speedreader: port to kuchikiki 0.8.2

We renamed our fork of the kuchiki tree-handling crate since upstream is unmaintained. It's easier to build against released crates now that we're vendoring, so move to the new publication crate name. This also addresses a `cargo audit` warning about the unmaintained dependency. Note however that we're still using a brave-specific fork from git here, so the crate itself is still installed from DEPS. This adds a new transitive dependency on `indexmap`, which is used to roundtrip html attributes with stable ordering so unit tests don't need trivial updates whenever we touch something. This dependency is already available in upstream chromium. kuchikiki 0.8.2 specifies newer versions of its other dependencies but seems to still work when built against the old releases, so those are left to another commit.
brave · Sep 19, 2023 · 32203ba · 32203ba
1 parent e44aea2
commit 32203ba
Show file tree

Hide file tree

Showing 19 changed files with 85 additions and 86 deletions.
diff --git a/DEPS b/DEPS
@@ -38,7 +38,7 @@ deps = {
   "third_party/rust/star_constellation/v0_2/crate": "https://github.com/brave/constellation.git@db575edec12509ce1bda6afe68bb58e538a21d3a",
   "third_party/rust/challenge_bypass_ristretto/v1/crate": "https://github.com/brave-intl/challenge-bypass-ristretto.git@a1da4641734adc8312215b38a8221962d2c8e045",
   "third_party/rust/futures_retry/v0_5/crate": "https://github.com/brave-intl/futures-retry.git@2aaaafbc3d394661534d4dbd14159d164243c20e",
-  "third_party/rust/kuchiki/v0_8/crate": "https://github.com/brave/kuchiki.git@589eadca2c1d06ddda2919354590bfe1ace88a43",
+  "third_party/rust/kuchikiki/v0_8/crate": "https://github.com/brave/kuchiki.git@436b4e21274e64c5f9bc142ba6fa330fe9a8c97d",
   "third_party/macholib": {
     "url": "https://github.com/ronaldoussoren/macholib.git@36a6777ccd0891c5d1b44ba885573d7c90740015",
     "condition": "checkout_mac",

diff --git a/components/speedreader/rust/ffi/Cargo.toml b/components/speedreader/rust/ffi/Cargo.toml
@@ -16,5 +16,5 @@ name = "speedreader_ffi"
 path = "lib.rs"
 crate-type = ["rlib"]
 
-[patch.crates-io.kuchiki]
-path = "../../../../third_party/rust/kuchiki/v0_8/crate"
+[patch.crates-io.kuchikiki]
+path = "../../../../third_party/rust/kuchikiki/v0_8/crate"
diff --git a/components/speedreader/rust/lib/BUILD.gn b/components/speedreader/rust/lib/BUILD.gn
@@ -18,7 +18,7 @@ rust_static_library("lib") {
   deps = [
     "src/readability:lib",
     "//brave/third_party/rust/html5ever/v0_25:lib",
-    "//brave/third_party/rust/kuchiki/v0_8:lib",
+    "//brave/third_party/rust/kuchikiki/v0_8:lib",
     "//brave/third_party/rust/lifeguard/v0_6:lib",
     "//brave/third_party/rust/lol_html/v0_3:lib",
     "//brave/third_party/rust/markup5ever/v0_10:lib",

diff --git a/components/speedreader/rust/lib/Cargo.toml b/components/speedreader/rust/lib/Cargo.toml
@@ -18,7 +18,7 @@ include = [
 
 [dependencies]
 html5ever = "0.25.1"
-kuchiki = "0.8.1"
+kuchikiki = "0.8.2"
 lol_html = "0.3.0"
 regex = "1"
 serde_json = "1.0.70"
@@ -44,6 +44,5 @@ criterion = "0.3"
 futures = "0.3"
 tokio = { version = "1.2", features = ["full"] }
 
-[patch.crates-io.kuchiki_v0_8]
-path = "../../../../third_party/rust/kuchiki/v0_8/crate"
-package = "kuchiki"
+[patch.crates-io.kuchikiki]
+path = "../../../../third_party/rust/kuchikiki/v0_8/crate"
diff --git a/components/speedreader/rust/lib/src/readability/BUILD.gn b/components/speedreader/rust/lib/src/readability/BUILD.gn
@@ -23,7 +23,7 @@ rust_static_library("lib") {
 
   deps = [
     "//brave/third_party/rust/html5ever/v0_25:lib",
-    "//brave/third_party/rust/kuchiki/v0_8:lib",
+    "//brave/third_party/rust/kuchikiki/v0_8:lib",
     "//brave/third_party/rust/thiserror/v1:lib",
     "//brave/third_party/rust/time/v0_3:lib",
     "//brave/third_party/rust/url/v2:lib",

diff --git a/components/speedreader/rust/lib/src/readability/Cargo.toml b/components/speedreader/rust/lib/src/readability/Cargo.toml
@@ -13,7 +13,7 @@ edition = "2018"
 [dependencies]
 bitflags         = "1.2"
 html5ever        = "0.25.1"
-kuchiki          = "0.8.1"
+kuchikiki        = "0.8.2"
 lazy_static      = "1.4"
 regex            = "1"
 serde_json       = "1.0"
@@ -25,6 +25,5 @@ url              = "2.2"
 assert_approx_eq = "1.1.0"
 reqwest = { version = "0.11.6", features = ["blocking", "stream"] }
 
