#include <stdio.h>
#include <stdarg.h>
#include "../utf8.h"
+#include "common.h"
#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++;
- }
- }
-}
+unsigned int as3_tokencount = 0;
-static int verbose = 1;
static void dbg(const char*format, ...)
{
char buf[1024];
int l;
va_list arglist;
- if(!verbose)
+ if(as3_verbosity<3)
return;
va_start(arglist, format);
- vsprintf(buf, format, arglist);
+ vsnprintf(buf, sizeof(buf)-1, format, arglist);
va_end(arglist);
l = strlen(buf);
while(l && buf[l-1]=='\n') {
fflush(stdout);
}
-void syntaxerror(const char*format, ...)
+#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)
{
- char buf[1024];
- int l;
- va_list arglist;
- if(!verbose)
- 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);
- fflush(stderr);
- exit(1);
+ as3_in = fi;
+ as3_buffer = 0;
}
-void warning(const char*format, ...)
+void as3_buffer_input(void*buffer, int len)
{
- char buf[1024];
- int l;
- va_list arglist;
- if(!verbose)
- return;
- va_start(arglist, format);
- vsprintf(buf, format, arglist);
- va_end(arglist);
- fprintf(stderr, "%s:%d:%d: warning: %s\n", current_filename_short, current_line, current_column, buf);
- fflush(stderr);
+ if(!buffer)
+ syntaxerror("trying to parse zero bytearray");
+ as3_buffer = buffer;
+ as3_buffer_len = len;
+ as3_buffer_pos = 0;
+ as3_in = 0;
}
-
-#ifndef YY_CURRENT_BUFFER
-#define YY_CURRENT_BUFFER yy_current_buffer
-#endif
+//#undef BEGIN
+//#define BEGIN(x) {(yy_start) = 1 + 2 *x;dbg("entering state %d", x);}
+
+#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)
{
} else {
int i1=0,i2=len;
// find start
- while(!strchr(" \n\r\t", text[i1])) i1++;
+ while(!strchr(" \n\r\t\xa0", text[i1])) i1++;
// strip
- while(strchr(" \n\r\t", text[i1])) i1++;
- while(strchr(" \n\r\t", text[i2-1])) i2--;
+ while(strchr(" \n\r\t\xa0", text[i1])) i1++;
+ while(strchr(" \n\r\t\xa0", text[i2-1])) i2--;
if(i2!=len) text[i2]=0;
filename = strdup(&text[i1]);
}
- char*fullfilename = enter_file(filename, YY_CURRENT_BUFFER);
+ char*fullfilename = find_file(filename, 1);
+ enter_file2(filename, fullfilename, YY_CURRENT_BUFFER);
yyin = fopen(fullfilename, "rb");
if (!yyin) {
syntaxerror("Couldn't open include file \"%s\"\n", fullfilename);
}
yy_switch_to_buffer(yy_create_buffer( yyin, YY_BUF_SIZE ) );
- //BEGIN(INITIAL); keep context
+ //BEGIN(DEFAULT); keep context
}
-string_t string_unescape(const char*in, int l)
+static int do_unescape(const char*s, const char*end, char*n)
{
- int len=0;
- const char*s = in;
- const char*end = &in[l];
- char*n = (char*)malloc(l);
char*o = n;
+ int len=0;
while(s<end) {
if(*s!='\\') {
- o[len++] = *s;
+ if(o) o[len] = *s;len++;
s++;
continue;
}
continue;
}
switch(*s) {
- case '\\': o[len++] = '\\';s++; break;
- case '"': o[len++] = '"';s++; break;
- case 'b': o[len++] = '\b';s++; break;
- case 'f': o[len++] = '\f';s++; break;
