check for a small parser bug, too

[swftools.git] / lib / as3 / tokenizer.lex

diff --git a/lib/as3/tokenizer.lex b/lib/as3/tokenizer.lex
index c06b2be..c84e3b3 100644 (file)
--- a/lib/as3/tokenizer.lex
+++ b/lib/as3/tokenizer.lex
@@ -32,8 +32,9 @@
 #include "files.h"
 
 int as3_pass = 0;
-
 int as3_verbosity = 1;
+unsigned int as3_tokencount = 0;
+
 void as3_error(const char*format, ...)
 {
     char buf[1024];
@@ -94,12 +95,43 @@ static void dbg(const char*format, ...)
     fflush(stdout);
 }
 
-
-
 #ifndef YY_CURRENT_BUFFER
 #define YY_CURRENT_BUFFER yy_current_buffer
 #endif
 
+static void*as3_buffer=0;
+static int as3_buffer_pos=0;
+static int as3_buffer_len=0;
+void as3_file_input(FILE*fi)
+{
+    as3_in = fi;
+    as3_buffer = 0;
+}
+void as3_buffer_input(void*buffer, int len)
+{
+    as3_buffer = buffer;
+    as3_buffer_len = len;
+    as3_buffer_pos = 0;
+    as3_in = 0;
+}
+
+#define YY_INPUT(buf,result,max_size) { \
+  if(!as3_buffer) { \
+      errno=0; \
+      while((result = fread(buf, 1, max_size, as3_in))==0 && ferror(as3_in)) \
+      { if(errno != EINTR) {YY_FATAL_ERROR("input in flex scanner failed"); break;} \
+        errno=0; clearerr(as3_in); \
+      } \
+  } else { \
+      int to_read = max_size; \
+      if(to_read + as3_buffer_pos > as3_buffer_len) \
+          to_read = as3_buffer_len - as3_buffer_pos; \
+      memcpy(buf, as3_buffer+as3_buffer_pos, to_read); \
+      as3_buffer_pos += to_read; \
+      result=to_read; \
+  } \
+}
+
 void handleInclude(char*text, int len, char quotes)
 {
     char*filename = 0;
@@ -122,7 +154,8 @@ void handleInclude(char*text, int len, char quotes)
         filename = strdup(&text[i1]);
     }
     
-    char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
+    char*fullfilename = find_file(filename);
+    enter_file2(filename, fullfilename, YY_CURRENT_BUFFER);
     yyin = fopen(fullfilename, "rb");
     if (!yyin) {
        syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
@@ -244,8 +277,21 @@ static string_t string_unescape(const char*in, int l)
     return out; 
 }
 
+static void handleCData(char*s, int len)
+{
+    a3_lval.str.str = s+9;    // <![CDATA[
+    a3_lval.str.len = len-9-3;// ]]>
+    a3_lval.str.str = strdup_n(a3_lval.str.str, a3_lval.str.len);
+}
+
 static void handleString(char*s, int len)
 {
+    if(as3_pass < 2) {
+        // don't bother decoding strings in pass 1
+        memset(&a3_lval, 0, sizeof(a3_lval));
+        return;
+    }
+
     if(s[0]=='"') {
         if(s[len-1]!='"') syntaxerror("String doesn't end with '\"'");
         s++;len-=2;
@@ -257,24 +303,24 @@ static void handleString(char*s, int len)
     else syntaxerror("String incorrectly terminated");
 
     
-    avm2_lval.str = string_unescape(s, len);
+    a3_lval.str = string_unescape(s, len);
 }
 
 
 char start_of_expression;
 
-static inline int mkid(int type)
+static inline int handleIdentifier(int type)
 {
     char*s = malloc(yyleng+1);
     memcpy(s, yytext, yyleng);
     s[yyleng]=0;
-    avm2_lval.id = s;
+    a3_lval.id = s;
     return type;
 }
 
 static inline int m(int type)
 {
-    avm2_lval.token = type;
+    a3_lval.token = type;
     return type;
 }
 
@@ -292,7 +338,7 @@ static char*nrbuf()
 
 static inline int setint(int v)
 {
-    avm2_lval.number_int = v;
+    a3_lval.number_int = v;
     if(v>-128)
         return T_BYTE;
     else if(v>=-32768)
@@ -302,7 +348,7 @@ static inline int setint(int v)
 }
 static inline int setuint(unsigned int v)
 {
-    avm2_lval.number_uint = v;
+    a3_lval.number_uint = v;
     if(v<128)
         return T_BYTE;
     else if(v<32768)
@@ -312,14 +358,14 @@ static inline int setuint(unsigned int v)
 }
 static inline int setfloat(double v)
 {
-    avm2_lval.number_float = v;
+    a3_lval.number_float = v;
     return T_FLOAT;
 }
 
 static inline int handlefloat()
 {
     char*s = nrbuf();
-    avm2_lval.number_float = atof(s);
+    a3_lval.number_float = atof(s);
     return T_FLOAT;
 }
 
