@natlibfi/melinda-record-matching 2.2.2 → 3.0.0-alpha.1
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/index.js +20 -3
- package/dist/index.js.map +1 -1
- package/dist/index.spec.js +3 -1
- package/dist/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/all-source-ids.js +6 -2
- package/dist/match-detection/features/bib/all-source-ids.js.map +1 -1
- package/dist/match-detection/features/bib/authors.js +28 -22
- package/dist/match-detection/features/bib/authors.js.map +1 -1
- package/dist/match-detection/features/bib/bibliographic-level.js +3 -1
- package/dist/match-detection/features/bib/bibliographic-level.js.map +1 -1
- package/dist/match-detection/features/bib/host-component.js +3 -1
- package/dist/match-detection/features/bib/host-component.js.map +1 -1
- package/dist/match-detection/features/bib/index.spec.js +3 -1
- package/dist/match-detection/features/bib/index.spec.js.map +1 -1
- package/dist/match-detection/features/bib/language.js +6 -2
- package/dist/match-detection/features/bib/language.js.map +1 -1
- package/dist/match-detection/features/bib/melinda-identifier-factory.js +5 -2
- package/dist/match-detection/features/bib/melinda-identifier-factory.js.map +1 -1
- package/dist/match-detection/features/bib/publication-time.js +3 -1
- package/dist/match-detection/features/bib/publication-time.js.map +1 -1
- package/dist/match-detection/features/bib/record-type.js +3 -1
- package/dist/match-detection/features/bib/record-type.js.map +1 -1
- package/dist/match-detection/features/bib/standard-identifier-factory.js +3 -1
- package/dist/match-detection/features/bib/standard-identifier-factory.js.map +1 -1
- package/dist/match-detection/features/bib/title.js +3 -1
- package/dist/match-detection/features/bib/title.js.map +1 -1
- package/dist/match-detection/index.js +47 -12
- package/dist/match-detection/index.js.map +1 -1
- package/dist/match-detection/index.spec.js +4 -1
- package/dist/match-detection/index.spec.js.map +1 -1
- package/package.json +1 -1
- package/src/index.js +5 -3
- package/src/index.spec.js +1 -1
- package/src/match-detection/features/bib/all-source-ids.js +3 -2
- package/src/match-detection/features/bib/authors.js +19 -15
- package/src/match-detection/features/bib/bibliographic-level.js +1 -1
- package/src/match-detection/features/bib/host-component.js +1 -1
- package/src/match-detection/features/bib/index.spec.js +1 -1
- package/src/match-detection/features/bib/language.js +4 -2
- package/src/match-detection/features/bib/melinda-identifier-factory.js +2 -2
- package/src/match-detection/features/bib/publication-time.js +1 -1
- package/src/match-detection/features/bib/record-type.js +1 -1
- package/src/match-detection/features/bib/standard-identifier-factory.js +1 -1
- package/src/match-detection/features/bib/title.js +1 -1
- package/src/match-detection/index.js +20 -12
- package/src/match-detection/index.spec.js +1 -1
|
@@ -47,12 +47,15 @@ var _default = () => {
|
|
|
47
47
|
extract,
|
|
48
48
|
compare
|
|
49
49
|
};
|
|
50
|
-
function extract(
|
|
50
|
+
function extract({
|
|
51
|
+
record,
|
|
52
|
+
recordExternal
|
|
53
|
+
}) {
|
|
51
54
|
const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');
|
|
52
55
|
const [f001] = record.get('001').map(field => field.value);
|
|
53
56
|
const f035MelindaIds = (0, _matchingUtils.getMelindaIdsF035)(record);
|
|
54
57
|
if (isMelindaRecord === undefined && f001 === undefined && f035MelindaIds.length < 1) {
|
|
55
|
-
debug(
|
|
58
|
+
debug(`${recordExternal.label} No Melinda-IDs found`);
|
|
56
59
|
return {};
|
|
57
60
|
}
|
|
58
61
|
return {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"melinda-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","extract","compare","record","isMelindaRecord","get","some","f003","value","f001","map","field","f035MelindaIds","getMelindaIdsF035","undefined","length","a","b","id","JSON","stringify","idA","idB"],"sources":["../../../../src/match-detection/features/bib/melinda-identifier-factory.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {getMelindaIdsF035} from '../../../matching-utils';\n\n// 003+001 FI-MELINDA <melinda-id>\n// 035 $a (FI-MELINDA)<melinda-id>\n// 035 $z (FI-MELINDA)<melinda-id>\n// 035 $a FCC<melinda-id>\n// 035 $z FCC<melinda-id>\n// melinda-id = 001234567\n\nexport default () => {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/melinda-id');\n const debugData = debug.extend('data');\n\n return {extract, compare};\n\n function extract(record) {\n\n const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');\n const [f001] = record.get('001').map(field => field.value);\n const f035MelindaIds = getMelindaIdsF035(record);\n\n if (\n isMelindaRecord === undefined &&\n f001 === undefined &&\n f035MelindaIds.length < 1) {\n\n debug(
|
|
1
|
+
{"version":3,"file":"melinda-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","extract","compare","record","recordExternal","isMelindaRecord","get","some","f003","value","f001","map","field","f035MelindaIds","getMelindaIdsF035","undefined","length","label","a","b","id","JSON","stringify","idA","idB"],"sources":["../../../../src/match-detection/features/bib/melinda-identifier-factory.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {getMelindaIdsF035} from '../../../matching-utils';\n\n// 003+001 FI-MELINDA <melinda-id>\n// 035 $a (FI-MELINDA)<melinda-id>\n// 035 $z (FI-MELINDA)<melinda-id>\n// 035 $a FCC<melinda-id>\n// 035 $z FCC<melinda-id>\n// melinda-id = 001234567\n\nexport default () => {\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/melinda-id');\n const debugData = debug.extend('data');\n\n return {extract, compare};\n\n function extract({record, recordExternal}) {\n\n const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');\n const [f001] = record.get('001').map(field => field.value);\n const f035MelindaIds = getMelindaIdsF035(record);\n\n if (\n isMelindaRecord === undefined &&\n f001 === undefined &&\n f035MelindaIds.length < 1) {\n\n debug(`${recordExternal.label} No Melinda-IDs found`);\n return {};\n }\n\n return {isMelindaRecord, f001, f035MelindaIds};\n\n }\n\n // eslint-disable-next-line max-statements\n function compare(a, b) {\n\n if (a.isMelindaRecord && b.isMelindaRecord &&\n a.f001 === b.f001) {\n debugData(`Melinda record's A f001 ${a.f001} matches Melinda record's B f001 ${a.f001}`);\n return 1;\n }\n\n if (a.isMelindaRecord && typeof b.f035MelindaIds !== 'undefined' &&\n b.f035MelindaIds.some(id => id === a.f001)) {\n debugData(`Melinda record's A f001 ${a.f001} matches record B f035 ${JSON.stringify(b.f035MelindaIds)}`);\n return 1;\n }\n\n if (b.isMelindaRecord && typeof a.f035MelindaIds !== 'undefined' &&\n a.f035MelindaIds.some(id => id === b.f001)) {\n debugData(`Melinda record's B f001 ${b.f001} matches record A f035 ${JSON.stringify(a.f035MelindaIds)}`);\n return 1;\n }\n\n if (typeof a.f035MelindaIds !== 'undefined' && typeof b.f035MelindaIds !== 'undefined' &&\n a.f035MelindaIds.some(idA => b.f035MelindaIds.some(idB => idB === idA))) {\n debugData(`Record A f035 ${JSON.stringify(a.f035MelindaIds)} matches record B f035 ${JSON.stringify(b.f035MelindaIds)}`);\n return 1;\n }\n debug(`No matching Melinda-IDs.