refactor: replace kuchikiki with dom_query

.changes/supersede-kuchikiki.md

@@ -0,0 +1,6 @@
+---
+"tauri-utils": minor:deps
+---
+
+Add new `html-manipulation-2` and `build-2` feature flags that use `dom_query` instead of `kuchikiki` for HTML parsing / manipulation.
+This allows downstream users to remove `kuchikiki` and its dependencies from their dependency tree.

crates/tauri-build/Cargo.toml

@@ -28,7 +28,7 @@ anyhow = "1"
 quote = { version = "1", optional = true }
 tauri-codegen = { version = "2.5.5", path = "../tauri-codegen", optional = true }
 tauri-utils = { version = "2.8.3", path = "../tauri-utils", features = [
-  "build",
+  "build-2",
   "resources",
 ] }
 cargo_toml = "0.22"

crates/tauri-cli/Cargo.toml

@@ -63,7 +63,7 @@ tauri-utils = { version = "2.8.3", path = "../tauri-utils", features = [
   "schema",
   "config-json5",
   "config-toml",
-  "html-manipulation",
+  "html-manipulation-2",
 ] }
 toml = "1"
 jsonschema = { version = "0.33", default-features = false }
@@ -89,8 +89,6 @@ env_logger = "0.11"
 icns = { package = "tauri-icns", version = "0.1" }
 image = { version = "0.25", default-features = false, features = ["ico"] }
 axum = { version = "0.8", features = ["ws"] }
-html5ever = "0.29"
-kuchiki = { package = "kuchikiki", version = "=0.8.8-speedreader" }
 tokio = { version = "1", features = ["macros", "sync"] }
 common-path = "1"
 serde-value = "0.7"

crates/tauri-cli/src/dev/builtin_dev_server.rs

@@ -7,8 +7,6 @@ use axum::{
   http::{header, StatusCode, Uri},
   response::{IntoResponse, Response},
 };
-use html5ever::{namespace_url, ns, LocalName, QualName};
-use kuchiki::{traits::TendrilSink, NodeRef};
 use std::{
   net::{IpAddr, SocketAddr},
   path::{Path, PathBuf},
@@ -128,30 +126,14 @@ async fn ws_handler(ws: WebSocketUpgrade, state: State<ServerState>) -> Response
 }
 
 fn inject_address(html_bytes: Vec<u8>, address: &SocketAddr) -> Vec<u8> {
-  fn with_html_head<F: FnOnce(&NodeRef)>(document: &mut NodeRef, f: F) {
-    if let Ok(ref node) = document.select_first("head") {
-      f(node.as_node())
-    } else {
-      let node = NodeRef::new_element(
-        QualName::new(None, ns!(html), LocalName::from("head")),
-        None,
-      );
-      f(&node);
-      document.prepend(node)
-    }
-  }
+  let document = tauri_utils::html2::parse_doc(String::from_utf8_lossy(&html_bytes).into_owned());
 
-  let mut document = kuchiki::parse_html()
-    .one(String::from_utf8_lossy(&html_bytes).into_owned())
-    .document_node;
-  with_html_head(&mut document, |head| {
-    let script = RELOAD_SCRIPT.replace("{{reload_url}}", &format!("ws://{address}/__tauri_cli"));
-    let script_el = NodeRef::new_element(QualName::new(None, ns!(html), "script".into()), None);
-    script_el.append(NodeRef::new_text(script));
-    head.prepend(script_el);
-  });
+  tauri_utils::html2::append_script_to_head(
+    &document,
+    &RELOAD_SCRIPT.replace("{{reload_url}}", &format!("ws://{address}/__tauri_cli")),
+  );
 
