X-Git-Url: http://git.asbjorn.biz/?a=blobdiff_plain;f=lib%2Fas3%2Ftokenizer.lex;h=bc4f01248265ed1d3763defc33378849f7b6c4a4;hb=0fd17f47ee30d90181e51f30fc17a6c31646137e;hp=c06b2be18d0f9aec98a75d906bde456e319669b0;hpb=1adc9bdd04d6c40fadf2211a4af8c281b46ca5ec;p=swftools.git diff --git a/lib/as3/tokenizer.lex b/lib/as3/tokenizer.lex index c06b2be..bc4f012 100644 --- 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, 1); + 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; // + 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 +305,15 @@ 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) -{ - 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; } @@ -292,7 +331,7 @@ 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 +341,7 @@ 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 +351,14 @@ 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; } @@ -330,14 +369,14 @@ static inline int handleint() char*max = l?"1073741824":"2147483647"; if(yyleng-l>10) { - as3_warning("integer overflow: %s (converted to Number)", s); + as3_softwarning("integer overflow: %s (converted to Number)", s); return handlefloat(); } if(yyleng-l==10) { int t; for(t=0;tmax[t]) { - as3_warning("integer overflow: %s (converted to Number)", s); + as3_softwarning("integer overflow: %s (converted to Number)", s); return handlefloat(); } else if(yytext[l+t]1073741824) { char*s = nrbuf(); - as3_warning("signed integer overflow: %s (converted to Number)", s); + as3_softwarning("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); + as3_softwarning("unsigned integer overflow: %s (converted to Number)", s); return setfloat(v); } @@ -432,7 +471,7 @@ 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 +487,11 @@ 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; } @@ -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 @@ -494,17 +555,18 @@ _ [^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 ])*\]*\]\]\> STRING ["](\\[\x00-\xff]|[^\\"\n])*["]|['](\\[\x00-\xff]|[^\\'\n])*['] S [ \n\r\t] MULTILINE_COMMENT [/][*]+([*][^/]|[^/*]|[^*][/]|[\x00-\x1f])*[*]+[/] -SINGLELINE_COMMENT \/\/[^\n]*\n +SINGLELINE_COMMENT \/\/[^\n\r]*[\n\r] REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]* %% @@ -516,34 +578,38 @@ 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;} { {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();} } +[\{] {c(); BEGIN(REGEXPOK);return m(T_DICTSTART);} +[\{] {c(); BEGIN(INITIAL); return m('{');} + \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");} -{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();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();BEGIN(INITIAL);handleLabel(yytext, yyleng-3);return T_FOR;} +{NAME}{S}*:{S}*do/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-2);return T_DO;} +{NAME}{S}*:{S}*while/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-5);return T_WHILE;} +{NAME}{S}*:{S}*switch/{_} {l();BEGIN(INITIAL);handleLabel(yytext, yyleng-6);return T_SWITCH;} +for {c();BEGIN(INITIAL);a3_lval.id="";return T_FOR;} +do {c();BEGIN(INITIAL);a3_lval.id="";return T_DO;} +while {c();BEGIN(INITIAL);a3_lval.id="";return T_WHILE;} +switch {c();BEGIN(INITIAL);a3_lval.id="";return T_SWITCH;} [&][&] {c();BEGIN(REGEXPOK);return m(T_ANDAND);} [|][|] {c();BEGIN(REGEXPOK);return m(T_OROR);} @@ -551,81 +617,84 @@ switch {c();avm2_lval.id="";return T_SWITCH;} [!][=][=] {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(REGEXPOK);return m(T_GE);} +[<][=] {c();BEGIN(REGEXPOK);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_ORBY);} -[>][>][=] {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(':');} -instanceof {c();return m(KW_INSTANCEOF);} -implements {c();return m(KW_IMPLEMENTS);} -interface {c();return m(KW_INTERFACE);} -namespace {c();return m(KW_NAMESPACE);} -protected {c();return m(KW_PROTECTED);} -undefined {c();return m(KW_UNDEFINED);} -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);} -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);} -typeof {c();return m(KW_TYPEOF);} -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);} -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);} -{NAME} {c();BEGIN(INITIAL);return mkid(T_IDENTIFIER);} - -[+-\/*^~@$!%&\(=\[\]\{\}|?:;,<>] {c();BEGIN(REGEXPOK);return m(yytext[0]);} +[+][=] {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_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(INITIAL);return m(KW_INTERFACE);} +namespace {c();BEGIN(INITIAL);return m(KW_NAMESPACE);} +protected {c();BEGIN(INITIAL);return m(KW_PROTECTED);} +undefined {c();BEGIN(INITIAL);return m(KW_UNDEFINED);} +continue {c();BEGIN(INITIAL);return m(KW_CONTINUE);} +override {c();BEGIN(INITIAL);return m(KW_OVERRIDE);} +internal {c();BEGIN(INITIAL);return m(KW_INTERNAL);} +function {c();BEGIN(INITIAL);return m(KW_FUNCTION);} +finally {c();BEGIN(INITIAL);return m(KW_FINALLY);} +default {c();BEGIN(INITIAL);return m(KW_DEFAULT);} +package {c();BEGIN(INITIAL);return m(KW_PACKAGE);} +private {c();BEGIN(INITIAL);return m(KW_PRIVATE);} +dynamic {c();BEGIN(INITIAL);return m(KW_DYNAMIC);} +extends {c();BEGIN(INITIAL);return m(KW_EXTENDS);} +delete {c();BEGIN(REGEXPOK);return m(KW_DELETE);} +return {c();BEGIN(REGEXPOK);return m(KW_RETURN);} +public {c();BEGIN(INITIAL);return m(KW_PUBLIC);} +native {c();BEGIN(INITIAL);return m(KW_NATIVE);} +static {c();BEGIN(INITIAL);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(INITIAL);return m(KW_CLASS);} +const {c();BEGIN(INITIAL);return m(KW_CONST);} +catch {c();BEGIN(INITIAL);return m(KW_CATCH);} +final {c();BEGIN(INITIAL);return m(KW_FINAL);} +false {c();BEGIN(INITIAL);return m(KW_FALSE);} +break {c();BEGIN(INITIAL);return m(KW_BREAK);} +super {c();BEGIN(INITIAL);return m(KW_SUPER);} +each {c();BEGIN(INITIAL);return m(KW_EACH);} +void {c();BEGIN(INITIAL);return m(KW_VOID);} +true {c();BEGIN(INITIAL);return m(KW_TRUE);} +null {c();BEGIN(INITIAL);return m(KW_NULL);} +else {c();BEGIN(INITIAL);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(INITIAL);return m(KW_GET);} +set {c();BEGIN(INITIAL);return m(KW_SET);} +var {c();BEGIN(INITIAL);return m(KW_VAR);} +try {c();BEGIN(INITIAL);return m(KW_TRY);} +is {c();BEGIN(REGEXPOK);return m(KW_IS) ;} +in {c();BEGIN(REGEXPOK);return m(KW_IN) ;} +if {c();BEGIN(INITIAL);return m(KW_IF) ;} +as {c();BEGIN(REGEXPOK);return m(KW_AS);} +$?{NAME} {c();BEGIN(INITIAL);return handleIdentifier();} + +[\]\}*] {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; @@ -666,12 +735,27 @@ int yywrap() static char mbuf[256]; char*token2string(enum yytokentype nr, YYSTYPE v) { - if(nr==T_STRING) return ""; + if(nr==T_STRING) { + char*s = malloc(v.str.len+10); + strcpy(s, ""); + 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, "%s", v.regexp.pattern); + return s; + } + else if(nr==T_IDENTIFIER) { + char*s = malloc(strlen(v.id)+10); + sprintf(s, "%s", v.id); + return s; + } else if(nr==T_INT) return ""; else if(nr==T_UINT) return ""; else if(nr==T_BYTE) return ""; else if(nr==T_FLOAT) return ""; - 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 "<="; @@ -710,7 +794,6 @@ char*token2string(enum yytokentype nr, YYSTYPE v) 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; @@ -722,3 +805,4 @@ void initialize_scanner() BEGIN(BEGINNING); } +