[Passwords] Parse zxcvbn dicts in background

This change updates the zxcvbn data component and library to support
parsing the data dictionaries on the background thread.

Fixed: 1112461
Change-Id: I1b553bf448c1baf9b0a1359a853fc5360716e52a
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2332609Reviewed-by: Joshua Pawlicki <waffles@chromium.org>
Commit-Queue: Jan Wilken Dörrie <jdoerrie@chromium.org>
Cr-Commit-Position: refs/heads/master@{#797629}

chrome/browser/component_updater/zxcvbn_data_component_installer.cc

@@ -19,6 +19,7 @@
 #include "base/location.h"
 #include "base/logging.h"
 #include "base/memory/scoped_refptr.h"
+#include "base/strings/string_split.h"
 #include "base/task/post_task.h"
 #include "base/task/thread_pool.h"
 #include "base/util/ranges/algorithm.h"
@@ -30,6 +31,7 @@
 #include "components/update_client/update_client_errors.h"
 #include "components/update_client/utils.h"
 #include "third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp"
+#include "third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists_common.hpp"
 
 namespace component_updater {
 
@@ -69,33 +71,27 @@ constexpr std::array<TagAndFileName, 6> kTagAndFileNamePairs = {{
      ZxcvbnDataComponentInstallerPolicy::kUsTvAndFilmTxtFileName},
 }};
 
-using DictionaryMap = base::flat_map<zxcvbn::DictionaryTag, std::string>;
-DictionaryMap ReadDictionaries(const base::FilePath& install_dir) {
-  DictionaryMap result;
-  result.reserve(kTagAndFileNamePairs.size());
+using RankedDictionaries =
+    std::unordered_map<zxcvbn::DictionaryTag, zxcvbn::RankedDict>;
+RankedDictionaries ParseRankedDictionaries(const base::FilePath& install_dir) {
+  RankedDictionaries result;
   for (const auto& pair : kTagAndFileNamePairs) {
     base::FilePath dictionary_path = install_dir.Append(pair.file_name);
     DVLOG(1) << "Reading Dictionary from file: " << dictionary_path;
 
     std::string dictionary;
-    if (base::ReadFileToString(dictionary_path, &dictionary))
-      result.try_emplace(result.end(), pair.tag, std::move(dictionary));
-    else
+    if (base::ReadFileToString(dictionary_path, &dictionary)) {
+      result.emplace(pair.tag, zxcvbn::build_ranked_dict(base::SplitStringPiece(
+                                   dictionary, "\r\n", base::TRIM_WHITESPACE,
+                                   base::SPLIT_WANT_NONEMPTY)));
+    } else {
       VLOG(1) << "Failed reading from " << dictionary_path;
+    }
   }
 
   return result;
 }
 
-void ParseDictionaries(const DictionaryMap& dictionaries) {
-  for (const auto& pair : dictionaries) {
-    if (!zxcvbn::ParseRankedDictionary(pair.first, pair.second)) {
-      VLOG(1) << "Failed to parse dictionary for tag "
-              << static_cast<int>(pair.first) << ". Contents: " << pair.second;
-    }
-  }
-}
-
 // The SHA256 of the SubjectPublicKeyInfo used to sign the extension.
 // The extension id is: ojhpjlocmbogdgmfpkhlaaeamibhnphh
 constexpr std::array<uint8_t, 32> kZxcvbnDataPublicKeySha256 = {
@@ -141,8 +137,8 @@ void ZxcvbnDataComponentInstallerPolicy::ComponentReady(
 
   base::ThreadPool::PostTaskAndReplyWithResult(
       FROM_HERE, {base::MayBlock(), base::TaskPriority::BEST_EFFORT},
-      base::BindOnce(&ReadDictionaries, install_dir),
-      base::BindOnce(&ParseDictionaries));
+      base::BindOnce(&ParseRankedDictionaries, install_dir),
+      base::BindOnce(&zxcvbn::SetRankedDicts));
 }
 
 base::FilePath ZxcvbnDataComponentInstallerPolicy::GetRelativeInstallDir()

chrome/browser/component_updater/zxcvbn_data_component_installer_unittest.cc

@@ -17,8 +17,8 @@ namespace component_updater {
 
 namespace {
 
-using ::testing::ElementsAre;
 using ::testing::Pair;
+using ::testing::UnorderedElementsAre;
 