- case 'n': o[len++] = '\n';s++; break;
- case 'r': o[len++] = '\r';s++; break;
- case 't': o[len++] = '\t';s++; break;
+ 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;
+ case 'r': if(o) o[len] = '\r';s++;len++; break;
+ case 't': if(o) o[len] = '\t';s++;len++; break;
case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': {
unsigned int num=0;
int nr = 0;
}
if(num>256)
syntaxerror("octal number out of range (0-255): %d", num);
- o[len++] = num;
+ if(o) o[len] = num;len++;
continue;
}
case 'x': case 'u': {
if(unicode) {
char*utf8 = getUTF8(num);
while(*utf8) {
- o[len++] = *utf8++;
+ if(o) o[len] = *utf8;utf8++;len++;
}
} else {
if(num>256)
syntaxerror("byte out of range (0-255): %d", num);
- o[len++] = num;
+ if(o) o[len] = num;len++;
}
break;
}
- default:
- syntaxerror("unknown escape sequence: \"\\%c\"", *s);
+ default: {
+ if(o) {
+ o[len+0] = '\\';
+ o[len+1] = *s;
+ }
+ s++;
+ len+=2;
+ break;
+ }
}
}
+ if(o) o[len]=0;
+ return len;
+}
+
+static string_t string_unescape(const char*in, int l)
+{
+ const char*s = in;
+ const char*end = &in[l];
+
+ int len = do_unescape(s, end, 0);
+ char*n = (char*)malloc(len+1);
+ do_unescape(s, end, n);
string_t out = string_new(n, len);
- o[len]=0;
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 handleRaw(char*s, int len)
+{
+ a3_lval.str.len = len;
+ a3_lval.str.str = strdup_n(s, a3_lval.str.len);
+}
+
static void handleString(char*s, int len)
{
if(s[0]=='"') {
s++;len-=2;
}
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)
-{
- char*s = malloc(yyleng+1);
- memcpy(s, yytext, yyleng);
- s[yyleng]=0;
- avm2_lval.id = s;
- return type;
-}
-
static inline int m(int type)
{
- avm2_lval.token = type;
+ a3_lval.token = type;
return type;
}
-
static char numberbuf[64];
-static inline int handlenumber()
+static char*nrbuf()
{
if(yyleng>sizeof(numberbuf)-1)
syntaxerror("decimal number overflow");
-
char*s = numberbuf;
memcpy(s, yytext, yyleng);
s[yyleng]=0;
+ return s;
+}
- int t;
- char is_float=0;
- for(t=0;t<yyleng;t++) {
- if(yytext[t]=='.') {
- if(is_float)
- syntaxerror("Invalid number");
- is_float=1;
- } else if(!strchr("-0123456789", yytext[t])) {
- syntaxerror("Invalid number");
- }
- }
- if(is_float) {
- avm2_lval.number_float = atof(s);
- return T_FLOAT;
- }
+static inline int setint(int v)
+{
+ a3_lval.number_int = v;
+ return T_INT;
+}
+static inline int setfloat(double v)
+{
+ a3_lval.number_float = v;
+ return T_FLOAT;
+}
+
+static inline int handlefloat()
+{
+ char*s = nrbuf();
+ a3_lval.number_float = atof(s);
+ return T_FLOAT;
+}
+
+static inline int handleint()
+{
+ char*s = nrbuf();
char l = (yytext[0]=='-');
- char*max = l?"1073741824":"2147483647";
- if(yyleng-l>10)
- syntaxerror("integer overflow");
+ //char*max = l?"1073741824":"2147483647";
+ char*max = l?"2147483648":"2147483647";
+
+ if(yyleng-l>10) {
