From 8f4788ca27c670357aed239b5acbf555e5e37e7f Mon Sep 17 00:00:00 2001
From: Yuichi Nishiwaki <yuichi.nishiwaki@gmail.com>
Date: Mon, 21 Oct 2013 10:48:03 +0900
Subject: [PATCH] read strings

---
 include/picrin.h        |  2 ++
 include/picrin/string.h |  2 --
 src/codegen.c           |  9 +++++++++
 src/parse.y             | 26 ++++++--------------------
 src/scan.l              | 31 +++++++++++++++++++++++++++++++
 5 files changed, 48 insertions(+), 22 deletions(-)

diff --git a/include/picrin.h b/include/picrin.h
index 12a01233..3cf3fb36 100644
--- a/include/picrin.h
+++ b/include/picrin.h
@@ -66,6 +66,8 @@ bool pic_eq_p(pic_state *, pic_value, pic_value);
 
 pic_value pic_intern_cstr(pic_state *, const char *);
 
+pic_value pic_str_new_cstr(pic_state *, const char *);
+
 bool pic_parse(pic_state *, const char *, pic_value *);
 
 pic_value pic_eval(pic_state *, pic_value, struct pic_env *);
diff --git a/include/picrin/string.h b/include/picrin/string.h
index ccef6efb..f4a1d39f 100644
--- a/include/picrin/string.h
+++ b/include/picrin/string.h
@@ -9,6 +9,4 @@ struct pic_string {
 
 #define pic_str_ptr(v) ((struct pic_string *)v.u.data)
 
-pic_value pic_str_new_cstr(pic_state *, const char *);
-
 #endif
diff --git a/src/codegen.c b/src/codegen.c
index 59a341e6..5e82cc6d 100644
--- a/src/codegen.c
+++ b/src/codegen.c
@@ -374,6 +374,15 @@ pic_gen(pic_state *pic, struct pic_irep *irep, pic_value obj, struct pic_env *en
     irep->clen++;
     break;
   }
+  case PIC_TT_STRING: {
+    int pidx;
+    pidx = pic->plen++;
+    pic->pool[pidx] = obj;
+    irep->code[irep->clen].insn = OP_PUSHCONST;
+    irep->code[irep->clen].u.i = pidx;
+    irep->clen++;
+    break;
+  }
   case PIC_TT_PROC:
   case PIC_TT_UNDEF:
   case PIC_TT_PORT: {
diff --git a/src/parse.y b/src/parse.y
index 0a76e282..885d055d 100644
--- a/src/parse.y
+++ b/src/parse.y
@@ -29,10 +29,10 @@ int yylex(struct parser_control *);
 
 %token tLPAREN tRPAREN tDOT
 %token tQUOTE
-%token <datum> tSYMBOL tNUMBER tBOOLEAN
+%token <datum> tSYMBOL tNUMBER tBOOLEAN tSTRING
 
-%type <datum> datum simple_datum symbol compound_datum abbrev
-%type <datum> number boolean list list_data
+%type <datum> datum simple_datum compound_datum abbrev
+%type <datum> list list_data
 
 %%
 
@@ -58,24 +58,10 @@ datum
 ;
 
 simple_datum
-	: symbol
-	| number
-	| boolean
-;
-
-symbol
 	: tSYMBOL
-	{
-	  $$ = $1;
-	}
-;
-
-number
-	: tNUMBER
-;
-
-boolean
-	: tBOOLEAN
+	| tNUMBER
+	| tBOOLEAN
+	| tSTRING
 ;
 
 compound_datum
diff --git a/src/scan.l b/src/scan.l
index 62cf0d2a..f23ebdc3 100644
--- a/src/scan.l
+++ b/src/scan.l
@@ -1,5 +1,6 @@
 %{
 #include <stdlib.h>
+#include <string.h>
 
 #include "picrin.h"
 #include "y.tab.h"
@@ -10,6 +11,8 @@ struct parser_control {
   bool incomp;
 };
 
+static pic_value escape_string(pic_state *, const char *);
+
 #define YY_DECL int yylex (struct parser_control *p)
 %}
 
@@ -29,6 +32,10 @@ ureal		{digit}+|\.{digit}+|{digit}+\.{digit}*
 sign		[+-]?
 infnan		"+inf.0"|"-inf.0"|"+nan.0"|"-nan.0"
 
+  /* string */
+string		\"{string_elem}*\"
+string_elem	[^\"\\]|"\\\""|"\\\\"
+
 %%
 
 [ \t\n\r]	/* skip whitespace */
@@ -39,6 +46,10 @@ infnan		"+inf.0"|"-inf.0"|"+nan.0"|"-nan.0"
 {boolean}	{ yylval.datum = pic_bool_value(strcmp(yytext, "#t") == 0 || strcmp(yytext, "#true") == 0); return tBOOLEAN; }
 {real} 		{ yylval.datum = pic_float_value(atof(yytext)); return tNUMBER; }
 {identifier}	{ yylval.datum = pic_intern_cstr(p->pic, yytext); return tSYMBOL; }
+{string}	{
+		  yylval.datum = escape_string(p->pic, yytext);
+		  return tSTRING;
+		}
 
 %%
 
@@ -48,6 +59,26 @@ yywrap()
   return 1;
 }
 
+static pic_value
+escape_string(pic_state *pic, const char *str)
+{
+  size_t len;
+  char *new_str;
+  int i;
+  pic_value v;
+
+  len = strlen(str);
+  new_str = (char *)pic_alloc(pic, len + 1);
+  for (i = 1; i < len - 1; ++i) {
+    new_str[i-1] = str[i];
+  }
+  new_str[i] = '\0';
+
+  v = pic_str_new_cstr(pic, new_str);
+  pic_free(pic, new_str);
+  return v;
+}
+
 void
 init_scanner(const char *str)
 {