 }  // namespace
 
@@ -103,15 +103,15 @@ TEST_F(ZxcvbnDataComponentInstallerPolicyTest, ComponentReady) {
   base::WriteFile(
       GetPath().Append(
           ZxcvbnDataComponentInstallerPolicy::kEnglishWikipediaTxtFileName),
-      "english_wikipedia");
+      "english\nwikipedia");
   base::WriteFile(
       GetPath().Append(
           ZxcvbnDataComponentInstallerPolicy::kFemaleNamesTxtFileName),
-      "female_names");
+      "female\nnames");
   base::WriteFile(
       GetPath().Append(
           ZxcvbnDataComponentInstallerPolicy::kMaleNamesTxtFileName),
-      "male_names");
+      "male\nnames");
   base::WriteFile(
       GetPath().Append(
           ZxcvbnDataComponentInstallerPolicy::kPasswordsTxtFileName),
@@ -122,26 +122,28 @@ TEST_F(ZxcvbnDataComponentInstallerPolicyTest, ComponentReady) {
   base::WriteFile(
       GetPath().Append(
           ZxcvbnDataComponentInstallerPolicy::kUsTvAndFilmTxtFileName),
-      "us_tv_and_film");
+      "us\ntv\nand\nfilm");
 
   policy().ComponentReady(version(), GetPath(), nullptr);
   task_env().RunUntilIdle();
 
   zxcvbn::RankedDicts ranked_dicts = zxcvbn::default_ranked_dicts();
-  EXPECT_THAT(zxcvbn::default_ranked_dicts(),
-              ::testing::UnorderedElementsAre(
-                  Pair(zxcvbn::DictionaryTag::ENGLISH_WIKIPEDIA,
-                       ElementsAre(Pair("english_wikipedia", 1))),
-                  Pair(zxcvbn::DictionaryTag::FEMALE_NAMES,
-                       ElementsAre(Pair("female_names", 1))),
-                  Pair(zxcvbn::DictionaryTag::MALE_NAMES,
-                       ElementsAre(Pair("male_names", 1))),
-                  Pair(zxcvbn::DictionaryTag::PASSWORDS,
-                       ElementsAre(Pair("passwords", 1))),
-                  Pair(zxcvbn::DictionaryTag::SURNAMES,
-                       ElementsAre(Pair("surnames", 1))),
-                  Pair(zxcvbn::DictionaryTag::US_TV_AND_FILM,
-                       ElementsAre(Pair("us_tv_and_film", 1)))));
+  EXPECT_THAT(
+      zxcvbn::default_ranked_dicts(),
+      UnorderedElementsAre(
+          Pair(zxcvbn::DictionaryTag::ENGLISH_WIKIPEDIA,
+               UnorderedElementsAre(Pair("english", 1), Pair("wikipedia", 2))),
+          Pair(zxcvbn::DictionaryTag::FEMALE_NAMES,
+               UnorderedElementsAre(Pair("female", 1), Pair("names", 2))),
+          Pair(zxcvbn::DictionaryTag::MALE_NAMES,
+               UnorderedElementsAre(Pair("male", 1), Pair("names", 2))),
+          Pair(zxcvbn::DictionaryTag::PASSWORDS,
+               UnorderedElementsAre(Pair("passwords", 1))),
+          Pair(zxcvbn::DictionaryTag::SURNAMES,
+               UnorderedElementsAre(Pair("surnames", 1))),
+          Pair(zxcvbn::DictionaryTag::US_TV_AND_FILM,
+               UnorderedElementsAre(Pair("us", 1), Pair("tv", 2),
+                                    Pair("and", 3), Pair("film", 4)))));
 }
 
