@natlibfi/melinda-record-matching 5.0.2-alpha.1 → 5.0.3-alpha.1

package/dist/match-detection/features/bib/publication-time.js

@@ -2,9 +2,51 @@ import { testStringOrNumber } from "../../../matching-utils.js";
 export default () => ({
   name: "Publication time",
   extract: ({ record }) => {
-    const value = record.get(/^008$/u)?.[0]?.value || void 0;
-    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
+    return getDateData();
+    function getDateData() {
+      const [f008] = record.get(/^008$/u);
+      if (!f008 || f008.value.length < 16) {
+        return splitDateData("|        ");
+      }
+      return splitDateData(f008.value.slice(6, 15));
+    }
+    function splitDateData(data) {
+      const typeOfDate = data[0];
+      const date1 = data.slice(1, 5);
+      const date2 = data.slice(5);
+      return { typeOfDate, date1, date2 };
+    }
   },
-  compare: (a, b) => a[0] === b[0] ? 0.1 : -1
+  compare: (aa, bb) => {
+    if (aa.typeOfDate === "b") {
+      if (bb.typeOfDate === "b") {
+        return 0;
+      }
+      return -1;
+    }
+    if (aa.typeOfDate === "n" || bb.typeOfDate === "n") {
+      return 0;
+    }
+    const skipList = ["    ", "||||", "uuuu"];
+    if (skipList.includes(aa.date1) || skipList.includes(bb.date1)) {
+      return 0;
+    }
+    if (matchingYears(aa.date1, bb.date1)) {
+      return 0.1;
+    }
+    return -1;
+    function matchingYears(yyyy1, yyyy2) {
+      if (yyyy1.length === 0) {
+        return true;
+      }
+      if (yyyy1[0] === "u" || yyyy2[0] === "u") {
+        return matchingYears(yyyy1.slice(1), yyyy2.slice(1));
+      }
+      if (yyyy1[0] !== yyyy2[0]) {
+        return false;
+      }
+      return matchingYears(yyyy1.slice(1), yyyy2.slice(1));
+    }
+  }
 });
 //# sourceMappingURL=publication-time.js.map

