Update Caesar Analyzers to be case-insensitive and testable.

src/cipher_analytical_machine/caesar.clj

@@ -14,21 +14,28 @@
        (+ key)
        (mod max-index)))
 
+(defn encrypt-or-skip
+  "If symbols do not have a character, then it will be skipped."
+  [index character key symbols max-index]
+  (if (nil? index) character
+      (get symbols
+           (encrypt-index index key max-index))))
+
 (defn encrypt-char
   "Calculates new index of a character. Uses a char from symbols, the key and max-index (exclusive)."
   [char key symbols max-index]
-  (get symbols
-       (-> char
-           (calculate-char-index symbols)
-           (encrypt-index key max-index))))
+  (-> char
+      (calculate-char-index symbols)
+      (encrypt-or-skip char key symbols max-index)))
 
 (defn encrypt-message
-  "Encrypt a message using the Caesar cipher."
+  "Encrypt a message using the Caesar cipher. The function is case-insensitive."
   [message key symbols]
   (let [max-index (count symbols)]
-    (cs/join
-     (map (fn [char] (encrypt-char char key symbols max-index))
-          message))))
+    (->> message
+         (cs/lower-case)
+         (map (fn [char] (encrypt-char char key symbols max-index)))
+         (cs/join))))
 
 (defn decrypt-message
   "Decrypt the ciphtext using the Caesar cipher."

src/cipher_analytical_machine/caesar_analyzers.clj

@@ -1,6 +1,7 @@
 (ns cipher-analytical-machine.caesar-analyzers
   (:require [cipher-analytical-machine.caesar :as caesar]
-            [cipher-analytical-machine.cipher_analyzers :as ca])
+            [clojure.string :as cs]
+            [cipher-analytical-machine.cipher-analyzers :as ca])
   (:gen-class))
 
 (defn get-all-texts
@@ -12,23 +13,36 @@
        (conj acc [key (caesar/decrypt-message ciphertext key symbols)]))
      '() keys)))
 
+(defn get-all-scores
+  "For pairs (key, plaintext) finds scores and return list of pairs [key, score]."
+  [letter-frequences pairs]
+  (map (fn [[key text]]
+         [key (ca/chi-squared-statistic text letter-frequences)])
+       pairs))
+
+(defn get-min-score-pair
+  "For pairs (key, plaintext) finds scores and return list of pairs [key, score]."
+  [pairs]
+  (reduce
+   (fn [[key value] [new-key new-value]]
+     (if (> value new-value) [new-key new-value]
+         [key value]))
+   [0 Double/MAX_VALUE]
+   pairs))
+
 (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])
+       (get-all-scores letter-frequences)
+       (get-min-score-pair)
        (first)))
 
 (defn get-plaintext
-  "Return the plaintext from a ciphertext."
+  "Return the plaintext from a ciphertext. The function is case-insensitive."
   [ciphertext symbols letter-frequences]
-  (caesar/decrypt-message ciphertext
-                          (get-key ciphertext symbols letter-frequences)
-                          symbols))
+  (let [ciphertext (cs/lower-case ciphertext)]
+    (caesar/decrypt-message ciphertext
+                            (get-key ciphertext symbols letter-frequences)
+                            symbols)))
 

src/cipher_analytical_machine/cipher_analyzers.clj

@@ -1,4 +1,4 @@
-(ns cipher-analytical-machine.cipher_analyzers
+(ns cipher-analytical-machine.cipher-analyzers
   (:require [clojure.string :as cs])
   (:gen-class))
 

src/cipher_analytical_machine/cli.clj

@@ -12,6 +12,9 @@
 (def cipher-options
   #{"Caesar"})
 
+(def language-options
+  #{"en", "uk"})
+
 (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."]
@@ -22,6 +25,8 @@
    ["-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")]
+   ["-l" "--language CODE" "The string will be used to set a default symbols for a cipher."
+    :validate [#(contains? language-options %) "Must be a code from the list: en, uk!!!"]]
    ["-C" "--cracking" "Cracking the encrypted message."]
    ["-h" "--help"]])
 
@@ -88,29 +93,72 @@
   (println exit-message)
   (System/exit (if status 0 1)))
 
-(defn get-message
-  "Return the message"
-  [options arguments]
-  (cond
-    (contains? options :message)
-    (:message options)
+(defn get-or-load-option
+  "Return the option value or load the value from a file."
+  [options arguments option]
+  (let [file-option (keyword (str (name option) "-file"))]
+    (cond
+      (contains? options option)
+      (option options)
 
