OLD | NEW |
(Empty) | |
| 1 // Copyright (c) 2017 The Chromium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. |
| 4 |
| 5 #include <algorithm> |
| 6 #include <iostream> |
| 7 #include <sstream> |
| 8 #include <string> |
| 9 #include <vector> |
| 10 |
| 11 #include "base/base_paths.h" |
| 12 #include "base/files/file_path.h" |
| 13 #include "base/files/file_util.h" |
| 14 #include "base/i18n/icu_util.h" |
| 15 #include "base/path_service.h" |
| 16 #include "base/strings/string_split.h" |
| 17 #include "base/strings/string_util.h" |
| 18 #include "third_party/icu/source/common/unicode/unistr.h" |
| 19 #include "third_party/icu/source/common/unicode/utypes.h" |
| 20 #include "third_party/icu/source/i18n/unicode/uspoof.h" |
| 21 |
| 22 std::string GetSkeleton(const std::string& domain, |
| 23 const USpoofChecker* spoof_checker) { |
| 24 UErrorCode status = U_ZERO_ERROR; |
| 25 icu::UnicodeString ustr_skeleton; |
| 26 uspoof_getSkeletonUnicodeString(spoof_checker, 0 /* not used */, |
| 27 icu::UnicodeString::fromUTF8(domain), |
| 28 ustr_skeleton, &status); |
| 29 std::string skeleton; |
| 30 return U_SUCCESS(status) ? ustr_skeleton.toUTF8String(skeleton) : skeleton; |
| 31 } |
| 32 |
| 33 base::FilePath GetPath(base::StringPiece basename) { |
| 34 base::FilePath path; |
| 35 base::PathService::Get(base::DIR_SOURCE_ROOT, &path); |
| 36 return path.Append(FILE_PATH_LITERAL("components")) |
| 37 .Append(FILE_PATH_LITERAL("url_formatter")) |
| 38 .Append(FILE_PATH_LITERAL("top_domains")) |
| 39 .AppendASCII(basename); |
| 40 } |
| 41 |
| 42 bool WriteToFile(const std::string& content, base::StringPiece basename) { |
| 43 base::FilePath path = GetPath(basename); |
| 44 size_t size = content.size(); |
| 45 bool succeeded = |
| 46 static_cast<size_t>(base::WriteFile(path, content.data(), size)) == size; |
| 47 if (!succeeded) |
| 48 std::cerr << "Failed to write to " << path.AsUTF8Unsafe() << '\n'; |
| 49 return succeeded; |
| 50 } |
| 51 |
| 52 int main(int argc, const char** argv) { |
| 53 if (argc != 1) { |
| 54 std::cerr << "Generates the list of top domain skeletons to use as input to" |
| 55 "\nbase/dafsa/make_dafsa.py.\nUsage: " |
| 56 << argv[0] << '\n'; |
| 57 return 1; |
| 58 } |
| 59 |
| 60 base::i18n::InitializeICU(); |
| 61 base::FilePath input_file = GetPath("alexa_domains.list"); |
| 62 std::string input_content; |
| 63 if (!base::ReadFileToString(input_file, &input_content)) { |
| 64 std::cerr << "Failed to read the input file " << input_file.AsUTF8Unsafe() |
| 65 << '\n'; |
| 66 return 1; |
| 67 } |
| 68 |
| 69 UErrorCode status = U_ZERO_ERROR; |
| 70 USpoofChecker* spoof_checker = uspoof_open(&status); |
| 71 if (U_FAILURE(status)) { |
| 72 std::cerr << "Failed to create an ICU uspoof_checker due to " |
| 73 << u_errorName(status) << ".\n"; |
| 74 return 1; |
| 75 } |
| 76 |
| 77 std::stringstream input(input_content); |
| 78 std::string output = |
| 79 R"(// Copyright 2017 The Chromium Authors. All rights reserved. |
| 80 // Use of this source code is governed by a BSD-style license that can be |
| 81 // found in the LICENSE file. |
| 82 |
| 83 // This file is generated by components/url_formatter/make_top_domain_gperf.cc |
| 84 // DO NOT MANUALLY EDIT! |
| 85 |
| 86 // Each entry is the skeleton of a top domain for the confusability check |
| 87 // in components/url_formatter/url_formatter.cc. |
| 88 %% |
| 89 )"; |
| 90 |
| 91 std::string domain; |
| 92 size_t max_labels = 0; |
| 93 std::string domain_with_max_labels; |
| 94 while (std::getline(input, domain)) { |
| 95 if (domain[0] == '#') |
| 96 continue; |
| 97 std::string skeleton = GetSkeleton(domain, spoof_checker); |
| 98 if (skeleton.empty()) { |
| 99 std::cerr << "Failed to generate the skeleton of " << domain << '\n'; |
| 100 output += "// " + domain + '\n'; |
| 101 } else { |
| 102 output += skeleton + ", 1\n"; |
| 103 } |
| 104 std::vector<base::StringPiece> labels = base::SplitStringPiece( |
| 105 domain, ".", base::TRIM_WHITESPACE, base::SPLIT_WANT_ALL); |
| 106 if (labels.size() > max_labels) { |
| 107 domain_with_max_labels = domain; |
| 108 max_labels = labels.size(); |
| 109 } |
| 110 } |
| 111 |
| 112 output += "%%\n"; |
| 113 |
| 114 if (!WriteToFile(output, "alexa_skeletons.gperf")) |
| 115 return 1; |
| 116 |
| 117 std::cout << "The first domain with the largest number of labels is " |
| 118 << domain_with_max_labels << " and has " << max_labels |
| 119 << " labels.\n"; |
| 120 |
| 121 return 0; |
| 122 } |
OLD | NEW |