From b2aec6be00863da7e48ea8152e4226f4fcb323f3 Mon Sep 17 00:00:00 2001
From: herbglitch <herbglitch@protonmail.com>
Date: Sat, 8 Mar 2025 02:30:23 -0700
Subject: [PATCH] basic tests for chemical parsing, still need to write create
 and destroy functions for passed back data

---
 src/std/chemical.c   | 88 +++++++++++++++++++++++++-------------------
 tests/std/chemical.c |  4 +-
 2 files changed, 53 insertions(+), 39 deletions(-)

diff --git a/src/std/chemical.c b/src/std/chemical.c
index 4e8a801..5561c62 100644
--- a/src/std/chemical.c
+++ b/src/std/chemical.c
@@ -70,23 +70,25 @@ static const uint32_t ARC_CHEMICAL_CLOSE_CURLY_BRACE = 0x25;
 static const uint32_t ARC_CHEMICAL_TILDE             = 0x26;
 
 static const uint32_t ARC_CHEMICAL_LANGUAGE       = 0x27;
-static const uint32_t ARC_CHEMICAL_VARIABLE_LINES = 0x28;
-static const uint32_t ARC_CHEMICAL_VARIABLE_LINE  = 0x29;
-static const uint32_t ARC_CHEMICAL_ALLOW_SPACE    = 0x2A;
-static const uint32_t ARC_CHEMICAL_GROUP_NAME     = 0x2B;
-static const uint32_t ARC_CHEMICAL_TYPE           = 0x2C;
-static const uint32_t ARC_CHEMICAL_VALUE          = 0x2D;
-static const uint32_t ARC_CHEMICAL_NESTED_VALUE   = 0x2E;
-static const uint32_t ARC_CHEMICAL_VALUE_ARGS     = 0x2F;
-static const uint32_t ARC_CHEMICAL_VARIABLE       = 0x30;
-static const uint32_t ARC_CHEMICAL_VARIABLE_NAME  = 0x31;
-static const uint32_t ARC_CHEMICAL_VARIABLE_CHAR  = 0x32;
-static const uint32_t ARC_CHEMICAL_STRING         = 0x33;
-static const uint32_t ARC_CHEMICAL_STRING_CHARS   = 0x34;
-static const uint32_t ARC_CHEMICAL_STRING_CHAR    = 0x35;
-static const uint32_t ARC_CHEMICAL_ESCAPE_CHAR    = 0x36;
-static const uint32_t ARC_CHEMICAL_WHITESPACE     = 0x37;
-static const uint32_t ARC_CHEMICAL_NUMBER_TAG     = 0x38;
+static const uint32_t ARC_CHEMICAL_GROUP          = 0x28;
+static const uint32_t ARC_CHEMICAL_GROUP_NAME     = 0x29;
+static const uint32_t ARC_CHEMICAL_GROUP_ARGS     = 0x2A;
+static const uint32_t ARC_CHEMICAL_VARIABLE_LINES = 0x2B;
+static const uint32_t ARC_CHEMICAL_VARIABLE_LINE  = 0x2C;
+static const uint32_t ARC_CHEMICAL_ALLOW_SPACE    = 0x2D;
+static const uint32_t ARC_CHEMICAL_TYPE           = 0x2E;
+static const uint32_t ARC_CHEMICAL_VALUE          = 0x2F;
+static const uint32_t ARC_CHEMICAL_NESTED_VALUE   = 0x30;
+static const uint32_t ARC_CHEMICAL_VALUE_ARGS     = 0x31;
+static const uint32_t ARC_CHEMICAL_VARIABLE       = 0x32;
+static const uint32_t ARC_CHEMICAL_VARIABLE_NAME  = 0x33;
+static const uint32_t ARC_CHEMICAL_VARIABLE_CHAR  = 0x34;
+static const uint32_t ARC_CHEMICAL_STRING         = 0x35;
+static const uint32_t ARC_CHEMICAL_STRING_CHARS   = 0x36;
+static const uint32_t ARC_CHEMICAL_STRING_CHAR    = 0x37;
+static const uint32_t ARC_CHEMICAL_ESCAPE_CHAR    = 0x38;
+static const uint32_t ARC_CHEMICAL_WHITESPACE     = 0x39;
+static const uint32_t ARC_CHEMICAL_NUMBER_TAG     = 0x3A;
 
