diff options
Diffstat (limited to 'src/utils/FrequencyMap.scala')
| -rw-r--r-- | src/utils/FrequencyMap.scala | 29 |
1 files changed, 15 insertions, 14 deletions
diff --git a/src/utils/FrequencyMap.scala b/src/utils/FrequencyMap.scala index 0bea25c..fffdabe 100644 --- a/src/utils/FrequencyMap.scala +++ b/src/utils/FrequencyMap.scala @@ -1,6 +1,6 @@ package ixee.cryptopals.utils -import ConversionUtils.tup +import FunctionUtils.tup class FrequencyMap(mapargs: Map[Char, Double]) { sealed trait DiffResult @@ -14,7 +14,7 @@ class FrequencyMap(mapargs: Map[Char, Double]) { if (c.isLower) mappings.get(c) else // TODO remove the bias here - mappings.get(c.toLower).map(_ * 0.002) // bias HARD against uppercase spam. should make this a per-map setting. + mappings.get(c.toLower).map(_ * 0.0002) // bias HARD against uppercase spam. should make this a per-map setting. def diff(other: FrequencyMap): Double = { // pseudoscience here @@ -36,29 +36,30 @@ class FrequencyMap(mapargs: Map[Char, Double]) { // TODO: don't hardcode these.. def diffAt(c: Char, charFreq: Double) = - c match { - case ' ' => 0 - case '{' | '}' | '`' | '|' => + Math.pow(c match { + case ' ' => + squared(0.10 - charFreq) + case '{' | '}' | '`' | '|' | '^' => squared(0.000001 - charFreq) // { } | and ` are very unlikely irl case '[' | ']' => - squared(0.000002 - charFreq) // [ ] are more likely - case '"' | '\''=> + squared(0.0000015 - charFreq) // [ ] are more likely + case '"' | '\'' => squared(0.00001 - charFreq) // " or ' are 100% unlikely 90% of the time case '~' | '+' | '=' | '<' | '>' | '/' | '\\' => - squared(0.000004 - charFreq) // math is weird kids + squared(0.00000175 - charFreq) // math is weird kids case ';' | ':' | '-' | '*' | '(' | '&' | ')' | '_' => - squared(0.000009 - charFreq) // getting into more common punctuation + squared(0.000003 - charFreq) // getting into more common punctuation case '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9' => - squared(0.00001 - charFreq) // numbers are KINDA uncommon + squared(0.000002 - charFreq) // numbers are KINDA uncommon case '$' | '%' | '#' | '@' | '!' => - squared(0.00003 - charFreq) // more punctuation... + squared(0.00002 - charFreq) // more punctuation... case '.' | ',' => squared(0.00007 - charFreq) // and the last of the punctuations case '\n' | '\r' => - squared(0.0000002 - charFreq) // explicit \r \n is rare in freeform text. + squared(0.000001 - charFreq) // explicit \r \n is rare in freeform text. case _ => - this.at(c).map(_ - charFreq).map(squared).getOrElse(0.0) - } + this.at(c).map(_ - charFreq).map(squared).getOrElse(0.4) + }, 0.5) def squared(x: Double) = x * x |