Optional CSS validation

Cargo.toml

@@ -37,8 +37,8 @@ base64 = "0.10"
 rmp-serde = "0.13.7"    # rmp-serde 0.14.0 breaks deserialization by changing how enums are deserialized
 hashbrown = { version = "0.6", features = ["serde"] }
 lifeguard = { version = "0.6", optional = true }
-cssparser = "0.25"
-selectors = "0.21"
+cssparser = { version = "0.25", optional = true }
+selectors = { version = "0.21", optional = true }
 psl = "0.4.1"
 
 [dev-dependencies]
@@ -81,3 +81,4 @@ full-domain-matching = [] # feature has no explicit dependencies
 metrics = []
 full-regex-handling = []
 object-pooling = ["lifeguard"]
+css-validation = ["cssparser", "selectors"]

src/cosmetic_filter_cache.rs

@@ -12,6 +12,19 @@ lazy_static! {
     static ref PUBLIC_SUFFIXES: psl::List = psl::List::new();
 }
 
+/// Contains cosmetic filter information intended to be injected into a particular hostname.
+///
+/// `hide_selectors` is a set of any CSS selector on the page that should be hidden, i.e. styled as
+/// `{ display: none !important; }`.
+///
+/// `style_selectors` is a map of CSS selectors on the page to respective non-hide style rules,
+/// i.e. any required styles other than `display: none`.
+///
+/// `exceptions` is a set of any class or id CSS selectors that should not have generic rules
+/// applied. In practice, these should be passed to `class_id_stylesheet` and not used otherwise.
+///
+/// `injected_script` is the Javascript code for any scriptlets that should be injected into the
+/// page.
 #[derive(Debug, PartialEq, Eq, Deserialize, Serialize)]
 pub struct HostnameSpecificResources {
     pub hide_selectors: HashSet<String>,

src/engine.rs

@@ -217,10 +217,18 @@ impl Engine {
 
     // Cosmetic filter functionality
 
-    pub fn class_id_stylesheet(&self, classes: &[String], ids: &[String], exceptions: HashSet<String>) -> Option<String> {
-        self.cosmetic_cache.class_id_stylesheet(classes, ids, &exceptions)
-    }
-
+    /// If any of the provided CSS classes or ids could cause a certain generic CSS hide rule
+    /// (i.e. `{ display: none !important; }`) to be required, this method will return a stylesheet
+    /// including it, providing that the corresponding rule does not have an exception.
+    ///
+    /// `exceptions` should be passed directly from `HostnameSpecificResources`.
+    pub fn class_id_stylesheet(&self, classes: &[String], ids: &[String], exceptions: &HashSet<String>) -> Option<String> {
+        self.cosmetic_cache.class_id_stylesheet(classes, ids, exceptions)
+    }
+
+    /// Returns a set of cosmetic filter resources required for a particular hostname. Once this
+    /// has been called, all CSS ids and classes on a page should be passed to
+    /// `class_id_stylesheet` to obtain any stylesheets consisting of generic rules.
     pub fn hostname_cosmetic_resources(&self, hostname: &str) -> HostnameSpecificResources {
         self.cosmetic_cache.hostname_cosmetic_resources(hostname)
     }

src/filters/cosmetic.rs

@@ -20,6 +20,7 @@ pub enum CosmeticFilterError {
     GenericScriptInject,
     GenericStyle,
     DoubleNegation,
+    EmptyRule,
 }
 
 bitflags! {
@@ -207,6 +208,10 @@ impl CosmeticFilter {
             };
 
             let mut selector = &line[suffix_start_index..];
+
+            if selector.trim().len() == 0 {
+                return Err(CosmeticFilterError::EmptyRule);
+            }
             let mut style = None;
             if line.len() - suffix_start_index > 4 && line[suffix_start_index..].starts_with("+js(") {
                 if sharp_index == 0 {
@@ -371,6 +376,18 @@ pub fn get_hostname_hashes_from_labels(hostname: &str, domain: &str) -> Vec<Hash
     get_hashes_from_labels(hostname, hostname.len(), hostname.len() - domain.len())
 }
 
+#[cfg(not(feature="css-validation"))]
+mod css_validation {
+    pub fn is_valid_css_selector(_selector: &str) -> bool {
+        true
+    }
+
+    pub fn is_valid_css_style(_style: &str) -> bool {
+        true
+    }
+}
+
+#[cfg(feature="css-validation")]
 mod css_validation {
     //! Methods for validating CSS selectors and style rules extracted from cosmetic filter rules.
     use cssparser::ParserInput;