parser working, needs more tests and an actual language to make sure that is true though

This commit is contained in:
herbglitch 2024-11-20 10:27:17 -07:00
parent 63dfb98aad
commit 606f8e4bad
10 changed files with 365 additions and 148 deletions

View file

@ -12,28 +12,28 @@ ARC_TEST(Lexer_Char_Match){
ARC_Lexer_LexString(lexer, &simple);
ARC_LexerToken token;
ARC_LexerToken *token;
token = ARC_Lexer_GetToken(lexer, 0);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_COLON_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_COLON_ID);
token = ARC_Lexer_GetToken(lexer, 1);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_COLON_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_COLON_ID);
token = ARC_Lexer_GetToken(lexer, 2);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_LEFT_CURLY_BRACE_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_LEFT_CURLY_BRACE_ID);
token = ARC_Lexer_GetToken(lexer, 3);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_RIGHT_CURLY_BRACE_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_RIGHT_CURLY_BRACE_ID);
token = ARC_Lexer_GetToken(lexer, 4);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_BANG_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_BANG_ID);
token = ARC_Lexer_GetToken(lexer, 5);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_FORWARD_SLASH_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_FORWARD_SLASH_ID);
token = ARC_Lexer_GetToken(lexer, 6);
ARC_CHECK(token.rule == ARC_LEXER_TOKEN_PERIOD_ID);
ARC_CHECK(token->rule == ARC_LEXER_TOKEN_PERIOD_ID);
ARC_Lexer_Destroy(lexer);
}

