Merge pull request #102 from jaboatman/master

cfvescovo · web-flow · commit ec91bf1ebc99 · 2023-02-28T19:38:25.000+01:00
Added feature flag `atomic` to make use of atomic `StrTendril` type.
diff --git a/Cargo.toml b/Cargo.toml
@@ -32,6 +32,7 @@ optional = true
 default = ["main"]
 deterministic = ["indexmap"]
 main = ["getopts"]
+atomic = []
 
 [[bin]]
 name = "scraper"
diff --git a/src/html/mod.rs b/src/html/mod.rs
@@ -204,4 +204,11 @@ mod tests {
             .collect();
         assert_eq!(result, vec!["element3", "element2", "element1"]);
     }
+
+    #[cfg(feature = "atomic")]
+    #[test]
+    fn html_is_send() {
+        fn send_sync<S: Send>() {}
+        send_sync::<Html>();
+    }
 }
diff --git a/src/html/tree_sink.rs b/src/html/tree_sink.rs
@@ -1,12 +1,12 @@
-use std::borrow::Cow;
-
 use super::Html;
 use crate::node::{Comment, Doctype, Element, Node, ProcessingInstruction, Text};
+use crate::tendril_util::make as make_tendril;
 use ego_tree::NodeId;
 use html5ever::tendril::StrTendril;
 use html5ever::tree_builder::{ElementFlags, NodeOrText, QuirksMode, TreeSink};
 use html5ever::Attribute;
 use html5ever::{ExpandedName, QualName};
+use std::borrow::Cow;
 
 /// Note: does not support the `<template>` element.
 impl TreeSink for Html {
@@ -74,7 +74,9 @@ impl TreeSink for Html {
     // Create a comment node.
     fn create_comment(&mut self, text: StrTendril) -> Self::Handle {
         self.tree
-            .orphan(Node::Comment(Comment { comment: text }))
+            .orphan(Node::Comment(Comment {
+                comment: make_tendril(text),
+            }))
             .id()
     }
 
@@ -85,6 +87,9 @@ impl TreeSink for Html {
         public_id: StrTendril,
         system_id: StrTendril,
     ) {
+        let name = make_tendril(name);
+        let public_id = make_tendril(public_id);
+        let system_id = make_tendril(system_id);
         let doctype = Doctype {
             name,
             public_id,
@@ -106,6 +111,7 @@ impl TreeSink for Html {
             }
 
             NodeOrText::AppendText(text) => {
+                let text = make_tendril(text);
                 let can_concat = parent
                     .last_child()
                     .map_or(false, |mut n| n.value().is_text());
@@ -148,6 +154,7 @@ impl TreeSink for Html {
                 }
 
                 NodeOrText::AppendText(text) => {
+                    let text = make_tendril(text);
                     let can_concat = sibling
                         .prev_sibling()
                         .map_or(false, |mut n| n.value().is_text());
@@ -189,7 +196,10 @@ impl TreeSink for Html {
         };
 
         for attr in attrs {
-            element.attrs.entry(attr.name).or_insert(attr.value);
+            element
+                .attrs
+                .entry(attr.name)
+                .or_insert_with(|| make_tendril(attr.value));
         }
     }
 
@@ -206,6 +216,8 @@ impl TreeSink for Html {
 
     // Create Processing Instruction.
     fn create_pi(&mut self, target: StrTendril, data: StrTendril) -> Self::Handle {
+        let target = make_tendril(target);
+        let data = make_tendril(data);
         self.tree
             .orphan(Node::ProcessingInstruction(ProcessingInstruction {
                 target,
diff --git a/src/lib.rs b/src/lib.rs
@@ -153,5 +153,31 @@ pub mod html;
 pub mod node;
 pub mod selector;
 
+#[cfg(feature = "atomic")]
+pub(crate) mod tendril_util {
+    use html5ever::tendril;
+    /// Atomic equivalent to the default `StrTendril` type.
+    pub type StrTendril = tendril::Tendril<tendril::fmt::UTF8, tendril::Atomic>;
+
+    /// Convert a standard tendril into an atomic one.
+    pub fn make(s: tendril::StrTendril) -> StrTendril {
+        s.into_send().into()
+    }
+}
+
+#[cfg(not(feature = "atomic"))]
+pub(crate) mod tendril_util {
+    use html5ever::tendril;
+    /// Primary string tendril type.
+    pub type StrTendril = tendril::StrTendril;
+
+    /// Return unaltered.
+    pub fn make(s: StrTendril) -> StrTendril {
+        s
+    }
+}
+
+pub use tendril_util::StrTendril;
+
 #[cfg(test)]
 mod test;
diff --git a/src/node.rs b/src/node.rs
@@ -6,7 +6,7 @@ use std::collections::{hash_set, HashSet};
 use std::fmt;
 use std::ops::Deref;
 
-use html5ever::tendril::StrTendril;
+use crate::StrTendril;
 use html5ever::{Attribute, LocalName, QualName};
 use selectors::attr::CaseSensitivity;
 
@@ -252,7 +252,7 @@ impl Element {
                 }
                 _ => (),
             };
-            attrs.insert(a.name, a.value);
+            attrs.insert(a.name, crate::tendril_util::make(a.value));
         }
 
         Element {

Original file line number	Diff line number	Diff line change
`@@ -204,4 +204,11 @@ mod tests {`
`204`	`204`	`.collect();`
`205`	`205`	`assert_eq!(result, vec!["element3", "element2", "element1"]);`
`206`	`206`	`}`
	`207`	`+`
	`208`	`+ #[cfg(feature = "atomic")]`
	`209`	`+ #[test]`
	`210`	`+ fn html_is_send() {`
	`211`	`+ fn send_sync<S: Send>() {}`
	`212`	`+ send_sync::<Html>();`
	`213`	`+ }`
`207`	`214`	`}`
Original file line number	Diff line number	Diff line change
`@@ -6,7 +6,7 @@ use std::collections::{hash_set, HashSet};`
`6`	`6`	`use std::fmt;`
`7`	`7`	`use std::ops::Deref;`
`8`	`8`
`9`		`-use html5ever::tendril::StrTendril;`
	`9`	`+use crate::StrTendril;`
`10`	`10`	`use html5ever::{Attribute, LocalName, QualName};`
`11`	`11`	`use selectors::attr::CaseSensitivity;`
`12`	`12`
`@@ -252,7 +252,7 @@ impl Element {`
`252`	`252`	`}`
`253`	`253`	`_ => (),`
`254`	`254`	`};`
`255`		`- attrs.insert(a.name, a.value);`
	`255`	`+ attrs.insert(a.name, crate::tendril_util::make(a.value));`
`256`	`256`	`}`
`257`	`257`
`258`	`258`	`Element {`