-    (contains? options :message-file)
-    (file/read-file (:message-file options))))
+      (contains? options file-option)
+      (file/read-file (file-option options)))))
+
+(defn set-option
+  "Set a option in the option map."
+  [options option value]
+  (-> options
+      (assoc option value)))
+
+(defn set-symbols
+  "Set defaults symbols for a language."
+  [options]
+  (set-option options :symbols
+              (sf/default-symbol-factory (:language options))))
+
+(defn load-and-set-option
+  "Load the option and set it."
+  [options option]
+  (->> (get-or-load-option options nil option)
+       (set-option options option)))
+
+(defn load-all-options
+  "Load all options."
+  [options]
+  (-> options
+      (set-symbols)
+      (load-and-set-option :message)))
+
+(defn cracking-actions
+  [options arguments action-type]
+  (let [message (:message options)
+        symbols (:symbols options)
+        frequencies symbol-frequences/english-letter-frequences]
+    (cond
+      (= action-type :cracking)
+      (caesar-analyzers/get-plaintext message symbols frequencies)
+      )))
+
+(defn crypt-actions
+  [options arguments action-type]
+  (let [message (:message options)
+        key (Integer/parseInt (:key options))
+        symbols (:symbols options)]
+    (cond
+      (= action-type :encrypt)
+      (caesar/encrypt-message message key symbols)
+
+      (= action-type :decrypt)
+      (caesar/decrypt-message message key symbols))))
 
 (defn actions
   [options arguments action-type]
-  (let [message (get-message options arguments)
-        key (Integer/parseInt (:key options))
-        symbols (:symbols options)]
+  (let [options (load-all-options options)]
     (println
      (cond
+       (contains? #{:encrypt :decrypt} action-type)
+       (crypt-actions options arguments action-type)
+
        (= action-type :cracking)
-       (caesar-analyzers/get-plaintext message symbols symbol-frequences/english-letter-frequences)
-
-       (= action-type :encrypt)
-       (caesar/encrypt-message message key symbols)
-
-       (= action-type :decrypt)
-       (caesar/decrypt-message message key symbols)))))
+       (cracking-actions options arguments action-type)))))
 

src/cipher_analytical_machine/symbol_factories.clj

@@ -2,7 +2,7 @@
   (:require [clojure.string :as cs])
   (:gen-class))
 
-(defn lower-english-alphabet-factory
+(defn english-alphabet-factory
   "Return the English alphabet with only lowercase letters"
   []
   (reduce
@@ -10,23 +10,11 @@
    ""
    (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
+(defn 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"
   []
@@ -48,14 +36,10 @@
   (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

