Bump selectors and cssparser to their current stable versions and adjust our usage as necessary.

Author: adamreichold

Cargo.lock

@@ -13,13 +13,14 @@ repository = "https://github.com/causal-agent/scraper"
 readme = "README.md"
 
 [dependencies]
-cssparser = "0.31.0"
+ahash = "0.8.0"
+cssparser = "0.34.0"
 ego-tree = "0.9.0"
-html5ever = "0.27"
-selectors = "0.25.0"
-tendril = "0.4.3"
-ahash = "0.8"
+html5ever = "0.27.0"
 indexmap = { version = "2.6.0", optional = true }
+precomputed-hash = "0.1.1"
+selectors = "0.26.0"
+tendril = "0.4.3"
 
 [dependencies.getopts]
 version = "0.2.21"

scraper/Cargo.toml

@@ -1,6 +1,7 @@
 use html5ever::Namespace;
 use selectors::{
     attr::{AttrSelectorOperation, CaseSensitivity, NamespaceConstraint},
+    bloom::BloomFilter,
     matching, Element, OpaqueElement,
 };
 
@@ -122,6 +123,10 @@ impl<'a> Element for ElementRef<'a> {
         self.value().has_class(&name.0, case_sensitivity)
     }
 
+    fn has_custom_state(&self, _name: &CssLocalName) -> bool {
+        false
+    }
+
     fn is_empty(&self) -> bool {
         !self
             .children()
@@ -134,6 +139,11 @@ impl<'a> Element for ElementRef<'a> {
     }
 
     fn apply_selector_flags(&self, _flags: matching::ElementSelectorFlags) {}
+
+    fn add_element_unique_hashes(&self, _filter: &mut BloomFilter) -> bool {
+        // FIXME: Do we want to add `self.node.id()` here?
+        false
+    }
 }
 
 #[cfg(test)]

scraper/src/element_ref/element.rs

@@ -7,7 +7,7 @@ use std::ops::Deref;
 use ego_tree::iter::{Edge, Traverse};
 use ego_tree::NodeRef;
 use html5ever::serialize::{serialize, SerializeOpts, TraversalScope};
-use selectors::NthIndexCache;
+use selectors::matching::SelectorCaches;
 
 use crate::node::Element;
 use crate::{Node, Selector};
@@ -49,7 +49,7 @@ impl<'a> ElementRef<'a> {
             scope: *self,
             inner,
             selector,
-            nth_index_cache: NthIndexCache::default(),
+            caches: Default::default(),
         }
     }
 
@@ -135,7 +135,7 @@ pub struct Select<'a, 'b> {
     scope: ElementRef<'a>,
     inner: Traverse<'a, Node>,
     selector: &'b Selector,
-    nth_index_cache: NthIndexCache,
+    caches: SelectorCaches,
 }
 
 impl Debug for Select<'_, '_> {
@@ -144,7 +144,7 @@ impl Debug for Select<'_, '_> {
             .field("scope", &self.scope)
             .field("inner", &self.inner)
             .field("selector", &self.selector)
-            .field("nth_index_cache", &"..")
+            .field("caches", &"..")
             .finish()
     }
 }
@@ -155,7 +155,7 @@ impl Clone for Select<'_, '_> {
             scope: self.scope,
             inner: self.inner.clone(),
             selector: self.selector,
-            nth_index_cache: NthIndexCache::default(),
+            caches: Default::default(),
         }
     }
 }
@@ -170,7 +170,7 @@ impl<'a, 'b> Iterator for Select<'a, 'b> {
                     if self.selector.matches_with_scope_and_cache(
                         &element,
                         Some(self.scope),
-                        &mut self.nth_index_cache,
+                        &mut self.caches,
                     ) {
                         return Some(element);
                     }

scraper/src/element_ref/mod.rs

@@ -10,7 +10,7 @@ use ego_tree::Tree;
 use html5ever::serialize::SerializeOpts;
 use html5ever::tree_builder::QuirksMode;
 use html5ever::{driver, serialize, QualName};
-use selectors::NthIndexCache;
+use selectors::matching::SelectorCaches;
 use tendril::TendrilSink;
 
 use crate::selector::Selector;
@@ -95,7 +95,7 @@ impl Html {
         Select {
             inner: self.tree.nodes(),
             selector,
-            nth_index_cache: NthIndexCache::default(),
+            caches: Default::default(),
         }
     }
 
