nva-language 1.0.3 → 1.0.7

package/.babelrc

@@ -1,4 +1,7 @@
 {
   "presets": ["@babel/preset-env"],
-  "ignore":  ["**/*.test.js"]
+  "ignore":  ["**/*.test.mjs"],
+  "plugins": [
+    ["replace-import-extension", { "extMapping": { ".mjs": ".js" }}]
+  ]
 }

package/README.md

@@ -1,2 +1,116 @@
 # nva-language-js
-Javascript implementation of NVA language handling
+Javascript implementation of NVA (the Norwegian national research archive) language handling.
+
+Languages in NVA are represented by URIs, this module allows users to find the correct URI for a language and find a name string in a given language for each URI without dereferencing the URI.
+
+NVA-language also routes non-preferred representations to preferred representations.
+
+## Using the library
+
+Install with either:
+
+```javascript
+    const { <function(s)> } = require('nva-language)
+```
+
+or
+
+```javascript
+    import { <function(s)> }  from 'nva-language'
+```
+
+(replace <function(s)> with the functions you actually use, see below).
+
+Use with:
+
+```javascript
+    const english = getLanguageByBokmaalName('Engelsk')
+```
+
+## The language object
+
+All functions return a language object, which has the structure:
+
+```json
+  {
+    "uri": "http://lexvo.org/id/iso639-3/deu",
+    "iso6391Code": "de",
+    "iso6392Codes": ["ger", "deu"],
+    "iso6393Code": "deu",
+    "eng": "German",
+    "nob": "Tysk",
+    "nno": "Tysk",
+    "sme": "Duiskkagiella"
+  }
+```
+
+The properties are structured as follows:
+
+-   uri: a Lexvo ISO 639-3 uri
+-   iso6391Code: an ISO 639-1, two-letter code
+-   iso6392Codes: an array containing ISO 639-2 B/T three-letter codes (min one, max. two items)
+-   iso6393Code: an ISO 639-3 code
+-   eng: the English name of the language
+-   nob: the Norwegian Bokmål name of the language
+-   nno: the Norwegian Nynorsk name of the language
+-   sme: the Northern Sami name of the language
+
+## Available functions
+
+There are a number of methods provided by nva-language, all of which return a Language object, these have self-explanatory names:
+
+- getLanguageByBokmaalName
+    -   example: `const english = getLanguageByBokmaalName('Engelsk')`
+    -   Given a Norwegian Bokmål token for a language, say "Engelsk", returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+- getLanguageByEnglishName
+    -   example: `const english = getLanguageByEnglishName('English')`
+    -   Given a Norwegian Bokmål token for a language, say "English", returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+- getLanguageByIso6391Code
+    -   example: `const english = getLanguageByIso6391Code('en')`
+    -   Given an ISO 639-1 code, returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+- getLanguageByIso6392Code
+    -   example: `const english = getLanguageByIso6392Code('eng')`
+    -   Given an ISO 639-2 code, returns a Language object that either matches the token, or is the Undefined language
+    -   Returns e.g. a Language object for German if either the B/T ("ger", "deu") variants are supplied
+    -   Matching is not case-sensitive
+- getLanguageByIso6393Code
+    -   example: `const english = getLanguageByIso6393Code('eng')`
+    -   Given an ISO 639-1 code, returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+- getLanguageByNynorskName
+    -   example: `const english = getLanguageByNynorskName('Engelsk')`
+    -   Given a Norwegian Nynorsk token for a language, say "Engelsk", returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+    -   The only difference between getLanguageByNynorskName and getLanguageByBokmaalName is Bokmål: "Flere språk", Nynorsk "Fleire språk"
+- getLanguageBySamiName
+    -   example: `const english = getLanguageBySamiName('Eaŋgalsgiella')`
+    -   Given a Northern Sami token for a language, say "Eaŋgalsgiella", returns a Language object that either matches the token, or is the Undefined language
+    -   Matching is not case-sensitive
+- getLanguageByUri
+    -   example: `const english = getLanguageByUri(' http://lexvo.org/id/iso639-3/eng')`
+    -   Given a Lexvo ISO 639-3 URI, returns the corresponding language object or Undefined language (in the case that the URI is not in use)
+    -   Note: Lexvo does *not* use HTTPS
+
+## Notes
+### Valid and invalid languages
+
+The languages supported in NVA are limited to the core languages used in Norwegian research publications, the list grows over time, but the list is still not long.
+
+Thus, some valid languages will return Undefined Language not because they are invalid but because they are undefined in the context of NVA.
+
+### Norwegian
+
+Since NVA is typically concerned with written language, codes and strings related to "Norwegian", as opposed to "Bokmål" and "Nynorsk" are not to be used.
+
+Using a term associated with "Norwegian" will return "Bokmål". No offense is intended here, this mechanism fits better for the majority of cases where texts are marked as "Norwegian".
+
+### Sami
+
+Since the primary source for the data for this module has been NSD DBH's publication channel data, a choice was made to map all "Samisk" to Northern Sami. This is unfortunate, but it solves the issue in view of a lack of Sami language skills. Sorry about that.
+
+### More than one language
+
+We use the ISO 639-3 `mul` code for these cases, which is preferred over the sometimes used `mis` code.

