2024-11-06 06:45:04 -07:00
|
|
|
#include "arc/std/parser/parserlang.h"
|
|
|
|
|
#include "arc/std/lexer.h"
|
|
|
|
|
#include "arc/std/parser.h"
|
|
|
|
|
#include "arc/std/string.h"
|
2024-11-20 10:27:17 -07:00
|
|
|
#include <stddef.h>
|
2024-11-06 06:45:04 -07:00
|
|
|
|
|
|
|
|
void ARC_ParserLang_InitLexerRulesFn(ARC_Lexer *lexer){
|
|
|
|
|
//null
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_NULL, 0));
|
|
|
|
|
|
|
|
|
|
//number
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharOrBetween(ARC_PARSERLANG_TOKEN_NUMBER, '0', '9'));
|
|
|
|
|
|
|
|
|
|
//alpha char
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharOrBetween(ARC_PARSERLANG_TOKEN_ALPHA_LOWER_CHAR, 'a', 'z'));
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharOrBetween(ARC_PARSERLANG_TOKEN_ALPHA_UPPER_CHAR, 'A', 'Z'));
|
|
|
|
|
|
|
|
|
|
//whitespace
|
|
|
|
|
ARC_String *whitespaceString;
|
|
|
|
|
ARC_String_CreateWithStrlen(&whitespaceString, " \t");
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharInStringRule(ARC_LEXER_TOKEN_WHITESPACE, whitespaceString));
|
|
|
|
|
ARC_String_Destroy(whitespaceString);
|
|
|
|
|
|
|
|
|
|
//single char tokens
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_NEWLINE_ID , ARC_PARSERLANG_TOKEN_NEWLINE_CHAR ));
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_LESS_THAN_ID , ARC_PARSERLANG_TOKEN_LESS_THAN_CHAR ));
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_GREATER_THAN_ID, ARC_PARSERLANG_TOKEN_GREATER_THAN_CHAR));
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_OR_ID , ARC_PARSERLANG_TOKEN_OR_CHAR ));
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_PARSERLANG_TOKEN_UNDERSCORE_ID , ARC_PARSERLANG_TOKEN_UNDERSCORE_CHAR ));
|
|
|
|
|
|
|
|
|
|
//arrow
|
|
|
|
|
ARC_String *arrowString;
|
|
|
|
|
ARC_String_CreateWithStrlen(&arrowString, ARC_PARSERLANG_TOKEN_ARROW_CSTRING);
|
|
|
|
|
ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchStringRule(ARC_PARSERLANG_TOKEN_ARROW_ID, arrowString));
|
|
|
|
|
ARC_String_Destroy(arrowString);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void ARC_Parser_CreateAsParserLang(ARC_Parser **parser){
|
2024-11-14 00:07:56 -07:00
|
|
|
//<line> -> <body> NEWLINE <line> | <body> | NEWLINE <line> | LAMBDA
|
|
|
|
|
uint32_t *line[] = { (uint32_t[]){ 3, ARC_PARSERLANG_BODY, ARC_PARSERLANG_TOKEN_NEWLINE_ID, ARC_PARSERLANG_LINE }, (uint32_t[]){ 1, ARC_PARSERLANG_BODY }, (uint32_t[]){ 2, ARC_PARSERLANG_TOKEN_NEWLINE_ID, ARC_PARSERLANG_LINE }, (uint32_t[]){ 1, ARC_PARSERLANG_LAMBDA } };
|
2024-11-06 06:45:04 -07:00
|
|
|
|
|
|
|
|
//<body> -> <tag> WHITESPACE ARROW WHITESPACE <arguments>
|
|
|
|
|
uint32_t *body[] = { (uint32_t[]){ 5, ARC_PARSERLANG_TAG, ARC_PARSERLANG_TOKEN_WHITESPACE, ARC_PARSERLANG_TOKEN_ARROW_ID, ARC_PARSERLANG_TOKEN_WHITESPACE, ARC_PARSERLANG_ARGUMENTS } };
|
|
|
|
|
|
|
|
|
|
//<arguments> -> <argument> WHITESPACE OR WHITESPACE <arguments> | <tagOrConstant>
|
|
|
|
|
uint32_t *arguments[] = { (uint32_t[]){ 5, ARC_PARSERLANG_ARGUMENT, ARC_PARSERLANG_TOKEN_WHITESPACE, ARC_PARSERLANG_TOKEN_OR_ID, ARC_PARSERLANG_TOKEN_WHITESPACE, ARC_PARSERLANG_ARGUMENTS }, (uint32_t[]){ 1, ARC_PARSERLANG_TAG_OR_CONSTANT } };
|
|
|
|
|
|
|
|
|
|
//<argument> -> <tagOrConstant> WHITESPACE <argument> | <tagOrConstant>
|
|
|
|
|
uint32_t *argument[] = { (uint32_t[]){ 3, ARC_PARSERLANG_TAG_OR_CONSTANT, ARC_PARSERLANG_TOKEN_WHITESPACE, ARC_PARSERLANG_ARGUMENT }, (uint32_t[]){ 1, ARC_PARSERLANG_TAG_OR_CONSTANT } };
|
|
|
|
|
|
|
|
|
|
