC,WAAW,GAAGqC,QAAQ,CAACV,WAAW,CAACd,UAAb,CAAX,GAAsCc,WAAW,CAACd,UAA/E,CAAH,GAAgG,EAAtI;AAEAV,QAAAA,SAAS,CAAE,8BAA6B4B,UAAU,CAACb,MAAO,EAAjD,CAAT,CAV2D,CAW3D;;AACA,YAAIjB,gBAAJ,EAAsB;AACpBE,UAAAA,SAAS,CAAE,iCAAgC6B,aAAa,CAACd,MAAO,EAAvD,CAAT;AACD;;AAED,eAAO;AAACa,UAAAA,UAAD;AAAaC,UAAAA;AAAb,SAAP;AACD;;AAED,eAASK,QAAT,CAAkB1B,OAAlB,EAA2B;AACzBR,QAAAA,SAAS,CAAE,gBAAec,KAAK,CAACO,KAAM,aAA7B,CAAT;AACA,eAAOb,OAAO,CAAC2B,GAAR,CAAaC,KAAD,KAAY,EAAC,GAAGA,KAAJ;AAAWC,UAAAA,UAAU,EAAEvB,KAAK,CAACO;AAA7B,SAAZ,CAAZ,CAAP;AACD;;AAED,eAASW,sBAAT,CAAgCvB,cAAhC,EAAgDd,aAAhD,EAA+D;AAC7DK,QAAAA,SAAS,CAAE,gDAA+CiB,iBAAkB,UAAStB,aAAc,GAA1F,CAAT;;AACA,YAAIA,aAAa,IAAIc,cAAc,IAAId,aAAvC,EAAsD;AACpDI,UAAAA,KAAK,CAAE,gEAA+DU,cAAe,MAAKd,aAAc,sBAAnG,CAAL;AACA,iBAAO,IAAP;AACD;AACF;AACF,KAvFc,CAyFf;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;AAEA;AACA;AACA;AACA;;;AAEA,aAAS2B,YAAT,CAAsB;AAACd,MAAAA,OAAD;AAAUM,MAAAA,KAAV;AAAiBS,MAAAA,UAAjB;AAA6Bb,MAAAA,UAA7B;AAAyCC,MAAAA,cAAzC;AAAyDF,MAAAA,cAAzD;AAAyEG,MAAAA;AAAzE,KAAtB,EAA+G;AAC7G0B,MAAAA,WAAW,CAAC;AAAC9B,QAAAA,OAAD;AAAUE,QAAAA,UAAV;AAAsBD,QAAAA,cAAtB;AAAsCE,QAAAA,cAAtC;AAAsDC,QAAAA;AAAtD,OAAD,CAAX;AACA,YAAM2B,WAAW,GAAGC,aAAa,CAAC1B,KAAD,EAAQS,UAAR,CAAjC,CAF6G,CAG7G;;AACA,YAAMkB,MAAM,GAAG3C,gBAAgB,GAAG;AAACU,QAAAA,OAAD;AAAU+B,QAAAA,WAAV;AAAuB7B,QAAAA;AAAvB,OAAH,GAAwC;AAACF,QAAAA,OAAD;AAAU+B,QAAAA;AAAV,OAAvE;AACAvC,MAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAesC,MAAf,CAAuB,EAA3B,CAAT;AACA,aAAOA,MAAP,CAN6G,CAQ7G;;AAEA,eAASH,WAAT,CAAqB;AAAC9B,QAAAA,OAAD;AAAUE,QAAAA,UAAV;AAAsBD,QAAAA,cAAtB;AAAsCE,QAAAA,cAAtC;AAAsDC,QAAAA;AAAtD,OAArB,EAA2F;AACzF,cAAM8B,UAAU,GAAGlC,OAAO,CAACO,MAA3B;AACAf,QAAAA,SAAS,CAAE,sBAAqBF,gBAAiB,EAAxC,CAAT;AACA,cAAM6C,mBAAmB,GAAG7C,gBAAgB,GAAGY,UAAU,CAACK,MAAd,GAAuBH,aAAnE;AACA,cAAMgC,YAAY,GAAGF,UAAU,GAAG9B,aAAb,GAA6BD,cAAlD;AACAZ,QAAAA,KAAK,CAAE,mBAAkBU,cAAe,cAAaiC,UAAW,iBAAgBC,mBAAoB,qBAAoBhC,cAAe,EAAlI,CAAL;AACAZ,QAAAA,KAAK,CAAE,qBAAoB6C,YAAa,MAAKnC,cAAe,uBAAvD,CAAL;;AACA,YAAImC,YAAY,KAAKnC,cAArB,EAAqC;AACnCV,UAAAA,KAAK,CAAE,gCAA+BU,cAAc,GAAGmC,YAAa,aAA/D,CAAL;AACA;AACD;;AACD;AACD;;AAED,eAASJ,aAAT,CAAuB1B,KAAvB,EAA8BS,UAA9B,EAA0C;AACxCvB,QAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAeW,KAAf,CAAsB,EAA1B,CAAT;AACAf,QAAAA,KAAK,CAAE,gBAAee,KAAK,CAACK,WAAY,sCAAqCL,KAAK,CAAC+B,eAAN,IAAyB/B,KAAK,CAAC+B,eAAN,IAAyB/B,KAAK,CAACgC,YAAa,4BAA2BhC,KAAK,CAACgC,YAAN,GAAqBhC,KAAK,CAACiC,qBAAsB,mBAAkBjC,KAAK,CAACkC,YAAN,CAAmBjC,MAAO,MAAKD,KAAK,CAACkC,YAAa,EAA7R,CAAL;AAEAhD,QAAAA,SAAS,CAAE,gBAAeuB,UAAW,GAA5B,CAAT;AAEA,cAAM0B,YAAY,GAAGnC,KAAK,CAAC+B,eAAN,IAAyB/B,KAAK,CAAC+B,eAAN,IAAyB/B,KAAK,CAACgC,YAA7E;AACA,cAAMI,YAAY,GAAGD,YAAY,GAAGnC,KAAK,CAACgC,YAAN,GAAqBhC,KAAK,CAACiC,qBAA9B,GAAsD,CAAvF;AACA/C,QAAAA,SAAS,CAAE,iBAAgBkD,YAAa,EAA/B,CAAT;;AAEA,YAAIpC,KAAK,CAACK,WAAN,GAAoB,CAApB,IAAyB+B,YAAY,GAAG,CAAxC,IAA6CpC,KAAK,CAACkC,YAAN,CAAmBjC,MAAnB,GAA4B,CAA7E,EAAgF;AAC9E,gBAAMoC,sBAAsB,GAAGrC,KAAK,CAACkC,YAAN,CAAmBjC,MAAnB,GAA4B,CAA5B,GAAgC,cAAhC,GAAiD,EAAhF;AACA,gBAAMqC,aAAa,GAAG7B,UAAU,KAAK,EAAf,GAAoB4B,sBAApB,GAA6C5B,UAAnE;AACAvB,UAAAA,SAAS,CAAE,4BAA2BmD,sBAAuB,GAApD,CAAT;AACAnD,UAAAA,SAAS,CAAE,mBAAkBoD,aAAc,GAAlC,CAAT;AACArD,UAAAA,KAAK,CAAE,qBAAF,CAAL;AACA,iBAAO;AAACsD,YAAAA,MAAM,EAAE,KAAT;AAAgB9B,YAAAA,UAAU,EAAE6B;AAA5B,WAAP;AACD;;AAEDrD,QAAAA,KAAK,CAAE,oBAAF,CAAL;AACA,eAAO;AAACsD,UAAAA,MAAM,EAAE,IAAT;AAAe9B,UAAAA;AAAf,SAAP;AACD;AACF;;AAED,aAASE,cAAT,CAAwB;AAACZ,MAAAA,OAAD;AAAUG,MAAAA,aAAV;AAAyBtB,MAAAA,UAAzB;AAAqCc,MAAAA,OAAO,GAAG,EAA/C;AAAmDE,MAAAA,UAAU,GAAG,EAAhE;AAAoE4C,MAAAA,aAAa,GAAG,EAApF;AAAwFC,MAAAA,gBAAgB,GAAG,EAA3G;AAA+GC,MAAAA,WAAW,GAAG,CAA7H;AAAgIC,MAAAA,oBAAoB,GAAG,CAAvJ;AAA0JC,MAAAA,mBAAmB,GAAG;AAAhL,KAAxB,EAA4M;AAE1M;AACA;AACA;AACA;AACA;AAEA;AACA;AACA;AACA;AAEA,YAAM,CAACC,SAAD,IAAc9C,OAApB;AACA,YAAM+C,cAAc,GAAGD,SAAS,GAAGH,WAAW,GAAG,CAAjB,GAAqBA,WAArD,CAd0M,CAgB1M;AACA;AACA;AACA;;AAEA,UAAIG,SAAJ,EAAe;AAEb,YAAIE,qBAAqB,CAACrD,OAAO,CAACyB,MAAR,CAAevB,UAAf,CAAD,EAA6BiD,SAA7B,CAAzB,EAAkE;AAChE,gBAAM;AAACtD,YAAAA,MAAM,EAAEyD,eAAT;AAA0BC,YAAAA,EAAE,EAAEC;AAA9B,cAA6CL,SAAnD;AACA5D,UAAAA,KAAK,CAAE,qCAAoCiE,WAAY,KAAIJ,cAAe,IAAG5C,aAAc,GAAtF,CAAL;AACA,gBAAMiD,eAAe,GAAG7D,MAAM,CAACC,MAAD,EAASyD,eAAT,CAA9B;AACA,iBAAOI,qBAAqB,CAACD,eAAD,EAAkBD,WAAlB,EAA+BF,eAA/B,CAA5B;AACD;;AACD,eAAOrC,cAAc,CAAC;AAACZ,UAAAA,OAAO,EAAEA,OAAO,CAACsD,KAAR,CAAc,CAAd,CAAV;AAA4BnD,UAAAA,aAA5B;AAA2CtB,UAAAA,UAA3C;AAAuDc,UAAAA,OAAvD;AAAgE8C,UAAAA,aAAhE;AAA+EE,UAAAA,WAAW,EAAEI,cAA5F;AAA4GL,UAAAA,gBAA5G;AAA8HE,UAAAA,oBAAoB,EAAEA,oBAAoB,GAAG,CAA3K;AAA8KC,UAAAA;AAA9K,SAAD,CAArB;AACD;;AAED3D,MAAAA,KAAK,CAAE,mCAAkCyD,WAAY,IAAGxC,aAAc,WAAUsC,aAAa,CAACvC,MAAO,mBAAkB0C,oBAAqB,gCAA+B3D,gBAAgB,GAAI,GAAEyD,gBAAgB,CAACxC,MAAO,EAA9B,GAAmC,GAAE2C,mBAAoB,EAAE,oBAAjP,CAAL;AACA,aAAO;AAAClD,QAAAA,OAAO,EAAE8C,aAAV;AAAyB5C,QAAAA,UAAU,EAAEZ,gBAAgB,GAAGyD,gBAAH,GAAsB,EAA3E;AAA+E5C,QAAAA,cAAc,EAAE8C,oBAA/F;AAAqH7C,QAAAA,aAAa,EAAE8C;AAApI,OAAP;;AAEA,eAASQ,qBAAT,CAA+BD,eAA/B,EAAgDD,WAAhD,EAA6DF,eAA7D,EAA8E;AAC5E9D,QAAAA,SAAS,CAAE,8BAA6BgE,WAAY,KAAIJ,cAAe,IAAG5C,aAAc,MAAKd,IAAI,CAACC,SAAL,CAAe8D,eAAf,CAAgC,EAApH,CAAT;;AAEA,YAAIA,eAAe,CAAC7B,KAAhB,IAAyBtC,gBAA7B,EAA+C;AAC7CC,UAAAA,KAAK,CAAE,GAAEkE,eAAe,CAAC7B,KAAhB,GAAyB,UAAS4B,WAAY,KAAIJ,cAAe,IAAG5C,aAAc,eAAlF,GAAoG,UAASgD,WAAY,KAAIJ,cAAe,IAAG5C,aAAc,mBAAmB,EAApL,CAAL;AACAhB,UAAAA,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAL,CAAe8D,eAAe,CAACG,QAA/B,CAAyC,eAAclE,IAAI,CAACC,SAAL,CAAe8D,eAAe,CAACI,QAA/B,CAAyC,EAA9G,CAAT;AAEA,gBAAM7C,WAAW,GAAG;AAClB8C,YAAAA,WAAW,EAAEL,eAAe,CAACK,WADX;AAElBX,YAAAA,SAAS,EAAE;AACTI,cAAAA,EAAE,EAAEC,WADK;AAET3D,cAAAA,MAAM,EAAEyD;AAFC;AAFO,WAApB;AAOA,gBAAMS,cAAc,GAAG;AACrBH,YAAAA,QAAQ,EAAEH,eAAe,CAACG,QADL;AAErBC,YAAAA,QAAQ,EAAEJ,eAAe,CAACI;AAFL,WAAvB;AAIA,gBAAMG,QAAQ,GAAG5E,cAAc,GAAG,EAAC,GAAG4B,WAAJ;AAAiB,eAAG+C;AAApB,WAAH,GAAyC,EAAC,GAAG/C;AAAJ,WAAxE;AAEAxB,UAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAeqE,QAAf,CAAyB,EAA7B,CAAT;AAEA,iBAAOC,iBAAiB,CAACR,eAAe,CAAC7B,KAAjB,EAAwBoC,QAAxB,CAAxB;AACD;;AAED,cAAME,sBAAsB,GAAGhB,mBAAmB,GAAG,CAArD;AACA1D,QAAAA,SAAS,CAAE,4CAA2C0E,sBAAuB,EAApE,CAAT;AAEA,eAAOjD,cAAc,CAAC;AAACZ,UAAAA,OAAO,EAAEA,OAAO,CAACsD,KAAR,CAAc,CAAd,CAAV;AAA4BnD,UAAAA,aAA5B;AAA2CtB,UAAAA,UAA3C;AAAuDc,UAAAA,OAAvD;AAAgE8C,UAAAA,aAAhE;AAA+EE,UAAAA,WAAW,EAAEI,cAA5F;AAA4GL,UAAAA,gBAA5G;AAA8HE,UAAAA,oBAA9H;AAAoJC,UAAAA,mBAAmB,EAAEgB;AAAzK,SAAD,CAArB;AACD;;AAED,eAASD,iBAAT,CAA2BE,OAA3B,EAAoCH,QAApC,EAA8C;AAC5C,cAAMI,gBAAgB,GAAGD,OAAO,GAAGrB,aAAa,CAACrB,MAAd,CAAqBuC,QAArB,CAAH,GAAoClB,aAApE;AACA,cAAMuB,mBAAmB,GAAGF,OAAO,GAAGpB,gBAAH,GAAsBA,gBAAgB,CAACtB,MAAjB,CAAwBuC,QAAxB,CAAzD;AAEAxE,QAAAA,SAAS,CAAE,yCAAwCQ,OAAO,CAACyB,MAAR,CAAe2C,gBAAf,EAAiC7D,MAAO,UAASrB,UAAW,GAAtG,CAAT,CAJ4C,CAM5C;;AACA,YAAII,gBAAJ,EAAsB;AACpBE,UAAAA,SAAS,CAAE,4CAA2CU,UAAU,CAACuB,MAAX,CAAkB4C,mBAAlB,EAAuC9D,MAAO,EAA3F,CAAT;AACD;;AAED,YAAIgB,eAAe,CAAC;AAACvB,UAAAA,OAAO,EAAEA,OAAO,CAACyB,MAAR,CAAe2C,gBAAf,CAAV;AAA4ClF,UAAAA;AAA5C,SAAD,CAAnB,EAA8E;AAC5EK,UAAAA,KAAK,CAAE,eAAcL,UAAW,gDAA+CkE,cAAe,yCAAwC5C,aAAa,GAAG4C,cAAe,EAAhK,CAAL;AACA,iBAAO;AAACpD,YAAAA,OAAO,EAAEoE,gBAAV;AAA4BlE,YAAAA,UAAU,EAAEZ,gBAAgB,GAAG+E,mBAAH,GAAyB,EAAjF;AAAqFlE,YAAAA,cAAc,EAAE8C,oBAArG;AAA2H7C,YAAAA,aAAa,EAAE8C;AAA1I,WAAP;AACD;;AAED,eAAOjC,cAAc,CAAC;AAACZ,UAAAA,OAAO,EAAEA,OAAO,CAACsD,KAAR,CAAc,CAAd,CAAV;AAA4BnD,UAAAA,aAA5B;AAA2CtB,UAAAA,UAA3C;AAAuDc,UAAAA,OAAvD;AAAgE8C,UAAAA,aAAa,EAAEsB,gBAA/E;AAAiGpB,UAAAA,WAAW,EAAEI,cAA9G;AAA8HL,UAAAA,gBAAgB,EAAEzD,gBAAgB,GAAG+E,mBAAH,GAAyB,EAAzL;AAA6LlE,UAAAA,cAAc,EAAE8C,oBAA7M;AAAmOC,UAAAA;AAAnO,SAAD,CAArB;AACD;;AAED,eAASG,qBAAT,CAA+BrD,OAA/B,EAAwCmD,SAAxC,EAAmD;AACjD5D,QAAAA,KAAK,CAAE,wBAAuB4D,SAAS,CAACI,EAAG,+BAA8BvD,OAAO,CAACO,MAAO,qBAAnF,CAAL;AACA,cAAM+D,cAAc,GAAGnB,SAAS,CAACI,EAAjC;AACA/D,QAAAA,SAAS,CAAE,mBAAkB8E,cAAe,EAAnC,CAAT;AACA,cAAMrC,MAAM,GAAGjC,OAAO,CAACuE,IAAR,CAAa,CAAC;AAACpB,UAAAA;AAAD,SAAD,KAAiBA,SAAS,CAACI,EAAV,KAAiBe,cAA/C,CAAf;AACA9E,QAAAA,SAAS,CAAE,WAAUyC,MAAO,EAAnB,CAAT;;AACA,YAAIA,MAAJ,EAAY;AACV1C,UAAAA,KAAK,CAAE,GAAE4D,SAAS,CAACI,EAAG,uBAAjB,CAAL;AACA,iBAAO,KAAP;AACD;;AACDhE,QAAAA,KAAK,CAAE,GAAE4D,SAAS,CAACI,EAAG,kCAAjB,CAAL;AACA,eAAO,IAAP;AACD;AACF;;AAED,aAAShC,eAAT,CAAyB;AAACvB,MAAAA,OAAD;AAAUd,MAAAA;AAAV,KAAzB,EAAgD;AAC9C,UAAIA,UAAU,IAAIc,OAAO,CAACO,MAAR,IAAkBrB,UAApC,EAAgD;AAC9CK,QAAAA,KAAK,CAAE,6CAA4CL,UAAW,0BAAzD,CAAL;AACA,eAAO,IAAP;AACD;AACF;AACF,GAnQD;AAoQD,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 