`);\n return 0;\n }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA0D;AA7B1D;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;AACA;AACA;AACA;AACA;AACA;AAAA,eAEe,MAAM;EAEnB,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,2EAA2E,CAAC;EAC5G,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtC,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAAC;IAACE,MAAM;IAAEC;EAAc,CAAC,EAAE;IAEzC,MAAMC,eAAe,GAAGF,MAAM,CAACG,GAAG,CAAC,KAAK,CAAC,CAACC,IAAI,CAACC,IAAI,IAAIA,IAAI,CAACC,KAAK,KAAK,YAAY,CAAC;IACnF,MAAM,CAACC,IAAI,CAAC,GAAGP,MAAM,CAACG,GAAG,CAAC,KAAK,CAAC,CAACK,GAAG,CAACC,KAAK,IAAIA,KAAK,CAACH,KAAK,CAAC;IAC1D,MAAMI,cAAc,GAAG,IAAAC,gCAAiB,EAACX,MAAM,CAAC;IAEhD,IACEE,eAAe,KAAKU,SAAS,IAC7BL,IAAI,KAAKK,SAAS,IAClBF,cAAc,CAACG,MAAM,GAAG,CAAC,EAAE;MAE3BnB,KAAK,CAAE,GAAEO,cAAc,CAACa,KAAM,uBAAsB,CAAC;MACrD,OAAO,CAAC,CAAC;IACX;IAEA,OAAO;MAACZ,eAAe;MAAEK,IAAI;MAAEG;IAAc,CAAC;EAEhD;;EAEA;EACA,SAASX,OAAO,CAACgB,CAAC,EAAEC,CAAC,EAAE;IAErB,IAAID,CAAC,CAACb,eAAe,IAAIc,CAAC,CAACd,eAAe,IACtCa,CAAC,CAACR,IAAI,KAAKS,CAAC,CAACT,IAAI,EAAE;MACrBX,SAAS,CAAE,2BAA0BmB,CAAC,CAACR,IAAK,oCAAmCQ,CAAC,CAACR,IAAK,EAAC,CAAC;MACxF,OAAO,CAAC;IACV;IAEA,IAAIQ,CAAC,CAACb,eAAe,IAAI,OAAOc,CAAC,CAACN,cAAc,KAAK,WAAW,IAC5DM,CAAC,CAACN,cAAc,CAACN,IAAI,CAACa,EAAE,IAAIA,EAAE,KAAKF,CAAC,CAACR,IAAI,CAAC,EAAE;MAC9CX,SAAS,CAAE,2BAA0BmB,CAAC,CAACR,IAAK,0BAAyBW,IAAI,CAACC,SAAS,CAACH,CAAC,CAACN,cAAc,CAAE,EAAC,CAAC;MACxG,OAAO,CAAC;IACV;IAEA,IAAIM,CAAC,CAACd,eAAe,IAAI,OAAOa,CAAC,CAACL,cAAc,KAAK,WAAW,IAC5DK,CAAC,CAACL,cAAc,CAACN,IAAI,CAACa,EAAE,IAAIA,EAAE,KAAKD,CAAC,CAACT,IAAI,CAAC,EAAE;MAC9CX,SAAS,CAAE,2BAA0BoB,CAAC,CAACT,IAAK,0BAAyBW,IAAI,CAACC,SAAS,CAACJ,CAAC,CAACL,cAAc,CAAE,EAAC,CAAC;MACxG,OAAO,CAAC;IACV;IAEA,IAAI,OAAOK,CAAC,CAACL,cAAc,KAAK,WAAW,IAAI,OAAOM,CAAC,CAACN,cAAc,KAAK,WAAW,IACjFK,CAAC,CAACL,cAAc,CAACN,IAAI,CAACgB,GAAG,IAAIJ,CAAC,CAACN,cAAc,CAACN,IAAI,CAACiB,GAAG,IAAIA,GAAG,KAAKD,GAAG,CAAC,CAAC,EAAE;MAC5ExB,SAAS,CAAE,iBAAgBsB,IAAI,CAACC,SAAS,CAACJ,CAAC,CAACL,cAAc,CAAE,0BAAyBQ,IAAI,CAACC,SAAS,CAACH,CAAC,CAACN,cAAc,CAAE,EAAC,CAAC;MACxH,OAAO,CAAC;IACV;IACAhB,KAAK,CAAE,0BAAyB,CAAC;IACjC,OAAO,CAAC;EACV;AACF,CAAC;AAAA"}
|
|
@@ -36,7 +36,9 @@ var _matchingUtils = require("../../../matching-utils");
|
|
|
36
36
|
// We could also enable matching of records which have consequent publication years
|
|
37
37
|
var _default = () => ({
|
|
38
38
|
name: 'Publication time',
|
|
39
|
-
extract:
|
|
39
|
+
extract: ({
|
|
40
|
+
record
|
|
41
|
+
}) => {
|
|
40
42
|
var _record$get, _record$get$;
|
|
41
43
|
const value = ((_record$get = record.get(/^008$/u)) === null || _record$get === void 0 ? void 0 : (_record$get$ = _record$get[0]) === null || _record$get$ === void 0 ? void 0 : _record$get$.value) || undefined;
|
|
42
44
|
return (0, _matchingUtils.testStringOrNumber)(value) ? [String(value).slice(7, 11)] : [];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport {testStringOrNumber} from '../../../matching-utils';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We could also enable matching of records which have consequent publication years\n\nexport default () => ({\n name: 'Publication time',\n extract: record => {\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n },\n compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;AA4BA;AA5BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA;AACA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,kBAAkB;EACxBC,OAAO,
|
|
1
|
+
{"version":3,"file":"publication-time.js","names":["name","extract","record","value","get","undefined","testStringOrNumber","String","slice","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/publication-time.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport {testStringOrNumber} from '../../../matching-utils';\n\n// We should also get copyright time and copyright/publication times from 26x\n// We could also enable matching of records which have consequent publication years\n\nexport default () => ({\n name: 'Publication time',\n extract: ({record}) => {\n const value = record.get(/^008$/u)?.[0]?.value || undefined;\n return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n },\n compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],"mappings":";;;;;;AA4BA;AA5BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAIA;AACA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,kBAAkB;EACxBC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IAAA;IACrB,MAAMC,KAAK,GAAG,gBAAAD,MAAM,CAACE,GAAG,CAAC,QAAQ,CAAC,gEAApB,YAAuB,CAAC,CAAC,iDAAzB,aAA2BD,KAAK,KAAIE,SAAS;IAC3D,OAAO,IAAAC,iCAAkB,EAACH,KAAK,CAAC,GAAG,CAACI,MAAM,CAACJ,KAAK,CAAC,CAACK,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE;EACtE,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAA"}
|
|
@@ -34,7 +34,9 @@ exports.default = void 0;
|
|
|
34
34
|
// we could handle the case of books/notes
|
|
35
35
|
var _default = () => ({
|
|
36
36
|
name: 'Record type',
|
|
37
|
-
extract:
|
|
37
|
+
extract: ({
|
|
38
|
+
record
|
|
39
|
+
}) => record.leader[6] ? [record.leader[6]] : [],
|
|
38
40
|
compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5
|
|
39
41
|
});
|
|
40
42
|
exports.default = _default;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"record-type.js","names":["name","extract","
|
|
1
|
+
{"version":3,"file":"record-type.js","names":["name","extract","record","leader","compare","a","b"],"sources":["../../../../src/match-detection/features/bib/record-type.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\n// we could handle the case of books/notes\n\nexport default () => ({\n name: 'Record type',\n extract: ({record}) => record.leader[6] ? [record.leader[6]] : [],\n compare: (a, b) => a[0] === b[0] ? 0.1 : -0.5\n});\n"],"mappings":";;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAEA;AAAA,eAEe,OAAO;EACpBA,IAAI,EAAE,aAAa;EACnBC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAKA,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,GAAG,CAACD,MAAM,CAACC,MAAM,CAAC,CAAC,CAAC,CAAC,GAAG,EAAE;EACjEC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAKD,CAAC,CAAC,CAAC,CAAC,KAAKC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,GAAG,CAAC;AAC5C,CAAC,CAAC;AAAA"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"standard-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","pattern","subfieldCodes","extract","compare","record","field","get","subfields","filter","code","includes","map","value","testStringOrNumber","String","replace","a","b","length","bothHaveValidIdentifiers","maxValues","matchingValues","getValueCount","aValues","bValues","validOnly","getIdentifiers","aValue","some","bValue","values"],"sources":["../../../../src/match-detection/features/bib/standard-identifier-factory.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');\nconst debugData = debug.extend('data');\n\nexport default ({pattern, subfieldCodes}) => {\n return {extract, compare};\n\n function extract(record) {\n const [field] = record.get(pattern);\n\n if (field) {\n return field.subfields\n .filter(({code}) => subfieldCodes.includes(code))\n .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))\n .filter(value => value);\n }\n\n return [];\n }\n\n function compare(a, b) {\n if (a.length === 0 || b.length === 0) {\n debugData(`No standardidentifiers to compare`);\n return 0;\n }\n\n if (bothHaveValidIdentifiers()) {\n const {maxValues, matchingValues} = getValueCount(true);\n return matchingValues / maxValues * 0.75;\n }\n\n const {maxValues, matchingValues} = getValueCount();\n return matchingValues / maxValues * 0.2;\n\n function bothHaveValidIdentifiers() {\n const aValues = a.filter(({code}) => code === 'a');\n const bValues = a.filter(({code}) => code === 'a');\n return aValues.length > 0 && bValues.length > 0;\n }\n\n function getValueCount(validOnly = false) {\n const aValues = getIdentifiers(a);\n const bValues = getIdentifiers(b);\n\n return {\n maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length\n };\n\n function getIdentifiers(values) {\n if (validOnly) {\n return values\n .filter(({code}) => code === 'a')\n .map(({value}) => value);\n }\n\n return values.map(({value}) => value);\n }\n }\n }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA2D;AA7B3D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAKA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;AAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,CAAC;EAACC,OAAO;EAAEC;AAAa,CAAC,KAAK;EAC3C,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,
