From f534f3e753182f37ec8ffbe1953eacbecb65abb3 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop+git@gmail.com>
Date: Tue, 30 Aug 2011 13:59:15 -0400
Subject: [PATCH] properly lex a string with escaped characters

---
 parser/lexrules.py | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/parser/lexrules.py b/parser/lexrules.py
index 686c9cc..4334a36 100644
--- a/parser/lexrules.py
+++ b/parser/lexrules.py
@@ -11,6 +11,10 @@ tokens = [
         'ID',
         ] + list(reserved.values())
 
+states = (
+        ('string', 'exclusive'),
+        )
+
 t_LPAREN = r'\('
 t_RPAREN = r'\)'
 t_SEMICOLON = r';'
@@ -23,8 +27,31 @@ def t_ID(t):
     return t
 
 def t_STRING(t):
-    r'"([^"])*"'
-    t.value = t.value[1:-1]
+    r'\"'
+    t.lexer.str_buildup = ''
+    t.lexer.begin('string')
+
+def t_string_1(t):
+    r'[^\\\n\"]'
+    t.lexer.str_buildup += t.value
+
+def t_string_2(t):
+    r'\\[^\n]'
+    c = {
+            't': '\t',
+            'r': '\r',
+            'n': '\n',
+            'f': '\f',
+            'b': '\b',
+            'v': '\v',
+        }.get(t.value[1], t.value[1])
+    t.lexer.str_buildup += c
+
+def t_string_3(t):
+    r'\"'
+    t.type = 'STRING'
+    t.value = t.lexer.str_buildup
+    t.lexer.begin('INITIAL')
     return t
 
 def t_newline(t):