-[patch.crates-io.kuchiki_v0_8]
-path = "../../../../../third_party/rust/kuchiki/v0_8/crate"
-package = "kuchiki"
+[patch.crates-io.kuchiki]
+path = "../../../../../third_party/rust/kuchikiki/v0_8/crate"
diff --git a/components/speedreader/rust/lib/src/readability/src/dom.rs b/components/speedreader/rust/lib/src/readability/src/dom.rs
@@ -10,10 +10,10 @@ use html5ever::tendril::TendrilSink;
 use html5ever::tree_builder::{ElementFlags, NodeOrText};
 use html5ever::{parse_document, ParseOpts};
 use html5ever::{Attribute, LocalName, QualName};
-use kuchiki::iter::NodeIterator;
-use kuchiki::NodeData::{Element, Text};
-use kuchiki::NodeRef as Handle;
-use kuchiki::Sink;
+use kuchikiki::iter::NodeIterator;
+use kuchikiki::NodeData::{Element, Text};
+use kuchikiki::NodeRef as Handle;
+use kuchikiki::Sink;
 use std::str::FromStr;
 
 /// A small wrapper function that creates a NodeOrText from a Text handle or an

diff --git a/components/speedreader/rust/lib/src/readability/src/extractor.rs b/components/speedreader/rust/lib/src/readability/src/extractor.rs
@@ -8,8 +8,8 @@ use html5ever::parse_document;
 use html5ever::tendril::TendrilSink;
 use html5ever::tree_builder::{ElementFlags, NodeOrText, TreeSink};
 use html5ever::QualName;
-use kuchiki::NodeRef as Handle;
-use kuchiki::Sink;
+use kuchikiki::NodeRef as Handle;
+use kuchikiki::Sink;
 use regex::Regex;
 use std::collections::{HashMap, HashSet};
 use std::default::Default;

diff --git a/components/speedreader/rust/lib/src/readability/src/lib.rs b/components/speedreader/rust/lib/src/readability/src/lib.rs
@@ -1,6 +1,6 @@
 #[macro_use]
 extern crate html5ever;
-extern crate kuchiki;
+extern crate kuchikiki;
 extern crate regex;
 extern crate url;
 #[macro_use]

diff --git a/components/speedreader/rust/lib/src/readability/src/scorer.rs b/components/speedreader/rust/lib/src/readability/src/scorer.rs
@@ -3,12 +3,12 @@ use html5ever::tendril::StrTendril;
 use html5ever::tree_builder::TreeSink;
 use html5ever::tree_builder::{ElementFlags, NodeOrText};
 use html5ever::{LocalName, QualName};
-use kuchiki::iter::NodeIterator;
-use kuchiki::NodeData::{
+use kuchikiki::iter::NodeIterator;
+use kuchikiki::NodeData::{
     Comment, Doctype, Document, DocumentFragment, Element, ProcessingInstruction, Text,
 };
-use kuchiki::NodeRef as Handle;
-use kuchiki::{ElementData, Sink};
+use kuchikiki::NodeRef as Handle;
+use kuchikiki::{ElementData, Sink};
 use regex::Regex;
 use std::cmp::Ordering;
 use std::collections::{HashMap, HashSet};

diff --git a/components/speedreader/rust/lib/src/readability/src/statistics.rs b/components/speedreader/rust/lib/src/readability/src/statistics.rs
@@ -1,7 +1,7 @@
 use crate::{dom, util};
-use kuchiki::iter::NodeIterator;
-use kuchiki::NodeRef as Handle;
-use kuchiki::{ElementData, Sink};
+use kuchikiki::iter::NodeIterator;
+use kuchikiki::NodeRef as Handle;
+use kuchikiki::{ElementData, Sink};
 use util::count_ignore_consecutive_whitespace;
 
 #[derive(Default)]

diff --git a/components/speedreader/rust/lib/src/speedreader_readability.rs b/components/speedreader/rust/lib/src/speedreader_readability.rs
@@ -1,6 +1,6 @@
 use html5ever::driver::{ParseOpts, Parser};
 use html5ever::tendril::{StrTendril, TendrilSink};
-use kuchiki::Sink;
+use kuchikiki::Sink;
 use lol_html::OutputSink;
 use std::collections::HashMap;
 use url::Url;

diff --git a/components/speedreader/rust/lib/tests/legacy.rs b/components/speedreader/rust/lib/tests/legacy.rs
@@ -19,9 +19,9 @@ use std::fs::File;
 use std::io::Read;
 use url::Url;
 