createSearchInterface, * as candidateSearch from './candidate-search';\nimport createDetectionInterface, * as matchDetection from './match-detection';\n\nexport {candidateSearch, matchDetection};\n\nexport default ({detection: detectionOptions, search: searchOptions, maxMatches = 1, maxCandidates = 25, returnStrategy = false, returnQuery = false, returnNonMatches = false}) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:index');\n const debugData = debug.extend('data');\n\n debugData(`DetectionOptions: ${JSON.stringify(detectionOptions)}`);\n debugData(`SearchOptions: ${JSON.stringify(searchOptions)}`);\n debugData(`MaxMatches: ${JSON.stringify(maxMatches)}`);\n debugData(`MaxCandidates: ${JSON.stringify(maxCandidates)}`);\n debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`ReturnQuery: ${JSON.stringify(returnQuery)}`);\n debugData(`ReturnNonMatches: ${JSON.stringify(returnNonMatches)}`);\n\n const detect = createDetectionInterface(detectionOptions, returnStrategy);\n\n return record => {\n const search = createSearchInterface({...searchOptions, record, maxCandidates});\n return iterate({});\n\n // candidateCount : amount of candidate records retrived from SRU for matching, NOT including current record set\n // matches : candidates that have been detected as matches by current matcher job\n // nonMatches : candidates that have been detected as non-matches by current matcher job (only if returnNonMatches is 'true')\n // duplicateCount : amount of candidate records that were retrieved from the SRU but not handled further because they were already found in the matches/nonMatches\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 async function iterate({initialState = {}, matches = [], candidateCount = 0, nonMatches = [], duplicateCount = 0, nonMatchCount = 0}) {\n debugData(`Starting next matcher iteration.`);\n const {records, ...state} = await search(initialState);\n\n debugData(`Current state: ${JSON.stringify(state)}, matches: ${matches.length}, candidateCount: ${candidateCount}, nonMatches: ${nonMatches.length}`);\n const recordSetSize = records.length;\n const newCandidateCount = candidateCount + recordSetSize;\n\n if (recordSetSize > 0) {\n return handleRecordSet();\n }\n\n if (state.queriesLeft > 0) {\n debug(`Empty record set ${state.searchCounter} for ${state.query}, but there are ${state.queriesLeft} queries left`);\n return iterate({initialState: state, matches, candidateCount: newCandidateCount, nonMatches, duplicateCount});\n }\n\n debug(`No (more) candidate records to check, no more queries left, matches: ${matches.length}`);\n return returnResult({matches, state, stopReason: '', nonMatches, candidateCount: newCandidateCount, duplicateCount});\n\n function handleRecordSet() {\n debug(`Checking record set of ${recordSetSize} candidate records for possible matches, found by ${state.searchCounter} search for ${state.query}`);\n\n const matchResult = iterateRecords({records, recordSetSize, maxMatches, matches, nonMatches});\n const newDuplicateCount = duplicateCount + matchResult.duplicateCount;\n const newNonMatchCount = nonMatchCount + matchResult.nonMatchCount;\n const {newMatches, newNonMatches} = handleMatchResult(matchResult, matches, nonMatches);\n\n if (maxMatchesFound({matches: newMatches, maxMatches})) {\n return returnResult({matches: newMatches, state, stopReason: 'maxMatches', nonMatches: newNonMatches, duplicateCount: newDuplicateCount, candidateCount: newCandidateCount, nonMatchCount: newNonMatchCount});\n }\n\n if (maxCandidatesRetrieved(newCandidateCount, maxCandidates)) {\n return returnResult({matches: newMatches, state, stopReason: 'maxCandidates', nonMatches: newNonMatches, duplicateCount: newDuplicateCount, candidateCount: newCandidateCount, nonMatchCount: newNonMatchCount});\n }\n\n return iterate({initialState: state, matches: newMatches, candidateCount: newCandidateCount, nonMatches: newNonMatches, duplicateCount: newDuplicateCount, nonMatchCount: newNonMatchCount});\n }\n\n function handleMatchResult(matchResult, matches, nonMatches) {\n debugData(`- Amount of new matches from record set: ${matchResult.matches.length}`);\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Amount of new nonMatches from record set: ${matchResult.nonMatches.length}`);\n }\n\n const newMatches = matches.concat(returnQuery ? addQuery(matchResult.matches) : matchResult.matches);\n const newNonMatches = returnNonMatches ? nonMatches.concat(returnQuery ? addQuery(matchResult.nonMatches) : matchResult.nonMatches) : [];\n\n debugData(`- Total amount of matches: ${newMatches.length}`);\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Total amount of nonMatches: ${newNonMatches.length}`);\n }\n\n return {newMatches, newNonMatches};\n }\n\n function addQuery(matches) {\n debugData(`Adding query ${state.query} to matches`);\n return matches.map((match) => ({...match, matchQuery: state.query}));\n }\n\n function maxCandidatesRetrieved(candidateCount, maxCandidates) {\n debugData(`Total amount of candidate records retrieved: ${newCandidateCount} (max: ${maxCandidates})`);\n if (maxCandidates && candidateCount >= maxCandidates) {\n debug(`Stopped matching because maximum number of candidate records ${candidateCount} / ${maxCandidates} have been retrieved`);\n return true;\n }\n }\n }\n\n // matches : array of matching candidate records\n // nonMatches : array of nonMatching candidate records (if returnNonMatches option is true, otherwise empty array)\n // - candidate.id\n // - candidate.record\n // - probability\n // - strategy (if returnStrategy option is true)\n // - treshold (if returnStrategy option is true)\n // - matchQuery (if returnQuery option is true)\n\n // we could have here also returnRecords/returnMatchRecords/returnNonMatchRecord options that could be turned false for not to return actual record data\n\n // matchStatus.status: boolean, true if matcher retrieved and handled all found candidate records, false if it did not\n // matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries',empty string/undefined), reason for stopping retrieving or handling the candidate records\n // - only one stopReason is returned (if there would be several possible stopReasons, stopReason is picked in the above order)\n // - currently stopReason can be non-empty also in cases where status is true, if matcher hit the stop reason when handling the last available candidate record\n\n function returnResult({matches, state, stopReason, nonMatches, duplicateCount, candidateCount, nonMatchCount}) {\n checkCounts({matches, nonMatches, candidateCount, duplicateCount, nonMatchCount});\n const matchStatus = getMatchState(state, stopReason);\n // add nonMatches to result only if returnNonMatches is 'true', otherwise nonMatches have not been gathered\n const result = returnNonMatches ? {matches, matchStatus, nonMatches} : {matches, matchStatus};\n debugData(`${JSON.stringify(result)}`);\n return result;\n\n // note that in cases where the matching has been stopped because of maxMatches checkCounts won't (in most cases) match\n\n function checkCounts({matches, nonMatches, candidateCount, duplicateCount, nonMatchCount}) {\n const matchCount = matches.length;\n debugData(`Return nonMatches: ${returnNonMatches}`);\n const chosenNonMatchCount = returnNonMatches ? nonMatches.length : nonMatchCount;\n const totalHandled = matchCount + nonMatchCount + duplicateCount;\n debug(`candidateCount: ${candidateCount}, matches: ${matchCount}, nonMatches: ${chosenNonMatchCount}, duplicateCount: ${duplicateCount}`);\n debug(`We got result for ${totalHandled} / ${candidateCount} retrieved candidates`);\n if (totalHandled !