@istex/istex-merge - npm Package Compare versions

Comparing version 2.1.5 to 2.2.0

index.js

 const generateMergedDocument = require('./src/generateMergedDocument');
 const defaultMapping = require('./mapping/default.json');
 const defaultRules = require('./rules/default.json');
-const generateHalTEI = require('./src/generateHalTEI');
+const generateHalTei = require('./src/generateHalTei');
 
@@ -10,3 +10,3 @@ module.exports = {
   defaultRules,
-  generateHalTEI,
+  generateHalTei,
 };

lib/SourcesManager.js

@@ -105,2 +105,47 @@ /* eslint-disable no-labels */
 
+  addRnsrInAuthors (authors) {
+    for (const currentAuthor of authors) {
+      // If the current author has no forename or no surname, go to the next author
+      if (!currentAuthor.forename || !currentAuthor.surname) continue;
+
+      for (const otherSource of this.getSourcesOrderedByPriority()) {
+        for (const docObjectFromOtherSource of this.docObjects[otherSource]) {
+          const authorsFromOtherSource = docObjectFromOtherSource.authors;
+          // If there is no author in this source, go to the next source
+          if (!authorsFromOtherSource) continue;
+          // Get the authors from this source that have at least one affiliation with an RNSR
+          const authorsFromOtherSourceThatHaveAffiliationsWithRnsr = _.filter(authorsFromOtherSource, author => {
+            // Make sure the author has affiliations
+            if (!SourceManager.isNonEmptyArray(author.affiliations)) return false;
+            // Look for affiliations with an RNSR (the RNSR can come from an enrichment)
+            for (const affiliation of author.affiliations) {
+              if (SourceManager.isNonEmptyArray(affiliation.rnsr) || SourceManager.isNonEmptyArray(_.get(affiliation, 'enrichments.rnsr'))) {
+                return true;
+              }
+            }
+          });
+          // If the authors from this source don't have affiliations with an RNSR, go to the next source
+          if (_.isEmpty(authorsFromOtherSourceThatHaveAffiliationsWithRnsr)) continue;
+          for (const authorFromOtherSourceThatHaveAffiliationsWithRnsr of authorsFromOtherSourceThatHaveAffiliationsWithRnsr) {
+            // If the current author from the other source is not our current author, go to the next one
+            if (!SourceManager.authorsEqual(authorFromOtherSourceThatHaveAffiliationsWithRnsr, currentAuthor)) continue;
+
+            if (!currentAuthor.rnsr) currentAuthor.rnsr = [];
+            for (const affiliation of authorFromOtherSourceThatHaveAffiliationsWithRnsr.affiliations) {
+              if (SourceManager.isNonEmptyArray(affiliation.rnsr)) {
+                currentAuthor.rnsr = currentAuthor.rnsr.concat(affiliation.rnsr);
+              }
+              const rnsrFromEnrichments = _.get(affiliation, 'enrichments.rnsr');
+              if (SourceManager.isNonEmptyArray(rnsrFromEnrichments)) {
+                currentAuthor.rnsr = currentAuthor.rnsr.concat(rnsrFromEnrichments);
+              }
+            }
+            // If multiple sources give the same RNSR multiple times we want to remove the potential duplicates
+            currentAuthor.rnsr = _.uniq(currentAuthor.rnsr);
+          }
+        }
+      }
+    }
+  }
+
   merge (source, mapping) {
@@ -118,3 +163,6 @@ const result = {
           // Special treatment for the authors
-          if (key === 'authors') this.addOrcIdInAuthors(value);
+          if (key === 'authors') {
+            this.addOrcIdInAuthors(value);
+            this.addRnsrInAuthors(value);
+          }
           result.properties[key] = source;
@@ -121,0 +169,0 @@ _.set(result.data, key, value);

