123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434 |
- #include "variant.h"
- class VariantConstruct {
- enum TokenType {
- TK_CURLY_BRACKET_OPEN,
- TK_CURLY_BRACKET_CLOSE,
- TK_BRACKET_OPEN,
- TK_BRACKET_CLOSE,
- TK_IDENTIFIER,
- TK_STRING,
- TK_NUMBER,
- TK_COLON,
- TK_COMMA,
- TK_EOF,
- TK_MAX
- };
- enum Expecting {
- EXPECT_OBJECT,
- EXPECT_OBJECT_KEY,
- EXPECT_COLON,
- EXPECT_OBJECT_VALUE,
- };
- struct Token {
- TokenType type;
- Variant value;
- };
- static const char * tk_name[TK_MAX];
- static String _print_var(const Variant& p_var);
- static Error _get_token(const CharType *p_str,int &index, int p_len,Token& r_token,int &line,String &r_err_str);
- static Error _parse_value(Variant &value,Token& token,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud);
- static Error _parse_array(Array &array,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud);
- static Error _parse_dict(Dictionary &object,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud);
- public:
- static Error parse(const String& p_string,Variant& r_ret,String &r_err_str,int &r_err_line,Variant::ObjectConstruct* p_construct,void* p_ud);
- };
- const char * VariantConstruct::tk_name[TK_MAX] = {
- "'{'",
- "'}'",
- "'['",
- "']'",
- "identifier",
- "string",
- "number",
- "':'",
- "','",
- "EOF",
- };
- Error VariantConstruct::_get_token(const CharType *p_str, int &idx, int p_len, Token& r_token,int &line,String &r_err_str) {
- while (true) {
- switch(p_str[idx]) {
- case '\n': {
- line++;
- idx++;
- break;
- };
- case 0: {
- r_token.type=TK_EOF;
- return OK;
- } break;
- case '{': {
- r_token.type=TK_CURLY_BRACKET_OPEN;
- idx++;
- return OK;
- };
- case '}': {
- r_token.type=TK_CURLY_BRACKET_CLOSE;
- idx++;
- return OK;
- };
- case '[': {
- r_token.type=TK_BRACKET_OPEN;
- idx++;
- return OK;
- };
- case ']': {
- r_token.type=TK_BRACKET_CLOSE;
- idx++;
- return OK;
- };
- case ':': {
- r_token.type=TK_COLON;
- idx++;
- return OK;
- };
- case ',': {
- r_token.type=TK_COMMA;
- idx++;
- return OK;
- };
- case '"': {
- idx++;
- String str;
- while(true) {
- if (p_str[idx]==0) {
- r_err_str="Unterminated String";
- return ERR_PARSE_ERROR;
- } else if (p_str[idx]=='"') {
- idx++;
- break;
- } else if (p_str[idx]=='\\') {
- //escaped characters...
- idx++;
- CharType next = p_str[idx];
- if (next==0) {
- r_err_str="Unterminated String";
- return ERR_PARSE_ERROR;
- }
- CharType res=0;
- switch(next) {
- case 'b': res=8; break;
- case 't': res=9; break;
- case 'n': res=10; break;
- case 'f': res=12; break;
- case 'r': res=13; break;
- case '\"': res='\"'; break;
- case '\\': res='\\'; break;
- case '/': res='/'; break; //wtf
- case 'u': {
- //hexnumbarh - oct is deprecated
- for(int j=0;j<4;j++) {
- CharType c = p_str[idx+j+1];
- if (c==0) {
- r_err_str="Unterminated String";
- return ERR_PARSE_ERROR;
- }
- if (!((c>='0' && c<='9') || (c>='a' && c<='f') || (c>='A' && c<='F'))) {
- r_err_str="Malformed hex constant in string";
- return ERR_PARSE_ERROR;
- }
- CharType v;
- if (c>='0' && c<='9') {
- v=c-'0';
- } else if (c>='a' && c<='f') {
- v=c-'a';
- v+=10;
- } else if (c>='A' && c<='F') {
- v=c-'A';
- v+=10;
- } else {
- ERR_PRINT("BUG");
- v=0;
- }
- res<<=4;
- res|=v;
- }
- idx+=4; //will add at the end anyway
- } break;
- default: {
- r_err_str="Invalid escape sequence";
- return ERR_PARSE_ERROR;
- } break;
- }
- str+=res;
- } else {
- if (p_str[idx]=='\n')
- line++;
- str+=p_str[idx];
- }
- idx++;
- }
- r_token.type=TK_STRING;
- r_token.value=str;
- return OK;
- } break;
- default: {
- if (p_str[idx]<=32) {
- idx++;
- break;
- }
- if (p_str[idx]=='-' || (p_str[idx]>='0' && p_str[idx]<='9')) {
- //a number
- const CharType *rptr;
- double number = String::to_double(&p_str[idx],&rptr);
- idx+=(rptr - &p_str[idx]);
- r_token.type=TK_NUMBER;
- r_token.value=number;
- return OK;
