Fix f540 and f856 generators (#186)

* Fix f540 generation: uses now only predefined mappings for generating copyright note * Fix f856 generation logic
NatLibFi · Nov 12, 2024 · 651c236 · 651c236
1 parent ee035c5
commit 651c236
Show file tree

Hide file tree

Showing 46 changed files with 433 additions and 105 deletions.
diff --git a/src/transform/convert/common/generate5xx.js b/src/transform/convert/common/generate5xx.js
@@ -178,49 +178,65 @@ export function generate506({getFieldValues}) {
  * @returns Empty array or array containing field 540
  */
 export function generate540({getFieldValues}) {
-  if (blockGeneration({getFieldValues})) {
+  const rightsValues = getFieldValues('dc.rights');
+
+  if (blockGeneration(rightsValues)) {
     return [];
   }
 
-  const rights = generateRights();
-  const uri = generateUri();
-  const url = generateUrl();
-
-  return rights.concat(uri, url);
-
-  function generateRights() {
-    const values = getFieldValues('dc.rights');
-    return values.map(value => {
-      const subfields = generateSubfields();
-      return {tag: '540', ind1: '', ind2: '', subfields};
-
-      function generateSubfields() {
-        return (/All rights reserved/u).test(value) ? [{code: 'a', value}] : [{code: 'c', value}];
-      }
-    });
-  }
+  const fieldMap = {
+    'cc by-nc-nd 4.0': {
+      value: 'CC BY-NC-ND 4.0',
+      url: 'https://creativecommons.org/licenses/by-nc-nd/4.0/deed.fi'
+    },
+    'cc by-sa 4.0': {
+      value: 'CC BY-SA 4.0',
+      url: 'https://creativecommons.org/licenses/by-sa/4.0/deed.fi'
+    },
+    'cc by-sa 4.0 nimeä-jaasamoin': {
+      value: 'CC BY-SA 4.0',
+      url: 'https://creativecommons.org/licenses/by-sa/4.0/deed.fi'
+    },
+    'cc by-nd 4.0': {
+      value: 'CC BY-ND 4.0',
+      url: 'https://creativecommons.org/licenses/by-nd/4.0/deed.fi'
+    },
+    'cc by-nc 4.0': {
+      value: 'CC BY-NC 4.0',
+      url: 'https://creativecommons.org/licenses/by-nc/4.0/deed.fi'
+    },
+    'cc by-nc-sa 4.0': {
+      value: 'CC BY-NC-SA 4.0',
+      url: 'https://creativecommons.org/licenses/by-nc-sa/4.0/deed.fi'
+    },
+    'cc by 4.0': {
+      value: 'CC BY 4.0',
+      url: 'https://creativecommons.org/licenses/by/4.0/deed.fi'
+    }
+  };
 
-  function generateUri() {
-    const values = getFieldValues('dc.rights.uri');
-    return values.map(value => ({
+  return rightsValues
+    .map(v => v.toLowerCase())
+    .filter(v => Object.keys(fieldMap).includes(v))
+    .map(k => ({
       tag: '540', ind1: '', ind2: '',
-      subfields: [{code: 'u', value}]
+      subfields: [
+        {code: 'f', value: fieldMap[k].value},
+        {code: '2', value: 'cc'},
+        {code: 'u', value: fieldMap[k].url}
+      ]
     }));
-  }
 
-  function generateUrl() {
-    const values = getFieldValues('dc.rights.url');
-    return values.map(value => ({
-      tag: '540', ind1: '', ind2: '',
-      subfields: [{code: 'u', value}]
-    }));
-  }
 
-  function blockGeneration({getFieldValues}) {
-    const rightsFields = getFieldValues('dc.rights');
-    const blockValues = ['This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.'];
+  function blockGeneration(rightsValues) {
+    const blockValues = [
+      'This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.',
+      'Tekijänoikeuslaki (404/1961) 9 § Tekijänoikeussuojaa vailla olevat teokset',
+      'Upphovsrättslag (404/1961) 9 § Verk utan upphovsrättsskydd',
+      'Copyright Act (404/1961) 9 § Works excluded from protection'
+    ];
 
-    return rightsFields.some(v => blockValues.includes(v));
+    return rightsValues.some(v => blockValues.includes(v));
   }
 }
 

diff --git a/src/transform/convert/common/generate8xx.js b/src/transform/convert/common/generate8xx.js
@@ -7,68 +7,82 @@ import {isOpenAccess, isValidLink} from '../util';
 import {sourceConfig} from '../../../config';
 