test/cipher_analytical_machine/caesar_analyzers_test.clj

@@ -0,0 +1,76 @@
+(ns cipher-analytical-machine.caesar-analyzers-test
+  (:require
+   [clojure.test :refer :all]
+   [cipher-analytical-machine.caesar-analyzers :refer :all]
+   [cipher-analytical-machine.symbol-frequences :as sf]
+   [cipher-analytical-machine.cipher-analyzers :as ca]
+   [clojure.string :as cs]))
+
+(deftest get-all-texts-test
+  (testing "There're three text when the text is 'abc' and the set of symbol is 'abc'."
+    (is (= 3 (count (get-all-texts "abc" "abc")))))
+
+  (testing "A pair must have an integer and a string."
+    (let [pairs (get-all-texts "abc" "abc")
+          pair (first pairs)]
+      (is (and (int? (first pair))
+               (string? (last pair))))))
+
+  (testing "There're 27 combinations for a 'Hello World'."
+    (let [ciphertext "khoorczruog"
+        symbols "abcdefghijklmnopqrstuvwxyz "]
+      (is (= 27 (count (get-all-texts ciphertext symbols)))))))
+
+(deftest get-all-scores-test
+  (testing "There're 27 combinations for a 'Hello World'."
+    (let [ciphertext "khoorczruog"
+          symbols "abcdefghijklmnopqrstuvwxyz "
+          frequences sf/english-letter-frequences
+          text-pairs (get-all-texts ciphertext symbols)]
+      (is (= 27 (count (get-all-scores frequences text-pairs)))))))
+
+(deftest get-min-score-pair-test
+  (testing "If a key is one, then the min score pair is 15."
+    (let [scores (list [0 20.0] [1 15.0] [2 30.0])
+          min-score-pair (get-min-score-pair scores)]
+      (is (= [1 15.0] min-score-pair)))))
+
+(deftest get-key-test
+  (let [plaintext "hello world"
+        ciphertext "khoorczruog"
+        key 3
+        symbols "abcdefghijklmnopqrstuvwxyz "
+        frequences sf/english-letter-frequences]
+    (testing "The plaintext is encrypted with 3 as the key."
+      (is (= key (get-key ciphertext symbols frequences)))))
+
+  (let [; З поеми "Кавказ" Тараса Григоровича Шевченка:
+        plaintext "Борітеся – поборете, Вам Бог помагає! За вас правда, за вас слава. І воля святая!"
+        ciphertext "жхчощйшде–ецхжхчйщй,езєуежхиецхуєиєк!емєезєшецчєзїє,емєезєшештєзє.еоезхтдешздщєд!"
+        key 7
+        symbols "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя "
+        frequences sf/ukrainian-letter-frequences]
+    (testing "The plaintext is encrypted with 3 as the key."
+      (is (= key (get-key ciphertext symbols frequences))))))
+
+(deftest get-plaintext-test
+  (let [plaintext "hello world"
+        ciphertext "khoorczruog"
+        symbols "abcdefghijklmnopqrstuvwxyz "
+        frequences sf/english-letter-frequences]
+    (testing "The plaintext is encrypted with 3 as the key."
+      (is (= plaintext (get-plaintext ciphertext symbols frequences))))
+
+    (testing "The ciphertext is case-insensitive."
+      (is (= plaintext (get-plaintext "KhoorcZruog" symbols frequences)))))
+
+  (let [; З поеми "Кавказ" Тараса Григоровича Шевченка:
+        plaintext "Борітеся – поборете, Вам Бог помагає! За вас правда, за вас слава. І воля святая!"
+        ciphertext "жхчощйшде–ецхжхчйщй,езєуежхиецхуєиєк!емєезєшецчєзїє,емєезєшештєзє.еоезхтдешздщєд!"
+        key 7
+        symbols "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя "
+        frequences sf/ukrainian-letter-frequences]
+    (testing "The ciphertext is case-insensitive."
+      (is (= (cs/lower-case plaintext)
+             (get-plaintext ciphertext symbols frequences))))))
+

test/cipher_analytical_machine/caesar_test.clj

@@ -51,14 +51,20 @@
     (testing "The message 'abc' is encrypted to 'bca'"
       (is (= "bca" (encrypt-message "abc" 1 symbols))))
 
+    (testing "The message 'ABC' is encrypted to 'bca', because of case-insensitive."
+      (is (= "bca" (encrypt-message "ABC" 1 symbols))))
+
     (testing "The message 'bca' is decrypted to 'abc'"
       (is (= "abc" (encrypt-message "bca" -1 symbols))))))
 
 (deftest decrypt-message-text
   (let [symbols "abc"]
     (testing "The message 'bca' is decrypted to 'abc'"
-      (is (= "abc" (decrypt-message "bca" 1 symbols)))
+      (is (= "abc" (decrypt-message "bca" 1 symbols))))
+
+    (testing "The message 'BCA' is decrypted to 'abc', because of case-insensitive."
+      (is (= "abc" (decrypt-message "BCA" 1 symbols))))
 
     (testing "The message 'abc' is encrypted to 'bca'"
-      (is (= "bca" (decrypt-message "abc" -1 symbols)))))))
+      (is (= "bca" (decrypt-message "abc" -1 symbols))))))
 

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/symbol_factories_test.clj

@@ -4,38 +4,22 @@
    [cipher-analytical-machine.symbol-factories :refer :all]
   ))
 
-(deftest lower-english-alphabet-factory-test
-  (let [symbols (lower-english-alphabet-factory)]
+(deftest english-alphabet-factory-test
+  (let [symbols (english-alphabet-factory)]
     (testing "It must have the 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 the 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)]
+(deftest ukrainian-alphabet-factory-test
+  (let [symbols (ukrainian-alphabet-factory)]
     (testing "It must have the 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 the 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 the same order."
@@ -61,15 +45,15 @@
       (is (= 3 (count symbols))))))
 
 (deftest default-symbol-factory-test
-  (testing "If language code is 'en', then it must have symbols."
+  (testing "If language code is 'en', then it must have 27 symbols."
     (let [symbols (default-symbol-factory "en")]
-      (is (= 70 (count symbols)))))
+      (is (= 27 (count symbols)))))
 
-  (testing "If language code is 'uk', then it must have symbols."
+  (testing "If language code is 'uk', then it must have 34 symbols."
     (let [symbols (default-symbol-factory "uk")]
-      (is (= 84 (count symbols)))))
+      (is (= 34 (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))))))
+      (is (= 27 (count symbols))))))