@natlibfi/melinda-record-matching 1.0.10 → 2.0.1

package/package.json

@@ -14,10 +14,10 @@
 		"url": "git@github.com:natlibfi/melinda-record-matching-js.git"
 	},
 	"license": "LGPL-3.0+",
-	"version": "1.0.10",
+	"version": "2.0.1",
 	"main": "./dist/index.js",
 	"engines": {
-		"node": ">=12"
+		"node": ">=14"
 	},
 	"publishConfig": {
 		"access": "public"
@@ -35,35 +35,34 @@
 		"watch:test": "cross-env DEBUG=1 NODE_ENV=test nodemon -w src -w test-fixtures --exec 'npm run test:dev'"
 	},
 	"dependencies": {
-		"@natlibfi/marc-record": "^6.1.1",
-		"@natlibfi/marc-record-serializers": "^7.2.3",
-		"@natlibfi/sru-client": "^4.0.0",
-		"debug": "^4.1.1",
-		"moment": "^2.28.0",
-		"mongodb": "^3.6.2",
-		"natural": "^2.1.5",
-		"uuid": "^8.3.0",
+		"@natlibfi/marc-record": "^7.0.0",
+		"@natlibfi/marc-record-serializers": "^8.0.4",
+		"@natlibfi/sru-client": "^5.0.0",
+		"debug": "^4.3.3",
+		"moment": "^2.29.1",
+		"natural": "^5.1.11",
+		"uuid": "^8.3.2",
 		"winston": "^3.3.3"
 	},
 	"devDependencies": {
-		"@babel/cli": "^7.11.6",
-		"@babel/core": "^7.11.6",
-		"@babel/node": "^7.10.5",
-		"@babel/preset-env": "^7.11.5",
-		"@babel/register": "^7.11.5",
-		"@natlibfi/eslint-config-melinda-backend": "^1.0.5",
-		"@natlibfi/fixugen": "^1.0.0",
-		"@natlibfi/fixugen-http-client": "^1.1.1",
-		"@natlibfi/fixura": "^2.1.3",
-		"babel-eslint": "^10.1.0",
-		"babel-plugin-istanbul": "^6.0.0",
+		"@babel/cli": "^7.16.0",
+		"@babel/core": "^7.16.0",
+		"@babel/node": "^7.16.0",
+		"@babel/preset-env": "^7.16.4",
+		"@babel/register": "^7.16.0",
+		"@babel/eslint-parser": "^7.16.3",
+		"@natlibfi/eslint-config-melinda-backend": "^1.1.1",
+		"@natlibfi/fixugen": "^1.0.2",
+		"@natlibfi/fixugen-http-client": "^1.1.3",
+		"@natlibfi/fixura": "^2.2.1",
+		"babel-plugin-istanbul": "^6.1.1",
 		"babel-plugin-rewire": "^1.2.0",
-		"chai": "^4.2.0",
+		"chai": "^4.3.4",
 		"chai-as-promised": "^7.1.1",
-		"cross-env": "^7.0.2",
-		"eslint": "^7.9.0",
-		"mocha": "^8.1.3",
-		"nodemon": "^2.0.4",
+		"cross-env": "^7.0.3",
+		"eslint": "^7.32.0",
+		"mocha": "^9.1.3",
+		"nodemon": "^2.0.15",
 		"nyc": "^15.1.0"
 	},
 	"eslintConfig": {
@@ -74,9 +73,7 @@
 			[
 				"@babel/preset-env",
 				{
-					"targets": {
-						"node": "12"
-					}
+					"targets": "maintained node versions"
 				}
 			]
 		],

package/src/candidate-search/candidate-search-utils.js

@@ -33,8 +33,10 @@ export function toQueries(identifiers, queryString) {
   const debug = createDebugLogger('@natlibfi/melinda-record-matching:toQueries');
   const debugData = debug.extend('data');
 