package/dist/match-detection/features/bib/publication-time.js.map

@@ -1,7 +1,7 @@
 {
   "version": 3,
   "sources": ["../../../../src/match-detection/features/bib/publication-time.js"],
-  "sourcesContent": ["\n\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// see publication-time-allow-cons-years for a version allowing consequent years to match\n\nexport default () => ({\n  name: 'Publication time',\n  extract: ({record}) => {\n    const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];\n  },\n  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0\n});\n"],
-  "mappings": "AAEA,SAAQ,0BAAyB;AAKjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,OAAM,MAAM;AACrB,UAAM,QAAQ,OAAO,IAAI,QAAQ,IAAI,CAAC,GAAG,SAAS;AAClD,WAAO,mBAAmB,KAAK,IAAI,CAAC,OAAO,KAAK,EAAE,MAAM,GAAG,EAAE,CAAC,IAAI,CAAC;AAAA,EACrE;AAAA,EACA,SAAS,CAAC,GAAG,MAAM,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,MAAM;AAC3C;",
+  "sourcesContent": ["\n\nimport {testStringOrNumber} from '../../../matching-utils.js';\n\n// We should also get copyright time and copyright/publication times from 26x\n// see publication-time-allow-cons-years for a version allowing consequent years to match\n\nexport default () => ({\n  name: 'Publication time',\n  extract: ({record}) => {\n    return getDateData();\n    //const value = record.get(/^008$/u)?.[0]?.value || undefined;\n    //return testStringOrNumber(value) ? [String(value).slice(6, 15)] : [];\n\n    function getDateData() {\n      const [f008] = record.get(/^008$/u);\n      if (!f008 || f008.value.length < 16) {\n        return splitDateData('|        ');\n      }\n      return splitDateData(f008.value.slice(6, 15));\n    }\n\n    function splitDateData(data) {\n      const typeOfDate = data[0]; // 008/06\n      const date1 = data.slice(1,5); // 008/07-10\n      const date2 = data.slice(5); // 008/11-14\n      return {typeOfDate, date1, date2};\n    }\n  },\n\n  compare: (aa, bb) => {\n    if (aa.typeOfDate === 'b') { // Berfore Christ. No really makes sense in our domain, though.\n      if (bb.typeOfDate === 'b') {\n        return 0;\n      }\n      return -1.0;\n    }\n    if (aa.typeOfDate === 'n' || bb.typeOfDate === 'n') { // n=unknown\n      return 0;\n    }\n\n    const skipList = ['    ', '||||', 'uuuu'];\n    if (skipList.includes(aa.date1) || skipList.includes(bb.date1)) { // 008/07-10 carries no information\n      return 0;\n    }\n\n    if (matchingYears(aa.date1, bb.date1)) { // 'u' support\n      return 0.1;\n    }\n\n    // TODO: add $q support here\n\n    return -1.0;\n\n    function matchingYears(yyyy1, yyyy2) {\n      if (yyyy1.length === 0) { // All digits have been succesfully consumed -> success\n        return true;\n      }\n      if (yyyy1[0] === 'u' || yyyy2[0] === 'u') { // Ignore 'u' (it refers to unknown millenia, century, decade or year)\n        return matchingYears(yyyy1.slice(1), yyyy2.slice(1));\n      }\n      if (yyyy1[0] !== yyyy2[0]) {\n        return false;\n      }\n      // Should we require that yyyy[0] is a digit at this point?\n      return matchingYears(yyyy1.slice(1), yyyy2.slice(1));\n    }\n\n    /*\n    function hasFourDigits(yyyy) { // Will be needed by 'q' support\n      return yyyy.match(/^[0-9]{4}$/u);\n    }\n      */\n  }\n\n});\n\n\n"],
+  "mappings": "AAEA,SAAQ,0BAAyB;AAKjC,eAAe,OAAO;AAAA,EACpB,MAAM;AAAA,EACN,SAAS,CAAC,EAAC,OAAM,MAAM;AACrB,WAAO,YAAY;AAInB,aAAS,cAAc;AACrB,YAAM,CAAC,IAAI,IAAI,OAAO,IAAI,QAAQ;AAClC,UAAI,CAAC,QAAQ,KAAK,MAAM,SAAS,IAAI;AACnC,eAAO,cAAc,WAAW;AAAA,MAClC;AACA,aAAO,cAAc,KAAK,MAAM,MAAM,GAAG,EAAE,CAAC;AAAA,IAC9C;AAEA,aAAS,cAAc,MAAM;AAC3B,YAAM,aAAa,KAAK,CAAC;AACzB,YAAM,QAAQ,KAAK,MAAM,GAAE,CAAC;AAC5B,YAAM,QAAQ,KAAK,MAAM,CAAC;AAC1B,aAAO,EAAC,YAAY,OAAO,MAAK;AAAA,IAClC;AAAA,EACF;AAAA,EAEA,SAAS,CAAC,IAAI,OAAO;AACnB,QAAI,GAAG,eAAe,KAAK;AACzB,UAAI,GAAG,eAAe,KAAK;AACzB,eAAO;AAAA,MACT;AACA,aAAO;AAAA,IACT;AACA,QAAI,GAAG,eAAe,OAAO,GAAG,eAAe,KAAK;AAClD,aAAO;AAAA,IACT;AAEA,UAAM,WAAW,CAAC,QAAQ,QAAQ,MAAM;AACxC,QAAI,SAAS,SAAS,GAAG,KAAK,KAAK,SAAS,SAAS,GAAG,KAAK,GAAG;AAC9D,aAAO;AAAA,IACT;AAEA,QAAI,cAAc,GAAG,OAAO,GAAG,KAAK,GAAG;AACrC,aAAO;AAAA,IACT;AAIA,WAAO;AAEP,aAAS,cAAc,OAAO,OAAO;AACnC,UAAI,MAAM,WAAW,GAAG;AACtB,eAAO;AAAA,MACT;AACA,UAAI,MAAM,CAAC,MAAM,OAAO,MAAM,CAAC,MAAM,KAAK;AACxC,eAAO,cAAc,MAAM,MAAM,CAAC,GAAG,MAAM,MAAM,CAAC,CAAC;AAAA,MACrD;AACA,UAAI,MAAM,CAAC,MAAM,MAAM,CAAC,GAAG;AACzB,eAAO;AAAA,MACT;AAEA,aAAO,cAAc,MAAM,MAAM,CAAC,GAAG,MAAM,MAAM,CAAC,CAAC;AAAA,IACrD;AAAA,EAOF;AAEF;",
   "names": []
 }