|
|
1
|
+
{"version":3,"file":"standard-identifier-factory.js","names":["debug","createDebugLogger","debugData","extend","pattern","subfieldCodes","extract","compare","record","field","get","subfields","filter","code","includes","map","value","testStringOrNumber","String","replace","a","b","length","bothHaveValidIdentifiers","maxValues","matchingValues","getValueCount","aValues","bValues","validOnly","getIdentifiers","aValue","some","bValue","values"],"sources":["../../../../src/match-detection/features/bib/standard-identifier-factory.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features:standard-identifiers');\nconst debugData = debug.extend('data');\n\nexport default ({pattern, subfieldCodes}) => {\n return {extract, compare};\n\n function extract({record}) {\n const [field] = record.get(pattern);\n\n if (field) {\n return field.subfields\n .filter(({code}) => subfieldCodes.includes(code))\n .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value).replace(/-/ug, '') : ''}))\n .filter(value => value);\n }\n\n return [];\n }\n\n function compare(a, b) {\n if (a.length === 0 || b.length === 0) {\n debugData(`No standardidentifiers to compare`);\n return 0;\n }\n\n if (bothHaveValidIdentifiers()) {\n const {maxValues, matchingValues} = getValueCount(true);\n return matchingValues / maxValues * 0.75;\n }\n\n const {maxValues, matchingValues} = getValueCount();\n return matchingValues / maxValues * 0.2;\n\n function bothHaveValidIdentifiers() {\n const aValues = a.filter(({code}) => code === 'a');\n const bValues = a.filter(({code}) => code === 'a');\n return aValues.length > 0 && bValues.length > 0;\n }\n\n function getValueCount(validOnly = false) {\n const aValues = getIdentifiers(a);\n const bValues = getIdentifiers(b);\n\n return {\n maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n matchingValues: aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length\n };\n\n function getIdentifiers(values) {\n if (validOnly) {\n return values\n .filter(({code}) => code === 'a')\n .map(({value}) => value);\n }\n\n return values.map(({value}) => value);\n }\n }\n }\n};\n"],"mappings":";;;;;;AA4BA;AACA;AAA2D;AA7B3D;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;;AAKA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,iFAAiF,CAAC;AAClH,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAAC,eAExB,CAAC;EAACC,OAAO;EAAEC;AAAa,CAAC,KAAK;EAC3C,OAAO;IAACC,OAAO;IAAEC;EAAO,CAAC;EAEzB,SAASD,OAAO,CAAC;IAACE;EAAM,CAAC,EAAE;IACzB,MAAM,CAACC,KAAK,CAAC,GAAGD,MAAM,CAACE,GAAG,CAACN,OAAO,CAAC;IAEnC,IAAIK,KAAK,EAAE;MACT,OAAOA,KAAK,CAACE,SAAS,CACnBC,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKR,aAAa,CAACS,QAAQ,CAACD,IAAI,CAAC,CAAC,CAChDE,GAAG,CAAC,CAAC;QAACF,IAAI;QAAEG;MAAK,CAAC,MAAM;QAACH,IAAI;QAAEG,KAAK,EAAE,IAAAC,iCAAkB,EAACD,KAAK,CAAC,GAAGE,MAAM,CAACF,KAAK,CAAC,CAACG,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG;MAAE,CAAC,CAAC,CAAC,CAC1GP,MAAM,CAACI,KAAK,IAAIA,KAAK,CAAC;IAC3B;IAEA,OAAO,EAAE;EACX;EAEA,SAAST,OAAO,CAACa,CAAC,EAAEC,CAAC,EAAE;IACrB,IAAID,CAAC,CAACE,MAAM,KAAK,CAAC,IAAID,CAAC,CAACC,MAAM,KAAK,CAAC,EAAE;MACpCpB,SAAS,CAAE,mCAAkC,CAAC;MAC9C,OAAO,CAAC;IACV;IAEA,IAAIqB,wBAAwB,EAAE,EAAE;MAC9B,MAAM;QAACC,SAAS;QAAEC;MAAc,CAAC,GAAGC,aAAa,CAAC,IAAI,CAAC;MACvD,OAAOD,cAAc,GAAGD,SAAS,GAAG,IAAI;IAC1C;IAEA,MAAM;MAACA,SAAS;MAAEC;IAAc,CAAC,GAAGC,aAAa,EAAE;IACnD,OAAOD,cAAc,GAAGD,SAAS,GAAG,GAAG;IAEvC,SAASD,wBAAwB,GAAG;MAClC,MAAMI,OAAO,GAAGP,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,MAAMe,OAAO,GAAGR,CAAC,CAACR,MAAM,CAAC,CAAC;QAACC;MAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC;MAClD,OAAOc,OAAO,CAACL,MAAM,GAAG,CAAC,IAAIM,OAAO,CAACN,MAAM,GAAG,CAAC;IACjD;IAEA,SAASI,aAAa,CAACG,SAAS,GAAG,KAAK,EAAE;MACxC,MAAMF,OAAO,GAAGG,cAAc,CAACV,CAAC,CAAC;MACjC,MAAMQ,OAAO,GAAGE,cAAc,CAACT,CAAC,CAAC;MAEjC,OAAO;QACLG,SAAS,EAAEG,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM,GAAGK,OAAO,CAACL,MAAM,GAAGM,OAAO,CAACN,MAAM;QAC5EG,cAAc,EAAEE,OAAO,CAACf,MAAM,CAACmB,MAAM,IAAIH,OAAO,CAACI,IAAI,CAACC,MAAM,IAAIF,MAAM,KAAKE,MAAM,CAAC,CAAC,CAACX;MACtF,CAAC;MAED,SAASQ,cAAc,CAACI,MAAM,EAAE;QAC9B,IAAIL,SAAS,EAAE;UACb,OAAOK,MAAM,CACVtB,MAAM,CAAC,CAAC;YAACC;UAAI,CAAC,KAAKA,IAAI,KAAK,GAAG,CAAC,CAChCE,GAAG,CAAC,CAAC;YAACC;UAAK,CAAC,KAAKA,KAAK,CAAC;QAC5B;QAEA,OAAOkB,MAAM,CAACnB,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAKA,KAAK,CAAC;MACvC;IACF;EACF;AACF,CAAC;AAAA"}
|
|
@@ -38,7 +38,9 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
38
38
|
treshold = 10
|
|
39
39
|
} = {}) => ({
|
|
40
40
|
name: 'Title',
|
|
41
|
-
extract:
|
|
41
|
+
extract: ({
|
|
42
|
+
record
|
|
43
|
+
}) => {
|
|
42
44
|
const title = getTitle();
|
|
43
45
|
if ((0, _matchingUtils.testStringOrNumber)(title)) {
|
|
44
46
|
return [String(title).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase()];
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"title.js","names":["treshold","name","extract","record","title","getTitle","testStringOrNumber","String","replace","toLowerCase","field","get","subfields","filter","code","includes","map","value","join","compare","a","b","debug","createDebugLogger","distance","leven","maxLength","getMaxLength","percentage","length"],"sources":["../../../../src/match-detection/features/bib/title.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nexport default ({treshold = 10} = {}) => ({\n name: 'Title',\n extract: record => {\n const title = getTitle();\n\n if (testStringOrNumber(title)) {\n return [String(title).replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase()];\n }\n\n return [];\n\n function getTitle() {\n const [field] = record.get(/^245$/u);\n\n if (field) {\n return field.subfields\n .filter(({code}) => ['a', 'b'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .join('');\n }\n return false;\n }\n },\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n const distance = leven(a[0], b[0]);\n\n if (distance === 0) {\n return 0.5;\n }\n\n const maxLength = getMaxLength();\n const percentage = distance / maxLength * 100;\n\n debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);\n\n if (percentage <= treshold) {\n return 0.3;\n }\n\n return -0.5;\n\n function getMaxLength() {\n return a[0].length > b[0].length ? a[0].length : b[0].length;\n }\n\n }\n});\n"],"mappings":";;;;;;AA4BA;AACA;AACA;AAA2D;AA9B3D;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,EA1BA,eAgCe,CAAC;EAACA,QAAQ,GAAG;AAAE,CAAC,GAAG,CAAC,CAAC,MAAM;EACxCC,IAAI,EAAE,OAAO;EACbC,OAAO,
|
|
1
|
+