View file

@ -1,7 +1,9 @@
#include "../test.h"
#include "arc/std/errno.h"
#include "arc/std/parser.h"
#include "arc/std/lexer.h"
#include "arc/std/parser/parserlang.h"
#include <stddef.h>
#define LAMBDA ARC_PARSER_TAG_LAMBDA
#define CHAR ARC_LEXER_TOKEN_ALPHALOWERCHAR
@ -10,6 +12,73 @@
#define VARIABLE_NAME 24
#define VARIABLE 25
void TEST_ParserLanguageTag_CreateStringFn(void **data){
ARC_String_Create((ARC_String **)data, NULL, 0);
}
void TEST_ParserLanguageTag_DestroyStringFn(void *data){
ARC_String_Destroy((ARC_String *)data);
}
//for this very basic example, the tagId does not matter
void TEST_ParserLanguageTag_AddFirstCharFn(void **data, uint32_t tagId, ARC_LexerToken *token){
if(*data == NULL){
return;
}
ARC_String *tokenData = NULL;
ARC_String_Copy(&tokenData, token->data);
ARC_String_Append(&tokenData, *data);
ARC_String_Destroy(*data);
*data = tokenData;
}
//for this very basic example, the tagId does not matter
void TEST_ParserLanguageTag_AddCharFn(void **data, uint32_t tagId, ARC_LexerToken *token){
if(*data == NULL){
return;
}
ARC_String_Append((ARC_String **)data, token->data);
}
uint32_t *charOrNumTokens[] = { (uint32_t[]){ 1, CHAR }, (uint32_t[]){ 1, NUM } };
uint32_t *variableNameTags[] = { (uint32_t[]){ 2, CHAR_OR_NUM, VARIABLE_NAME }, (uint32_t[]){ 1, LAMBDA } };
uint32_t *variableTokensOrTags[] = { (uint32_t[]){ 2, CHAR, VARIABLE_NAME } };
//TODO: note how language function callbacks work, and how they use the parentData if createDataFn is NULL
ARC_ParserData_CreateFn createStringFn = TEST_ParserLanguageTag_CreateStringFn;
ARC_ParserData_DestroyFn destroyStringFn = TEST_ParserLanguageTag_DestroyStringFn;
ARC_ParserLanguageTag_AddDataFn addCharFn = TEST_ParserLanguageTag_AddCharFn;
ARC_ParserLanguageTag_AddDataFn addFirstCharFn = TEST_ParserLanguageTag_AddFirstCharFn;
ARC_ParserLanguageTag testTags[3] = {
{
VARIABLE, //tagId
variableTokensOrTags, //tokensOrTags
1, //tokenOrTagsSize
&addFirstCharFn //addDataFn
},
{
VARIABLE_NAME, //tagId
variableNameTags, //tokensOrTags
2, //tokenOrTagsSize
NULL, //addDataFn
},
{
CHAR_OR_NUM, //tagId
charOrNumTokens, //tokensOrTags
2, //tokenOrTagsSize
&addCharFn //addDataFn
}
};
ARC_Array languageArray = {
3, //size
testTags //data
};
void TEST_Parser_InitLexerRulesFn(ARC_Lexer *lexer){
ARC_Lexer_InitBasicTokenRules(lexer);
}
@ -17,33 +86,7 @@ void TEST_Parser_InitLexerRulesFn(ARC_Lexer *lexer){
ARC_TEST(Parser_Init){
ARC_Parser *parser;
uint32_t *charOrNumTokens[] = { (uint32_t[]){ 1, CHAR }, (uint32_t[]){ 1, NUM } };
uint32_t *variableNameTags[] = { (uint32_t[]){ 2, CHAR_OR_NUM, VARIABLE_NAME }, (uint32_t[]){ 1, LAMBDA } };
uint32_t *variableTokensOrTags[] = { (uint32_t[]){ 2, CHAR, VARIABLE_NAME } };
ARC_ParserLanguageTag testTags[3] = {
{
VARIABLE, //tagId
variableTokensOrTags, //tokensOrTags
1 //tokenOrTagsSize
},
{
VARIABLE_NAME, //tagId
variableNameTags, //tokensOrTags
2 //tokenOrTagsSize
},
{
CHAR_OR_NUM, //tagId
charOrNumTokens, //tokensOrTags
2 //tokenOrTagsSize
}
};
ARC_Array languageArray = {
3, //size
testTags //data
};
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn);
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn, NULL, NULL);
ARC_Parser_Destroy(parser);
@ -53,37 +96,11 @@ ARC_TEST(Parser_Init){
ARC_TEST(Parser_Basic_Parse){
ARC_Parser *parser;
uint32_t *charOrNumTokens[] = { (uint32_t[]){ 1, CHAR }, (uint32_t[]){ 1, NUM } };
uint32_t *variableNameTags[] = { (uint32_t[]){ 2, CHAR_OR_NUM, VARIABLE_NAME }, (uint32_t[]){ 1, LAMBDA } };
uint32_t *variableTokensOrTags[] = { (uint32_t[]){ 2, CHAR, VARIABLE_NAME } };
ARC_ParserLanguageTag testTags[3] = {
{
