Add a simple language analyzers.

.drone.yaml

@@ -0,0 +1,10 @@
+kind: pipeline
+type: docker
+name: default
+
+steps:
+  - name: test
+    image: clojure:temurin-11-lein-2.10.0-alpine
+    commands:
+      - lein test
+

project.clj

@@ -4,7 +4,9 @@
   :license {:name "LGPL"
             :url "http://www.gnu.org/licenses/lgpl-3.0.txt"}
   :dependencies [[org.clojure/clojure "1.11.1"]
-                 [org.clojure/tools.cli "1.0.219"]]
+                 [org.clojure/tools.cli "1.0.219"]
+                 [org.apache.tika/tika-core "1.28.5"]
+                 [org.apache.tika/tika-langdetect "1.28.5"]]
   :main ^:skip-aot cipher-analytical-machine.core
   :target-path "target/%s"
   :profiles {:uberjar {:aot :all

src/cipher_analytical_machine/cli.clj

@@ -1,5 +1,6 @@
 (ns cipher-analytical-machine.cli
   (:require
+   [cipher-analytical-machine.file :as file]
    [cipher-analytical-machine.caesar :as caesar]
    [clojure.string :as cs]
    [clojure.tools.cli :refer [parse-opts]])
@@ -10,6 +11,7 @@
 
 (def cli-options
   [["-m" "--message MESSAGE" "The message will be encrypted or decrypted."]
+   ["-M" "--message-file MESSAGE-FILE" "The file contains the message that will be encrypted or decrypted."]
    ["-k" "--key KEY" "The key will be used to encrypt or decrypt."]
    ["-c" "--cipher CIPHER" "The cipher will be used to encrypt or decrypt a message."
     :default "Caesar"]
@@ -52,8 +54,12 @@
       (map-has-keys? options [:encrypt, :decrypt])
       {:exit-message (error-msg ["You can't use enctypt and decrypt mode at the same time!!!"])}
 
+      (map-has-keys? options [:message, :message-file])
+      {:exit-message (error-msg ["You can't use message and message-file options at the same time!!!"])}
+
       (and (map-has-keys? options [:cipher, :key])
-           (contains? options :message)
+           (or (contains? options :message)
+               (contains? options :message-file))
            (or (contains? options :encrypt)
                (contains? options :decrypt)))
       {:options options :arguments arguments}
@@ -66,10 +72,20 @@
   (println exit-message)
   (System/exit (if status 0 1)))
 
+(defn get-message
+  "Return the message"
+  [options arguments]
+  (cond
+    (contains? options :message)
+    (:message options)
+
+    (contains? options :message-file)
+    (file/read-file (:message-file options))))
+
 (defn actions
   [options arguments]
   (println (str options arguments))
-  (let [message (:message options)
+  (let [message (get-message options arguments)
         key (Integer/parseInt (:key options))
         symbols "abc"]
     (println

src/cipher_analytical_machine/file.clj

@@ -0,0 +1,18 @@
+(ns cipher-analytical-machine.file
+  (:require [clojure.java.io :as io])
+  (:gen-class))
+
+(defn is-file-exists?
+  "Checks that a file is exists"
+  [file-path]
+  (-> (io/file file-path)
+      (.exists)))
+
+(defn is-file-not-exists?
+  "Checks that a file is not exists"
+  [file-path]
+  (not (is-file-exists? file-path)))
+
+(defn read-file [file-path]
+  (slurp file-path))
+

src/cipher_analytical_machine/language_analyzer.clj

@@ -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)))
+

src/cipher_analytical_machine/symbol_factories.clj

@@ -0,0 +1,34 @@
+(ns cipher-analytical-machine.symbol_factories
+  (:require [clojure.string :as cs])
+  (:gen-class))
+
+(defn lower-english-alphabet-factory
+  "Return the English alphabet with only lowercase letters"
+  []
+  (reduce
+   (fn [acc val] (str acc (char val)))
+   ""
+   (range (int \a) (inc (int \z)))))
+
+(defn english-alphabet-factory
+  "Return the English alphabet with lower and upper case letters"
+  []
+  (let [letters (lower-english-alphabet-factory)]
+    (str letters (cs/upper-case letters))))
+
+(defn lower-ukrainian-alphabet-factory
+  "Return the Ukrainian alphabet with only lowercase letters"
+  []
+  "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя")
+
+(defn ukrainian-alphabet-factory
+  "Return the Ukrainian alphabet with lower and upper case letters"
+  []
+  (let [letters (lower-ukrainian-alphabet-factory)]
+    (str letters (cs/upper-case letters))))
+
+(defn digit-set-factory
+  "Return digits"
+  []
+  "0123456789")
+

