regex-engine/tests/parse_tests.c

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

#include "parse.h"
#include "testing.h"

#define PARSE_EXPR_STRING(s, r) parse_expr(s, strlen(s), r)

static void a_has_1_alternative(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("a", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	regex_free(&t);
}

static void a_pipe_b_has_2_alternatives(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("a|b", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(2, t.count);
	regex_free(&t);
}

static void a_pipe_b_pipe_c_has_3_alternatives(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("a|b|c", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(3, t.count);
	regex_free(&t);
}

static void a_is_parsed_as_unquantified_literal(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("a", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[0].type);
	ASSERT_EQ('a', t.contents[0].contents[0].literal);

	regex_free(&t);
}

static void b_is_parsed_as_unquantified_literal(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("b", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[0].type);
	ASSERT_EQ('b', t.contents[0].contents[0].literal);

	regex_free(&t);
}

static void abc_is_parsed_as_sequence_of_unquantified_literals(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("abc", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(3, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[0].type);
	ASSERT_EQ('a', t.contents[0].contents[0].literal);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[1].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[1].type);
	ASSERT_EQ('b', t.contents[0].contents[1].literal);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[2].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[2].type);
	ASSERT_EQ('c', t.contents[0].contents[2].literal);

	regex_free(&t);
}

static void dot_is_parsed_as_unquantified_wildcard_term(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING(".", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_WILDCARD, t.contents[0].contents[0].type);

	regex_free(&t);
}

static void backslash_dot_is_parsed_as_unquantified_literal(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("\\.", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[0].type);
	ASSERT_EQ('.', t.contents[0].contents[0].literal);

	regex_free(&t);
}

static void backslash_backslash_is_parsed_as_unquantified_literal(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("\\\\", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[0].type);
	ASSERT_EQ('\\', t.contents[0].contents[0].literal);

	regex_free(&t);
}

static void a_pipe_b_in_parens_is_parsed_as_subexpr_term(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("(a|b)", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_SUBEXPR, t.contents[0].contents[0].type);

	const regex_t *inner = &t.contents[0].contents[0].subexpr;
	ASSERT_EQ(2, inner->count);

	ASSERT_EQ(1, inner->contents[0].count);
	ASSERT_EQ(
	    REGEX_QUANTIFIER_NONE, inner->contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, inner->contents[0].contents[0].type);
	ASSERT_EQ('a', inner->contents[0].contents[0].literal);

	ASSERT_EQ(1, inner->contents[1].count);
	ASSERT_EQ(
	    REGEX_QUANTIFIER_NONE, inner->contents[1].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, inner->contents[1].contents[0].type);
	ASSERT_EQ('b', inner->contents[1].contents[0].literal);

	regex_free(&t);
}

static void a_in_parens_b_is_parsed_as_sequence_with_subexpr_term(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("(a)b", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(2, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_SUBEXPR, t.contents[0].contents[0].type);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[1].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, t.contents[0].contents[1].type);
	ASSERT_EQ('b', t.contents[0].contents[1].literal);

	const regex_t *inner = &t.contents[0].contents[0].subexpr;
	ASSERT_EQ(1, inner->contents[0].count);
	ASSERT_EQ(
	    REGEX_QUANTIFIER_NONE, inner->contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_LITERAL, inner->contents[0].contents[0].type);
	ASSERT_EQ('a', inner->contents[0].contents[0].literal);

	regex_free(&t);
}

static void dot_star_is_parsed_as_star_quantified_wildcard(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING(".*", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_STAR, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_WILDCARD, t.contents[0].contents[0].type);

	regex_free(&t);
}

static void dot_plus_is_parsed_as_plus_quantified_wildcard(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING(".+", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_PLUS, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_WILDCARD, t.contents[0].contents[0].type);

	regex_free(&t);
}

static void dot_question_mark_is_parsed_as_qmrk_quantified_wildcard(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING(".?", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_QMARK, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_WILDCARD, t.contents[0].contents[0].type);

	regex_free(&t);
}

static void a_in_brackets_is_parsed_as_class_containing_only_a(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("[a]", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_CLASS, t.contents[0].contents[0].type);
	ASSERT_FALSE(t.contents[0].contents[0].class.negated);
	ASSERT_EQ(1, t.contents[0].contents[0].class.count);
	ASSERT_NOT_NULL(t.contents[0].contents[0].class.contents);
	ASSERT_EQ('a', t.contents[0].contents[0].class.contents[0]);

	regex_free(&t);
}

static void caret_a_in_brackets_parses_as_negated_class(void)
{
	regex_t t;
	const int result = PARSE_EXPR_STRING("[^a]", &t);
	ASSERT_NE(-1, result);
	ASSERT_EQ(1, t.count);
	ASSERT_NOT_NULL(t.contents);

	ASSERT_EQ(1, t.contents[0].count);
	ASSERT_EQ(REGEX_QUANTIFIER_NONE, t.contents[0].contents[0].quantifier);
	ASSERT_EQ(REGEX_TERM_CLASS, t.contents[0].contents[0].type);
	ASSERT_TRUE(t.contents[0].contents[0].class.negated);
	ASSERT_EQ(1, t.contents[0].contents[0].class.count);
	ASSERT_NOT_NULL(t.contents[0].contents[0].class.contents);
	ASSERT_EQ('a', t.contents[0].contents[0].class.contents[0]);

	regex_free(&t);
}

int main(void)
{
	TESTING_BEGIN();
	a_has_1_alternative();
	a_pipe_b_has_2_alternatives();
	a_pipe_b_pipe_c_has_3_alternatives();
	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_star_quantified_wildcard();
	dot_plus_is_parsed_as_plus_quantified_wildcard();
	dot_question_mark_is_parsed_as_qmrk_quantified_wildcard();
	a_in_brackets_is_parsed_as_class_containing_only_a();
	caret_a_in_brackets_parses_as_negated_class();
	return TESTING_END();
}