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_EXPR_STRING(s, r) parse_expr(s, strlen(s), r)

static void a_has_no_alternative(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("a", &t);
	ASSERT_NE(-1, result);
	ASSERT_NULL(t.alternative);
	parse_tree_free_children(&t);
}

static void a_pipe_b_has_alternative(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("a|b", &t);
	ASSERT_NE(-1, result);
	ASSERT_NOT_NULL(t.alternative);
	parse_tree_free_children(&t);
}

static void a_pipe_b_pipe_c_result_alternative_has_alternative(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("a|b|c", &t);
	ASSERT_NE(-1, result);
	ASSERT_NOT_NULL(t.alternative);
	ASSERT_NOT_NULL(t.alternative->alternative);
	parse_tree_free_children(&t);
}

static void a_is_parsed_as_unquantified_literal(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("a", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[0].type);
	ASSERT_EQ('a', t.sequence.contents[0].literal);

	parse_tree_free_children(&t);
}

static void b_is_parsed_as_unquantified_literal(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("b", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[0].type);
	ASSERT_EQ('b', t.sequence.contents[0].literal);

	parse_tree_free_children(&t);
}

static void abc_is_parsed_as_sequence_of_unquantified_literals(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("abc", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(3, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[0].type);
	ASSERT_EQ('a', t.sequence.contents[0].literal);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[1].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[1].type);
	ASSERT_EQ('b', t.sequence.contents[1].literal);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[2].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[2].type);
	ASSERT_EQ('c', t.sequence.contents[2].literal);

	parse_tree_free_children(&t);
}

static void dot_is_parsed_as_unquantified_wildcard_term(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING(".", &t);
	ASSERT_NE(-1, result);

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

	parse_tree_free_children(&t);
}

static void backslash_dot_is_parsed_as_unquantified_literal(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("\\.", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[0].type);
	ASSERT_EQ('.', t.sequence.contents[0].literal);

	parse_tree_free_children(&t);
}

static void backslash_backslash_is_parsed_as_unquantified_literal(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("\\\\", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[0].type);
	ASSERT_EQ('\\', t.sequence.contents[0].literal);

	parse_tree_free_children(&t);
}

static void a_pipe_b_in_parens_is_parsed_as_subexpr_term(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("(a|b)", &t);
	ASSERT_NE(-1, result);

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

	const parse_tree_t *inner = &t.sequence.contents[0].subexpr;
	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 parse_tree_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);

	parse_tree_free_children(&t);
}

static void a_in_parens_b_is_parsed_as_sequence_with_subexpr_term(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("(a)b", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(2, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_SUBEXPR, t.sequence.contents[0].type);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[1].quantifier);
	ASSERT_EQ(TERM_TYPE_LITERAL, t.sequence.contents[1].type);
	ASSERT_EQ('b', t.sequence.contents[1].literal);

	const parse_tree_t *inner = &t.sequence.contents[0].subexpr;
	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);

	parse_tree_free_children(&t);
}

static void dot_star_is_parsed_as_zero_or_more_wildcard(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING(".*", &t);
	ASSERT_NE(-1, result);

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

	parse_tree_free_children(&t);
}

static void dot_plus_is_parsed_as_one_or_more_wildcard(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING(".+", &t);
	ASSERT_NE(-1, result);

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

	parse_tree_free_children(&t);
}

static void dot_question_mark_is_parsed_as_zero_or_one_wildcard(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING(".?", &t);
	ASSERT_NE(-1, result);

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

	parse_tree_free_children(&t);
}

static void a_in_brackets_is_parsed_as_class_containing_only_a(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("[a]", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_CLASS, t.sequence.contents[0].type);
	ASSERT_FALSE(t.sequence.contents[0].class.negated);
	ASSERT_EQ(1, t.sequence.contents[0].class.count);
	ASSERT_NOT_NULL(t.sequence.contents[0].class.contents);
	ASSERT_EQ('a', t.sequence.contents[0].class.contents[0]);

	parse_tree_free_children(&t);
}

static void caret_a_in_brackets_parses_as_negated_class(void)
{
	parse_tree_t t;
	const int result = PARSE_EXPR_STRING("[^a]", &t);
	ASSERT_NE(-1, result);

	ASSERT_EQ(1, t.sequence.len);
	ASSERT_EQ(QUANTIFIER_NONE, t.sequence.contents[0].quantifier);
	ASSERT_EQ(TERM_TYPE_CLASS, t.sequence.contents[0].type);
	ASSERT_TRUE(t.sequence.contents[0].class.negated);
	ASSERT_EQ(1, t.sequence.contents[0].class.count);
	ASSERT_NOT_NULL(t.sequence.contents[0].class.contents);
	ASSERT_EQ('a', t.sequence.contents[0].class.contents[0]);

	parse_tree_free_children(&t);
}

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_subexpr_term();
	a_in_parens_b_is_parsed_as_sequence_with_subexpr_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();
}