X-Git-Url: http://git.asbjorn.biz/?a=blobdiff_plain;f=lib%2Fas3%2Ftokenizer.lex;h=63f586dda0e60cbc55ad3cc53ff9690bb32293b1;hb=063a400cad92488e0cc96f8362a3d54d894e0422;hp=d2e277c91029c34c3d9287c439aafb04af5bcb11;hpb=6a35c7253ba161c50a18ecc86e46d0b3ead9a2a0;p=swftools.git

diff --git a/lib/as3/tokenizer.lex b/lib/as3/tokenizer.lex
index d2e277c..63f586d 100644
--- a/lib/as3/tokenizer.lex
+++ b/lib/as3/tokenizer.lex
@@ -31,59 +31,43 @@
 #include "tokenizer.h"
 #include "files.h"
 
-static void countlines(char*text, int len) {
-    int t;
-    for(t=0;t<len;t++) {
-	if(text[t]=='\n') {
-	    current_line++;
-	    current_column=0;
-	} else {
-	    current_column++;
-	}
-    }
-}
+int as3_pass = 0;
+int as3_verbosity = 1;
+unsigned int as3_tokencount = 0;
 
-static int verbose = 1;
-static void dbg(const char*format, ...)
+void as3_error(const char*format, ...)
 {
     char buf[1024];
     int l;
     va_list arglist;
-    if(!verbose)
-	return;
+    if(as3_verbosity<0)
+        exit(1);
     va_start(arglist, format);
     vsprintf(buf, format, arglist);
     va_end(arglist);
-    l = strlen(buf);
-    while(l && buf[l-1]=='\n') {
-	buf[l-1] = 0;
-	l--;
-    }
-    printf("(tokenizer) ");
-    printf("%s\n", buf);
-    fflush(stdout);
+    fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
+    fflush(stderr);
+    exit(1);
 }
-
-void syntaxerror(const char*format, ...)
+void as3_warning(const char*format, ...)
 {
     char buf[1024];
     int l;
     va_list arglist;
-    if(!verbose)
-	return;
+    if(as3_verbosity<1)
+        return;
     va_start(arglist, format);
     vsprintf(buf, format, arglist);
     va_end(arglist);
-    fprintf(stderr, "%s:%d:%d: error: %s\n", current_filename_short, current_line, current_column, buf);
+    fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
     fflush(stderr);
-    exit(1);
 }
-void warning(const char*format, ...)
+void as3_softwarning(const char*format, ...)
 {
     char buf[1024];
     int l;
     va_list arglist;
-    if(!verbose)
+    if(as3_verbosity<2)
 	return;
     va_start(arglist, format);
     vsprintf(buf, format, arglist);
@@ -91,12 +75,63 @@ void warning(const char*format, ...)
     fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
     fflush(stderr);
 }
-
+static void dbg(const char*format, ...)
+{
+    char buf[1024];
+    int l;
+    va_list arglist;
+    if(as3_verbosity<3)
+	return;
+    va_start(arglist, format);
+    vsprintf(buf, format, arglist);
+    va_end(arglist);
+    l = strlen(buf);
+    while(l && buf[l-1]=='\n') {
+	buf[l-1] = 0;
+	l--;
+    }
+    printf("(tokenizer) ");
+    printf("%s\n", buf);
+    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;
@@ -119,7 +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);
@@ -157,6 +193,7 @@ static int do_unescape(const char*s, const char*end, char*n)
         switch(*s) {
 	    case '\\': if(o) o[len] = '\\';s++;len++; break;
 	    case '"': if(o) o[len] = '"';s++;len++; break;
+	    case '\'': if(o) o[len] = '\'';s++;len++; break;
 	    case 'b': if(o) o[len] = '\b';s++;len++; break;
 	    case 'f': if(o) o[len] = '\f';s++;len++; break;
 	    case 'n': if(o) o[len] = '\n';s++;len++; break;
@@ -240,8 +277,20 @@ 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;// ]]>
+}
+
 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;
@@ -253,24 +302,24 @@ 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;
 }
 
@@ -288,7 +337,7 @@ 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)
@@ -298,7 +347,7 @@ 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)
@@ -308,14 +357,14 @@ 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;
 }
 
