From 0959c9cc5d3573c7807e4dc9ede3616a0a4381fb Mon Sep 17 00:00:00 2001
From: Markus Unterwaditzer <markus-tarpit+git@unterwaditzer.net>
Date: Tue, 29 Oct 2024 04:14:17 +0100
Subject: [PATCH] add callbacks emitter and update readme (#91)

---
 .github/dependabot.yml            |   6 +
 Cargo.toml                        |   6 +
 README.md                         |  32 +-
 examples/build_tree.rs            |   4 +-
 examples/callback_emitter.rs      |  53 +++
 examples/scraper.rs               |   4 +-
 fuzz/Cargo.lock                   | 569 +++++++++++++-----------------
 fuzz/Cargo.toml                   |  14 +-
 fuzz/README.md                    |   1 +
 fuzz/src/testcase/html5ever.rs    |   8 +-
 fuzz/src/testcase/old_html5gum.rs |  24 +-
 src/callbacks.rs                  | 497 ++++++++++++++++++++++++++
 src/default_emitter.rs            | 266 ++++++++++++++
 src/emitter.rs                    | 446 -----------------------
 src/html5ever_emitter.rs          | 272 +++++++-------
 src/htmlstring.rs                 | 130 +++++++
 src/lib.rs                        |   9 +-
 src/tokenizer.rs                  |  29 +-
 tests/html5lib_tree_builder.rs    |   6 +-
 19 files changed, 1428 insertions(+), 948 deletions(-)
 create mode 100644 examples/callback_emitter.rs
 create mode 100644 src/callbacks.rs
 create mode 100644 src/default_emitter.rs
 create mode 100644 src/htmlstring.rs

diff --git a/.github/dependabot.yml b/.github/dependabot.yml
index 093f76c..4508697 100644
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -10,6 +10,12 @@ updates:
   schedule:
     interval: daily
   open-pull-requests-limit: 10
+
+- package-ecosystem: cargo
+  directory: "/fuzz"
+  schedule:
+    interval: daily
+  open-pull-requests-limit: 10
   
 - package-ecosystem: gitsubmodule
   directory: "/"
diff --git a/Cargo.toml b/Cargo.toml
index 279eea5..414bb21 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -59,6 +59,12 @@ harness = false
 name = "build_tree"
 required-features = ["tree-builder"]
 
+[[example]]
+name = "custom_emitter"
+
+[[example]]
+name = "callback_emitter"
+
 [[example]]
 name = "scraper"
 required-features = ["tree-builder"]
diff --git a/README.md b/README.md
index 4890712..9c7562e 100644
--- a/README.md
+++ b/README.md
@@ -30,6 +30,13 @@ for token in Tokenizer::new(html).infallible() {
 assert_eq!(new_html, "<title>hello world</title>");
 ```
 
+`html5gum` provides multiple kinds of APIs:
+
+* Iterating over tokens as shown above.
+* Implementing your own `Emitter` for maximum performance, see [the `custom_emitter.rs` example](examples/custom_emitter.rs).
+* A callbacks-based API for a middleground between convenience and performance, see [the `callback_emitter.rs` example](examples/callback_emitter.rs).
+* With the `tree-builder` feature, html5gum can be integrated with `html5ever` and `scraper`. See [the `scraper.rs` example](examples/scraper.rs).
+
 ## What a tokenizer does and what it does not do
 
 `html5gum` fully implements [13.2.5 of the WHATWG HTML
@@ -42,9 +49,6 @@ test suite](https://github.com/html5lib/html5lib-tests/tree/master/tokenizer). S
   gracefully from invalid UTF-8.
 * `html5gum` **does not** [correct mis-nested
   tags.](https://html.spec.whatwg.org/#an-introduction-to-error-handling-and-strange-cases-in-the-parser)
-* `html5gum` **does not** recognize implicitly self-closing elements like
-  `<img>`, as a tokenizer it will simply emit a start token. It does however
-  emit a self-closing tag for `<img .. />`.
 * `html5gum` doesn't implement the DOM, and unfortunately in the HTML spec,
   constructing the DOM ("tree construction") influences how tokenization is
   done. For an example of which problems this causes see [this example
@@ -54,23 +58,9 @@ test suite](https://github.com/html5lib/html5lib-tests/tree/master/tokenizer). S
   21](https://github.com/untitaker/html5gum/issues/21).
 
 With those caveats in mind, `html5gum` can pretty much ~parse~ _tokenize_
-anything that browsers can.
-
-## The `Emitter` trait
-
-A distinguishing feature of `html5gum` is that you can bring your own token
-datastructure and hook into token creation by implementing the `Emitter` trait.
-This allows you to:
-
-* Rewrite all per-HTML-tag allocations to use a custom allocator or datastructure.
-
-* Efficiently filter out uninteresting categories data without ever allocating
-  for it. For example if any plaintext between tokens is not of interest to
-  you, you can implement the respective trait methods as noop and therefore
-  avoid any overhead creating plaintext tokens.
-
-See [the `custom_emitter` example][examples/custom_emitter.rs] for how this
-looks like in practice.
+anything that browsers can. However, using the experimental `tree-builder`
+feature, html5gum can be integrated with `html5ever` and `scraper`. See [the
+`scraper.rs` example](examples/scraper.rs).
 
 ## Other features
 
@@ -116,3 +106,5 @@ Licensed under the MIT license, see [`./LICENSE`][LICENSE].
 [LICENSE]: ./LICENSE
 [examples/tokenize_with_state_switches.rs]: ./examples/tokenize_with_state_switches.rs
 [examples/custom_emitter.rs]: ./examples/custom_emitter.rs
+[examples/callback_emitter.rs]: ./examples/callback_emitter.rs
+[examples/scraper.rs]: ./examples/scraper.rs
diff --git a/examples/build_tree.rs b/examples/build_tree.rs
index 4a567a2..97ca5b7 100644
--- a/examples/build_tree.rs
+++ b/examples/build_tree.rs
@@ -2,8 +2,6 @@
 /// building logic and DOM implementation. The result is a technically complete HTML5 parser.
 ///
 /// You may want to refer to `examples/scraper.rs` for better ergonomics.
-use std::iter::repeat;
-
 use html5ever::tree_builder::TreeBuilder;
 use html5gum::{Html5everEmitter, IoReader, Tokenizer};
 use markup5ever_rcdom::{Handle, NodeData, RcDom};
@@ -11,7 +9,7 @@ use markup5ever_rcdom::{Handle, NodeData, RcDom};
 fn walk(indent: usize, handle: &Handle) {
     let node = handle;
     // FIXME: don't allocate
-    print!("{}", repeat(" ").take(indent).collect::<String>());
+    print!("{}", " ".repeat(indent));
     match node.data {
         NodeData::Document => println!("#Document"),
 
diff --git a/examples/callback_emitter.rs b/examples/callback_emitter.rs
new file mode 100644
index 0000000..de392d6
--- /dev/null
+++ b/examples/callback_emitter.rs
@@ -0,0 +1,53 @@
+//! A slightly simpler, but less performant version of the link extractor that can be found in
+//! `examples/custom_emitter.rs`.
+//!
+//! ```text
+//! printf '<h1>Hello world!</h1><a href="foo">bar</a>' | cargo run --example=custom_emitter
+//! ```
+//!
+//! Output:
+//!
+//! ```text
+//! link: foo
+//! ```
+use html5gum::callbacks::{CallbackEmitter, CallbackEvent};
+use html5gum::{Emitter, IoReader, Tokenizer};
+
+fn get_emitter() -> impl Emitter<Token = String> {
+    let mut is_anchor_tag = false;
+    let mut is_href_attr = false;
+
+    CallbackEmitter::new(move |event: CallbackEvent<'_>| match event {
+        CallbackEvent::OpenStartTag { name } => {
+            is_anchor_tag = name == b"a";
+            is_href_attr = false;
+            None
+        }
+        CallbackEvent::AttributeName { name } => {
+            is_href_attr = name == b"href";
+            None
+        }
+        CallbackEvent::AttributeValue { value } if is_anchor_tag && is_href_attr => {
+            Some(String::from_utf8_lossy(value).into_owned())
+        }
+        _ => None,
+    })
+}
+
+fn main() {
+    for token in
+        Tokenizer::new_with_emitter(IoReader::new(std::io::stdin().lock()), get_emitter()).flatten()
+    {
+        println!("link: {}", token);
+    }
+}
+
+#[test]
+fn basic() {
+    let tokens: Vec<_> =
+        Tokenizer::new_with_emitter("<h1>Hello world</h1><a href=foo>bar</a>", get_emitter())
+            .flatten()
+            .collect();
+
+    assert_eq!(tokens, vec!["foo".to_owned()]);
+}
diff --git a/examples/scraper.rs b/examples/scraper.rs
index d6b476f..fd3a04d 100644
--- a/examples/scraper.rs
+++ b/examples/scraper.rs
@@ -6,7 +6,9 @@
 /// echo '<h1><span class=hello>Hello</span></h1>' | cargo run --all-features --example scraper
 /// ```
 ///
-/// Essentially, your HTML parsing will be powered by a combination of html5gum and html5ever.
+/// Essentially, your HTML parsing will be powered by a combination of html5gum and html5ever. This
+/// has no immediate benefit over using scraper normally and is mostly done as a transitionary step
+/// until html5gum has its own implementation of tree building and the DOM.
 ///
 /// Requires the tree-builder feature.
 use std::io::{stdin, Read};
diff --git a/fuzz/Cargo.lock b/fuzz/Cargo.lock
index 8bb0df3..b4743d2 100644
--- a/fuzz/Cargo.lock
+++ b/fuzz/Cargo.lock
@@ -14,12 +14,11 @@ dependencies = [
 
 [[package]]
 name = "afl"
-version = "0.11.1"
+version = "0.15.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6c0f4180b6d7095a1c3130f1aadead4ece7fb3094dc724c701adb30a92a95228"
+checksum = "80bb240a3b9ff18002142c1a736e98046461d51a694d687c3e7329b456ab0fe4"
 dependencies = [
- "cc",
- "clap",
+ "home",
  "libc",
  "rustc_version",
  "xdg",
@@ -27,75 +26,54 @@ dependencies = [
 
 [[package]]
 name = "ahash"
-version = "0.8.3"
+version = "0.8.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2c99f64d1e06488f620f932677e24bc6e2897582980441ae90a671415bd7ec2f"
+checksum = "e89da841a80418a9b391ebaea17f5c112ffaaa96f621d2c285b5174da76b9011"
 dependencies = [
  "cfg-if",
- "getrandom 0.2.10",
  "once_cell",
  "version_check",
+ "zerocopy",
 ]
 
 [[package]]
 name = "aho-corasick"
-version = "1.0.2"
+version = "1.1.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "43f6cb1bf222025340178f382c426f13757b2960e89779dfcb319c32542a5a41"
+checksum = "8e60d3430d3a69478ad0993f19238d2df97c507009a52b3c10addcd7f6bcb916"
 dependencies = [
  "memchr",
 ]
 
-[[package]]
-name = "ansi_term"
-version = "0.12.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d52a9bb7ec0cf484c551830a7ce27bd20d67eac647e1befb56b0be4ee39a55d2"
-dependencies = [
- "winapi",
-]
-
 [[package]]
 name = "arbitrary"
-version = "1.3.0"
+version = "1.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2d098ff73c1ca148721f37baad5ea6a465a13f9573aba8641fbbbae8164a54e"
+checksum = "7d5a26814d8dcb93b0e5a0ff3c6d80a8843bafb21b39e8e18a6f05471870e110"
 
 [[package]]
 name = "ast_node"
-version = "0.9.5"
+version = "0.9.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c09c69dffe06d222d072c878c3afe86eee2179806f20503faec97250268b4c24"
+checksum = "f9184f2b369b3e8625712493c89b785881f27eedc6cde480a81883cef78868b2"
 dependencies = [
- "pmutil",
  "proc-macro2",
  "quote",
  "swc_macros_common",
- "syn 2.0.26",
-]
-
-[[package]]
-name = "atty"
-version = "0.2.14"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d9b39be18770d11421cdb1b9947a45dd3f37e93092cbf377614828a319d5fee8"
-dependencies = [
- "hermit-abi",
- "libc",
- "winapi",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "autocfg"
-version = "1.1.0"
+version = "1.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
+checksum = "ace50bade8e6234aa140d9a2f552bbee1db4d353f69b8217bc503490fc1a9f26"
 
 [[package]]
 name = "better_scoped_tls"
-version = "0.1.1"
+version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "794edcc9b3fb07bb4aecaa11f093fd45663b4feadb782d68303a2268bc2701de"
+checksum = "297b153aa5e573b5863108a6ddc9d5c968bd0b20e75cc614ee9821d2f45679c7"
 dependencies = [
  "scoped-tls",
 ]
@@ -108,29 +86,31 @@ checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
 
 [[package]]
 name = "bitflags"
-version = "2.3.3"
+version = "2.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "630be753d4e58660abd17930c71b647fe46c27ea6b63cc59e1e3851406972e42"
+checksum = "b048fb63fd8b5923fc5aa7b340d8e156aec7ec02f0c78fa8a6ddc2613f6f71de"
 
 [[package]]
 name = "byteorder"
-version = "1.4.3"
+version = "1.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "14c189c53d098945499cdfa7ecc63567cf3886b3332b312a5b4585d8d3a6a610"
+checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
 
 [[package]]
 name = "bytes"
-version = "1.4.0"
+version = "1.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "89b2fd2a0dcf38d7971e2194b6b6eebab45ae01067456a7fd93d5547a61b70be"
+checksum = "9ac0150caa2ae65ca5bd83f25c7de183dea78d4d366469f148435e2acfbad0da"
 
 [[package]]
 name = "cc"
-version = "1.0.79"
+version = "1.1.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "50d30906286121d95be3d479533b458f87493b30a4b5f79a607db8f5d11aa91f"
+checksum = "c2e7962b54006dcfcc61cb72735f4d89bb97061dd6a7ed882ec6b8ee53714c6f"
 dependencies = [
  "jobserver",
+ "libc",
+ "shlex",
 ]
 
 [[package]]
@@ -139,21 +119,6 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
 
-[[package]]
-name = "clap"
-version = "2.34.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a0610544180c38b88101fecf2dd634b174a62eef6946f84dfc6a7127512b381c"
-dependencies = [
- "ansi_term",
- "atty",
- "bitflags 1.3.2",
- "strsim",
- "textwrap",
- "unicode-width",
- "vec_map",
-]
-
 [[package]]
 name = "convert_case"
 version = "0.4.0"