+ as3_softwarning("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_softwarning("integer overflow: %s (converted to Number)", s);
+ return handlefloat();
+ }
else if(yytext[l+t]<max[t])
break;
}
}
if(yytext[0]=='-') {
int v = atoi(s);
- avm2_lval.number_int = v;
- if(v>-128)
- return T_BYTE;
- else if(v>=-32768)
- return T_SHORT;
- else
- return T_INT;
+ return setint(v);
} else {
unsigned int v = 0;
+ int t;
for(t=0;t<yyleng;t++) {
v*=10;
v+=yytext[t]-'0';
}
- avm2_lval.number_uint = v;
- if(v<128)
- return T_BYTE;
- else if(v<32768)
- return T_SHORT;
- else
- return T_UINT;
+ return setint(v);
+ }
+}
+
+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(len-l>8) {
+ char*s = nrbuf();
+ syntaxerror("integer overflow %s", s);
+ }
+
+ int t;
+ unsigned int v = 0;
+ 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'))
+ v|=(c&0x0f)+9;
+ }
+ if(l && v>=0x80000000) {
+ char*s = nrbuf();
+ as3_softwarning("integer overflow: %s (converted to Number)", s);
+ return setfloat(v);
+ }
+ if(!l && v>0x7fffffff) {
+ char*s = nrbuf();
+ as3_softwarning("integer overflow: %s (converted to Number)", s);
+ return setfloat(v);
+ }
+
+ if(l==3) {
+ return setint(-(int)v);
+ } else {
+ return setint(v);
+ }
+}
+
+void handleLabel(char*text, int len)
+{
+ int t;
+ for(t=len-1;t>=0;--t) {
+ if(text[t]!=' ' &&
+ text[t]!=':')
+ break;
+ }
+ char*s = malloc(t+1);
+ memcpy(s, yytext, t);
+ s[t]=0;
+ 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;
+}
+
+trie_t*active_namespaces = 0;
+/*void tokenizer_register_namespace(const char*id)
+{
+ trie_put(namespaces, id, 0);
+}
+void tokenizer_unregister_namespace(const char*id)
+{
+ trie_remove(namespaces, id);
+}*/
+static inline char tokenizer_is_namespace(const char*id)
+{
+ return trie_contains(active_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;
+}
+static int tokenerror();
+
//Boolean {c();return m(KW_BOOLEAN);}
//int {c();return m(KW_INT);}
//uint {c();return m(KW_UINT);}
//Number {c();return m(KW_NUMBER);}
+//XMLCOMMENT <!--([^->]|(-/[^-])|(--/[^>]))*-->
+
+//{XMLCOMMENT}
+
%}
%s REGEXPOK
%s BEGINNING
+%s DEFAULT
+%x XMLTEXT
+%x XML
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]+)(e[0-9]+)?
-NUMBER [0-9]+(\.[0-9]*)?|-?\.[0-9]+
-NUMBERWITHSIGN [+-]?({NUMBER})
+HEXWITHSIGN [+-]?({HEXINT})
+HEXFLOATWITHSIGN [+-]?({HEXFLOAT})
+INTWITHSIGN [+-]?({INT})
+FLOATWITHSIGN [+-]?({FLOAT})
+
+CDATA <!\[CDATA\[([^]]|\][^]]|\]\][^>])*\]*\]\]\>
+XMLCOMMENT <!--([^->]|[-]+[^>-]|>)*-*-->
+XML <[^>]+{S}>
+XMLID [A-Za-z0-9_\x80-\xff]+([:][A-Za-z0-9_\x80-\xff]+)?
+XMLSTRING ["][^"]*["]
STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*[']
-S [ \n\r\t]
+S [ \n\r\t\xa0]
MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/]
-SINGLELINE_COMMENT \/\/[^\n]*\n
+SINGLELINE_COMMENT \/\/[^\n\r]*[\n\r]
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\xa0\r\n][\x20-\xff]*{S}*/\n {l();handleInclude(yytext, yyleng, 0);}
