parser.c File Reference

#include "arc/std/parser/parserlang.h"
#include "arc/std/parser.h"
#include "arc/std/bool.h"
#include "arc/std/errno.h"
#include "arc/std/io.h"
#include "arc/std/lexer.h"
#include "arc/std/vector.h"
#include <stdint.h>
#include <stdlib.h>
#include <string.h>

Go to the source code of this file.

Data Structures
struct	ARC_Parser

Functions
void	ARC_Parser_Create (ARC_Parser **parser, ARC_Array *language, ARC_Parser_InitLexerRulesFn initLexerRulesFn, ARC_ParserData_CreateFn *createDataFn, ARC_ParserData_DestroyFn *destroyDataFn, void *userData)
	creates an ARC_Parser type
void	ARC_Parser_CreateFromVector (ARC_Parser **parser, ARC_Vector *language, ARC_Parser_InitLexerRulesFn initLexerRulesFn, ARC_ParserData_CreateFn *createDataFn, ARC_ParserData_DestroyFn *destroyDataFn, void *userData)
	creates an ARC_Parser type from an arc vector
void	ARC_Parser_CreateFromString (ARC_Parser **parser, ARC_String *languageString, ARC_Parser_InitLexerRulesFn initLexerRulesFn, ARC_Parser_GetStringIdFn getStringIdFn, ARC_ParserData_CreateFn *createDataFn, ARC_ParserData_DestroyFn *destroyDataFn, void *userData)
	creates an ARC_Parser type from a string
void	ARC_Parser_Destroy (ARC_Parser *parser)
	destroys an ARC_Parser type
void	ARC_ParserTagToken_Create (ARC_ParserTagToken **tagToken, uint32_t id)
void	ARC_ParserTagToken_Destroy (ARC_ParserTagToken *tagToken)
void	ARC_ParserTagToken_VectorDestroyDataFn (void *data)
ARC_Bool	ARC_Parser_ParseTag (ARC_Parser *parser, ARC_ParserTagToken *tagToken, uint32_t *lexerIndex)
void	ARC_Parser_Parse (ARC_Parser *parser, ARC_String **data)
void	ARC_Parser_ParseFile (ARC_Parser *parser, ARC_String *path)
void	ARC_Parser_ClearData (ARC_Parser *parser)
void *	ARC_Parser_GetData (ARC_Parser *parser)

Function Documentation

ARC_Parser_ClearData()

void ARC_Parser_ClearData

(

ARC_Parser *

parser

)

Parameters

[in]

parser

Definition at line 358 of file parser.c.

                                             {
    //check if that data exists and the destructor exists to make sure they can be run, set to true to indicate clear
    if(parser->data != NULL && parser->destroyDataFn != NULL){
        (*(parser->destroyDataFn))(parser->data, ARC_True, parser->userData);
    }
 
    //TODO: might want to error here
    parser->data = NULL;
}

References ARC_True, ARC_Parser::data, ARC_Parser::destroyDataFn, and ARC_Parser::userData.

Referenced by ARC_TEST().

ARC_Parser_Create()

void ARC_Parser_Create	(	ARC_Parser **	parser,
		ARC_Array *	language,
		ARC_Parser_InitLexerRulesFn	initLexerRulesFn,
		ARC_ParserData_CreateFn *	createDataFn,
		ARC_ParserData_DestroyFn *	destroyDataFn,
		void *	userData )

creates an ARC_Parser type

@TODO: fix this documentation to reflect changes

@TODO: probs want to move the note to another file

Note

array of tokens for langauge? like ARC_ParserTag tag = { VARIABLE_NAME, //tagId { { 2, CHAR_OR_NUM, VARIABLE_NAME }, { 1, LAMBDA }, }, //components 2 //componentsSize };

Parameters

[out]	parser	ARC_Parser to create
[in]	language	an arry of ARC_ParserLanguageTags defining a langauge
[in]	initLexerRulesFn	a callback used to initalize the token rules the lexer within the parser will use

Definition at line 24 of file parser.c.