@@ -184,20 +149,20 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "13b588ba4ac1a99f7f2964d24b3d896ddc6bf847ee3855dbd4366f058cfcd331"
 dependencies = [
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "derive_more"
-version = "0.99.17"
+version = "0.99.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4fb810d30a7c1953f91334de7244731fc3f3c10d7fe163338a35b9f640960321"
+checksum = "5f33878137e4dafd7fa914ad4e259e18a4e8e532b9617a2d0150262bf53abfce"
 dependencies = [
  "convert_case",
  "proc-macro2",
  "quote",
  "rustc_version",
- "syn 1.0.109",
+ "syn 2.0.85",
 ]
 
 [[package]]
@@ -214,47 +179,46 @@ checksum = "dcbb2bf8e87535c23f7a8a321e364ce21462d0ff10cb6407820e8e96dfff6653"
 
 [[package]]
 name = "dtoa-short"
-version = "0.3.4"
+version = "0.3.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dbaceec3c6e4211c79e7b1800fb9680527106beb2f9c51904a3210c03a448c74"
+checksum = "cd1511a7b6a56299bd043a9c167a6d2bfb37bf84a6dfceaba651168adfb43c87"
 dependencies = [
  "dtoa",
 ]
 
 [[package]]
 name = "either"
-version = "1.9.0"
+version = "1.13.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a26ae43d7bcc3b814de94796a5e736d4029efb0ee900c12e2d54c993ad1a1e07"
+checksum = "60b1af1c220855b6ceac025d3f6ecdd2b7c4894bfe9cd9bda4fbb4bc7c0d4cf0"
 
 [[package]]
 name = "encoding_rs"
-version = "0.8.32"
+version = "0.8.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "071a31f4ee85403370b58aca746f01041ede6f0da2730960ad001edc2b71b394"
+checksum = "75030f3c4f45dafd7586dd6780965a8c7e8e285a5ecb86713e63a79c5b2766f3"
 dependencies = [
  "cfg-if",
 ]
 
 [[package]]
 name = "form_urlencoded"
-version = "1.2.0"
+version = "1.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a62bc1cf6f830c2ec14a513a9fb124d0a213a629668a4186f329db21fe045652"
+checksum = "e13624c2627564efccf4934284bdd98cbaa14e79b0b5a141218e507b3a823456"
 dependencies = [
  "percent-encoding",
 ]
 
 [[package]]
 name = "from_variant"
-version = "0.1.6"
+version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "03ec5dc38ee19078d84a692b1c41181ff9f94331c76cee66ff0208c770b5e54f"
+checksum = "32016f1242eb82af5474752d00fd8ebcd9004bd69b462b1c91de833972d08ed4"
 dependencies = [
- "pmutil",
  "proc-macro2",
  "swc_macros_common",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
@@ -289,9 +253,9 @@ dependencies = [
 
 [[package]]
 name = "getrandom"
-version = "0.2.10"
+version = "0.2.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "be4136b2a15dd319360be1c07d9933517ccf0be8f16bf62a3bee4f0d618df427"
+checksum = "c4567c8db10ae91089c99af84c68c38da3ec2f087c3f82960bcdbf3656b6f4d7"
 dependencies = [
  "cfg-if",
  "libc",
@@ -307,49 +271,41 @@ dependencies = [
  "ahash",
 ]
 
-[[package]]
-name = "hermit-abi"
-version = "0.1.19"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "62b467343b94ba476dcb2500d242dadbb39557df889310ac77c5d99100aaac33"
-dependencies = [
- "libc",
-]
-
 [[package]]
 name = "home"
-version = "0.5.5"
+version = "0.5.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5444c27eef6923071f7ebcc33e3444508466a76f7a2b93da00ed6e19f30c1ddb"
+checksum = "e3d1354bf6b7235cb4a0576c2619fd4ed18183f689b12b006a0ee7329eeff9a5"
 dependencies = [
  "windows-sys",
 ]
 
 [[package]]
 name = "html5ever"
-version = "0.26.0"
-source = "git+https://github.com/untitaker/html5ever?branch=html5lib-tests-update#2a5a62647c079e39a3d9b1b5029a069429db049b"
+version = "0.27.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c13771afe0e6e846f1e67d038d4cb29998a6779f93c809212e4e9c32efd244d4"
 dependencies = [
  "log",
  "mac",
  "markup5ever",
  "proc-macro2",
  "quote",
- "syn 1.0.109",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "html5gum"
-version = "0.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2dad48b66db55322add2819ae1d7bda0c32f3415269a08330679dbc8b0afeb30"
+version = "0.6.0"
 dependencies = [
  "jetscii",
 ]
 
 [[package]]
 name = "html5gum"
-version = "0.5.7"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5d4c69595eca3d5d246be4c0be51577b0c39c5f8c75d27a06c8a69c7b594cd37"
 dependencies = [
  "jetscii",
 ]
@@ -362,8 +318,8 @@ dependencies = [
  "bytes",
  "encoding_rs",
  "html5ever",
- "html5gum 0.4.0",
- "html5gum 0.5.7",
+ "html5gum 0.6.0",
+ "html5gum 0.6.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "libfuzzer-sys",
  "lol_html",
  "pretty_assertions",
@@ -374,9 +330,9 @@ dependencies = [
 
 [[package]]
 name = "idna"
-version = "0.4.0"
+version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7d20d6b07bfbc108882d88ed8e37d39636dcc260e15e30c45e6ba089610b917c"
+checksum = "634d9b1461af396cad843f47fdba5597a4f9e6ddd4bfb6ff5d85028c25cb12f6"
 dependencies = [
  "unicode-bidi",
  "unicode-normalization",
@@ -384,15 +340,14 @@ dependencies = [
 
 [[package]]
 name = "is-macro"
-version = "0.3.0"
+version = "0.3.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f4467ed1321b310c2625c5aa6c1b1ffc5de4d9e42668cf697a08fb033ee8265e"
+checksum = "2069faacbe981460232f880d26bf3c7634e322d49053aa48c27e3ae642f728f1"
 dependencies = [
  "Inflector",
- "pmutil",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
@@ -403,9 +358,9 @@ checksum = "b71991ff56294aa922b450139ee08b3bfc70982c6b2c7562771375cf73542dd4"
 
 [[package]]
 name = "itoa"
-version = "1.0.9"
+version = "1.0.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "af150ab688ff2122fcef229be89cb50dd66af9e01a4ff320cc137eecc9bacc38"
+checksum = "49f1f14873335454500d59611f1cf4a4b0f786f9ac11f4312a78e4cf2566695b"
 
 [[package]]
 name = "jetscii"
@@ -415,18 +370,18 @@ checksum = "47f142fe24a9c9944451e8349de0a56af5f3e7226dc46f3ed4d4ecc0b85af75e"
 
 [[package]]
 name = "jobserver"
-version = "0.1.26"
+version = "0.1.32"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "936cfd212a0155903bcbc060e316fb6cc7cbf2e1907329391ebadc1fe0ce77c2"
+checksum = "48d1dbcbbeb6a7fec7e059840aa538bd62aaccf972c7346c4d9d2059312853d0"
 dependencies = [
  "libc",
 ]
 
 [[package]]
 name = "lazy_static"
-version = "1.4.0"
+version = "1.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
+checksum = "bbd2bcb4c963f2ddae06a2efc7e9f3591312473c50c6685e1f298068316e66fe"
 
 [[package]]
 name = "lazycell"
@@ -436,15 +391,15 @@ checksum = "830d08ce1d1d941e6b30645f1a0eb5643013d835ce3779a5fc208261dbe10f55"
 
 [[package]]
 name = "libc"
-version = "0.2.147"
+version = "0.2.161"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b4668fb0ea861c1df094127ac5f1da3409a82116a4ba74fca2e58ef927159bb3"
+checksum = "8e9489c2807c139ffd9c1794f4af0ebe86a828db53ecdc7fea2111d0fed085d1"
 
 [[package]]
 name = "libfuzzer-sys"
-version = "0.4.6"
+version = "0.4.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "beb09950ae85a0a94b27676cccf37da5ff13f27076aa1adbc6545dd0d0e1bd4e"
+checksum = "a96cfd5557eb82f2b83fed4955246c988d331975a002961b07c81584d107e7f7"
 dependencies = [
  "arbitrary",
  "cc",
@@ -453,9 +408,9 @@ dependencies = [
 
 [[package]]
 name = "lock_api"
-version = "0.4.10"
+version = "0.4.12"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c1cc9717a20b1bb222f333e6a92fd32f7d8a18ddc5a3191a11af45dcbf4dcd16"
+checksum = "07af8b9cdd281b7915f413fa73f29ebd5d55d0d3f0155584dade1ff18cea1b17"
 dependencies = [
  "autocfg",
  "scopeguard",
@@ -463,9 +418,9 @@ dependencies = [
 
 [[package]]
 name = "log"
-version = "0.4.19"
+version = "0.4.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b06a4cde4c0f271a446782e3eff8de789548ce57dbc8eca9292c27f4a42004b4"
+checksum = "a7a70ba024b9dc04c27ea2f0c0548feb474ec5c54bba33a7f72f873a39d07b24"
 
 [[package]]
 name = "lol_html"
@@ -473,7 +428,7 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1610d7994d67a05bb35861cd733b069b1171de8693bc8452849c59361a1bb87b"
 dependencies = [
- "bitflags 2.3.3",
+ "bitflags 2.6.0",
  "cfg-if",
  "cssparser",
  "encoding_rs",
@@ -495,8 +450,9 @@ checksum = "c41e0c4fef86961ac6d6f8a82609f55f31b05e4fce149ac5710e439df7619ba4"
 
 [[package]]
 name = "markup5ever"
-version = "0.11.0"
-source = "git+https://github.com/untitaker/html5ever?branch=html5lib-tests-update#2a5a62647c079e39a3d9b1b5029a069429db049b"
+version = "0.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "16ce3abbeba692c8b8441d036ef91aea6df8da2c6b6e21c7e14d3c18e526be45"
 dependencies = [
  "log",
  "phf 0.11.2",
@@ -514,9 +470,9 @@ checksum = "2532096657941c2fea9c289d370a250971c689d4f143798ff67113ec042024a5"
 
 [[package]]
 name = "memchr"
-version = "2.5.0"
+version = "2.7.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2dffe52ecf27772e601905b7522cb4ef790d2cc203488bbd0e2fe85fcb74566d"
+checksum = "78ca9ab1a0babb1e7d5695e3530886289c18cf2f87ec19a575a0abdce112e3a3"
 
 [[package]]
 name = "mime"
@@ -526,9 +482,9 @@ checksum = "6877bb514081ee2a7ff5ef9de3281f14a4dd4bceac4c09388074a6b5df8a139a"
 
 [[package]]
 name = "new_debug_unreachable"
-version = "1.0.4"
+version = "1.0.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e4a24736216ec316047a1fc4252e27dabb04218aa4a3f37c6e7ddbf1f9782b54"
+checksum = "650eef8c711430f1a879fdd01d4745a7deea475becfb90269c06775983bbf086"
 
 [[package]]
 name = "nodrop"
@@ -538,45 +494,43 @@ checksum = "72ef4a56884ca558e5ddb05a1d1e7e1bfd9a68d9ed024c21704cc98872dae1bb"
 
 [[package]]
 name = "num-bigint"
-version = "0.4.3"
+version = "0.4.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f93ab6289c7b344a8a9f60f88d80aa20032336fe78da341afc91c8a2341fc75f"
+checksum = "a5e44f723f1133c9deac646763579fdb3ac745e418f2a7af9cd0c431da1f20b9"
 dependencies = [
- "autocfg",
  "num-integer",
  "num-traits",
 ]
 
 [[package]]
 name = "num-integer"
-version = "0.1.45"
+version = "0.1.46"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "225d3389fb3509a24c93f5c29eb6bde2586b98d9f016636dff58d7c6f7569cd9"
+checksum = "7969661fd2958a5cb096e56c8e1ad0444ac2bbcd0061bd28660485a44879858f"
 dependencies = [
- "autocfg",
  "num-traits",
 ]
 
 [[package]]
 name = "num-traits"
-version = "0.2.16"
+version = "0.2.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f30b0abd723be7e2ffca1272140fac1a2f084c77ec3e123c192b66af1ee9e6c2"
+checksum = "071dfc062690e90b734c0b2273ce72ad0ffa95f0c74596bc250dcfd960262841"
 dependencies = [
  "autocfg",
 ]
 
 [[package]]
 name = "once_cell"
-version = "1.18.0"
+version = "1.20.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dd8b5dd2ae5ed71462c540258bedcb51965123ad7e7ccf4b9a8cafaa4a63576d"
+checksum = "1261fe7e33c73b354eab43b1273a57c8f967d0391e80353e51f764ac02cf6775"
 
 [[package]]
 name = "parking_lot"
-version = "0.12.1"
+version = "0.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3742b2c103b9f06bc9fff0a37ff4912935851bee6d36f3c02bcc755bcfec228f"
+checksum = "f1bf18183cf54e8d6059647fc3063646a1801cf30896933ec2311622cc4b9a27"
 dependencies = [
  "lock_api",
  "parking_lot_core",
@@ -584,9 +538,9 @@ dependencies = [
 
 [[package]]
 name = "parking_lot_core"
-version = "0.9.8"
+version = "0.9.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "93f00c865fe7cabf650081affecd3871070f26767e7b2070a3ffae14c654b447"
+checksum = "1e401f977ab385c9e4e3ab30627d6f26d00e2c73eef317493c4ec6d468726cf8"
 dependencies = [
  "cfg-if",
  "libc",
@@ -597,9 +551,9 @@ dependencies = [
 
 [[package]]
 name = "percent-encoding"
-version = "2.3.0"
+version = "2.3.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9b2a4787296e9989611394c33f193f676704af1686e70b8f8033ab5ba9a35a94"
+checksum = "e3148f5046208a5d56bcfc03053e3ca6334e51da8dfb19b6cdc8b306fae3283e"
 
 [[package]]
 name = "phf"
@@ -714,27 +668,19 @@ dependencies = [
 
 [[package]]
 name = "pin-project-lite"
-version = "0.2.10"
+version = "0.2.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4c40d25201921e5ff0c862a505c6557ea88568a4e3ace775ab55e93f2f4f9d57"
+checksum = "915a1e146535de9163f3987b8944ed8cf49a18bb0056bcebcdcece385cece4ff"
 
 [[package]]
-name = "pmutil"
-version = "0.6.1"
+name = "ppv-lite86"
+version = "0.2.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "52a40bc70c2c58040d2d8b167ba9a5ff59fc9dab7ad44771cfde3dcfde7a09c6"
+checksum = "77957b295656769bb8ad2b6a6b09d897d94f05c41b069aede1fcdaa675eaea04"
 dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.26",
+ "zerocopy",
 ]
 
-[[package]]
-name = "ppv-lite86"
-version = "0.2.17"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
-
 [[package]]
 name = "precomputed-hash"
 version = "0.1.1"
@@ -743,9 +689,9 @@ checksum = "925383efa346730478fb4838dbe9137d2a47675ad789c546d150a6e1dd4ab31c"
 
 [[package]]
 name = "pretty_assertions"
-version = "1.4.0"
+version = "1.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "af7cee1a6c8a5b9208b3cb1061f10c0cb689087b3d8ce85fb9d2dd7a29b6ba66"
+checksum = "3ae130e2f271fbc2ac3a40fb1d07180839cdbbe443c7a27e1e3c13c5cac0116d"
 dependencies = [
  "diff",
  "yansi",
@@ -759,18 +705,18 @@ checksum = "dc375e1527247fe1a97d8b7156678dfe7c1af2fc075c9a4db3690ecd2a148068"
 
 [[package]]
 name = "proc-macro2"
-version = "1.0.65"
+version = "1.0.89"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "92de25114670a878b1261c79c9f8f729fb97e95bac93f6312f583c60dd6a1dfe"
+checksum = "f139b0662de085916d1fb67d2b4169d1addddda1919e696f3252b740b629986e"
 dependencies = [
  "unicode-ident",
 ]
 
 [[package]]
 name = "quote"
-version = "1.0.30"