//<tagOrConstant> -> <parserLangageTag> | <constant>
|
|
|
|
|
uint32_t *tagOrConstant[] = { (uint32_t[]){ 1, ARC_PARSERLANG_TAG }, (uint32_t[]){ 1, ARC_PARSERLANG_CONSTANT } };
|
|
|
|
|
|
|
|
|
|
//<constant> -> ALPHA_UPPER_CHAR <constantBody>
|
|
|
|
|
uint32_t *constant[] = { (uint32_t[]){ 2, ARC_PARSERLANG_TOKEN_ALPHA_UPPER_CHAR, ARC_PARSERLANG_CONSTANT_BODY } };
|
|
|
|
|
|
|
|
|
|
//<constantBody> -> <constantChar> <constantBody> | LAMBDA
|
|
|
|
|
uint32_t *constantBody[] = { (uint32_t[]){ 2, ARC_PARSERLANG_CONSTANT_CHAR, ARC_PARSERLANG_CONSTANT_BODY }, (uint32_t[]){ 1, ARC_PARSERLANG_LAMBDA } };
|
|
|
|
|
|
|
|
|
|
//<constantChar> -> ALPHA_UPPER_CHAR | UNDERSCORE
|
|
|
|
|
uint32_t *constantChar[] = { (uint32_t[]){ 2, ARC_PARSERLANG_TOKEN_ALPHA_UPPER_CHAR, ARC_PARSERLANG_TOKEN_UNDERSCORE_ID } };
|
|
|
|
|
|
|
|
|
|
//<tag> -> LESS_THAN <variable> GREATER_THAN
|
|
|
|
|
uint32_t *tag[] = { (uint32_t[]){ 3, ARC_PARSERLANG_TOKEN_LESS_THAN_ID, ARC_PARSERLANG_VARIABLE, ARC_PARSERLANG_TOKEN_GREATER_THAN_ID } };
|
|
|
|
|
|
|
|
|
|
//<variable> -> <alphaChar> <variableBody> | UNDERSCORE <variableBody>
|
|
|
|
|
uint32_t *variable[] = { (uint32_t[]){ 2, ARC_PARSERLANG_ALPHA_CHAR, ARC_PARSERLANG_VARIABLE_BODY }, (uint32_t[]){ 2, ARC_PARSERLANG_TOKEN_UNDERSCORE_ID, ARC_PARSERLANG_VARIABLE_BODY } };
|
|
|
|
|
|
|
|
|
|
//<variableBody> -> <variableChar> <variableBody> | LAMBDA
|
|
|
|
|
uint32_t *variableBody[] = { (uint32_t[]){ 2, ARC_PARSERLANG_VARIABLE_CHAR, ARC_PARSERLANG_VARIABLE_BODY }, (uint32_t[]){ 1, ARC_PARSERLANG_LAMBDA } };
|
|
|
|
|
|
|
|
|
|
//<variableChar> -> <alphaChar> | NUMBER | UNDERSCORE
|
|
|
|
|
uint32_t *variableChar[] = { (uint32_t[]){ 1, ARC_PARSERLANG_ALPHA_CHAR }, (uint32_t[]){ 1, ARC_PARSERLANG_TOKEN_NUMBER }, (uint32_t[]){ 1, ARC_PARSERLANG_TOKEN_UNDERSCORE_ID } };
|
|
|
|
|
|
|
|
|
|
//<alphaChar> -> ALPHA_LOWER_CHAR | ALPHA_UPPER_CHAR
|
|
|
|
|
uint32_t *alphaChar[] = { (uint32_t[]){ 1, ARC_PARSERLANG_TOKEN_ALPHA_LOWER_CHAR }, (uint32_t[]){ 1, ARC_PARSERLANG_TOKEN_ALPHA_UPPER_CHAR }};
|
|
|
|
|
|
|
|
|
|
ARC_ParserLanguageTag parserLangTags[13] = {
|
2024-11-20 10:27:17 -07:00
|
|
|
{ ARC_PARSERLANG_LINE , line , 4, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_BODY , body , 1, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_ARGUMENTS , arguments , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_ARGUMENT , argument , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_TAG_OR_CONSTANT, tagOrConstant, 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_CONSTANT , constant , 1, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_CONSTANT_BODY , constantBody , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_CONSTANT_CHAR , constantChar , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_TAG , tag , 1, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_VARIABLE , variable , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_VARIABLE_BODY , variableBody , 2, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_VARIABLE_CHAR , variableChar , 3, NULL },
|
|
|
|
|
{ ARC_PARSERLANG_ALPHA_CHAR , alphaChar , 2, NULL }
|
2024-11-06 06:45:04 -07:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
ARC_Array parserLanguageArray = {
|
|
|
|
|
13, //size
|
|
|
|
|
parserLangTags //data
|
|
|
|
|
};
|
|
|
|
|
|
2024-11-20 10:27:17 -07:00
|
|
|
//TODO: add the create, destroy, and add callbacks
|
|
|
|
|
ARC_Parser_Create(parser, &parserLanguageArray, ARC_ParserLang_InitLexerRulesFn, NULL, NULL);
|
2024-11-06 06:45:04 -07:00
|
|
|
}
|