                                                                                                                                                                                                              {
    *parser = (ARC_Parser *)malloc(sizeof(ARC_Parser));
 
    //set the language size to 0 and data to NULL in case the language is NULL
    (*parser)->language.size = 0;
    (*parser)->language.data = NULL;
 
    //if the language exists, copy the language
    if(language != NULL){
        (*parser)->language.size = language->size;
        (*parser)->language.data = malloc(sizeof(ARC_ParserTag) * language->size);
 
        for(uint32_t index = 0; index < language->size; index++){
            ARC_ParserTag *languageTag = ((ARC_ParserTag *)language->data) + index;
            ARC_ParserTag *currentTag  = ((ARC_ParserTag *)(*parser)->language.data) + index;
 
            //copy the language tag into the current tag
            currentTag->tagId            = languageTag->tagId;
            currentTag->tokensOrTagsSize = languageTag->tokensOrTagsSize;
 
            //create place to store tokens or tags
            currentTag->tokensOrTags = (uint32_t **)malloc(sizeof(uint32_t *) * languageTag->tokensOrTagsSize);
 
            //copy each or section into the tokensOrTags
            for(uint32_t orIndex = 0; orIndex < languageTag->tokensOrTagsSize; orIndex++){
                currentTag->tokensOrTags[orIndex] = (uint32_t *)malloc(sizeof(uint32_t) * (languageTag->tokensOrTags[orIndex][0] + 1));
 
                for(uint32_t tokenOrTagIndex = 0; tokenOrTagIndex < languageTag->tokensOrTags[orIndex][0] + 1; tokenOrTagIndex++){
                    currentTag->tokensOrTags[orIndex][tokenOrTagIndex] = languageTag->tokensOrTags[orIndex][tokenOrTagIndex];
                }
            }
        }
    }
 
    //create the lexer
    ARC_Lexer_Create(&((*parser)->lexer));
 
    //register instructions to the lexer
    initLexerRulesFn(((*parser)->lexer));
 
    //set the data to null (the parse function is what creates it)
    (*parser)->data = NULL;
 
    //set the userData for the create callback function
    (*parser)->userData = userData;
 
    //init the create function callback with null, then copy the callback if it exists
    (*parser)->createDataFn = NULL;
    if(createDataFn != NULL){
        (*parser)->createDataFn = (ARC_ParserData_CreateFn *)malloc(sizeof(ARC_ParserData_CreateFn));
        *((*parser)->createDataFn) = *createDataFn;
    }
 
    //init the destroy function callback with null, then copy the callback if it exists
    (*parser)->destroyDataFn = NULL;
    if(createDataFn != NULL){
        (*parser)->destroyDataFn = (ARC_ParserData_DestroyFn *)malloc(sizeof(ARC_ParserData_DestroyFn));
        *((*parser)->destroyDataFn) = *destroyDataFn;
    }
}

References ARC_Lexer_Create(), ARC_Array::data, ARC_Parser::language, ARC_Array::size, ARC_ParserTag::tagId, ARC_ParserTag::tokensOrTags, and ARC_ParserTag::tokensOrTagsSize.

Referenced by ARC_Parser_CreateFromVector(), ARC_ParserLang_CreateAsParser(), ARC_TEST(), ARC_TEST(), ARC_TEST(), and ARC_TEST().

ARC_Parser_CreateFromString()

void ARC_Parser_CreateFromString	(	ARC_Parser **	parser,
		ARC_String *	languageString,
		ARC_Parser_InitLexerRulesFn	initLexerRulesFn,
		ARC_Parser_GetStringIdFn	getStringIdFn,
		ARC_ParserData_CreateFn *	createDataFn,
		ARC_ParserData_DestroyFn *	destroyDataFn,
		void *	userData )

creates an ARC_Parser type from a string

@TODO: probs want to move the note to another file

Note

the syntax looks like: <variable> -> CHAR <variableName> EOF <variableName> -> <charOrNum> <variableName> | LAMBDA <charOrNum> -> CHAR | NUM

Parameters

[out]	parser	ARC_Parser to create
[in]	language	an arry of ARC_ParserLanguageTags defining a langauge
[in]	initLexerRulesFn	a callback used to initalize the token rules the lexer within the parser will use

Definition at line 105 of file parser.c.

