regex-engine/tests/parser_tests.c

/*
 * Copyright (c) Camden Dixie O'Brien
 * SPDX-License-Identifier: AGPL-3.0-only
 */

#include "parser.h"
#include "testing.h"

#define PARSE_REGEX_STRING(s, r) parse_regex(s, strlen(s), r)

static void a_has_no_alternative(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("a", &r);
	ASSERT_NE(-1, result);
	ASSERT_NULL(r.alternative);
	regex_free_children(&r);
}

static void a_pipe_b_has_alternative(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("a|b", &r);
	ASSERT_NE(-1, result);
	ASSERT_NOT_NULL(r.alternative);
	regex_free_children(&r);
}

static void a_pipe_b_pipe_c_result_alternative_has_alternative(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("a|b|c", &r);
	ASSERT_NE(-1, result);
	ASSERT_NOT_NULL(r.alternative);
	ASSERT_NOT_NULL(r.alternative->alternative);
	regex_free_children(&r);
}

static void a_is_parsed_as_unquantified_literal(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("a", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[0].type);
	ASSERT_EQ('a', r.sequence.contents[0].literal);

	regex_free_children(&r);
}

static void b_is_parsed_as_unquantified_literal(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("b", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[0].type);
	ASSERT_EQ('b', r.sequence.contents[0].literal);

	regex_free_children(&r);
}

static void abc_is_parsed_as_sequence_of_unquantified_literals(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("abc", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(3, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[0].type);
	ASSERT_EQ('a', r.sequence.contents[0].literal);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[1].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[1].type);
	ASSERT_EQ('b', r.sequence.contents[1].literal);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[2].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[2].type);
	ASSERT_EQ('c', r.sequence.contents[2].literal);

	regex_free_children(&r);
}

static void dot_is_parsed_as_unquantified_wildcard_term(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING(".", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_WILDCARD, r.sequence.contents[0].type);

	regex_free_children(&r);
}

static void backslash_dot_is_parsed_as_unquantified_literal(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("\\.", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[0].type);
	ASSERT_EQ('.', r.sequence.contents[0].literal);

	regex_free_children(&r);
}

static void backslash_backslash_is_parsed_as_unquantified_literal(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("\\\\", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[0].type);
	ASSERT_EQ('\\', r.sequence.contents[0].literal);

	regex_free_children(&r);
}

static void a_pipe_b_in_parens_is_parsed_as_regex_term(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("(a|b)", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_REGEX, r.sequence.contents[0].type);

	const regex_t *inner = &r.sequence.contents[0].regex;
	ASSERT_EQ(1, inner->sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, inner->sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, inner->sequence.contents[0].type);
	ASSERT_EQ('a', inner->sequence.contents[0].literal);

	const regex_t *inner_alt = inner->alternative;
	ASSERT_EQ(1, inner->sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, inner_alt->sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, inner_alt->sequence.contents[0].type);
	ASSERT_EQ('b', inner_alt->sequence.contents[0].literal);

	regex_free_children(&r);
}

static void a_in_parens_b_is_parsed_as_sequence_with_regex_term(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("(a)b", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(2, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_REGEX, r.sequence.contents[0].type);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[1].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, r.sequence.contents[1].type);
	ASSERT_EQ('b', r.sequence.contents[1].literal);

	const regex_t *inner = &r.sequence.contents[0].regex;
	ASSERT_EQ(1, inner->sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, inner->sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, inner->sequence.contents[0].type);
	ASSERT_EQ('a', inner->sequence.contents[0].literal);

	regex_free_children(&r);
}

static void dot_star_is_parsed_as_zero_or_more_wildcard(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING(".*", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_ZERO_OR_MORE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_WILDCARD, r.sequence.contents[0].type);

	regex_free_children(&r);
}

static void dot_plus_is_parsed_as_one_or_more_wildcard(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING(".+", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_ONE_OR_MORE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_WILDCARD, r.sequence.contents[0].type);

	regex_free_children(&r);
}

static void dot_question_mark_is_parsed_as_zero_or_one_wildcard(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING(".?", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_ZERO_OR_ONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_WILDCARD, r.sequence.contents[0].type);

	regex_free_children(&r);
}

static void a_in_brackets_is_parsed_as_class_containing_only_a(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("[a]", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_CLASS, r.sequence.contents[0].type);
	ASSERT_FALSE(r.sequence.contents[0].class.negated);
	ASSERT_EQ(1, r.sequence.contents[0].class.count);
	ASSERT_NOT_NULL(r.sequence.contents[0].class.contents);
	ASSERT_EQ('a', r.sequence.contents[0].class.contents[0]);

	regex_free_children(&r);
}

static void caret_a_in_brackets_parses_as_negated_class(void)
{
	regex_t r = { 0 };
	const int result = PARSE_REGEX_STRING("[^a]", &r);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, r.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, r.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_CLASS, r.sequence.contents[0].type);
	ASSERT_TRUE(r.sequence.contents[0].class.negated);
	ASSERT_EQ(1, r.sequence.contents[0].class.count);
	ASSERT_NOT_NULL(r.sequence.contents[0].class.contents);
	ASSERT_EQ('a', r.sequence.contents[0].class.contents[0]);

	regex_free_children(&r);
}

int main(void)
{
	TESTING_BEGIN();
	a_has_no_alternative();
	a_pipe_b_has_alternative();
	a_pipe_b_pipe_c_result_alternative_has_alternative();
	a_is_parsed_as_unquantified_literal();
	b_is_parsed_as_unquantified_literal();
	abc_is_parsed_as_sequence_of_unquantified_literals();
	dot_is_parsed_as_unquantified_wildcard_term();
	backslash_dot_is_parsed_as_unquantified_literal();
	backslash_backslash_is_parsed_as_unquantified_literal();
	a_pipe_b_in_parens_is_parsed_as_regex_term();
	a_in_parens_b_is_parsed_as_sequence_with_regex_term();
	dot_star_is_parsed_as_zero_or_more_wildcard();
	dot_plus_is_parsed_as_one_or_more_wildcard();
	dot_question_mark_is_parsed_as_zero_or_one_wildcard();
	a_in_brackets_is_parsed_as_class_containing_only_a();
	caret_a_in_brackets_parses_as_negated_class();
	return TESTING_END();
}