{"version":3,"file":"title.js","names":["treshold","name","extract","record","title","getTitle","testStringOrNumber","String","replace","toLowerCase","field","get","subfields","filter","code","includes","map","value","join","compare","a","b","debug","createDebugLogger","distance","leven","maxLength","getMaxLength","percentage","length"],"sources":["../../../../src/match-detection/features/bib/title.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020-2022 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport {LevenshteinDistance as leven} from 'natural';\nimport {testStringOrNumber} from '../../../matching-utils';\n\nexport default ({treshold = 10} = {}) => ({\n name: 'Title',\n extract: ({record}) => {\n const title = getTitle();\n\n if (testStringOrNumber(title)) {\n return [String(title).replace(/[^\\p{Letter}\\p{Number}]/gu, '').toLowerCase()];\n }\n\n return [];\n\n function getTitle() {\n const [field] = record.get(/^245$/u);\n\n if (field) {\n return field.subfields\n .filter(({code}) => ['a', 'b'].includes(code))\n .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n .join('');\n }\n return false;\n }\n },\n compare: (a, b) => {\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:features/bib/title');\n const distance = leven(a[0], b[0]);\n\n if (distance === 0) {\n return 0.5;\n }\n\n const maxLength = getMaxLength();\n const percentage = distance / maxLength * 100;\n\n debug(`'${a}' vs '${b}': Max length = ${maxLength}, distance = ${distance}, percentage = ${percentage}`);\n\n if (percentage <= treshold) {\n return 0.3;\n }\n\n return -0.5;\n\n function getMaxLength() {\n return a[0].length > b[0].length ? a[0].length : b[0].length;\n }\n\n }\n});\n"],"mappings":";;;;;;AA4BA;AACA;AACA;AAA2D;AA9B3D;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,EA1BA,eAgCe,CAAC;EAACA,QAAQ,GAAG;AAAE,CAAC,GAAG,CAAC,CAAC,MAAM;EACxCC,IAAI,EAAE,OAAO;EACbC,OAAO,EAAE,CAAC;IAACC;EAAM,CAAC,KAAK;IACrB,MAAMC,KAAK,GAAGC,QAAQ,EAAE;IAExB,IAAI,IAAAC,iCAAkB,EAACF,KAAK,CAAC,EAAE;MAC7B,OAAO,CAACG,MAAM,CAACH,KAAK,CAAC,CAACI,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CAACC,WAAW,EAAE,CAAC;IAC/E;IAEA,OAAO,EAAE;IAET,SAASJ,QAAQ,GAAG;MAClB,MAAM,CAACK,KAAK,CAAC,GAAGP,MAAM,CAACQ,GAAG,CAAC,QAAQ,CAAC;MAEpC,IAAID,KAAK,EAAE;QACT,OAAOA,KAAK,CAACE,SAAS,CACnBC,MAAM,CAAC,CAAC;UAACC;QAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAACC,QAAQ,CAACD,IAAI,CAAC,CAAC,CAC7CE,GAAG,CAAC,CAAC;UAACC;QAAK,CAAC,KAAK,IAAAX,iCAAkB,EAACW,KAAK,CAAC,GAAGV,MAAM,CAACU,KAAK,CAAC,GAAG,EAAE,CAAC,CAChEC,IAAI,CAAC,EAAE,CAAC;MACb;MACA,OAAO,KAAK;IACd;EACF,CAAC;EACDC,OAAO,EAAE,CAACC,CAAC,EAAEC,CAAC,KAAK;IACjB,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,sEAAsE,CAAC;IACvG,MAAMC,QAAQ,GAAG,IAAAC,4BAAK,EAACL,CAAC,CAAC,CAAC,CAAC,EAAEC,CAAC,CAAC,CAAC,CAAC,CAAC;IAElC,IAAIG,QAAQ,KAAK,CAAC,EAAE;MAClB,OAAO,GAAG;IACZ;IAEA,MAAME,SAAS,GAAGC,YAAY,EAAE;IAChC,MAAMC,UAAU,GAAGJ,QAAQ,GAAGE,SAAS,GAAG,GAAG;IAE7CJ,KAAK,CAAE,IAAGF,CAAE,SAAQC,CAAE,mBAAkBK,SAAU,gBAAeF,QAAS,kBAAiBI,UAAW,EAAC,CAAC;IAExG,IAAIA,UAAU,IAAI5B,QAAQ,EAAE;MAC1B,OAAO,GAAG;IACZ;IAEA,OAAO,CAAC,GAAG;IAEX,SAAS2B,YAAY,GAAG;MACtB,OAAOP,CAAC,CAAC,CAAC,CAAC,CAACS,MAAM,GAAGR,CAAC,CAAC,CAAC,CAAC,CAACQ,MAAM,GAAGT,CAAC,CAAC,CAAC,CAAC,CAACS,MAAM,GAAGR,CAAC,CAAC,CAAC,CAAC,CAACQ,MAAM;IAC9D;EAEF;AACF,CAAC,CAAC;AAAA"}
|
|
@@ -41,22 +41,51 @@ function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { de
|
|
|
41
41
|
var _default = ({
|
|
42
42
|
strategy,
|
|
43
43
|
treshold = 0.9
|
|
44
|
-
}, returnStrategy = false) => (
|
|
44
|
+
}, returnStrategy = false) => ({
|
|
45
|
+
recordA,
|
|
46
|
+
recordB,
|
|
47
|
+
recordAExternal = {},
|
|
48
|
+
recordBExternal = {}
|
|
49
|
+
}) => {
|
|
45
50
|
const minProbabilityQuantifier = 0.5;
|
|
46
51
|
const debug = (0, _debug.default)('@natlibfi/melinda-record-matching:match-detection');
|
|
47
52
|
const debugData = debug.extend('data');
|
|
48
|
-
debugData(`Strategy: ${JSON.stringify(strategy)}`);
|
|
49
|
-
debugData(`
|
|
50
|
-
|
|
51
|
-
const featuresA = extractFeatures(
|
|
53
|
+
debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);
|
|
54
|
+
debugData(`Records: A: ${recordA}\nB: ${recordB}`);
|
|
55
|
+
debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);
|
|
56
|
+
const featuresA = extractFeatures({
|
|
57
|
+
record: recordA,
|
|
58
|
+
recordExternal: recordAExternal
|
|
59
|
+
});
|
|
52
60
|
debug(`We got an array of records: ${Array.isArray(recordB)}`);
|
|
53
61
|
const recordsB = Array.isArray(recordB) ? recordB : [recordB];
|
|
54
|
-
const
|
|
62
|
+
const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];
|
|
63
|
+
const detectionResults = recordsB.map((record, index) => actualDetection({
|
|
64
|
+
featuresA,
|
|
65
|
+
recordAExternal,
|
|
66
|
+
record,
|
|
67
|
+
recordExternal: recordsBExternal[index],
|
|
68
|
+
index
|
|
69
|
+
}));
|
|
70
|
+
|
|
71
|
+
// if we got array of records, we return an array of result
|
|
72
|
+
// if we got a singular record, we return a singular result
|
|
55
73
|
return Array.isArray(recordB) ? detectionResults : detectionResults[0];
|
|
56
|
-
function actualDetection(
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
74
|
+
function actualDetection({
|
|
75
|
+
featuresA,
|
|
76
|
+
record,
|
|
77
|
+
recordExternal,
|
|
78
|
+
index
|
|
79
|
+
}) {
|
|
80
|
+
const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';
|
|
81
|
+
const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';
|
|
82
|
+
debug(`Actual detection for record ${index + 1} ${labelB}`);
|
|
83
|
+
const featuresB = extractFeatures({
|
|
84
|
+
record,
|
|
85
|
+
recordExternal
|
|
86
|
+
});
|
|
87
|
+
debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);
|
|
88
|
+
debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);
|
|
60
89
|
const featurePairs = generateFeaturePairs(featuresA, featuresB);
|
|
61
90
|
const similarityVector = generateSimilarityVector(featurePairs);
|
|
62
91
|
if (similarityVector.some(v => v >= minProbabilityQuantifier)) {
|
|
@@ -73,13 +102,19 @@ var _default = ({
|
|
|
73
102
|
probability: 0.0
|
|
74
103
|
});
|
|
75
104
|
}
|
|
76
|
-
function extractFeatures(
|
|
105
|
+
function extractFeatures({
|
|
106
|
+
record,
|
|
107
|
+
recordExternal
|
|
108
|
+
}) {
|
|
77
109
|
return strategy.reduce((acc, {
|
|
78
110
|
name,
|
|
79
111
|
extract
|
|
80
112
|
}) => acc.concat({
|
|
81
113
|
name,
|
|
82
|
-
value: extract(
|
|
114
|
+
value: extract({
|
|
115
|
+
record,
|
|
116
|
+
recordExternal
|
|
117
|
+
})
|
|
83
118
|
}), []);
|
|
84
119
|
}
|
|
85
120
|
function returnResult(result) {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","names":["strategy","treshold","returnStrategy","recordA","recordB","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","Array","isArray","recordsB","detectionResults","map","record","actualDetection","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateprobability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","a","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","index","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["/* eslint-disable no-console */\n/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport * as features from './features';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false) => (recordA, recordB) => {\n const minProbabilityQuantifier = 0.5;\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n const debugData = debug.extend('data');\n\n debugData(`Strategy: ${JSON.stringify(strategy)}`);\n debugData(`Treshold: ${JSON.stringify(treshold)}`);\n debugData(`ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n\n const featuresA = extractFeatures(recordA);\n\n debug(`We got an array of records: ${Array.isArray(recordB)}`);\n const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n const detectionResults = recordsB.map(record => actualDetection(featuresA, record));\n\n return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n function actualDetection(featuresA, record) {\n\n const featuresB = extractFeatures(record);\n\n debugData(`Features (a): ${JSON.stringify(featuresA)}`);\n debugData(`Features (b): ${JSON.stringify(featuresB)}`);\n\n const featurePairs = generateFeaturePairs(featuresA, featuresB);\n const similarityVector = generateSimilarityVector(featurePairs);\n\n if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n const probability = calculateprobability(similarityVector);\n debug(`probability: ${probability} (Treshold: ${treshold})`);\n return returnResult({match: probability >= treshold, probability});\n }\n\n debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n return returnResult({match: false, probability: 0.0});\n }\n\n function extractFeatures(record) {\n return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract(record)}), []);\n }\n\n function returnResult(result) {\n if (returnStrategy) {\n debug(`Returning detection strategy with the result`);\n const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n debugData(`${JSON.stringify(resultWithStrategy)}`);\n return resultWithStrategy;\n }\n return result;\n }\n\n function formatStrategy(strategy) {\n const strategyNames = strategy.map(element => element.name);\n return strategyNames || [];\n }\n\n function calculateprobability(similarityVector) {\n const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n return probability > 1.0 ? 