                                                                                                                                                                                                                                                                       {
    //create the language from the string
    ARC_Parser *parserlangParser;
    ARC_ParserLang_CreateAsParser(&parserlangParser, getStringIdFn);
    if(arc_errno){
        ARC_DEBUG_LOG_ERROR_WITH_VARIABLES("ARC_Parser_CreateFromString(parser, languageString, initLexerRulesFn, getStringIdFn, createDataFn, destroyDataFn, userData), failed to create language from: %s", languageString->data);
        return;
    }
 
    //copy the string because parse will destroy the string that is passed in
    ARC_String *parserLangString;
    ARC_String_Copy(&parserLangString, languageString);
    ARC_Parser_Parse(parserlangParser, &parserLangString);
 
    //create the parser from the parsed language
    ARC_Vector *language = (ARC_Vector *)ARC_Parser_GetData(parserlangParser);
    ARC_Parser_CreateFromVector(parser, language, initLexerRulesFn, createDataFn, destroyDataFn, userData);
 
    //cleanup
    ARC_Parser_Destroy(parserlangParser);
}

References ARC_DEBUG_LOG_ERROR_WITH_VARIABLES, arc_errno, ARC_Parser_CreateFromVector(), ARC_Parser_Destroy(), ARC_Parser_GetData(), ARC_Parser_Parse(), ARC_ParserLang_CreateAsParser(), ARC_String_Copy(), and ARC_String::data.

Referenced by ARC_ParserCSV_CreateAsParser(), and ARC_TEST().

ARC_Parser_CreateFromVector()

void ARC_Parser_CreateFromVector	(	ARC_Parser **	parser,
		ARC_Vector *	language,
		ARC_Parser_InitLexerRulesFn	initLexerRulesFn,
		ARC_ParserData_CreateFn *	createDataFn,
		ARC_ParserData_DestroyFn *	destroyDataFn,
		void *	userData )

creates an ARC_Parser type from an arc vector

Parameters

[out]	parser	ARC_Parser to create
[in]	language	an vector of ARC_ParserLanguageTags defining a langauge
[in]	initLexerRulesFn	a callback used to initalize the token rules the lexer within the parser will use

Definition at line 85 of file parser.c.

                                                                                                                                                                                                                         {
    //creates the variables to copy the vector into
    const uint32_t languageSize = ARC_Vector_GetSize(language);
    ARC_ParserTag languageArray[languageSize];
 
    //copy the language from a vector into an array
    for(uint32_t index = 0; index < languageSize; index++){
        languageArray[index] = *(ARC_ParserTag *)ARC_Vector_Get(language, index);
    }
 
    //set the vector data as an ARC_Array
    ARC_Array languageAsArray = {
        languageSize,
        languageArray
    };
 
    //create the parser
    ARC_Parser_Create(parser, &languageAsArray, initLexerRulesFn, createDataFn, destroyDataFn, userData);
}

References ARC_Parser_Create(), ARC_Vector_Get(), ARC_Vector_GetSize(), and languageArray.

Referenced by ARC_Parser_CreateFromString(), and ARC_TEST().

ARC_Parser_Destroy()

void ARC_Parser_Destroy

(

ARC_Parser *

parser

)

destroys an ARC_Parser type

Parameters

[in]

parser

ARC_Parser to free

Definition at line 127 of file parser.c.

                                           {
    //clear all the copied token or tags from memory
    for(uint32_t index = 0; index < parser->language.size; index++){
        ARC_ParserTag *currentTag  = ((ARC_ParserTag *)parser->language.data) + index;
 
        //free the orIndex vlues
        for(uint32_t orIndex = 0; orIndex < currentTag->tokensOrTagsSize; orIndex++){
            free(currentTag->tokensOrTags[orIndex]);
        }
 
        //free the tokens or tags
        free(currentTag->tokensOrTags);
    }
 
    //free the creation function callback
    if(parser->createDataFn != NULL){
        free(parser->createDataFn);
    }
 
    //do the same thing as clear but this time pass in the userData as well to clean that up
    if(parser->destroyDataFn != NULL){
        //set to false to indicate full destroy instead of clear
        (*(parser->destroyDataFn))(parser->data, ARC_False, parser->userData);
        free(parser->destroyDataFn);
    }
 
    //clear the copied language from memory
    free(parser->language.data);
 
    ARC_Lexer_Destroy(parser->lexer);
 
    free(parser);
}

References ARC_False, ARC_Lexer_Destroy(), ARC_Parser::createDataFn, ARC_Array::data, ARC_Parser::data, ARC_Parser::destroyDataFn, ARC_Parser::language, ARC_Parser::lexer, ARC_Array::size, ARC_ParserTag::tokensOrTags, ARC_ParserTag::tokensOrTagsSize, and ARC_Parser::userData.

