OLD | NEW |
1 // Copyright 2013 The Chromium Authors. All rights reserved. | 1 // Copyright 2013 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "components/autofill/core/common/autofill_regexes.h" | 5 #include "components/autofill/core/common/autofill_regexes.h" |
6 | 6 |
7 #include "base/containers/scoped_ptr_hash_map.h" | 7 #include "base/containers/scoped_ptr_hash_map.h" |
8 #include "base/logging.h" | 8 #include "base/logging.h" |
9 #include "base/memory/scoped_ptr.h" | 9 #include "base/memory/scoped_ptr.h" |
10 #include "base/memory/singleton.h" | 10 #include "base/memory/singleton.h" |
11 #include "base/strings/utf_string_conversions.h" | 11 #include "base/strings/string16.h" |
12 #include "third_party/re2/re2/re2.h" | 12 #include "third_party/icu/source/i18n/unicode/regex.h" |
13 | 13 |
14 namespace { | 14 namespace { |
15 | 15 |
16 // A singleton class that serves as a cache of compiled regex patterns. | 16 // A singleton class that serves as a cache of compiled regex patterns. |
17 class AutofillRegexes { | 17 class AutofillRegexes { |
18 public: | 18 public: |
19 static AutofillRegexes* GetInstance(); | 19 static AutofillRegexes* GetInstance(); |
20 | 20 |
21 // Returns the compiled regex matcher corresponding to |pattern|. | 21 // Returns the compiled regex matcher corresponding to |pattern|. |
22 re2::RE2* GetMatcher(const std::string& pattern); | 22 icu::RegexMatcher* GetMatcher(const base::string16& pattern); |
23 | 23 |
24 private: | 24 private: |
25 AutofillRegexes(); | 25 AutofillRegexes(); |
26 ~AutofillRegexes(); | 26 ~AutofillRegexes(); |
27 friend struct base::DefaultSingletonTraits<AutofillRegexes>; | 27 friend struct base::DefaultSingletonTraits<AutofillRegexes>; |
28 | 28 |
29 // Maps patterns to their corresponding regex matchers. | 29 // Maps patterns to their corresponding regex matchers. |
30 base::ScopedPtrHashMap<std::string, scoped_ptr<re2::RE2>> matchers_; | 30 base::ScopedPtrHashMap<base::string16, scoped_ptr<icu::RegexMatcher>> |
| 31 matchers_; |
31 | 32 |
32 DISALLOW_COPY_AND_ASSIGN(AutofillRegexes); | 33 DISALLOW_COPY_AND_ASSIGN(AutofillRegexes); |
33 }; | 34 }; |
34 | 35 |
35 // static | 36 // static |
36 AutofillRegexes* AutofillRegexes::GetInstance() { | 37 AutofillRegexes* AutofillRegexes::GetInstance() { |
37 return base::Singleton<AutofillRegexes>::get(); | 38 return base::Singleton<AutofillRegexes>::get(); |
38 } | 39 } |
39 | 40 |
40 AutofillRegexes::AutofillRegexes() { | 41 AutofillRegexes::AutofillRegexes() { |
41 } | 42 } |
42 | 43 |
43 AutofillRegexes::~AutofillRegexes() { | 44 AutofillRegexes::~AutofillRegexes() { |
44 } | 45 } |
45 | 46 |
46 re2::RE2* AutofillRegexes::GetMatcher(const std::string& pattern) { | 47 icu::RegexMatcher* AutofillRegexes::GetMatcher(const base::string16& pattern) { |
47 auto it = matchers_.find(pattern); | 48 auto it = matchers_.find(pattern); |
48 if (it == matchers_.end()) { | 49 if (it == matchers_.end()) { |
49 re2::RE2::Options options; | 50 const icu::UnicodeString icu_pattern(pattern.data(), pattern.length()); |
50 options.set_case_sensitive(false); | 51 |
51 scoped_ptr<re2::RE2> matcher(new re2::RE2(pattern, options)); | 52 UErrorCode status = U_ZERO_ERROR; |
52 DCHECK(matcher->ok()); | 53 scoped_ptr<icu::RegexMatcher> matcher( |
| 54 new icu::RegexMatcher(icu_pattern, UREGEX_CASE_INSENSITIVE, status)); |
| 55 DCHECK(U_SUCCESS(status)); |
| 56 |
53 auto result = matchers_.add(pattern, matcher.Pass()); | 57 auto result = matchers_.add(pattern, matcher.Pass()); |
54 DCHECK(result.second); | 58 DCHECK(result.second); |
55 it = result.first; | 59 it = result.first; |
56 } | 60 } |
57 return it->second; | 61 return it->second; |
58 } | 62 } |
59 | 63 |
60 } // namespace | 64 } // namespace |
61 | 65 |
62 namespace autofill { | 66 namespace autofill { |
63 | 67 |
64 bool MatchesPattern(const base::string16& input, const std::string& pattern) { | 68 bool MatchesPattern(const base::string16& input, |
65 // TODO(isherman): Run performance tests to determine whether caching regex | 69 const base::string16& pattern) { |
66 // matchers is still useful now that we've switched from ICU to RE2. | 70 icu::RegexMatcher* matcher = |
67 // http://crbug.com/470065 | 71 AutofillRegexes::GetInstance()->GetMatcher(pattern); |
68 re2::RE2* matcher = AutofillRegexes::GetInstance()->GetMatcher(pattern); | 72 icu::UnicodeString icu_input(input.data(), input.length()); |
69 return re2::RE2::PartialMatch(base::UTF16ToUTF8(input), *matcher); | 73 matcher->reset(icu_input); |
| 74 |
| 75 UErrorCode status = U_ZERO_ERROR; |
| 76 UBool match = matcher->find(0, status); |
| 77 DCHECK(U_SUCCESS(status)); |
| 78 return match == TRUE; |
70 } | 79 } |
71 | 80 |
72 } // namespace autofill | 81 } // namespace autofill |
OLD | NEW |