package/dist/LanguageConstants.js

@@ -5,7 +5,7 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.ALL_LANGUAGES = exports.UNDEFINED_LANGUAGE = exports.HUNGARIAN = exports.CZECH = exports.POLISH = exports.CHINESE = exports.CATALAN = exports.GREEK = exports.BULGARIAN = exports.AFRIKAANS = exports.MULTIPLE = exports.SWEDISH = exports.SPANISH = exports.RUSSIAN = exports.PORTUGUESE = exports.NORWEGIAN = exports.ITALIAN = exports.ICELANDIC = exports.FINNISH = exports.DANISH = exports.DUTCH = exports.GERMAN = exports.FRENCH = exports.SAMI = exports.NYNORSK = exports.BOKMAAL = exports.ENGLISH = void 0;
 
-var _Language = _interopRequireDefault(require("./Language.mjs"));
+var _Language = _interopRequireDefault(require("./Language.js"));
 
 function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { "default": obj }; }
 

package/dist/LanguageMapper.js

@@ -5,7 +5,7 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.getLanguageByUri = exports.getLanguageBySamiName = exports.getLanguageByNynorskName = exports.getLanguageByBokmaalName = exports.getLanguageByEnglishName = exports.getLanguageByIso6393Code = exports.getLanguageByIso6392Code = exports.getLanguageByIso6391Code = void 0;
 
-var _LanguageConstants = require("./LanguageConstants.mjs");
+var _LanguageConstants = require("./LanguageConstants.js");
 
 var convertString = function convertString(string) {
   return string !== null ? string.trim().toLowerCase() : '';

package/package.json

@@ -1,9 +1,9 @@
 {
   "name": "nva-language",
-  "version": "1.0.3",
+  "version": "1.0.7",
   "description": "",
   "main": "./dist/LanguageMapper.js",
-  "module": "./src/languageMapper.mjs",
+  "module": "./src/LanguageMapper.mjs",
   "scripts": {
     "lint": "eslint . --ext .mjs --ext .js",
     "lint:fix": "npm run lint -- --fix",
@@ -29,6 +29,7 @@
     "@babel/cli": "^7.15.7",
     "@babel/core": "^7.15.5",
     "@babel/preset-env": "^7.15.6",
+    "babel-plugin-replace-import-extension": "^1.1.1",
     "c8": "^7.9.0",
     "chai": "^4.3.4",
     "eslint-config-standard": "^16.0.3",

package/src/Language.mjs

@@ -0,0 +1,12 @@
+export default (iso6391Code, iso6392Code, iso6393Code, eng, nob, nno, sme) => {
+  return {
+    uri: `http://lexvo.org/id/iso639-3/${iso6393Code}`,
+    iso6391Code: iso6391Code,
+    iso6392Codes: iso6392Code,
+    iso6393Code: iso6393Code,
+    eng: eng,
+    nob: nob,
+    nno: nno,
+    sme: sme
+  }
+}