Referenced by ARC_Parser_CreateFromString(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), and ARC_TEST().

ARC_Parser_GetData()

void * ARC_Parser_GetData

(

ARC_Parser *

parser

)

Parameters

[in]

parser

Definition at line 368 of file parser.c.

                                            {
    return parser->data;
}

References ARC_Parser::data.

Referenced by ARC_Parser_CreateFromString(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), and ARC_TEST().

ARC_Parser_Parse()

void ARC_Parser_Parse	(	ARC_Parser *	parser,
		ARC_String **	data )

Parameters

[in]	parser
	[in/out]	data the string to parse, will be freed and set to NULL by the end of this function

Definition at line 297 of file parser.c.

                                                            {
    //make sure the parser has a language
    if(parser->language.size == 0){
        ARC_DEBUG_LOG_ERROR("ARC_Parser_Parse(parser, data), no parser language defined");
        return;
    }
 
    //lex the subdata
    ARC_Lexer_LexString(parser->lexer, data);
    if(arc_errno){
        ARC_DEBUG_LOG_ERROR("ARC_Parser_Parse(parser, data), could not lex the given data");
        ARC_Lexer_Clear(parser->lexer);
        return;
    }
 
    //set the lexer index to start and get the first tag
    uint32_t lexerIndex = 0;
    ARC_ParserTag *startTag = parser->language.data;
 
    //setup a tag token that will be passed to the creation callback on success
    ARC_ParserTagToken *tagToken;
    ARC_ParserTagToken_Create(&tagToken, startTag->tagId);
 
    //TODO: handle error checks for if parsing fails
    //recursivly parse from the inital start tag
    ARC_Bool parsed = ARC_Parser_ParseTag(parser, tagToken, &lexerIndex);
    ARC_Bool allTokensParsed = lexerIndex == ARC_Lexer_GetTokensSize(parser->lexer);
 
    //error if anything went wrong
    if(parsed == ARC_False || allTokensParsed == ARC_False || arc_errno){
        ARC_Lexer_Clear(parser->lexer);
        ARC_ParserTagToken_Destroy(tagToken);
 
        arc_errno = ARC_ERRNO_DATA;
        ARC_DEBUG_LOG_ERROR_WITH_VARIABLES("ARC_Parser_Parse(parser, data), could not parse the given data at lexer index: %u", lexerIndex);
        return;
    }
 
    //create the data if the creation callback exists
    if(parser->createDataFn != NULL){
        (*(parser->createDataFn))(&(parser->data), tagToken, parser->userData);
    }
 
    ARC_ParserTagToken_Destroy(tagToken);
    //cleanup
    ARC_Lexer_Clear(parser->lexer);
}

References ARC_Bool, ARC_DEBUG_LOG_ERROR, ARC_DEBUG_LOG_ERROR_WITH_VARIABLES, arc_errno, ARC_ERRNO_DATA, ARC_False, ARC_Lexer_Clear(), ARC_Lexer_GetTokensSize(), ARC_Lexer_LexString(), ARC_Parser_ParseTag(), ARC_ParserTagToken_Create(), ARC_ParserTagToken_Destroy(), ARC_Parser::createDataFn, ARC_Array::data, ARC_Parser::data, ARC_Parser::language, ARC_Parser::lexer, ARC_Array::size, ARC_ParserTag::tagId, and ARC_Parser::userData.

Referenced by ARC_Parser_CreateFromString(), ARC_Parser_ParseFile(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), ARC_TEST(), and ARC_TEST().