+{STRING} {l(); BEGIN(DEFAULT);handleString(yytext, yyleng);return T_STRING;}
+{CDATA} {l(); BEGIN(DEFAULT);handleCData(yytext, yyleng);return T_STRING;}
+
+<DEFAULT,BEGINNING,REGEXPOK>{
+{XMLCOMMENT} {l(); BEGIN(DEFAULT);handleRaw(yytext, yyleng);return T_STRING;}
+}
+
+<XML>{
+{XMLSTRING} {l(); handleRaw(yytext, yyleng);return T_STRING;}
+[{] {c(); BEGIN(REGEXPOK);return m('{');}
+[<] {c(); return m('<');}
+[/] {c(); return m('/');}
+[>] {c(); return m('>');}
+[=] {c(); return m('=');}
+{XMLID} {c(); handleRaw(yytext, yyleng);return T_IDENTIFIER;}
+{S} {l();}
+<<EOF>> {syntaxerror("unexpected end of file");}
+}
+
+<XMLTEXT>{
+[^<>{]+ {l(); handleRaw(yytext, yyleng);return T_STRING;}
+[{] {c(); BEGIN(REGEXPOK);return m('{');}
+[<] {c(); BEGIN(XML);return m('<');}
+[>] {c(); return m('>');}
+{XMLCOMMENT} {l(); handleRaw(yytext, yyleng);return T_STRING;}
+{CDATA} {l(); handleRaw(yytext, yyleng);return T_STRING;}
+<<EOF>> {syntaxerror("unexpected end of file");}
+}
<BEGINNING,REGEXPOK>{
-{REGEXP} {c(); BEGIN(INITIAL);return m(T_REGEXP);}
-{NUMBERWITHSIGN} {c(); BEGIN(INITIAL);return handlenumber();}
+{REGEXP} {c(); BEGIN(DEFAULT);return handleregexp();}
+{HEXWITHSIGN}/{_} {c(); BEGIN(DEFAULT);return handlehex();}
+{HEXFLOATWITHSIGN}/{_} {c(); BEGIN(DEFAULT);return handlehexfloat();}
+{INTWITHSIGN}/{_} {c(); BEGIN(DEFAULT);return handleint();}
+{FLOATWITHSIGN}/{_} {c(); BEGIN(DEFAULT);return handlefloat();}
}
+<REGEXPOK>[\{] {c(); BEGIN(REGEXPOK);return m(T_DICTSTART);}
+[\{] {c(); BEGIN(DEFAULT); return m('{');}
+
\xef\xbb\xbf {/* utf 8 bom */}
-{S} {c();}
+{S} {l();}
-{NUMBER} {c(); BEGIN(INITIAL);return handlenumber();}
+{HEXINT}/{_} {c(); BEGIN(DEFAULT);return handlehex();}
+{HEXFLOAT}/{_} {c(); BEGIN(DEFAULT);return handlehexfloat();}
+{INT}/{_} {c(); BEGIN(DEFAULT);return handleint();}
+{FLOAT}/{_} {c(); BEGIN(DEFAULT);return handlefloat();}
+NaN {c(); BEGIN(DEFAULT);return m(KW_NAN);}
3rr0r {/* for debugging: generates a tokenizer-level error */
syntaxerror("3rr0r");}
+{NAME}{S}*:{S}*for/{_} {l();BEGIN(DEFAULT);handleLabel(yytext, yyleng-3);return T_FOR;}
+{NAME}{S}*:{S}*do/{_} {l();BEGIN(DEFAULT);handleLabel(yytext, yyleng-2);return T_DO;}
+{NAME}{S}*:{S}*while/{_} {l();BEGIN(DEFAULT);handleLabel(yytext, yyleng-5);return T_WHILE;}
+{NAME}{S}*:{S}*switch/{_} {l();BEGIN(DEFAULT);handleLabel(yytext, yyleng-6);return T_SWITCH;}
+default{S}xml {l();BEGIN(DEFAULT);return m(KW_DEFAULT_XML);}
+for {c();BEGIN(DEFAULT);a3_lval.id="";return T_FOR;}
+do {c();BEGIN(DEFAULT);a3_lval.id="";return T_DO;}
+while {c();BEGIN(DEFAULT);a3_lval.id="";return T_WHILE;}
+switch {c();BEGIN(DEFAULT);a3_lval.id="";return T_SWITCH;}
+
[&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);}
[|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);}
[!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
