tricu

commit 221cef8e919583af6a42b331061a155ad58f8124
parent d89f406026b6e5c9d19938ff59c27b1bb28f2c28
Author: James Eversole <james@eversole.co>
Date:   Tue, 15 Apr 2025 10:52:53 -0500

String escaping using backslash

Diffstat:
M
src/Lexer.hs
 | 
23
+++++++++++++++++------
M
src/Research.hs
 | 
1
-
M
test/Spec.hs
 | 
17
++++++++++++++++-

3 files changed, 33 insertions(+), 8 deletions(-)
diff --git a/src/Lexer.hs b/src/Lexer.hs
@@ -41,7 +41,6 @@ tricuLexer = do
       , try stringLiteral
       , assign
       , colon
-      , backslash
       , openParen
       , closeParen
       , openBracket
@@ -94,9 +93,6 @@ assign = char '=' $> LAssign
 colon :: Lexer LToken
 colon = char ':' $> LColon
 
-backslash :: Lexer LToken
-backslash = char '\\' $> LBackslash
-
 openParen :: Lexer LToken
 openParen = char '(' $> LOpenParen
 
@@ -126,7 +122,22 @@ integerLiteral = do
 stringLiteral :: Lexer LToken
 stringLiteral = do
   char '"'
-  content <- many (noneOf ['"'])
-  char '"' --"
+  content <- manyTill Lexer.charLiteral (char '"')
   return (LStringLiteral content)
 
+charLiteral :: Lexer Char
+charLiteral = escapedChar <|> normalChar
+  where
+    normalChar = noneOf ['"', '\\']
+    escapedChar = do
+      void $ char '\\'
+      c <- oneOf ['n', 't', 'r', 'f', 'b', '\\', '"', '\'']
+      return $ case c of
+        'n'  -> '\n'
+        't'  -> '\t'
+        'r'  -> '\r'
+        'f'  -> '\f'
+        'b'  -> '\b'
+        '\\' -> '\\'
+        '"'  -> '"'
+        '\'' -> '\''
diff --git a/src/Research.hs b/src/Research.hs
@@ -38,7 +38,6 @@ data LToken
   | LAssign
   | LColon
   | LDot
-  | LBackslash
   | LOpenParen
   | LCloseParen
   | LOpenBracket
diff --git a/test/Spec.hs b/test/Spec.hs
@@ -51,7 +51,22 @@ lexer = testGroup "Lexer Tests"
 
   , testCase "Lex escaped characters in strings" $ do
       let input = "\"hello\\nworld\""
-          expect = Right [LStringLiteral "hello\\nworld"]
+          expect = Right [LStringLiteral "hello\nworld"]
+      runParser tricuLexer "" input @?= expect
+
+  , testCase "Lex multiple escaped characters in strings" $ do
+      let input = "\"tab:\\t newline:\\n quote:\\\" backslash:\\\\\""
+          expect = Right [LStringLiteral "tab:\t newline:\n quote:\" backslash:\\"]
+      runParser tricuLexer "" input @?= expect
+
+  , testCase "Lex escaped characters in string literals" $ do
+      let input = "x = \"line1\\nline2\\tindented\""
+          expect = Right [LIdentifier "x", LAssign, LStringLiteral "line1\nline2\tindented"]
+      runParser tricuLexer "" input @?= expect
+
+  , testCase "Lex empty string with escape sequence" $ do
+      let input = "\"\\\"\""
+          expect = Right [LStringLiteral "\""]
       runParser tricuLexer "" input @?= expect
 
   , testCase "Lex mixed literals" $ do