@@ -326,14 +375,16 @@ static inline int handleint()
 
     char*max = l?"1073741824":"2147483647";
     if(yyleng-l>10) {
-        warning("integer overflow: %s", s);
+        as3_warning("integer overflow: %s (converted to Number)", s);
         return handlefloat();
     }
     if(yyleng-l==10) {
         int t;
         for(t=0;t<yyleng-l;t++) {
-            if(yytext[l+t]>max[t])
-                syntaxerror("integer overflow %s > %s", s+l,max);
+            if(yytext[l+t]>max[t]) {
+                as3_warning("integer overflow: %s (converted to Number)", s);
+                return handlefloat();
+            }
             else if(yytext[l+t]<max[t])
                 break;
         }
@@ -352,34 +403,59 @@ static inline int handleint()
     }
 }
 
+static inline int handlehexfloat()
+{
+    char l = (yytext[0]=='-')+2;
+    double d=0;
+    char dot=0;
+    double base=1;
+    int t;
+    for(t=l;t<yyleng;t++) {
+        char c = yytext[t];
+        if(c=='.') {
+            dot=1;
+            continue;
+        }
+        if(!dot) {
+            d*=16;
+        } else {
+            base*=1/16.0;
+        }
+        if(c>='0' && c<='9')
+            d+=(c&15)*base;
+        else if((c>='a' && c<='f') || (c>='A' && c<='F'))
+            d+=((c&0x0f)+9)*base;
+    }
+    return setfloat(d);
+}
 static inline int handlehex()
 {
     char l = (yytext[0]=='-')+2;
+    int len = yyleng;
 
-    if(yyleng-l>8) {
+    if(len-l>8) {
         char*s = nrbuf();
         syntaxerror("integer overflow %s", s);
     }
 
     int t;
     unsigned int v = 0;
-    for(t=l;t<yyleng;t++) {
+    for(t=l;t<len;t++) {
         v<<=4;
         char c = yytext[t];
         if(c>='0' && c<='9')
             v|=(c&15);
-        else if(c>='a' && c<='f' ||
-                c>='A' && c<='F')
+        else if((c>='a' && c<='f') || (c>='A' && c<='F'))
             v|=(c&0x0f)+9;
     }
     if(l && v>1073741824) {
         char*s = nrbuf();
-        warning("signed integer overflow: %s", s);
+        as3_warning("signed integer overflow: %s (converted to Number)", s);
         return setfloat(v);
     }
     if(!l && v>2147483647) {
         char*s = nrbuf();
-        warning("unsigned integer overflow: %s", s);
+        as3_warning("unsigned integer overflow: %s (converted to Number)", s);
         return setfloat(v);
     }
 
@@ -395,26 +471,62 @@ void handleLabel(char*text, int len)
     int t;
     for(t=len-1;t>=0;--t) {
         if(text[t]!=' ' &&
-           text[t]!='.')
+           text[t]!=':')
             break;
     }
     char*s = malloc(t+1);
     memcpy(s, yytext, t);
     s[t]=0;
-    avm2_lval.id = s;
+    a3_lval.id = s;
+}
+
+static int handleregexp()
+{
+    char*s = malloc(yyleng);
+    int len=yyleng-1;
+    memcpy(s, yytext+1, len);
+    s[len] = 0;
+    int t;
+    for(t=len;t>=0;--t) {
+        if(s[t]=='/') {
+            s[t] = 0;
+            break;
+        }
+    }
+    a3_lval.regexp.pattern = s;
+    if(t==len) {
+        a3_lval.regexp.options = 0;
+    } else {
+        a3_lval.regexp.options = s+t+1;
+    }
+    return T_REGEXP;
 }
 
 void initialize_scanner();
 #define YY_USER_INIT initialize_scanner();
 
-#define c() {countlines(yytext, yyleng);}
+/* count the number of lines+columns consumed by this token */
+static inline void l() {
+    int t;
+    for(t=0;t<yyleng;t++) {
+	if(yytext[t]=='\n') {
+	    current_line++;
+	    current_column=0;
+	} else {
+	    current_column++;
+	}
+    }
+}
+/* count the number of columns consumed by this token */
+static inline void c() {
+    current_column+=yyleng;
+}
 
 //Boolean                      {c();return m(KW_BOOLEAN);}
 //int                          {c();return m(KW_INT);}
 //uint                         {c();return m(KW_UINT);}
 //Number                       {c();return m(KW_NUMBER);}
 
-
 %}
 
 %s REGEXPOK
@@ -424,13 +536,16 @@ NAME	 [a-zA-Z_][a-zA-Z0-9_\\]*
 _        [^a-zA-Z0-9_\\]
 