package/src/LanguageConstants.mjs

@@ -0,0 +1,57 @@
+import Language from './Language.mjs'
+
+export const ENGLISH = Language('en', ['eng'], 'eng', 'English', 'Engelsk', 'Engelsk', 'Eaŋgalsgiella')
+export const BOKMAAL = Language('nb', ['nob'], 'nob', 'Bokmål', 'Bokmål', 'Bokmål', 'Girjedárogiella')
+export const NYNORSK = Language('nn', ['nno'], 'nno', 'Nynorsk', 'Nynorsk', 'Nynorsk', 'Ođđadárogiella')
+export const SAMI = Language('se', [], 'sme', 'Sami', 'Samisk', 'Samisk', 'Davvisámegiella')
+export const FRENCH = Language('fr', ['fra', 'fre'], 'fra', 'French', 'Fransk', 'Fransk', 'Fránskkagiella')
+export const GERMAN = Language('de', ['deu', 'ger'], 'deu', 'German', 'Tysk', 'Tysk', 'Duiskkagiella')
+export const DUTCH = Language('nl', ['nld', 'dut'], 'nld', 'Dutch', 'Nederlandsk', 'Nederlandsk', 'Hollánddagiella')
+export const DANISH = Language('da', ['dan'], 'dan', 'Danish', 'Dansk', 'Dansk', 'Dánskkagiella')
+export const FINNISH = Language('fi', ['fin'], 'fin', 'Finnish', 'Finsk', 'Finsk', 'Suomagiella')
+export const ICELANDIC = Language('is', ['isl', 'ice'], 'isl', 'Icelandic', 'Islandsk', 'Islandsk', 'Islánddagiella')
+export const ITALIAN = Language('it', ['ita'], 'ita', 'Italian', 'Italiensk', 'Italiensk', 'Itáliagiella')
+export const NORWEGIAN = Language('no', ['nor'], 'nor', 'Norwegian', 'Norsk', 'Norsk', 'Dárogiella')
+export const PORTUGUESE = Language('pt', ['por'], 'por', 'Portuguese', 'Portugisisk', 'Portugisisk', 'Portugálagiella')
+export const RUSSIAN = Language('ru', ['rus'], 'rus', 'Russian', 'Russisk', 'Russisk', 'Ruoššagiella')
+export const SPANISH = Language('es', ['spa'], 'spa', 'Spanish', 'Spansk', 'Spansk', 'Spánskkagiella')
+export const SWEDISH = Language('sv', ['swe'], 'swe', 'Swedish', 'Svensk', 'Svensk', 'Ruoŧagiella')
+export const MULTIPLE = Language(null, ['mul'], 'mul', 'Multiple languages', 'Flere språk', 'Fleire språk', 'gielat')
+export const AFRIKAANS = Language('af', ['afr'], 'afr', 'Afrikaans', 'Afrikaans', 'Afrikaans', 'Afrikánsagiella')
+export const BULGARIAN = Language('bg', ['bul'], 'bul', 'Bulgarian', 'Bulgarsk', 'Bulgarsk', 'Bulgáriagiella')
+export const GREEK = Language('el', ['gre', 'ell'], 'ell', 'Greek', 'Gresk', 'Gresk', 'Greikkagiella')
+export const CATALAN = Language('ca', ['cat'], 'car', 'Catalan', 'Katalansk', 'Katalansk', 'Katalánagiella')
+export const CHINESE = Language('zh', ['zho', 'chi'], 'zho', 'Chinese', 'Kinesisk', 'Kinesisk', 'Kiinnágiella')
+export const POLISH = Language('pl', ['pol'], 'pol', 'Polish', 'Polsk', 'Polsk', 'Polskkagiella')
+export const CZECH = Language('cs', ['ces', 'cze'], 'ces', 'Czech', 'Tsjekkisk', 'Tsjekkisk', 'Čeahkagiella')
+export const HUNGARIAN = Language('hu', ['hun'], 'hun', 'Hungarian', 'Ungarsk', 'Ungarsk', 'Ungárgiella')
+export const UNDEFINED_LANGUAGE = Language(null, ['und'], 'und', 'Unknown language', 'Ukjent språk', 'Ukjent språk', 'Dovdameahttun giella')
+
+export const ALL_LANGUAGES = Object.freeze([
+  ENGLISH,
+  BOKMAAL,
+  NYNORSK,
+  SAMI,
+  FRENCH,
+  GERMAN,
+  DUTCH,
+  DANISH,
+  FINNISH,
+  ICELANDIC,
+  ITALIAN,
+  NORWEGIAN,
+  PORTUGUESE,
+  RUSSIAN,
+  SPANISH,
+  SWEDISH,
+  MULTIPLE,
+  UNDEFINED_LANGUAGE,
+  AFRIKAANS,
+  BULGARIAN,
+  GREEK,
+  CATALAN,
+  CHINESE,
+  POLISH,
+  CZECH,
+  HUNGARIAN
+])

package/src/LanguageMapper.mjs

@@ -0,0 +1,45 @@
+import { ALL_LANGUAGES, UNDEFINED_LANGUAGE, BOKMAAL, NORWEGIAN } from './LanguageConstants.mjs'
+
+const convertString = string => string !== null ? string.trim().toLowerCase() : ''
+
+const mappedValue = language => {
+  if (language === null || language === undefined) {
+    return UNDEFINED_LANGUAGE
+  }
+  if (NORWEGIAN === language) {
+    return BOKMAAL
+  }
+  return language
+}
+
+export const getLanguageByIso6391Code = code => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(code) === language.iso6391Code)
+)
+
+export const getLanguageByIso6392Code = code => mappedValue(
+  ALL_LANGUAGES.find(language => language.iso6392Codes.find(variant => convertString(code) === variant))
+)
+
+export const getLanguageByIso6393Code = code => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(code) === language.iso6393Code)
+)
+
+export const getLanguageByEnglishName = name => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(name) === language.eng.toLowerCase())
+)
+
+export const getLanguageByBokmaalName = name => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(name) === language.nob.toLowerCase())
+)
+
+export const getLanguageByNynorskName = name => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(name) === language.nno.toLowerCase())
+)
+
+export const getLanguageBySamiName = name => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(name) === language.sme.toLowerCase())
+)
+
+export const getLanguageByUri = (uri) => mappedValue(
+  ALL_LANGUAGES.find(language => convertString(uri) === language.uri)
+)