 }  // namespace component_updater

third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.cpp

 #include <zxcvbn/frequency_lists.hpp>
 
 #include <unordered_map>
+#include <utility>
 
 #include "base/no_destructor.h"
-#include "base/strings/string_split.h"
-#include "base/strings/string_piece.h"
 
 namespace zxcvbn {
 
@@ -16,16 +15,10 @@ std::unordered_map<DictionaryTag, RankedDict>& ranked_dicts() {
   return *ranked_dicts;
 }
 
-}
-
-bool ParseRankedDictionary(DictionaryTag tag, base::StringPiece str) {
-  RankedDict& dict = ranked_dicts()[tag];
-  if (!dict.empty())
-    return false;
+}  // namespace
 
-  dict = build_ranked_dict(base::SplitStringPiece(
-      str, "\r\n", base::TRIM_WHITESPACE, base::SPLIT_WANT_NONEMPTY));
-  return true;
+void SetRankedDicts(std::unordered_map<DictionaryTag, RankedDict> dicts) {
+  ranked_dicts() = std::move(dicts);
 }
 
 RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict> & ranked_dicts) {

third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp

@@ -38,7 +38,7 @@ namespace zxcvbn {
 
 using RankedDicts = std::unordered_map<DictionaryTag, const RankedDict &>;
 
-bool ParseRankedDictionary(DictionaryTag tag, base::StringPiece str);
+void SetRankedDicts(std::unordered_map<DictionaryTag, RankedDict> dicts);
 
 RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict> & ranked_dicts);
 RankedDicts default_ranked_dicts();

third_party/zxcvbn-cpp/patches/dictionary_component.diff

 diff --git a/third_party/zxcvbn-cpp/data-scripts/build_frequency_lists.py b/third_party/zxcvbn-cpp/data-scripts/build_frequency_lists.py
-index d5504e073cb3..9be1391c5aa9 100755
+index d5504e0..9be1391 100755
 --- a/third_party/zxcvbn-cpp/data-scripts/build_frequency_lists.py
 +++ b/third_party/zxcvbn-cpp/data-scripts/build_frequency_lists.py
 @@ -122,8 +122,7 @@ def to_kv(lst, lst_name):
@@ -37,20 +37,19 @@ index d5504e073cb3..9be1391c5aa9 100755
          output_fn = output_coffee
  
 diff --git a/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.cpp b/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.cpp
-index 4e4b72756e34..17c73a67bd0a 100644
+index 4e4b727..a7adbe9 100644
 --- a/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.cpp
 +++ b/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.cpp
-@@ -1,11 +1,33 @@
+@@ -1,11 +1,26 @@
  #include <zxcvbn/frequency_lists.hpp>
  
 -#include <zxcvbn/_frequency_lists.hpp>
 -
  #include <unordered_map>
- 
-+#include "base/no_destructor.h"
-+#include "base/strings/string_split.h"
-+#include "base/strings/string_piece.h"
++#include <utility>
 +
++#include "base/no_destructor.h"
+ 
  namespace zxcvbn {
  
 +namespace {
@@ -61,22 +60,16 @@ index 4e4b72756e34..17c73a67bd0a 100644
 +  return *ranked_dicts;
 +}
 +
-+}
-+
-+bool ParseRankedDictionary(DictionaryTag tag, base::StringPiece str) {
-+  RankedDict& dict = ranked_dicts()[tag];
-+  if (!dict.empty())
-+    return false;
++}  // namespace
 +
-+  dict = build_ranked_dict(base::SplitStringPiece(
-+      str, "\r\n", base::TRIM_WHITESPACE, base::SPLIT_WANT_NONEMPTY));
-+  return true;
++void SetRankedDicts(std::unordered_map<DictionaryTag, RankedDict> dicts) {
++  ranked_dicts() = std::move(dicts);
 +}
 +
  RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict> & ranked_dicts) {
    RankedDicts build;
  
