herbglitch/archeus
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

got tags parsing in parserlang, though constants are still borked

Browse source
This commit is contained in:
herbglitch 2024-12-02 18:34:19 -07:00
parent 16821e8411
commit c9a06e50ca
2 changed files with 111 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

136
src/std/parser/parserlang.c
View file

@ -39,24 +39,15 @@ void ARC_ParserLang_InitLexerRulesFn(ARC_Lexer *lexer){
ARC_String_Destroy(arrowString); ARC_String_Destroy(arrowString);
} }
/* void ARC_ParserLang_VectorDestroyVector(void *data){
<line> -> <body> NEWLINE <line> | <body> | NEWLINE <line> | LAMBDA ARC_Vector *vector = (ARC_Vector *)data;
<body> -> <tag> WHITESPACE ARROW WHITESPACE <arguments> ARC_Vector_Destroy(vector);
}
<arguments> -> <argument> WHITESPACE OR WHITESPACE <arguments> | <tagOrConstant> void ARC_ParserLang_VectorDestroyUInt32(void *data){
<argument> -> <tagOrConstant> WHITESPACE <argument> | <tagOrConstant> uint32_t *uint = (uint32_t *)data;
<tagOrConstant> -> <tag> | <constant> free(uint);
}
<constant> -> ALPHA_UPPER_CHAR <constantBody>
<constantBody> -> <constantChar> <constantBody> | LAMBDA
<constantChar> -> ALPHA_UPPER_CHAR | UNDERSCORE
<tag> -> LESS_THAN <variable> GREATER_THAN
<variable> -> <alphaChar> <variableBody> | UNDERSCORE <variableBody>
<variableBody> -> <variableChar> <variableBody> | LAMBDA
<variableChar> -> <alphaChar> | NUMBER | UNDERSCORE
<alphaChar> -> ALPHA_LOWER_CHAR | ALPHA_UPPER_CHAR
*/
void ARC_ParserLang_VectorDestroyParserTagFn(void *data){ void ARC_ParserLang_VectorDestroyParserTagFn(void *data){
ARC_ParserTag *currentTag = (ARC_ParserTag *)data; ARC_ParserTag *currentTag = (ARC_ParserTag *)data;
@ -113,11 +104,98 @@ void ARC_ParserLangParsedData_CreateTagString(ARC_String **tagString, ARC_Parser
} }
} }
/*
<argument> -> <tagOrConstant> WHITESPACE <argument> | <tagOrConstant>
*/
void ARC_ParserLangParsedData_GetArgumentTag(ARC_Vector *orTokensOrTags, ARC_ParserTagToken *tagToken, ARC_ParserLang_GetIdFn *getIdFn){
for(uint32_t index = 0; index < ARC_Vector_GetSize(tagToken->tagTokens); index++){
ARC_ParserTagToken *childTagToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, index);
//switch variables
ARC_String *tagOrConstantString = NULL;
uint32_t *id = NULL;
switch(childTagToken->id){
case ARC_PARSERLANG_ARGUMENT:
//recurse to check all the arguments
ARC_ParserLangParsedData_GetArgumentTag(orTokensOrTags, childTagToken, getIdFn);
continue;
case ARC_PARSERLANG_TAG_OR_CONSTANT:
//get the id of the tag/constant
ARC_ParserLangParsedData_RecurseStringAdd(&tagOrConstantString, childTagToken);
id = (uint32_t *)malloc(sizeof(uint32_t));
*id = (*getIdFn)(tagOrConstantString);
//add the id to the matching or vector
ARC_Vector_Add(orTokensOrTags, (void *)id);
//cleanup
ARC_String_Destroy(tagOrConstantString);
continue;
default:
//this should only be whitespace
continue;
}
}
}
/*
<arguments> -> <argument> WHITESPACE OR WHITESPACE <arguments> | <tagOrConstant>
*/
void ARC_ParserLangParsedData_GetArgumentsTag(ARC_Vector *tokensOrTags, ARC_ParserTagToken *tagToken, ARC_ParserLang_GetIdFn *getIdFn){ void ARC_ParserLangParsedData_GetArgumentsTag(ARC_Vector *tokensOrTags, ARC_ParserTagToken *tagToken, ARC_ParserLang_GetIdFn *getIdFn){
for(uint32_t index = 0; index < ARC_Vector_GetSize(tagToken->tagTokens); index++){ for(uint32_t index = 0; index < ARC_Vector_GetSize(tagToken->tagTokens); index++){
//TODO: write this ARC_ParserTagToken *childTagToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, index);
//ARC_ParserTagToken *childTagToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, index);
//switch variables
ARC_String *tagOrConstantString;
ARC_String_Create(&tagOrConstantString, NULL, 0);
uint32_t *id = NULL;
uint32_t tokensOrTagsIndex = 0;
ARC_Vector *orTokensOrTags = NULL;
ARC_Vector_DestroyDataFn destroyUint32Fn = ARC_ParserLang_VectorDestroyUInt32;
switch(childTagToken->id){