1.0 : probability;\n }\n\n function generateSimilarityVector(featurePairs) {\n const compared = featurePairs.map(({name, a, b}) => {\n const {compare} = strategy.find(({name: featureName}) => name === featureName);\n const points = compare(a, b);\n return {name, points};\n });\n\n debugData(`Points: ${JSON.stringify(compared)}`);\n return compared.map(({points}) => points);\n }\n\n function generateFeaturePairs(featuresA, featuresB) {\n const pairs = generatePairs();\n const missingFeatures = findMissing();\n\n debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n return pairs;\n\n function generatePairs() {\n return featuresA\n .reduce((acc, {name, value}, index) => acc.concat({\n name,\n a: value,\n b: featuresB[index].value\n }), [])\n .filter(({a, b}) => {\n if (a.length === 0 || b.length === 0) {\n return false;\n }\n\n return true;\n });\n }\n\n function findMissing() {\n return featuresA\n .map(({name}) => name)\n .filter(v => pairs.some(({name}) => name === v) === false);\n }\n }\n\n};\n"],"mappings":";;;;;;AA6BA;AACA;AAAuC;AAAA;AAAA;AAAA;AA9BvC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,eAiCe,CAAC;EAACA,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,KAAK,CAACC,OAAO,EAAEC,OAAO,KAAK;EAC3F,MAAMC,wBAAwB,GAAG,GAAG;EAEpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,mDAAmD,CAAC;EACpF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtCD,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACX,QAAQ,CAAE,EAAC,CAAC;EAClDQ,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACV,QAAQ,CAAE,EAAC,CAAC;EAClDO,SAAS,CAAE,mBAAkBE,IAAI,CAACC,SAAS,CAACT,cAAc,CAAE,EAAC,CAAC;EAE9D,MAAMU,SAAS,GAAGC,eAAe,CAACV,OAAO,CAAC;EAE1CG,KAAK,CAAE,+BAA8BQ,KAAK,CAACC,OAAO,CAACX,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMY,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACX,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMa,gBAAgB,GAAGD,QAAQ,CAACE,GAAG,CAACC,MAAM,IAAIC,eAAe,CAACR,SAAS,EAAEO,MAAM,CAAC,CAAC;EAEnF,OAAOL,KAAK,CAACC,OAAO,CAACX,OAAO,CAAC,GAAGa,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAe,CAACR,SAAS,EAAEO,MAAM,EAAE;IAE1C,MAAME,SAAS,GAAGR,eAAe,CAACM,MAAM,CAAC;IAEzCX,SAAS,CAAE,iBAAgBE,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IACvDJ,SAAS,CAAE,iBAAgBE,IAAI,CAACC,SAAS,CAACU,SAAS,CAAE,EAAC,CAAC;IAEvD,MAAMC,YAAY,GAAGC,oBAAoB,CAACX,SAAS,EAAES,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAItB,wBAAwB,CAAC,EAAE;MAC7D,MAAMuB,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DlB,KAAK,CAAE,gBAAesB,WAAY,eAAc3B,QAAS,GAAE,CAAC;MAC5D,OAAO6B,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAI3B,QAAQ;QAAE2B;MAAW,CAAC,CAAC;IACpE;IAEApB,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAOyB,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASf,eAAe,CAACM,MAAM,EAAE;IAC/B,OAAOnB,QAAQ,CAACgC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAChB,MAAM;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EAClG;EAEA,SAASW,YAAY,CAACQ,MAAM,EAAE;IAC5B,IAAIpC,cAAc,EAAE;MAClBI,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMiC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAE5B,QAAQ,EAAEwC,cAAc,CAACxC,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HO,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAAC4B,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAc,CAACxC,QAAQ,EAAE;IAChC,MAAMyC,aAAa,GAAGzC,QAAQ,CAACkB,GAAG,CAACwB,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,oBAAoB,CAACL,gBAAgB,EAAE;IAC9C,MAAMI,WAAW,GAAGJ,gBAAgB,CAACQ,MAAM,CAAC,CAACC,GAAG,EAAEN,CAAC,KAAKM,GAAG,GAAGN,CAAC,EAAE,GAAG,CAAC;IACrE,OAAOC,WAAW,GAAG,GAAG,GAAG,GAAG,GAAGA,WAAW;EAC9C;EAEA,SAASH,wBAAwB,CAACH,YAAY,EAAE;IAC9C,MAAMqB,QAAQ,GAAGrB,YAAY,CAACJ,GAAG,CAAC,CAAC;MAACgB,IAAI;MAAEU,CAAC;MAAEC;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAG9C,QAAQ,CAAC+C,IAAI,CAAC,CAAC;QAACb,IAAI,EAAEc;MAAW,CAAC,KAAKd,IAAI,KAAKc,WAAW,CAAC;MAC9E,MAAMC,MAAM,GAAGH,OAAO,CAACF,CAAC,EAAEC,CAAC,CAAC;MAC5B,OAAO;QAACX,IAAI;QAAEe;MAAM,CAAC;IACvB,CAAC,CAAC;IAEFzC,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACgC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAACzB,GAAG,CAAC,CAAC;MAAC+B;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAAS1B,oBAAoB,CAACX,SAAS,EAAES,SAAS,EAAE;IAClD,MAAM6B,KAAK,GAAGC,aAAa,EAAE;IAC7B,MAAMC,eAAe,GAAGC,WAAW,EAAE;IAErC/C,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAACyC,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAa,GAAG;MACvB,OAAOvC,SAAS,CACboB,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAEiB,KAAK,KAAKrB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJU,CAAC,EAAEP,KAAK;QACRQ,CAAC,EAAExB,SAAS,CAACiC,KAAK,CAAC,CAACjB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNkB,MAAM,CAAC,CAAC;QAACX,CAAC;QAAEC;MAAC,CAAC,KAAK;QAClB,IAAID,CAAC,CAACY,MAAM,KAAK,CAAC,IAAIX,CAAC,CAACW,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASH,WAAW,GAAG;MACrB,OAAOzC,SAAS,CACbM,GAAG,CAAC,CAAC;QAACgB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBqB,MAAM,CAAC5B,CAAC,IAAIuB,KAAK,CAACxB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA"}
|
|
1
|
+
{"version":3,"file":"index.js","names":["strategy","treshold","returnStrategy","recordA","recordB","recordAExternal","recordBExternal","minProbabilityQuantifier","debug","createDebugLogger","debugData","extend","JSON","stringify","featuresA","extractFeatures","record","recordExternal","Array","isArray","recordsB","recordsBExternal","detectionResults","map","index","actualDetection","labelA","label","labelB","featuresB","featurePairs","generateFeaturePairs","similarityVector","generateSimilarityVector","some","v","probability","calculateprobability","returnResult","match","reduce","acc","name","extract","concat","value","result","resultWithStrategy","formatStrategy","strategyNames","element","compared","a","b","compare","find","featureName","points","pairs","generatePairs","missingFeatures","findMissing","filter","length"],"sources":["../../src/match-detection/index.js"],"sourcesContent":["/* eslint-disable no-console */\n/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport createDebugLogger from 'debug';\nimport * as features from './features';\n\nexport {features};\n\nexport default ({strategy, treshold = 0.9}, returnStrategy = false) => ({recordA, recordB, recordAExternal = {}, recordBExternal = {}}) => {\n const minProbabilityQuantifier = 0.5;\n\n const debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection');\n const debugData = debug.extend('data');\n\n debugData(`Strategy: ${JSON.stringify(strategy)}, Treshold: ${JSON.stringify(treshold)}, ReturnStrategy: ${JSON.stringify(returnStrategy)}`);\n debugData(`Records: A: ${recordA}\\nB: ${recordB}`);\n debug(`Externals: A: ${JSON.stringify(recordAExternal)}, B: ${JSON.stringify(recordBExternal)}`);\n\n const featuresA = extractFeatures({record: recordA, recordExternal: recordAExternal});\n\n debug(`We got an array of records: ${Array.isArray(recordB)}`);\n const recordsB = Array.isArray(recordB) ? recordB : [recordB];\n const recordsBExternal = Array.isArray(recordB) ? recordBExternal : [recordBExternal];\n\n const detectionResults = recordsB.map((record, index) => actualDetection({featuresA, recordAExternal, record, recordExternal: recordsBExternal[index], index}));\n\n // if we got array of records, we return an array of result\n // if we got a singular record, we return a singular result\n return Array.isArray(recordB) ? detectionResults : detectionResults[0];\n\n function actualDetection({featuresA, record, recordExternal, index}) {\n const labelA = recordAExternal && recordAExternal.label ? recordAExternal.label : 'a';\n const labelB = recordExternal && recordExternal.label ? recordExternal.label : 'b';\n\n\n debug(`Actual detection for record ${index + 1} ${labelB}`);\n const featuresB = extractFeatures({record, recordExternal});\n\n debugData(`Features (a: ${labelA}): ${JSON.stringify(featuresA)}`);\n debugData(`Features (b: ${labelB}): ${JSON.stringify(featuresB)}`);\n\n const featurePairs = generateFeaturePairs(featuresA, featuresB);\n const similarityVector = generateSimilarityVector(featurePairs);\n\n if (similarityVector.some(v => v >= minProbabilityQuantifier)) {\n const probability = calculateprobability(similarityVector);\n debug(`probability: ${probability} (Treshold: ${treshold})`);\n return returnResult({match: probability >= treshold, probability});\n }\n\n debugData(`No feature yielded minimum probability amount of points (${minProbabilityQuantifier})`);\n return returnResult({match: false, probability: 0.0});\n }\n\n function extractFeatures({record, recordExternal}) {\n return strategy.reduce((acc, {name, extract}) => acc.concat({name, value: extract({record, recordExternal})}), []);\n }\n\n function returnResult(result) {\n if (returnStrategy) {\n debug(`Returning detection strategy with the result`);\n const resultWithStrategy = {match: result.match, probability: result.probability, strategy: formatStrategy(strategy), treshold};\n debugData(`${JSON.stringify(resultWithStrategy)}`);\n return resultWithStrategy;\n }\n return result;\n }\n\n function formatStrategy(strategy) {\n const strategyNames = strategy.map(element => element.name);\n return strategyNames || [];\n }\n\n function calculateprobability(similarityVector) {\n const probability = similarityVector.reduce((acc, v) => acc + v, 0.0);\n return probability > 1.0 ? 1.0 : probability;\n }\n\n function generateSimilarityVector(featurePairs) {\n const compared = featurePairs.map(({name, a, b}) => {\n const {compare} = strategy.find(({name: featureName}) => name === featureName);\n const points = compare(a, b);\n return {name, points};\n });\n\n debugData(`Points: ${JSON.stringify(compared)}`);\n return compared.map(({points}) => points);\n }\n\n function generateFeaturePairs(featuresA, featuresB) {\n const pairs = generatePairs();\n const missingFeatures = findMissing();\n\n debug(`Not comparing the following features because one, or both records are missing features: ${JSON.stringify(missingFeatures)}`);\n return pairs;\n\n function generatePairs() {\n return featuresA\n .reduce((acc, {name, value}, index) => acc.concat({\n name,\n a: value,\n b: featuresB[index].value\n }), [])\n .filter(({a, b}) => {\n if (a.length === 0 || b.length === 0) {\n return false;\n }\n\n return true;\n });\n }\n\n function findMissing() {\n return featuresA\n .map(({name}) => name)\n .filter(v => pairs.some(({name}) => name === v) === false);\n }\n }\n\n};\n"],"mappings":";;;;;;AA6BA;AACA;AAAuC;AAAA;AAAA;AAAA;AA9BvC;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AA1BA,eAiCe,CAAC;EAACA,QAAQ;EAAEC,QAAQ,GAAG;AAAG,CAAC,EAAEC,cAAc,GAAG,KAAK,KAAK,CAAC;EAACC,OAAO;EAAEC,OAAO;EAAEC,eAAe,GAAG,CAAC,CAAC;EAAEC,eAAe,GAAG,CAAC;AAAC,CAAC,KAAK;EACzI,MAAMC,wBAAwB,GAAG,GAAG;EAEpC,MAAMC,KAAK,GAAG,IAAAC,cAAiB,EAAC,mDAAmD,CAAC;EACpF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;EAEtCD,SAAS,CAAE,aAAYE,IAAI,CAACC,SAAS,CAACb,QAAQ,CAAE,eAAcY,IAAI,CAACC,SAAS,CAACZ,QAAQ,CAAE,qBAAoBW,IAAI,CAACC,SAAS,CAACX,cAAc,CAAE,EAAC,CAAC;EAC5IQ,SAAS,CAAE,eAAcP,OAAQ,QAAOC,OAAQ,EAAC,CAAC;EAClDI,KAAK,CAAE,iBAAgBI,IAAI,CAACC,SAAS,CAACR,eAAe,CAAE,QAAOO,IAAI,CAACC,SAAS,CAACP,eAAe,CAAE,EAAC,CAAC;EAEhG,MAAMQ,SAAS,GAAGC,eAAe,CAAC;IAACC,MAAM,EAAEb,OAAO;IAAEc,cAAc,EAAEZ;EAAe,CAAC,CAAC;EAErFG,KAAK,CAAE,+BAA8BU,KAAK,CAACC,OAAO,CAACf,OAAO,CAAE,EAAC,CAAC;EAC9D,MAAMgB,QAAQ,GAAGF,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGA,OAAO,GAAG,CAACA,OAAO,CAAC;EAC7D,MAAMiB,gBAAgB,GAAGH,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGE,eAAe,GAAG,CAACA,eAAe,CAAC;EAErF,MAAMgB,gBAAgB,GAAGF,QAAQ,CAACG,GAAG,CAAC,CAACP,MAAM,EAAEQ,KAAK,KAAKC,eAAe,CAAC;IAACX,SAAS;IAAET,eAAe;IAAEW,MAAM;IAAEC,cAAc,EAAEI,gBAAgB,CAACG,KAAK,CAAC;IAAEA;EAAK,CAAC,CAAC,CAAC;;EAE/J;EACA;EACA,OAAON,KAAK,CAACC,OAAO,CAACf,OAAO,CAAC,GAAGkB,gBAAgB,GAAGA,gBAAgB,CAAC,CAAC,CAAC;EAEtE,SAASG,eAAe,CAAC;IAACX,SAAS;IAAEE,MAAM;IAAEC,cAAc;IAAEO;EAAK,CAAC,EAAE;IACnE,MAAME,MAAM,GAAGrB,eAAe,IAAIA,eAAe,CAACsB,KAAK,GAAGtB,eAAe,CAACsB,KAAK,GAAG,GAAG;IACrF,MAAMC,MAAM,GAAGX,cAAc,IAAIA,cAAc,CAACU,KAAK,GAAGV,cAAc,CAACU,KAAK,GAAG,GAAG;IAGlFnB,KAAK,CAAE,+BAA8BgB,KAAK,GAAG,CAAE,IAAGI,MAAO,EAAC,CAAC;IAC3D,MAAMC,SAAS,GAAGd,eAAe,CAAC;MAACC,MAAM;MAAEC;IAAc,CAAC,CAAC;IAE3DP,SAAS,CAAE,gBAAegB,MAAO,MAAKd,IAAI,CAACC,SAAS,CAACC,SAAS,CAAE,EAAC,CAAC;IAClEJ,SAAS,CAAE,gBAAekB,MAAO,MAAKhB,IAAI,CAACC,SAAS,CAACgB,SAAS,CAAE,EAAC,CAAC;IAElE,MAAMC,YAAY,GAAGC,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,CAAC;IAC/D,MAAMG,gBAAgB,GAAGC,wBAAwB,CAACH,YAAY,CAAC;IAE/D,IAAIE,gBAAgB,CAACE,IAAI,CAACC,CAAC,IAAIA,CAAC,IAAI5B,wBAAwB,CAAC,EAAE;MAC7D,MAAM6B,WAAW,GAAGC,oBAAoB,CAACL,gBAAgB,CAAC;MAC1DxB,KAAK,CAAE,gBAAe4B,WAAY,eAAcnC,QAAS,GAAE,CAAC;MAC5D,OAAOqC,YAAY,CAAC;QAACC,KAAK,EAAEH,WAAW,IAAInC,QAAQ;QAAEmC;MAAW,CAAC,CAAC;IACpE;IAEA1B,SAAS,CAAE,4DAA2DH,wBAAyB,GAAE,CAAC;IAClG,OAAO+B,YAAY,CAAC;MAACC,KAAK,EAAE,KAAK;MAAEH,WAAW,EAAE;IAAG,CAAC,CAAC;EACvD;EAEA,SAASrB,eAAe,CAAC;IAACC,MAAM;IAAEC;EAAc,CAAC,EAAE;IACjD,OAAOjB,QAAQ,CAACwC,MAAM,CAAC,CAACC,GAAG,EAAE;MAACC,IAAI;MAAEC;IAAO,CAAC,KAAKF,GAAG,CAACG,MAAM,CAAC;MAACF,IAAI;MAAEG,KAAK,EAAEF,OAAO,CAAC;QAAC3B,MAAM;QAAEC;MAAc,CAAC;IAAC,CAAC,CAAC,EAAE,EAAE,CAAC;EACpH;EAEA,SAASqB,YAAY,CAACQ,MAAM,EAAE;IAC5B,IAAI5C,cAAc,EAAE;MAClBM,KAAK,CAAE,8CAA6C,CAAC;MACrD,MAAMuC,kBAAkB,GAAG;QAACR,KAAK,EAAEO,MAAM,CAACP,KAAK;QAAEH,WAAW,EAAEU,MAAM,CAACV,WAAW;QAAEpC,QAAQ,EAAEgD,cAAc,CAAChD,QAAQ,CAAC;QAAEC;MAAQ,CAAC;MAC/HS,SAAS,CAAE,GAAEE,IAAI,CAACC,SAAS,CAACkC,kBAAkB,CAAE,EAAC,CAAC;MAClD,OAAOA,kBAAkB;IAC3B;IACA,OAAOD,MAAM;EACf;EAEA,SAASE,cAAc,CAAChD,QAAQ,EAAE;IAChC,MAAMiD,aAAa,GAAGjD,QAAQ,CAACuB,GAAG,CAAC2B,OAAO,IAAIA,OAAO,CAACR,IAAI,CAAC;IAC3D,OAAOO,aAAa,IAAI,EAAE;EAC5B;EAEA,SAASZ,oBAAoB,CAACL,gBAAgB,EAAE;IAC9C,MAAMI,WAAW,GAAGJ,gBAAgB,CAACQ,MAAM,CAAC,CAACC,GAAG,EAAEN,CAAC,KAAKM,GAAG,GAAGN,CAAC,EAAE,GAAG,CAAC;IACrE,OAAOC,WAAW,GAAG,GAAG,GAAG,GAAG,GAAGA,WAAW;EAC9C;EAEA,SAASH,wBAAwB,CAACH,YAAY,EAAE;IAC9C,MAAMqB,QAAQ,GAAGrB,YAAY,CAACP,GAAG,CAAC,CAAC;MAACmB,IAAI;MAAEU,CAAC;MAAEC;IAAC,CAAC,KAAK;MAClD,MAAM;QAACC;MAAO,CAAC,GAAGtD,QAAQ,CAACuD,IAAI,CAAC,CAAC;QAACb,IAAI,EAAEc;MAAW,CAAC,KAAKd,IAAI,KAAKc,WAAW,CAAC;MAC9E,MAAMC,MAAM,GAAGH,OAAO,CAACF,CAAC,EAAEC,CAAC,CAAC;MAC5B,OAAO;QAACX,IAAI;QAAEe;MAAM,CAAC;IACvB,CAAC,CAAC;IAEF/C,SAAS,CAAE,WAAUE,IAAI,CAACC,SAAS,CAACsC,QAAQ,CAAE,EAAC,CAAC;IAChD,OAAOA,QAAQ,CAAC5B,GAAG,CAAC,CAAC;MAACkC;IAAM,CAAC,KAAKA,MAAM,CAAC;EAC3C;EAEA,SAAS1B,oBAAoB,CAACjB,SAAS,EAAEe,SAAS,EAAE;IAClD,MAAM6B,KAAK,GAAGC,aAAa,EAAE;IAC7B,MAAMC,eAAe,GAAGC,WAAW,EAAE;IAErCrD,KAAK,CAAE,2FAA0FI,IAAI,CAACC,SAAS,CAAC+C,eAAe,CAAE,EAAC,CAAC;IACnI,OAAOF,KAAK;IAEZ,SAASC,aAAa,GAAG;MACvB,OAAO7C,SAAS,CACb0B,MAAM,CAAC,CAACC,GAAG,EAAE;QAACC,IAAI;QAAEG;MAAK,CAAC,EAAErB,KAAK,KAAKiB,GAAG,CAACG,MAAM,CAAC;QAChDF,IAAI;QACJU,CAAC,EAAEP,KAAK;QACRQ,CAAC,EAAExB,SAAS,CAACL,KAAK,CAAC,CAACqB;MACtB,CAAC,CAAC,EAAE,EAAE,CAAC,CACNiB,MAAM,CAAC,CAAC;QAACV,CAAC;QAAEC;MAAC,CAAC,KAAK;QAClB,IAAID,CAAC,CAACW,MAAM,KAAK,CAAC,IAAIV,CAAC,CAACU,MAAM,KAAK,CAAC,EAAE;UACpC,OAAO,KAAK;QACd;QAEA,OAAO,IAAI;MACb,CAAC,CAAC;IACN;IAEA,SAASF,WAAW,GAAG;MACrB,OAAO/C,SAAS,CACbS,GAAG,CAAC,CAAC;QAACmB;MAAI,CAAC,KAAKA,IAAI,CAAC,CACrBoB,MAAM,CAAC3B,CAAC,IAAIuB,KAAK,CAACxB,IAAI,CAAC,CAAC;QAACQ;MAAI,CAAC,KAAKA,IAAI,KAAKP,CAAC,CAAC,KAAK,KAAK,CAAC;IAC9D;EACF;AAEF,CAAC;AAAA"}
