1*ccdc9c3eSSadaf Ebrahimi // Copyright 2009 The RE2 Authors. All Rights Reserved.
2*ccdc9c3eSSadaf Ebrahimi // Use of this source code is governed by a BSD-style
3*ccdc9c3eSSadaf Ebrahimi // license that can be found in the LICENSE file.
4*ccdc9c3eSSadaf Ebrahimi
5*ccdc9c3eSSadaf Ebrahimi #include <string>
6*ccdc9c3eSSadaf Ebrahimi
7*ccdc9c3eSSadaf Ebrahimi #include "util/test.h"
8*ccdc9c3eSSadaf Ebrahimi #include "util/logging.h"
9*ccdc9c3eSSadaf Ebrahimi #include "re2/regexp.h"
10*ccdc9c3eSSadaf Ebrahimi
11*ccdc9c3eSSadaf Ebrahimi namespace re2 {
12*ccdc9c3eSSadaf Ebrahimi
13*ccdc9c3eSSadaf Ebrahimi struct PrefixTest {
14*ccdc9c3eSSadaf Ebrahimi const char* regexp;
15*ccdc9c3eSSadaf Ebrahimi bool return_value;
16*ccdc9c3eSSadaf Ebrahimi const char* prefix;
17*ccdc9c3eSSadaf Ebrahimi bool foldcase;
18*ccdc9c3eSSadaf Ebrahimi const char* suffix;
19*ccdc9c3eSSadaf Ebrahimi };
20*ccdc9c3eSSadaf Ebrahimi
21*ccdc9c3eSSadaf Ebrahimi static PrefixTest tests[] = {
22*ccdc9c3eSSadaf Ebrahimi // If the regexp is missing a ^, there's no required prefix.
23*ccdc9c3eSSadaf Ebrahimi { "abc", false },
24*ccdc9c3eSSadaf Ebrahimi { "", false },
25*ccdc9c3eSSadaf Ebrahimi { "(?m)^", false },
26*ccdc9c3eSSadaf Ebrahimi
27*ccdc9c3eSSadaf Ebrahimi // If the regexp immediately goes into
28*ccdc9c3eSSadaf Ebrahimi // something not a literal match, there's no required prefix.
29*ccdc9c3eSSadaf Ebrahimi { "^(abc)", false },
30*ccdc9c3eSSadaf Ebrahimi { "^a*", false },
31*ccdc9c3eSSadaf Ebrahimi
32*ccdc9c3eSSadaf Ebrahimi // Otherwise, it should work.
33*ccdc9c3eSSadaf Ebrahimi { "^abc$", true, "abc", false, "(?-m:$)" },
34*ccdc9c3eSSadaf Ebrahimi { "^abc", true, "abc", false, "" },
35*ccdc9c3eSSadaf Ebrahimi { "^(?i)abc", true, "abc", true, "" },
36*ccdc9c3eSSadaf Ebrahimi { "^abcd*", true, "abc", false, "d*" },
37*ccdc9c3eSSadaf Ebrahimi { "^[Aa][Bb]cd*", true, "ab", true, "cd*" },
38*ccdc9c3eSSadaf Ebrahimi { "^ab[Cc]d*", true, "ab", false, "[Cc]d*" },
39*ccdc9c3eSSadaf Ebrahimi { "^☺abc", true, "☺abc", false, "" },
40*ccdc9c3eSSadaf Ebrahimi };
41*ccdc9c3eSSadaf Ebrahimi
TEST(RequiredPrefix,SimpleTests)42*ccdc9c3eSSadaf Ebrahimi TEST(RequiredPrefix, SimpleTests) {
43*ccdc9c3eSSadaf Ebrahimi for (int i = 0; i < arraysize(tests); i++) {
44*ccdc9c3eSSadaf Ebrahimi const PrefixTest& t = tests[i];
45*ccdc9c3eSSadaf Ebrahimi for (int j = 0; j < 2; j++) {
46*ccdc9c3eSSadaf Ebrahimi Regexp::ParseFlags flags = Regexp::LikePerl;
47*ccdc9c3eSSadaf Ebrahimi if (j == 0)
48*ccdc9c3eSSadaf Ebrahimi flags = flags | Regexp::Latin1;
49*ccdc9c3eSSadaf Ebrahimi Regexp* re = Regexp::Parse(t.regexp, flags, NULL);
50*ccdc9c3eSSadaf Ebrahimi ASSERT_TRUE(re != NULL) << " " << t.regexp;
51*ccdc9c3eSSadaf Ebrahimi
52*ccdc9c3eSSadaf Ebrahimi string p;
53*ccdc9c3eSSadaf Ebrahimi bool f;
54*ccdc9c3eSSadaf Ebrahimi Regexp* s;
55*ccdc9c3eSSadaf Ebrahimi ASSERT_EQ(t.return_value, re->RequiredPrefix(&p, &f, &s))
56*ccdc9c3eSSadaf Ebrahimi << " " << t.regexp << " " << (j==0 ? "latin1" : "utf")
57*ccdc9c3eSSadaf Ebrahimi << " " << re->Dump();
58*ccdc9c3eSSadaf Ebrahimi if (t.return_value) {
59*ccdc9c3eSSadaf Ebrahimi ASSERT_EQ(p, string(t.prefix))
60*ccdc9c3eSSadaf Ebrahimi << " " << t.regexp << " " << (j==0 ? "latin1" : "utf");
61*ccdc9c3eSSadaf Ebrahimi ASSERT_EQ(f, t.foldcase)
62*ccdc9c3eSSadaf Ebrahimi << " " << t.regexp << " " << (j==0 ? "latin1" : "utf");
63*ccdc9c3eSSadaf Ebrahimi ASSERT_EQ(s->ToString(), string(t.suffix))
64*ccdc9c3eSSadaf Ebrahimi << " " << t.regexp << " " << (j==0 ? "latin1" : "utf");
65*ccdc9c3eSSadaf Ebrahimi s->Decref();
66*ccdc9c3eSSadaf Ebrahimi }
67*ccdc9c3eSSadaf Ebrahimi re->Decref();
68*ccdc9c3eSSadaf Ebrahimi }
69*ccdc9c3eSSadaf Ebrahimi }
70*ccdc9c3eSSadaf Ebrahimi }
71*ccdc9c3eSSadaf Ebrahimi
72*ccdc9c3eSSadaf Ebrahimi } // namespace re2
73