case ARC_PARSERLANG_ARGUMENT:
continue;
case ARC_PARSERLANG_TOKEN_OR_ID:
//add a new vector to tagsOrTokens for the or statment
ARC_Vector_Create(&orTokensOrTags, NULL, &destroyUint32Fn);
//add the first or vector to the tokensOrTags
ARC_Vector_Add(tokensOrTags, (void *)orTokensOrTags);
continue;
case ARC_PARSERLANG_ARGUMENTS:
//recurse to check all the arguments
ARC_ParserLangParsedData_GetArgumentsTag(tokensOrTags, childTagToken, getIdFn);
continue;
case ARC_PARSERLANG_TAG_OR_CONSTANT:
//get the id of the tag/constant
ARC_ParserLangParsedData_RecurseStringAdd(&tagOrConstantString, childTagToken);
id = (uint32_t *)malloc(sizeof(uint32_t));
*id = (*getIdFn)(tagOrConstantString);
//get the last vector within tokens or tags to add the tag/constant to
tokensOrTagsIndex = ARC_Vector_GetSize(tokensOrTags);
orTokensOrTags = (ARC_Vector *)ARC_Vector_Get(tokensOrTags, tokensOrTagsIndex - 1);
//add the id to the last or vector
ARC_Vector_Add(orTokensOrTags, (void *)id);
//cleanup
ARC_String_Destroy(tagOrConstantString);
continue;
default:
//this should only be whitespace
continue;
}
} }
} }
@ -140,20 +218,24 @@ void ARC_ParserLangParsedData_CreateBodyTag(ARC_ParserTag **tag, ARC_ParserTagTo
ARC_String_Destroy(tagIdString); ARC_String_Destroy(tagIdString);
/* ~ Tokens Or Tags Array ~ */ /* ~ Tokens Or Tags Array ~ */
//create a vector to store the tokens and tags in, we will not add a destroy as we will move those values to an array (instead of copying them) //create a vector to store another vector of data
ARC_Vector *tokensOrTags; ARC_Vector *tokensOrTags;
ARC_Vector_Create(&tokensOrTags, NULL, NULL); ARC_Vector_DestroyDataFn destroyVectorFn = ARC_ParserLang_VectorDestroyVector;
ARC_Vector_Create(&tokensOrTags, NULL, &destroyVectorFn);
//create vector within the tokens or tags vector to store the or rule in
ARC_Vector *orTokensOrTags;
ARC_Vector_DestroyDataFn destroyUint32Fn = ARC_ParserLang_VectorDestroyUInt32;
ARC_Vector_Create(&orTokensOrTags, NULL, &destroyUint32Fn);
//add the first or vector to the tokensOrTags
ARC_Vector_Add(tokensOrTags, (void *)orTokensOrTags);
//skipping whitespace and arrow tokens, the arguments index starts at 4 //skipping whitespace and arrow tokens, the arguments index starts at 4
ARC_ParserTagToken *argumentsToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, 4); ARC_ParserTagToken *argumentsToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, 4);
ARC_ParserLangParsedData_GetArgumentsTag(tokensOrTags, argumentsToken, getIdFn); ARC_ParserLangParsedData_GetArgumentsTag(tokensOrTags, argumentsToken, getIdFn);
//move the data from the vector to an array //TODO: move the data from the vector to an array
bodyTag->tokensOrTagsSize = ARC_Vector_GetSize(tokensOrTags);
bodyTag->tokensOrTags = (uint32_t **)malloc(sizeof(uint32_t *) * bodyTag->tokensOrTagsSize);
for(uint32_t index = 0; index < bodyTag->tokensOrTagsSize; index++){
bodyTag->tokensOrTags[index] = (uint32_t *)ARC_Vector_Get(tokensOrTags, index);
}
//cleanup //cleanup
ARC_Vector_Destroy(tokensOrTags); ARC_Vector_Destroy(tokensOrTags);

3
tests/std/parser.c
View file

@ -201,6 +201,7 @@ ARC_TEST(Parser_Basic_GetParsedValue){
/* ~ parserlang tests ~ */ /* ~ parserlang tests ~ */
uint32_t TEST_ParserLang_GetIdFn(ARC_String *constant){ uint32_t TEST_ParserLang_GetIdFn(ARC_String *constant){
printf("tag: %s\n", constant->data);
return 0; return 0;
} }
@ -209,7 +210,7 @@ ARC_TEST(Parser_ParserLang_BasicTest){
ARC_Parser_CreateAsParserLang(&parser, TEST_ParserLang_GetIdFn); ARC_Parser_CreateAsParserLang(&parser, TEST_ParserLang_GetIdFn);
ARC_String *tempString; ARC_String *tempString;
ARC_String_CreateWithStrlen(&tempString, "<test> -> <testingStuffs>\n"); ARC_String_CreateWithStrlen(&tempString, "<test> -> CHAR <testingStuffs>\n");
//this destroys string, so no need for cleanup //this destroys string, so no need for cleanup
ARC_Parser_Parse(parser, &tempString); ARC_Parser_Parse(parser, &tempString);