[!][=][=] {c();BEGIN(REGEXPOK);return m(T_NEE);}
[=][=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQEQ);}
[=][=] {c();BEGIN(REGEXPOK);return m(T_EQEQ);}
-[>][=] {c();return m(T_GE);}
-[<][=] {c();return m(T_LE);}
-[-][-] {c();BEGIN(INITIAL);return m(T_MINUSMINUS);}
-[+][+] {c();BEGIN(INITIAL);return m(T_PLUSPLUS);}
-[+][=] {c();return m(T_PLUSBY);}
-[-][=] {c();return m(T_MINUSBY);}
-[/][=] {c();return m(T_DIVBY);}
-[%][=] {c();return m(T_MODBY);}
-[*][=] {c();return m(T_MULBY);}
-[>][>][=] {c();return m(T_SHRBY);}
-[<][<][=] {c();return m(T_SHLBY);}
-[>][>][>][=] {c();return m(T_USHRBY);}
-[<][<] {c();return m(T_SHL);}
-[>][>][>] {c();return m(T_USHR);}
-[>][>] {c();return m(T_SHR);}
-\.\.\. {c();return m(T_DOTDOTDOT);}
-\.\. {c();return m(T_DOTDOT);}
-\. {c();return m('.');}
-:: {c();return m(T_COLONCOLON);}
-: {c();return m(':');}
-implements {c();return m(KW_IMPLEMENTS);}
-interface {c();return m(KW_INTERFACE);}
-namespace {c();return m(KW_NAMESPACE);}
-protected {c();return m(KW_PROTECTED);}
-override {c();return m(KW_OVERRIDE);}
-internal {c();return m(KW_INTERNAL);}
-function {c();return m(KW_FUNCTION);}
-package {c();return m(KW_PACKAGE);}
-private {c();return m(KW_PRIVATE);}
-dynamic {c();return m(KW_DYNAMIC);}
-extends {c();return m(KW_EXTENDS);}
-delete {c();return m(KW_DELETE);}
-return {c();return m(KW_RETURN);}
-public {c();return m(KW_PUBLIC);}
-native {c();return m(KW_NATIVE);}
-static {c();return m(KW_STATIC);}
-import {c();return m(KW_IMPORT);}
-while {c();return m(KW_WHILE);}
-class {c();return m(KW_CLASS);}
-const {c();return m(KW_CONST);}
-final {c();return m(KW_FINAL);}
-false {c();return m(KW_FALSE);}
-break {c();return m(KW_BREAK);}
-true {c();return m(KW_TRUE);}
-null {c();return m(KW_NULL);}
-else {c();return m(KW_ELSE);}
-use {c();return m(KW_USE);}
-new {c();return m(KW_NEW);}
-get {c();return m(KW_GET);}
-for {c();return m(KW_FOR);}
-set {c();return m(KW_SET);}
-var {c();return m(KW_VAR);}
-is {c();return m(KW_IS) ;}
-if {c();return m(KW_IF) ;}
-as {c();return m(KW_AS);}
-{NAME} {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);}
-
-[+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
-[\)\]] {c();BEGIN(INITIAL);return m(yytext[0]);}
-
-. {char c1=yytext[0];
- char buf[128];
- buf[0] = yytext[0];
- int t;
- for(t=1;t<128;t++) {
- char c = buf[t]=input();
- if(c=='\n' || c==EOF) {
- buf[t] = 0;
- break;
- }
- }
- if(c1>='0' && c1<='9')
- syntaxerror("syntax error: %s (identifiers must not start with a digit)");
- else
- syntaxerror("syntax error: %s", buf);
- printf("\n");
- exit(1);
- yyterminate();
- }
-<<EOF>> {c();
+[>][=] {c();BEGIN(REGEXPOK);return m(T_GE);}
+[<][=] {c();BEGIN(REGEXPOK);return m(T_LE);}
+[-][-] {c();BEGIN(DEFAULT);return m(T_MINUSMINUS);}
+[+][+] {c();BEGIN(DEFAULT);return m(T_PLUSPLUS);}
+[+][=] {c();BEGIN(REGEXPOK);return m(T_PLUSBY);}