== candidateCount) {\n debug(`WARNING: Missing results for ${candidateCount - totalHandled} candidates`);\n return;\n }\n return;\n }\n\n function getMatchState(state, stopReason) {\n debugData(`${JSON.stringify(state)}`);\n debug(`Queries left ${state.queriesLeft}, Searches for current query left: ${state.resultSetOffset && state.resultSetOffset <= state.totalRecords}, non-retrieved records: ${state.totalRecords - state.queryCandidateCounter}, maxedQueries (${state.maxedQueries.length}): ${state.maxedQueries}`);\n\n debugData(`StopReason: <${stopReason}>`);\n\n const searchesLeft = state.resultSetOffset && state.resultSetOffset <= state.totalRecords;\n const nonRetrieved = searchesLeft ? state.totalRecords - state.queryCandidateCounter : 0;\n debugData(`nonRetrieved: ${nonRetrieved}`);\n\n if (state.queriesLeft > 0 || nonRetrieved > 0 || state.maxedQueries.length > 0) {\n const maxedQueriesStopReason = state.maxedQueries.length > 0 ? 'maxedQueries' : '';\n const newStopReason = stopReason === '' ? maxedQueriesStopReason : stopReason;\n debugData(`MaxedQueriesStopReason: <${maxedQueriesStopReason}>`);\n debugData(`NewStopReason: <${newStopReason}>`);\n debug(`Match status: false`);\n return {status: false, stopReason: newStopReason};\n }\n\n debug(`Match status: true`);\n return {status: true, stopReason};\n }\n }\n\n function iterateRecords({records, recordSetSize, maxMatches, matches = [], nonMatches = [], recordMatches = [], recordNonMatches = [], recordCount = 0, recordDuplicateCount = 0, recordNonMatchCount = 0}) {\n\n // recordSetSize : total amount of records in the current record set\n // recordCount : amount of records from the current record set that have been handled\n // maxMatches : setting for maximum amount found by current matcher job before the matcher job is stopped\n // recordDuplicateCount : amount of records from the current record set that are already included in matches/nonMatches results\n // recordNonMatchCount: amount of records from the current record set that are nonMatches (only is returnNonMatches setting is false)\n\n // records : non-handled records in the current record set\n // matches : found matches in the current matcher job\n // recordMatches : found matches in the current record set\n // recordNonMatches : found nonMatches in the current record set (only if returnNonMatches setting is true)\n\n const [candidate] = records;\n const newRecordCount = candidate ? recordCount + 1 : recordCount;\n\n // The matcher uses same matchDetection strategy for candidates from all candidate-searches -> matchDetection result for the same candidate is always same\n // Exceptions would happen if the candidate would have been updated in the database between candidate searches\n // Note that if returnNonMatches is false, matcher won't remember candidates that didn't match, so they will be matched again everytime they are retrieved by\n // different candidate search queries. Same candidate search query won't have duplicate records.\n\n if (candidate) {\n\n if (candidateNotInMatches(matches.concat(nonMatches), candidate)) {\n const {record: candidateRecord, id: candidateId} = candidate;\n debug(`Running matchDetection for record ${candidateId} (${newRecordCount}/${recordSetSize})`);\n const detectionResult = detect(record, candidateRecord);\n return handleDetectionResult(detectionResult, candidateId, candidateRecord);\n }\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches, recordCount: newRecordCount, recordNonMatches, recordDuplicateCount: recordDuplicateCount + 1, recordNonMatchCount});\n }\n\n debug(`No more candidates, record set (${recordCount}/${recordSetSize}) done, ${recordMatches.length} matches found, ${recordDuplicateCount} candidates already handled, ${returnNonMatches ? `${recordNonMatches.length}` : `${recordNonMatchCount}`} nonMatches found.`);\n return {matches: recordMatches, nonMatches: returnNonMatches ? recordNonMatches : [], duplicateCount: recordDuplicateCount, nonMatchCount: recordNonMatchCount};\n\n function handleDetectionResult(detectionResult, candidateId, candidateRecord) {\n debugData(`MatchDetection results for ${candidateId} (${newRecordCount}/${recordSetSize}): ${JSON.stringify(detectionResult)}`);\n\n if (detectionResult.match || returnNonMatches) {\n debug(`${detectionResult.match ? `Record ${candidateId} (${newRecordCount}/${recordSetSize}) is a match!` : `Record ${candidateId} (${newRecordCount}/${recordSetSize}) is NOT a match!`}`);\n debugData(`Strategy: ${JSON.stringify(detectionResult.strategy)}, Treshold: ${JSON.stringify(detectionResult.treshold)}`);\n\n const matchResult = {\n probability: detectionResult.probability,\n candidate: {\n id: candidateId,\n record: candidateRecord\n }\n };\n const strategyResult = {\n strategy: detectionResult.strategy,\n treshold: detectionResult.treshold\n };\n const newMatch = returnStrategy ? {...matchResult, ...strategyResult} : {...matchResult};\n\n debugData(`${JSON.stringify(newMatch)}`);\n\n return handleRecordMatch(detectionResult.match, newMatch);\n }\n\n const newRecordNonMatchCount = recordNonMatchCount + 1;\n debugData(`- Total nonMatches after this detection: ${newRecordNonMatchCount}`);\n\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches, recordCount: newRecordCount, recordNonMatches, recordDuplicateCount, recordNonMatchCount: newRecordNonMatchCount});\n }\n\n function handleRecordMatch(isMatch, newMatch) {\n const newRecordMatches = isMatch ? recordMatches.concat(newMatch) : recordMatches;\n const newRecordNonMatches = isMatch ? recordNonMatches : recordNonMatches.concat(newMatch);\n\n debugData(`- Total matches after this detection: ${matches.concat(newRecordMatches).length} (max: ${maxMatches})`);\n\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Total nonMatches after this detection: ${nonMatches.concat(newRecordNonMatches).length}`);\n }\n\n if (maxMatchesFound({matches: matches.concat(newRecordMatches), maxMatches})) {\n debug(`MaxMatches (${maxMatches}) reached, handled candidates in record set: ${newRecordCount} non-handled candidates in record set ${recordSetSize - newRecordCount}`);\n return {matches: newRecordMatches, nonMatches: returnNonMatches ? newRecordNonMatches : [], duplicateCount: recordDuplicateCount, nonMatchCount: recordNonMatchCount};\n }\n\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches: newRecordMatches, recordCount: newRecordCount, recordNonMatches: returnNonMatches ? newRecordNonMatches : [], duplicateCount: recordDuplicateCount, recordNonMatchCount});\n }\n\n function candidateNotInMatches(matches, candidate) {\n debug(`Checking that record ${candidate.id} is not already included in ${matches.length} matches/nonMatches`);\n const newCandidateId = candidate.id;\n debugData(`newCandidateId: ${newCandidateId}`);\n const result = matches.find(({candidate}) => candidate.id === newCandidateId);\n debugData(`Result: ${result}`);\n if (result) {\n debug(`${candidate.id} was already handled.`);\n return false;\n }\n debug(`${candidate.id} not found in matches/nonMatches`);\n return true;\n }\n }\n\n function maxMatchesFound({matches, maxMatches}) {\n if (maxMatches && matches.length >= maxMatches) {\n debug(`Stopping recordSet iteration: maxMatches (${maxMatches}) for matcher job found.`);\n return true;\n }\n }\n };\n};\n"],"file":"index.js"}
|
|
1
|
+