+  const quotedIdentifiers = identifiers.map(identifier => identifier.match(/\//u) ? `"${identifier}"` : `${identifier}`);
+
   // Aleph supports only two queries with or -operator (This is not actually true)
-  const pairs = toPairs(identifiers);
+  const pairs = toPairs(quotedIdentifiers);
   const queries = pairs.map(([a, b]) => b ? `${queryString}=${a} or ${queryString}=${b}` : `${queryString}=${a}`);
 
   debugData(`Pairs (${pairs.length}): ${JSON.stringify(pairs)}`);

package/src/candidate-search/index.js

@@ -27,7 +27,7 @@
 */
 
 import createDebugLogger from 'debug';
-import createClient from '@natlibfi/sru-client';
+import createClient, {SruSearchError} from '@natlibfi/sru-client';
 import {MarcRecord} from '@natlibfi/marc-record';
 import {MARCXML} from '@natlibfi/marc-record-serializers';
 import generateQueryList from './query-list';
@@ -48,6 +48,7 @@ export default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {
     retrieveAll: false
   });
 
+  debug(`Searching matches for ${inputRecordId}`);
   debug(`Generated queryList ${JSON.stringify(queryList)}`);
   if (queryList.length === 0) { // eslint-disable-line functional/no-conditional-statement
     throw new CandidateSearchError(`Generated query list contains no queries`);
@@ -79,6 +80,12 @@ export default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {
 
         client.searchRetrieve(query, {startRecord: resultSetOffset})
           .on('error', err => {
+            // eslint-disable-next-line functional/no-conditional-statement
+            if (err instanceof SruSearchError) {
+              debug(`SRU SruSearchError for query: ${query}: ${err}`);
+              reject(new CandidateSearchError(`SRU SruSearchError for query: ${query}: ${err}`));
+            }
+            debug(`SRU error for query: ${query}: ${err}`);
             reject(new CandidateSearchError(`SRU error for query: ${query}: ${err}`));
           })
           .on('end', async nextOffset => {
@@ -101,6 +108,9 @@ export default ({record, searchSpec, url, maxRecordsPerRequest = 50}) => {
                 const foundRecordMarc = await MARCXML.from(foundRecord, {subfieldValues: false});
                 const foundRecordId = getRecordId(foundRecordMarc);
 
+                // This does not work and might cause problems:
+                // Record *should* match itself AND in REST the input record is given id 000000001 always
+                debug(`Checking ${inputRecordId} vs ${foundRecordId}`);
                 if (inputRecordId === foundRecordId) {
                   debug(`Input and candidate are the same record per 001. Discarding candidate`);
                   return;

package/src/candidate-search/query-list/bib.js

@@ -39,12 +39,15 @@ export function bibSourceIds(record) {
      SID__ $c 123457 $b helka
      SID__ $c (ANDL100020)1077305 $b sata
      SID__ $c VER999999 $ FI-KV
+     SID__ $c /10024/508126 $ REPO_THESEUS
 
     In melinda.sourceid -index case is kept, sourceprefixes in brackets and hyphens are normalized away:
+    Note: slashes are not normalized away, but a SRU-search-string including slashes needs to be quoted
 
     1234567helka
     1077305sata
     VER999999FIKV
+    /10024/508126REPO_THESEUS
 
     Note: All Melinda records that have a matching records in a local db do NOT have SID for that local records,
           existence of a SID field depends on how the record has been added to Melinda and how it has been handled
@@ -56,6 +59,8 @@ export function bibSourceIds(record) {
 
   const debug = createDebugLogger('@natlibfi/melinda-record-matching:candidate-search:query:source-ids');
   const debugData = debug.extend('data');
+  //const debugInfo = debug.extend('info');
+
   debug(`Creating queries for sourceid's`);
 
   const fSids = record.get('SID');
@@ -168,12 +173,20 @@ export function bibHostComponents(record) {
   }
 }
 
+// SRU search dc.title with a search phrase starting with ^ maps currently in Melinda to
+// (probably) to *headings* index TIT
+// Headings index TIT drops articles etc. from the start of the title according to the filing indicator
+// Currently filing indicator is not implemented - if the title starts with an article and the Melinda
+// record is correctly catalogued using a filing indicator -> dc.title search won't match
+
 export function bibTitle(record) {
   const title = getTitle();
 
   if (title) {
     const formatted = title
       .replace(/[^\w\s\p{Alphabetic}]/gu, '')
+      // Clean up concurrent spaces from fe. subfield changes
+      .replace(/ +/gu, ' ')
       .trim()
       .slice(0, 30)
       .trim();
@@ -191,7 +204,8 @@ export function bibTitle(record) {
       return field.subfields
         .filter(({code}) => ['a', 'b'].includes(code))
         .map(({value}) => value)
-        .join('');
+        // In Melinda's index subfield separators are indexed as ' '
+        .join(' ');
     }
     return false;
   }