mark objects without kind as 'undefined'

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

diff --git a/lib/as3/tokenizer.lex b/lib/as3/tokenizer.lex
index 11b1162..d9eb6da 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 +95,45 @@ 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)
+{
+    if(!buffer)
+        syntaxerror("trying to parse zero bytearray");
+    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 +156,8 @@ 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 +279,21 @@ 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;
@@ -263,15 +311,6 @@ static void handleString(char*s, int len)
 
 char start_of_expression;
 
-static inline int mkid(int type)
-{
-    char*s = malloc(yyleng+1);
-    memcpy(s, yytext, yyleng);
-    s[yyleng]=0;
-    a3_lval.id = s;
-    return type;
-}
-
 static inline int m(int type)
 {
     a3_lval.token = type;
@@ -477,12 +516,34 @@ static inline void c() {
     current_column+=yyleng;
 }
 
+static trie_t*namespaces = 0;
+void tokenizer_register_namespace(const char*id)
+{
+    trie_put(&namespaces, id);
+}
+static inline tokenizer_is_namespace(const char*id)
+{
+    return trie_lookup(namespaces, id);
+}
+
+static inline int handleIdentifier()
+{
+    char*s = malloc(yyleng+1);
+    memcpy(s, yytext, yyleng);
+    s[yyleng]=0;
+    a3_lval.id = s;
+    if(tokenizer_is_namespace(s)) 
+        return T_NAMESPACE;
+    else
+        return T_IDENTIFIER;
+}
+
+
 //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
@@ -494,13 +555,14 @@ _        [^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]+
+FLOAT     ([0-9]+(\.[0-9]*)?|\.[0-9]+)(e[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])*[*]+[/]
@@ -516,22 +578,23 @@ 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();}
-{INTWITHSIGN}                {c(); BEGIN(INITIAL);return handleint();}
-{FLOATWITHSIGN}              {c(); BEGIN(INITIAL);return handlefloat();}
+{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}                          {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();}
+{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");}
@@ -620,12 +683,14 @@ 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();}
 
-[+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
+[\]\}]                       {c();BEGIN(INITIAL);return m(yytext[0]);}
+[+-\/*^~@$!%&\(=\[\{|?:;,<>] {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 +787,4 @@ void initialize_scanner()
     BEGIN(BEGINNING);
 }
 
+