Sync for validator cpp engine and cpp htmlparser (#36271)

* Add glog/logging.h for LOG(FATAL) and LOG_IF(FATAL, ...)

PiperOrigin-RevId: 400277046

* Fix build rules to regenerate states.h

PiperOrigin-RevId: 400280277

* Fix missing flag declaration in htmldataset_test.cc

PiperOrigin-RevId: 400288157

* Use com_googlesource_re2 instead of com_github_re2

PiperOrigin-RevId: 400295263

* Add sha256 in WORKSPACE files and other clean up.
Moved com_google_protobuf to first in WORKSPACEs.
Removed unused com_github_re2 http_archive.

PiperOrigin-RevId: 400304559

validator/cpp/engine/BUILD

@@ -8,7 +8,11 @@ load("embed_data.bzl", "embed_data")
 # Usage:
 # To build the library, use the following bazel command:
 #
-# bazel build --repo_env=CC=clang --cxxopt='-std=c++17' <build_target>
+# bazel build --repo_env=CC=clang --cxxopt='-std=c++17' validator
+#
+# To run the tests, use the following bazel command:
+#
+# bazel test --cxxopt='-std=c++17' validator_test
 
 package(default_visibility = ["//visibility:public"])
 
@@ -36,13 +40,13 @@ cc_library(
     name = "parse-layout",
     srcs = ["parse-layout.cc"],
     hdrs = ["parse-layout.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/strings",
         "@com_google_absl//absl/strings:str_format",
         "@com_googlesource_code_re2//:re2",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -60,12 +64,12 @@ cc_library(
     name = "parse-layout-sizes",
     srcs = ["parse-layout-sizes.cc"],
     hdrs = ["parse-layout-sizes.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/strings",
         "@com_googlesource_code_re2//:re2",
         "@htmlparser//:strings",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -82,13 +86,13 @@ cc_library(
     name = "parse-srcset",
     srcs = ["parse-srcset.cc"],
     hdrs = ["parse-srcset.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/base",
         "@com_google_absl//absl/strings",
         "@com_googlesource_code_re2//:re2",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -105,13 +109,13 @@ cc_library(
     name = "parse-viewport",
     srcs = ["parse-viewport.cc"],
     hdrs = ["parse-viewport.h"],
+    copts = ["-std=c++17"],
     deps = [
         ":utf8-util",
         "@com_google_absl//absl/base",
         "@com_google_absl//absl/strings",
         "@htmlparser//:strings",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -128,13 +132,13 @@ cc_library(
     name = "utf8-util",
     srcs = ["utf8-util.cc"],
     hdrs = ["utf8-util.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/base",
         "@com_google_absl//absl/strings",
         "@com_google_absl//absl/strings:cord",
         "@htmlparser//:strings",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -152,13 +156,13 @@ cc_library(
     name = "keyframes-parse-css",
     srcs = ["keyframes-parse-css.cc"],
     hdrs = ["keyframes-parse-css.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/base",
         "@com_google_absl//absl/memory",
         "@com_google_absl//absl/strings",
         "@htmlparser//css:parse-css",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -177,11 +181,11 @@ cc_library(
     name = "type-identifier",
     srcs = ["type-identifier.cc"],
     hdrs = ["type-identifier.h"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_absl//absl/strings",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -198,14 +202,14 @@ cc_library(
     name = "error-formatter",
     srcs = ["error-formatter.cc"],
     hdrs = ["error-formatter.h"],
+    copts = ["-std=c++17"],
     deps = [
         ":validator_pb",
-        "@com_googlesource_code_re2//:re2",
         "@com_github_glog//:glog",
+        "@com_googlesource_code_re2//:re2",
         "@htmlparser//:strings",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(
@@ -221,13 +225,13 @@ cc_test(
 cc_binary(
     name = "validator-convert",
     srcs = ["validator-convert.cc"],
+    copts = ["-std=c++17"],
     deps = [
         "@com_google_protobuf//:protobuf",
         "@htmlparser//:defer",
         "@htmlparser//:fileutil",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 genrule(
@@ -244,6 +248,7 @@ cc_library(
     srcs = [
         "validator-internal.cc",
     ],
+    copts = ["-std=c++17"],
     visibility = ["//visibility:private"],
     deps = [
         ":keyframes-parse-css",
@@ -279,20 +284,19 @@ cc_library(
         "@htmlparser//json:parser",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_library(
     name = "validator",
     hdrs = [
         "validator.h",
     ],
+    copts = ["-std=c++17"],
     deps = [
         ":validator-internal",
         "@validator//:validator_cc_proto",
     ],
     alwayslink = 1,
-    copts = ["-std=c++17"],
 )
 
