Sync htmlparser changes from g3

.circleci/config.yml

@@ -289,19 +289,6 @@ jobs:
           name: '⭐⭐⭐ Bundle Size ⭐⭐⭐'
           command: node build-system/pr-check/bundle-size.js
       - teardown_vm
-  validator_tests:
-    executor:
-      name: jdk-docker-2xlarge
-    steps:
-      - setup_vm
-      - run:
-          name: 'Install Validator Dependencies'
-          command: ./.circleci/install_validator_dependencies.sh
-      - run:
-          name: '⭐⭐⭐ Validator Tests ⭐⭐⭐'
-          command: node build-system/pr-check/validator-tests.js
-      - store_test_output
-      - teardown_vm
   visual_diff_tests:
     executor:
       name: node-docker-large
@@ -565,11 +552,6 @@ workflows:
             - 'Nomodule 3p Build (Bundle Size)'
             - '⛓️ Module Build (Bundle Size)'
             - 'Module 3p Build (Bundle Size)'
-      - validator_tests:
-          name: 'Validator Tests'
-          <<: *push_and_pr_builds
-          requires:
-            - 'Initialize Repository'
       - visual_diff_tests:
           name: 'Visual Diff Tests'
           <<: *push_and_pr_builds

validator/WORKSPACE

@@ -18,9 +18,8 @@ http_archive(
 
 http_archive(
     name = "com_google_protobuf",
-    sha256 = "750428a8c7f1a75a8e6027e30b46a1c2f0475205f44563589736e0f279b938c0",
-    strip_prefix = "protobuf-f0dc78d7e6e331b8c6bb2d5283e06aa26883ca7c",
-    urls = ["https://github.com/protocolbuffers/protobuf/archive/f0dc78d7e6e331b8c6bb2d5283e06aa26883ca7c.zip"],
+    strip_prefix = "protobuf-43e1626812c1b543e56a7bec59dc09eb18248bd2",
+    urls = ["https://github.com/protocolbuffers/protobuf/archive/43e1626812c1b543e56a7bec59dc09eb18248bd2.zip"],
 )
 load("@com_google_protobuf//:protobuf_deps.bzl", "protobuf_deps")
 protobuf_deps()

validator/cpp/htmlparser/BUILD

@@ -8,7 +8,10 @@ load("@rules_cc//cc:defs.bzl", "cc_library", "cc_test")
 #
 # bazel build --repo_env=CC=clang --cxxopt='-std=c++17' <build_target>
 
-package(default_visibility = ["//visibility:public"])
+package(
+    default_compatible_with = ["//buildenv/target:non_prod"],
+    default_visibility = ["//visibility:public"],
+)
 
 licenses(["notice"])
 
@@ -195,10 +198,10 @@ cc_library(
     hdrs = [
         "logging.h",
     ],
+    copts = ["-std=c++17"],
     deps = [
         ":glog_polyfill",
     ],
-    copts = ["-std=c++17"],
 )
 
 # Defines token type and token structures, used during tokenization.
@@ -463,6 +466,7 @@ cc_test(
         "tokenizer_test.cc",
     ],
     deps = [
+        ":atom",
         ":token",
         ":tokenizer",
         "@com_google_googletest//:gtest_main",
@@ -551,7 +555,7 @@ cc_test(
         ":strings",
         ":testconstants",
         ":tokenizer",
-        "@com_google_googletest//:gtest_main",
         "@com_google_absl//absl/flags:flag",
+        "@com_google_googletest//:gtest_main",
     ],
 )

validator/cpp/htmlparser/allocator.h

@@ -138,7 +138,7 @@
 // IMPORTANT: Tree like structure must not destroy the child nodes or sibling
 // nodes. Allocator destroys all the objects and call its destructor, it is an
 // error to invoke destructors on objects allocated by this allocator. Allocator
-// is the master owner of all the objects. Client treats all objects as const
+// is the primary owner of all the objects. Client treats all objects as const
 // pointer as far as destruction goes.
 //
 // It is not possible to destroy random objects or free up the slots to be
@@ -174,6 +174,7 @@
 #include <cstring>
 #include <memory>
 #include <tuple>
+#include <type_traits>
 #include <vector>
 
 namespace htmlparser {

validator/cpp/htmlparser/allocator_test.cc

@@ -1,5 +1,10 @@
 #include "cpp/htmlparser/allocator.h"
 
+#include <array>
+#include <cstdint>
+#include <string>
+#include <vector>
+
 #include "gtest/gtest.h"
 
 // Memory leaks will automatically be detected by the test framework.
@@ -12,9 +17,9 @@ TEST(AllocatorTest, BasicTest) {
     int32_t a;
     int64_t b;
     std::string c;
-    short d;
+    uint16_t d;
     Data() {}
-    Data(int32_t a_, int64_t b_, std::string c_, short d_)
+    Data(int32_t a_, int64_t b_, std::string c_, uint16_t d_)
         : a(a_), b(b_), c(c_), d(d_) {}
   };
 