@@ -127,15 +127,15 @@ impl Html {
 pub struct Select<'a, 'b> {
     inner: Nodes<'a, Node>,
     selector: &'b Selector,
-    nth_index_cache: NthIndexCache,
+    caches: SelectorCaches,
 }
 
 impl fmt::Debug for Select<'_, '_> {
     fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
         fmt.debug_struct("Select")
             .field("inner", &self.inner)
             .field("selector", &self.selector)
-            .field("nth_index_cache", &"..")
+            .field("caches", &"..")
             .finish()
     }
 }
@@ -145,7 +145,7 @@ impl Clone for Select<'_, '_> {
         Self {
             inner: self.inner.clone(),
             selector: self.selector,
-            nth_index_cache: NthIndexCache::default(),
+            caches: Default::default(),
         }
     }
 }
@@ -157,11 +157,9 @@ impl<'a, 'b> Iterator for Select<'a, 'b> {
         for node in self.inner.by_ref() {
             if let Some(element) = ElementRef::wrap(node) {
                 if element.parent().is_some()
-                    && self.selector.matches_with_scope_and_cache(
-                        &element,
-                        None,
-                        &mut self.nth_index_cache,
-                    )
+                    && self
+                        .selector
+                        .matches_with_scope_and_cache(&element, None, &mut self.caches)
                 {
                     return Some(element);
                 }
@@ -182,11 +180,9 @@ impl<'a, 'b> DoubleEndedIterator for Select<'a, 'b> {
         for node in self.inner.by_ref().rev() {
             if let Some(element) = ElementRef::wrap(node) {
                 if element.parent().is_some()
-                    && self.selector.matches_with_scope_and_cache(
-                        &element,
-                        None,
-                        &mut self.nth_index_cache,
-                    )
+                    && self
+                        .selector
+                        .matches_with_scope_and_cache(&element, None, &mut self.caches)
                 {
                     return Some(element);
                 }

scraper/src/html/mod.rs

@@ -5,10 +5,10 @@ use std::fmt;
 
 pub use cssparser::ToCss;
 use html5ever::{LocalName, Namespace};
+use precomputed_hash::PrecomputedHash;
 use selectors::{
     matching,
     parser::{self, ParseRelative, SelectorList, SelectorParseErrorKind},
-    NthIndexCache,
 };
 
 use crate::error::SelectorErrorKind;
@@ -43,7 +43,7 @@ impl Selector {
     /// The optional `scope` argument is used to specify which element has `:scope` pseudo-class.
     /// When it is `None`, `:scope` will match the root element.
     pub fn matches_with_scope(&self, element: &ElementRef, scope: Option<ElementRef>) -> bool {
-        self.matches_with_scope_and_cache(element, scope, &mut NthIndexCache::default())
+        self.matches_with_scope_and_cache(element, scope, &mut Default::default())
     }
 
     // The `nth_index_cache` must not be used after `self` is dropped
@@ -53,19 +53,19 @@ impl Selector {
         &self,
         element: &ElementRef,
         scope: Option<ElementRef>,
-        nth_index_cache: &mut NthIndexCache,
+        caches: &mut matching::SelectorCaches,
     ) -> bool {
         let mut context = matching::MatchingContext::new(
             matching::MatchingMode::Normal,
             None,
-            nth_index_cache,
+            caches,
             matching::QuirksMode::NoQuirks,
             matching::NeedsSelectorFlags::No,
-            matching::IgnoreNthChildForInvalidation::No,
+            matching::MatchingForInvalidation::No,
         );
         context.scope_element = scope.map(|x| selectors::Element::opaque(&x));
         self.selectors
-            .0
+            .slice()
             .iter()
             .any(|s| matching::matches_selector(s, 0, None, element, &mut context))
     }
@@ -160,6 +160,12 @@ impl ToCss for CssLocalName {
     }
 }
 
+impl PrecomputedHash for CssLocalName {
+    fn precomputed_hash(&self) -> u32 {
+        self.0.precomputed_hash()
+    }
+}
+
 /// Non Tree-Structural Pseudo-Class.
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum NonTSPseudoClass {}