{"version":3,"sources":["../src/index.js"],"names":["detection","detectionOptions","search","searchOptions","maxMatches","maxCandidates","returnStrategy","returnQuery","returnNonMatches","returnFailures","debug","debugData","extend","JSON","stringify","detect","record","iterate","initialState","matches","candidateCount","nonMatches","duplicateCount","nonMatchCount","conversionFailures","matchErrors","records","failures","state","length","recordSetSize","failureSetSize","newCandidateCount","newConversionFailures","concat","handleRecordSet","queriesLeft","searchCounter","query","returnResult","stopReason","matchResult","iterateRecords","newDuplicateCount","newNonMatchCount","newMatches","newNonMatches","newMatchErrors","handleMatchResult","maxMatchesFound","maxCandidatesRetrieved","addQuery","map","match","matchQuery","conversionFailureCount","matchErrorCount","checkCounts","matchStatus","getMatchState","matchesResult","result","matchCount","chosenNonMatchCount","totalHandled","conversionFailuresCount","resultSetOffset","totalRecords","queryCandidateCounter","maxedQueries","searchesLeft","nonRetrieved","maxedQueriesStopReason","undefined","conversionFailuresStopReason","matchErrorsStopReason","newStopReason","status","recordMatches","recordNonMatches","recordCount","recordDuplicateCount","recordNonMatchCount","recordMatchErrors","candidate","newRecordCount","candidateNotInMatches","candidateRecord","id","candidateId","detectionResult","handleDetectionResult","error","matchError","payload","message","newRecordMatchErrors","slice","strategy","treshold","probability","strategyResult","newMatch","handleRecordMatch","newRecordNonMatchCount","isMatch","newRecordMatches","newRecordNonMatches","newCandidateId","find"],"mappings":";;;;;;;AA4BA;;AACA;;;;AACA;;;;;;;;;;AA9BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAKA;eAIe,CAAC;AAACA,EAAAA,SAAS,EAAEC,gBAAZ;AAA8BC,EAAAA,MAAM,EAAEC,aAAtC;AAAqDC,EAAAA,UAAU,GAAG,CAAlE;AAAqEC,EAAAA,aAAa,GAAG,EAArF;AAAyFC,EAAAA,cAAc,GAAG,KAA1G;AAAiHC,EAAAA,WAAW,GAAG,KAA/H;AAAsIC,EAAAA,gBAAgB,GAAG,KAAzJ;AAAgKC,EAAAA,cAAc,GAAG;AAAjL,CAAD,KAA6L;AAC1M,QAAMC,KAAK,GAAG,oBAAkB,yCAAlB,CAAd;AACA,QAAMC,SAAS,GAAGD,KAAK,CAACE,MAAN,CAAa,MAAb,CAAlB;AAEAD,EAAAA,SAAS,CAAE,qBAAoBE,IAAI,CAACC,SAAL,CAAeb,gBAAf,CAAiC,EAAvD,CAAT;AACAU,EAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAeX,aAAf,CAA8B,EAAjD,CAAT;AACAQ,EAAAA,SAAS,CAAE,eAAcE,IAAI,CAACC,SAAL,CAAeV,UAAf,CAA2B,EAA3C,CAAT;AACAO,EAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAeT,aAAf,CAA8B,EAAjD,CAAT;AACAM,EAAAA,SAAS,CAAE,mBAAkBE,IAAI,CAACC,SAAL,CAAeR,cAAf,CAA+B,EAAnD,CAAT;AACAK,EAAAA,SAAS,CAAE,gBAAeE,IAAI,CAACC,SAAL,CAAeP,WAAf,CAA4B,EAA7C,CAAT;AACAI,EAAAA,SAAS,CAAE,qBAAoBE,IAAI,CAACC,SAAL,CAAeN,gBAAf,CAAiC,EAAvD,CAAT;AACAG,EAAAA,SAAS,CAAE,mBAAkBE,IAAI,CAACC,SAAL,CAAeL,cAAf,CAA+B,EAAnD,CAAT;AAGA,QAAMM,MAAM,GAAG,4BAAyBd,gBAAzB,EAA2CK,cAA3C,CAAf;AAEA,SAAOU,MAAM,IAAI;AACf,UAAMd,MAAM,GAAG,6BAAsB,EAAC,GAAGC,aAAJ;AAAmBa,MAAAA,MAAnB;AAA2BX,MAAAA;AAA3B,KAAtB,CAAf;AACA,WAAOY,OAAO,CAAC,EAAD,CAAd,CAFe,CAIf;AACA;AACA;AACA;AAEA;AACA;AACA;AACA;AACA;AACA;AACA;;AAEA,mBAAeA,OAAf,CAAuB;AAACC,MAAAA,YAAY,GAAG,EAAhB;AAAoBC,MAAAA,OAAO,GAAG,EAA9B;AAAkCC,MAAAA,cAAc,GAAG,CAAnD;AAAsDC,MAAAA,UAAU,GAAG,EAAnE;AAAuEC,MAAAA,cAAc,GAAG,CAAxF;AAA2FC,MAAAA,aAAa,GAAG,CAA3G;AAA8GC,MAAAA,kBAAkB,GAAG,EAAnI;AAAuIC,MAAAA,WAAW,GAAG;AAArJ,KAAvB,EAAiL;AAC/Kd,MAAAA,SAAS,CAAE,kCAAF,CAAT;AACA,YAAM;AAACe,QAAAA,OAAD;AAAUC,QAAAA,QAAV;AAAoB,WAAGC;AAAvB,UAAgC,MAAM1B,MAAM,CAACgB,YAAD,CAAlD;AAEAP,MAAAA,SAAS,CAAE,kBAAiBE,IAAI,CAACC,SAAL,CAAec,KAAf,CAAsB,cAAaT,OAAO,CAACU,MAAO,qBAAoBT,cAAe,iBAAgBC,UAAU,CAACQ,MAAO,oBAAmBN,aAAc,yBAAwBC,kBAAmB,kBAAiBC,WAAW,CAACI,MAAO,EAA1P,CAAT;AACA,YAAMC,aAAa,GAAGJ,OAAO,CAACG,MAA9B;AACA,YAAME,cAAc,GAAGJ,QAAQ,CAACE,MAAhC;AACA,YAAMG,iBAAiB,GAAGZ,cAAc,GAAGU,aAAjB,GAAiCC,cAA3D;AAEA,YAAME,qBAAqB,GAAGT,kBAAkB,CAACU,MAAnB,CAA0BP,QAA1B,CAA9B;AACAhB,MAAAA,SAAS,CAAE,aAAYgB,QAAQ,CAACE,MAAO,yBAAwBL,kBAAkB,CAACK,MAAO,4BAA2BI,qBAAqB,CAACJ,MAAO,EAAxI,CAAT;;AAEA,UAAIC,aAAa,GAAG,CAApB,EAAuB;AACrB,eAAOK,eAAe,EAAtB;AACD;;AAED,UAAIP,KAAK,CAACQ,WAAN,GAAoB,CAAxB,EAA2B;AACzB1B,QAAAA,KAAK,CAAE,oBAAmBkB,KAAK,CAACS,aAAc,QAAOT,KAAK,CAACU,KAAM,mBAAkBV,KAAK,CAACQ,WAAY,eAAhG,CAAL;AACA,eAAOnB,OAAO,CAAC;AAACC,UAAAA,YAAY,EAAEU,KAAf;AAAsBT,UAAAA,OAAtB;AAA+BC,UAAAA,cAAc,EAAEY,iBAA/C;AAAkEX,UAAAA,UAAlE;AAA8EE,UAAAA,aAA9E;AAA6FD,UAAAA,cAA7F;AAA6GE,UAAAA,kBAAkB,EAAES,qBAAjI;AAAwJR,UAAAA;AAAxJ,SAAD,CAAd;AACD;;AAEDf,MAAAA,KAAK,CAAE,wEAAuES,OAAO,CAACU,MAAO,EAAxF,CAAL;AACA,aAAOU,YAAY,CAAC;AAACpB,QAAAA,OAAD;AAAUS,QAAAA,KAAV;AAAiBY,QAAAA,UAAU,EAAE,EAA7B;AAAiCnB,QAAAA,UAAjC;AAA6CE,QAAAA,aAA7C;AAA4DH,QAAAA,cAAc,EAAEY,iBAA5E;AAA+FV,QAAAA,cAA/F;AAA+GE,QAAAA,kBAAkB,EAAES,qBAAnI;AAA0JR,QAAAA;AAA1J,OAAD,CAAnB;;AAEA,eAASU,eAAT,GAA2B;AACzBzB,QAAAA,KAAK,CAAE,0BAAyBoB,aAAc,qDAAoDF,KAAK,CAACS,aAAc,eAAcT,KAAK,CAACU,KAAM,EAA3I,CAAL;AAEA,cAAMG,WAAW,GAAGC,cAAc,CAAC;AAAChB,UAAAA,OAAD;AAAUI,UAAAA,aAAV;AAAyB1B,UAAAA,UAAzB;AAAqCe,UAAAA,OAArC;AAA8CE,UAAAA,UAA9C;AAA0DE,UAAAA;AAA1D,SAAD,CAAlC;AAEA,cAAMoB,iBAAiB,GAAGrB,cAAc,GAAGmB,WAAW,CAACnB,cAAvD;AACA,cAAMsB,gBAAgB,GAAGrB,aAAa,GAAGkB,WAAW,CAAClB,aAArD;AACA,cAAM;AAACsB,UAAAA,UAAD;AAAaC,UAAAA,aAAb;AAA4BC,UAAAA;AAA5B,YAA8CC,iBAAiB,CAACP,WAAD,EAActB,OAAd,EAAuBE,UAAvB,EAAmCI,WAAnC,CAArE;;AAEA,YAAIwB,eAAe,CAAC;AAAC9B,UAAAA,OAAO,EAAE0B,UAAV;AAAsBzC,UAAAA;AAAtB,SAAD,CAAnB,EAAwD;AACtD,iBAAOmC,YAAY,CAAC;AAACpB,YAAAA,OAAO,EAAE0B,UAAV;AAAsBjB,YAAAA,KAAtB;AAA6BY,YAAAA,UAAU,EAAE,YAAzC;AAAuDnB,YAAAA,UAAU,EAAEyB,aAAnE;AAAkFxB,YAAAA,cAAc,EAAEqB,iBAAlG;AAAqHvB,YAAAA,cAAc,EAAEY,iBAArI;AAAwJT,YAAAA,aAAa,EAAEqB,gBAAvK;AAAyLpB,YAAAA,kBAAkB,EAAES,qBAA7M;AAAoOR,YAAAA,WAAW,EAAEsB;AAAjP,WAAD,CAAnB;AACD;;AAED,YAAIG,sBAAsB,CAAClB,iBAAD,EAAoB3B,aAApB,CAA1B,EAA8D;AAC5D,iBAAOkC,YAAY,CAAC;AAACpB,YAAAA,OAAO,EAAE0B,UAAV;AAAsBjB,YAAAA,KAAtB;AAA6BY,YAAAA,UAAU,EAAE,eAAzC;AAA0DnB,YAAAA,UAAU,EAAEyB,aAAtE;AAAqFxB,YAAAA,cAAc,EAAEqB,iBAArG;AAAwHvB,YAAAA,cAAc,EAAEY,iBAAxI;AAA2JT,YAAAA,aAAa,EAAEqB,gBAA1K;AAA4LpB,YAAAA,kBAAkB,EAAES,qBAAhN;AAAuOR,YAAAA,WAAW,EAAEsB;AAApP,WAAD,CAAnB;AACD;;AAED,eAAO9B,OAAO,CAAC;AAACC,UAAAA,YAAY,EAAEU,KAAf;AAAsBT,UAAAA,OAAO,EAAE0B,UAA/B;AAA2CzB,UAAAA,cAAc,EAAEY,iBAA3D;AAA8EX,UAAAA,UAAU,EAAEyB,aAA1F;AAAyGxB,UAAAA,cAAc,EAAEqB,iBAAzH;AAA4IpB,UAAAA,aAAa,EAAEqB,gBAA3J;AAA6KpB,UAAAA,kBAAkB,EAAES,qBAAjM;AAAwNR,UAAAA,WAAW,EAAEsB;AAArO,SAAD,CAAd;AACD;;AAED,eAASC,iBAAT,CAA2BP,WAA3B,EAAwCtB,OAAxC,EAAiDE,UAAjD,EAA6DI,WAA7D,EAA0E;AACxEd,QAAAA,SAAS,CAAE,4CAA2C8B,WAAW,CAACtB,OAAZ,CAAoBU,MAAO,EAAxE,CAAT,CADwE,CAExE;;AACA,YAAIrB,gBAAJ,EAAsB;AACpBG,UAAAA,SAAS,CAAE,+CAA8C8B,WAAW,CAACpB,UAAZ,CAAuBQ,MAAO,EAA9E,CAAT;AACD;;AAED,cAAMgB,UAAU,GAAG1B,OAAO,CAACe,MAAR,CAAe3B,WAAW,GAAG4C,QAAQ,CAACV,WAAW,CAACtB,OAAb,CAAX,GAAmCsB,WAAW,CAACtB,OAAzE,CAAnB;AACA,cAAM2B,aAAa,GAAGtC,gBAAgB,GAAGa,UAAU,CAACa,MAAX,CAAkB3B,WAAW,GAAG4C,QAAQ,CAACV,WAAW,CAACpB,UAAb,CAAX,GAAsCoB,WAAW,CAACpB,UAA/E,CAAH,GAAgG,EAAtI;AACA,cAAM0B,cAAc,GAAGtB,WAAW,CAACS,MAAZ,CAAmBO,WAAW,CAAChB,WAA/B,CAAvB;AAEAd,QAAAA,SAAS,CAAE,8BAA6BkC,UAAU,CAAChB,MAAO,EAAjD,CAAT,CAXwE,CAYxE;;AACA,YAAIrB,gBAAJ,EAAsB;AACpBG,UAAAA,SAAS,CAAE,iCAAgCmC,aAAa,CAACjB,MAAO,EAAvD,CAAT;AACD;;AAEDlB,QAAAA,SAAS,CAAE,gBAAeE,IAAI,CAACC,SAAL,CAAe2B,WAAf,CAA4B,EAA7C,CAAT;AACA9B,QAAAA,SAAS,CAAE,oBAAmBE,IAAI,CAACC,SAAL,CAAeW,WAAf,CAA4B,mCAAkCZ,IAAI,CAACC,SAAL,CAAe2B,WAAW,CAAChB,WAA3B,CAAwC,sBAAqBZ,IAAI,CAACC,SAAL,CAAeiC,cAAf,CAA+B,EAA/K,CAAT;AAEApC,QAAAA,SAAS,CAAE,kCAAiCoC,cAAc,CAAClB,MAAO,EAAzD,CAAT;AAEA,eAAO;AAACgB,UAAAA,UAAD;AAAaC,UAAAA,aAAb;AAA4BC,UAAAA;AAA5B,SAAP;AACD;;AAED,eAASI,QAAT,CAAkBhC,OAAlB,EAA2B;AACzBR,QAAAA,SAAS,CAAE,gBAAeiB,KAAK,CAACU,KAAM,aAA7B,CAAT;AACA,eAAOnB,OAAO,CAACiC,GAAR,CAAaC,KAAD,KAAY,EAAC,GAAGA,KAAJ;AAAWC,UAAAA,UAAU,EAAE1B,KAAK,CAACU;AAA7B,SAAZ,CAAZ,CAAP;AACD;;AAED,eAASY,sBAAT,CAAgC9B,cAAhC,EAAgDf,aAAhD,EAA+D;AAC7DM,QAAAA,SAAS,CAAE,gDAA+CqB,iBAAkB,UAAS3B,aAAc,GAA1F,CAAT;;AACA,YAAIA,aAAa,IAAIe,cAAc,IAAIf,aAAvC,EAAsD;AACpDK,UAAAA,KAAK,CAAE,gEAA+DU,cAAe,MAAKf,aAAc,sBAAnG,CAAL;AACA,iBAAO,IAAP;AACD;AACF;AACF,KAlGc,CAoGf;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;AAEA;AACA;AACA;AACA;;;AAEA,aAASkC,YAAT,CAAsB;AAACpB,MAAAA,OAAD;AAAUS,MAAAA,KAAV;AAAiBY,MAAAA,UAAjB;AAA6BnB,MAAAA,UAA