+version = "1.0.37"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5907a1b7c277254a8b15170f6e7c97cfa60ee7872a3217663bb81151e48184bb"
+checksum = "b5b9d34b8991d19d98081b46eacdd8eb58c6f2b201139f7c5f643cc155a633af"
 dependencies = [
  "proc-macro2",
 ]
@@ -835,7 +781,7 @@ version = "0.6.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ec0be4795e2f6a28069bec0b5ff3e2ac9bafc99e6a9a7dc3547996c5c816922c"
 dependencies = [
- "getrandom 0.2.10",
+ "getrandom 0.2.15",
 ]
 
 [[package]]
@@ -858,18 +804,18 @@ dependencies = [
 
 [[package]]
 name = "redox_syscall"
-version = "0.3.5"
+version = "0.5.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "567664f262709473930a4bf9e51bf2ebf3348f2e748ccc50dea20646858f8f29"
+checksum = "9b6dfecf2c74bce2466cabf93f6664d6998a69eb21e39f4207930065b27b771f"
 dependencies = [
- "bitflags 1.3.2",
+ "bitflags 2.6.0",
 ]
 
 [[package]]
 name = "regex"
-version = "1.9.1"
+version = "1.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b2eae68fc220f7cf2532e4494aded17545fce192d59cd996e0fe7887f4ceb575"
+checksum = "b544ef1b4eac5dc2db33ea63606ae9ffcfac26c1416a2806ae0bf5f56b201191"
 dependencies = [
  "aho-corasick",
  "memchr",
@@ -879,9 +825,9 @@ dependencies = [
 
 [[package]]
 name = "regex-automata"
-version = "0.3.3"
+version = "0.4.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "39354c10dd07468c2e73926b23bb9c2caca74c5501e38a35da70406f1d923310"
+checksum = "368758f23274712b504848e9d5a6f010445cc8b87a7cdb4d7cbee666c1288da3"
 dependencies = [
  "aho-corasick",
  "memchr",
@@ -890,9 +836,9 @@ dependencies = [
 
 [[package]]
 name = "regex-syntax"
-version = "0.7.4"
+version = "0.8.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e5ea92a5b6195c6ef2a0295ea818b312502c6fc94dde986c5553242e18fd4ce2"
+checksum = "2b15c43186be67a4fd63bee50d0303afffcef381492ebe2c5d87f324e1b8815c"
 
 [[package]]
 name = "rustc-hash"
@@ -902,18 +848,18 @@ checksum = "08d43f7aa6b08d49f382cde6a7982047c3426db949b1424bc4b7ec9ae12c6ce2"
 
 [[package]]
 name = "rustc_version"
-version = "0.4.0"
+version = "0.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bfa0f585226d2e68097d4f95d113b15b83a82e819ab25717ec0590d9584ef366"
+checksum = "cfcb3a22ef46e85b45de6ee7e79d063319ebb6594faafcf1c225ea92ab6e9b92"
 dependencies = [
  "semver",
 ]
 
 [[package]]
 name = "ryu"
-version = "1.0.15"
+version = "1.0.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1ad4cc8da4ef723ed60bced201181d83791ad433213d8c24efffda1eec85d741"
+checksum = "f3cb5ba0dc43242ce17de99c180e96db90b235b8a9fdc9543c96d2209116bd9f"
 
 [[package]]
 name = "safemem"
@@ -929,9 +875,9 @@ checksum = "e1cf6437eb19a8f4a6cc0f7dca544973b0b78843adbfeb3683d1a94a0024a294"
 
 [[package]]
 name = "scopeguard"
-version = "1.1.0"
+version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd"
+checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
 
 [[package]]
 name = "selectors"
@@ -955,37 +901,38 @@ dependencies = [
 
 [[package]]
 name = "semver"
-version = "1.0.18"
+version = "1.0.23"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b0293b4b29daaf487284529cc2f5675b8e57c61f70167ba415a463651fd6a918"
+checksum = "61697e0a1c7e512e84a621326239844a24d8207b4669b41bc18b32ea5cbf988b"
 
 [[package]]
 name = "serde"
-version = "1.0.171"
+version = "1.0.214"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "30e27d1e4fd7659406c492fd6cfaf2066ba8773de45ca75e855590f856dc34a9"
+checksum = "f55c3193aca71c12ad7890f1785d2b73e1b9f63a0bbc353c08ef26fe03fc56b5"
 dependencies = [
  "serde_derive",
 ]
 
 [[package]]
 name = "serde_derive"
-version = "1.0.171"
+version = "1.0.214"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "389894603bd18c46fa56231694f8d827779c0951a667087194cf9de94ed24682"
+checksum = "de523f781f095e28fa605cdce0f8307e451cc0fd14e2eb4cd2e98a355b147766"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "serde_json"
-version = "1.0.103"
+version = "1.0.132"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d03b412469450d4404fe8499a268edd7f8b79fecb074b0d812ad64ca21f4031b"
+checksum = "d726bfaff4b320266d395898905d0eba0345aae23b54aee3a737e260fd46db03"
 dependencies = [
- "itoa 1.0.9",
+ "itoa 1.0.11",
+ "memchr",
  "ryu",
  "serde",
 ]
@@ -1000,17 +947,23 @@ dependencies = [
  "stable_deref_trait",
 ]
 
+[[package]]
+name = "shlex"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0fda2ff0d084019ba4d7c6f371c95d8fd75ce3524c3cb8fb653a3023f6323e64"
+
 [[package]]
 name = "siphasher"
-version = "0.3.10"
+version = "0.3.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7bd3e3206899af3f8b12af284fafc038cc1dc2b41d1b89dd17297221c5d225de"
+checksum = "38b58827f4464d87d377d175e90bf58eb00fd8716ff0a62f80356b5e61555d0d"
 
 [[package]]
 name = "smallvec"
-version = "1.11.0"
+version = "1.13.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "62bb4feee49fdd9f707ef802e22365a35de4b7b299de4763d44bfea899442ff9"
+checksum = "3c5e1a9a646d36c3599cd173a41282daf47c44583ad367b8e6837255952e5c67"
 
 [[package]]
 name = "stable_deref_trait"
@@ -1046,28 +999,21 @@ dependencies = [
 
 [[package]]
 name = "string_enum"
-version = "0.4.1"
+version = "0.4.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8fa4d4f81d7c05b9161f8de839975d3326328b8ba2831164b465524cc2f55252"
+checksum = "05e383308aebc257e7d7920224fa055c632478d92744eca77f99be8fa1545b90"
 dependencies = [
- "pmutil",
  "proc-macro2",
  "quote",
  "swc_macros_common",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
-[[package]]
-name = "strsim"
-version = "0.8.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ea5119cdb4c55b55d432abb513a0429384878c15dde60cc77b1c99de1a95a6a"
-
 [[package]]
 name = "swc_atoms"
-version = "0.5.8"
+version = "0.5.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b8066e17abb484602da673e2d35138ab32ce53f26368d9c92113510e1659220b"
+checksum = "9f54563d7dcba626d4acfe14ed12def7ecc28e004debe3ecd2c3ee07cc47e449"
 dependencies = [
  "once_cell",
  "rustc-hash",
@@ -1079,11 +1025,10 @@ dependencies = [
 
 [[package]]
 name = "swc_common"
-version = "0.31.18"
+version = "0.31.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e30cd01afa791b15263fcfe8f77ecbbd020ddef659f0f58d3c7b794ad65c1738"
+checksum = "88d00f960c667c59c133f30492f4d07f26242fcf988a066d3871e6d3d838d528"
 dependencies = [
- "ahash",
  "ast_node",
  "better_scoped_tls",
  "cfg-if",
@@ -1106,21 +1051,20 @@ dependencies = [
 
 [[package]]
 name = "swc_eq_ignore_macros"
-version = "0.1.2"
+version = "0.1.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "05a95d367e228d52484c53336991fdcf47b6b553ef835d9159db4ba40efb0ee8"
+checksum = "63db0adcff29d220c3d151c5b25c0eabe7e32dd936212b84cdaa1392e3130497"
 dependencies = [
- "pmutil",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "swc_html_ast"
-version = "0.31.18"
+version = "0.31.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0edeedae12497a0f6e5bb97d77c4bc459495fc644a1a71f34d24b8cb988a14bc"
+checksum = "bb2c7b319ef5ff810dffb876f772ebb7d6aac934454998c00cd41fd525e3c8ed"
 dependencies = [
  "is-macro",
  "string_enum",
@@ -1130,9 +1074,9 @@ dependencies = [
 
 [[package]]
 name = "swc_html_parser"
-version = "0.37.21"
+version = "0.37.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d449b38d99056aad88f6a7743ce65ccb0c4ab8863089b2a781262d72acb1e3dd"
+checksum = "89b0e2371bc52709de74dd73f2aad2a49f9f5fbf1030351d7d5815e849a91caf"
 dependencies = [
  "swc_atoms",
  "swc_common",
@@ -1142,9 +1086,9 @@ dependencies = [
 
 [[package]]
 name = "swc_html_utils"
-version = "0.16.18"
+version = "0.16.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4b703fedf29d2347ab77fe98bbf506e4a7f5a34462c7de8a966fe6ff00c79c78"
+checksum = "1ac1904e208373342021df5c0ac501fdf825a93ca06cce2854f5921f1279e430"
 dependencies = [
  "once_cell",
  "serde",
@@ -1155,21 +1099,20 @@ dependencies = [
 
 [[package]]
 name = "swc_macros_common"
-version = "0.3.8"
+version = "0.3.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7a273205ccb09b51fabe88c49f3b34c5a4631c4c00a16ae20e03111d6a42e832"
+checksum = "27e18fbfe83811ffae2bb23727e45829a0d19c6870bced7c0f545cc99ad248dd"
 dependencies = [
- "pmutil",
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "swc_visit"
-version = "0.5.7"
+version = "0.5.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e87c337fbb2d191bf371173dea6a957f01899adb8f189c6c31b122a6cfc98fc3"
+checksum = "043d11fe683dcb934583ead49405c0896a5af5face522e4682c16971ef7871b9"
 dependencies = [
  "either",
  "swc_visit_macros",
@@ -1177,16 +1120,15 @@ dependencies = [
 
 [[package]]
 name = "swc_visit_macros"
-version = "0.5.8"
+version = "0.5.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0f322730fb82f3930a450ac24de8c98523af7d34ab8cb2f46bcb405839891a99"
+checksum = "92807d840959f39c60ce8a774a3f83e8193c658068e6d270dbe0a05e40e90b41"
 dependencies = [
  "Inflector",
- "pmutil",
  "proc-macro2",
  "quote",
  "swc_macros_common",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
@@ -1202,9 +1144,9 @@ dependencies = [
 
 [[package]]
 name = "syn"
-version = "2.0.26"
+version = "2.0.85"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "45c3457aacde3c65315de5031ec191ce46604304d2446e803d71ade03308d970"
+checksum = "5023162dfcd14ef8f32034d8bcd4cc5ddc61ef7a247c024a33e24e1f24d21b56"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -1222,15 +1164,6 @@ dependencies = [
  "utf-8",
 ]
 
-[[package]]
-name = "textwrap"
-version = "0.11.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d326610f408c7a4eb6f51c37c330e496b08506c9457c9d34287ecc38809fb060"
-dependencies = [
- "unicode-width",
-]
-
 [[package]]
 name = "thin-slice"
 version = "0.1.1"
@@ -1239,29 +1172,29 @@ checksum = "8eaa81235c7058867fa8c0e7314f33dcce9c215f535d1913822a2b3f5e289f3c"
 
 [[package]]
 name = "thiserror"
-version = "1.0.43"
+version = "1.0.65"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a35fc5b8971143ca348fa6df4f024d4d55264f3468c71ad1c2f365b0a4d58c42"
+checksum = "5d11abd9594d9b38965ef50805c5e469ca9cc6f197f883f717e0269a3057b3d5"
 dependencies = [
  "thiserror-impl",
 ]
 
 [[package]]
 name = "thiserror-impl"
-version = "1.0.43"
+version = "1.0.65"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "463fe12d7993d3b327787537ce8dd4dfa058de32fc2b195ef3cde03dc4771e8f"
+checksum = "ae71770322cbd277e69d762a16c444af02aa0575ac0d174f0b9562d3b37f8602"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "tinyvec"
-version = "1.6.0"
+version = "1.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "87cc5ceb3875bb20c2890005a4e226a4651264a5c75edb2421b52861a0a0cb50"
+checksum = "445e881f4f6d382d5f27c034e25eb92edd7c784ceab92a0937db7f2e9471b938"
 dependencies = [
  "tinyvec_macros",
 ]
@@ -1274,11 +1207,10 @@ checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
 
 [[package]]
 name = "tracing"
-version = "0.1.37"
+version = "0.1.40"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ce8c33a8d48bd45d624a6e523445fd21ec13d3653cd51f681abf67418f54eb8"
+checksum = "c3523ab5a71916ccf420eebdf5521fcef02141234bbc0b8a49f2fdc4544364ef"
 dependencies = [
- "cfg-if",
  "pin-project-lite",
  "tracing-attributes",
  "tracing-core",
@@ -1286,29 +1218,29 @@ dependencies = [
 
 [[package]]
 name = "tracing-attributes"
-version = "0.1.26"
+version = "0.1.27"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5f4f31f56159e98206da9efd823404b79b6ef3143b4a7ab76e67b1751b25a4ab"
+checksum = "34704c8d6ebcbc939824180af020566b01a7c01f80641264eba0999f6c2b6be7"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 2.0.26",
+ "syn 2.0.85",
 ]
 
 [[package]]
 name = "tracing-core"
-version = "0.1.31"
+version = "0.1.32"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0955b8137a1df6f1a2e9a37d8a6656291ff0297c1a97c24e0d8425fe2312f79a"
+checksum = "c06d3da6113f116aaee68e4d601191614c9053067f9ab7f6edbcb161237daa54"
 dependencies = [
  "once_cell",
 ]
 
 [[package]]
 name = "triomphe"
-version = "0.1.9"
+version = "0.1.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0eee8098afad3fb0c54a9007aab6804558410503ad676d4633f9c2559a00ac0f"
+checksum = "ef8f7726da4807b58ea5c96fdc122f80702030edc33b35aff9190a51148ccc85"
 dependencies = [
  "serde",
  "stable_deref_trait",
@@ -1316,36 +1248,36 @@ dependencies = [
 
 [[package]]
 name = "unicode-bidi"
-version = "0.3.13"
+version = "0.3.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "92888ba5573ff080736b3648696b70cafad7d250551175acbaa4e0385b3e1460"
+checksum = "5ab17db44d7388991a428b2ee655ce0c212e862eff1768a455c58f9aad6e7893"
 
 [[package]]
 name = "unicode-ident"
-version = "1.0.11"
+version = "1.0.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "301abaae475aa91687eb82514b328ab47a211a533026cb25fc3e519b86adfc3c"
+checksum = "e91b56cd4cadaeb79bbf1a5645f6b4f8dc5bde8834ad5894a8db35fda9efa1fe"
 
 [[package]]
 name = "unicode-normalization"
-version = "0.1.22"
+version = "0.1.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5c5713f0fc4b5db668a2ac63cdb7bb4469d8c9fed047b1d0292cc7b0ce2ba921"
+checksum = "5033c97c4262335cded6d6fc3e5c18ab755e1a3dc96376350f3d8e9f009ad956"
 dependencies = [
  "tinyvec",
 ]
 
 [[package]]
 name = "unicode-width"
-version = "0.1.10"