-  tauri_utils::html::serialize_node(&document)
+  tauri_utils::html2::serialize_doc(&document)
 }
 
 fn fs_read_scoped(path: PathBuf, scope: &Path) -> crate::Result<Vec<u8>> {

crates/tauri-codegen/Cargo.toml

@@ -21,7 +21,7 @@ syn = "2"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 tauri-utils = { version = "2.8.3", path = "../tauri-utils", features = [
-  "build",
+  "build-2",
 ] }
 thiserror = "2"
 walkdir = "2"

crates/tauri-codegen/src/context.rs

@@ -25,7 +25,7 @@ use tauri_utils::{
   },
   assets::AssetKey,
   config::{Config, FrontendDist, PatternKind},
-  html::{inject_nonce_token, parse as parse_html, serialize_node as serialize_html_node, NodeRef},
+  html2::{inject_nonce_token, parse_doc, serialize_doc, Document},
   platform::Target,
   tokens::{map_lit, str_lit},
 };
@@ -44,27 +44,25 @@ pub struct ContextData {
   pub test: bool,
 }
 
-fn inject_script_hashes(document: &NodeRef, key: &AssetKey, csp_hashes: &mut CspHashes) {
-  if let Ok(inline_script_elements) = document.select("script:not(:empty)") {
-    let mut scripts = Vec::new();
-    for inline_script_el in inline_script_elements {
-      let script = inline_script_el.as_node().text_contents();
-      let mut hasher = Sha256::new();
-      hasher.update(tauri_utils::html::normalize_script_for_csp(
-        script.as_bytes(),
-      ));
-      let hash = hasher.finalize();
-      scripts.push(format!(
-        "'sha256-{}'",
-        base64::engine::general_purpose::STANDARD.encode(hash)
-      ));
-    }
-    csp_hashes
-      .inline_scripts
-      .entry(key.clone().into())
-      .or_default()
-      .append(&mut scripts);
-  }
+fn inject_script_hashes(document: &Document, key: &AssetKey, csp_hashes: &mut CspHashes) {
+  let script_elements = document.select("script:not(:empty)");
+
+  let scripts = script_elements
+    .iter()
+    .map(|element| {
+      let script = tauri_utils::html2::normalize_script_for_csp(element.text().as_bytes());
+      let script_hash = Sha256::digest(script);
+      let hash_base64 = base64::engine::general_purpose::STANDARD.encode(script_hash);
+
+      format!("'sha256-{hash_base64}'")
+    })
+    .collect::<Vec<_>>();
+
+  csp_hashes
+    .inline_scripts
+    .entry(key.clone().into())
+    .or_default()
+    .extend(scripts);
 }
 
 fn map_core_assets(
@@ -77,15 +75,15 @@ fn map_core_assets(
     if path.extension() == Some(OsStr::new("html")) {
       #[allow(clippy::collapsible_if)]
       if csp {
-        let document = parse_html(String::from_utf8_lossy(input).into_owned());
+        let document = parse_doc(String::from_utf8_lossy(input).into_owned());
 
         inject_nonce_token(&document, &dangerous_disable_asset_csp_modification);
 
         if dangerous_disable_asset_csp_modification.can_modify("script-src") {
           inject_script_hashes(&document, key, csp_hashes);
         }
 
-        *input = serialize_html_node(&document);
+        *input = serialize_doc(&document);
       }
     }
     Ok(())
@@ -108,13 +106,13 @@ fn map_isolation(
 
   move |key, path, input, csp_hashes| {
     if path.extension() == Some(OsStr::new("html")) {
-      let isolation_html = parse_html(String::from_utf8_lossy(input).into_owned());
+      let isolation_html = parse_doc(String::from_utf8_lossy(input).into_owned());
 
       // this is appended, so no need to reverse order it
-      tauri_utils::html::inject_codegen_isolation_script(&isolation_html);
+      tauri_utils::html2::inject_codegen_isolation_script(&isolation_html);
 
       // temporary workaround for windows not loading assets
-      tauri_utils::html::inline_isolation(&isolation_html, &dir);
+      tauri_utils::html2::inline_isolation(&isolation_html, &dir);
 
       inject_nonce_token(
         &isolation_html,
@@ -125,7 +123,7 @@ fn map_isolation(
 
       csp_hashes.styles.push(iframe_style_csp_hash.clone());
 