@@ -146,7 +151,7 @@ TEST(AllocatorTest, DestructorCalled) {
 
 TEST(AllocatorTest, BitFields) {
   struct HasBitFields {
-    short s;
+    uint16_t s;
     char c;
     int flip : 1;
     int nybble : 4;

validator/cpp/htmlparser/atomutil.cc

@@ -1,5 +1,8 @@
 #include "cpp/htmlparser/atomutil.h"
 
+#include <string>
+#include <string_view>
+
 #include "cpp/htmlparser/hash.h"
 
 namespace htmlparser {
@@ -20,7 +23,7 @@ Atom AtomUtil::ToAtom(const std::string& s) {
   table_index = (hash >> 16) & (kNamesHashTable.size() - 1);
   atom_value = kNamesHashTable[table_index];
   atom_len = atom_value & 0xff;
-  if (atom_len == s.size() && ToString(atom_value).compare(s) == 0) {
+  if (atom_len == s.size() && ToString(atom_value) == s) {
     return CastToAtom(atom_value);
   }
 

validator/cpp/htmlparser/bin/atomgen.cc

@@ -13,6 +13,7 @@
 #include <memory>
 #include <sstream>
 #include <string>
+#include <string_view>
 #include <utility>
 #include <vector>
 
@@ -55,12 +56,12 @@ class TableBuilder {
 
   bool Insert(const std::string& s) {
     std::pair<uint32_t, uint32_t> hashes = Hash(s);
-    if (table_[hashes.first] == "") {
+    if (table_[hashes.first].empty()) {
       table_[hashes.first] = s;
       return true;
     }
 
-    if (table_[hashes.second] == "") {
+    if (table_[hashes.second].empty()) {
       table_[hashes.second] = s;
       return true;
     }
@@ -100,7 +101,7 @@ class TableBuilder {
     std::pair<uint32_t, uint32_t> hashes = Hash(entry);
     uint32_t new_location = hashes.first + hashes.second - i;
 
-    if (table_[new_location] != "" && !Push(new_location, depth + 1)) {
+    if (!table_[new_location].empty() && !Push(new_location, depth + 1)) {
       return false;
     }
 
@@ -190,17 +191,16 @@ int main(int argc, char** argv) {
   // Find hash that minimizes table size.
   std::unique_ptr<TableBuilder> table{nullptr};
   for (int i = 0; i < 1; i++) {
-    if (table.get() != nullptr
-        && (1 << (table->hash_num_bits() - 1)) < all_names.size()) {
+    if (table != nullptr &&
+        (1 << (table->hash_num_bits() - 1)) < all_names.size()) {
       break;
     }
 
     srand(time(nullptr));
     uint32_t rand_state;
     uint32_t hash0 = rand_r(&rand_state) % 1000000000;
     for (uint32_t k = 0; k <= 16; k++) {
-      if (table.get() != nullptr && k >= table->hash_num_bits())
-        break;
+      if (table != nullptr && k >= table->hash_num_bits()) break;
       std::unique_ptr<TableBuilder> base(new TableBuilder());
       if (base->Init(hash0, k, all_names)) {
         table.reset(base.release());
@@ -209,7 +209,7 @@ int main(int argc, char** argv) {
     }
   }
 
-  if (table.get() == nullptr) {
+  if (table == nullptr) {
     std::cerr << "Failed to construct string table." << std::endl;
     std::cerr <<  all_names.size() << ": elements." << std::endl;
     return EXIT_FAILURE;
@@ -224,11 +224,11 @@ int main(int argc, char** argv) {
   while (changed) {
     changed = false;
     for (std::size_t i = 0; i < layout.size(); i++) {
-      if (layout[i] == "") continue;
+      if (layout[i].empty()) continue;
 
       for (std::size_t j = 0; j < layout.size(); j++) {
-        if (i != j && layout[j] != ""
-            && layout[i].find(layout[j]) != std::string::npos) {
+        if (i != j && !layout[j].empty() &&
+            layout[i].find(layout[j]) != std::string::npos) {
           changed = true;
           layout[j] = "";
         }
@@ -244,7 +244,7 @@ int main(int argc, char** argv) {
     int bestj = -1;
     int bestk = 0;
     for (std::size_t i = 0; i < layout.size(); i++) {
-      if (layout[i] == "") continue;
+      if (layout[i].empty()) continue;
 
       for (std::size_t j = 0; j < layout.size(); j++) {
         if (i == j) continue;

validator/cpp/htmlparser/bin/casetablegen.cc

@@ -1,7 +1,9 @@
 #include <fstream>
 #include <iostream>
 #include <sstream>
+#include <string>
 #include <utility>
+#include <vector>
 