+version = "0.1.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c0edd1e5b14653f783770bce4a4dabb4a5108a5370a5f5d8cfe8710c361f6c8b"
+checksum = "7dd6e30e90baa6f72411720665d41d89b9a3d039dc45b8faea1ddd07f617f6af"
 
 [[package]]
 name = "url"
-version = "2.4.0"
+version = "2.5.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "50bff7831e19200a85b17131d085c25d7811bc4e186efdaf54bbd132994a88cb"
+checksum = "22784dbdf76fdde8af1aeda5622b546b422b6fc585325248a2bf9f5e41e94d6c"
 dependencies = [
  "form_urlencoded",
  "idna",
@@ -1358,17 +1290,11 @@ version = "0.7.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "09cc8ee72d2a9becf2f2febe0205bbed8fc6615b7cb429ad062dc7b7ddd036a9"
 
-[[package]]
-name = "vec_map"
-version = "0.8.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1bddf1187be692e79c5ffeab891132dfb0f236ed36a43c7ed39f1165ee20191"
-
 [[package]]
 name = "version_check"
-version = "0.9.4"
+version = "0.9.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "49874b5167b65d7193b8aba1567f5c7d93d001cafc34600cee003eda787e483f"
+checksum = "0b928f33d975fc6ad9f86c8f283853ad26bdd5b10b7f1542aa2fa15e2289105a"
 
 [[package]]
 name = "wasi"
@@ -1382,46 +1308,25 @@ version = "0.11.0+wasi-snapshot-preview1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
 
-[[package]]
-name = "winapi"
-version = "0.3.9"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5c839a674fcd7a98952e593242ea400abe93992746761e38641405d28b00f419"
-dependencies = [
- "winapi-i686-pc-windows-gnu",
- "winapi-x86_64-pc-windows-gnu",
-]
-
-[[package]]
-name = "winapi-i686-pc-windows-gnu"
-version = "0.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
-
-[[package]]
-name = "winapi-x86_64-pc-windows-gnu"
-version = "0.4.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
-
 [[package]]
 name = "windows-sys"
-version = "0.48.0"
+version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "677d2418bec65e3338edb076e806bc1ec15693c5d0104683f2efe857f61056a9"
+checksum = "282be5f36a8ce781fad8c8ae18fa3f9beff57ec1b52cb3de0789201425d9a33d"
 dependencies = [
  "windows-targets",
 ]
 
 [[package]]
 name = "windows-targets"
-version = "0.48.1"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "05d4b17490f70499f20b9e791dcf6a299785ce8af4d709018206dc5b4953e95f"
+checksum = "9b724f72796e036ab90c1021d4780d4d3d648aca59e491e6b98e725b84e99973"
 dependencies = [
  "windows_aarch64_gnullvm",
  "windows_aarch64_msvc",
  "windows_i686_gnu",
+ "windows_i686_gnullvm",
  "windows_i686_msvc",
  "windows_x86_64_gnu",
  "windows_x86_64_gnullvm",
@@ -1430,57 +1335,81 @@ dependencies = [
 
 [[package]]
 name = "windows_aarch64_gnullvm"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "91ae572e1b79dba883e0d315474df7305d12f569b400fcf90581b06062f7e1bc"
+checksum = "32a4622180e7a0ec044bb555404c800bc9fd9ec262ec147edd5989ccd0c02cd3"
 
 [[package]]
 name = "windows_aarch64_msvc"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b2ef27e0d7bdfcfc7b868b317c1d32c641a6fe4629c171b8928c7b08d98d7cf3"
+checksum = "09ec2a7bb152e2252b53fa7803150007879548bc709c039df7627cabbd05d469"
 
 [[package]]
 name = "windows_i686_gnu"
-version = "0.48.0"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e9b5ad5ab802e97eb8e295ac6720e509ee4c243f69d781394014ebfe8bbfa0b"
+
+[[package]]
+name = "windows_i686_gnullvm"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "622a1962a7db830d6fd0a69683c80a18fda201879f0f447f065a3b7467daa241"
+checksum = "0eee52d38c090b3caa76c563b86c3a4bd71ef1a819287c19d586d7334ae8ed66"
 
 [[package]]
 name = "windows_i686_msvc"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4542c6e364ce21bf45d69fdd2a8e455fa38d316158cfd43b3ac1c5b1b19f8e00"
+checksum = "240948bc05c5e7c6dabba28bf89d89ffce3e303022809e73deaefe4f6ec56c66"
 
 [[package]]
 name = "windows_x86_64_gnu"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ca2b8a661f7628cbd23440e50b05d705db3686f894fc9580820623656af974b1"
+checksum = "147a5c80aabfbf0c7d901cb5895d1de30ef2907eb21fbbab29ca94c5b08b1a78"
 
 [[package]]
 name = "windows_x86_64_gnullvm"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7896dbc1f41e08872e9d5e8f8baa8fdd2677f29468c4e156210174edc7f7b953"
+checksum = "24d5b23dc417412679681396f2b49f3de8c1473deb516bd34410872eff51ed0d"
 
 [[package]]
 name = "windows_x86_64_msvc"
-version = "0.48.0"
+version = "0.52.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1a515f5799fe4961cb532f983ce2b23082366b898e52ffbce459c86f67c8378a"
+checksum = "589f6da84c646204747d1270a2a5661ea66ed1cced2631d546fdfb155959f9ec"
 
 [[package]]
 name = "xdg"
-version = "2.5.0"
+version = "2.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "213b7324336b53d2414b2db8537e56544d981803139155afa84f76eeebb7a546"
+
+[[package]]
+name = "yansi"
+version = "1.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cfe53a6657fd280eaa890a3bc59152892ffa3e30101319d168b781ed6529b049"
+
+[[package]]
+name = "zerocopy"
+version = "0.7.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "688597db5a750e9cad4511cb94729a078e274308099a0382b5b8203bbc767fee"
+checksum = "1b9b4fd18abc82b8136838da5d50bae7bdea537c574d8dc1a34ed098d6c166f0"
 dependencies = [
- "home",
+ "byteorder",
+ "zerocopy-derive",
 ]
 
 [[package]]
-name = "yansi"
-version = "0.5.1"
+name = "zerocopy-derive"
+version = "0.7.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09041cd90cf85f7f8b2df60c646f853b7f535ce68f85244eb6731cf89fa498ec"
+checksum = "fa4f8080344d4671fb4e831a13ad1e68092748387dfc4f55e356242fae12ce3e"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.85",
+]
diff --git a/fuzz/Cargo.toml b/fuzz/Cargo.toml
index 3deb140..a32f1a4 100644
--- a/fuzz/Cargo.toml
+++ b/fuzz/Cargo.toml
@@ -8,21 +8,21 @@ cargo-fuzz = true
 
 [dependencies]
 libfuzzer-sys = "0.4"
+afl = { version = "0.15.0", optional = true }
 html5gum = { path = "../" }
-html5gum_old = { version = "0.4.0", package = "html5gum" }
-afl = { version = "0.11.0", optional = true }
-# https://github.com/servo/html5ever/pull/460
-html5ever = { version = "*", git = "https://github.com/untitaker/html5ever", branch = "html5lib-tests-update" }
 pretty_assertions = "1.0.0"
 
+# thirdparty crates to fuzz against
+html5gum_old = { version = "0.6.0", package = "html5gum" }
+html5ever = "0.27.0"
+swc_common = "0.31.18"
+swc_html_parser = "0.37.21"
+swc_html_ast = "0.31.18"
 
 # lol-html and its dependencies
 lol_html = { version = "0.4", features = ["integration_test"] }
 encoding_rs = "0.8"
 bytes = "1"
-swc_common = "0.31.18"
-swc_html_parser = "0.37.21"
-swc_html_ast = "0.31.18"
 
 
 # Prevent this from interfering with workspaces
diff --git a/fuzz/README.md b/fuzz/README.md
index acc2108..206dc88 100644
--- a/fuzz/README.md
+++ b/fuzz/README.md
@@ -22,6 +22,7 @@ target.
   * `FUZZ_IGNORE_PARSE_ERRORS=order` will sort errors from both parsers such that order can be ignored.
   * `FUZZ_IGNORE_PARSE_ERRORS=1` will delete all errors so that parsing errors are not compared at all.
   * `FUZZ_IGNORE_PARSE_ERRORS=if-reference-contains:duplicate-attribute` will delete all errors _if_ any of them _in the old version of html5gum_ contains the string `duplicate-attribute`.
+  * `FUZZ_IGNORE_PARSE_ERRORS=if-testing-contains:duplicate-attribute` will delete all errors _if_ any of them _in the new version of html5gum_ contains the string `duplicate-attribute`.
 
   This envvar is a comma-separated list of instructions. For example,
   `FUZZ_IGNORE_PARSE_ERRORS=order,if-reference-contains:foo` means "ignore
diff --git a/fuzz/src/testcase/html5ever.rs b/fuzz/src/testcase/html5ever.rs
index 0c45b22..88a9bc6 100644
--- a/fuzz/src/testcase/html5ever.rs
+++ b/fuzz/src/testcase/html5ever.rs
@@ -64,22 +64,22 @@ impl<R: Reader, E: Emitter<Token = Token>> html5ever::tokenizer::TokenSink for T
         match (token, reference_token) {
             (Some(Token::StartTag(tag)), Token2::TagToken(tag2)) => {
                 assert_eq!(tag2.kind, TagKind::StartTag);
-                assert_eq!(tag.name, tag2.name.as_ref().as_bytes().to_owned().into());
+                assert_eq!(tag.name, tag2.name.as_ref().as_bytes().to_owned());
             }
             (Some(Token::EndTag(tag)), Token2::TagToken(tag2)) => {
                 assert_eq!(tag2.kind, TagKind::EndTag);
-                assert_eq!(tag.name, tag2.name.as_ref().as_bytes().to_owned().into());
+                assert_eq!(tag.name, tag2.name.as_ref().as_bytes().to_owned());
             }
             (None, Token2::EOFToken) => {}
             (Some(Token::Comment(comment)), Token2::CommentToken(comment2)) => {
-                assert_eq!(comment, comment2.as_ref().as_bytes().to_owned().into());
+                assert_eq!(comment, comment2.as_ref().as_bytes().to_owned());
             }
             (Some(Token::Doctype(doctype)), Token2::DoctypeToken(doctype2)) => {
                 assert_eq!(
                     doctype.name,
                     doctype2
                         .name
-                        .map(|x| x.as_ref().to_owned().into_bytes().into())
+                        .map(|x| x.as_ref().to_owned().into_bytes())
                         .unwrap_or_default()
                 );
                 assert_eq!(
diff --git a/fuzz/src/testcase/old_html5gum.rs b/fuzz/src/testcase/old_html5gum.rs
index 69681bf..3f50be4 100644
--- a/fuzz/src/testcase/old_html5gum.rs
+++ b/fuzz/src/testcase/old_html5gum.rs
@@ -43,6 +43,14 @@ pub fn run_old_html5gum(s: &str) {
                     testing_tokens.retain(isnt_error);
                 }
             }
+            x if x.starts_with("if-testing-contains:") => {
+                if testing_tokens.contains(&html5gum::Token::Error(
+                    x["if-testing-contains:".len()..].parse().unwrap(),
+                )) {
+                    reference_tokens.retain(isnt_old_error);
+                    testing_tokens.retain(isnt_error);
+                }
+            }
             x => panic!("unknown FUZZ_IGNORE_PARSE_ERRORS instruction: {}", x),
         }
     }
@@ -50,26 +58,26 @@ pub fn run_old_html5gum(s: &str) {
     let reference_tokens: Vec<_> = reference_tokens
         .into_iter()
         .map(|x| match x {
-            html5gum_old::Token::String(x) => Token::String(x.into()),
-            html5gum_old::Token::Comment(x) => Token::Comment(x.into()),
+            html5gum_old::Token::String(x) => Token::String(Vec::from(x).into()),
+            html5gum_old::Token::Comment(x) => Token::Comment(Vec::from(x).into()),
             html5gum_old::Token::StartTag(x) => Token::StartTag(StartTag {
-                name: x.name.into(),
+                name: Vec::from(x.name).into(),
                 attributes: x
                     .attributes
                     .into_iter()
-                    .map(|(k, v)| (k.into(), v.into()))
+                    .map(|(k, v)| (Vec::from(k).into(), Vec::from(v).into()))
                     .collect(),
                 self_closing: x.self_closing,
             }),
             html5gum_old::Token::EndTag(x) => Token::EndTag(EndTag {
-                name: x.name.into(),
+                name: Vec::from(x.name).into(),
             }),
             html5gum_old::Token::Error(x) => Token::Error(x.to_string().parse().unwrap()),
             html5gum_old::Token::Doctype(x) => Token::Doctype(Doctype {
-                name: x.name.into(),
+                name: Vec::from(x.name).into(),
                 force_quirks: x.force_quirks,
-                public_identifier: x.public_identifier.map(From::from),
-                system_identifier: x.system_identifier.map(From::from),
+                public_identifier: x.public_identifier.map(|x| Vec::from(x).into()),
+                system_identifier: x.system_identifier.map(|x| Vec::from(x).into())
             }),
         })
         .collect();
diff --git a/src/callbacks.rs b/src/callbacks.rs
new file mode 100644
index 0000000..5b3f33c
--- /dev/null
+++ b/src/callbacks.rs
@@ -0,0 +1,497 @@
+//! Consume the parsed HTML as a series of events through a callback.
+//!
+//! While using the [DefaultEmitter] provides an easy-to-use API with low performance, and
+//! implementing your own [Emitter] brings maximal performance and maximal pain, this is a middle
+//! ground. All strings are borrowed from some intermediate buffer instead of individually
+//! allocated.
+//!
+//! ```
+//! // Extract all text between span tags, in a naive (but fast) way. Does not handle tags inside of the span. See `examples/` as well.
+//! use html5gum::Tokenizer;
+//! use html5gum::callbacks::{CallbackEvent, CallbackEmitter};
+//!
+//! let mut is_in_span = false;
+//! let emitter = CallbackEmitter::new(move |event: CallbackEvent<'_>| -> Option<Vec<u8>> {
+//!     match event {
+//!         CallbackEvent::OpenStartTag { name } => {
+//!             is_in_span = name == b"span";
+//!         },
+//!         CallbackEvent::String { value } if is_in_span => {
+//!             return Some(value.to_vec());
+//!         }
+//!         CallbackEvent::EndTag { .. } => {
+//!             is_in_span = false;
+//!         }
+//!         _ => {}
+//!     }
+//!