mapping/default.json

@@ -75,21 +75,21 @@ {
   "sciencespo": true,
-  "_technical": false,
-  "_business.name": false,
-  "_business.duplicateGenre": true,
-  "_business.sourceUidChain": true,
-  "_business.sources": false,
-  "_business.duplicates": {
+  "technical": false,
+  "business.name": false,
+  "business.duplicateGenre": true,
+  "business.sourceUidChain": true,
+  "business.sources": false,
+  "business.duplicates": {
     "action": "merge",
     "id": "sourceUid"
   },
-  "_business.duplicateRules": true,
-  "_business.isDuplicate": false,
-  "_business.xPublicationDate": false,
-  "_business.xissn": false,
-  "_business.first3AuthorNames": false,
-  "_business.first3AuthorNamesWithInitials": false,
-  "_business.hasDoi": false,
-  "_business.hasFulltext": false,
-  "_business.isDeduplicable": false,
-  "_business.hasTransDuplicate": false,
+  "business.duplicateRules": true,
+  "business.isDuplicate": false,
+  "business.xPublicationDate": false,
+  "business.xissn": false,
+  "business.first3AuthorNames": false,
+  "business.first3AuthorNamesWithInitials": false,
+  "business.hasDoi": false,
+  "business.hasFulltext": false,
+  "business.isDeduplicable": false,
+  "business.hasTransDuplicate": false,
   "title.default": true,
@@ -96,0 +96,0 @@ "title.en": true,

package.json

 {
   "name": "@istex/istex-merge",
-  "version": "2.1.5",
+  "version": "2.2.0",
   "description": "Library to build merged documents and generate Hal TEIs from them.",
@@ -5,0 +5,0 @@ "main": "index.js",

README.md

@@ -13,3 +13,3 @@ # istex-merge
   - [Example](#example)
-- [generateHalTEI](#generatehaltei)
+- [generateHalTei](#generatehaltei)
   - [Prerequisites](#prerequisites-1)
@@ -47,3 +47,3 @@ - [Usage](#usage-1)
   // ...
-  "_business.duplicates": {
+  "business.duplicates": {
     "action": "merge",
@@ -53,3 +53,3 @@ "id": "sourceUid"
   // ...
-  "_business.hasFulltext": false,
+  "business.hasFulltext": false,
   "fulltextUrl": true
@@ -71,3 +71,3 @@ }
 - Fields with a simple value (like a string): you can specify a path to where the merged data will be in the final object. In the example above, the `sourceUid` field is merged and placed into `sourceUids` (we make it plurial because the value becomes an array).
-- Fields with an array value (like `_business.duplicates`): a property (`sourceUid` in the example above) must be used to discriminate the values and remove potential duplicates if the values are objects.
+- Fields with an array value (like `business.duplicates`): a property (`sourceUid` in the example above) must be used to discriminate the values and remove potential duplicates if the values are objects.
 
@@ -98,3 +98,3 @@
     // ...
-    "_business.hasFulltext": [/*...*/],
+    "business.hasFulltext": [/*...*/],
     "fulltextUrl": [/*...*/]
@@ -242,3 +242,3 @@ }
 
-## generateHalTEI
+## generateHalTei
 Function to generate a Hal TEI from a merged document.
@@ -251,3 +251,3 @@
 ```JS
-const { generateMergedDocument, generateHalTEI } = require('@istex/istex-merge');
+const { generateMergedDocument, generateHalTei } = require('@istex/istex-merge');
 const docObjects = [{...}, {...}, {...}];
@@ -257,9 +257,9 @@
 
-const halTEIAsString = generateHalTEI(mergedDocument);
+const halTeiAsString = generateHalTei(mergedDocument);
 ```
 