- } else if ((p_str[idx]>='A' && p_str[idx]<='Z') || (p_str[idx]>='a' && p_str[idx]<='z')) {
- String id;
- while((p_str[idx]>='A' && p_str[idx]<='Z') || (p_str[idx]>='a' && p_str[idx]<='z')) {
- id+=p_str[idx];
- idx++;
- }
- r_token.type=TK_IDENTIFIER;
- r_token.value=id;
- return OK;
- } else {
- r_err_str="Unexpected character.";
- return ERR_PARSE_ERROR;
- }
- }
- }
- }
- return ERR_PARSE_ERROR;
- }
- Error VariantConstruct::_parse_value(Variant &value,Token& token,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud) {
- if (token.type==TK_CURLY_BRACKET_OPEN) {
- Dictionary d;
- Error err = _parse_dict(d,p_str,index,p_len,line,r_err_str,p_construct,p_ud);
- if (err)
- return err;
- value=d;
- return OK;
- } else if (token.type==TK_BRACKET_OPEN) {
- Array a;
- Error err = _parse_array(a,p_str,index,p_len,line,r_err_str,p_construct,p_ud);
- if (err)
- return err;
- value=a;
- return OK;
- } else if (token.type==TK_IDENTIFIER) {
- String id = token.value;
- if (id=="true")
- value=true;
- else if (id=="false")
- value=false;
- else if (id=="null")
- value=Variant();
- else {
- r_err_str="Expected 'true','false' or 'null', got '"+id+"'.";
- return ERR_PARSE_ERROR;
- }
- return OK;
- } else if (token.type==TK_NUMBER) {
- value=token.value;
- return OK;
- } else if (token.type==TK_STRING) {
- value=token.value;
- return OK;
- } else {
- r_err_str="Expected value, got "+String(tk_name[token.type])+".";
- return ERR_PARSE_ERROR;
- }
- return ERR_PARSE_ERROR;
- }
- Error VariantConstruct::_parse_array(Array &array,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud) {
- Token token;
- bool need_comma=false;
- while(index<p_len) {
- Error err = _get_token(p_str,index,p_len,token,line,r_err_str);
- if (err!=OK)
- return err;
- if (token.type==TK_BRACKET_CLOSE) {
- return OK;
- }
- if (need_comma) {
- if (token.type!=TK_COMMA) {
- r_err_str="Expected ','";
- return ERR_PARSE_ERROR;
- } else {
- need_comma=false;
- continue;
- }
- }
- Variant v;
- err = _parse_value(v,token,p_str,index,p_len,line,r_err_str,p_construct,p_ud);
- if (err)
- return err;
- array.push_back(v);
- need_comma=true;
- }
- return OK;
- }
- Error VariantConstruct::_parse_dict(Dictionary &dict,const CharType *p_str,int &index, int p_len,int &line,String &r_err_str,Variant::ObjectConstruct* p_construct,void* p_ud) {
- bool at_key=true;
- Variant key;
- Token token;
- bool need_comma=false;
- while(index<p_len) {
- if (at_key) {
- Error err = _get_token(p_str,index,p_len,token,line,r_err_str);
- if (err!=OK)
- return err;
- if (token.type==TK_CURLY_BRACKET_CLOSE) {
- return OK;
- }
- if (need_comma) {
- if (token.type!=TK_COMMA) {
- r_err_str="Expected '}' or ','";
- return ERR_PARSE_ERROR;
- } else {
- need_comma=false;
- continue;
- }
- }
- err = _parse_value(key,token,p_str,index,p_len,line,r_err_str,p_construct,p_ud);
- if (err!=OK)
- return err;
- err = _get_token(p_str,index,p_len,token,line,r_err_str);
- if (err!=OK)
- return err;
- if (token.type!=TK_COLON) {
- r_err_str="Expected ':'";
- return ERR_PARSE_ERROR;
- }
- at_key=false;
- } else {
- Error err = _get_token(p_str,index,p_len,token,line,r_err_str);
- if (err!=OK)
- return err;
- Variant v;
- err = _parse_value(v,token,p_str,index,p_len,line,r_err_str,p_construct,p_ud);
- if (err)
- return err;
- dict[key]=v;
- need_comma=true;
- at_key=true;
- }
- }
- return OK;
- }
- Error VariantConstruct::parse(const String& p_string,Variant& r_ret,String &r_err_str,int &r_err_line,Variant::ObjectConstruct* p_construct,void* p_ud) {
- const CharType *str = p_string.ptr();
- int idx = 0;
- int len = p_string.length();
- Token token;
- r_err_line=0;
- String aux_key;
- Error err = _get_token(str,idx,len,token,r_err_line,r_err_str);
- if (err)
- return err;
- return _parse_value(r_ret,token,str,idx,len,r_err_line,r_err_str,p_construct,p_ud);
- }
|