Compare commits

...
3 Commits
Author SHA1 Message Date
KKlochko 4607f7f013 Update Caesar Analyzers to be case-insensitive and testable.
continuous-integration/drone/push Build is passing
2023-09-22 21:04:11 +03:00
KKlochko 79e316c51b Update the Caesar to be case-insensitive and skip unknown symbols.
continuous-integration/drone/push Build is passing
2023-09-21 21:05:22 +03:00
KKlochko 20c1d1ef1b Update symbol defaults, factories to have only lower alphabets.
continuous-integration/drone/push Build is passing
2023-09-21 21:01:52 +03:00
6 changed files with 135 additions and 64 deletions
+13 -6
View File
@@ -14,21 +14,28 @@
(+ key) (+ key)
(mod max-index))) (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 (defn encrypt-char
"Calculates new index of a character. Uses a char from symbols, the key and max-index (exclusive)." "Calculates new index of a character. Uses a char from symbols, the key and max-index (exclusive)."
[char key symbols max-index] [char key symbols max-index]
(get symbols
(-> char (-> char
(calculate-char-index symbols) (calculate-char-index symbols)
(encrypt-index key max-index)))) (encrypt-or-skip char key symbols max-index)))
(defn encrypt-message (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] [message key symbols]
(let [max-index (count symbols)] (let [max-index (count symbols)]
(cs/join (->> message
(map (fn [char] (encrypt-char char key symbols max-index)) (cs/lower-case)
message)))) (map (fn [char] (encrypt-char char key symbols max-index)))
(cs/join))))
(defn decrypt-message (defn decrypt-message
"Decrypt the ciphtext using the Caesar cipher." "Decrypt the ciphtext using the Caesar cipher."
@@ -1,5 +1,6 @@
(ns cipher-analytical-machine.caesar-analyzers (ns cipher-analytical-machine.caesar-analyzers
(:require [cipher-analytical-machine.caesar :as caesar] (:require [cipher-analytical-machine.caesar :as caesar]
[clojure.string :as cs]
[cipher-analytical-machine.cipher-analyzers :as ca]) [cipher-analytical-machine.cipher-analyzers :as ca])
(:gen-class)) (:gen-class))
@@ -12,23 +13,36 @@
(conj acc [key (caesar/decrypt-message ciphertext key symbols)])) (conj acc [key (caesar/decrypt-message ciphertext key symbols)]))
'() keys))) '() keys)))
(defn get-key (defn get-all-scores
"To find the key with frequencies of letters." "For pairs (key, plaintext) finds scores and return list of pairs [key, score]."
[ciphertext symbols letter-frequences] [letter-frequences pairs]
(->> (get-all-texts ciphertext symbols)
(map (fn [[key text]] (map (fn [[key text]]
[key (ca/chi-squared-statistic text letter-frequences)])) [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 (reduce
(fn [[key value] [new-key new-value]] (fn [[key value] [new-key new-value]]
(if (> value new-value) [new-key new-value] (if (> value new-value) [new-key new-value]
[key value])) [key value]))
[0 Double/MAX_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)
(get-all-scores letter-frequences)
(get-min-score-pair)
(first))) (first)))
(defn get-plaintext (defn get-plaintext
"Return the plaintext from a ciphertext." "Return the plaintext from a ciphertext. The function is case-insensitive."
[ciphertext symbols letter-frequences] [ciphertext symbols letter-frequences]
(let [ciphertext (cs/lower-case ciphertext)]
(caesar/decrypt-message ciphertext (caesar/decrypt-message ciphertext
(get-key ciphertext symbols letter-frequences) (get-key ciphertext symbols letter-frequences)
symbols)) symbols)))
@@ -2,7 +2,7 @@
(:require [clojure.string :as cs]) (:require [clojure.string :as cs])
(:gen-class)) (:gen-class))
(defn lower-english-alphabet-factory (defn english-alphabet-factory
"Return the English alphabet with only lowercase letters" "Return the English alphabet with only lowercase letters"
[] []
(reduce (reduce
@@ -10,23 +10,11 @@
"" ""
(range (int \a) (inc (int \z))))) (range (int \a) (inc (int \z)))))
(defn english-alphabet-factory (defn ukrainian-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" "Return the Ukrainian alphabet with only lowercase letters"
[] []
"абвгґдеєжзиіїйклмнопрстуфхцчшщьюя") "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 (defn digit-set-factory
"Return digits" "Return digits"
[] []
@@ -48,14 +36,10 @@
(cond (cond
(= language-code "en") (= language-code "en")
(str (english-alphabet-factory) (str (english-alphabet-factory)
(punctuation-symbol-factory)
(digit-set-factory)
\ ) \ )
(= language-code "uk") (= language-code "uk")
(str (ukrainian-alphabet-factory) (str (ukrainian-alphabet-factory)
(punctuation-symbol-factory)
(digit-set-factory)
\ ) \ )
:else :else
@@ -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))))))
@@ -51,14 +51,20 @@
(testing "The message 'abc' is encrypted to 'bca'" (testing "The message 'abc' is encrypted to 'bca'"
(is (= "bca" (encrypt-message "abc" 1 symbols)))) (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'" (testing "The message 'bca' is decrypted to 'abc'"
(is (= "abc" (encrypt-message "bca" -1 symbols)))))) (is (= "abc" (encrypt-message "bca" -1 symbols))))))
(deftest decrypt-message-text (deftest decrypt-message-text
(let [symbols "abc"] (let [symbols "abc"]
(testing "The message 'bca' is decrypted to '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'" (testing "The message 'abc' is encrypted to 'bca'"
(is (= "bca" (decrypt-message "abc" -1 symbols))))))) (is (= "bca" (decrypt-message "abc" -1 symbols))))))
@@ -4,38 +4,22 @@
[cipher-analytical-machine.symbol-factories :refer :all] [cipher-analytical-machine.symbol-factories :refer :all]
)) ))
(deftest lower-english-alphabet-factory-test (deftest english-alphabet-factory-test
(let [symbols (lower-english-alphabet-factory)] (let [symbols (english-alphabet-factory)]
(testing "It must have the same order." (testing "It must have the same order."
(is (= "abcdefghijklmnopqrstuvwxyz" symbols))) (is (= "abcdefghijklmnopqrstuvwxyz" symbols)))
(testing "It must have 26 letters." (testing "It must have 26 letters."
(is (= 26 (count symbols)))))) (is (= 26 (count symbols))))))
(deftest english-alphabet-factory-test (deftest ukrainian-alphabet-factory-test
(let [symbols (english-alphabet-factory)] (let [symbols (ukrainian-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)]
(testing "It must have the same order." (testing "It must have the same order."
(is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя" symbols))) (is (= "абвгґдеєжзиіїйклмнопрстуфхцчшщьюя" symbols)))
(testing "It must have 33 letters." (testing "It must have 33 letters."
(is (= 33 (count symbols)))))) (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 (deftest digit-set-factory-test
(let [symbols (digit-set-factory)] (let [symbols (digit-set-factory)]
(testing "It must have the same order." (testing "It must have the same order."
@@ -61,15 +45,15 @@
(is (= 3 (count symbols)))))) (is (= 3 (count symbols))))))
(deftest default-symbol-factory-test (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")] (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")] (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'." (testing "If language code is unidentified, then it must have the 'en'."
(let [symbols (default-symbol-factory "Maybe a code")] (let [symbols (default-symbol-factory "Maybe a code")]
(is (= 70 (count symbols)))))) (is (= 27 (count symbols))))))