Compare commits

..
3 Commits
Author SHA1 Message Date
KKlochko a7ccad8e1e Add the caesar-analyzers to get the plaintext from a ciphertext.
continuous-integration/drone/push Build is failing
2023-09-15 21:30:37 +03:00
KKlochko 79e2957402 Add Chi-squared Statistic function to analyze texts.
continuous-integration/drone/push Build is failing
2023-09-14 21:09:28 +03:00
KKlochko 4750242a1d Add the frequences of symbols as defaults. 2023-09-14 20:50:28 +03:00
3 changed files with 141 additions and 0 deletions
@@ -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))
@@ -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)))
@@ -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
0.046
\г 0.013
0.0001
0.027
\е 0.042
0.005
0.007
0.020
0.055
\і 0.044
0.010
0.009
0.033
0.027
0.029
0.068
\о 0.086
\п 0.025
\р 0.043
\с 0.037
0.045
\у 0.027
0.003
\х 0.011
0.010
0.011
0.005
0.004
0.016
0.008
0.019
\ 0.138
})