7B;AAAyCC,MAAAA,cAAzC;AAAyDF,MAAAA,cAAzD;AAAyEG,MAAAA,aAAzE;AAAwFC,MAAAA,kBAAxF;AAA4GC,MAAAA;AAA5G,KAAtB,EAAgJ;AAC9I,YAAM8B,sBAAsB,GAAG/B,kBAAkB,CAACK,MAAlD;AACA,YAAM2B,eAAe,GAAG/B,WAAW,CAACI,MAApC;AACA4B,MAAAA,WAAW,CAAC;AAACtC,QAAAA,OAAD;AAAUE,QAAAA,UAAV;AAAsBD,QAAAA,cAAtB;AAAsCE,QAAAA,cAAtC;AAAsDC,QAAAA,aAAtD;AAAqEgC,QAAAA,sBAArE;AAA6FC,QAAAA;AAA7F,OAAD,CAAX;AACA,YAAME,WAAW,GAAGC,aAAa,CAAC/B,KAAD,EAAQY,UAAR,EAAoBe,sBAApB,EAA4CC,eAA5C,CAAjC,CAJ8I,CAK9I;;AACA,YAAMI,aAAa,GAAGpD,gBAAgB,GAAG;AAACW,QAAAA,OAAD;AAAUuC,QAAAA,WAAV;AAAuBrC,QAAAA;AAAvB,OAAH,GAAwC;AAACF,QAAAA,OAAD;AAAUuC,QAAAA;AAAV,OAA9E;AACA,YAAM/B,QAAQ,GAAG,CAAC,GAAGH,kBAAJ,EAAwB,GAAGC,WAA3B,CAAjB;AACA,YAAMoC,MAAM,GAAGpD,cAAc,GAAG,EAAC,GAAGmD,aAAJ;AAAmBpC,QAAAA,kBAAkB,EAAEG;AAAvC,OAAH,GAAsDiC,aAAnF;AACAjD,MAAAA,SAAS,CAAE,kBAAiBF,cAAe,EAAlC,CAAT;AACAE,MAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAe+C,MAAf,CAAuB,EAA3B,CAAT;AACA,aAAOA,MAAP,CAX8I,CAa9I;;AAEA,eAASJ,WAAT,CAAqB;AAACtC,QAAAA,OAAD;AAAUE,QAAAA,UAAV;AAAsBD,QAAAA,cAAtB;AAAsCE,QAAAA,cAAtC;AAAsDC,QAAAA,aAAtD;AAAqEgC,QAAAA,sBAArE;AAA6FC,QAAAA;AAA7F,OAArB,EAAoI;AAClI,cAAMM,UAAU,GAAG3C,OAAO,CAACU,MAA3B;AACAlB,QAAAA,SAAS,CAAE,sBAAqBH,gBAAiB,EAAxC,CAAT;AACA,cAAMuD,mBAAmB,GAAGvD,gBAAgB,GAAGa,UAAU,CAACQ,MAAd,GAAuBN,aAAnE;AACA,cAAMyC,YAAY,GAAGF,UAAU,GAAGC,mBAAb,GAAmCzC,cAAxD;AACAZ,QAAAA,KAAK,CAAE,mBAAkBU,cAAe,cAAa0C,UAAW,iBAAgBC,mBAAoB,qBAAoBzC,cAAe,6BAA4BiC,sBAAuB,sBAAqBC,eAAgB,EAA1N,CAAL;AACA9C,QAAAA,KAAK,CAAE,qBAAoBsD,YAAa,MAAK5C,cAAe,uBAAvD,CAAL;;AACA,YAAI4C,YAAY,KAAK5C,cAArB,EAAqC;AACnCV,UAAAA,KAAK,CAAE,gCAA+BU,cAAc,GAAG4C,YAAa,aAA/D,CAAL;AACA;AACD;;AACD;AACD,OA3B6I,CA6B9I;;;AACA,eAASL,aAAT,CAAuB/B,KAAvB,EAA8BY,UAA9B,EAA0CyB,uBAA1C,EAAmET,eAAnE,EAAoF;AAClF7C,QAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAec,KAAf,CAAsB,EAA1B,CAAT;AACAlB,QAAAA,KAAK,CAAE,UAASuD,uBAAwB,oDAAnC,CAAL;AACAvD,QAAAA,KAAK,CAAE,UAAS8C,eAAgB,uDAA3B,CAAL;AACA9C,QAAAA,KAAK,CAAE,gBAAekB,KAAK,CAACQ,WAAY,sCAAqCR,KAAK,CAACsC,eAAN,IAAyBtC,KAAK,CAACsC,eAAN,IAAyBtC,KAAK,CAACuC,YAAa,4BAA2BvC,KAAK,CAACuC,YAAN,GAAqBvC,KAAK,CAACwC,qBAAsB,mBAAkBxC,KAAK,CAACyC,YAAN,CAAmBxC,MAAO,MAAKD,KAAK,CAACyC,YAAa,EAA7R,CAAL;AAEA1D,QAAAA,SAAS,CAAE,gBAAe6B,UAAW,GAA5B,CAAT;AAEA,cAAM8B,YAAY,GAAG1C,KAAK,CAACsC,eAAN,IAAyBtC,KAAK,CAACsC,eAAN,IAAyBtC,KAAK,CAACuC,YAA7E;AACA,cAAMI,YAAY,GAAGD,YAAY,GAAG1C,KAAK,CAACuC,YAAN,GAAqBvC,KAAK,CAACwC,qBAA9B,GAAsD,CAAvF;AACAzD,QAAAA,SAAS,CAAE,iBAAgB4D,YAAa,EAA/B,CAAT,CAVkF,CAYlF;AACA;;AAEA,YAAI3C,KAAK,CAACQ,WAAN,GAAoB,CAApB,IAAyBmC,YAAY,GAAG,CAAxC,IAA6C3C,KAAK,CAACyC,YAAN,CAAmBxC,MAAnB,GAA4B,CAAzE,IAA8E0B,sBAAsB,GAAG,CAAvG,IAA4GC,eAAe,GAAG,CAAlI,EAAqI;AACnI,gBAAMgB,sBAAsB,GAAG5C,KAAK,CAACyC,YAAN,CAAmBxC,MAAnB,GAA4B,CAA5B,GAAgC,cAAhC,GAAiD4C,SAAhF;AACA,gBAAMC,4BAA4B,GAAGnB,sBAAsB,GAAG,CAAzB,GAA6B,oBAA7B,GAAoDkB,SAAzF;AACA,gBAAME,qBAAqB,GAAGnB,eAAe,GAAG,CAAlB,GAAsB,aAAtB,GAAsCiB,SAApE;AACA,gBAAMG,aAAa,GAAGpC,UAAU,KAAK,EAAf,IAAqBA,UAAU,KAAKiC,SAApC,GAAgDD,sBAAsB,IAAIE,4BAA1B,IAA0DC,qBAA1G,GAAkInC,UAAxJ;AACA7B,UAAAA,SAAS,CAAE,4BAA2B6D,sBAAuB,GAApD,CAAT;AACA7D,UAAAA,SAAS,CAAE,gCAA+B+D,4BAA6B,GAA9D,CAAT;AACA/D,UAAAA,SAAS,CAAE,0BAAyBgE,qBAAsB,GAAjD,CAAT;AACAhE,UAAAA,SAAS,CAAE,mBAAkBiE,aAAc,GAAlC,CAAT;AACAlE,UAAAA,KAAK,CAAE,qBAAF,CAAL;AACA,iBAAO;AAACmE,YAAAA,MAAM,EAAE,KAAT;AAAgBrC,YAAAA,UAAU,EAAEoC;AAA5B,WAAP;AACD;;AAEDlE,QAAAA,KAAK,CAAE,oBAAF,CAAL;AACA,eAAO;AAACmE,UAAAA,MAAM,EAAE,IAAT;AAAerC,UAAAA;AAAf,SAAP;AACD;AACF,KAlLc,CAoLf;AACA;AACA;AACA;AAEA;;;AACA,aAASE,cAAT,CAAwB;AAAChB,MAAAA,OAAD;AAAUI,MAAAA,aAAV;AAAyB1B,MAAAA,UAAzB;AAAqCe,MAAAA,OAAO,GAAG,EAA/C;AAAmDE,MAAAA,UAAU,GAAG,EAAhE;AAAoEyD,MAAAA,aAAa,GAAG,EAApF;AAAwFC,MAAAA,gBAAgB,GAAG,EAA3G;AAA+GC,MAAAA,WAAW,GAAG,CAA7H;AAAgIC,MAAAA,oBAAoB,GAAG,CAAvJ;AAA0JC,MAAAA,mBAAmB,GAAG,CAAhL;AAAmLC,MAAAA,iBAAiB,GAAG;AAAvM,KAAxB,EAAoO;AAElO;AACA;AACA;AACA;AACA;AAEA;AACA;AACA;AACA;AACA;AAEA,YAAM,CAACC,SAAD,IAAc1D,OAApB;AACA,YAAM2D,cAAc,GAAGD,SAAS,GAAGJ,WAAW,GAAG,CAAjB,GAAqBA,WAArD,CAfkO,CAiBlO;AACA;AACA;AACA;;AAEA;AACN;AACA;AACA;AACA;AACA;AACA;;AAEM,UAAII,SAAJ,EAAe;AAEb;AACA,YAAIE,qBAAqB,CAACnE,OAAO,CAACe,MAAR,CAAeb,UAAf,CAAD,EAA6B+D,SAA7B,CAAzB,EAAkE;AAChE,gBAAM;AAACpE,YAAAA,MAAM,EAAEuE,eAAT;AAA0BC,YAAAA,EAAE,EAAEC;AAA9B,cAA6CL,SAAnD;;AACA,cAAI;AACF1E,YAAAA,KAAK,CAAE,qCAAoC+E,WAAY,KAAIJ,cAAe,IAAGvD,aAAc,GAAtF,CAAL,CADE,CAEF;;AACA,kBAAM4D,eAAe,GAAG3E,MAAM,CAACC,MAAD,EAASuE,eAAT,CAA9B;AAEA,mBAAOI,qBAAqB,CAACD,eAAD,EAAkBD,WAAlB,EAA+BF,eAA/B,CAA5B;AACD,WAND,CAME,OAAOK,KAAP,EAAc;AACdlF,YAAAA,KAAK,CAAE,2CAA0C+E,WAAY,KAAIG,KAAM,EAAlE,CAAL;AAEA,kBAAMC,UAAU,GAAG;AAAChB,cAAAA,MAAM,EAAE,GAAT;AAAciB,cAAAA,OAAO,EAAE;AAACC,gBAAAA,OAAO,EAAG,wCAAuCN,WAAY,KAAIG,KAAK,CAACG,OAAQ,GAAhF;AAAoFP,gBAAAA,EAAE,EAAEC;AAAxF;AAAvB,aAAnB;AACA,kBAAMO,oBAAoB,GAAGb,iBAAiB,CAACjD,MAAlB,CAAyB2D,UAAzB,CAA7B;AACA,mBAAOnD,cAAc,CAAC;AAAChB,cAAAA,OAAO,EAAEA,OAAO,CAACuE,KAAR,CAAc,CAAd,CAAV;AAA4BnE,cAAAA,aAA5B;AAA2C1B,cAAAA,UAA3C;AAAuDe,cAAAA,OAAvD;AAAgE2D,cAAAA,aAAhE;AAA+EE,cAAAA,WAAW,EAAEK,cAA5F;AAA4GN,cAAAA,gBAA5G;AAA8HE,cAAAA,oBAA9H;AAAoJC,cAAAA,mBAApJ;AAAyKC,cAAAA,iBAAiB,EAAEa;AAA5L,aAAD,CAArB;AACD;AACF;;AAED,eAAOtD,cAAc,CAAC;AAAChB,UAAAA,OAAO,EAAEA,OAAO,CAACuE,KAAR,CAAc,CAAd,CAAV;AAA4BnE,UAAAA,aAA5B;AAA2C1B,UAAAA,UAA3C;AAAuDe,UAAAA,OAAvD;AAAgE2D,UAAAA,aAAhE;AAA+EE,UAAAA,WAAW,EAAEK,cAA5F;AAA4GN,UAAAA,gBAA5G;AAA8HE,UAAAA,oBAAoB,EAAEA,oBAAoB,GAAG,CAA3K;AAA8KC,UAAAA,mBAA9K;AAAmMC,UAAAA;AAAnM,SAAD,CAArB;AACD;;AAEDzE,MAAAA,KAAK,CAAE,mCAAkCsE,WAAY,IAAGlD,aAAc,WAAUgD,aAAa,CAACjD,MAAO,mBAAkBoD,oBAAqB,gCAA+BzE,gBAAgB,GAAI,GAAEuE,gBAAgB,CAAClD,MAAO,EAA9B,GAAmC,GAAEqD,mBAAoB,EAAE,oBAAjP,CAAL;AACA,aAAO;AAAC/D,QAAAA,OAAO,EAAE2D,aAAV;AAAyBzD,QAAAA,UAAU,EAAEb,gBAAgB,GAAGuE,gBAAH,GAAsB,EAA3E;AAA+EzD,QAAAA,cAAc,EAAE2D,oBAA/F;AAAqH1D,QAAAA,aAAa,EAAE2D,mBAApI;AAAyJzD,QAAAA,WAAW,EAAE0D;AAAtK,OAAP;;AAEA,eAASQ,qBAAT,CAA+BD,eAA/B,EAAgDD,WAAhD,EAA6DF,eAA7D,EAA8E;AAC5E5E,QAAAA,SAAS,CAAE,8BAA6B8E,WAAY,KAAIJ,cAAe,IAAGvD,aAAc,MAAKjB,IAAI,CAACC,SAAL,CAAe4E,eAAf,CAAgC,EAApH,CAAT;;AAEA,YAAIA,eAAe,CAACrC,KAAhB,IAAyB7C,gBAA7B,EAA+C;AAC7CE,UAAAA,KAAK,CAAE,GAAEgF,eAAe,CAACrC,KAAhB,GAAyB,UAASoC,WAAY,KAAIJ,cAAe,IAAGvD,aAAc,eAAlF,GAAoG,UAAS2D,WAAY,KAAIJ,cAAe,IAAGvD,aAAc,mBAAmB,EAApL,CAAL;AACAnB,UAAAA,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAL,CAAe4E,eAAe,CAACQ,QAA/B,CAAyC,eAAcrF,IAAI,CAACC,SAAL,CAAe4E,eAAe,CAACS,QAA/B,CAAyC,EAA9G,CAAT;AAEA,gBAAM1D,WAAW,GAAG;AAClB2D,YAAAA,WAAW,EAAEV,eAAe,CAACU,WADX;AAElBhB,YAAAA,SAAS,EAAE;AACTI,cAAAA,EAAE,EAAEC,WADK;AAETzE,cAAAA,MAAM,EAAEuE;AAFC;AAFO,WAApB;AAOA,gBAAMc,cAAc,GAAG;AACrBH,YAAAA,QAAQ,EAAER,eAAe,CAACQ,QADL;AAErBC,YAAAA,QAAQ,EAAET,eAAe,CAACS;AAFL,WAAvB;AAIA,gBAAMG,QAAQ,GAAGhG,cAAc,GAAG,EAAC,GAAGmC,WAAJ;AAAiB,eAAG4D;AAApB,WAAH,GAAyC,EAAC,GAAG5D;AAAJ,WAAxE;AAEA9B,UAAAA,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAL,CAAewF,QAAf,CAAyB,EAA7B,CAAT;AAEA,iBAAOC,iBAAiB,CAACb,eAAe,CAACrC,KAAjB,EAAwBiD,QAAxB,CAAxB;AACD;;AAED,cAAME,sBAAsB,GAAGtB,mBAAmB,GAAG,CAArD;AACAvE,QAAAA,SAAS,CAAE,4CAA2C6F,sBAAuB,EAApE,CAAT;AAEA,eAAO9D,cAAc,CAAC;AAAChB,UAAAA,OAAO,EAAEA,OAAO,CAACuE,KAAR,CAAc,CAAd,CAAV;AAA4BnE,UAAAA,aAA5B;AAA2C1B,UAAAA,UAA3C;AAAuDe,UAAAA,OAAvD;AAAgE2D,UAAAA,aAAhE;AAA+EE,UAAAA,WAAW,EAAEK,cAA5F;AAA4GN,UAAAA,gBAA5G;AAA8HE,UAAAA,oBAA9H;AAAoJC,UAAAA,mBAAmB,EAAEsB,sBAAzK;AAAiMrB,UAAAA;AAAjM,SAAD,CAArB;AACD;;AAED,eAASoB,iBAAT,CAA2BE,OAA3B,EAAoCH,QAApC,EAA8C;AAC5C,cAAMI,gBAAgB,GAAGD,OAAO,GAAG3B,aAAa,CAAC5C,MAAd,CAAqBoE,QAArB,CAAH,GAAoCxB,aAApE;AACA,cAAM6B,mBAAmB,GAAGF,OAAO,GAAG1B,gBAAH,GAAsBA,gBAAgB,CAAC7C,MAAjB,CAAwBoE,QAAxB,CAAzD;AACA,cAAME,sBAAsB,GAAGC,OAAO,GAAGvB,mBAAH,GAAyBA,mBAAmB,GAAG,CAArF;AAEAvE,QAAAA,SAAS,CAAE,yCAAwCQ,OAAO,CAACe,MAAR,CAAewE,gBAAf,EAAiC7E,MAAO,UAASzB,UAAW,GAAtG,CAAT,CAL4C,CAO5C;;AACA,YAAII,gBAAJ,EAAsB;AACpBG,UAAAA,SAAS,CAAE,4CAA2CU,UAAU,CAACa,MAAX,CAAkByE,mBAAlB,EAAuC9E,MAAO,EAA3F,CAAT;AACD;;AACDlB,QAAAA,SAAS,CAAE,+CAA8CuE,mBAAoB,EAApE,CAAT;;AAEA,YAAIjC,eAAe,CAAC;AAAC9B,UAAAA,OAAO,EAAEA,OAAO,CAACe,MAAR,CAAewE,gBAAf,CAAV;AAA4CtG,UAAAA;AAA5C,SAAD,CAAnB,EAA8E;AAC5