test/cipher_analytical_machine/language_analyzer_test.clj

@@ -0,0 +1,45 @@
+(ns cipher-analytical-machine.language_analyzer_test
+  (:require
+   [clojure.test :refer :all]
+   [cipher-analytical-machine.language_analyzer :refer :all]
+  ))
+
+(deftest detect-language-test
+  (testing "Checking the English text"
+    (is (= "en" (detect-language "This is a sentence that uses English."))))
+
+  (testing "Checking the Ukrainian text"
+    (is (= "uk" (detect-language "Текст, що написаний українською."))))
+
+  (testing "Checking a gibberish that uses English letters"
+    (is (= "lt" (detect-language "dfgjldfjgdfl gjdfg idfjig hdgesr khs e."))))
+
+  (testing "Checking a gibberish that uses Ukrainian letters"
+    (is (= "uk" (detect-language "іаврпшшцді врадів аргів аріл варї йцґ.")))))
+
+(deftest is-language?-test
+  (testing "Checking the English text"
+    (is (is-language? "This is a sentence that uses English." "en")))
+
+  (testing "Checking the Ukrainian text"
+    (is (is-language? "Текст, що написаний українською." "uk")))
+
+  (testing "Checking a gibberish that uses English letters"
+    (is (is-language? "dfgjldfjgdfl gjdfg idfjig hdgesr khs e." "lt")))
+
+  (testing "Checking a gibberish that uses Ukrainian letters"
+    (is (is-language? "іаврпшшцді врадів аргів аріл варї йцґ." "uk"))))
+
+(deftest is-nonsense?-test
+  (testing "Checking the English text"
+    (is (not (is-nonsense? "This is a sentence that uses English." "en"))))
+
+  (testing "Checking the Ukrainian text"
+    (is (not (is-nonsense? "Текст, що написаний українською." "uk"))))
+
+  (testing "Checking a gibberish that uses English letters"
+    (is (is-nonsense? "dfgjldfjgdfl gjdfg idfjig hdgesr khs e." "en")))
+
+  (testing "Checking a gibberish that uses Ukrainian letters"
+    (is (is-nonsense? "іаврпшшцді врадів аргів аріл варї йцґ." "uk"))))
+

test/cipher_analytical_machine/symbol_factories_test.clj

@@ -0,0 +1,46 @@
+(ns cipher-analytical-machine.symbol-factories-test
+  (:require
+   [clojure.test :refer :all]
+   [cipher-analytical-machine.symbol_factories :refer :all]
+  ))
+
+(deftest lower-english-alphabet-factory-test
+  (let [symbols (lower-english-alphabet-factory)]
+    (testing "It must have same order"
+      (is (= "abcdefghijklmnopqrstuvwxyz" symbols)))
+
+    (testing "It must have 26 letters"
+      (is (= 26 (count symbols))))))
+
+(deftest english-alphabet-factory-test
+  (let [symbols (english-alphabet-factory)]
+    (testing "It must have same order"
+      (is (= "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ" symbols)))
+
+    (testing "It must have 2*26 letters"
+      (is (= (* 2 26) (count symbols))))))
+
+(deftest lower-ukrainian-alphabet-factory-test
+  (let [symbols (lower-ukrainian-alphabet-factory)]
+    (testing "It must have same order"
+      (is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя" symbols)))
+
+    (testing "It must have 33 letters"
+      (is (= 33 (count symbols))))))
+
+(deftest ukrainian-alphabet-factory-test
+  (let [symbols (ukrainian-alphabet-factory)]
+    (testing "It must have same order"
+      (is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюяАБВГҐДЕЄЖЗИІЇЙКЛМНОПРСТУФХЦЧШЩЬЮЯ" symbols)))
+
+    (testing "It must have 2*33 letters"
+      (is (= (* 2 33) (count symbols))))))
+
+(deftest digit-set-factory-test
+  (let [symbols (digit-set-factory)]
+    (testing "It must have same order"
+      (is (= "0123456789" symbols)))
+
+    (testing "It must have 10 digits"
+      (is (= 10 (count symbols))))))
+