ARC_Parser_ParseFile()

void ARC_Parser_ParseFile	(	ARC_Parser *	parser,
		ARC_String *	path )

Parameters

[in]	parser
[in]	language

Definition at line 345 of file parser.c.

                                                               {
    ARC_String *fileString;
    ARC_IO_FileToStr(path, &fileString);
 
    if(fileString == NULL){
        arc_errno = ARC_ERRNO_NULL;
        ARC_DEBUG_LOG_ERROR_WITH_VARIABLES("ARC_Parser_ParseFile(parser, data), could not read file \"%s\"", path->data);
        return;
    }
 
    ARC_Parser_Parse(parser, &fileString);
}

References ARC_DEBUG_LOG_ERROR_WITH_VARIABLES, arc_errno, ARC_ERRNO_NULL, ARC_IO_FileToStr(), ARC_Parser_Parse(), and ARC_String::data.

ARC_Parser_ParseTag()

ARC_Bool ARC_Parser_ParseTag	(	ARC_Parser *	parser,
		ARC_ParserTagToken *	tagToken,
		uint32_t *	lexerIndex )

Definition at line 188 of file parser.c.

                                                                                                    {
    //get the current tag
    ARC_ParserTag *tag = NULL;
    for(uint32_t index = 0; index < parser->language.size; index++){
        ARC_ParserTag *foundTag = ((ARC_ParserTag *)parser->language.data) + index;
        if(foundTag->tagId == tagToken->id){
            tag = foundTag;
            break;
        }
    }
 
    //if the tag was not found can't do much, so throw an error
    if(tag == NULL){
        arc_errno = ARC_ERRNO_NULL;
        ARC_DEBUG_LOG_ERROR_WITH_VARIABLES("ARC_Parser_ParseTag(parser, subdata, tagId), could not find tag with id: %u", tagToken->id);
        return ARC_False;
    }
 
    //create a vector of tag token to use if a rule is validated, a comparison function is not needed as it will be iterated through
    ARC_Vector_DestroyDataFn destroyTokenTagFn = ARC_ParserTagToken_VectorDestroyDataFn;
    ARC_Vector_Create(&(tagToken->tagTokens), NULL, &destroyTokenTagFn);
 
    //loop through each or section of the tags and tokens
    for(uint32_t orIndex = 0; orIndex < tag->tokensOrTagsSize; orIndex++){
        //reset the tag tokens for each or index
        ARC_Vector_Clear(tagToken->tagTokens);
 
        //loop through each token or tag to check if the lexed data matches
        uint32_t lexerCheckIndex = *lexerIndex;
        ARC_Bool foundRule = ARC_True;
        for(uint32_t tokenOrTagIndex = 1; tokenOrTagIndex < tag->tokensOrTags[orIndex][0] + 1; tokenOrTagIndex++){
            //get next tag id to check
            uint32_t nextTagId = tag->tokensOrTags[orIndex][tokenOrTagIndex];
 
            //check if it is lambda (can return safely)
            if(nextTagId == ARC_PARSER_TAG_LAMBDA){
                break;
            }
 
            //if the value isn't a token it is a tag, so recurs if it isn't a token
            ARC_Bool isToken = ARC_Lexer_IsTokenId(parser->lexer, nextTagId);
            if(isToken == ARC_False){
                //create a temporary lexer index in case the rule does not exist
                uint32_t tempLexerCheckIndex = lexerCheckIndex;
 
                //create tag token for if the rule works
                ARC_ParserTagToken *nextTagToken;
                ARC_ParserTagToken_Create(&nextTagToken, nextTagId);
 
                //check if the tag works if not break to continue checking next or index
                foundRule = ARC_Parser_ParseTag(parser, nextTagToken, &tempLexerCheckIndex);
                if(foundRule == ARC_False){
                    //clean up the tag token
                    ARC_ParserTagToken_Destroy(nextTagToken);
                    break;
                }
 
                //add the tag token because rule was found
                ARC_Vector_Add(tagToken->tagTokens, nextTagToken);
 
                //increase the lexer check index as a recursed rule was found, and continue checking
                lexerCheckIndex = tempLexerCheckIndex;
 
                continue;
            }
 
            //check if there is another token that can be used
            if(lexerCheckIndex >= ARC_Lexer_GetTokensSize(parser->lexer)){
                //out of tokens to the current or does not work, so break
                foundRule = ARC_False;
                break;
            }
 
