From 1f78f8b3adb9505b54fc986b756a13c486307a38 Mon Sep 17 00:00:00 2001
From: Sean D Gillespie <sean@mistersg.net>
Date: Sat, 16 Mar 2024 13:58:35 -0400
Subject: [PATCH] refactor(lib): Extract normalizeWord from genTrigraph

---
 src/Data/Gibberish/Trigraph.hs | 20 +++++++++++++++-----
 1 file changed, 15 insertions(+), 5 deletions(-)

diff --git a/src/Data/Gibberish/Trigraph.hs b/src/Data/Gibberish/Trigraph.hs
index 38a4f08..3589e55 100644
--- a/src/Data/Gibberish/Trigraph.hs
+++ b/src/Data/Gibberish/Trigraph.hs
@@ -35,15 +35,11 @@ newtype TrigraphConfig = TrigraphConfig
 
 -- | Generate trigraphs from a list of words
 genTrigraph :: [Text] -> Trigraph
-genTrigraph = Trigraph . foldr (foldWord . transform) Map.empty
+genTrigraph = Trigraph . foldr (foldWord . normalizeWord) Map.empty
   where
     foldWord = Map.unionWith combine . mkTrigraph
     combine (Frequencies f1) (Frequencies f2) = Frequencies $ Map.unionWith (+) f1 f2
 
-    transform word' =
-      Text.map (toQwertyKey . toLower) $
-        Text.filter (not . isPunctuation) word'
-
 -- | Generate a trigraph from a single word
 mkTrigraph :: Text -> Map Digram Frequencies
 mkTrigraph word = foldr insert' Map.empty $ scanTrigrams word
@@ -54,6 +50,20 @@ mkTrigraph word = foldr insert' Map.empty $ scanTrigrams word
       Frequencies (Map.unionWith (+) m1 m2)
     mkFrequencies c = Frequencies $ Map.singleton (Unigram c) 1
 
+-- | Normalize a word before calculating the trigraph:
+--
+--  1. Remove punctuation (quotes, dashes, and so on)
+--  2. Lower case all letters
+--  3. Translate non-qwerty chars to qwerty keys (eg, à -> a)
+normalizeWord :: Text -> Text
+normalizeWord = Text.map transformChar . Text.filter filterChar
+  where
+    transformChar :: Char -> Char
+    transformChar = toQwertyKey . toLower
+
+    filterChar :: Char -> Bool
+    filterChar = not . isPunctuation
+
 scanTrigrams :: Text -> [Trigram]
 scanTrigrams word = case Text.take 3 word of
   [a, b, c] -> Trigram a b c : scanTrigrams (Text.tail word)