VARIABLE, //tagId
variableTokensOrTags, //tokensOrTags
1 //tokenOrTagsSize
},
{
VARIABLE_NAME, //tagId
variableNameTags, //tokensOrTags
2 //tokenOrTagsSize
},
{
CHAR_OR_NUM, //tagId
charOrNumTokens, //tokensOrTags
2 //tokenOrTagsSize
}
};
ARC_Array languageArray = {
3, //size
testTags //data
};
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn);
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn, NULL, NULL);
ARC_String *tempString;
//first variable test
/* ~ first test ~ */
ARC_String_CreateWithStrlen(&tempString, "myvar1");
//this destroys string, so no need for cleanup
@ -92,7 +109,7 @@ ARC_TEST(Parser_Basic_Parse){
ARC_CHECK(arc_errno == 0);
//second variable test
/* ~ second test ~ */
ARC_String_CreateWithStrlen(&tempString, "z1xwvq");
//this destroys string, so no need for cleanup
@ -101,7 +118,7 @@ ARC_TEST(Parser_Basic_Parse){
ARC_CHECK(arc_errno == 0);
//third variable test
/* ~ third test ~ */
ARC_String_CreateWithStrlen(&tempString, "z1234");
//this destroys string, so no need for cleanup
@ -110,7 +127,7 @@ ARC_TEST(Parser_Basic_Parse){
ARC_CHECK(arc_errno == 0);
//fourth variable test
/* ~ fourth test ~ */
ARC_String_CreateWithStrlen(&tempString, "aaaaa");
//this destroys string, so no need for cleanup
@ -119,44 +136,19 @@ ARC_TEST(Parser_Basic_Parse){
ARC_CHECK(arc_errno == 0);
/* ~ cleanup ~ */
ARC_Parser_Destroy(parser);
}
ARC_TEST(Parser_Basic_ParseError){
ARC_Parser *parser;
uint32_t *charOrNumTokens[] = { (uint32_t[]){ 1, CHAR }, (uint32_t[]){ 1, NUM } };
uint32_t *variableNameTags[] = { (uint32_t[]){ 2, CHAR_OR_NUM, VARIABLE_NAME }, (uint32_t[]){ 1, LAMBDA } };
uint32_t *variableTokensOrTags[] = { (uint32_t[]){ 2, CHAR, VARIABLE_NAME } };
ARC_ParserLanguageTag testTags[3] = {
{
VARIABLE, //tagId
variableTokensOrTags, //tokensOrTags
1 //tokenOrTagsSize
},
{
VARIABLE_NAME, //tagId
variableNameTags, //tokensOrTags
2 //tokenOrTagsSize
},
{
CHAR_OR_NUM, //tagId
charOrNumTokens, //tokensOrTags
2 //tokenOrTagsSize
}
};
ARC_Array languageArray = {
3, //size
testTags //data
};
//TODO: remove this
arc_errno = 0;
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn);
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn, NULL, NULL);
ARC_String *tempString;
/* ~ first test ~ */
ARC_String_CreateWithStrlen(&tempString, "!myVar1");
//this destroys string, so no need for cleanup
@ -164,6 +156,8 @@ ARC_TEST(Parser_Basic_ParseError){
ARC_CHECK(arc_errno == ARC_ERRNO_DATA);
/* ~ second test ~ */
//check again with moved character
arc_errno = 0;
ARC_String_CreateWithStrlen(&tempString, "my!Var1");
@ -173,6 +167,8 @@ ARC_TEST(Parser_Basic_ParseError){
ARC_CHECK(arc_errno == ARC_ERRNO_DATA);
/* ~ third test ~ */
//check again with moved character
arc_errno = 0;
ARC_String_CreateWithStrlen(&tempString, "myVar1!");
@ -182,12 +178,39 @@ ARC_TEST(Parser_Basic_ParseError){
ARC_CHECK(arc_errno == ARC_ERRNO_DATA);
/* ~ cleanup ~ */
ARC_Parser_Destroy(parser);
//reset for next test
arc_errno = 0;
}
ARC_TEST(Parser_Basic_GetParsedValue){
ARC_Parser *parser;
ARC_Parser_Create(&parser, &languageArray, TEST_Parser_InitLexerRulesFn, &createStringFn, &destroyStringFn);
ARC_String *tempString;
/* ~ first test ~ */
ARC_String_CreateWithStrlen(&tempString, "myvar1");
//this destroys string, so no need for cleanup
ARC_Parser_Parse(parser, &tempString);
ARC_CHECK(arc_errno == 0);
ARC_String *checkValue = (ARC_String *)ARC_Parser_GetData(parser);
ARC_CHECK(ARC_String_EqualsCStringWithStrlen(checkValue, "myvar1"));
/* ~ cleanup ~ */
ARC_Parser_Destroy(parser);
}
ARC_TEST(Parser_ParserLang_BasicTest){
ARC_Parser *parser;
ARC_Parser_CreateAsParserLang(&parser);