            //get the next token in the lexer and increment the lexers index
            ARC_LexerToken *token = ARC_Lexer_GetToken(parser->lexer, lexerCheckIndex);
            lexerCheckIndex++;
 
            //if the token rule does not match the current token in the current or statement the token rule could not be found for the current or index so break
            if(token->rule != tag->tokensOrTags[orIndex][tokenOrTagIndex]){
                foundRule = ARC_False;
                break;
            }
 
            //the rule was a match so create a tag token to store the token in
            ARC_ParserTagToken *nextTagToken;
            ARC_ParserTagToken_Create(&nextTagToken, nextTagId);
            nextTagToken->token = token;
 
            //add the token to the tag tokens
            ARC_Vector_Add(tagToken->tagTokens, nextTagToken);
        }
 
        //if the rule is found we don't need to check anymore so we can return out
        if(foundRule == ARC_True){
            *lexerIndex = lexerCheckIndex;
 
            //cleanup
            return ARC_True;
        }
    }
 
    //cleanup
    ARC_Vector_Destroy(tagToken->tagTokens);
    tagToken->tagTokens = NULL;
 
    //no rule was found, so return false
    return ARC_False;
}

References ARC_Bool, ARC_DEBUG_LOG_ERROR_WITH_VARIABLES, arc_errno, ARC_ERRNO_NULL, ARC_False, ARC_Lexer_GetToken(), ARC_Lexer_GetTokensSize(), ARC_Lexer_IsTokenId(), ARC_Parser_ParseTag(), ARC_PARSER_TAG_LAMBDA, ARC_ParserTagToken_Create(), ARC_ParserTagToken_Destroy(), ARC_ParserTagToken_VectorDestroyDataFn(), ARC_True, ARC_Vector_Add(), ARC_Vector_Clear(), ARC_Vector_Create(), ARC_Vector_Destroy(), ARC_Array::data, ARC_ParserTagToken::id, ARC_Parser::language, ARC_Parser::lexer, ARC_LexerToken::rule, ARC_Array::size, ARC_ParserTag::tagId, ARC_ParserTagToken::tagTokens, ARC_ParserTagToken::token, ARC_ParserTag::tokensOrTags, and ARC_ParserTag::tokensOrTagsSize.

Referenced by ARC_Parser_Parse(), and ARC_Parser_ParseTag().

ARC_ParserTagToken_Create()

void ARC_ParserTagToken_Create	(	ARC_ParserTagToken **	tagToken,
		uint32_t	id )

Definition at line 163 of file parser.c.

                                                                          {
    *tagToken = (ARC_ParserTagToken *)malloc(sizeof(ARC_ParserTagToken));
 
    (*tagToken)->id        = id;
    (*tagToken)->token     = NULL;
    (*tagToken)->tagTokens = NULL;
}

References ARC_ParserTagToken::id.

Referenced by ARC_Parser_Parse(), and ARC_Parser_ParseTag().

ARC_ParserTagToken_Destroy()

void ARC_ParserTagToken_Destroy

(

ARC_ParserTagToken *

tagToken

)

Definition at line 172 of file parser.c.

                                                             {
    //destroy the tag token vector
    if(tagToken->tagTokens != NULL){
        ARC_Vector_Destroy(tagToken->tagTokens);
    }
 
    free(tagToken);
}

References ARC_Vector_Destroy(), and ARC_ParserTagToken::tagTokens.

Referenced by ARC_Parser_Parse(), ARC_Parser_ParseTag(), and ARC_ParserTagToken_VectorDestroyDataFn().

ARC_ParserTagToken_VectorDestroyDataFn()

void ARC_ParserTagToken_VectorDestroyDataFn

(

void *

data

)

Definition at line 182 of file parser.c.

                                                       {
    //we can just use the destroy function with casted data
    ARC_ParserTagToken_Destroy((ARC_ParserTagToken *)data);
}

References ARC_ParserTagToken_Destroy().

Referenced by ARC_Parser_ParseTag().