+[\^][=] {c();BEGIN(REGEXPOK);return m(T_XORBY);}
+[-][=] {c();BEGIN(REGEXPOK);return m(T_MINUSBY);}
+[/][=] {c();BEGIN(REGEXPOK);return m(T_DIVBY);}
+[%][=] {c();BEGIN(REGEXPOK);return m(T_MODBY);}
+[*][=] {c();BEGIN(REGEXPOK);return m(T_MULBY);}
+[|][=] {c();BEGIN(REGEXPOK);return m(T_ORBY);}
+[&][=] {c();BEGIN(REGEXPOK);return m(T_ANDBY);}
+[>][>][=] {c();BEGIN(REGEXPOK);return m(T_SHRBY);}
+[<][<][=] {c();BEGIN(REGEXPOK);return m(T_SHLBY);}
+[>][>][>][=] {c();BEGIN(REGEXPOK);return m(T_USHRBY);}
+[<][<] {c();BEGIN(REGEXPOK);return m(T_SHL);}
+[>][>][>] {c();BEGIN(REGEXPOK);return m(T_USHR);}
+[>][>] {c();BEGIN(REGEXPOK);return m(T_SHR);}
+\.\.\. {c();BEGIN(REGEXPOK);return m(T_DOTDOTDOT);}
+\.\. {c();BEGIN(REGEXPOK);return m(T_DOTDOT);}
+\. {c();BEGIN(REGEXPOK);return m('.');}
+:: {c();BEGIN(REGEXPOK);return m(T_COLONCOLON);}
+: {c();BEGIN(REGEXPOK);return m(':');}
+instanceof {c();BEGIN(REGEXPOK);return m(KW_INSTANCEOF);}
+implements {c();BEGIN(REGEXPOK);return m(KW_IMPLEMENTS);}
+interface {c();BEGIN(DEFAULT);return m(KW_INTERFACE);}
+protected {c();BEGIN(DEFAULT);return m(KW_PROTECTED);}
+namespace {c();BEGIN(DEFAULT);return m(KW_NAMESPACE);}
+undefined {c();BEGIN(DEFAULT);return m(KW_UNDEFINED);}
+arguments {c();BEGIN(DEFAULT);return m(KW_ARGUMENTS);}
+continue {c();BEGIN(DEFAULT);return m(KW_CONTINUE);}
+override {c();BEGIN(DEFAULT);return m(KW_OVERRIDE);}
+internal {c();BEGIN(DEFAULT);return m(KW_INTERNAL);}
+function {c();BEGIN(DEFAULT);return m(KW_FUNCTION);}
+finally {c();BEGIN(DEFAULT);return m(KW_FINALLY);}
+default {c();BEGIN(DEFAULT);return m(KW_DEFAULT);}
+package {c();BEGIN(DEFAULT);return m(KW_PACKAGE);}
+private {c();BEGIN(DEFAULT);return m(KW_PRIVATE);}
+dynamic {c();BEGIN(DEFAULT);return m(KW_DYNAMIC);}
+extends {c();BEGIN(DEFAULT);return m(KW_EXTENDS);}
+delete {c();BEGIN(REGEXPOK);return m(KW_DELETE);}
+return {c();BEGIN(REGEXPOK);return m(KW_RETURN);}
+public {c();BEGIN(DEFAULT);return m(KW_PUBLIC);}
+native {c();BEGIN(DEFAULT);return m(KW_NATIVE);}
+static {c();BEGIN(DEFAULT);return m(KW_STATIC);}
+import {c();BEGIN(REGEXPOK);return m(KW_IMPORT);}
+typeof {c();BEGIN(REGEXPOK);return m(KW_TYPEOF);}
+throw {c();BEGIN(REGEXPOK);return m(KW_THROW);}
+class {c();BEGIN(DEFAULT);return m(KW_CLASS);}
+const {c();BEGIN(DEFAULT);return m(KW_CONST);}
+catch {c();BEGIN(DEFAULT);return m(KW_CATCH);}
+final {c();BEGIN(DEFAULT);return m(KW_FINAL);}
+false {c();BEGIN(DEFAULT);return m(KW_FALSE);}
+break {c();BEGIN(DEFAULT);return m(KW_BREAK);}
+super {c();BEGIN(DEFAULT);return m(KW_SUPER);}
+each {c();BEGIN(DEFAULT);return m(KW_EACH);}
+void {c();BEGIN(DEFAULT);return m(KW_VOID);}
+true {c();BEGIN(DEFAULT);return m(KW_TRUE);}
+null {c();BEGIN(DEFAULT);return m(KW_NULL);}
+else {c();BEGIN(DEFAULT);return m(KW_ELSE);}
+case {c();BEGIN(REGEXPOK);return m(KW_CASE);}
+with {c();BEGIN(REGEXPOK);return m(KW_WITH);}
