older lexer stuff
This commit is contained in:
herbglitch
parent
7bd7cc4aa5
commit
380e74a0e6
3 changed files with 211 additions and 72 deletions
|
|
@ -5,7 +5,6 @@
|
|||
extern "C" {
|
||||
#endif
|
||||
|
||||
#include "arc/std/bool.h"
|
||||
#include "arc/std/string.h"
|
||||
#include <stdint.h>
|
||||
|
||||
|
|
@ -25,16 +24,15 @@ typedef struct ARC_LexerToken {
|
|||
/**
|
||||
* @brief checks to see if a string is a type of token
|
||||
*
|
||||
* @note do not set tokenData if this function returns ARC_False, doing so will create a memory leak
|
||||
* @note do not set tokenData if this function returns 0, doing so will create a memory leak
|
||||
*
|
||||
* @param[in/out] string a string to be checked to see if it matches a token,
|
||||
* this needs to srip the token out for the lexer to avoid an infinite loop
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] string a string to be checked to see if it matches a token
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
*
|
||||
* @return if a token was successfully found ARC_True, otherwise ARC_False
|
||||
* @return the size of the token found, or 0 if the token was not found
|
||||
*/
|
||||
typedef ARC_Bool (* ARC_LexerTokenRule_AutomataFn)(ARC_String **string, ARC_String **tokenData, void *automataData);
|
||||
typedef uint32_t (* ARC_LexerTokenRule_AutomataFn)(ARC_String **tokenData, ARC_String *string, void *automataData);
|
||||
|
||||
/**
|
||||
* @brief a callback function to clean up ARC_LexerTokenRule's automataData
|
||||
|
|
@ -124,14 +122,13 @@ uint32_t ARC_Lexer_GetTokensSize(ARC_Lexer *lexer);
|
|||
* @note this is intended as a helper callback
|
||||
* @note this function is a ARC_Lexer_AutomataFn callback
|
||||
*
|
||||
* @param[in/out] string a string to be checked to see if it matches a token,
|
||||
* this needs to srip the token out for the lexer to avoid an infinite loop
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] string a string to be checked to see if it matches a token
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
*
|
||||
* @return if a token was successfully found ARC_True, otherwise ARC_False
|
||||
* @return the size of the token found, or 0 if the token was not found
|
||||
*/
|
||||
ARC_Bool ARC_Lexer_AutomataMatchCharFn(ARC_String **string, ARC_String **tokenData, void *automataData);
|
||||
uint32_t ARC_Lexer_AutomataMatchCharFn(ARC_String **tokenData, ARC_String *string, void *automataData);
|
||||
|
||||
/**
|
||||
* @brief checks if the substring automataData as an ARC_String matches the first part of string
|
||||
|
|
@ -139,14 +136,27 @@ ARC_Bool ARC_Lexer_AutomataMatchCharFn(ARC_String **string, ARC_String **tokenDa
|
|||
* @note this is intended as a helper callback
|
||||
* @note this function is a ARC_Lexer_AutomataFn callback
|
||||
*
|
||||
* @param[in/out] string a string to be checked to see if it matches a token,
|
||||
* this needs to srip the token out for the lexer to avoid an infinite loop
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] string a string to be checked to see if it matches a token
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
*
|
||||
* @return if a token was successfully found ARC_True, otherwise ARC_False
|
||||
* @return the size of the token found, or 0 if the token was not found
|
||||
*/
|
||||
ARC_Bool ARC_Lexer_AutomataMatchStringFn(ARC_String **string, ARC_String **tokenData, void *automataData);
|
||||
uint32_t ARC_Lexer_AutomataMatchStringFn(ARC_String **tokenData, ARC_String *string, void *automataData);
|
||||
|
||||
/**
|
||||
* @brief checks if the first part of string is a character in substring
|
||||
*
|
||||
* @note this is intended as a helper callback
|
||||
* @note this function is a ARC_Lexer_AutomataFn callback
|
