Add tests for the cipher-analyzers module.

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/cli.clj

@@ -2,6 +2,9 @@
   (:require
    [cipher-analytical-machine.file :as file]
    [cipher-analytical-machine.caesar :as caesar]
+   [cipher-analytical-machine.caesar-analyzers :as caesar-analyzers]
+   [cipher-analytical-machine.symbol-factories :as sf]
+   [cipher-analytical-machine.symbol-frequences :as symbol-frequences]
    [clojure.string :as cs]
    [clojure.tools.cli :refer [parse-opts]])
   (:gen-class))
@@ -17,6 +20,9 @@
     :default "Caesar"]
    ["-e" "--encrypt" "Encrypt the message."]
    ["-d" "--decrypt" "Decrypt the message."]
+   ["-s" "--symbols" "The string will be used as a set of symbols for a cipher."
+    :default (sf/default-symbol-factory "en")]
+   ["-C" "--cracking" "Cracking the encrypted message."]
    ["-h" "--help"]])
 
 (defn usage [options-summary]
@@ -62,7 +68,17 @@
                (contains? options :message-file))
            (or (contains? options :encrypt)
                (contains? options :decrypt)))
-      {:options options :arguments arguments}
+      (cond
+        (contains? options :encrypt)
+        {:options options :arguments arguments :action-type :encrypt}
+
+        (contains? options :decrypt)
+        {:options options :arguments arguments :action-type :decrypt})
+
+      (and (map-has-keys? options [:cipher, :cracking])
+           (or (contains? options :message)
+               (contains? options :message-file)))
+      {:options options :arguments arguments :action-type :cracking}
 
       :else
       {:exit-message (usage summary)})))
@@ -83,16 +99,18 @@
     (file/read-file (:message-file options))))
 
 (defn actions
-  [options arguments]
-  (println (str options arguments))
+  [options arguments action-type]
   (let [message (get-message options arguments)
         key (Integer/parseInt (:key options))
-        symbols "abc"]
+        symbols (:symbols options)]
     (println
      (cond
-       (contains? options :encrypt)
+       (= action-type :cracking)
+       (caesar-analyzers/get-plaintext message symbols symbol-frequences/english-letter-frequences)
+
+       (= action-type :encrypt)
        (caesar/encrypt-message message key symbols)
 
-       (contains? options :decrypt)
+       (= action-type :decrypt)
        (caesar/decrypt-message message key symbols)))))
 

src/cipher_analytical_machine/core.clj

@@ -5,8 +5,8 @@
 
 (defn -main
   [& args]
-  (let [{:keys [options arguments exit-message ok?]} (cli/validate-args args)]
+  (let [{:keys [options arguments action-type exit-message ok?]} (cli/validate-args args)]
     (if exit-message
       (cli/exit exit-message ok?)
-      (cli/actions options arguments))))
+      (cli/actions options arguments action-type))))
 

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

@@ -1,4 +1,4 @@
-(ns cipher-analytical-machine.symbol_factories
+(ns cipher-analytical-machine.symbol-factories
   (:require [clojure.string :as cs])
   (:gen-class))
 
@@ -32,3 +32,32 @@
   []
   "0123456789")
 
+(defn punctuation-symbol-factory
+  "Return punctuation symbols"
+  []
+  ".,:;-!?")
+
+(defn space-symbol-factory
+  "Return punctuation symbols"
+  []
+  " \t\n")
+
+(defn default-symbol-factory
+  "Return a default string of symbols for a language"
+  [language-code]
+  (cond
+    (= language-code "en")
+    (str (english-alphabet-factory)
+         (punctuation-symbol-factory)
+         (digit-set-factory)
+         \ )
+
+    (= language-code "uk")
+    (str (ukrainian-alphabet-factory)
+         (punctuation-symbol-factory)
+         (digit-set-factory)
+         \ )
+
+    :else
+    (default-symbol-factory "en")))
+

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/cipher_analyzers_test.clj

@@ -0,0 +1,25 @@
+(ns cipher-analytical-machine.cipher-analyzers-test
+  (:require
+   [clojure.test :refer :all]
+   [cipher-analytical-machine.cipher-analyzers :refer :all]))
+
+(deftest count-characters-test
+  (testing "Count two dublicates and one uniq character from a string."
+    (is (= {\a 2 \b 1}
+           (count-characters "aba")))))
+
+(deftest chi-squared-letter-statistic-test
+  (testing "If the frequence of A is 0.1, the text length is 100 and contains 20 times, then the chi-squared is 10."
+    (is (= 10.0
+           (chi-squared-letter-statistic 20 0.1 100))))
+
+  (testing "If the frequence of A is 0.1, the text length is 10 and contains 5 times, then the chi-squared is 16.0."
+    (is (= 16.0
+           (chi-squared-letter-statistic 5 0.1 10)))))
+
+(deftest chi-squared-statistic-test
+  (let [text "aaaaabbbbb"]
+  (testing "If the frequences of a and b is 0.1, the text length is 10 and contains them 5 times, then score is 32."
+    (is (= 32.0
+           (chi-squared-statistic text {\a 0.1 \b 0.1}))))))
+

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 івафі лоінїмч йцувіа івапґ." "uk"))))
+

test/cipher_analytical_machine/symbol_factories_test.clj

@@ -1,46 +1,75 @@
 (ns cipher-analytical-machine.symbol-factories-test
   (:require
    [clojure.test :refer :all]
-   [cipher-analytical-machine.symbol_factories :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"
+    (testing "It must have the same order."
       (is (= "abcdefghijklmnopqrstuvwxyz" symbols)))
 
-    (testing "It must have 26 letters"
+    (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"
+    (testing "It must have the same order."
       (is (= "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ" symbols)))
 
-    (testing "It must have 2*26 letters"
+    (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"
+    (testing "It must have the same order."
       (is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя" symbols)))
 
-    (testing "It must have 33 letters"
+    (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"
+    (testing "It must have the same order."
       (is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюяАБВГҐДЕЄЖЗИІЇЙКЛМНОПРСТУФХЦЧШЩЬЮЯ" symbols)))
 
-    (testing "It must have 2*33 letters"
+    (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"
+    (testing "It must have the same order."
       (is (= "0123456789" symbols)))
 
-    (testing "It must have 10 digits"
+    (testing "It must have 10 digits."
       (is (= 10 (count symbols))))))
 
+(deftest punctuation-symbol-factory-test
+  (let [symbols (punctuation-symbol-factory)]
+    (testing "It must have the same order."
+      (is (= ".,:;-!?" symbols)))
+
+    (testing "It must have 7 symbols."
+      (is (= 7 (count symbols))))))
+
+(deftest space-symbol-factory-test
+  (let [symbols (space-symbol-factory)]
+    (testing "It must have the same order."
+      (is (= " \t\n" symbols)))
+
+    (testing "It must have 3 symbols."
+      (is (= 3 (count symbols))))))
+
+(deftest default-symbol-factory-test
+  (testing "If language code is 'en', then it must have symbols."
+    (let [symbols (default-symbol-factory "en")]
+      (is (= 70 (count symbols)))))
+
+  (testing "If language code is 'uk', then it must have symbols."
+    (let [symbols (default-symbol-factory "uk")]
+      (is (= 84 (count symbols)))))
+
+  (testing "If language code is unidentified, then it must have the 'en'."
+    (let [symbols (default-symbol-factory "Maybe a code")]
+      (is (= 70 (count symbols))))))
+