+use {c();BEGIN(REGEXPOK);return m(KW_USE);}
+new {c();BEGIN(REGEXPOK);return m(KW_NEW);}
+get {c();BEGIN(DEFAULT);return m(KW_GET);}
+set {c();BEGIN(DEFAULT);return m(KW_SET);}
+var {c();BEGIN(DEFAULT);return m(KW_VAR);}
+try {c();BEGIN(DEFAULT);return m(KW_TRY);}
+is {c();BEGIN(REGEXPOK);return m(KW_IS) ;}
+in {c();BEGIN(REGEXPOK);return m(KW_IN) ;}
+if {c();BEGIN(DEFAULT);return m(KW_IF) ;}
+as {c();BEGIN(REGEXPOK);return m(KW_AS);}
+$?{NAME} {c();BEGIN(DEFAULT);return handleIdentifier();}
+
+[\]\}*] {c();BEGIN(DEFAULT);return m(yytext[0]);}
+[+-\/^~@$!%&\(=\[|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);}
+[\)\]] {c();BEGIN(DEFAULT);return m(yytext[0]);}
+
+<DEFAULT,BEGINNING,REGEXPOK,XML,XMLTEXT>{
+. {tokenerror();}
+}
+<<EOF>> {l();
void*b = leave_file();
if (!b) {
yyterminate();
return 1;
}
+static int tokenerror()
+{
+ char c1=yytext[0];
+ char buf[128];
+ buf[0] = yytext[0];
+ int t;
+ for(t=1;t<128;t++) {
+ char c = buf[t]=input();
+ if(c=='\n' || c==EOF) {
+ buf[t] = 0;
+ break;
+ }
+ }
+ if(c1>='0' && c1<='9')
+ syntaxerror("syntax error: %s (identifiers must not start with a digit)");
+ else
+ syntaxerror("syntax error [%d]: %s", (yy_start-1)/2, buf);
+ printf("\n");
+ exit(1);
+ yyterminate();
+}
+
+
static char mbuf[256];
char*token2string(enum yytokentype nr, YYSTYPE v)
{
- if(nr==T_STRING) return "<string>";
+ if(nr==T_STRING) {
+ char*s = malloc(v.str.len+10);
+ strcpy(s, "<string>");
+ memcpy(s+8, v.str.str, v.str.len);
+ sprintf(s+8+v.str.len, " (%d bytes)", v.str.len);
+ return s;
+ }
+ else if(nr==T_REGEXP) {
+ char*s = malloc(strlen(v.regexp.pattern)+10);
+ sprintf(s, "<regexp>%s", v.regexp.pattern);
+ return s;
+ }
+ else if(nr==T_IDENTIFIER) {
+ char*s = malloc(strlen(v.id)+10);
+ sprintf(s, "<ID>%s", v.id);
+ return s;
+ }
else if(nr==T_INT) return "<int>";
else if(nr==T_UINT) return "<uint>";
- else if(nr==T_BYTE) return "<byte>";
else if(nr==T_FLOAT) return "<float>";
- else if(nr==T_REGEXP) return "REGEXP";
else if(nr==T_EOF) return "***END***";
else if(nr==T_GE) return ">=";
else if(nr==T_LE) return "<=";
else if(nr==KW_INT) return "int";
else if(nr==KW_NEW) return "new";
else if(nr==KW_GET) return "get";
- else if(nr==KW_FOR) return "for";
else if(nr==KW_SET) return "set";
else if(nr==KW_VAR) return "var";
else if(nr==KW_IS) return "is";
else if(nr==KW_AS) return "as";
- else if(nr==T_IDENTIFIER) return "ID";
else {
sprintf(mbuf, "%d", nr);
return mbuf;
}
}
+void tokenizer_begin_xml()
+{
+ dbg("begin reading xml");
+ BEGIN(XML);
+}
+void tokenizer_begin_xmltext()
+{
+ dbg("begin reading xml text");
+ BEGIN(XMLTEXT);
+}
+void tokenizer_end_xmltext()
+{
+ dbg("end reading xml text");
+ BEGIN(XML);
+}
+void tokenizer_end_xml()
+{
+ dbg("end reading xml");
+ BEGIN(DEFAULT);
+}
+
void initialize_scanner()
{
BEGIN(BEGINNING);
}
+
git.asbjorn.biz / swftools.git / blobdiff