+//!     None
+//! });
+//!
+//! let input = r#"<h1><span class=hello>Hello</span> world!</h1>"#;
+//! let text_fragments = Tokenizer::new_with_emitter(input, emitter)
+//!     .infallible()
+//!     .collect::<Vec<_>>();
+//!
+//! assert_eq!(text_fragments, vec![b"Hello".to_vec()]);
+//! ```
+
+use std::collections::VecDeque;
+use std::convert::Infallible;
+use std::mem::swap;
+
+use crate::utils::trace_log;
+use crate::{naive_next_state, Emitter, Error, State};
+
+/// Events used by [CallbackEmitter].
+///
+/// This operates at a slightly lower level than [Token], as start tags are split up into multiple
+/// events.
+#[derive(Debug)]
+pub enum CallbackEvent<'a> {
+    /// Visit the `"<mytag"` in `"<mytag mykey=myvalue>"`. Signifies the beginning of a new start
+    /// tag.
+    ///
+    /// Attributes have not yet been read.
+    OpenStartTag {
+        /// The name of the start tag.
+        name: &'a [u8],
+    },
+
+    /// Visit an attribute name, for example `"mykey"` in `"<mytag mykey=myvalue>"`.
+    ///
+    /// The attribute value has not yet been read.
+    AttributeName {
+        /// The name of the attribute.
+        name: &'a [u8],
+    },
+
+    /// Visit an attribute value, for example `"myvalue"` in `"<mytag mykey=myvalue>"`.
+    ///
+    /// Things like whitespace, quote handling is taken care of.
+    ///
+    /// After this event, the start tag may be closed using [CloseStartTag], or another
+    /// [AttributeName] may follow.
+    AttributeValue {
+        /// The value of the attribute.
+        value: &'a [u8],
+    },
+
+    /// Visit the end of the start tag, for example `">"` in `"<mytag mykey=myvalue>"`.
+    ///
+    CloseStartTag {
+        /// Whether the tag ended with `"/>"`.
+        ///
+        /// Note that in HTML5 this difference is largely ignored, and tags are considered
+        /// self-closing based on a hardcoded list of names, not based on syntax.
+        self_closing: bool,
+    },
+
+    /// Visit `"</mytag>".
+    ///
+    /// Note: Because of strangeness in the HTML spec, attributes may be observed outside of start
+    /// tags, before this event. It's best to ignore them as they are not valid HTML, but can still
+    /// be observed through most HTML parsers.
+    EndTag {
+        /// The name of the end tag.
+        name: &'a [u8],
+    },
+
+    /// Visit a string, as in, the actual text between tags. The content. Remember actual content
+    /// in HTML, before SPAs took over? I remember.
+    ///
+    /// It's guaranteed that all consecutive "character tokens" (as the spec calls them) are folded
+    /// into one string event.
+    String {
+        /// A series of character tokens.
+        value: &'a [u8],
+    },
+
+    /// Visit a comment, like `<!-- DON'T HACK THIS WEBSITE -->`
+    Comment {
+        /// The contents of the comment.
+        value: &'a [u8],
+    },
+
+    /// Visit `<!DOCTYPE html>`.
+    Doctype {
+        /// Name of the docstring.
+        name: &'a [u8],
+        /// Public identifier (see spec)
+        public_identifier: Option<&'a [u8]>,
+        /// System identifier (see spec)
+        system_identifier: Option<&'a [u8]>,
+        /// Enable quirksmode
+        force_quirks: bool,
+    },
+
+    /// Visit a parsing error.
+    Error(Error),
+}
+
+#[derive(Debug, Clone, Copy)]
+enum CurrentTag {
+    Start,
+    End,
+}
+
+#[derive(Debug)]
+struct CallbackState<F, T> {
+    callback: F,
+    emitted_tokens: VecDeque<T>,
+}
+
+/// This trait is implemented for all functions that have the same signature as
+/// [Callback::handle_event]. The trait only exists in case you want to implement it on a nameable
+/// type.
+pub trait Callback<T> {
+    /// Perform some action on a parsing event, and, optionally, return a value that can be yielded
+    /// from the [Tokenizer] iterator.
+    fn handle_event(&mut self, event: CallbackEvent<'_>) -> Option<T>;
+}
+
+impl<T, F> Callback<T> for F
+where
+    F: FnMut(CallbackEvent<'_>) -> Option<T>,
+{
+    fn handle_event(&mut self, event: CallbackEvent<'_>) -> Option<T> {
+        self(event)
+    }
+}
+
+impl<F, T> CallbackState<F, T>
+where
+    F: Callback<T>,
+{
+    fn emit_event(&mut self, event: CallbackEvent<'_>) {
+        let res = self.callback.handle_event(event);
+        if let Some(token) = res {
+            self.emitted_tokens.push_front(token);
+        }
+    }
+}
+
+impl<F, T> Default for CallbackState<F, T>
+where
+    F: Default,
+{
+    fn default() -> Self {
+        CallbackState {
+            callback: F::default(),
+            emitted_tokens: VecDeque::default(),
+        }
+    }
+}
+
+#[derive(Debug, Default)]
+struct EmitterState {
+    naively_switch_states: bool,
+
+    current_characters: Vec<u8>,
+    current_comment: Vec<u8>,
+
+    last_start_tag: Vec<u8>,
+    current_tag_had_attributes: bool,
+    current_tag_type: Option<CurrentTag>,
+    current_tag_self_closing: bool,
+    current_tag_name: Vec<u8>,
+    current_attribute_name: Vec<u8>,
+    current_attribute_value: Vec<u8>,
+
+    // strings related to doctype
+    doctype_name: Vec<u8>,
+    doctype_has_public_identifier: bool,
+    doctype_has_system_identifier: bool,
+    doctype_public_identifier: Vec<u8>,
+    doctype_system_identifier: Vec<u8>,
+    doctype_force_quirks: bool,
+}
+
+/// The emitter class to pass to [Tokenizer::new_with_emitter]
+#[derive(Debug)]
+pub struct CallbackEmitter<F, T = Infallible> {
+    // this struct is only split out so [CallbackState::emit_event] can borrow things concurrently
+    // with other attributes.
+    callback_state: CallbackState<F, T>,
+    emitter_state: EmitterState,
+}
+
+impl<F, T> Default for CallbackEmitter<F, T>
+where
+    F: Default,
+{
+    fn default() -> Self {
+        CallbackEmitter {
+            callback_state: CallbackState::default(),
+            emitter_state: EmitterState::default(),
+        }
+    }
+}
+
+impl<F, T> CallbackEmitter<F, T>
+where
+    F: Callback<T>,
+{
+    /// Create a new emitter. See type-level docs to understand basic usage.
+    ///
+    /// The given callback may return optional tokens that then become available through the
+    /// [Tokenizer]'s iterator. If that's not used, return [Option<Infallible>].
+    pub fn new(callback: F) -> Self {
+        CallbackEmitter {
+            callback_state: CallbackState {
+                callback,
+                emitted_tokens: VecDeque::new(),
+            },
+            emitter_state: EmitterState::default(),
+        }
+    }
+
+    /// Get mutable access to the inner callback.
+    pub fn callback_mut(&mut self) -> &mut F {
+        &mut self.callback_state.callback
+    }
+
+    /// Whether to use [`naive_next_state`] to switch states automatically.
+    ///
+    /// The default is off.
+    pub fn naively_switch_states(&mut self, yes: bool) {
+        self.emitter_state.naively_switch_states = yes;
+    }
+
+    fn flush_attribute_name(&mut self) {
+        if !self.emitter_state.current_attribute_name.is_empty() {
+            self.callback_state
+                .emit_event(CallbackEvent::AttributeName {
+                    name: &self.emitter_state.current_attribute_name,
+                });
+            self.emitter_state.current_attribute_name.clear();
+        }
+    }
+
+    fn flush_attribute(&mut self) {
+        self.flush_attribute_name();
+
+        if !self.emitter_state.current_attribute_value.is_empty() {
+            self.callback_state
+                .emit_event(CallbackEvent::AttributeValue {
+                    value: &self.emitter_state.current_attribute_value,
+                });
+            self.emitter_state.current_attribute_value.clear();
+        }
+    }
+
+    fn flush_open_start_tag(&mut self) {
+        if matches!(self.emitter_state.current_tag_type, Some(CurrentTag::Start))
+            && !self.emitter_state.current_tag_name.is_empty()
+        {
+            self.callback_state.emit_event(CallbackEvent::OpenStartTag {
+                name: &self.emitter_state.current_tag_name,
+            });
+
+            self.emitter_state.last_start_tag.clear();
+            swap(
+                &mut self.emitter_state.last_start_tag,
+                &mut self.emitter_state.current_tag_name,
+            );
+        }
+    }
+
+    fn flush_current_characters(&mut self) {
+        if self.emitter_state.current_characters.is_empty() {
+            return;
+        }
+
+        self.callback_state.emit_event(CallbackEvent::String {
+            value: &self.emitter_state.current_characters,
+        });
+        self.emitter_state.current_characters.clear();
+    }
+}
+impl<F, T> Emitter for CallbackEmitter<F, T>
+where
+    F: Callback<T>,
+{
+    type Token = T;
+
+    fn set_last_start_tag(&mut self, last_start_tag: Option<&[u8]>) {
+        self.emitter_state.last_start_tag.clear();
+        self.emitter_state
+            .last_start_tag
+            .extend(last_start_tag.unwrap_or_default());
+    }
+
+    fn emit_eof(&mut self) {
+        self.flush_current_characters();
+    }
+
+    fn emit_error(&mut self, error: Error) {
+        self.callback_state.emit_event(CallbackEvent::Error(error));
+    }
+
+    fn pop_token(&mut self) -> Option<Self::Token> {
+        self.callback_state.emitted_tokens.pop_back()
+    }
+
+    fn emit_string(&mut self, s: &[u8]) {
+        crate::utils::trace_log!("callbacks: emit_string, len={}", s.len());
+        self.emitter_state.current_characters.extend(s);
+    }
+
+    fn init_start_tag(&mut self) {
+        self.emitter_state.current_tag_name.clear();
+        self.emitter_state.current_tag_type = Some(CurrentTag::Start);
+        self.emitter_state.current_tag_self_closing = false;
+    }
+
+    fn init_end_tag(&mut self) {
+        self.emitter_state.current_tag_name.clear();
+        self.emitter_state.current_tag_type = Some(CurrentTag::End);
+        self.emitter_state.current_tag_had_attributes = false;
+    }
+
+    fn init_comment(&mut self) {
+        self.flush_current_characters();
+        self.emitter_state.current_comment.clear();
+    }
+
+    fn emit_current_tag(&mut self) -> Option<State> {
+        self.flush_attribute();
+        self.flush_current_characters();
+        match self.emitter_state.current_tag_type {
+            Some(CurrentTag::Start) => {
+                self.flush_open_start_tag();
+                self.callback_state
+                    .emit_event(CallbackEvent::CloseStartTag {
+                        self_closing: self.emitter_state.current_tag_self_closing,
+                    });
+            }
+            Some(CurrentTag::End) => {
+                if self.emitter_state.current_tag_had_attributes {
+                    self.emit_error(Error::EndTagWithAttributes);
+                }
+                self.emitter_state.last_start_tag.clear();
+                self.callback_state.emit_event(CallbackEvent::EndTag {
+                    name: &self.emitter_state.current_tag_name,
+                });
+            }
+            _ => {}
+        }
+
+        if self.emitter_state.naively_switch_states {
+            naive_next_state(&self.emitter_state.last_start_tag)
+        } else {
+            None
+        }
+    }
+    fn emit_current_comment(&mut self) {
+        self.callback_state.emit_event(CallbackEvent::Comment {
+            value: &self.emitter_state.current_comment,
+        });
+        self.emitter_state.current_comment.clear();
+    }
+
+    fn emit_current_doctype(&mut self) {
+        self.callback_state.emit_event(CallbackEvent::Doctype {
+            name: &self.emitter_state.doctype_name,
+            public_identifier: if self.emitter_state.doctype_has_public_identifier {
+                Some(&self.emitter_state.doctype_public_identifier)
+            } else {
+                None
+            },
+            system_identifier: if self.emitter_state.doctype_has_system_identifier {
+                Some(&self.emitter_state.doctype_system_identifier)
+            } else {
+                None
+            },
+            force_quirks: self.emitter_state.doctype_force_quirks,
+        });
+    }
+
+    fn set_self_closing(&mut self) {
+        trace_log!("set_self_closing");
+        if matches!(self.emitter_state.current_tag_type, Some(CurrentTag::End)) {
+            self.callback_state
+                .emit_event(CallbackEvent::Error(Error::EndTagWithTrailingSolidus));
+        } else {
+            self.emitter_state.current_tag_self_closing = true;
+        }
+    }
+
+    fn set_force_quirks(&mut self) {
+        self.emitter_state.doctype_force_quirks = true;
+    }
+
+    fn push_tag_name(&mut self, s: &[u8]) {
+        self.emitter_state.current_tag_name.extend(s);
+    }
+
+    fn push_comment(&mut self, s: &[u8]) {
+        self.emitter_state.current_comment.extend(s);
+    }
+
+    fn push_doctype_name(&mut self, s: &[u8]) {
+        self.emitter_state.doctype_name.extend(s);
+    }
+
+    fn init_doctype(&mut self) {
+        self.flush_current_characters();
+        self.emitter_state.doctype_name.clear();
+        self.emitter_state.doctype_has_public_identifier = false;
+        self.emitter_state.doctype_has_system_identifier = false;
+        self.emitter_state.doctype_public_identifier.clear();
+        self.emitter_state.doctype_system_identifier.clear();
+        self.emitter_state.doctype_force_quirks = false;
+    }
+
+    fn init_attribute(&mut self) {
+        self.flush_open_start_tag();
+        self.flush_attribute();
+        self.emitter_state.current_tag_had_attributes = true;
+    }
+
+    fn push_attribute_name(&mut self, s: &[u8]) {
+        self.emitter_state.current_attribute_name.extend(s);
+    }
+
+    fn push_attribute_value(&mut self, s: &[u8]) {
+        self.flush_attribute_name();
+        self.emitter_state.current_attribute_value.extend(s);
+    }
+
+    fn set_doctype_public_identifier(&mut self, value: &[u8]) {
+        self.emitter_state.doctype_has_public_identifier = true;
+        self.emitter_state.doctype_public_identifier.clear();
+        self.emitter_state.doctype_public_identifier.extend(value);
+    }
+    fn set_doctype_system_identifier(&mut self, value: &[u8]) {
+        self.emitter_state.doctype_has_system_identifier = true;
+        self.emitter_state.doctype_system_identifier.clear();
+        self.emitter_state.doctype_system_identifier.extend(value);
+    }
+    fn push_doctype_public_identifier(&mut self, value: &[u8]) {
+        self.emitter_state.doctype_public_identifier.extend(value);
+    }
+    fn push_doctype_system_identifier(&mut self, value: &[u8]) {
+        self.emitter_state.doctype_system_identifier.extend(value);
+    }
+
+    fn current_is_appropriate_end_tag_token(&mut self) -> bool {
+        if self.emitter_state.last_start_tag.is_empty() {
+            crate::utils::trace_log!(
+                "current_is_appropriate_end_tag_token: no, because last_start_tag is empty"
+            );
+            return false;
+        }
+
+        if !matches!(self.emitter_state.current_tag_type, Some(CurrentTag::End)) {
+            crate::utils::trace_log!(
+                "current_is_appropriate_end_tag_token: no, because current_tag_type is not end"
+            );
+            return false;
+        }
+
+        crate::utils::trace_log!("last_start_tag = {:?}", self.emitter_state.last_start_tag);
+        crate::utils::trace_log!("current_tag = {:?}", self.emitter_state.current_tag_name);
+        self.emitter_state.last_start_tag == self.emitter_state.current_tag_name
+    }
+}
diff --git a/src/default_emitter.rs b/src/default_emitter.rs
new file mode 100644
index 0000000..b245795
--- /dev/null
+++ b/src/default_emitter.rs
@@ -0,0 +1,266 @@
+use std::collections::btree_map::Entry;
+use std::collections::BTreeMap;
+use std::mem::take;
+
+use crate::{Emitter, Error, HtmlString, State};
+
+use crate::callbacks::{Callback, CallbackEmitter, CallbackEvent};
+
+#[derive(Debug, Default)]
+struct OurCallback {
+    tag_name: Vec<u8>,
+    attribute_name: HtmlString,
+    attribute_map: BTreeMap<HtmlString, HtmlString>,
+}
+
+impl Callback<Token> for OurCallback {
+    fn handle_event(&mut self, event: CallbackEvent<'_>) -> Option<Token> {
+        crate::utils::trace_log!("event: {:?}", event);
+        match event {
+            CallbackEvent::OpenStartTag { name } => {
+                self.tag_name.clear();
+                self.tag_name.extend(name);
+                None
+            }
+            CallbackEvent::AttributeName { name } => {
+                self.attribute_name.clear();
+                match self.attribute_map.entry(name.to_owned().into()) {
+                    Entry::Occupied(_) => Some(Token::Error(Error::DuplicateAttribute)),
+                    Entry::Vacant(vacant) => {
+                        self.attribute_name.extend(name);
+                        vacant.insert(Default::default());
+                        None
+                    }
+                }
+            }
+            CallbackEvent::AttributeValue { value } => {
+                if !self.attribute_name.is_empty() {
+                    self.attribute_map
+                        .get_mut(&self.attribute_name)
+                        .unwrap()
+                        .extend(value);
+                }
+                None
+            }
+            CallbackEvent::CloseStartTag { self_closing } => Some(Token::StartTag(StartTag {
+                self_closing,
+                name: take(&mut self.tag_name).into(),
+                attributes: take(&mut self.attribute_map),
+            })),
+            CallbackEvent::EndTag { name } => {
+                self.attribute_map.clear();
+                Some(Token::EndTag(EndTag {
+                    name: name.to_owned().into(),
+                }))
+            }
+            CallbackEvent::String { value } => Some(Token::String(value.to_owned().into())),
+            CallbackEvent::Comment { value } => Some(Token::Comment(value.to_owned().into())),
+            CallbackEvent::Doctype {
+                name,
+                public_identifier,
+                system_identifier,
+                force_quirks,
+            } => Some(Token::Doctype(Doctype {
+                force_quirks,
+                name: name.to_owned().into(),
+                public_identifier: public_identifier.map(|x| x.to_owned().into()),
+                system_identifier: system_identifier.map(|x| x.to_owned().into()),
+            })),
+            CallbackEvent::Error(error) => Some(Token::Error(error)),
+        }
+    }
+}
+
+/// The default implementation of [`crate::Emitter`], used to produce ("emit") tokens.
