Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
2cc5be57c8 | ||
|
|
ddfd5c473d |
+2
-1
@@ -7,7 +7,8 @@
|
||||
[org.clojure/tools.cli "1.0.219"]
|
||||
[org.apache.tika/tika-core "1.28.5"]
|
||||
[org.apache.tika/tika-langdetect "1.28.5"]
|
||||
[cheshire "5.12.0"]]
|
||||
[cheshire "5.12.0"]
|
||||
[org.clojure/math.combinatorics "0.2.0"]]
|
||||
:main ^:skip-aot cipher-analytical-machine.core
|
||||
:java-source-paths ["src/main/java"]
|
||||
:target-path "target/%s"
|
||||
|
||||
@@ -0,0 +1,22 @@
|
||||
(ns cipher-analytical-machine.analyzers.simple-substitution
|
||||
(:require [clojure.string :as cs]
|
||||
[clojure.math.combinatorics :as comb]
|
||||
[cipher-analytical-machine.ciphers.simple-substitution :as ss]
|
||||
[cipher-analytical-machine.analyzers.analyzers :as analyzers]
|
||||
[cipher-analytical-machine.analyzers.caesar :as caesar])
|
||||
(:gen-class))
|
||||
|
||||
(defn get-possible-key-vector
|
||||
"Conver a possible key map to a vector of possible keys. For example, if you have a map: {\\a [\\a \\b \\c] \\b [\\b \\c] \\c [\\a \\c]}, then the vector is [[\\a \\b \\c] [\\b \\c] [\\a \\c]]."
|
||||
[possible-key-map symbols]
|
||||
(reduce
|
||||
(fn [acc el]
|
||||
(conj acc (get possible-key-map el)))
|
||||
[] symbols))
|
||||
|
||||
(defn get-possible-keys
|
||||
"Return possible keys for a possible key vector. For example, if you have a map: {\\a [\\a \\b \\c] \\b [\\b \\c] \\c [\\a \\c]}, then the vector is [[\\a \\b \\c] [\\b \\c] [\\a \\c]] and keys (\"aba\" \"abc\" ...)."
|
||||
[possible-key-vector]
|
||||
(->> (apply comb/cartesian-product possible-key-vector)
|
||||
(map cs/join)))
|
||||
|
||||
@@ -9,7 +9,8 @@
|
||||
(let [analyzer (:analyzer options)
|
||||
message (:message options)
|
||||
symbols (:symbols options)
|
||||
frequencies symbol-frequencies/english-letter-frequencies]
|
||||
frequencies (-> (get options :language "en")
|
||||
(symbol-frequencies/default-frequency-factory))]
|
||||
(cond
|
||||
(= analyzer "Chi^2")
|
||||
(caesar-analyzers/get-plaintext message symbols frequencies)
|
||||
|
||||
@@ -77,3 +77,16 @@
|
||||
\ 0.138
|
||||
})
|
||||
|
||||
(defn default-frequency-factory
|
||||
"Returns a default map of letter frequencies for a language"
|
||||
[language-code]
|
||||
(cond
|
||||
(= language-code "en")
|
||||
english-letter-frequencies
|
||||
|
||||
(= language-code "uk")
|
||||
ukrainian-letter-frequencies
|
||||
|
||||
:else
|
||||
english-letter-frequencies))
|
||||
|
||||
|
||||
@@ -0,0 +1,23 @@
|
||||
(ns cipher-analytical-machine.analyzers.simple-substitution-test
|
||||
(:require
|
||||
[clojure.test :refer :all]
|
||||
[cipher-analytical-machine.symbols.frequencies :as sf]
|
||||
[cipher-analytical-machine.ciphers.simple-substitution :as ss]
|
||||
[cipher-analytical-machine.analyzers.simple-substitution :refer :all])
|
||||
(:gen-class))
|
||||
|
||||
(deftest get-possible-keys-vector-test
|
||||
(testing "The vector must be ordered by symbols"
|
||||
(are [possible-keys-map symbols expected-key-vector]
|
||||
(= expected-key-vector (get-possible-key-vector possible-keys-map symbols))
|
||||
{\a [\a \b] \b [\b \a] \c [\a \c]} "abc" [[\a \b] [\b \a] [\a \c]]
|
||||
{\a [\a \b] \c [\b \a] \b [\a \c]} "abc" [[\a \b] [\a \c] [\b \a]])))
|
||||
|
||||
(deftest get-possible-keys-test
|
||||
(testing "The keys must be the all combinations"
|
||||
(are [possible-keys-vector expected-keys]
|
||||
(= expected-keys (get-possible-keys possible-keys-vector))
|
||||
[[\a \b] [\f] [\a \c] [\g]] '("afag" "afcg" "bfag" "bfcg")
|
||||
[[\a \b] [\a \c]] '("aa" "ac" "ba" "bc")
|
||||
[[\a \b \c] [\a \c] [\b \c]] '("aab" "aac" "acb" "acc" "bab" "bac" "bcb" "bcc" "cab" "cac" "ccb" "ccc"))))
|
||||
|
||||
@@ -9,3 +9,13 @@
|
||||
(is (= "etaoinsrhldcumfpgwybvkxjqz"
|
||||
(map-to-string symbol-map))))))
|
||||
|
||||
(deftest default-frequency-factory-test
|
||||
(testing "Factory supports languages: English (en), Ukrainian (uk). If language code is unidentified, then it must have the 'en'."
|
||||
(are [language-code expected-map-size]
|
||||
(->> (default-frequency-factory language-code)
|
||||
count
|
||||
(= expected-map-size))
|
||||
"en" 26
|
||||
"uk" 34
|
||||
"Maybe a code" 26)))
|
||||
|
||||
|
||||
Reference in New Issue
Block a user