@natlibfi/melinda-record-matching 5.0.0-alpha.1 → 5.0.0-alpha.3

package/dist/matching-utils.js

@@ -0,0 +1,79 @@
+import createDebugLogger from "debug";
+const debug = createDebugLogger("@natlibfi/melinda-record-matching:utils");
+const debugData = debug.extend("data");
+export function getMelindaIdsF035(record) {
+  const debug2 = createDebugLogger("@natlibfi/melinda-record-matching:melinda-id");
+  const debugData2 = debug2.extend("data");
+  const f035s = record.getFields("035");
+  if (f035s.length < 1) {
+    debug2(`No f035s found.`);
+    return [];
+  }
+  const allF035MelindaIds = [].concat(...f035s.map((field) => toMelindaIds(field)));
+  const f035MelindaIds = [...new Set(allF035MelindaIds)];
+  debugData2(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);
+  debugData2(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);
+  debugData2(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);
+  return f035MelindaIds;
+}
+export function toMelindaIds({ subfields }, subfieldsToParse = ["a", "z"]) {
+  const melindaIdRegExp = /^(?<prefix>\(FI-MELINDA\)|FCC)(?<id>\d{9})$/u;
+  return subfields.filter((sub) => subfieldsToParse.includes(sub.code)).filter((sub) => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value))).map(({ value }) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, "$<id>") : "");
+}
+export function validateSidFieldSubfieldCounts(field) {
+  debugData(`Validating SID field ${JSON.stringify(field)}`);
+  const countC = countSubfields(field, "c");
+  const countB = countSubfields(field, "b");
+  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);
+  return countC === 1 && countB === 1;
+}
+function countSubfields(field, subfieldCode) {
+  return field.subfields.filter(({ code }) => code === subfieldCode).length;
+}
+export function getSubfieldValues(field, subfieldCode) {
+  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);
+  return field.subfields.filter(({ code }) => code === subfieldCode).map(({ value }) => testStringOrNumber(value) ? String(value) : "").filter((value) => value);
+}
+export function testStringOrNumber(value) {
+  if (typeof value === "string" || typeof value === "number") {
+    return true;
+  }
+  return false;
+}
+export function extractSubfieldsFromField(field, subfieldCodes) {
+  if (field === void 0 || field.subfields === void 0) {
+    return [];
+  }
+  const resultSubfields = field.subfields.filter(({ code }) => subfieldCodes.includes(code)).map(({ code, value }) => ({ code, value: testStringOrNumber(value) ? String(value) : "" })).filter((value) => value);
+  return resultSubfields;
+}
+export function uniqueSubfields(subfields) {
+  return subfields.reduce((arr, e) => {
+    if (!arr.find((item) => item.code === e.code && item.value === e.value)) {
+      const newArr = arr.concat(e);
+      return newArr;
+    }
+    return arr;
+  }, []);
+}
+export function getMatchCounts(aValues, bValues) {
+  const matchingValues = getMatchingValuesAmount(aValues, bValues);
+  return {
+    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,
+    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)
+    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,
+    matchingValues
+  };
+  function getMatchingValuesAmount(aValues2, bValues2) {
