#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);
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
//BEGIN(INITIAL); 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;
}
syntaxerror("unknown escape sequence: \"\\%c\"", *s);
}
}
+ 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 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;
else syntaxerror("String incorrectly terminated");
- avm2_lval.str = string_unescape(s, len);
+ a3_lval.str = string_unescape(s, len);
}
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;
}
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)
}
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)
else
return T_UINT;
}
+static inline int setfloat(double 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;
}
char l = (yytext[0]=='-');
char*max = l?"1073741824":"2147483647";
- if(yyleng-l>10)
- syntaxerror("integer overflow");
+ if(yyleng-l>10) {
+ 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;
}
}
}
+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);
+ }
- if(yyleng-l>8)
- syntaxerror("integer overflow");
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)
- syntaxerror("signed integer overflow");
- if(!l && v>2147483647)
- syntaxerror("unsigned integer overflow");
+ if(l && v>1073741824) {
+ char*s = nrbuf();
+ as3_warning("signed integer overflow: %s (converted to Number)", s);
+ return setfloat(v);
+ }
+ if(!l && v>2147483647) {
+ char*s = nrbuf();
+ as3_warning("unsigned integer overflow: %s (converted to Number)", s);
+ return setfloat(v);
+ }
if(l==3) {
return setint(-(int)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;
+}
//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
%s BEGINNING
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})
%%
-{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;}
<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/{_} {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);}
[!][=] {c();BEGIN(REGEXPOK);return m(T_NE);}
[/][=] {c();return m(T_DIVBY);}
[%][=] {c();return m(T_MODBY);}
[*][=] {c();return m(T_MULBY);}
+[|][=] {c();return m(T_ORBY);}
[>][>][=] {c();return m(T_SHRBY);}
[<][<][=] {c();return m(T_SHLBY);}
[>][>][>][=] {c();return m(T_USHRBY);}
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);}
dynamic {c();return m(KW_DYNAMIC);}
static {c();return m(KW_STATIC);}
import {c();return m(KW_IMPORT);}
typeof {c();return m(KW_TYPEOF);}
-while {c();return m(KW_WHILE);}
+throw {c();return m(KW_THROW);}
class {c();return m(KW_CLASS);}
const {c();return m(KW_CONST);}
+catch {c();return m(KW_CATCH);}
final {c();return m(KW_FINAL);}
false {c();return m(KW_FALSE);}
break {c();return m(KW_BREAK);}
super {c();return m(KW_SUPER);}
+each {c();return m(KW_EACH);}
void {c();return m(KW_VOID);}
true {c();return m(KW_TRUE);}
null {c();return m(KW_NULL);}
else {c();return m(KW_ELSE);}
+case {c();return m(KW_CASE);}
+with {c();return m(KW_WITH);}
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);}
+try {c();return m(KW_TRY);}
is {c();return m(KW_IS) ;}
+in {c();return m(KW_IN) ;}
if {c();return m(KW_IF) ;}
as {c();return m(KW_AS);}
-do {c();return m(KW_DO);}
{NAME} {c();BEGIN(INITIAL);return mkid(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;
exit(1);
yyterminate();
}
-<<EOF>> {c();
+<<EOF>> {l();
void*b = leave_file();
if (!b) {
yyterminate();
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";
git.asbjorn.biz / swftools.git / blobdiff