EM,UAAAA,KAAK,CAAE,eAAcN,UAAW,gDAA+CiF,cAAe,yCAAwCvD,aAAa,GAAGuD,cAAe,EAAhK,CAAL;AACA,iBAAO;AAAClE,YAAAA,OAAO,EAAEuF,gBAAV;AAA4BrF,YAAAA,UAAU,EAAEb,gBAAgB,GAAGmG,mBAAH,GAAyB,EAAjF;AAAqFrF,YAAAA,cAAc,EAAE2D,oBAArG;AAA2H1D,YAAAA,aAAa,EAAEiF,sBAA1I;AAAkK/E,YAAAA,WAAW,EAAE0D;AAA/K,WAAP;AACD;;AAED,eAAOzC,cAAc,CAAC;AAAChB,UAAAA,OAAO,EAAEA,OAAO,CAACuE,KAAR,CAAc,CAAd,CAAV;AAA4BnE,UAAAA,aAA5B;AAA2C1B,UAAAA,UAA3C;AAAuDe,UAAAA,OAAvD;AAAgE2D,UAAAA,aAAa,EAAE4B,gBAA/E;AAAiG1B,UAAAA,WAAW,EAAEK,cAA9G;AAA8HN,UAAAA,gBAAgB,EAAEvE,gBAAgB,GAAGmG,mBAAH,GAAyB,EAAzL;AAA6LrF,UAAAA,cAAc,EAAE2D,oBAA7M;AAAmOC,UAAAA,mBAAmB,EAAEsB,sBAAxP;AAAgR/E,UAAAA,WAAW,EAAE0D;AAA7R,SAAD,CAArB;AACD;;AAED,eAASG,qBAAT,CAA+BnE,OAA/B,EAAwCiE,SAAxC,EAAmD;AACjD1E,QAAAA,KAAK,CAAE,wBAAuB0E,SAAS,CAACI,EAAG,+BAA8BrE,OAAO,CAACU,MAAO,qBAAnF,CAAL;AACA,cAAM+E,cAAc,GAAGxB,SAAS,CAACI,EAAjC;AACA7E,QAAAA,SAAS,CAAE,mBAAkBiG,cAAe,EAAnC,CAAT;AACA,cAAM/C,MAAM,GAAG1C,OAAO,CAAC0F,IAAR,CAAa,CAAC;AAACzB,UAAAA;AAAD,SAAD,KAAiBA,SAAS,CAACI,EAAV,KAAiBoB,cAA/C,CAAf;AACAjG,QAAAA,SAAS,CAAE,WAAUkD,MAAO,EAAnB,CAAT;;AACA,YAAIA,MAAJ,EAAY;AACVnD,UAAAA,KAAK,CAAE,GAAE0E,SAAS,CAACI,EAAG,uBAAjB,CAAL;AACA,iBAAO,KAAP;AACD;;AACD9E,QAAAA,KAAK,CAAE,GAAE0E,SAAS,CAACI,EAAG,kCAAjB,CAAL;AACA,eAAO,IAAP;AACD;AACF;;AAED,aAASvC,eAAT,CAAyB;AAAC9B,MAAAA,OAAD;AAAUf,MAAAA;AAAV,KAAzB,EAAgD;AAC9C,UAAIA,UAAU,IAAIe,OAAO,CAACU,MAAR,IAAkBzB,UAApC,EAAgD;AAC9CM,QAAAA,KAAK,CAAE,6CAA4CN,UAAW,0BAAzD,CAAL;AACA,eAAO,IAAP;AACD;AACF;AACF,GA3TD;AA4TD,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 createSearchInterface, * as candidateSearch from './candidate-search';\nimport createDetectionInterface, * as matchDetection from './match-detection';\n//import inspect from 'util';\n\nexport {candidateSearch, matchDetection};\n\nexport default ({detection: detectionOptions, search: searchOptions, maxMatches = 1, maxCandidates = 25, returnStrategy = false, returnQuery = false, returnNonMatches = false, returnFailures = false}) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:index');\n const debugData = debug.extend('data');\n\n debugData(`DetectionOptions: ${JSON.stringify(detectionOptions)}`);\n debugData(`SearchOptions: ${JSON.stringify(searchOptions)}`);\n debugData(`MaxMatches: ${JSON.stringify(maxMatches)}`);\n debugData(`MaxCandidates: ${JSON.stringify(maxCandidates)}`);\n debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`ReturnQuery: ${JSON.stringify(returnQuery)}`);\n debugData(`ReturnNonMatches: ${JSON.stringify(returnNonMatches)}`);\n debugData(`ReturnFailures: ${JSON.stringify(returnFailures)}`);\n\n\n const detect = createDetectionInterface(detectionOptions, returnStrategy);\n\n return record => {\n const search = createSearchInterface({...searchOptions, record, maxCandidates});\n return iterate({});\n\n // candidateCount : amount of candidate records retrived from SRU for matching, NOT including current record set\n // matches : candidates that have been detected as matches by current matcher job\n // nonMatches : candidates that have been detected as non-matches by current matcher job (only if returnNonMatches is 'true')\n // duplicateCount : amount of candidate records that were retrieved from the SRU but not handled further because they were already found in the matches/nonMatches\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 async function iterate({initialState = {}, matches = [], candidateCount = 0, nonMatches = [], duplicateCount = 0, nonMatchCount = 0, conversionFailures = [], matchErrors = []}) {\n debugData(`Starting next matcher iteration.`);\n const {records, failures, ...state} = await search(initialState);\n\n debugData(`Current state: ${JSON.stringify(state)}, matches: ${matches.length}, candidateCount: ${candidateCount}, nonMatches: ${nonMatches.length}, nonMatchCount: ${nonMatchCount}, conversionFailures: ${conversionFailures}, matchErrors: ${matchErrors.length}`);\n const recordSetSize = records.length;\n const failureSetSize = failures.length;\n const newCandidateCount = candidateCount + recordSetSize + failureSetSize;\n\n const newConversionFailures = conversionFailures.concat(failures);\n debugData(`Failures: ${failures.length}, ConversionFailures: ${conversionFailures.length}, NewConversionFailures: ${newConversionFailures.length}`);\n\n if (recordSetSize > 0) {\n return handleRecordSet();\n }\n\n if (state.queriesLeft > 0) {\n debug(`Empty record set ${state.searchCounter} for ${state.query}, but there are ${state.queriesLeft} queries left`);\n return iterate({initialState: state, matches, candidateCount: newCandidateCount, nonMatches, nonMatchCount, duplicateCount, conversionFailures: newConversionFailures, matchErrors});\n }\n\n debug(`No (more) candidate records to check, no more queries left, matches: ${matches.length}`);\n return returnResult({matches, state, stopReason: '', nonMatches, nonMatchCount, candidateCount: newCandidateCount, duplicateCount, conversionFailures: newConversionFailures, matchErrors});\n\n function handleRecordSet() {\n debug(`Checking record set of ${recordSetSize} candidate records for possible matches, found by ${state.searchCounter} search for ${state.query}`);\n\n const matchResult = iterateRecords({records, recordSetSize, maxMatches, matches, nonMatches, nonMatchCount});\n\n const newDuplicateCount = duplicateCount + matchResult.duplicateCount;\n const newNonMatchCount = nonMatchCount + matchResult.nonMatchCount;\n const {newMatches, newNonMatches, newMatchErrors} = handleMatchResult(matchResult, matches, nonMatches, matchErrors);\n\n if (maxMatchesFound({matches: newMatches, maxMatches})) {\n return returnResult({matches: newMatches, state, stopReason: 'maxMatches', nonMatches: newNonMatches, duplicateCount: newDuplicateCount, candidateCount: newCandidateCount, nonMatchCount: newNonMatchCount, conversionFailures: newConversionFailures, matchErrors: newMatchErrors});\n }\n\n if (maxCandidatesRetrieved(newCandidateCount, maxCandidates)) {\n return returnResult({matches: newMatches, state, stopReason: 'maxCandidates', nonMatches: newNonMatches, duplicateCount: newDuplicateCount, candidateCount: newCandidateCount, nonMatchCount: newNonMatchCount, conversionFailures: newConversionFailures, matchErrors: newMatchErrors});\n }\n\n return iterate({initialState: state, matches: newMatches, candidateCount: newCandidateCount, nonMatches: newNonMatches, duplicateCount: newDuplicateCount, nonMatchCount: newNonMatchCount, conversionFailures: newConversionFailures, matchErrors: newMatchErrors});\n }\n\n function handleMatchResult(matchResult, matches, nonMatches, matchErrors) {\n debugData(`- Amount of new matches from record set: ${matchResult.matches.length}`);\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Amount of new nonMatches from record set: ${matchResult.nonMatches.length}`);\n }\n\n const newMatches = matches.concat(returnQuery ? addQuery(matchResult.matches) : matchResult.matches);\n const newNonMatches = returnNonMatches ? nonMatches.concat(returnQuery ? addQuery(matchResult.nonMatches) : matchResult.nonMatches) : [];\n const newMatchErrors = matchErrors.concat(matchResult.matchErrors);\n\n debugData(`- Total amount of matches: ${newMatches.length}`);\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Total amount of nonMatches: ${newNonMatches.length}`);\n }\n\n debugData(`MatchResult: ${JSON.stringify(matchResult)}`);\n debugData(`Old matchErrors: ${JSON.stringify(matchErrors)}, matchErrors from matchResult: ${JSON.stringify(matchResult.matchErrors)}, New matchErrors: ${JSON.stringify(newMatchErrors)}`);\n\n debugData(`- Total amount of matchErrors: ${newMatchErrors.length}`);\n\n return {newMatches, newNonMatches, newMatchErrors};\n }\n\n function addQuery(matches) {\n debugData(`Adding query ${state.query} to matches`);\n return matches.map((match) => ({...match, matchQuery: state.query}));\n }\n\n function