From bf163e52fe5212bdcaeab97c0a9af1472319b622 Mon Sep 17 00:00:00 2001
From: Jonny <j.sinteix@gmail.com>
Date: Sat, 1 Feb 2025 02:21:39 +0000
Subject: [PATCH] feat: initial lexer implementation

---
 src/main/wacc/lexer.scala | 51 ++++++++++++++++++++++++++++++++++++---
 1 file changed, 48 insertions(+), 3 deletions(-)

diff --git a/src/main/wacc/lexer.scala b/src/main/wacc/lexer.scala
index f633bf6..907053b 100644
--- a/src/main/wacc/lexer.scala
+++ b/src/main/wacc/lexer.scala
@@ -1,16 +1,61 @@
 package wacc
 
 import parsley.Parsley
-import parsley.token.Lexer
+import parsley.token.{Basic, Lexer}
 import parsley.token.descriptions.*
 
 object lexer {
   private val desc = LexicalDesc.plain.copy(
-    // your configuration goes here
+    nameDesc = NameDesc.plain.copy(
+      identifierStart = Basic(c => c.isLetter || c == '_'),
+      identifierLetter = Basic(c => c.isLetterOrDigit || c == '_')
+    ),
+    symbolDesc = SymbolDesc.plain.copy(
+      hardKeywords = Set(
+        "begin", "end", "is", "skip", "if", "then", "else", "fi", "while", "do",
+        "done", "read", "free", "return", "exit", "print", "println", "true",
+        "false", "int", "bool", "char", "string", "pair", "newpair", "fst",
+        "snd", "call", "chr", "ord", "len", "null"
+      ),
+      hardOperators = Set(
+        "+", "-", "*", "/", "%", ">", "<", ">=", "<=", "==", "!=", "&&", "||",
+        "!"
+      )
+    ),
+    spaceDesc = SpaceDesc.plain.copy(
+      lineCommentStart = "#"
+    ),
+    // TODO - See BNF 1.1 and Table 5 2.3.6
+    textDesc = TextDesc.plain.copy(
+      graphicCharacter =
+        Basic(c => c >= ' ' && c != '\\' && c != '\'' && c != '"'),
+      escapeSequences = EscapeDesc.plain.copy(
+        literals = Set('\\', '"', '\''),
+        mapping = Map(
+          "0" -> '\u0000',
+          "b" -> '\b',
+          "t" -> '\t',
+          "n" -> '\n',
+          "f" -> '\f',
+          "r" -> '\r'
+        )
+      )
+    )
   )
+
   private val lexer = Lexer(desc)
 
-  val integer = lexer.lexeme.integer.decimal
+  // Enforce 32-bit signed integer range - see 1.5
+  // TODO By default leadingZerosAllowed = true in NumericDesc - Wacc doesnt specify (I think) but should reach consensus
+  val integer = lexer.lexeme.integer.decimal32[Int]
+
+  // TODO Check if textDesc can handle this
+  val charLit = lexer.lexeme.character.ascii
+
+  // TODO Check if textDesc can handle this
+  val stringLit = lexer.lexeme.string.ascii
+
   val implicits = lexer.lexeme.symbol.implicits
+
   def fully[A](p: Parsley[A]): Parsley[A] = lexer.fully(p)
 }