-use kuchiki::NodeData::{Element, Text};
-use kuchiki::NodeRef as Handle;
-use kuchiki::Sink;
+use kuchikiki::NodeData::{Element, Text};
+use kuchikiki::NodeRef as Handle;
+use kuchikiki::Sink;
 use regex::Regex;
 use std::vec::Vec;
 

diff --git a/components/speedreader/rust/lib/tests/lib.rs b/components/speedreader/rust/lib/tests/lib.rs
@@ -13,8 +13,8 @@ use url::Url;
 
 use distance::damerau_levenshtein;
 use html5ever::LocalName;
-use kuchiki::NodeData::{Element, Text};
-use kuchiki::NodeRef as Handle;
+use kuchikiki::NodeData::{Element, Text};
+use kuchikiki::NodeRef as Handle;
 use std::vec::Vec;
 
 static SAMPLES_PATH: &str = "data/tests-samples/";

diff --git a/third_party/rust/Cargo.lock b/third_party/rust/Cargo.lock
diff --git a/third_party/rust/Cargo.toml b/third_party/rust/Cargo.toml
@@ -31,7 +31,7 @@ fvm_shared = "3"
 hex = "0.4"
 hex-literal = "0.4"
 hmac = "0.10"
-kuchiki = "0.8"
+kuchikiki = "0.8"
 lazy_static = "1"
 libc = "0.2"
 lifeguard = "0.6"
@@ -692,9 +692,9 @@ package = "itoa"
 path = "keccak/v0_1/crate"
 package = "keccak"
 
-[patch.crates-io.kuchiki_v0_8]
-path = "kuchiki/v0_8/crate"
-package = "kuchiki"
+[patch.crates-io.kuchikiki_v0_8]
+path = "kuchikiki/v0_8/crate"
+package = "kuchikiki"
 
 [patch.crates-io.lazy_static_v1]
 path = "../../../third_party/rust/lazy_static/v1/crate"

diff --git a/third_party/rust/kuchiki/v0_8/BUILD.gn b/third_party/rust/kuchiki/v0_8/BUILD.gn
diff --git a/third_party/rust/kuchikiki/v0_8/BUILD.gn b/third_party/rust/kuchikiki/v0_8/BUILD.gn
@@ -0,0 +1,46 @@
+# Copyright 2023 The Chromium Authors
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+import("//build/rust/cargo_crate.gni")
+
+cargo_crate("lib") {
+  crate_name = "kuchikiki"
+  epoch = "0.8"
+  crate_type = "rlib"
+  crate_root = "crate/src/lib.rs"
+  sources = [
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/examples/find_matches.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/examples/stack-overflow.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/attributes.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/cell_extras.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/iter.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/lib.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/node_data_ref.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/parser.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/select.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/serializer.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/tests.rs",
+    "//brave/third_party/rust/kuchikiki/v0_8/crate/src/tree.rs",
+  ]
+  inputs = [ "//brave/third_party/rust/kuchikiki/v0_8/crate/README.md" ]
+
+  # Unit tests skipped. Generate with --with-tests to include them.
+  build_native_rust_unit_tests = false
+  edition = "2018"
+  cargo_pkg_version = "0.8.2"
+  cargo_pkg_authors = "Simon Sapin <[email protected]>"
+  cargo_pkg_name = "kuchikiki"
+  cargo_pkg_description = "(口利き) HTML tree manipulation library"
+  library_configs -= [ "//build/config/compiler:chromium_code" ]
+  library_configs += [ "//build/config/compiler:no_chromium_code" ]
+  executable_configs -= [ "//build/config/compiler:chromium_code" ]
+  executable_configs += [ "//build/config/compiler:no_chromium_code" ]
+  deps = [
+    "//brave/third_party/rust/cssparser/v0_27:lib",
+    "//brave/third_party/rust/html5ever/v0_25:lib",
+    "//third_party/rust/indexmap/v1:lib",
+    "//brave/third_party/rust/matches/v0_1:lib",
+    "//brave/third_party/rust/selectors/v0_22:lib",
+  ]
+}
diff --git a/third_party/rust/third_party.toml b/third_party/rust/third_party.toml
@@ -63,7 +63,7 @@ skrifa = "0.4"
 libc = "0.2.107"
 
 # speedreader
-kuchiki = "0.8.1"
+kuchikiki = "0.8.1"
 lol_html = "0.3.0"
 # regex = "1"
 # serde_json = "1.0.70"
@@ -74,7 +74,7 @@ lazy_static = "1.4"
 
 # readability
 bitflags         = "1.2"
-kuchiki          = "0.8.1"
+kuchikiki        = "0.8.1"
 lazy_static      = "1.4"
 # regex            = "1"
 # serde_json       = "1.0"