package/.github/workflows/node.js.yaml

@@ -1,22 +0,0 @@
-name: Node.js CI
-
-on: [push,pull_request]
-
-jobs:
-  build:
-
-    runs-on: ubuntu-latest
-
-    steps:
-    - uses: actions/checkout@v2
-    - uses: volta-cli/action@v1
-    - run: npm ci
-    - run: npm run build --if-present
-    - run: npm test
-    - run: npm run coverage
-# Disabling for now since codaacy-coverage-reporter does not currently support ECMAScript modules
-#    - name: codacy-coverage-reporter
-#      uses: codacy/codacy-coverage-reporter-action@v1.1.0
-#      with:
-#        project-token: ${{ secrets.CODACY_PROJECT_TOKEN }}
-#        coverage-reports: coverage/cobertura-coverage.xml

package/.github/workflows/npm.publish.yaml

@@ -1,21 +0,0 @@
-name: Publish
-
-on:
-  push:
-    branches:
-      - main
-
-jobs:
-  build:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v1
-      - uses: volta-cli/action@v1.7.0
-      - name: Publishing steps
-        run: |
-          npm install
-          npm config set //registry.npmjs.org/:_authToken ${NPM_TOKEN}
-          npm run build
-          npm publish --ignore-scripts
-        env:
-          NPM_TOKEN: ${{secrets.NPM_TOKEN}}

package/dist/tests/Behaviour.test.js

@@ -1,57 +0,0 @@
-"use strict";
-
-var _mocha = require("mocha");
-
-var _LanguageMapper = require("../LanguageMapper.mjs");
-
-var _LanguageConstants = require("../LanguageConstants.mjs");
-
-var _chai = _interopRequireDefault(require("chai"));
-
-function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { "default": obj }; }
-
-var expect = _chai["default"].expect;
-(0, _mocha.describe)('String values exist', function () {
-  (0, _mocha.it)('should return expected URI"', function () {
-    _LanguageConstants.ALL_LANGUAGES.forEach(function (language) {
-      return expect(language.uri).to.equal('http://lexvo.org/id/iso639-3/' + language.iso6393Code);
-    });
-  });
-  (0, _mocha.it)('should return English when input is valid', function () {
-    expect((0, _LanguageMapper.getLanguageByIso6391Code)('en')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageByIso6392Code)('eng')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageByIso6393Code)('eng')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageByEnglishName)('english')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageByBokmaalName)('engelsk')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageByNynorskName)('engelsk')).to.equal(_LanguageConstants.ENGLISH);
-    expect((0, _LanguageMapper.getLanguageBySamiName)('eaŋgalsgiella')).to.equal(_LanguageConstants.ENGLISH);
-  });
-  (0, _mocha.it)('should return German when input is ISO 639-2 variant', function () {
-    ['deu', 'ger'].forEach(function (variant) {
-      expect((0, _LanguageMapper.getLanguageByIso6392Code)(variant)).to.equal(_LanguageConstants.GERMAN);
-    });
-  });
-  (0, _mocha.it)('should return Bokmål when input is Norwegian', function () {
-    expect((0, _LanguageMapper.getLanguageByIso6391Code)('no')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageByIso6392Code)('nor')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageByIso6393Code)('nor')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageByEnglishName)('norwegian')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageByBokmaalName)('norsk')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageByNynorskName)('norsk')).to.equal(_LanguageConstants.BOKMAAL);
-    expect((0, _LanguageMapper.getLanguageBySamiName)('dárogiella')).to.equal(_LanguageConstants.BOKMAAL);
-  });
-  (0, _mocha.it)('should return Undefined language when input is ISO 639-1 Unknown', function () {
-    expect((0, _LanguageMapper.getLanguageByIso6391Code)('øl')).to.equal(_LanguageConstants.UNDEFINED_LANGUAGE);
-  });
-  (0, _mocha.it)('should return Undefined language when input is ISO 639-1 empty', function () {
-    [null, '', '  ', '\t', '\r\n'].forEach(function (input) {
-      expect((0, _LanguageMapper.getLanguageByIso6391Code)(input)).to.equal(_LanguageConstants.UNDEFINED_LANGUAGE);
-    });
-  });
-  (0, _mocha.it)('should return language when input is valid URI', function () {
-    expect((0, _LanguageMapper.getLanguageByUri)('http://lexvo.org/id/iso639-3/eng')).to.equal(_LanguageConstants.ENGLISH);
-  });
-  (0, _mocha.it)('should return undefined when input is unknown URI', function () {
-    expect((0, _LanguageMapper.getLanguageByUri)('http://example.org/spambot')).to.equal(_LanguageConstants.UNDEFINED_LANGUAGE);
-  });
-});