regex/re_test.go

package main

import (
	"slices"
	"testing"
)

var reTests = []struct {
	re     string
	str    string
	result []matchIndex
}{
	{"a", "abc", []matchIndex{{0, 1}}},
	{"a", "bca", []matchIndex{{2, 3}}},
	{"l", "ggllgg", []matchIndex{{2, 3}, {3, 4}}},
	{"(b|c)", "abdceb", []matchIndex{{1, 2}, {3, 4}, {5, 6}}},
	{"a+", "brerereraaaaabbbbb", []matchIndex{{8, 13}}},
	{"ab+", "qweqweqweaqweqweabbbbbr", []matchIndex{{16, 22}}},
	{"(b|c|A)", "ooaoobocA", []matchIndex{{5, 6}, {7, 8}, {8, 9}}},
	{"ab*", "a", []matchIndex{{0, 1}}},
	{"ab*", "abb", []matchIndex{{0, 3}}},
	{"a*b", "aaab", []matchIndex{{0, 4}}},
	{"a*b", "qwqw", []matchIndex{}},
	{"(abc)*", "abcabcabc", []matchIndex{{0, 9}}},
	{"((abc)|(def))*", "abcdef", []matchIndex{{0, 6}}},
	{"(abc)*|(def)*", "abcdef", []matchIndex{{0, 3}, {3, 3}, {3, 6}, {6, 6}}},
	{"b*a*a", "bba", []matchIndex{{0, 3}}},
	{"(ab)+", "abcabddd", []matchIndex{{0, 2}, {3, 5}}},
	{"a(b(c|d)*)*", "abccbd", []matchIndex{{0, 6}}},
	{"a(b|c)*d+", "abccdd", []matchIndex{{0, 6}}},
	{"a*", "", []matchIndex{{0, 0}}},
	{"a|b", "c", []matchIndex{}},
	{"(a|b)*c", "aabbc", []matchIndex{{0, 5}}},
	{"a(b|b)", "ab", []matchIndex{{0, 2}}},
	{"a*", "aaaaaaaa", []matchIndex{{0, 8}, {8, 8}}},

	{"ab?", "ab", []matchIndex{{0, 2}}},
	{"a?b", "ab", []matchIndex{{0, 2}}},
	{"a?", "", []matchIndex{{0, 0}}},
	{"a?b?c", "a", []matchIndex{}},
	{"a?b?c?", "ab", []matchIndex{{0, 2}}},
	{"a?b?c?", "ac", []matchIndex{{0, 2}}},
	{"a?b?c", "abc", []matchIndex{{0, 3}}},
	{"a?b?c", "acb", []matchIndex{{0, 2}}},

	{"[abc]", "defadefbdefce", []matchIndex{{3, 4}, {7, 8}, {11, 12}}},
	{"[ab]c", "ab", []matchIndex{}},
	{"g[ab]c", "gac", []matchIndex{{0, 3}}},
	{"g[ab]c", "gbc", []matchIndex{{0, 3}}},
	{"g[ab]c", "gc", []matchIndex{}},
	{"g[ab]c", "gfc", []matchIndex{}},
	{"[ab]*", "aabbbabaababab", []matchIndex{{0, 14}}},
	{"[ab]+", "aabbbabaababab", []matchIndex{{0, 14}}},
	{"[Ff]r[Uu]it", "fruit", []matchIndex{{0, 5}}},
	{"[Ff]r[Uu]it", "FrUit", []matchIndex{{0, 5}}},
	{"[Ff]r[Uu|]it", "Fr|it", []matchIndex{{0, 5}}},
	{"[Ff]r([Uu]|[pP])it", "Frpit", []matchIndex{{0, 5}}},
	{"[Ff]r[Uu]|[pP]it", "Frpit", []matchIndex{{2, 5}}},
	{"[a-zA-Z]+", "Hello, how is it going?", []matchIndex{{0, 5}, {7, 10}, {11, 13}, {14, 16}, {17, 22}}},

	{".+", "Hello, how is it going?", []matchIndex{{0, 23}}},
	{"a.", "a ", []matchIndex{{0, 2}}},
	{"a.b", "a/b", []matchIndex{{0, 3}}},
	{".", "a ", []matchIndex{{0, 2}}},
	{"a.", "a ", []matchIndex{{0, 2}}},
}

func TestFindAllMatches(t *testing.T) {
	for _, test := range reTests {
		t.Run(test.re+"	"+test.str, func(t *testing.T) {
			re_postfix := shuntingYard(test.re)
			startState := thompson(re_postfix)
			matchIndices := findAllMatches(startState, test.str)
			if !slices.Equal(test.result, matchIndices) {
				t.Errorf("Wanted %v	Got %v\n", test.result, matchIndices)
			}
		})
	}
}