+    if (bValues2.length > aValues2.length) {
+      return aValues2.filter((aValue) => bValues2.some((bValue) => aValue === bValue)).length;
+    }
+    if (aValues2.length > bValues2.length) {
+      return bValues2.filter((bValue) => aValues2.some((aValue) => bValue === aValue)).length;
+    }
+    const aToB = aValues2.filter((aValue) => bValues2.some((bValue) => aValue === bValue)).length;
+    const bToA = bValues2.filter((bValue) => aValues2.some((aValue) => bValue === aValue)).length;
+    return aToB < bToA ? aToB : bToA;
+  }
+}
+//# sourceMappingURL=matching-utils.js.map

package/dist/matching-utils.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/matching-utils.js"],
+  "sourcesContent": ["import createDebugLogger from 'debug';\n\nconst debug = createDebugLogger('@natlibfi/melinda-record-matching:utils');\nconst debugData = debug.extend('data');\n\nexport function getMelindaIdsF035(record) {\n\n  const debug = createDebugLogger('@natlibfi/melinda-record-matching:melinda-id');\n  const debugData = debug.extend('data');\n\n  const f035s = record.getFields('035');\n\n  if (f035s.length < 1) {\n    debug(`No f035s found.`);\n    return [];\n  }\n\n  const allF035MelindaIds = [].concat(...f035s.map(field => toMelindaIds(field)));\n  const f035MelindaIds = [...new Set(allF035MelindaIds)];\n\n  debugData(`Fields (${f035s.length}): ${JSON.stringify(f035s)}`);\n  debugData(`Ids (${allF035MelindaIds.length}): ${JSON.stringify(allF035MelindaIds)}`);\n  debugData(`Unique ids (${f035MelindaIds.length}): ${JSON.stringify(f035MelindaIds)}`);\n\n  return f035MelindaIds;\n}\n\nexport function toMelindaIds({subfields}, subfieldsToParse = ['a', 'z']) {\n  const melindaIdRegExp = /^(?<prefix>\\(FI-MELINDA\\)|FCC)(?<id>\\d{9})$/u;\n\n  return subfields\n    .filter(sub => subfieldsToParse.includes(sub.code))\n    .filter(sub => testStringOrNumber(sub.value) && melindaIdRegExp.test(String(sub.value)))\n    .map(({value}) => testStringOrNumber(value) ? String(value).replace(melindaIdRegExp, '$<id>') : '');\n}\n\nexport function validateSidFieldSubfieldCounts(field) {\n  // Valid SID-fields have just one $c and one $b\n  debugData(`Validating SID field ${JSON.stringify(field)}`);\n  const countC = countSubfields(field, 'c');\n  const countB = countSubfields(field, 'b');\n  debug(`Found ${countC} sf $cs and ${countB} sf $bs. IsValid: ${countC === 1 && countB === 1}`);\n\n  return countC === 1 && countB === 1;\n}\n\nfunction countSubfields(field, subfieldCode) {\n  // debug(`Counting subfields ${subfieldCode}`);\n  return field.subfields.filter(({code}) => code === subfieldCode).length;\n}\n\nexport function getSubfieldValues(field, subfieldCode) {\n  debugData(`Get subfield(s) $${subfieldCode} from ${JSON.stringify(field)}`);\n  return field.subfields\n    .filter(({code}) => code === subfieldCode)\n    .map(({value}) => testStringOrNumber(value) ? String(value) : '')\n    .filter(value => value);\n}\n\nexport function testStringOrNumber(value) {\n  if (typeof value === 'string' || typeof value === 'number') {\n    return true;\n  }\n  return false;\n}\n\nexport function extractSubfieldsFromField(field, subfieldCodes) {\n  if (field === undefined || field.subfields === undefined) {\n    return [];\n  }\n  const resultSubfields = field.subfields\n    .filter(({code}) => subfieldCodes.includes(code))\n    .map(({code, value}) => ({code, value: testStringOrNumber(value) ? String(value) : ''}))\n    .filter(value => value);\n  return resultSubfields;\n}\n\nexport function uniqueSubfields(subfields) {\n  return subfields.reduce((arr, e) => {\n    if (!arr.find(item => item.code === e.code && item.value === e.value)) {\n      const newArr = arr.concat(e);\n      return newArr;\n    }\n    return arr;\n  }, []);\n}\n\nexport function getMatchCounts(aValues, bValues) {\n\n  const matchingValues = getMatchingValuesAmount(aValues, bValues);\n\n  return {\n    maxValues: aValues.length > bValues.length ? aValues.length : bValues.length,\n    // possibleMatchingValues: amount of identifiers in set of less identifiers (we cannot more values than these)\n    possibleMatchValues: aValues.length > bValues.length ? bValues.length : aValues.length,\n    matchingValues\n  };\n\n  function getMatchingValuesAmount(aValues, bValues) {\n    if (bValues.length > aValues.length) {\n      return aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    }\n    if (aValues.length > bValues.length) {\n      return bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n    }\n\n    // If we have same amount of values, we'll check matches both ways, to avoid mixups in cases\n    // there would be duplicate values\n    const aToB = aValues.filter(aValue => bValues.some(bValue => aValue === bValue)).length;\n    const bToA = bValues.filter(bValue => aValues.some(aValue => bValue === aValue)).length;\n\n    return aToB < bToA ? aToB : bToA;\n  }\n}\n"],
+  "mappings": "AAAA,OAAO,uBAAuB;AAE9B,MAAM,QAAQ,kBAAkB,yCAAyC;AACzE,MAAM,YAAY,MAAM,OAAO,MAAM;AAE9B,gBAAS,kBAAkB,QAAQ;AAExC,QAAMA,SAAQ,kBAAkB,8CAA8C;AAC9E,QAAMC,aAAYD,OAAM,OAAO,MAAM;AAErC,QAAM,QAAQ,OAAO,UAAU,KAAK;AAEpC,MAAI,MAAM,SAAS,GAAG;AACpB,IAAAA,OAAM,iBAAiB;AACvB,WAAO,CAAC;AAAA,EACV;AAEA,QAAM,oBAAoB,CAAC,EAAE,OAAO,GAAG,MAAM,IAAI,WAAS,aAAa,KAAK,CAAC,CAAC;AAC9E,QAAM,iBAAiB,CAAC,GAAG,IAAI,IAAI,iBAAiB,CAAC;AAErD,EAAAC,WAAU,WAAW,MAAM,MAAM,MAAM,KAAK,UAAU,KAAK,CAAC,EAAE;AAC9D,EAAAA,WAAU,QAAQ,kBAAkB,MAAM,MAAM,KAAK,UAAU,iBAAiB,CAAC,EAAE;AACnF,EAAAA,WAAU,eAAe,eAAe,MAAM,MAAM,KAAK,UAAU,cAAc,CAAC,EAAE;AAEpF,SAAO;AACT;AAEO,gBAAS,aAAa,EAAC,UAAS,GAAG,mBAAmB,CAAC,KAAK,GAAG,GAAG;AACvE,QAAM,kBAAkB;AAExB,SAAO,UACJ,OAAO,SAAO,iBAAiB,SAAS,IAAI,IAAI,CAAC,EACjD,OAAO,SAAO,mBAAmB,IAAI,KAAK,KAAK,gBAAgB,KAAK,OAAO,IAAI,KAAK,CAAC,CAAC,EACtF,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,EAAE,QAAQ,iBAAiB,OAAO,IAAI,EAAE;AACtG;AAEO,gBAAS,+BAA+B,OAAO;AAEpD,YAAU,wBAAwB,KAAK,UAAU,KAAK,CAAC,EAAE;AACzD,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,eAAe,OAAO,GAAG;AACxC,QAAM,SAAS,MAAM,eAAe,MAAM,qBAAqB,WAAW,KAAK,WAAW,CAAC,EAAE;AAE7F,SAAO,WAAW,KAAK,WAAW;AACpC;AAEA,SAAS,eAAe,OAAO,cAAc;AAE3C,SAAO,MAAM,UAAU,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EAAE;AACnE;AAEO,gBAAS,kBAAkB,OAAO,cAAc;AACrD,YAAU,oBAAoB,YAAY,SAAS,KAAK,UAAU,KAAK,CAAC,EAAE;AAC1E,SAAO,MAAM,UACV,OAAO,CAAC,EAAC,KAAI,MAAM,SAAS,YAAY,EACxC,IAAI,CAAC,EAAC,MAAK,MAAM,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,EAAE,EAC/D,OAAO,WAAS,KAAK;AAC1B;AAEO,gBAAS,mBAAmB,OAAO;AACxC,MAAI,OAAO,