From 5dab7df6830018c5c28ebcc7cc3b242ccad2736b Mon Sep 17 00:00:00 2001
From: Teddy <ted.sybil@gmail.com>
Date: Mon, 5 Aug 2013 23:11:39 +0800
Subject: added string quoting support in the parser

---
 parser.cpp | 73 ++++++++++++++++++++++++++++++++++++++++++++++----------------
 1 file changed, 55 insertions(+), 18 deletions(-)

(limited to 'parser.cpp')

diff --git a/parser.cpp b/parser.cpp
index 499b832..cb8695f 100644
--- a/parser.cpp
+++ b/parser.cpp
@@ -12,7 +12,7 @@ static char buff[TOKEN_BUFF_SIZE];
 static EvalObj *parse_stack[PARSE_STACK_SIZE];
 extern Cons *empty_list;
 
-Tokenizor::Tokenizor() : stream(stdin), buff_ptr(buff) {}
+Tokenizor::Tokenizor() : stream(stdin), buff_ptr(buff), escaping(false) {}
 void Tokenizor::set_stream(FILE *_stream) {
     stream = _stream;
 }
@@ -27,37 +27,74 @@ void Tokenizor::set_stream(FILE *_stream) {
     (IS_BRACKET(ch) || IS_SPACE(ch))
 #define IS_COMMENT(ch) \
     ((ch) == ';')
+#define IS_QUOTE(ch) \
+    ((ch) == '\"')
+#define IS_SLASH(ch) \
+    ((ch) == '\\')
 
 #define POP \
-do { \
-    *buff_ptr = 0; \
-    ret = string(buff); \
-    buff_ptr = buff; \
-} while (0)
+    do { \
+        *buff_ptr = 0; \
+        ret = string(buff); \
+        buff_ptr = buff; \
+    } while (0)
 
 bool Tokenizor::get_token(string &ret) {
     char ch;
     bool flag = false;
     while (fread(&ch, 1, 1, stream))
     {
-        if (buff_ptr != buff && 
-            (IS_BRACKET(*buff) || 
-             IS_DELIMITER(ch) ||
-             IS_COMMENT(ch)))
+        if (escaping)
         {
-            if (IS_COMMENT(*buff))
+            escaping = false;
+            switch (ch)
             {
-                if (IS_NEWLINE(ch)) buff_ptr = buff;
-                else buff_ptr = buff + 1;
+                case '\\': *buff_ptr++ = '\\'; break;
+                case '\"': *buff_ptr++ = '\"'; break;
+                case 'n': *buff_ptr++ = '\n'; break;
+                case 't': *buff_ptr++ = '\t'; break;
+                default: {
+                             buff_ptr = buff;
+                             throw TokenError(string("") + ch, 
+                                     PAR_ERR_ILLEGAL_CHAR_IN_ESC);
+                         }
             }
-            else
+        }
+        else
+        {
+            bool in_quote = buff_ptr != buff && IS_QUOTE(*buff);
+            if (buff_ptr != buff && 
+                    (IS_BRACKET(*buff) || 
+                     IS_DELIMITER(ch) ||
+                     IS_COMMENT(ch) ||
+                     IS_QUOTE(ch)))
+            {
+                if (IS_COMMENT(*buff))
+                {
+                    if (IS_NEWLINE(ch)) buff_ptr = buff;
+                    else buff_ptr = buff + 1;
+                }
+                else if (!in_quote)
+                {
+                    POP;
+                    flag = true;
+                }
+                else if (IS_QUOTE(ch))
+                {
+                    *buff_ptr++ = '\"';
+                    POP;
+                    return true;    // discard current slash
+                }
+            }
+            if (in_quote || !IS_SPACE(ch)) 
             {
-                POP;
-                flag = true;
+                if (in_quote && IS_SLASH(ch))
+                    escaping = true;
+                else 
+                    *buff_ptr++ = ch;
             }
+            if (flag) return true;
         }
-        if (!IS_SPACE(ch)) *buff_ptr++ = ch;
-        if (flag) return true;
     }
     if (buff_ptr != buff) POP;
     return false; // can't read more 
-- 
cgit v1.2.3