-      *input = isolation_html.to_string().as_bytes().to_vec()
+      *input = serialize_doc(&isolation_html)
     }
 
     Ok(())

crates/tauri-codegen/src/embedded_assets.rs

@@ -181,7 +181,7 @@ impl CspHashes {
         let mut hasher = Sha256::new();
         hasher.update(
           &std::fs::read(path)
-            .map(|b| tauri_utils::html::normalize_script_for_csp(&b))
+            .map(|b| tauri_utils::html2::normalize_script_for_csp(&b))
             .map_err(|error| EmbeddedAssetsError::AssetRead {
               path: path.to_path_buf(),
               error,

crates/tauri-plugin/Cargo.toml

@@ -28,7 +28,7 @@ runtime = []
 anyhow = { version = "1", optional = true }
 serde = { version = "1", optional = true }
 tauri-utils = { version = "2.8.3", default-features = false, features = [
-  "build",
+  "build-2",
 ], path = "../tauri-utils" }
 serde_json = { version = "1", optional = true }
 glob = { version = "0.3", optional = true }

crates/tauri-utils/Cargo.toml

@@ -24,6 +24,7 @@ brotli = { version = "8", optional = true, default-features = false, features =
 url = { version = "2", features = ["serde"] }
 html5ever = { version = "0.29", optional = true }
 kuchiki = { package = "kuchikiki", version = "0.8.8-speedreader", optional = true }
+dom_query = { version = "0.27", optional = true, default-features = false }
 proc-macro2 = { version = "1", optional = true }
 quote = { version = "1", optional = true }
 # Our code requires at least 0.8.21 so don't change this to 0.8
@@ -59,6 +60,7 @@ swift-rs = { version = "1", optional = true, features = ["build"] }
 getrandom = { version = "0.3", features = ["std"] }
 serial_test = "3"
 tauri = { path = "../tauri" }
+tempfile = "3.15.0"
 
 [features]
 build = [
@@ -69,6 +71,15 @@ build = [
   "swift-rs",
   "html-manipulation",
 ]
+# Same as `build` but uses `html-manipulation-2` to avoid the `kuchikiki` dependency.
+build-2 = [
+  "proc-macro2",
+  "quote",
+  "cargo_metadata",
+  "schema",
+  "swift-rs",
+  "html-manipulation-2",
+]
 compression = ["brotli"]
 schema = ["schemars"]
 isolation = ["aes-gcm", "getrandom", "serialize-to-javascript"]
@@ -77,3 +88,4 @@ config-json5 = ["json5"]
 config-toml = []
 resources = ["walkdir"]
 html-manipulation = ["dep:html5ever", "dep:kuchiki"]
+html-manipulation-2 = ["dep:dom_query"]

crates/tauri-utils/src/acl/capability.rs

@@ -322,7 +322,7 @@ impl FromStr for CapabilityFile {
   }
 }
 
-#[cfg(feature = "build")]
+#[cfg(any(feature = "build", feature = "build-2"))]
 mod build {
   use std::convert::identity;
 

crates/tauri-utils/src/acl/identifier.rs

@@ -283,7 +283,7 @@ mod tests {
   }
 }
 
-#[cfg(feature = "build")]
+#[cfg(any(feature = "build", feature = "build-2"))]
 mod build {
   use proc_macro2::TokenStream;
   use quote::{quote, ToTokens, TokenStreamExt};

crates/tauri-utils/src/acl/manifest.rs

@@ -126,7 +126,7 @@ impl Manifest {
   }
 }
 
-#[cfg(feature = "build")]
+#[cfg(any(feature = "build", feature = "build-2"))]
 mod build {
   use proc_macro2::TokenStream;
   use quote::{quote, ToTokens, TokenStreamExt};