@@ -432,7 +478,7 @@ void handleLabel(char*text, int len)
     char*s = malloc(t+1);
     memcpy(s, yytext, t);
     s[t]=0;
-    avm2_lval.id = s;
+    a3_lval.id = s;
 }
 
 static int handleregexp()
@@ -448,11 +494,11 @@ static int handleregexp()
             break;
         }
     }
-    avm2_lval.regexp.pattern = s;
+    a3_lval.regexp.pattern = s;
     if(t==len) {
-        avm2_lval.regexp.options = 0;
+        a3_lval.regexp.options = 0;
     } else {
-        avm2_lval.regexp.options = s+t+1;
+        a3_lval.regexp.options = s+t+1;
     }
     return T_REGEXP;
 }
@@ -482,7 +528,6 @@ static inline void c() {
 //uint                         {c();return m(KW_UINT);}
 //Number                       {c();return m(KW_NUMBER);}
 
-
 %}
 
 %s REGEXPOK
@@ -501,6 +546,7 @@ HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
 INTWITHSIGN [+-]?({INT})
 FLOATWITHSIGN [+-]?({FLOAT})
 
+CDATA    <!\[CDATA\[([^]]|\][^]]|\]\][^>])*\]*\]\]\>
 STRING   ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
 S       [ \n\r\t]
 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
@@ -516,11 +562,12 @@ REGEXP   [/]([^/\n]|\\[/])*[/][a-zA-Z]*
 ^include{S}+{STRING}{S}*/\n    {l();handleInclude(yytext, yyleng, 1);}
 ^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n    {l();handleInclude(yytext, yyleng, 0);}
 {STRING}                     {l(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
+{CDATA}                      {l(); BEGIN(INITIAL);handleCData(yytext, yyleng);return T_STRING;}
 
 <BEGINNING,REGEXPOK>{
 {REGEXP}                     {c(); BEGIN(INITIAL);return handleregexp();} 
 {HEXWITHSIGN}                {c(); BEGIN(INITIAL);return handlehex();}
-{HEXFLOATWITHSIGN}                {c(); BEGIN(INITIAL);return handlehexfloat();}
+{HEXFLOATWITHSIGN}           {c(); BEGIN(INITIAL);return handlehexfloat();}
 {INTWITHSIGN}                {c(); BEGIN(INITIAL);return handleint();}
 {FLOATWITHSIGN}              {c(); BEGIN(INITIAL);return handlefloat();}
 }
@@ -540,10 +587,10 @@ REGEXP   [/]([^/\n]|\\[/])*[/][a-zA-Z]*
 {NAME}{S}*:{S}*do/{_}         {l();handleLabel(yytext, yyleng-2);return T_DO;}
 {NAME}{S}*:{S}*while/{_}      {l();handleLabel(yytext, yyleng-5);return T_WHILE;}
 {NAME}{S}*:{S}*switch/{_}     {l();handleLabel(yytext, yyleng-6);return T_SWITCH;}
-for                          {c();avm2_lval.id="";return T_FOR;}
-do                           {c();avm2_lval.id="";return T_DO;}
-while                        {c();avm2_lval.id="";return T_WHILE;}
-switch                       {c();avm2_lval.id="";return T_SWITCH;}
+for                          {c();a3_lval.id="";return T_FOR;}
+do                           {c();a3_lval.id="";return T_DO;}
+while                        {c();a3_lval.id="";return T_WHILE;}
+switch                       {c();a3_lval.id="";return T_SWITCH;}
 
 [&][&]                       {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
 [|][|]                       {c();BEGIN(REGEXPOK);return m(T_OROR);}
@@ -620,12 +667,13 @@ is                           {c();return m(KW_IS) ;}
 in                           {c();return m(KW_IN) ;}
 if                           {c();return m(KW_IF) ;}
 as                           {c();return m(KW_AS);}
-{NAME}                       {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);}
+{NAME}                       {c();BEGIN(INITIAL);return handleIdentifier(T_IDENTIFIER);}
 
 [+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
 [\)\]]                           {c();BEGIN(INITIAL);return m(yytext[0]);}
 
-.                           {char c1=yytext[0];
+.                           {/* ERROR */
+                              char c1=yytext[0];
                               char buf[128];
                               buf[0] = yytext[0];
                               int t;
@@ -722,3 +770,4 @@ void initialize_scanner()
     BEGIN(BEGINNING);
 }
 
+