This commit is contained in:
@@ -0,0 +1,38 @@
|
||||
(ns cipher-analytical-machine.analyzers.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 +1,59 @@
|
||||
(ns cipher-analytical-machine.analyzers.caesar
|
||||
(:require [cipher-analytical-machine.ciphers.caesar :as caesar]
|
||||
[clojure.string :as cs]
|
||||
[cipher-analytical-machine.analyzers.analyzers :as ca])
|
||||
(:import [cipher_analytical_machine.ciphers.caesar Decrypted]
|
||||
[cipher_analytical_machine.analyzers.caesar FrequencyAnalyzer])
|
||||
(: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-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)
|
||||
(get-all-scores letter-frequences)
|
||||
(get-min-score-pair)
|
||||
(first)))
|
||||
|
||||
(defn get-plaintext
|
||||
"Return the plaintext from a ciphertext. The function is case-insensitive."
|
||||
[ciphertext symbols letter-frequences]
|
||||
(let [ciphertext (cs/lower-case ciphertext)]
|
||||
(caesar/decrypt-message ciphertext
|
||||
(get-key ciphertext symbols letter-frequences)
|
||||
symbols)))
|
||||
|
||||
(defn frequency-analizer-get-plaintext
|
||||
"Return the plaintext from a ciphertext using simple analizer. The function is case-insensitive."
|
||||
[ciphertext symbols letter-frequences]
|
||||
(let [decrypt (reify Decrypted
|
||||
(decrypt [this message key symbols]
|
||||
(caesar/decrypt-message message key symbols)))]
|
||||
(-> (new FrequencyAnalyzer ciphertext symbols letter-frequences decrypt)
|
||||
.crack)))
|
||||
|
||||
@@ -0,0 +1,21 @@
|
||||
(ns cipher-analytical-machine.analyzers.language
|
||||
(: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)))
|
||||
|
||||
Reference in New Issue
Block a user