-@@ -17,8 +39,7 @@ RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict
+@@ -17,8 +32,7 @@ RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict
  }
  
  RankedDicts default_ranked_dicts() {
@@ -87,10 +80,10 @@ index 4e4b72756e34..17c73a67bd0a 100644
 -
  }
 diff --git a/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp b/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp
-index c75ea30d5a91..191842fb530c 100644
+index c75ea30..634ecb2 100644
 --- a/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp
 +++ b/third_party/zxcvbn-cpp/native-src/zxcvbn/frequency_lists.hpp
-@@ -2,15 +2,24 @@
+@@ -2,7 +2,6 @@
  #define __ZXCVBN__FREQUENCY_LISTS_HPP
  
  #include <zxcvbn/frequency_lists_common.hpp>
@@ -98,10 +91,8 @@ index c75ea30d5a91..191842fb530c 100644
  
  #include <unordered_map>
  
- #include <cstdint>
+@@ -10,7 +9,15 @@
  
-+#include "base/strings/string_piece.h"
-+
  namespace zxcvbn {
  
 -using DictionaryTag = _frequency_lists::DictionaryTag;
@@ -117,11 +108,11 @@ index c75ea30d5a91..191842fb530c 100644
  
  }
  
-@@ -29,6 +38,8 @@ namespace zxcvbn {
+@@ -29,6 +36,8 @@ namespace zxcvbn {
  
  using RankedDicts = std::unordered_map<DictionaryTag, const RankedDict &>;
  
-+bool ParseRankedDictionary(DictionaryTag tag, base::StringPiece str);
++void SetRankedDicts(std::unordered_map<DictionaryTag, RankedDict> dicts);
 +
  RankedDicts convert_to_ranked_dicts(std::unordered_map<DictionaryTag, RankedDict> & ranked_dicts);
  RankedDicts default_ranked_dicts();

third_party/zxcvbn-cpp/test/matching_unittest.cc

@@ -304,7 +304,7 @@ TEST(ZxcvbnTest, DictionaryMatching) {
 
   {
     // default dictionaries
-    ParseRankedDictionary(DictionaryTag::US_TV_AND_FILM, "wow");
+    SetRankedDicts({{DictionaryTag::US_TV_AND_FILM, {{"wow", 1}}}});
     std::vector<Match> matches =
         dictionary_match("wow", default_ranked_dicts());
     EXPECT_THAT(matches, ElementsAre(ExpectedDictionaryMatch{
@@ -319,7 +319,7 @@ TEST(ZxcvbnTest, DictionaryMatching) {
 
   {
     // matches with provided user input dictionary
-    ParseRankedDictionary(DictionaryTag::USER_INPUTS, "foo\nbar");
+    SetRankedDicts({{DictionaryTag::USER_INPUTS, {{"foo", 1}, {"bar", 2}}}});
     std::vector<Match> matches =
         dictionary_match("foobar", default_ranked_dicts());
     EXPECT_THAT(matches, ElementsAre(
@@ -1004,7 +1004,8 @@ TEST(ZxcvbnTest, DateMatching) {
 TEST(ZxcvbnTest, Omnimatch) {
   EXPECT_THAT(omnimatch(""), IsEmpty());
 
-  ParseRankedDictionary(DictionaryTag::ENGLISH_WIKIPEDIA, "rosebud\nmaelstrom");
+  SetRankedDicts(
+      {{DictionaryTag::ENGLISH_WIKIPEDIA, {{"rosebud", 1}, {"maelstrom", 2}}}});
   std::string password = "r0sebudmaelstrom11/20/91aaaa";
   std::vector<Match> matches = omnimatch(password);