+#[derive(Default, Debug)]
+pub struct DefaultEmitter {
+    inner: CallbackEmitter<OurCallback, Token>,
+}
+
+impl DefaultEmitter {
+    /// Whether to use [`naive_next_state`] to switch states automatically.
+    ///
+    /// The default is off.
+    pub fn naively_switch_states(&mut self, yes: bool) {
+        self.inner.naively_switch_states(yes)
+    }
+}
+
+// opaque type around inner emitter
+impl Emitter for DefaultEmitter {
+    type Token = Token;
+
+    fn set_last_start_tag(&mut self, last_start_tag: Option<&[u8]>) {
+        self.inner.set_last_start_tag(last_start_tag)
+    }
+
+    fn emit_eof(&mut self) {
+        self.inner.emit_eof()
+    }
+
+    fn emit_error(&mut self, error: Error) {
+        self.inner.emit_error(error)
+    }
+
+    fn should_emit_errors(&mut self) -> bool {
+        self.inner.should_emit_errors()
+    }
+
+    fn pop_token(&mut self) -> Option<Self::Token> {
+        self.inner.pop_token()
+    }
+    fn emit_string(&mut self, c: &[u8]) {
+        self.inner.emit_string(c)
+    }
+
+    fn init_start_tag(&mut self) {
+        self.inner.init_start_tag()
+    }
+
+    fn init_end_tag(&mut self) {
+        self.inner.init_end_tag()
+    }
+
+    fn init_comment(&mut self) {
+        self.inner.init_comment()
+    }
+
+    fn emit_current_tag(&mut self) -> Option<State> {
+        self.inner.emit_current_tag()
+    }
+
+    fn emit_current_comment(&mut self) {
+        self.inner.emit_current_comment()
+    }
+
+    fn emit_current_doctype(&mut self) {
+        self.inner.emit_current_doctype()
+    }
+
+    fn set_self_closing(&mut self) {
+        self.inner.set_self_closing()
+    }
+
+    fn set_force_quirks(&mut self) {
+        self.inner.set_force_quirks()
+    }
+
+    fn push_tag_name(&mut self, s: &[u8]) {
+        self.inner.push_tag_name(s)
+    }
+
+    fn push_comment(&mut self, s: &[u8]) {
+        self.inner.push_comment(s)
+    }
+
+    fn push_doctype_name(&mut self, s: &[u8]) {
+        self.inner.push_doctype_name(s)
+    }
+
+    fn init_doctype(&mut self) {
+        self.inner.init_doctype()
+    }
+
+    fn init_attribute(&mut self) {
+        self.inner.init_attribute()
+    }
+
+    fn push_attribute_name(&mut self, s: &[u8]) {
+        self.inner.push_attribute_name(s)
+    }
+
+    fn push_attribute_value(&mut self, s: &[u8]) {
+        self.inner.push_attribute_value(s)
+    }
+
+    fn set_doctype_public_identifier(&mut self, value: &[u8]) {
+        self.inner.set_doctype_public_identifier(value)
+    }
+
+    fn set_doctype_system_identifier(&mut self, value: &[u8]) {
+        self.inner.set_doctype_system_identifier(value)
+    }
+
+    fn push_doctype_public_identifier(&mut self, s: &[u8]) {
+        self.inner.push_doctype_public_identifier(s)
+    }
+
+    fn push_doctype_system_identifier(&mut self, s: &[u8]) {
+        self.inner.push_doctype_system_identifier(s)
+    }
+
+    fn current_is_appropriate_end_tag_token(&mut self) -> bool {
+        self.inner.current_is_appropriate_end_tag_token()
+    }
+
+    fn adjusted_current_node_present_but_not_in_html_namespace(&mut self) -> bool {
+        self.inner
+            .adjusted_current_node_present_but_not_in_html_namespace()
+    }
+}
+
+/// A HTML end/close tag, such as `<p>` or `<a>`.
+#[derive(Debug, Default, Eq, PartialEq, Clone)]
+pub struct StartTag {
+    /// Whether this tag is self-closing. If it is self-closing, no following [`EndTag`] should be
+    /// expected.
+    pub self_closing: bool,
+
+    /// The start tag's name, such as `"p"` or `"a"`.
+    pub name: HtmlString,
+
+    /// A mapping for any HTML attributes this start tag may have.
+    ///
+    /// Duplicate attributes are ignored after the first one as per WHATWG spec. Implement your own
+    /// [`Emitter`] to tweak this behavior.
+    pub attributes: BTreeMap<HtmlString, HtmlString>,
+}
+
+/// A HTML end/close tag, such as `</p>` or `</a>`.
+#[derive(Debug, Default, Eq, PartialEq, Clone)]
+pub struct EndTag {
+    /// The ending tag's name, such as `"p"` or `"a"`.
+    pub name: HtmlString,
+}
+
+/// A doctype. Some examples:
+///
+/// * `<!DOCTYPE {name}>`
+/// * `<!DOCTYPE {name} PUBLIC '{public_identifier}'>`
+/// * `<!DOCTYPE {name} SYSTEM '{system_identifier}'>`
+/// * `<!DOCTYPE {name} PUBLIC '{public_identifier}' '{system_identifier}'>`
+#[derive(Debug, Eq, PartialEq, Clone)]
+pub struct Doctype {
+    /// The ["force quirks"](https://html.spec.whatwg.org/#force-quirks-flag) flag.
+    pub force_quirks: bool,
+
+    /// The doctype's name. For HTML documents this is "html".
+    pub name: HtmlString,
+
+    /// The doctype's public identifier.
+    pub public_identifier: Option<HtmlString>,
+
+    /// The doctype's system identifier.
+    pub system_identifier: Option<HtmlString>,
+}
+
+/// The token type used by default. You can define your own token type by implementing the
+/// [`crate::Emitter`] trait and using [`crate::Tokenizer::new_with_emitter`].
+#[derive(Debug, Eq, PartialEq, Clone)]
+pub enum Token {
+    /// A HTML start tag.
+    StartTag(StartTag),
+    /// A HTML end tag.
+    EndTag(EndTag),
+    /// A literal string.
+    String(HtmlString),
+    /// A HTML comment.
+    Comment(HtmlString),
+    /// A HTML doctype declaration.
+    Doctype(Doctype),
+    /// A HTML parsing error.
+    ///
+    /// Can be skipped over, the tokenizer is supposed to recover from the error and continues with
+    /// more tokens afterward.
+    Error(Error),
+}
diff --git a/src/emitter.rs b/src/emitter.rs
index 0040fce..9e911e1 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -1,127 +1,5 @@
-use std::borrow::{Borrow, BorrowMut};
-use std::collections::BTreeMap;
-use std::collections::BTreeSet;
-use std::collections::VecDeque;
-use std::fmt::{Debug, Formatter};
-use std::mem;
-use std::ops::{Deref, DerefMut};
-
 use crate::{Error, State};
 
-/// A wrapper around a bytestring.
-///
-/// This newtype only exists to provide a nicer `Debug` impl
-#[derive(Clone, Default, Eq, PartialEq, Ord, PartialOrd, Hash)]
-pub struct HtmlString(pub Vec<u8>);
-
-impl Deref for HtmlString {
-    type Target = Vec<u8>;
-
-    fn deref(&self) -> &Self::Target {
-        &self.0
-    }
-}
-
-impl DerefMut for HtmlString {
-    fn deref_mut(&mut self) -> &mut Self::Target {
-        &mut self.0
-    }
-}
-
-impl Debug for HtmlString {
-    fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), std::fmt::Error> {
-        write!(f, "b\"")?;
-        for &byte in &self.0 {
-            for ch in std::ascii::escape_default(byte) {
-                write!(f, "{}", ch as char)?;
-            }
-        }
-
-        write!(f, "\"")
-    }
-}
-
-impl Borrow<[u8]> for HtmlString {
-    fn borrow(&self) -> &[u8] {
-        &self.0
-    }
-}
-
-impl BorrowMut<[u8]> for HtmlString {
-    fn borrow_mut(&mut self) -> &mut [u8] {
-        &mut self.0
-    }
-}
-
-impl AsRef<[u8]> for HtmlString {
-    fn as_ref(&self) -> &[u8] {
-        &self.0
-    }
-}
-
-impl<const N: usize> PartialEq<&[u8; N]> for HtmlString {
-    fn eq(&self, other: &&[u8; N]) -> bool {
-        self.0 == *other
-    }
-}
-
-impl<const N: usize> PartialEq<HtmlString> for &[u8; N] {
-    fn eq(&self, other: &HtmlString) -> bool {
-        other.0 == *self
-    }
-}
-
-impl PartialEq<&[u8]> for HtmlString {
-    fn eq(&self, other: &&[u8]) -> bool {
-        self.0 == *other
-    }
-}
-
-impl PartialEq<HtmlString> for &[u8] {
-    fn eq(&self, other: &HtmlString) -> bool {
-        *self == other.0
-    }
-}
-
-#[test]
-fn test_eq_html_str_and_byte_literal() {
-    assert!(HtmlString(b"hello world".to_vec()) == b"hello world");
-}
-
-#[test]
-fn test_eq_byte_literal_and_html_str() {
-    assert!(b"hello world" == HtmlString(b"hello world".to_vec()));
-}
-
-#[test]
-fn test_eq_html_str_and_byte_slice() {
-    assert!(HtmlString(b"hello world".to_vec()) == b"hello world".as_slice());
-}
-
-#[test]
-fn test_eq_byte_slice_and_html_str() {
-    assert!(b"hello world".as_slice() == HtmlString(b"hello world".to_vec()));
-}
-
-#[test]
-fn test_borrowing() {
-    // demonstrate a usecase for Borrow/BorrowMut
-    let tag = StartTag::default();
-    assert!(!tag.attributes.contains_key(b"href".as_slice()));
-}
-
-impl From<Vec<u8>> for HtmlString {
-    fn from(vec: Vec<u8>) -> HtmlString {
-        HtmlString(vec)
-    }
-}
-
-impl From<HtmlString> for Vec<u8> {
-    fn from(other: HtmlString) -> Vec<u8> {
-        other.0
-    }
-}
-
 /// An emitter is an object providing methods to the tokenizer to produce tokens.
 ///
 /// Domain-specific applications of the HTML tokenizer can manually implement this trait to
@@ -349,327 +227,3 @@ pub fn naive_next_state(tag_name: &[u8]) -> Option<State> {
         _ => None,
     }
 }
-
-/// The default implementation of [`crate::Emitter`], used to produce ("emit") tokens.
-#[derive(Debug, Default)]
-pub struct DefaultEmitter {
-    current_characters: HtmlString,
-    current_token: Option<Token>,
-    last_start_tag: HtmlString,
-    current_attribute: Option<(HtmlString, HtmlString)>,
-    seen_attributes: BTreeSet<HtmlString>,
-    emitted_tokens: VecDeque<Token>,
-    naively_switch_states: bool,
-}
-
-impl DefaultEmitter {
-    /// Whether to use [`naive_next_state`] to switch states automatically.
-    ///
-    /// The default is off.