maxCandidatesRetrieved(candidateCount, maxCandidates) {\n debugData(`Total amount of candidate records retrieved: ${newCandidateCount} (max: ${maxCandidates})`);\n if (maxCandidates && candidateCount >= maxCandidates) {\n debug(`Stopped matching because maximum number of candidate records ${candidateCount} / ${maxCandidates} have been retrieved`);\n return true;\n }\n }\n }\n\n // matches : array of matching candidate records\n // nonMatches : array of nonMatching candidate records (if returnNonMatches option is true, otherwise empty array)\n // - candidate.id\n // - candidate.record\n // - probability\n // - strategy (if returnStrategy option is true)\n // - treshold (if returnStrategy option is true)\n // - matchQuery (if returnQuery option is true)\n // failures: array of conversionFailures from candidate-search and matchErrors from matchDetection in error format {status, payload: {message, id}} if returnFailures is true\n\n // we could have here also returnRecords/returnMatchRecords/returnNonMatchRecord options that could be turned false for not to return actual record data\n\n // matchStatus.status: boolean, true if matcher retrieved and handled all found candidate records, false if it did not\n // matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries','conversionFailures', empty string/undefined), reason for stopping retrieving or handling the candidate records\n // - only one stopReason is returned (if there would be several possible stopReasons, stopReason is picked in the above order)\n // - currently stopReason can be non-empty also in cases where status is true, if matcher hit the stop reason when handling the last available candidate record\n\n function returnResult({matches, state, stopReason, nonMatches, duplicateCount, candidateCount, nonMatchCount, conversionFailures, matchErrors}) {\n const conversionFailureCount = conversionFailures.length;\n const matchErrorCount = matchErrors.length;\n checkCounts({matches, nonMatches, candidateCount, duplicateCount, nonMatchCount, conversionFailureCount, matchErrorCount});\n const matchStatus = getMatchState(state, stopReason, conversionFailureCount, matchErrorCount);\n // add nonMatches to result only if returnNonMatches is 'true', otherwise nonMatches have not been gathered\n const matchesResult = returnNonMatches ? {matches, matchStatus, nonMatches} : {matches, matchStatus};\n const failures = [...conversionFailures, ...matchErrors];\n const result = returnFailures ? {...matchesResult, conversionFailures: failures} : matchesResult;\n debugData(`ReturnFailures ${returnFailures}`);\n debugData(`${JSON.stringify(result)}`);\n return result;\n\n // note that in cases where the matching has been stopped because of maxMatches checkCounts won't (in most cases) match\n\n function checkCounts({matches, nonMatches, candidateCount, duplicateCount, nonMatchCount, conversionFailureCount, matchErrorCount}) {\n const matchCount = matches.length;\n debugData(`Return nonMatches: ${returnNonMatches}`);\n const chosenNonMatchCount = returnNonMatches ? nonMatches.length : nonMatchCount;\n const totalHandled = matchCount + chosenNonMatchCount + duplicateCount;\n debug(`candidateCount: ${candidateCount}, matches: ${matchCount}, nonMatches: ${chosenNonMatchCount}, duplicateCount: ${duplicateCount}, conversionFailureCount: ${conversionFailureCount}, matchErrorCount: ${matchErrorCount}`);\n debug(`We got result for ${totalHandled} / ${candidateCount} retrieved candidates`);\n if (totalHandled !== candidateCount) {\n debug(`WARNING: Missing results for ${candidateCount - totalHandled} candidates`);\n return;\n }\n return;\n }\n\n // eslint-disable-next-line max-statements\n function getMatchState(state, stopReason, conversionFailuresCount, matchErrorCount) {\n debugData(`${JSON.stringify(state)}`);\n debug(`We had ${conversionFailuresCount} retrieved candidates that could not be converted.`);\n debug(`We had ${matchErrorCount} retrieved candidates that errored in matchDetection.`);\n debug(`Queries left ${state.queriesLeft}, Searches for current query left: ${state.resultSetOffset && state.resultSetOffset <= state.totalRecords}, non-retrieved records: ${state.totalRecords - state.queryCandidateCounter}, maxedQueries (${state.maxedQueries.length}): ${state.maxedQueries}`);\n\n debugData(`StopReason: <${stopReason}>`);\n\n const searchesLeft = state.resultSetOffset && state.resultSetOffset <= state.totalRecords;\n const nonRetrieved = searchesLeft ? state.totalRecords - state.queryCandidateCounter : 0;\n debugData(`nonRetrieved: ${nonRetrieved}`);\n\n // matchStatus.stopReason: string ('maxMatches','maxCandidates','maxedQueries','conversionFailures', empty string/undefined), reason for stopping retrieving or handling the candidate records\n // 'maxMatches' and 'maxCandidates' are in stopReason, 'maxedQueries', 'conversionFailures' and 'matchErrors' are created here\n\n if (state.queriesLeft > 0 || nonRetrieved > 0 || state.maxedQueries.length > 0 || conversionFailureCount > 0 || matchErrorCount > 0) {\n const maxedQueriesStopReason = state.maxedQueries.length > 0 ? 'maxedQueries' : undefined;\n const conversionFailuresStopReason = conversionFailureCount > 0 ? 'conversionFailures' : undefined;\n const matchErrorsStopReason = matchErrorCount > 0 ? 'matchErrors' : undefined;\n const newStopReason = stopReason === '' || stopReason === undefined ? maxedQueriesStopReason || conversionFailuresStopReason || matchErrorsStopReason : stopReason;\n debugData(`MaxedQueriesStopReason: <${maxedQueriesStopReason}>`);\n debugData(`ConversionFailureStopReason <${conversionFailuresStopReason}>`);\n debugData(`MatchErrorsStopReason <${matchErrorsStopReason}>`);\n debugData(`NewStopReason: <${newStopReason}>`);\n debug(`Match status: false`);\n return {status: false, stopReason: newStopReason};\n }\n\n debug(`Match status: true`);\n return {status: true, stopReason};\n }\n }\n\n // NOTES:\n // - we could optimize by creating the featureSet for the incoming record once and using it for all database/candidateRecords\n // - if creating the featureSet for the incoming record fails we have an unprocessable entity\n // - if creating the featureSet for a candidate record fails we could skip that candidate - but list the case as a detectionFailure, same as conversionFailures\n\n // eslint-disable-next-line max-statements\n function iterateRecords({records, recordSetSize, maxMatches, matches = [], nonMatches = [], recordMatches = [], recordNonMatches = [], recordCount = 0, recordDuplicateCount = 0, recordNonMatchCount = 0, recordMatchErrors = []}) {\n\n // recordSetSize : total amount of records in the current record set\n // recordCount : amount of records from the current record set that have been handled\n // maxMatches : setting for maximum amount found by current matcher job before the matcher job is stopped\n // recordDuplicateCount : amount of records from the current record set that are already included in matches/nonMatches results\n // recordNonMatchCount: amount of records from the current record set that are nonMatches (only is returnNonMatches setting is false)\n\n // records : non-handled records in the current record set\n // matches : found matches in the current matcher job\n // recordMatches : found matches in the current record set\n // recordNonMatches : found nonMatches in the current record set (only if returnNonMatches setting is true)\n // recordMatchErrors: errored matchDetection in the current record set\n\n const [candidate] = records;\n const newRecordCount = candidate ? recordCount + 1 : recordCount;\n\n // The matcher uses same matchDetection strategy for candidates from all candidate-searches -> matchDetection result for the same candidate is always same\n // Exceptions would happen if the candidate would have been updated in the database between candidate searches\n // Note that if returnNonMatches is false, matcher won't remember candidates that didn't match, so they will be matched again everytime they are retrieved by\n // different candidate search queries. Same candidate search query won't have duplicate records.