|
|
@@ -72,7 +72,10 @@ describe('match-detection', () => {
|
|
|
72
72
|
subfieldValues: false
|
|
73
73
|
});
|
|
74
74
|
debugData((0, _util.inspect)(recordB));
|
|
75
|
-
const results = detect(
|
|
75
|
+
const results = detect({
|
|
76
|
+
recordA,
|
|
77
|
+
recordB
|
|
78
|
+
});
|
|
76
79
|
debugData(`${JSON.stringify(results)}`);
|
|
77
80
|
(0, _chai.expect)(results).to.eql(expectedResults);
|
|
78
81
|
function formatOptions() {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.spec.js","names":["debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","recurse","fixura","reader","READERS","JSON","callback","getFixture","options","expectedResults","array","enabled","detect","createDetectionInterface","formatOptions","recordA","MarcRecord","subfieldValues","inspect","recordB","map","recordJson","results","stringify","expect","to","eql","contextFeatures","features","strategy","type","v"],"sources":["../../src/match-detection/index.spec.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './features';\nimport createDetectionInterface from '.';\nimport {inspect} from 'util';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:test');\nconst debugData = debug.extend('data');\n\ndescribe('match-detection', () => {\n generateTests({\n path: [__dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],\n useMetadataFile: true,\n recurse: false,\n fixura: {\n reader: READERS.JSON\n },\n callback: ({getFixture, options, expectedResults, array, enabled = true}) => {\n\n if (!enabled) {\n debug(`*** DISABLED TEST! ***`);\n return;\n }\n\n const detect = createDetectionInterface(formatOptions());\n const recordA = new MarcRecord(getFixture('recordA.json'), {subfieldValues: false});\n debugData(inspect(recordA));\n\n debug(`Our recordB is an array of records: ${array}`);\n const recordB = array\n ? getFixture('recordB.json').map(recordJson => new MarcRecord(recordJson, {subfieldValues: false}))\n : new MarcRecord(getFixture('recordB.json'), {subfieldValues: false});\n debugData(inspect(recordB));\n\n const results = detect(recordA, recordB);\n debugData(`${JSON.stringify(results)}`);\n\n expect(results).to.eql(expectedResults);\n\n function formatOptions() {\n const contextFeatures = features[options.strategy.type];\n\n return {\n ...options,\n strategy: options.strategy.features.map(v => contextFeatures[v]())\n };\n }\n }\n });\n});\n"],"mappings":";;AA4BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAsC;AAAA;AAAA;AAnCtC;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;;AAWA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,wDAAwD,CAAC;AACzF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAEtCC,QAAQ,CAAC,iBAAiB,EAAE,MAAM;EAChC,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,OAAO,CAAC;IAC1EC,eAAe,EAAE,IAAI;IACrBC,OAAO,EAAE,KAAK;IACdC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IACDC,QAAQ,EAAE,CAAC;MAACC,UAAU;MAAEC,OAAO;MAAEC,eAAe;MAAEC,KAAK;MAAEC,OAAO,GAAG;IAAI,CAAC,KAAK;MAE3E,IAAI,CAACA,OAAO,EAAE;QACZnB,KAAK,CAAE,wBAAuB,CAAC;QAC/B;MACF;MAEA,MAAMoB,MAAM,GAAG,IAAAC,SAAwB,EAACC,aAAa,EAAE,CAAC;MACxD,MAAMC,OAAO,GAAG,IAAIC,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACnFvB,SAAS,CAAC,IAAAwB,aAAO,EAACH,OAAO,CAAC,CAAC;MAE3BvB,KAAK,CAAE,uCAAsCkB,KAAM,EAAC,CAAC;MACrD,MAAMS,OAAO,GAAGT,KAAK,GACjBH,UAAU,CAAC,cAAc,CAAC,CAACa,GAAG,CAACC,UAAU,IAAI,IAAIL,sBAAU,CAACK,UAAU,EAAE;QAACJ,cAAc,EAAE;MAAK,CAAC,CAAC,CAAC,GACjG,IAAID,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACvEvB,SAAS,CAAC,IAAAwB,aAAO,EAACC,OAAO,CAAC,CAAC;MAE3B,MAAMG,OAAO,GAAGV,MAAM,
|
|
1
|
+
{"version":3,"file":"index.spec.js","names":["debug","createDebugLogger","debugData","extend","describe","generateTests","path","__dirname","useMetadataFile","recurse","fixura","reader","READERS","JSON","callback","getFixture","options","expectedResults","array","enabled","detect","createDetectionInterface","formatOptions","recordA","MarcRecord","subfieldValues","inspect","recordB","map","recordJson","results","stringify","expect","to","eql","contextFeatures","features","strategy","type","v"],"sources":["../../src/match-detection/index.spec.js"],"sourcesContent":["/**\n*\n* @licstart The following is the entire license notice for the JavaScript code in this file.\n*\n* Melinda record matching modules for Javascript\n*\n* Copyright (C) 2020 University Of Helsinki (The National Library Of Finland)\n*\n* This file is part of melinda-record-matching-js\n*\n* melinda-record-matching-js program is free software: you can redistribute it and/or modify\n* it under the terms of the GNU Lesser General Public License as\n* published by the Free Software Foundation, either version 3 of the\n* License, or (at your option) any later version.\n*\n* melinda-record-matching-js is distributed in the hope that it will be useful,\n* but WITHOUT ANY WARRANTY; without even the implied warranty of\n* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n* GNU Lesser General Public License for more details.\n*\n* You should have received a copy of the GNU Affero General Public License\n* along with this program. If not, see <http://www.gnu.org/licenses/>.\n*\n* @licend The above is the entire license notice\n* for the JavaScript code in this file.\n*\n*/\n\nimport generateTests from '@natlibfi/fixugen';\nimport {READERS} from '@natlibfi/fixura';\nimport {expect} from 'chai';\nimport {MarcRecord} from '@natlibfi/marc-record';\nimport * as features from './features';\nimport createDetectionInterface from '.';\nimport {inspect} from 'util';\nimport createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:match-detection:test');\nconst debugData = debug.extend('data');\n\ndescribe('match-detection', () => {\n generateTests({\n path: [__dirname, '..', '..', 'test-fixtures', 'match-detection', 'index'],\n useMetadataFile: true,\n recurse: false,\n fixura: {\n reader: READERS.JSON\n },\n callback: ({getFixture, options, expectedResults, array, enabled = true}) => {\n\n if (!enabled) {\n debug(`*** DISABLED TEST! ***`);\n return;\n }\n\n const detect = createDetectionInterface(formatOptions());\n const recordA = new MarcRecord(getFixture('recordA.json'), {subfieldValues: false});\n debugData(inspect(recordA));\n\n debug(`Our recordB is an array of records: ${array}`);\n const recordB = array\n ? getFixture('recordB.json').map(recordJson => new MarcRecord(recordJson, {subfieldValues: false}))\n : new MarcRecord(getFixture('recordB.json'), {subfieldValues: false});\n debugData(inspect(recordB));\n\n const results = detect({recordA, recordB});\n debugData(`${JSON.stringify(results)}`);\n\n expect(results).to.eql(expectedResults);\n\n function formatOptions() {\n const contextFeatures = features[options.strategy.type];\n\n return {\n ...options,\n strategy: options.strategy.features.map(v => contextFeatures[v]())\n };\n }\n }\n });\n});\n"],"mappings":";;AA4BA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AAAsC;AAAA;AAAA;AAnCtC;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;;AAWA,MAAMA,KAAK,GAAG,IAAAC,cAAiB,EAAC,wDAAwD,CAAC;AACzF,MAAMC,SAAS,GAAGF,KAAK,CAACG,MAAM,CAAC,MAAM,CAAC;AAEtCC,QAAQ,CAAC,iBAAiB,EAAE,MAAM;EAChC,IAAAC,gBAAa,EAAC;IACZC,IAAI,EAAE,CAACC,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE,iBAAiB,EAAE,OAAO,CAAC;IAC1EC,eAAe,EAAE,IAAI;IACrBC,OAAO,EAAE,KAAK;IACdC,MAAM,EAAE;MACNC,MAAM,EAAEC,eAAO,CAACC;IAClB,CAAC;IACDC,QAAQ,EAAE,CAAC;MAACC,UAAU;MAAEC,OAAO;MAAEC,eAAe;MAAEC,KAAK;MAAEC,OAAO,GAAG;IAAI,CAAC,KAAK;MAE3E,IAAI,CAACA,OAAO,EAAE;QACZnB,KAAK,CAAE,wBAAuB,CAAC;QAC/B;MACF;MAEA,MAAMoB,MAAM,GAAG,IAAAC,SAAwB,EAACC,aAAa,EAAE,CAAC;MACxD,MAAMC,OAAO,GAAG,IAAIC,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACnFvB,SAAS,CAAC,IAAAwB,aAAO,EAACH,OAAO,CAAC,CAAC;MAE3BvB,KAAK,CAAE,uCAAsCkB,KAAM,EAAC,CAAC;MACrD,MAAMS,OAAO,GAAGT,KAAK,GACjBH,UAAU,CAAC,cAAc,CAAC,CAACa,GAAG,CAACC,UAAU,IAAI,IAAIL,sBAAU,CAACK,UAAU,EAAE;QAACJ,cAAc,EAAE;MAAK,CAAC,CAAC,CAAC,GACjG,IAAID,sBAAU,CAACT,UAAU,CAAC,cAAc,CAAC,EAAE;QAACU,cAAc,EAAE;MAAK,CAAC,CAAC;MACvEvB,SAAS,CAAC,IAAAwB,aAAO,EAACC,OAAO,CAAC,CAAC;MAE3B,MAAMG,OAAO,GAAGV,MAAM,CAAC;QAACG,OAAO;QAAEI;MAAO,CAAC,CAAC;MAC1CzB,SAAS,CAAE,GAAEW,IAAI,CAACkB,SAAS,CAACD,OAAO,CAAE,EAAC,CAAC;MAEvC,IAAAE,YAAM,EAACF,OAAO,CAAC,CAACG,EAAE,CAACC,GAAG,CAACjB,eAAe,CAAC;MAEvC,SAASK,aAAa,GAAG;QACvB,MAAMa,eAAe,GAAGC,QAAQ,CAACpB,OAAO,CAACqB,QAAQ,CAACC,IAAI,CAAC;QAEvD,OAAO;UACL,GAAGtB,OAAO;UACVqB,QAAQ,EAAErB,OAAO,CAACqB,QAAQ,CAACD,QAAQ,CAACR,GAAG,CAACW,CAAC,IAAIJ,eAAe,CAACI,CAAC,CAAC,EAAE;QACnE,CAAC;MACH;IACF;EACF,CAAC,CAAC;AACJ,CAAC,CAAC"}