-    pub fn naively_switch_states(&mut self, yes: bool) {
-        self.naively_switch_states = yes;
-    }
-
-    fn emit_token(&mut self, token: Token) {
-        self.flush_current_characters();
-        self.emitted_tokens.push_front(token);
-    }
-
-    fn flush_current_attribute(&mut self) {
-        if let Some((k, v)) = self.current_attribute.take() {
-            match self.current_token {
-                Some(Token::StartTag(ref mut tag)) => {
-                    let mut error = None;
-                    tag.attributes
-                        .entry(k)
-                        .and_modify(|_| {
-                            error = Some(Error::DuplicateAttribute);
-                        })
-                        .or_insert(v);
-
-                    if let Some(e) = error {
-                        self.emit_error(e);
-                    }
-                }
-                Some(Token::EndTag(_)) => {
-                    if !self.seen_attributes.insert(k) {
-                        self.emit_error(Error::DuplicateAttribute);
-                    }
-                }
-                _ => {
-                    debug_assert!(false);
-                }
-            }
-        }
-    }
-
-    fn flush_current_characters(&mut self) {
-        if self.current_characters.is_empty() {
-            return;
-        }
-
-        let s = mem::take(&mut self.current_characters);
-        self.emit_token(Token::String(s));
-    }
-}
-
-impl Emitter for DefaultEmitter {
-    type Token = Token;
-
-    fn set_last_start_tag(&mut self, last_start_tag: Option<&[u8]>) {
-        self.last_start_tag.clear();
-        self.last_start_tag
-            .extend(last_start_tag.unwrap_or_default());
-    }
-
-    fn emit_eof(&mut self) {
-        self.flush_current_characters();
-    }
-
-    fn emit_error(&mut self, error: Error) {
-        // bypass character flushing in self.emit_token: we don't need the error location to be
-        // that exact
-        self.emitted_tokens.push_front(Token::Error(error));
-    }
-
-    fn pop_token(&mut self) -> Option<Self::Token> {
-        self.emitted_tokens.pop_back()
-    }
-
-    fn emit_string(&mut self, s: &[u8]) {
-        self.current_characters.extend(s);
-    }
-
-    fn init_start_tag(&mut self) {
-        self.current_token = Some(Token::StartTag(StartTag::default()));
-    }
-    fn init_end_tag(&mut self) {
-        self.current_token = Some(Token::EndTag(EndTag::default()));
-        self.seen_attributes.clear();
-    }
-
-    fn init_comment(&mut self) {
-        self.current_token = Some(Token::Comment(HtmlString::default()));
-    }
-    fn emit_current_tag(&mut self) -> Option<State> {
-        self.flush_current_attribute();
-        let mut token = self.current_token.take().unwrap();
-        match token {
-            Token::EndTag(_) => {
-                if !self.seen_attributes.is_empty() {
-                    self.emit_error(Error::EndTagWithAttributes);
-                }
-                self.seen_attributes.clear();
-                self.set_last_start_tag(None);
-            }
-            Token::StartTag(ref mut tag) => {
-                self.set_last_start_tag(Some(&tag.name));
-            }
-            _ => debug_assert!(false),
-        }
-        self.emit_token(token);
-        if self.naively_switch_states {
-            naive_next_state(&self.last_start_tag)
-        } else {
-            None
-        }
-    }
-    fn emit_current_comment(&mut self) {
-        let comment = self.current_token.take().unwrap();
-        debug_assert!(matches!(comment, Token::Comment(_)));
-        self.emit_token(comment);
-    }
-
-    fn emit_current_doctype(&mut self) {
-        let doctype = self.current_token.take().unwrap();
-        debug_assert!(matches!(doctype, Token::Doctype(_)));
-        self.emit_token(doctype);
-    }
-
-    fn set_self_closing(&mut self) {
-        let tag = self.current_token.as_mut().unwrap();
-        match tag {
-            Token::StartTag(StartTag {
-                ref mut self_closing,
-                ..
-            }) => {
-                *self_closing = true;
-            }
-            Token::EndTag(_) => {
-                self.emit_error(Error::EndTagWithTrailingSolidus);
-            }
-            _ => {
-                debug_assert!(false);
-            }
-        }
-    }
-    fn set_force_quirks(&mut self) {
-        match self.current_token {
-            Some(Token::Doctype(ref mut doctype)) => doctype.force_quirks = true,
-            _ => debug_assert!(false),
-        }
-    }
-    fn push_tag_name(&mut self, s: &[u8]) {
-        match self.current_token {
-            Some(
-                Token::StartTag(StartTag { ref mut name, .. })
-                | Token::EndTag(EndTag { ref mut name, .. }),
-            ) => {
-                name.extend(s);
-            }
-            _ => debug_assert!(false),
-        }
-    }
-
-    fn push_comment(&mut self, s: &[u8]) {
-        match self.current_token {
-            Some(Token::Comment(ref mut data)) => data.extend(s),
-            _ => debug_assert!(false),
-        }
-    }
-
-    fn push_doctype_name(&mut self, s: &[u8]) {
-        match self.current_token {
-            Some(Token::Doctype(ref mut doctype)) => doctype.name.extend(s),
-            _ => debug_assert!(false),
-        }
-    }
-    fn init_doctype(&mut self) {
-        self.current_token = Some(Token::Doctype(Doctype {
-            name: HtmlString::default(),
-            force_quirks: false,
-            public_identifier: None,
-            system_identifier: None,
-        }));
-    }
-
-    fn init_attribute(&mut self) {
-        self.flush_current_attribute();
-        self.current_attribute = Some(Default::default());
-    }
-    fn push_attribute_name(&mut self, s: &[u8]) {
-        self.current_attribute.as_mut().unwrap().0.extend(s);
-    }
-    fn push_attribute_value(&mut self, s: &[u8]) {
-        self.current_attribute.as_mut().unwrap().1.extend(s);
-    }
-    fn set_doctype_public_identifier(&mut self, value: &[u8]) {
-        if let Some(Token::Doctype(Doctype {
-            ref mut public_identifier,
-            ..
-        })) = self.current_token
-        {
-            *public_identifier = Some(value.to_vec().into());
-        } else {
-            debug_assert!(false);
-        }
-    }
-    fn set_doctype_system_identifier(&mut self, value: &[u8]) {
-        if let Some(Token::Doctype(Doctype {
-            ref mut system_identifier,
-            ..
-        })) = self.current_token
-        {
-            *system_identifier = Some(value.to_vec().into());
-        } else {
-            debug_assert!(false);
-        }
-    }
-    fn push_doctype_public_identifier(&mut self, s: &[u8]) {
-        if let Some(Token::Doctype(Doctype {
-            public_identifier: Some(ref mut id),
-            ..
-        })) = self.current_token
-        {
-            id.extend(s);
-        } else {
-            debug_assert!(false);
-        }
-    }
-    fn push_doctype_system_identifier(&mut self, s: &[u8]) {
-        if let Some(Token::Doctype(Doctype {
-            system_identifier: Some(ref mut id),
-            ..
-        })) = self.current_token
-        {
-            id.extend(s);
-        } else {
-            debug_assert!(false);
-        }
-    }
-
-    fn current_is_appropriate_end_tag_token(&mut self) -> bool {
-        match self.current_token {
-            Some(Token::EndTag(ref tag)) => {
-                !self.last_start_tag.is_empty() && self.last_start_tag == tag.name
-            }
-            _ => false,
-        }
-    }
-}
-
-/// A HTML end/close tag, such as `<p>` or `<a>`.
-#[derive(Debug, Default, Eq, PartialEq, Clone)]
-pub struct StartTag {
-    /// Whether this tag is self-closing. If it is self-closing, no following [`EndTag`] should be
-    /// expected.
-    pub self_closing: bool,
-
-    /// The start tag's name, such as `"p"` or `"a"`.
-    pub name: HtmlString,
-
-    /// A mapping for any HTML attributes this start tag may have.
-    ///
-    /// Duplicate attributes are ignored after the first one as per WHATWG spec. Implement your own
-    /// [`Emitter`] to tweak this behavior.
-    pub attributes: BTreeMap<HtmlString, HtmlString>,
-}
-
-/// A HTML end/close tag, such as `</p>` or `</a>`.
-#[derive(Debug, Default, Eq, PartialEq, Clone)]
-pub struct EndTag {
-    /// The ending tag's name, such as `"p"` or `"a"`.
-    pub name: HtmlString,
-}
-
-/// A doctype. Some examples:
-///
-/// * `<!DOCTYPE {name}>`
-/// * `<!DOCTYPE {name} PUBLIC '{public_identifier}'>`
-/// * `<!DOCTYPE {name} SYSTEM '{system_identifier}'>`
-/// * `<!DOCTYPE {name} PUBLIC '{public_identifier}' '{system_identifier}'>`
-#[derive(Debug, Eq, PartialEq, Clone)]
-pub struct Doctype {
-    /// The ["force quirks"](https://html.spec.whatwg.org/#force-quirks-flag) flag.
-    pub force_quirks: bool,
-
-    /// The doctype's name. For HTML documents this is "html".
-    pub name: HtmlString,
-
-    /// The doctype's public identifier.
-    pub public_identifier: Option<HtmlString>,
-
-    /// The doctype's system identifier.
-    pub system_identifier: Option<HtmlString>,
-}
-
-/// The token type used by default. You can define your own token type by implementing the
-/// [`crate::Emitter`] trait and using [`crate::Tokenizer::new_with_emitter`].
-#[derive(Debug, Eq, PartialEq, Clone)]
-pub enum Token {
-    /// A HTML start tag.
-    StartTag(StartTag),
-    /// A HTML end tag.
-    EndTag(EndTag),
-    /// A literal string.
-    String(HtmlString),
-    /// A HTML comment.
-    Comment(HtmlString),
-    /// A HTML doctype declaration.
-    Doctype(Doctype),
-    /// A HTML parsing error.
-    ///
-    /// Can be skipped over, the tokenizer is supposed to recover from the error and continues with
-    /// more tokens afterward.
-    Error(Error),
-}
diff --git a/src/html5ever_emitter.rs b/src/html5ever_emitter.rs
index 4f792ae..4398427 100644
--- a/src/html5ever_emitter.rs
+++ b/src/html5ever_emitter.rs
@@ -1,78 +1,161 @@
 use std::convert::Infallible;
 
-use crate::{DefaultEmitter, Emitter, Error, State, Token};
-use html5ever::tokenizer::states::RawKind;
+use crate::callbacks::{Callback, CallbackEmitter, CallbackEvent};
+use crate::utils::trace_log;
+use crate::{Emitter, Error, State};
+
 use html5ever::tokenizer::{
-    Doctype, Tag, TagKind, Token as Html5everToken, TokenSink, TokenSinkResult,
+    states::RawKind, Doctype, Tag, TagKind, Token as Html5everToken, TokenSink, TokenSinkResult,
 };
 use html5ever::{Attribute, QualName};
 
 const BOGUS_LINENO: u64 = 1;
 
+#[derive(Debug)]
+struct OurCallback<'a, S> {
+    sink: &'a mut S,
+    current_start_tag: Option<Tag>,
+    next_state: Option<State>,
+}
+
+impl<'a, S: TokenSink> OurCallback<'a, S> {
+    fn handle_sink_result<H>(&mut self, result: TokenSinkResult<H>) {
+        match result {
+            TokenSinkResult::Continue => {}
+            TokenSinkResult::Script(_) => {
+                self.next_state = Some(State::Data);
+                // TODO: suspend tokenizer for script
+            }
+            TokenSinkResult::Plaintext => {
+                self.next_state = Some(State::PlainText);
+            }
+            TokenSinkResult::RawData(RawKind::Rcdata) => {
+                self.next_state = Some(State::RcData);
+            }
+            TokenSinkResult::RawData(RawKind::Rawtext) => {
+                self.next_state = Some(State::RawText);
+            }
+            TokenSinkResult::RawData(RawKind::ScriptData) => {
+                self.next_state = Some(State::ScriptData);
+            }
+            TokenSinkResult::RawData(RawKind::ScriptDataEscaped(_)) => {
+                todo!()
+            }
+        }
+    }
+
+    fn sink_token(&mut self, token: Html5everToken) {
+        trace_log!("sink_token: {:?}", token);
+        let result = self.sink.process_token(token, BOGUS_LINENO);
+        self.handle_sink_result(result);
+    }
+}
+
+impl<'a, S: TokenSink> Callback<Infallible> for OurCallback<'a, S> {
+    fn handle_event(&mut self, event: CallbackEvent<'_>) -> Option<Infallible> {
+        trace_log!("Html5everEmitter::handle_event: {:?}", event);
+        match event {
+            CallbackEvent::OpenStartTag { name } => {
+                self.current_start_tag = Some(Tag {
+                    kind: TagKind::StartTag,
+                    name: String::from_utf8_lossy(name).into_owned().into(),
+                    self_closing: false,
+                    attrs: Default::default(),
+                });
+            }
+            CallbackEvent::AttributeName { name } => {
+                if let Some(ref mut tag) = self.current_start_tag {
+                    tag.attrs.push(Attribute {
+                        name: QualName::new(
+                            None,
+                            Default::default(),
+                            String::from_utf8_lossy(name).into_owned().into(),
+                        ),
+                        value: Default::default(),
+                    });
+                }
+            }
+            CallbackEvent::AttributeValue { value } => {
+                if let Some(ref mut tag) = self.current_start_tag {
+                    if let Some(attr) = tag.attrs.last_mut() {
+                        attr.value.push_slice(&String::from_utf8_lossy(value));
+                    }
+                }
+            }
+            CallbackEvent::CloseStartTag { self_closing } => {
+                if let Some(mut tag) = self.current_start_tag.take() {
+                    tag.self_closing = self_closing;
+                    self.sink_token(Html5everToken::TagToken(tag));
+                }
+            }
+            CallbackEvent::EndTag { name } => {
+                self.sink_token(Html5everToken::TagToken(Tag {
+                    kind: TagKind::EndTag,
+                    name: String::from_utf8_lossy(name).into_owned().into(),
+                    self_closing: false,
+                    attrs: Default::default(),
+                }));
+            }
+            CallbackEvent::String { value } => {
+                let mut first = true;
+                for part in String::from_utf8_lossy(value).split('\0') {
+                    if !first {
+                        self.sink_token(Html5everToken::NullCharacterToken);
+                    }
+
+                    first = false;
+                    self.sink_token(Html5everToken::CharacterTokens(part.to_owned().into()));
+                }
+            }
+            CallbackEvent::Comment { value } => {
+                self.sink_token(Html5everToken::CommentToken(
+                    String::from_utf8_lossy(value).into_owned().into(),
+                ));
+            }
+            CallbackEvent::Doctype {
+                name,
+                public_identifier,
+                system_identifier,
+                force_quirks,
+            } => {
+                self.sink_token(Html5everToken::DoctypeToken(Doctype {
+                    name: Some(name)
+                        .filter(|x| !x.is_empty())
+                        .map(|x| String::from_utf8_lossy(x).into_owned().into()),
+                    public_id: public_identifier
+                        .map(|x| String::from_utf8_lossy(x).into_owned().into()),
+                    system_id: system_identifier
+                        .map(|x| String::from_utf8_lossy(x).into_owned().into()),
+                    force_quirks,
+                }));
+            }
+            CallbackEvent::Error(error) => {
+                self.sink_token(Html5everToken::ParseError(error.as_str().into()));
+            }
+        }
+
+        None
+    }
+}
+
 /// A compatibility layer that allows you to plug the TreeBuilder from html5ever into the tokenizer
 /// from html5gum.
 ///
 /// See [`examples/scraper.rs`] for usage.