\n\n /* We could optimize and detect all retrieved candidates at once\n const candidateRecords = records.map(record => record.record);\n const recordsIsArray = Array.isArray(candidateRecords);\n debug(`records is an array: ${recordsIsArray}`);\n const result = detect(record, candidateRecords);\n debug(`${JSON.stringify(result)}`);\n */\n\n if (candidate) {\n\n // eslint-disable-next-line functional/no-conditional-statement\n if (candidateNotInMatches(matches.concat(nonMatches), candidate)) {\n const {record: candidateRecord, id: candidateId} = candidate;\n try {\n debug(`Running matchDetection for record ${candidateId} (${newRecordCount}/${recordSetSize})`);\n // we should handle errors from detection somehow - ie. cases where either record or candidateRecord errors\n const detectionResult = detect(record, candidateRecord);\n\n return handleDetectionResult(detectionResult, candidateId, candidateRecord);\n } catch (error) {\n debug(`MatchDetection errored: database record ${candidateId}: ${error}`);\n\n const matchError = {status: 422, payload: {message: `Matching errored for database record ${candidateId}. ${error.message}.`, id: candidateId}};\n const newRecordMatchErrors = recordMatchErrors.concat(matchError);\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches, recordCount: newRecordCount, recordNonMatches, recordDuplicateCount, recordNonMatchCount, recordMatchErrors: newRecordMatchErrors});\n }\n }\n\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches, recordCount: newRecordCount, recordNonMatches, recordDuplicateCount: recordDuplicateCount + 1, recordNonMatchCount, recordMatchErrors});\n }\n\n debug(`No more candidates, record set (${recordCount}/${recordSetSize}) done, ${recordMatches.length} matches found, ${recordDuplicateCount} candidates already handled, ${returnNonMatches ? `${recordNonMatches.length}` : `${recordNonMatchCount}`} nonMatches found.`);\n return {matches: recordMatches, nonMatches: returnNonMatches ? recordNonMatches : [], duplicateCount: recordDuplicateCount, nonMatchCount: recordNonMatchCount, matchErrors: recordMatchErrors};\n\n function handleDetectionResult(detectionResult, candidateId, candidateRecord) {\n debugData(`MatchDetection results for ${candidateId} (${newRecordCount}/${recordSetSize}): ${JSON.stringify(detectionResult)}`);\n\n if (detectionResult.match || returnNonMatches) {\n debug(`${detectionResult.match ? `Record ${candidateId} (${newRecordCount}/${recordSetSize}) is a match!` : `Record ${candidateId} (${newRecordCount}/${recordSetSize}) is NOT a match!`}`);\n debugData(`Strategy: ${JSON.stringify(detectionResult.strategy)}, Treshold: ${JSON.stringify(detectionResult.treshold)}`);\n\n const matchResult = {\n probability: detectionResult.probability,\n candidate: {\n id: candidateId,\n record: candidateRecord\n }\n };\n const strategyResult = {\n strategy: detectionResult.strategy,\n treshold: detectionResult.treshold\n };\n const newMatch = returnStrategy ? {...matchResult, ...strategyResult} : {...matchResult};\n\n debugData(`${JSON.stringify(newMatch)}`);\n\n return handleRecordMatch(detectionResult.match, newMatch);\n }\n\n const newRecordNonMatchCount = recordNonMatchCount + 1;\n debugData(`- Total nonMatches after this detection: ${newRecordNonMatchCount}`);\n\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches, recordCount: newRecordCount, recordNonMatches, recordDuplicateCount, recordNonMatchCount: newRecordNonMatchCount, recordMatchErrors});\n }\n\n function handleRecordMatch(isMatch, newMatch) {\n const newRecordMatches = isMatch ? recordMatches.concat(newMatch) : recordMatches;\n const newRecordNonMatches = isMatch ? recordNonMatches : recordNonMatches.concat(newMatch);\n const newRecordNonMatchCount = isMatch ? recordNonMatchCount : recordNonMatchCount + 1;\n\n debugData(`- Total matches after this detection: ${matches.concat(newRecordMatches).length} (max: ${maxMatches})`);\n\n // eslint-disable-next-line functional/no-conditional-statement\n if (returnNonMatches) {\n debugData(`- Total nonMatches after this detection: ${nonMatches.concat(newRecordNonMatches).length}`);\n }\n debugData(`- Total nonMatchCount after this detection: ${recordNonMatchCount}`);\n\n if (maxMatchesFound({matches: matches.concat(newRecordMatches), maxMatches})) {\n debug(`MaxMatches (${maxMatches}) reached, handled candidates in record set: ${newRecordCount} non-handled candidates in record set ${recordSetSize - newRecordCount}`);\n return {matches: newRecordMatches, nonMatches: returnNonMatches ? newRecordNonMatches : [], duplicateCount: recordDuplicateCount, nonMatchCount: newRecordNonMatchCount, matchErrors: recordMatchErrors};\n }\n\n return iterateRecords({records: records.slice(1), recordSetSize, maxMatches, matches, recordMatches: newRecordMatches, recordCount: newRecordCount, recordNonMatches: returnNonMatches ? newRecordNonMatches : [], duplicateCount: recordDuplicateCount, recordNonMatchCount: newRecordNonMatchCount, matchErrors: recordMatchErrors});\n }\n\n function candidateNotInMatches(matches, candidate) {\n debug(`Checking that record ${candidate.id} is not already included in ${matches.length} matches/nonMatches`);\n const newCandidateId = candidate.id;\n debugData(`newCandidateId: ${newCandidateId}`);\n const result = matches.find(({candidate}) => candidate.id === newCandidateId);\n debugData(`Result: ${result}`);\n if (result) {\n debug(`${candidate.id} was already handled.`);\n return false;\n }\n debug(`${candidate.id} not found in matches/nonMatches`);\n return true;\n }\n }\n\n function maxMatchesFound({matches, maxMatches}) {\n if (maxMatches && matches.length >= maxMatches) {\n debug(`Stopping recordSet iteration: maxMatches (${maxMatches}) for matcher job found.`);\n return true;\n }\n }\n };\n};\n"],"file":"index.js"}
|
package/dist/index.spec.js
CHANGED
|
@@ -62,7 +62,8 @@ describe('INDEX', () => {
|
|
|
62
62
|
options,
|
|
63
63
|
enabled = true,
|
|
64
64
|
expectedMatchStatus,
|
|
65
|
-
expectedStopReason
|
|
65
|
+
expectedStopReason,
|
|
66
|
+
expectedFailures
|
|
66
67
|
}) {
|
|
67
68
|
if (!enabled) {
|
|
68
69
|
debug(`Disabled test!`);
|
|
@@ -78,15 +79,20 @@ describe('INDEX', () => {
|
|
|
78
79
|
const {
|
|
79
80
|
matches,
|
|
80
81
|
matchStatus,
|
|
81
|
-
nonMatches
|
|
82
|
+
nonMatches,
|
|
83
|
+
conversionFailures
|
|
82
84
|
} = await match(record);
|
|
83
|
-
debugData(
|
|
85
|
+
debugData(`Matches: ${matches.length}, Status: ${matchStatus.status}/${matchStatus.stopReason}, NonMatches: ${nonMatches ? nonMatches.length : 'not returned'}, ConversionFailures: ${conversionFailures ? conversionFailures.length : 'not returned'}`);
|
|
86
|
+
(0, _chai.expect)(matchStatus.status).to.eql(expectedMatchStatus);
|
|
87
|
+
(0, _chai.expect)(matchStatus.stopReason).to.eql(expectedStopReason);
|
|
84
88
|
const formattedMatchResult = formatRecordResults(matches);
|
|
85
89
|
(0, _chai.expect)(formattedMatchResult).to.eql(expectedMatches);
|
|
86
90
|
const formattedNonMatchResult = formatRecordResults(nonMatches);
|
|
87
|
-
(0, _chai.expect)(formattedNonMatchResult).to.eql(expectedNonMatches);
|
|
88
|
-
|
|
89
|
-
|
|
91
|
+
(0, _chai.expect)(formattedNonMatchResult).to.eql(expectedNonMatches); // eslint-disable-next-line functional/no-conditional-statement
|
|
92
|
+
|
|
93
|
+
if (expectedFailures) {
|
|
94
|
+
(0, _chai.expect)(conversionFailures).to.eql(expectedFailures);
|
|
95
|
+
}
|
|
90
96
|
|
|
91
97
|
function formatOptions() {
|
|
92
98
|
const contextFeatures = _.matchDetection.features[options.detection.strategy.type];
|