|
package/package.json
CHANGED
package/src/index.js
CHANGED
|
@@ -49,8 +49,9 @@ export default ({detection: detectionOptions, search: searchOptions, maxMatches
|
|
|
49
49
|
|
|
50
50
|
const detect = createDetectionInterface(detectionOptions, returnStrategy);
|
|
51
51
|
|
|
52
|
-
return record => {
|
|
53
|
-
|
|
52
|
+
return ({record, recordExternal = {recordSource: 'incomingRecord', label: 'ic'}}) => {
|
|
53
|
+
|
|
54
|
+
const search = createSearchInterface({...searchOptions, record, maxCandidates, recordExternal});
|
|
54
55
|
return iterate({});
|
|
55
56
|
|
|
56
57
|
// candidateCount : amount of candidate records retrived from SRU for matching, NOT including current record set
|
|
@@ -270,10 +271,11 @@ export default ({detection: detectionOptions, search: searchOptions, maxMatches
|
|
|
270
271
|
// eslint-disable-next-line functional/no-conditional-statement
|
|
271
272
|
if (candidateNotInMatches(matches.concat(nonMatches), candidate)) {
|
|
272
273
|
const {record: candidateRecord, id: candidateId} = candidate;
|
|
274
|
+
const recordBExternal = {id: candidate.id, recordSource: 'databaseRecord', label: `db-${candidate.id}`};
|
|
273
275
|
try {
|
|
274
276
|
debug(`Running matchDetection for record ${candidateId} (${newRecordCount}/${recordSetSize})`);
|
|
275
277
|
// we should handle errors from detection somehow - ie. cases where either record or candidateRecord errors
|
|
276
|
-
const detectionResult = detect(record, candidateRecord);
|
|
278
|
+
const detectionResult = detect({recordA: record, recordB: candidateRecord, recordAExternal: recordExternal, recordBExternal});
|
|
277
279
|
|
|
278
280
|
return handleDetectionResult(detectionResult, candidateId, candidateRecord);
|
|
279
281
|
} catch (error) {
|
package/src/index.spec.js
CHANGED
|
@@ -59,7 +59,7 @@ describe('INDEX', () => {
|
|
|
59
59
|
|
|
60
60
|
|
|
61
61
|
const match = createMatchInterface(formatOptions());
|
|
62
|
-
const {matches, matchStatus, nonMatches, conversionFailures} = await match(record);
|
|
62
|
+
const {matches, matchStatus, nonMatches, conversionFailures} = await match({record});
|
|
63
63
|
debugData(`Matches: ${matches.length}, Status: ${matchStatus.status}/${matchStatus.stopReason}, NonMatches: ${nonMatches ? nonMatches.length : 'not returned'}, ConversionFailures: ${conversionFailures ? conversionFailures.length : 'not returned'}`);
|
|
64
64
|
|
|
65
65
|
expect(matchStatus.status).to.eql(expectedMatchStatus);
|
|
@@ -56,8 +56,9 @@ Records with matching local ids for matching local db are a very good match.
|
|
|
56
56
|
|
|
57
57
|
export default () => ({
|
|
58
58
|
name: 'All source IDs',
|
|
59
|
-
extract: record => {
|
|
60
|
-
|
|
59
|
+
extract: ({record, recordExternal}) => {
|
|
60
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
61
|
+
debug(`Creating match detection features for all local source id's for ${label}`);
|
|
61
62
|
|
|
62
63
|
const fSids = record.get('SID');
|
|
63
64
|
debugData(`SID-fields (${fSids.length}): ${JSON.stringify(fSids)}`);
|
|
@@ -33,22 +33,26 @@ import {testStringOrNumber} from '../../../matching-utils';
|
|
|
33
33
|
|
|
34
34
|
export default ({nameTreshold = 10} = {}) => ({
|
|
35
35
|
name: 'Authors',
|
|
36
|
-
extract: record =>
|
|
37
|
-
.
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
36
|
+
extract: ({record}) => {
|
|
37
|
+
//const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
38
|
+
const authors = record.get(/^(?<def>100|700)$/u)
|
|
39
|
+
.map(({subfields}) => {
|
|
40
|
+
return subfields
|
|
41
|
+
.filter(({code, value}) => code && value)
|
|
42
|
+
.filter(({code}) => ['a', '0'].includes(code))
|
|
43
|
+
.map(toObj)
|
|
44
|
+
.reduce((acc, v) => ({...acc, ...v}), {});
|
|
45
|
+
|
|
46
|
+
function toObj({code, value}) {
|
|
47
|
+
if (code === 'a') {
|
|
48
|
+
return {name: testStringOrNumber(value) ? String(value).replace(/[^\p{Letter}\p{Number}]/gu, '').toLowerCase() : ''};
|
|
49
|
+
}
|
|
50
|
+
|
|
51
|
+
return {id: value};
|
|
47
52
|
}
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
}),
|
|
53
|
+
});
|
|
54
|
+
return authors;
|
|
55
|
+
},
|
|
52
56
|
compare: (a, b) => {
|
|
53
57
|
const maxAuthors = a.length > b.length ? a.length : b.length;
|
|
54
58
|
const matchingIds = findMatchingIds();
|
|
@@ -28,6 +28,6 @@
|
|
|
28
28
|
|
|
29
29
|
export default () => ({
|
|
30
30
|
name: 'Host/Component record',
|
|
31
|
-
extract:
|
|
31
|
+
extract: ({record}) => record.get(/^773$/u).length > 0 ? ['component'] : ['host'],
|
|
32
32
|
compare: (a, b) => a[0] === b[0] ? 0.0 : -1.0
|
|
33
33
|
});
|
|
@@ -52,7 +52,7 @@ describe('match-detection/features/bib/', () => {
|
|
|
52
52
|
const record = new MarcRecord(inputRecord, {subfieldValues: false});
|
|
53
53
|
const {extract} = features[feature](options);
|
|
54
54
|
|
|
55
|
-
expect(extract(record)).to.eql(expectedFeatures);
|
|
55
|
+
expect(extract({record})).to.eql(expectedFeatures);
|
|
56
56
|
return;
|
|
57
57
|
}
|
|
58
58
|
|
|
@@ -34,10 +34,12 @@ const debugData = debug.extend('data');
|
|
|
34
34
|
|
|
35
35
|
export default () => ({
|
|
36
36
|
name: 'Language',
|
|
37
|
-
extract: record => {
|
|
37
|
+
extract: ({record, recordExternal}) => {
|
|
38
|
+
const label = recordExternal && recordExternal.label ? recordExternal.label : 'record';
|
|
39
|
+
|
|
38
40
|
const value008 = get008Value();
|
|
39
41
|
const values041 = get041Values();
|
|
40
|
-
debugData(
|
|
42
|
+
debugData(`${label} 008: ${JSON.stringify(value008)}, 041: ${JSON.stringify(values041)}`);
|
|
41
43
|
|
|
42
44
|
if (value008 && values041.length > 0) {
|
|
43
45
|
debugData(`There's both 008 and 041, searching for value in both`);
|
|
@@ -43,7 +43,7 @@ export default () => {
|
|
|
43
43
|
|
|
44
44
|
return {extract, compare};
|
|
45
45
|
|
|
46
|
-
function extract(record) {
|
|
46
|
+
function extract({record, recordExternal}) {
|
|
47
47
|
|
|
48
48
|
const isMelindaRecord = record.get('003').some(f003 => f003.value === 'FI-MELINDA');
|
|
49
49
|
const [f001] = record.get('001').map(field => field.value);
|
|
@@ -54,7 +54,7 @@ export default () => {
|
|
|
54
54
|
f001 === undefined &&
|
|
55
55
|
f035MelindaIds.length < 1) {
|
|
56
56
|
|
|
57
|
-
debug(
|
|
57
|
+
debug(`${recordExternal.label} No Melinda-IDs found`);
|
|
58
58
|
return {};
|
|
59
59
|
}
|
|
60
60
|
|
|
@@ -33,7 +33,7 @@ import {testStringOrNumber} from '../../../matching-utils';
|
|
|
33
33
|
|
|
34
34
|
export default () => ({
|
|
35
35
|
name: 'Publication time',
|
|
36
|
-
extract: record => {
|
|
36
|
+
extract: ({record}) => {
|
|
37
37
|
const value = record.get(/^008$/u)?.[0]?.value || undefined;
|
|
38
38
|
return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
|
|
39
39
|
},
|