UAAU,YAAY,OAAO,UAAU,UAAU;AAC1D,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAEO,gBAAS,0BAA0B,OAAO,eAAe;AAC9D,MAAI,UAAU,UAAa,MAAM,cAAc,QAAW;AACxD,WAAO,CAAC;AAAA,EACV;AACA,QAAM,kBAAkB,MAAM,UAC3B,OAAO,CAAC,EAAC,KAAI,MAAM,cAAc,SAAS,IAAI,CAAC,EAC/C,IAAI,CAAC,EAAC,MAAM,MAAK,OAAO,EAAC,MAAM,OAAO,mBAAmB,KAAK,IAAI,OAAO,KAAK,IAAI,GAAE,EAAE,EACtF,OAAO,WAAS,KAAK;AACxB,SAAO;AACT;AAEO,gBAAS,gBAAgB,WAAW;AACzC,SAAO,UAAU,OAAO,CAAC,KAAK,MAAM;AAClC,QAAI,CAAC,IAAI,KAAK,UAAQ,KAAK,SAAS,EAAE,QAAQ,KAAK,UAAU,EAAE,KAAK,GAAG;AACrE,YAAM,SAAS,IAAI,OAAO,CAAC;AAC3B,aAAO;AAAA,IACT;AACA,WAAO;AAAA,EACT,GAAG,CAAC,CAAC;AACP;AAEO,gBAAS,eAAe,SAAS,SAAS;AAE/C,QAAM,iBAAiB,wBAAwB,SAAS,OAAO;AAE/D,SAAO;AAAA,IACL,WAAW,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA;AAAA,IAEtE,qBAAqB,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ;AAAA,IAChF;AAAA,EACF;AAEA,WAAS,wBAAwBC,UAASC,UAAS;AACjD,QAAIA,SAAQ,SAASD,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AACA,QAAID,SAAQ,SAASC,SAAQ,QAAQ;AACnC,aAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAAA,IAC7E;AAIA,UAAM,OAAOA,SAAQ,OAAO,YAAUC,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AACjF,UAAM,OAAOA,SAAQ,OAAO,YAAUD,SAAQ,KAAK,YAAU,WAAW,MAAM,CAAC,EAAE;AAEjF,WAAO,OAAO,OAAO,OAAO;AAAA,EAC9B;AACF;",
+  "names": ["debug", "debugData", "aValues", "bValues"]
+}

package/package.json

@@ -5,16 +5,15 @@
 		"name": "The National Library of Finland"
 	},
 	"keywords": [],
-	"homepage": "https://github.com/natlibfi/melinda-record-matching-js",
+	"homepage": "https://www.npmjs.com/package/@natlibfi/melinda-record-matching",
 	"bugs": {
 		"url": "https://github.com/natlibfi/melinda-record-matching-js/issues"
 	},
 	"repository": {
-		"type": "git",
-		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
+		"url": "https://github.com/NatLibFi/melinda-record-matching-js"
 	},
 	"license": "MIT",
-	"version": "5.0.0-alpha.1",
+	"version": "5.0.0-alpha.3",
 	"main": "./dist/index.js",
 	"type": "module",
 	"bin": "./dist/cli.js",
@@ -25,6 +24,7 @@
 		"access": "public"
 	},
 	"scripts": {
+		"prepare": "npm run build",
 		"start": "node dist/index.js",
 		"build": "esbuild \"src/**/*.js\" --sourcemap --platform=node --outdir=dist",
 		"cli": "node dist/cli.js",
@@ -37,22 +37,22 @@
 		"dev:debug": "cross-env LOG_LEVEL=debug DEBUG=@natlibfi/* NODE_ENV=test npm run watch:test"
 	},
 	"dependencies": {
-		"@natlibfi/marc-record": "^10.0.0-alpha.1",
-		"@natlibfi/marc-record-serializers": "^11.0.0-alpha.2",
-		"@natlibfi/melinda-commons": "^14.0.0-alpha.3",
-		"@natlibfi/sru-client": "^7.0.0-alpha.1",
+		"@natlibfi/marc-record": "next",
+		"@natlibfi/marc-record-serializers": "next",
+		"@natlibfi/melinda-commons": "next",
+		"@natlibfi/sru-client": "next",
 		"debug": "^4.4.1",
 		"isbn3": "^1.2.14",
 		"moment": "^2.30.1",
 		"natural": "^8.0.1",
-		"uuid": "^11.1.0",
+		"uuid": "^13.0.0",
 		"winston": "^3.17.0",
 		"yargs": "^18.0.0"
 	},
 	"devDependencies": {
-		"@natlibfi/fixugen": "^3.0.0-alpha.6",
-		"@natlibfi/fixugen-http-client": "^4.0.0-alpha.1",
-		"@natlibfi/fixura": "^4.0.0-alpha.13",
+		"@natlibfi/fixugen": "next",
+		"@natlibfi/fixugen-http-client": "next",
+		"@natlibfi/fixura": "next",
 		"cross-env": "^7.0.3",
 		"esbuild": "^0.25.9",
 		"eslint": "^9.35.0"