This commit is contained in:
@@ -0,0 +1,21 @@
|
||||
(ns cipher-analytical-machine.language_analyzer
|
||||
(:import [org.apache.tika.language LanguageIdentifier])
|
||||
(:gen-class))
|
||||
|
||||
(defn detect-language
|
||||
"Return a identified language."
|
||||
[text]
|
||||
(-> (new LanguageIdentifier text)
|
||||
.getLanguage))
|
||||
|
||||
(defn is-language?
|
||||
"Check if the text written a language."
|
||||
[text language-id]
|
||||
(= (detect-language text)
|
||||
language-id))
|
||||
|
||||
(defn is-nonsense?
|
||||
"Check if the text written a language is not a language."
|
||||
[text language-id]
|
||||
(not (is-language? text language-id)))
|
||||
|
||||
Reference in New Issue
Block a user