Add the caesar-analyzers to get the plaintext from a ciphertext.

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/caesar_analyzers.clj

@@ -0,0 +1,34 @@
+(ns cipher-analytical-machine.caesar-analyzers
+  (:require [cipher-analytical-machine.caesar :as caesar]
+            [cipher-analytical-machine.cipher_analyzers :as ca])
+  (:gen-class))
+
+(defn get-all-texts
+  "Return a list of pairs which have a key and a second posible plaintext."
+  [ciphertext symbols]
+  (let [keys (range (count symbols))]
+    (reduce
+     (fn [acc key]
+       (conj acc [key (caesar/decrypt-message ciphertext key symbols)]))
+     '() keys)))
+
+(defn get-key
+  "To find the key with frequencies of letters."
+  [ciphertext symbols letter-frequences]
+  (->> (get-all-texts ciphertext symbols)
+       (map (fn [[key text]]
+              [key (ca/chi-squared-statistic text letter-frequences)]))
+       (reduce
+        (fn [[key value] [new-key new-value]]
+          (if (> value new-value) [new-key new-value]
+               [key value]))
+        [0 Double/MAX_VALUE])
+       (first)))
+
+(defn get-plaintext
+  "Return the plaintext from a ciphertext."
+  [ciphertext symbols letter-frequences]
+  (caesar/decrypt-message ciphertext
+                          (get-key ciphertext symbols letter-frequences)
+                          symbols))
+

src/cipher_analytical_machine/cipher_analyzers.clj

@@ -0,0 +1,38 @@
+(ns cipher-analytical-machine.cipher_analyzers
+  (:require [clojure.string :as cs])
+  (:gen-class))
+
+(defn count-characters
+  "Count the letters and return map."
+  [text]
+  (reduce
+   (fn [acc val]
+     (assoc acc val
+            (inc (get acc val 0))))
+   {} text))
+
+(defn chi-squared-letter-statistic
+  "To calculate the frequency distribution of a letter."
+  [letter-count letter-frequence length-text]
+  (let [letter-frequence (* length-text letter-frequence)]
+    (/ (Math/pow
+        (- letter-count letter-frequence) 2)
+       letter-frequence)))
+
+(defn chi-squared-statistic
+  "To calculate the frequency distribution of a language. More detail about the algorithm here: http://practicalcryptography.com/cryptanalysis/text-characterisation/chi-squared-statistic/"
+  [text letter-frequences]
+  (let [length-text (count text)
+        letter-counts (count-characters (cs/lower-case text))]
+    ;letter-counts
+    (reduce
+     (fn [acc [char count]]
+       (+ acc
+          (if (contains? letter-frequences char)
+            (chi-squared-letter-statistic count
+                                          (get letter-frequences char)
+                                          length-text)
+            0)))
+            0
+            letter-counts)))
+

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_frequences.clj

@@ -0,0 +1,69 @@
+(ns cipher-analytical-machine.symbol_frequences
+  (:gen-class))
+
+(def english-letter-frequences {
+   \a 0.0804
+   \b 0.0154
+   \c 0.0306
+   \d 0.0399
+   \e 0.1251
+   \f 0.0230
+   \g 0.0196
+   \h 0.0549
+   \i 0.0726
+   \j 0.0016
+   \k 0.0067
+   \l 0.0414
+   \m 0.0253
+   \n 0.0709
+   \o 0.0760
+   \p 0.0200
+   \q 0.0011
+   \r 0.0612
+   \s 0.0654
+   \t 0.0925
+   \u 0.0271
+   \v 0.0099
+   \w 0.0192
+   \x 0.0019
+   \y 0.0173
+   \z 0.0009
+})
+
+(def ukrainian-letter-frequences {
+   \а 0.064
+   \б 0.013
+   \в 0.046
+   \г 0.013
+   \ґ 0.0001
+   \д 0.027
+   \е 0.042
+   \є 0.005
+   \ж 0.007
+   \з 0.020
+   \и 0.055
+   \і 0.044
+   \ї 0.010
+   \й 0.009
+   \к 0.033
+   \л 0.027
+   \м 0.029
+   \н 0.068
+   \о 0.086
+   \п 0.025
+   \р 0.043
+   \с 0.037
+   \т 0.045
+   \у 0.027
+   \ф 0.003
+   \х 0.011
+   \ц 0.010
+   \ч 0.011
+   \ш 0.005
+   \щ 0.004
+   \ь 0.016
+   \ю 0.008
+   \я 0.019
+   \  0.138
+})
+

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"))))
+