 #include "cpp/htmlparser/defer.h"
 #include "cpp/htmlparser/fileutil.h"

validator/cpp/htmlparser/bin/entitytablegen.cc

@@ -6,6 +6,7 @@
 #include <fstream>
 #include <iostream>
 #include <sstream>
+#include <string>
 #include <utility>
 #include <vector>
 

validator/cpp/htmlparser/bin/validatorgen.cc

@@ -15,6 +15,7 @@
 //
 
 #include <iostream>
+#include <string>
 
 #include "glog/logging.h"
 #include "absl/flags/flag.h"

validator/cpp/htmlparser/bluze.textproto

@@ -0,0 +1,44 @@
+# proto-file: devtools/blueprint/blueprint_file.proto
+# proto-message: BlueprintFile
+# DO NOT EDIT! Regenerate the contents by running go/bluze after changing any BUILD file or the Blueprint.
+# Override the default values in third_party.htmlparser.blueprint instead.
+
+buildable_unit: {
+  name: "third_party.htmlparser"
+  build_pattern: "//third_party/htmlparser/..."
+  test_pattern: "//third_party/htmlparser/..."
+  test_tag_filter: "-nofastbuild"
+  build_tag_filter: "-nofastbuild"
+  enable_coverage: true
+  enable_presubmit: true
+  enable_continuous_build: true
+  continuous_build_email: {
+    build_cop_email_id: "other-3p-community-maintainers+htmlparser+buildgardener@google.com"
+  }
+  enable_release: false
+}
+buildable_unit: {
+  name: "third_party.htmlparser.opt"
+  test_pattern: "//third_party/htmlparser/..."
+  test_tag_filter: "-noopt"
+  build_flag: "--compilation_mode=opt"
+  enable_coverage: false
+  enable_presubmit: false
+  enable_continuous_build: false
+  continuous_build_email: {
+    build_cop_email_id: "other-3p-community-maintainers+htmlparser+buildgardener@google.com"
+  }
+  enable_release: false
+  [tap.tap_settings]: {
+    on_demand: true
+    on_demand_frequency: EVERY_4_HOURS
+  }
+}
+continuous_tests: {
+  name: "third_party.htmlparser"
+  buildable_unit_name: "third_party.htmlparser"
+}
+continuous_tests: {
+  name: "third_party.htmlparser.opt"
+  buildable_unit_name: "third_party.htmlparser.opt"
+}

validator/cpp/htmlparser/casetable_test.cc

@@ -1,5 +1,7 @@
 #include "cpp/htmlparser/casetable.h"
 
+#include <string_view>
+
 #include "gtest/gtest.h"
 #include "cpp/htmlparser/strings.h"
 

validator/cpp/htmlparser/css/amp4ads-parse-css.cc

@@ -1,13 +1,14 @@
 #include "cpp/htmlparser/css/amp4ads-parse-css.h"
 
-#include "absl/memory/memory.h"
+#include <memory>
+#include <string>
+#include <vector>
+
 #include "absl/strings/str_cat.h"
 #include "absl/strings/str_join.h"
 #include "cpp/htmlparser/css/parse-css.h"
 #include "re2/re2.h"
 
-using absl::make_unique;
-using std::string_view;
 using amp::validator::ValidationError;
 using std::unique_ptr;
 
@@ -18,7 +19,7 @@ namespace {
 unique_ptr<ErrorToken> CreateParseErrorTokenAt(
     const Token& position_token, ValidationError::Code code,
     const std::vector<std::string>& params) {
-  auto token = make_unique<ErrorToken>(code, params);
+  auto token = std::make_unique<ErrorToken>(code, params);
   position_token.CopyStartPositionTo(token.get());
   return token;
 }

validator/cpp/htmlparser/css/parse-css-urls.cc

@@ -1,6 +1,8 @@
 #include "cpp/htmlparser/css/parse-css-urls.h"
 
 #include <memory>
+#include <string>
+#include <utility>
 
 #include "absl/algorithm/container.h"
 #include "absl/memory/memory.h"

validator/cpp/htmlparser/css/parse-css-urls_test.cc

@@ -1,13 +1,14 @@
 #include "cpp/htmlparser/css/parse-css-urls.h"
 
 #include <ostream>
+#include <string>
+#include <vector>
 
 #include <gmock/gmock.h>
 #include "gtest/gtest.h"
 #include "absl/strings/str_cat.h"
 #include "cpp/htmlparser/strings.h"
 
-using testing::Eq;
 using testing::Pointwise;
 
 namespace htmlparser::css::url {

validator/cpp/htmlparser/css/parse-css.cc

@@ -2,6 +2,11 @@
 
 #include <deque>
 #include <memory>
+#include <sstream>
+#include <string>
+#include <string_view>
+#include <utility>
+#include <vector>
 
 #include "absl/algorithm/container.h"
 #include "absl/memory/memory.h"

validator/cpp/htmlparser/css/parse-css_test.cc

@@ -1,6 +1,10 @@
 #include "cpp/htmlparser/css/parse-css.h"
 
+#include <algorithm>
 #include <memory>
+#include <string>
+#include <string_view>
+#include <utility>
 #include <vector>
 
 #include <gmock/gmock.h>