 HEXINT    0x[a-zA-Z0-9]+
+HEXFLOAT  0x[a-zA-Z0-9]*\.[a-zA-Z0-9]*
 INT       [0-9]+
 FLOAT     [0-9]+(\.[0-9]*)?|\.[0-9]+
 
 HEXWITHSIGN [+-]?({HEXINT})
+HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
 INTWITHSIGN [+-]?({INT})
 FLOATWITHSIGN [+-]?({FLOAT})
 
+CDATA    <!\[CDATA\[([^]]|\][^]]|\]\][^>])*\]*\]\]\>
 STRING   ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
 S 	 [ \n\r\t]
 MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
@@ -439,39 +554,42 @@ REGEXP   [/]([^/\n]|\\[/])*[/][a-zA-Z]*
 %%
 
 
-{SINGLELINE_COMMENT}         {c(); /* single line comment */}
-{MULTILINE_COMMENT}          {c(); /* multi line comment */}
+{SINGLELINE_COMMENT}         {l(); /* single line comment */}
+{MULTILINE_COMMENT}          {l(); /* multi line comment */}
 [/][*]                       {syntaxerror("syntax error: unterminated comment", yytext);}
 
-^include{S}+{STRING}{S}*/\n    {c();handleInclude(yytext, yyleng, 1);}
-^include{S}+[^" \t\r\n][\x20-\xff]*{S}*/\n    {c();handleInclude(yytext, yyleng, 0);}
-{STRING}                     {c(); BEGIN(INITIAL);handleString(yytext, yyleng);return T_STRING;}
+^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 m(T_REGEXP);} 
+{REGEXP}                     {c(); BEGIN(INITIAL);return handleregexp();} 
 {HEXWITHSIGN}                {c(); BEGIN(INITIAL);return handlehex();}
+{HEXFLOATWITHSIGN}           {c(); BEGIN(INITIAL);return handlehexfloat();}
 {INTWITHSIGN}                {c(); BEGIN(INITIAL);return handleint();}
 {FLOATWITHSIGN}              {c(); BEGIN(INITIAL);return handlefloat();}
 }
 
 \xef\xbb\xbf                 {/* utf 8 bom */}
-{S}                          {c();}
+{S}                          {l();}
 
 {HEXINT}                     {c(); BEGIN(INITIAL);return handlehex();}
+{HEXFLOAT}                   {c(); BEGIN(INITIAL);return handlehexfloat();}
 {INT}                        {c(); BEGIN(INITIAL);return handleint();}
 {FLOAT}                      {c(); BEGIN(INITIAL);return handlefloat();}
 
 3rr0r                        {/* for debugging: generates a tokenizer-level error */
                               syntaxerror("3rr0r");}
 
-{NAME}{S}*:{S}*for/{_}        {c();handleLabel(yytext, yyleng-3);return T_FOR;}
-{NAME}{S}*:{S}*do/{_}         {c();handleLabel(yytext, yyleng-2);return T_DO;}
-{NAME}{S}*:{S}*while/{_}      {c();handleLabel(yytext, yyleng-5);return T_WHILE;}
-{NAME}{S}*:{S}*switch/{_}     {c();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;}
+{NAME}{S}*:{S}*for/{_}        {l();handleLabel(yytext, yyleng-3);return T_FOR;}
+{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();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);}
@@ -510,6 +628,7 @@ continue                     {c();return m(KW_CONTINUE);}
 override                     {c();return m(KW_OVERRIDE);}
 internal                     {c();return m(KW_INTERNAL);}
 function                     {c();return m(KW_FUNCTION);}
+finally                      {c();return m(KW_FINALLY);}
 default                      {c();return m(KW_DEFAULT);}
 package                      {c();return m(KW_PACKAGE);}
 private                      {c();return m(KW_PRIVATE);}
@@ -547,12 +666,13 @@ 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]);}
+[\)\]]                           {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;
@@ -571,7 +691,7 @@ as                           {c();return m(KW_AS);}
 			      exit(1);
 		              yyterminate();
 		             }
-<<EOF>>		             {c();
+<<EOF>>		             {l();
                               void*b = leave_file();
 			      if (!b) {
 			         yyterminate();
@@ -649,3 +769,4 @@ void initialize_scanner()
     BEGIN(BEGINNING);
 }
 
+