-You can also pass an `options` object to `generateHalTEI`. This object is passed as is to [xmlbuilder2](https://oozcitak.github.io/xmlbuilder2/) (the XML builder used by `istex-merge`). You can find all the available options [here](https://oozcitak.github.io/xmlbuilder2/serialization.html#serialization-settings).
+You can also pass an `options` object to `generateHalTei`. This object is passed as is to [xmlbuilder2](https://oozcitak.github.io/xmlbuilder2/) (the XML builder used by `istex-merge`). You can find all the available options [here](https://oozcitak.github.io/xmlbuilder2/serialization.html#serialization-settings).
 For example, you can use this `options` object to pretty print the TEI like so:
 ```JS
-const prettyPrintedTEI = generateHalTEI(mergedDocument, { prettyPrint: true });
+const prettyPrintedTei = generateHalTei(mergedDocument, { prettyPrint: true });
 ```

rules/default.json

@@ -115,18 +115,18 @@ {
     "sciencespo": [],
-    "_technical": [],
-    "_business.name": [],
-    "_business.duplicateGenre": [],
-    "_business.sourceUidChain": [],
-    "_business.sources": [],
-    "_business.duplicates": [],
-    "_business.duplicateRules": [],
-    "_business.isDuplicate": [],
-    "_business.xPublicationDate": [],
-    "_business.xissn": [],
-    "_business.first3AuthorNames": [],
-    "_business.first3AuthorNamesWithInitials": [],
-    "_business.hasDoi": [],
-    "_business.hasFulltext": [],
-    "_business.isDeduplicable": [],
-    "_business.hasTransDuplicate": [],
+    "technical": [],
+    "business.name": [],
+    "business.duplicateGenre": [],
+    "business.sourceUidChain": [],
+    "business.sources": [],
+    "business.duplicates": [],
+    "business.duplicateRules": [],
+    "business.isDuplicate": [],
+    "business.xPublicationDate": [],
+    "business.xissn": [],
+    "business.first3AuthorNames": [],
+    "business.first3AuthorNamesWithInitials": [],
+    "business.hasDoi": [],
+    "business.hasFulltext": [],
+    "business.isDeduplicable": [],
+    "business.hasTransDuplicate": [],
     "title.default": [
@@ -133,0 +133,0 @@ "sudoc-theses",

rules/halWithoutFulltext.json

@@ -55,18 +55,18 @@ {
     "sciencespo": [],
-    "_technical": [],
-    "_business.name": [],
-    "_business.duplicateGenre": [],
-    "_business.sourceUidChain": [],
-    "_business.sources": [],
-    "_business.duplicates": [],
-    "_business.duplicateRules": [],
-    "_business.isDuplicate": [],
-    "_business.xPublicationDate": [],
-    "_business.xissn": [],
-    "_business.first3AuthorNames": [],
-    "_business.first3AuthorNamesWithInitials": [],
-    "_business.hasDoi": [],
-    "_business.hasFulltext": [],
-    "_business.isDeduplicable": [],
-    "_business.hasTransDuplicate": [],
+    "technical": [],
+    "business.name": [],
+    "business.duplicateGenre": [],
+    "business.sourceUidChain": [],
+    "business.sources": [],
+    "business.duplicates": [],
+    "business.duplicateRules": [],
+    "business.isDuplicate": [],
+    "business.xPublicationDate": [],
+    "business.xissn": [],
+    "business.first3AuthorNames": [],
+    "business.first3AuthorNamesWithInitials": [],
+    "business.hasDoi": [],
+    "business.hasFulltext": [],
+    "business.isDeduplicable": [],
+    "business.hasTransDuplicate": [],
     "title.default": [],
@@ -73,0 +73,0 @@ "title.en": [],

src/generateMergedDocument.js

@@ -32,3 +32,3 @@ const _ = require('lodash');
 
-  if (sourceManager.hasSource('hal') && !sourceManager.getPropertyOf('hal', '_business.hasFulltext')) {
+  if (sourceManager.hasSource('hal') && !sourceManager.getPropertyOf('hal', 'business.hasFulltext')) {
     rules = halWithoutFulltextRules;
@@ -35,0 +35,0 @@ }

No alert changes

Improved metrics

Total package byte prevSize

84966

increased by7.05%

Lines of code

964

increased by14.22%

No dependency changes