package/package.json

@@ -13,7 +13,7 @@
 		"url": "https://github.com/NatLibFi/melinda-record-matching-js"
 	},
 	"license": "MIT",
-	"version": "5.0.2-alpha.1",
+	"version": "5.0.3-alpha.1",
 	"type": "module",
 	"main": "./dist/index.js",
 	"bin": "./dist/cli.js",
@@ -39,13 +39,13 @@
 	"dependencies": {
 		"@natlibfi/marc-record": "^10.0.1",
 		"@natlibfi/marc-record-serializers": "^11.0.1",
-		"@natlibfi/marc-record-validators-melinda": "^12.0.6",
+		"@natlibfi/marc-record-validators-melinda": "^12.0.7",
 		"@natlibfi/melinda-commons": "^14.0.2",
 		"@natlibfi/sru-client": "^7.0.1",
 		"debug": "^4.4.3",
-		"isbn3": "^2.0.4",
+		"isbn3": "^2.0.6",
 		"moment": "^2.30.1",
-		"natural": "^8.1.0",
+		"natural": "^8.1.1",
 		"uuid": "^13.0.0",
 		"yargs": "^18.0.0"
 	},
@@ -54,8 +54,8 @@
 		"@natlibfi/fixugen-http-client": "^4.0.2",
 		"@natlibfi/fixura": "^4.0.0",
 		"cross-env": "^10.1.0",
-		"esbuild": "^0.27.2",
-		"eslint": "^9.39.2"
+		"esbuild": "^0.27.3",
+		"eslint": "^10.0.2"
 	},
 	"overrides": {
 		"nanoid": "^3.3.8"

package/src/match-detection/features/bib/publication-time.js

@@ -8,8 +8,71 @@ import {testStringOrNumber} from '../../../matching-utils.js';
 export default () => ({
   name: 'Publication time',
   extract: ({record}) => {
-    const value = record.get(/^008$/u)?.[0]?.value || undefined;
-    return testStringOrNumber(value) ? [String(value).slice(7, 11)] : [];
+    return getDateData();
+    //const value = record.get(/^008$/u)?.[0]?.value || undefined;
+    //return testStringOrNumber(value) ? [String(value).slice(6, 15)] : [];
+
+    function getDateData() {
+      const [f008] = record.get(/^008$/u);
+      if (!f008 || f008.value.length < 16) {
+        return splitDateData('|        ');
+      }
+      return splitDateData(f008.value.slice(6, 15));
+    }
+
+    function splitDateData(data) {
+      const typeOfDate = data[0]; // 008/06
+      const date1 = data.slice(1,5); // 008/07-10
+      const date2 = data.slice(5); // 008/11-14
+      return {typeOfDate, date1, date2};
+    }
   },
-  compare: (a, b) => a[0] === b[0] ? 0.1 : -1.0
+
+  compare: (aa, bb) => {
+    if (aa.typeOfDate === 'b') { // Berfore Christ. No really makes sense in our domain, though.
+      if (bb.typeOfDate === 'b') {
+        return 0;
+      }
+      return -1.0;
+    }
+    if (aa.typeOfDate === 'n' || bb.typeOfDate === 'n') { // n=unknown
+      return 0;
+    }
+
+    const skipList = ['    ', '||||', 'uuuu'];
+    if (skipList.includes(aa.date1) || skipList.includes(bb.date1)) { // 008/07-10 carries no information
+      return 0;
+    }
+
+    if (matchingYears(aa.date1, bb.date1)) { // 'u' support
+      return 0.1;
+    }
+
+    // TODO: add $q support here
+
+    return -1.0;
+
+    function matchingYears(yyyy1, yyyy2) {
+      if (yyyy1.length === 0) { // All digits have been succesfully consumed -> success
+        return true;
+      }
+      if (yyyy1[0] === 'u' || yyyy2[0] === 'u') { // Ignore 'u' (it refers to unknown millenia, century, decade or year)
+        return matchingYears(yyyy1.slice(1), yyyy2.slice(1));
+      }
+      if (yyyy1[0] !== yyyy2[0]) {
+        return false;
+      }
+      // Should we require that yyyy[0] is a digit at this point?
+      return matchingYears(yyyy1.slice(1), yyyy2.slice(1));
+    }
+
+    /*
+    function hasFourDigits(yyyy) { // Will be needed by 'q' support
+      return yyyy.match(/^[0-9]{4}$/u);
+    }
+      */
+  }
+
 });
+
+