 /**
- * Generates field 856 ($u: optional, $y: optional).
- * Field generation is based on dc.rights.accesslevel value.
- * 856 fields relating to publicly available access prioritization:
- * URN and DOI > URI and URL.
+ * Generates field 856 ($u, $y: optional).
  * @param {Object} ValueInterface containing getFieldValues function
  * @returns Empty array or array containing field 856 ($u, $y)
  */
 export function generate856({getFieldValues}) {
-  const publicAccessFields = generatePublicAccessFields();
-  const otherUrnFields = generateOtherUrnFields();
+  const publicAccessFields = generatePublicAccessFields({getFieldValues});
+  const otherUrnFields = generateOtherUrnFields({getFieldValues});
 
   return publicAccessFields.concat(otherUrnFields);
 
-  function generatePublicAccessFields() {
+  /**
+   * Generation prioritization if open access fields are to be generated:
+   *   1. If dc.identifier.urn has value(s), generate subfields using all these values
+   *   2. If dc.identifier.doi/dc.identifier.uri have value(s), generate subfields using all these values
+  */
+  function generatePublicAccessFields({getFieldValues}) {
     const openAccess = isOpenAccess({getFieldValues});
 
-    // NB: not having access level fields generates f856 public access field
     if (openAccess) {
-      const subfields = generateSubfields();
+      const subfields = generateSubfields({getFieldValues});
       return subfields.length > 0 ? [{tag: '856', ind1: '4', ind2: '0', subfields}] : [];
     }
 
     return [];
 
-    function generateSubfields() {
-      const linkSubfields = generateLinkSubfields();
+
+    function generateSubfields({getFieldValues}) {
+      const linkSubfields = generateLinkSubfields({getFieldValues});
       return linkSubfields.length > 0 ? linkSubfields.concat({code: 'y', value: 'Linkki verkkoaineistoon'}) : [];
     }
-  }
 
-  function generateOtherUrnFields() {
-    const urn = getFieldValues('dc.relation.urn');
+    function generateLinkSubfields({getFieldValues}) {
+      const urn = generateUrn({getFieldValues});
+      const doi = generateU('dc.identifier.doi', {getFieldValues});
+      const uri = generateU('dc.identifier.uri', {getFieldValues});
 
-    return urn.length > 0 ? [
-      {
-        tag: '856', ind1: '4', ind2: '2',
-        subfields: [{code: 'u', value: urn[0]}]
+      if (urn.length > 0) {
+        return urn;
       }
-    ] : [];
-  }
 
-  function generateLinkSubfields() {
-    const urn = generateUrn();
-    const doi = generateU('dc.identifier.doi');
-    const uri = generateU('dc.identifier.uri');
-    const url = generateU('dc.identifier.url');
+      return doi.concat(uri);
 
-    if (urn.length > 0) {
-      return doi ? urn.concat(doi) : urn;
+      // NB: URN needs to be in proper format (http|https://urn.fi/<value>) or otherwise it will not be used in subfield generation
+      function generateUrn({getFieldValues}) {
+        const values = getFieldValues('dc.identifier.urn');
+        return values
+          .filter(urnIsValid)
+          .map(mapUrnValue)
+          .map(v => ({code: 'u', value: v}));
+      }
+
+      function generateU(path, {getFieldValues}) {
+        const values = getFieldValues(path);
+        return values.filter(value => isValidLink(value)).map(value => ({code: 'u', value}));
+      }
     }
+  }
 
-    return doi.concat(uri, url);
+  function generateOtherUrnFields({getFieldValues}) {
+    const urn = getFieldValues('dc.relation.urn');
+    const firstUrnValue = urn.length > 0 ? urn[0] : null;
+    const validUrn = urnIsValid(firstUrnValue);
 
-    function generateUrn() {
-      const values = getFieldValues('dc.identifier.urn');
-      return values.map(v => ({code: 'u', value: (/http:\/\/urn.fi\//u).test(v) ? v : `http://urn.fi/${v}`}));
+    if (!firstUrnValue || !validUrn) {
+      return [];
     }
 
-    function generateU(path) {
-      const values = getFieldValues(path);
-      return values.filter(value => isValidLink(value)).map(value => ({code: 'u', value}));
-    }
+    return [{tag: '856', ind1: '4', ind2: '2', subfields: [{code: 'u', value: mapUrnValue(firstUrnValue)}]}];
+  }
+
+  function urnIsValid(urn) {
+    return (/https?:\/\/urn.fi\//u).test(urn) || (/^URN:/u).test(urn);
+  }
+
+  function mapUrnValue(urn) {
+    return (/https?:\/\/urn.fi\//u).test(urn) ? urn : `http://urn.fi/${urn}`;
   }
 }
 

diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/01/input.json b/test-fixtures/transform/convert/common/generate5xx/generate540/01/input.json
@@ -5,21 +5,5 @@
       "element": "rights",
       "value": "CC BY 4.0"
     }
-  },
-  {
-    "$": {
-      "schema": "dc",
-      "element": "rights",
-      "qualifier": "url",
-      "value": "https://creativecommons.org/licenses/by/4.0/"
-    }
-  },
-  {
-    "$": {
-      "schema": "dc",
-      "element": "rights",
-      "qualifier": "uri",
-      "value": "https://creativecommons.org/licenses/by/4.0/deed.en"
-    }
   }
 ]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/01/metadata.json b/test-fixtures/transform/convert/common/generate5xx/generate540/01/metadata.json
@@ -1,4 +1,4 @@
 {
-  "description": "Generates f540 from each dc.rights, dc.rights.url and dc.rights.uri information",
+  "description": "Generates f540 for CC BY 4.0 using predefined mapping",
   "only": false
 }
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/01/output.json b/test-fixtures/transform/convert/common/generate5xx/generate540/01/output.json
@@ -2,19 +2,9 @@
   {
     "tag": "540", "ind1": "", "ind2": "",
     "subfields": [
-      {"code": "c", "value": "CC BY 4.0"}
-    ]
-  },
-  {
-    "tag": "540", "ind1": "", "ind2": "",
-    "subfields": [
-      {"code": "u", "value": "https://creativecommons.org/licenses/by/4.0/deed.en"}
-    ]
-  },
-  {
-    "tag": "540", "ind1": "", "ind2": "",
-    "subfields": [
-      {"code": "u", "value": "https://creativecommons.org/licenses/by/4.0/"}
+      {"code": "f", "value": "CC BY 4.0"},
+      {"code": "2", "value": "cc"},
+      {"code": "u", "value": "https://creativecommons.org/licenses/by/4.0/deed.fi"}
     ]
   }
 ]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/04/input.json b/test-fixtures/transform/convert/common/generate5xx/generate540/04/input.json
@@ -0,0 +1,32 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "CC BY 4.0"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "url",
+      "value": "https://creativecommons.org/licenses/by/4.0/"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "uri",
+      "value": "https://creativecommons.org/licenses/by/4.0/deed.en"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "Tekijänoikeuslaki (404/1961) 9 § Tekijänoikeussuojaa vailla olevat teokset"
+    }
+  }
+]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/04/metadata.json b/test-fixtures/transform/convert/common/generate5xx/generate540/04/metadata.json
@@ -0,0 +1,4 @@
+{
+  "description": "Blocks generation of f540 if dc.rights has note regarding exclusion from protection (Finnish)",
+  "only": false
+}
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/04/output.json b/test-fixtures/transform/convert/common/generate5xx/generate540/04/output.json
@@ -0,0 +1 @@
+[]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/05/input.json b/test-fixtures/transform/convert/common/generate5xx/generate540/05/input.json
@@ -0,0 +1,32 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "CC BY 4.0"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "url",
+      "value": "https://creativecommons.org/licenses/by/4.0/"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "uri",
+      "value": "https://creativecommons.org/licenses/by/4.0/deed.en"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "Upphovsrättslag (404/1961) 9 § Verk utan upphovsrättsskydd"
+    }
+  }
+]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/05/metadata.json b/test-fixtures/transform/convert/common/generate5xx/generate540/05/metadata.json
@@ -0,0 +1,4 @@
+{
+  "description": "Blocks generation of f540 if dc.rights has note regarding exclusion from protection (Swedish)",
+  "only": false
+}
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/05/output.json b/test-fixtures/transform/convert/common/generate5xx/generate540/05/output.json
@@ -0,0 +1 @@
+[]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/06/input.json b/test-fixtures/transform/convert/common/generate5xx/generate540/06/input.json
@@ -0,0 +1,32 @@
+[
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "CC BY 4.0"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "url",
+      "value": "https://creativecommons.org/licenses/by/4.0/"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "qualifier": "uri",
+      "value": "https://creativecommons.org/licenses/by/4.0/deed.en"
+    }
+  },
+  {
+    "$": {
+      "schema": "dc",
+      "element": "rights",
+      "value": "Copyright Act (404/1961) 9 § Works excluded from protection"
+    }
+  }
+]
diff --git a/test-fixtures/transform/convert/common/generate5xx/generate540/06/metadata.json b/test-fixtures/transform/convert/common/generate5xx/generate540/06/metadata.json
@@ -0,0 +1,4 @@
+{
+  "description": "Blocks generation of f540 if dc.rights has note regarding exclusion from protection (English)",
+  "only": false
+}