 cc_library(
@@ -303,9 +307,10 @@ cc_library(
     hdrs = [
         "testing-utils.h",
     ],
+    copts = ["-std=c++17"],
     data = [
-        "@validator//:testdata_files",
         "@amphtml-extensions//:testdata_files",
+        "@validator//:testdata_files",
     ],
     deps = [
         ":error-formatter",
@@ -314,7 +319,6 @@ cc_library(
         "@htmlparser//:strings",
         "@validator//:validator_cc_proto",
     ],
-    copts = ["-std=c++17"],
 )
 
 cc_test(

validator/cpp/engine/scripts/BUILD

@@ -1,7 +1,7 @@
-licenses(["notice"])
-
 package(default_visibility = ["//visibility:public"])
 
+licenses(["notice"])
+
 cc_binary(
     name = "filecontents_to_cpp_header",
     srcs = [

validator/cpp/htmlparser/bin/BUILD

@@ -54,15 +54,17 @@ cc_binary(
 )
 
 cc_binary(
-    name = "jsongrammargen",
+    name = "statetablegen",
     srcs = [
-        "jsongrammargen.cc",
+        "statetablegen.cc",
     ],
     data = [
         "@htmlparser//data:jsongrammar.txt",
         "@htmlparser//json:states.h",
     ],
     deps = [
+        "@com_google_absl//absl/flags:flag",
+        "@com_google_absl//absl/flags:parse",
         "@htmlparser//grammar:tablebuilder",
     ],
 )

validator/cpp/htmlparser/css/BUILD

@@ -96,9 +96,9 @@ cc_library(
     hdrs = ["amp4ads-parse-css.h"],
     deps = [
         ":parse-css",
-        "@com_github_re2//:re2",
         "@com_google_absl//absl/memory",
         "@com_google_absl//absl/strings",
+        "@com_googlesource_code_re2//:re2",
     ],
 )
 

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

@@ -8,6 +8,7 @@
 #include "absl/strings/numbers.h"
 #include "absl/strings/str_join.h"
 #include "absl/types/variant.h"
+#include "glog/logging.h"
 #include "logging.h"
 #include "strings.h"
 

validator/cpp/htmlparser/grammar/tablebuilder.cc

@@ -165,7 +165,7 @@ bool TableBuilder::OutputHeaderFile(
   fd << "// -*- C++ -*-\n";
 
   fd << "\n// AUTO GENERATED; DO NOT EDIT.\n";
-  fd << "// To regenerate this file, see comments in bin/jsongrammargen\n\n";
+  fd << "// To regenerate this file, see comments in bin/statetablegen\n\n";
   if (!header_options_.ifdef_guard.empty()) {
     fd << "#ifndef " << header_options_.ifdef_guard << "\n";
     fd << "#define " << header_options_.ifdef_guard << "\n\n";

validator/cpp/htmlparser/htmldataset_test.cc

@@ -20,6 +20,8 @@
 #include "testconstants.h"
 #include "tokenizer.h"
 
+ABSL_FLAG(std::string, test_srcdir, "", "Testdata directory");
+
 using namespace htmlparser;
 
 // Represents a single test case.

validator/cpp/htmlparser/json/states.h

@@ -1,7 +1,7 @@
 // -*- C++ -*-
 
 // AUTO GENERATED; DO NOT EDIT.
-// To regenerate this file, see comments in bin/jsongrammargen
+// To regenerate this file, see comments in bin/statestablegen
 
 #ifndef HTMLPARSER__JSON_STATES_H_
 #define HTMLPARSER__JSON_STATES_H_