X-Git-Url: http://git.asbjorn.biz/?a=blobdiff_plain;f=lib%2Fas3%2Ftokenizer.lex;h=bc4f01248265ed1d3763defc33378849f7b6c4a4;hb=0fd17f47ee30d90181e51f30fc17a6c31646137e;hp=63f586dda0e60cbc55ad3cc53ff9690bb32293b1;hpb=063a400cad92488e0cc96f8362a3d54d894e0422;p=swftools.git diff --git a/lib/as3/tokenizer.lex b/lib/as3/tokenizer.lex index 63f586d..bc4f012 100644 --- a/lib/as3/tokenizer.lex +++ b/lib/as3/tokenizer.lex @@ -109,6 +109,8 @@ void as3_file_input(FILE*fi) } 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; @@ -154,7 +156,7 @@ void handleInclude(char*text, int len, char quotes) filename = strdup(&text[i1]); } - char*fullfilename = find_file(filename); + char*fullfilename = find_file(filename, 1); enter_file2(filename, fullfilename, YY_CURRENT_BUFFER); yyin = fopen(fullfilename, "rb"); if (!yyin) { @@ -281,6 +283,7 @@ 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) @@ -308,15 +311,6 @@ static void handleString(char*s, int len) char start_of_expression; -static inline int handleIdentifier(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; @@ -375,14 +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); } @@ -522,6 +516,29 @@ 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);} @@ -538,7 +555,7 @@ _ [^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}) @@ -549,7 +566,7 @@ 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]* %% @@ -565,31 +582,34 @@ REGEXP [/]([^/\n]|\\[/])*[/][a-zA-Z]* { {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();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;} +{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);} @@ -597,78 +617,80 @@ switch {c();a3_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 handleIdentifier(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]);} . {/* ERROR */ @@ -713,12 +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 "<="; @@ -757,7 +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;