 void ARC_Chemical_InitLexerRulesFn(ARC_Lexer *lexer){
     ARC_Lexer_RegisterTokenRule(lexer, ARC_LexerTokenRule_CreateAndReturnMatchCharRule(ARC_CHEMICAL_TAB    , '\t'));
@@ -262,6 +264,15 @@ uint32_t ARC_Chemical_GetStringIdFn(ARC_String *string){
     if(ARC_String_EqualsCStringWithStrlen(string, "<language>")){
         return ARC_CHEMICAL_LANGUAGE;
     }
+    if(ARC_String_EqualsCStringWithStrlen(string, "<group>")){
+        return ARC_CHEMICAL_GROUP;
+    }
+    if(ARC_String_EqualsCStringWithStrlen(string, "<groupName>")){
+        return ARC_CHEMICAL_GROUP_NAME;
+    }
+    if(ARC_String_EqualsCStringWithStrlen(string, "<groupArgs>")){
+        return ARC_CHEMICAL_GROUP_ARGS;
+    }
     if(ARC_String_EqualsCStringWithStrlen(string, "<variableLines>")){
         return ARC_CHEMICAL_VARIABLE_LINES;
     }
@@ -271,9 +282,6 @@ uint32_t ARC_Chemical_GetStringIdFn(ARC_String *string){
     if(ARC_String_EqualsCStringWithStrlen(string, "<allowSpace>")){
         return ARC_CHEMICAL_ALLOW_SPACE;
     }
-    if(ARC_String_EqualsCStringWithStrlen(string, "<groupName>")){
-        return ARC_CHEMICAL_GROUP_NAME;
-    }
     if(ARC_String_EqualsCStringWithStrlen(string, "<type>")){
         return ARC_CHEMICAL_TYPE;
     }
@@ -321,6 +329,7 @@ void ARC_ChemicalData_CreateFn(void **data, ARC_ParserTagToken *parsedData, void
     printf("HERE??\n");
     *data = NULL;
     if(parsedData == NULL || userData == NULL){
+        printf("Parsed Data was NULL\n");
         //TODO: error here?
         *data = NULL;
         return;
@@ -364,30 +373,33 @@ void ARC_Chemical_Create(ARC_Chemical **chemical){
 
     /* ~ define the language as a string ~ */
     char *languageCString =
-        "<language>     -> <variableLine> <language> | <whitespace> <language> | LAMBDA\n"
+        "<language>      -> <group> <language> | <variableLines> <language> | LAMBDA\n"
 
-        "<variableLine> -> <allowSpace> <type> <whitespace> <variable> <whitespace> EQUAL <whitespace> <value> <whitespace> SEMICOLON <allowSpace>\n"
-        "<allowSpace>   -> SPACE <allowSpace> | TAB <allowSpace> | LAMBDA\n"
+        "<group>         -> <whitespace> <groupName> <whitespace> <variable> <whitespace> OPEN_CURLY_BRACE <groupArgs> <whitespace> CLOSE_CURLY_BRACE\n"
+        "<groupName>     -> <variable>\n"
+        "<groupArgs>     -> <whitespace> <variableLines> | LAMBDA\n"
 
-        "<groupName>    -> <variable>\n"
-        "<type>         -> <variable>\n"
-        "<value>        -> <variable> | <number>\n"
-        //"<value>        -> <variable> | <number> | <string> | <nestedValue>\n"
-        //"<nestedValue>  -> OPEN_CURLY_BRACE <whitespace> <valueArgs> <whitespace> CLOSE_CURLY_BRACE\n"
-        //"<valueArgs>    -> <value> COMMA <valueArgs> | <value>\n"
+        "<variableLines> -> <variableLine> <whitespace> <variableLines> | <variableLine>\n"
+        "<variableLine>  -> <whitespace> <type> <whitespace> <variable> <whitespace> EQUAL <whitespace> <value> <whitespace> SEMICOLON\n"
+        "<allowSpace>    -> SPACE <allowSpace> | TAB <allowSpace> | LAMBDA\n"
 
