chore(nlp): Czech tokenizer, stemmer and stopwords added

hubtype · Nov 19, 2020 · 8138d47 · 8138d47
1 parent 4f37d1c
commit 8138d47
Show file tree

Hide file tree

Showing 9 changed files with 469 additions and 0 deletions.
diff --git a/packages/botonic-plugin-contentful/package-lock.json b/packages/botonic-plugin-contentful/package-lock.json
diff --git a/packages/botonic-plugin-contentful/package.json b/packages/botonic-plugin-contentful/package.json
@@ -46,6 +46,7 @@
     "@nlpjs/lang-ru": "^4.14.0",
     "@nlpjs/lang-tr": "^4.14.0",
     "@nlpjs/lang-el": "^4.14.0",
+    "@nlpjs/lang-cs": "^4.14.0",
     "@nlpjs/ner": "^4.14.0",
     "@nlpjs/similarity": "^4.14.0",
     "@types/joi": "^14.3.4",

diff --git a/packages/botonic-plugin-contentful/src/nlp/locales.ts b/packages/botonic-plugin-contentful/src/nlp/locales.ts
@@ -14,6 +14,7 @@ export const FRENCH = 'fr'
 export const GERMAN = 'de'
 export const ROMANIAN = 'ro'
 export const GREEK = 'el'
+export const CZECH = 'cs'
 
 export const SUPPORTED_LOCALES = [
   SPANISH,
@@ -28,6 +29,7 @@ export const SUPPORTED_LOCALES = [
   GERMAN,
   ROMANIAN,
   GREEK,
+  CZECH,
 ]
 
 export function checkLocale(locale: Locale): Locale {

diff --git a/packages/botonic-plugin-contentful/src/nlp/stemmer.ts b/packages/botonic-plugin-contentful/src/nlp/stemmer.ts
@@ -1,5 +1,6 @@
 import { Stemmer } from '@nlpjs/core/src'
 import StemmerCa from '@nlpjs/lang-ca/src/stemmer-ca'
+import StemmerCs from '@nlpjs/lang-cs/src/stemmer-cs'
 import StemmerDe from '@nlpjs/lang-de/src/stemmer-de'
 import StemmerEl from '@nlpjs/lang-el/src/stemmer-el'
 import StemmerEn from '@nlpjs/lang-en-min/src/stemmer-en'
@@ -32,6 +33,7 @@ export const stemmers: { [key: string]: Stemmer } = {
   de: new StemmerDe(),
   ro: new StemmerRo(),
   el: new StemmerEl(),
+  cs: new StemmerCs(),
 }
 
 export function stemmerFor(locale: Locale): Stemmer {