Compare commits

..
2 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
2 changed files with 72 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)))