 #[derive(Debug)]
 pub struct Html5everEmitter<'a, S: TokenSink> {
-    next_state: Option<State>,
-    sink: &'a mut S,
-    // TODO: get rid of default emitter, construct html5ever tokens directly
-    emitter_inner: DefaultEmitter,
+    emitter_inner: CallbackEmitter<OurCallback<'a, S>>,
 }
 
 impl<'a, S: TokenSink> Html5everEmitter<'a, S> {
     /// Construct the compatibility layer.
     pub fn new(sink: &'a mut S) -> Self {
         Html5everEmitter {
-            next_state: None,
-            sink,
-            emitter_inner: DefaultEmitter::default(),
-        }
-    }
-
-    fn pop_token_inner(&mut self) {
-        while let Some(token) = self.emitter_inner.pop_token() {
-            token_to_html5ever(token, |token| {
-                crate::utils::trace_log!("tree builder token: {:?}", token);
-                match self.sink.process_token(token, BOGUS_LINENO) {
-                    TokenSinkResult::Continue => {}
-                    TokenSinkResult::Script(_) => {
-                        if self.next_state.is_some() {
-                            crate::utils::trace_log!("dropping state: {:?}", self.next_state);
-                        }
-                        self.next_state = Some(State::Data);
-                        // TODO: suspend tokenizer for script
-                    }
-                    TokenSinkResult::Plaintext => {
-                        if self.next_state.is_some() {
-                            crate::utils::trace_log!("dropping state: {:?}", self.next_state);
-                        }
-                        self.next_state = Some(State::PlainText);
-                    }
-                    TokenSinkResult::RawData(RawKind::Rcdata) => {
-                        if self.next_state.is_some() {
-                            crate::utils::trace_log!("dropping state: {:?}", self.next_state);
-                        }
-                        self.next_state = Some(State::RcData);
-                    }
-                    TokenSinkResult::RawData(RawKind::Rawtext) => {
-                        if self.next_state.is_some() {
-                            crate::utils::trace_log!("dropping state: {:?}", self.next_state);
-                        }
-                        self.next_state = Some(State::RawText);
-                    }
-                    TokenSinkResult::RawData(RawKind::ScriptData) => {
-                        if self.next_state.is_some() {
-                            crate::utils::trace_log!("dropping state: {:?}", self.next_state);
-                        }
-                        self.next_state = Some(State::ScriptData);
-                    }
-                    TokenSinkResult::RawData(RawKind::ScriptDataEscaped(_)) => {
-                        todo!()
-                    }
-                }
-            });
+            emitter_inner: CallbackEmitter::new(OurCallback {
+                sink,
+                current_start_tag: None,
+                next_state: None,
+            }),
         }
     }
 }
@@ -86,16 +169,13 @@ impl<'a, S: TokenSink> Emitter for Html5everEmitter<'a, S> {
 
     fn emit_eof(&mut self) {
         self.emitter_inner.emit_eof();
-        self.pop_token_inner();
-        let _ignored = self
-            .sink
-            .process_token(Html5everToken::EOFToken, BOGUS_LINENO);
-        self.sink.end();
+        let sink = &mut self.emitter_inner.callback_mut().sink;
+        let _ignored = sink.process_token(Html5everToken::EOFToken, BOGUS_LINENO);
+        sink.end();
     }
 
     fn emit_error(&mut self, error: Error) {
         self.emitter_inner.emit_error(error);
-        self.pop_token_inner();
     }
 
     fn pop_token(&mut self) -> Option<Infallible> {
@@ -104,38 +184,31 @@ impl<'a, S: TokenSink> Emitter for Html5everEmitter<'a, S> {
 
     fn emit_string(&mut self, c: &[u8]) {
         self.emitter_inner.emit_string(c);
-        self.pop_token_inner();
     }
 
     fn init_start_tag(&mut self) {
         self.emitter_inner.init_start_tag();
-        self.pop_token_inner();
     }
 
     fn init_end_tag(&mut self) {
         self.emitter_inner.init_end_tag();
-        self.pop_token_inner();
     }
 
     fn init_comment(&mut self) {
         self.emitter_inner.init_comment();
-        self.pop_token_inner();
     }
 
     fn emit_current_tag(&mut self) -> Option<State> {
         assert!(self.emitter_inner.emit_current_tag().is_none());
-        self.pop_token_inner();
-        self.next_state.take()
+        self.emitter_inner.callback_mut().next_state.take()
     }
 
     fn emit_current_comment(&mut self) {
         self.emitter_inner.emit_current_comment();
-        self.pop_token_inner();
     }
 
     fn emit_current_doctype(&mut self) {
         self.emitter_inner.emit_current_doctype();
-        self.pop_token_inner();
     }
 
     fn set_self_closing(&mut self) {
@@ -160,7 +233,6 @@ impl<'a, S: TokenSink> Emitter for Html5everEmitter<'a, S> {
 
     fn init_doctype(&mut self) {
         self.emitter_inner.init_doctype();
-        self.pop_token_inner();
     }
 
     fn init_attribute(&mut self) {
@@ -196,63 +268,9 @@ impl<'a, S: TokenSink> Emitter for Html5everEmitter<'a, S> {
     }
 
     fn adjusted_current_node_present_but_not_in_html_namespace(&mut self) -> bool {
-        self.sink
+        self.emitter_inner
+            .callback_mut()
+            .sink
             .adjusted_current_node_present_but_not_in_html_namespace()
     }
 }
-
-fn token_to_html5ever(token: Token, mut foreach_fn: impl FnMut(Html5everToken)) {
-    match token {
-        Token::StartTag(tag) => foreach_fn(Html5everToken::TagToken(Tag {
-            kind: TagKind::StartTag,
-            name: String::from_utf8_lossy(&tag.name).into_owned().into(),
-            self_closing: tag.self_closing,
-            attrs: tag
-                .attributes
-                .into_iter()
-                .map(|(key, value)| Attribute {
-                    name: QualName::new(
-                        None,
-                        Default::default(),
-                        String::from_utf8_lossy(&key).into_owned().into(),
-                    ),
-                    value: String::from_utf8_lossy(&value).into_owned().into(),
-                })
-                .collect(),
-        })),
-        Token::EndTag(tag) => foreach_fn(Html5everToken::TagToken(Tag {
-            kind: TagKind::EndTag,
-            name: String::from_utf8_lossy(&tag.name).into_owned().into(),
-            self_closing: false,
-            attrs: Vec::new(),
-        })),
-        Token::String(s) => {
-            let s = String::from_utf8_lossy(&s);
-            let mut first = true;
-            for part in s.split('\0') {
-                if !first {
-                    foreach_fn(Html5everToken::NullCharacterToken);
-                }
-
-                first = false;
-                foreach_fn(Html5everToken::CharacterTokens(part.to_owned().into()));
-            }
-        }
-        Token::Comment(c) => foreach_fn(Html5everToken::CommentToken(
-            String::from_utf8_lossy(&c).into_owned().into(),
-        )),
-        Token::Doctype(doctype) => foreach_fn(Html5everToken::DoctypeToken(Doctype {
-            name: Some(&*doctype.name)
-                .filter(|x| !x.is_empty())
-                .map(|x| String::from_utf8_lossy(x).into_owned().into()),
-            public_id: doctype
-                .public_identifier
-                .map(|x| String::from_utf8_lossy(&x).into_owned().into()),
-            system_id: doctype
-                .system_identifier
-                .map(|x| String::from_utf8_lossy(&x).into_owned().into()),
-            force_quirks: doctype.force_quirks,
-        })),
-        Token::Error(err) => foreach_fn(Html5everToken::ParseError(err.as_str().into())),
-    }
-}
diff --git a/src/htmlstring.rs b/src/htmlstring.rs
new file mode 100644
index 0000000..c789343
--- /dev/null
+++ b/src/htmlstring.rs
@@ -0,0 +1,130 @@
+use std::borrow::{Borrow, BorrowMut};
+use std::fmt::{Debug, Formatter};
+use std::ops::{Deref, DerefMut};
+
+/// A wrapper around a bytestring.
+///
+/// This newtype only exists to provide a nicer `Debug` impl
+#[derive(Clone, Default, Eq, PartialEq, Ord, PartialOrd, Hash)]
+pub struct HtmlString(pub Vec<u8>);
+
+impl Deref for HtmlString {
+    type Target = Vec<u8>;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+impl DerefMut for HtmlString {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        &mut self.0
+    }
+}
+
+impl Debug for HtmlString {
+    fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), std::fmt::Error> {
+        write!(f, "b\"")?;
+        for &byte in &self.0 {
+            for ch in std::ascii::escape_default(byte) {
+                write!(f, "{}", ch as char)?;
+            }
+        }
+
+        write!(f, "\"")
+    }
+}
+
+impl Borrow<[u8]> for HtmlString {
+    fn borrow(&self) -> &[u8] {
+        &self.0
+    }
+}
+
+impl BorrowMut<[u8]> for HtmlString {
+    fn borrow_mut(&mut self) -> &mut [u8] {
+        &mut self.0
+    }
+}
+
+impl AsRef<[u8]> for HtmlString {
+    fn as_ref(&self) -> &[u8] {
+        &self.0
+    }
+}
+
+impl<const N: usize> PartialEq<&[u8; N]> for HtmlString {
+    fn eq(&self, other: &&[u8; N]) -> bool {
+        self.0 == *other
+    }
+}
+
+impl<const N: usize> PartialEq<HtmlString> for &[u8; N] {
+    fn eq(&self, other: &HtmlString) -> bool {
+        other.0 == *self
+    }
+}
+
+impl PartialEq<&[u8]> for HtmlString {
+    fn eq(&self, other: &&[u8]) -> bool {
+        self.0 == *other
+    }
+}
+
+impl PartialEq<HtmlString> for &[u8] {
+    fn eq(&self, other: &HtmlString) -> bool {
+        *self == other.0
+    }
+}
+
+impl PartialEq<Vec<u8>> for HtmlString {
+    fn eq(&self, other: &Vec<u8>) -> bool {
+        self.0 == *other
+    }
+}
+
+impl PartialEq<HtmlString> for Vec<u8> {
+    fn eq(&self, other: &HtmlString) -> bool {
+        *self == other.0
+    }
+}
+
+#[test]
+fn test_eq_html_str_and_byte_literal() {
+    assert!(HtmlString(b"hello world".to_vec()) == b"hello world");
+}
+
+#[test]
+fn test_eq_byte_literal_and_html_str() {
+    assert!(b"hello world" == HtmlString(b"hello world".to_vec()));
+}
+
+#[test]
+fn test_eq_html_str_and_byte_slice() {
+    assert!(HtmlString(b"hello world".to_vec()) == b"hello world".as_slice());
+}
+
+#[test]
+fn test_eq_byte_slice_and_html_str() {
+    assert!(b"hello world".as_slice() == HtmlString(b"hello world".to_vec()));
+}
+
+#[test]
+fn test_borrowing() {
+    use crate::StartTag;
+    // demonstrate a usecase for Borrow/BorrowMut
+    let tag = StartTag::default();
+    assert!(!tag.attributes.contains_key(b"href".as_slice()));
+}
+
+impl From<Vec<u8>> for HtmlString {
+    fn from(vec: Vec<u8>) -> HtmlString {
+        HtmlString(vec)
+    }
+}
+
+impl From<HtmlString> for Vec<u8> {
+    fn from(other: HtmlString) -> Vec<u8> {
+        other.0
+    }
+}
diff --git a/src/lib.rs b/src/lib.rs
index 340fcbf..d1a1421 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -36,12 +36,15 @@ macro_rules! blob_url_prefix {
 use blob_url_prefix;
 
 mod arrayvec;
+pub mod callbacks;
 mod char_validator;
+mod default_emitter;
 mod emitter;
 mod entities;
 mod error;
 #[cfg(feature = "html5ever")]
 mod html5ever_emitter;
+mod htmlstring;
 mod machine;
 mod machine_helper;
 mod read_helper;
@@ -54,10 +57,10 @@ mod utils;
 #[doc(hidden)]
 pub mod testutils;
 
-pub use emitter::{
-    naive_next_state, DefaultEmitter, Doctype, Emitter, EndTag, HtmlString, StartTag, Token,
-};
+pub use default_emitter::{DefaultEmitter, Doctype, EndTag, StartTag, Token};
+pub use emitter::{naive_next_state, Emitter};
 pub use error::Error;
+pub use htmlstring::HtmlString;
 pub use reader::{IoReader, Readable, Reader, StringReader};
 pub use state::State;
 pub use tokenizer::{InfallibleTokenizer, Tokenizer};
diff --git a/src/tokenizer.rs b/src/tokenizer.rs
index 95d60f5..0d1908a 100644
--- a/src/tokenizer.rs
+++ b/src/tokenizer.rs
@@ -62,13 +62,32 @@ impl<R: Reader, E: Emitter<Token = Infallible>> Tokenizer<R, E> {
     /// Some emitters don't ever produce any tokens and instead have other side effects. In those
     /// cases, you will find yourself writing code like this to handle errors:
     ///
-    /// ```norun
-    /// for _ in tokenizer {
-    ///     result.unwrap();
-    /// }
     /// ```
+    /// use std::convert::Infallible;
     ///
-    /// This is a bit silly, so instead you can use `tokenizer.finish()`.
+    /// use html5gum::Tokenizer;
+    /// use html5gum::callbacks::{CallbackEvent, CallbackEmitter};
+    ///
+    /// let emitter = CallbackEmitter::new(move |event: CallbackEvent<'_>| -> Option<Infallible> {
+    ///     if let CallbackEvent::String { value } = event {
+    ///         println!("{}", String::from_utf8_lossy(value));
+    ///     }
+    ///
+    ///     // We may choose to return any Option<T> (such as errors, or our own tokens), but since
+    ///     // we do all the real work in the callback itself, we choose to use Option<Infallible>.
+    ///     None
+    /// });
+    ///
+    /// let tokenizer = Tokenizer::new_with_emitter("hello <div><div><div> world!", emitter);
+    ///
+    /// // this is a bit silly
+    /// // for _ in tokenizer {
+    /// //     result.unwrap();
+    /// // }
+    ///
+    /// // much better:
+    /// tokenizer.finish();
+    /// ```
     pub fn finish(self) -> Result<(), R::Error> {
         for result in self {
             result?;
diff --git a/tests/html5lib_tree_builder.rs b/tests/html5lib_tree_builder.rs
index eadc3ae..e5d1c3d 100644
--- a/tests/html5lib_tree_builder.rs
+++ b/tests/html5lib_tree_builder.rs
@@ -175,9 +175,7 @@ fn build_test(testcase: Testcase, fname: &str, i: usize, scripting: bool) -> Tri
                 tokenizer.set_state(state);
             }
 
-            for result in tokenizer {
-                result.unwrap();
-            }
+            tokenizer.finish().unwrap();
 
             let rcdom = tree_builder.sink;
             let mut actual = String::new();
@@ -195,7 +193,7 @@ fn build_test(testcase: Testcase, fname: &str, i: usize, scripting: bool) -> Tri
             }
 
             let expected = testcase.document.unwrap();
-            assert_eq!(expected, actual);
+            assert_eq!(actual, expected);
         })
     })
 }