-        "<variable>     -> ALPHA_UPPER_CHAR <variableName> | ALPHA_LOWER_CHAR <variableName> | UNDERSCORE <variableName>\n"
-        "<variableName> -> <variableChar> <variableName> | LAMBDA\n"
-        "<variableChar> -> ALPHA_UPPER_CHAR | ALPHA_LOWER_CHAR | UNDERSCORE | NUMBER\n"
+        "<type>          -> <variable>\n"
+        "<value>         -> <variable> | <number> | <string> | <nestedValue>\n"
+        "<nestedValue>   -> OPEN_CURLY_BRACE <whitespace> <valueArgs> <whitespace> CLOSE_CURLY_BRACE\n"
+        "<valueArgs>     -> <value> COMMA <valueArgs> | <value>\n"
 
-        //"<string>       -> QUOTE <stringChars> QUOTE\n"
-        //"<stringChars>  -> <stringChar> <stringChars> | <escapeChar> <stringChars> | LAMBDA\n"
-        //"<stringChar>   -> TAB | SPACE | BANG | HASH | DOLLAR | PERCENT | AMPERSAND | SINGLE_QUOTE | OPEN_PAREN | CLOSE_PAREN | ASTERISK | PLUS | COMMA | MINUS | PERIOD | SLASH | NUMBER | COLON | SEMICOLON | LESS_THAN | GREATER_THAN | EQUAL | QUESTION_MARK | AT | ALPHA_UPPER_CHAR | OPEN_BRACKET | CLOSE_BRACKET | CARET | UNDERSCORE | GRAVE | ALPHA_LOWER_CHAR | OPEN_CURLY_BRACE | VERTICAL_LINE | CLOSE_CURLY_BRACE | TILDE\n"
-        //"<escapeChar>   -> BACKSLASH BACKSLASH | BACKSLASH QUOTE | BACKSLASH ALPHA_UPPER_CHAR | BACKSLASH ALPHA_LOWER_CHAR\n"
+        "<variable>      -> ALPHA_UPPER_CHAR <variableName> | ALPHA_LOWER_CHAR <variableName> | UNDERSCORE <variableName>\n"
+        "<variableName>  -> <variableChar> <variableName> | LAMBDA\n"
+        "<variableChar>  -> ALPHA_UPPER_CHAR | ALPHA_LOWER_CHAR | UNDERSCORE | NUMBER\n"
 
-        "<number>       -> NUMBER <number> | NUMBER\n"
+        "<string>        -> QUOTE <stringChars> QUOTE\n"
+        "<stringChars>   -> <stringChar> <stringChars> | <escapeChar> <stringChars> | LAMBDA\n"
+        "<stringChar>    -> TAB | SPACE | BANG | HASH | DOLLAR | PERCENT | AMPERSAND | SINGLE_QUOTE | OPEN_PAREN | CLOSE_PAREN | ASTERISK | PLUS | COMMA | MINUS | PERIOD | SLASH | NUMBER | COLON | SEMICOLON | LESS_THAN | GREATER_THAN | EQUAL | QUESTION_MARK | AT | ALPHA_UPPER_CHAR | OPEN_BRACKET | CLOSE_BRACKET | CARET | UNDERSCORE | GRAVE | ALPHA_LOWER_CHAR | OPEN_CURLY_BRACE | VERTICAL_LINE | CLOSE_CURLY_BRACE | TILDE\n"
+        "<escapeChar>    -> BACKSLASH BACKSLASH | BACKSLASH QUOTE | BACKSLASH ALPHA_UPPER_CHAR | BACKSLASH ALPHA_LOWER_CHAR\n"
 
-        "<whitespace>   -> SPACE <whitespace> | TAB <whitespace> | NEWLINE <whitespace> | LAMBDA\n";
+        "<number>        -> NUMBER <number> | NUMBER\n"
+
+        "<whitespace>    -> SPACE <whitespace> | TAB <whitespace> | NEWLINE <whitespace> | LAMBDA\n";
 
     /* ~ define the language as a string ~ */
     ARC_String *languageString;
diff --git a/tests/std/chemical.c b/tests/std/chemical.c
index edbdf56..228574f 100644
--- a/tests/std/chemical.c
+++ b/tests/std/chemical.c
@@ -10,7 +10,9 @@ ARC_TEST(Chemical_BasicTest){
 
     ARC_CHECK(arc_errno == 0);
 
-    char *tempCString = "int32 test = 5; ";
+    //char *tempCString = "int32 test = 5;";
+    char *tempCString = "group test { int32 test = 5; }";
+    //char *tempCString = "group test {  }";
     ARC_String *tempString;
     ARC_String_CreateWithStrlen(&tempString, tempCString);
     ARC_Chemical_LoadFromString(chemical, &tempString);