/**
* compilerkit_alpha_numeric_character_class_new:
* @fn compilerkit_alpha_numeric_character_class_new
*
* Constructs an alphanumeric character class object (internally the equivalent CompilerKitAlternation).
*
* For example, `compilerkit_character_class_new('0','z')` produces the regex `[0-9A-Za-z]`.
* Note that only alphanumeric characters will be included in the character class, and nothing else.
*
* @pre None.
* @param gunichar `lo` The low end of the character class. If this is `'\0'`, `lo` is the next higher character.
* @param gunichar `hi` The high end of the character class. If this is `'\0'`, return the empty string.
* @return GObject* a pointer to the new character class regex, or NULL if `lo` or `hi` are non-alphanumeric.
*/
GObject* compilerkit_alpha_numeric_character_class_new(gunichar lo, gunichar hi)
{
GObject *result;
if (!compilerkit_character_is_alpha_numeric(lo)) return NULL;
if (!compilerkit_character_is_alpha_numeric(hi)) return NULL;
sort_chars (&lo, &hi);
result = compilerkit_empty_set_get_instance ();
if (lo <= '9' && hi >= 'A')
{
result = compilerkit_character_class_new (lo, '9');
lo = 'A';
}
if ('A' <= lo && lo <= 'Z' && hi >= 'a')
{
result = compilerkit_alternation_new (result, compilerkit_character_class_new (lo, 'Z'));
lo = 'a';
}
if ('a' <= lo)
{
result = compilerkit_alternation_new (result, compilerkit_character_class_new (lo, hi));
}
return result;
}
/**
* compilerkit_regex_punct:
* @fn compilerkit_regex_punct
*
* Return a character class corresponding to all punctuation.
*
* @pre None
* @param None
* @return GObject* regex that matches all punctuation.
*/
GObject *compilerkit_regex_punct(void)
{
return compilerkit_alternation_vlist_new(
compilerkit_character_class_new('!', '/'),
compilerkit_character_class_new('@',':'),
compilerkit_character_class_new('[','`'),
compilerkit_character_class_new('{','~'),
NULL);
}
void test_convenience_alternation(void)
{
GObject* expression1 = compilerkit_character_class_new('0','h');
GObject* expression2 = compilerkit_character_class_new(33,'h');
GObject* expression3 = compilerkit_character_class_new('0',137);
g_assert(expression1 != NULL);
g_assert(expression2 == NULL);
g_assert(expression3 == NULL);
g_object_unref(expression1);
g_object_unref(expression2);
g_object_unref(expression3);
}
/** @todo Describe what task main will accomplish. */
int main (int argc, char ** argv)
{
CompilerKitScanner* scanner;
CompilerKitToken *token;
g_type_init();
scanner = compilerkit_scanner_new();
// Specify lexical structure of the language.
compilerkit_scanner_register(scanner, compilerkit_token_new("IDENTIFIER"),
compilerkit_positive_closure_new(compilerkit_character_class_new('A','Z')));
// Attach it to a parser
compilerkit_scanner_open_filename(scanner, "scanner-demo.c");
while (token = compilerkit_scanner_next_token(scanner))
{
// do something with the token
}
compilerkit_scanner_close_file(scanner);
g_object_unref (scanner);
}
/**
* compilerkit_regex_digits:
* @fn compilerkit_regex_digits
*
* Return a character class corresponding to `[0-9]`
*
* @pre None
* @param None
* @return GObject* regex that matches the digits 0 through 9.
*/
GObject *compilerkit_regex_digits()
{
return compilerkit_character_class_new('0', '9');
}
/**
* compilerkit_regex_whitespace:
* @fn compilerkit_regex_whitespace
*
* Return a character class corresponding to all whitespace.
*
* @pre None
* @param None
* @return GObject* regex that matches all whitespace.
*/
GObject *compilerkit_regex_whitespace(void)
{
return compilerkit_character_class_new(' ', ' ');;
}
/**
* compilerkit_regex_upper:
* @fn compilerkit_regex_upper
*
* Return a character class corresponding to `[A-Z]`
*
* @pre None
* @param None
* @return GObject* regex that matches the symbols A through Z.
*/
GObject *compilerkit_regex_upper(void)
{
return compilerkit_character_class_new('A', 'Z');
}
/**
* compilerkit_regex_lower:
* @fn compilerkit_regex_lower
*
* Return a character class corresponding to `[a-z]`
*
* @pre None
* @param None
* @return GObject* regex that matches the symbols a through z.
*/
GObject *compilerkit_regex_lower(void)
{
return compilerkit_character_class_new('a', 'z');
}