||||
*
|
||||
* @param[out] tokenData a place to store token data (like a variable name), can be NULL if not needed
|
||||
* @param[in] string a string to be checked to see if it matches a token
|
||||
* @param[in] automataData any data that needs to be used for the ARC_Lexer_AutomataFn
|
||||
*
|
||||
* @return the size of the token found, or 0 if the token was not found
|
||||
*/
|
||||
uint32_t ARC_Lexer_AutomataMatchCharInStringFn(ARC_String **tokenData, ARC_String *string, void *automataData);
|
||||
|
||||
/**
|
||||
* @brief creates a ARC_LexerTokenRule with a given id and character
|
||||
|
|
@ -173,6 +183,28 @@ ARC_LexerTokenRule ARC_LexerTokenRule_CreateAndReturnMatchCharRule(uint32_t id,
|
|||
*/
|
||||
ARC_LexerTokenRule ARC_LexerTokenRule_CreateAndReturnMatchStringRule(uint32_t id, ARC_String *string);
|
||||
|
||||
/**
|
||||
* @brief creates a ARC_LexerTokenRule with a given id and string
|
||||
*
|
||||
* @note this is intended as a helper funtion
|
||||
* #note string will not be freed (it will be copied and the copy will be freed)
|
||||
*
|
||||
* @param[in] id a tokens id (basically the token value)
|
||||
* @param[in] character the string to match against, will be copied
|
||||
*
|
||||
* @return a token rule based in the id and string
|
||||
*/
|
||||
ARC_LexerTokenRule ARC_LexerTokenRule_CreateAndReturnMatchCharInStringRule(uint32_t id, ARC_String *string);
|
||||
|
||||
/**
|
||||
* @brief basic tokens
|
||||
*/
|
||||
#define ARC_LEXER_TOKEN_NULL 0
|
||||
#define ARC_LEXER_TOKEN_EOF 1
|
||||
#define ARC_LEXER_TOKEN_NUMBER 2
|
||||
#define ARC_LEXER_TOKEN_ALPHACHAR 3
|
||||
#define ARC_LEXER_TOKEN_WHITESPACE 4
|
||||
|
||||
/**
|
||||
* @brief basic token type ids, chars, and tags
|
||||
*/
|
||||
|
|
|
|||
|
|
@ -0,0 +1,63 @@
|
|||
#ifndef ARC_STD_PARSER_H_
|
||||
#define ARC_STD_PARSER_H_
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
#include "arc/std/string.h"
|
||||
|
||||
/**
|
||||
* @brief a parser type
|
||||
*/
|
||||
typedef struct ARC_Parser ARC_Parser;
|
||||
|
||||
/**
|
||||
* @brief a parser node
|
||||
*/
|
||||
typedef struct ARC_ParserNode ARC_ParserNode;
|
||||
|
||||
/**
|
||||
* @brief creates an ARC_Parser type
|
||||
*
|
||||
* @param[out] parser
|
||||
* @param[in] language ..., can be NULL
|
||||
*/
|
||||
void ARC_Parser_Create(ARC_Parser **parser, ARC_String *language);
|
||||
|
||||
/**
|
||||
* @brief destroys an ARC_Parser type
|
||||
*
|
||||
* @param[in] parser ARC_Parser to free
|
||||
*/
|
||||
void ARC_Parser_Destroy(ARC_Parser *parser);
|
||||
|
||||
/**
|
||||
* @brief sets the definition of the parser, the language itself is parsed and will throw an error if invalid
|
||||
*
|
||||
* @param[in] parser ARC_Parser to set the language to
|
||||
* @param[in] language the language as a string the parser should use
|
||||
*/
|
||||
void ARC_Parser_SetLanguage(ARC_Parser *parser, ARC_String *language);
|
||||
|
||||
/**
|
||||
* @brief sets the definition of the parser, the language itself is parsed and will throw an error if invalid
|
||||
*
|
||||
* @param[in] parser ARC_Parser to set the language to
|
||||
* @param[in] language the language as a string the parser should use
|
||||
*/
|
||||
void ARC_Parser_Parse(ARC_Parser *parser, ARC_String *data);
|
||||
|
||||
/**
|
||||
* @brief sets the definition of the parser, the language itself is parsed and will throw an error if invalid
|
||||
*
|
||||
* @param[in] parser ARC_Parser to set the language to
|
||||
* @param[in] language the language as a string the parser should use
|
||||
*/
|
||||
void ARC_Parser_ParseFile(ARC_Parser *parser, ARC_String *path);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif // !ARC_STD_LEXER_H_
|
||||
Loading…
Add table
Add a link
Reference in a new issue