Skip to content

YC (S26) | Open Computer History | Continuously record your company computer work, map your workflows, help you find work worth automating, and power your agents' context

Cache a11y ignored URL patterns (#4050)

Louis Beaumont0ca5bae9ae88
diff --git a/crates/screenpipe-a11y/src/tree/mod.rs b/crates/screenpipe-a11y/src/tree/mod.rs
index de782e189..c6067a33d 100644
--- a/crates/screenpipe-a11y/src/tree/mod.rs
+++ b/crates/screenpipe-a11y/src/tree/mod.rs
@@ -593,10 +593,7 @@ pub fn create_tree_walker(config: TreeWalkerConfig) -> Box<dyn TreeWalkerPlatfor
if ignored_urls.is_empty() {
walker
} else {
- Box::new(UrlFilteredWalker {
- inner: walker,
- ignored_urls,
- })
+ Box::new(UrlFilteredWalker::new(walker, ignored_urls))
}
}
@@ -605,7 +602,16 @@ pub fn create_tree_walker(config: TreeWalkerConfig) -> Box<dyn TreeWalkerPlatfor
/// platform walker) applies the filter uniformly across macOS/Windows/Linux.
struct UrlFilteredWalker {
inner: Box<dyn TreeWalkerPlatform>,
- ignored_urls: Vec<String>,
+ ignored_url_patterns: Vec<String>,
+}
+
+impl UrlFilteredWalker {
+ fn new(inner: Box<dyn TreeWalkerPlatform>, ignored_urls: Vec<String>) -> Self {
+ Self {
+ inner,
+ ignored_url_patterns: crate::url_filter::normalize_blocked_patterns(&ignored_urls),
+ }
+ }
}
impl TreeWalkerPlatform for UrlFilteredWalker {
@@ -613,7 +619,10 @@ impl TreeWalkerPlatform for UrlFilteredWalker {
let result = self.inner.walk_focused_window()?;
if let TreeWalkResult::Found(ref snapshot) = result {
if let Some(ref url) = snapshot.browser_url {
- if crate::url_filter::is_url_blocked(url, &self.ignored_urls) {
+ if crate::url_filter::is_url_blocked_by_normalized_patterns(
+ url,
+ &self.ignored_url_patterns,
+ ) {
return Ok(TreeWalkResult::Skipped(SkipReason::BlockedUrl));
}
}
@@ -665,10 +674,10 @@ mod tests {
#[test]
fn test_url_filtered_walker_blocks_matching_url() {
- let walker = UrlFilteredWalker {
- inner: Box::new(FakeWalker(Some("https://www.chase.com/login".into()))),
- ignored_urls: vec!["chase.com".into()],
- };
+ let walker = UrlFilteredWalker::new(
+ Box::new(FakeWalker(Some("https://www.chase.com/login".into()))),
+ vec!["chase.com".into()],
+ );
assert!(matches!(
walker.walk_focused_window().unwrap(),
TreeWalkResult::Skipped(SkipReason::BlockedUrl)
@@ -677,10 +686,10 @@ mod tests {
#[test]
fn test_url_filtered_walker_passes_other_urls() {
- let walker = UrlFilteredWalker {
- inner: Box::new(FakeWalker(Some("https://example.com".into()))),
- ignored_urls: vec!["chase.com".into()],
- };
+ let walker = UrlFilteredWalker::new(
+ Box::new(FakeWalker(Some("https://example.com".into()))),
+ vec!["chase.com".into()],
+ );
assert!(matches!(
walker.walk_focused_window().unwrap(),
TreeWalkResult::Found(_)
@@ -689,10 +698,7 @@ mod tests {
#[test]
fn test_url_filtered_walker_passes_non_browser_windows() {
- let walker = UrlFilteredWalker {
- inner: Box::new(FakeWalker(None)),
- ignored_urls: vec!["chase.com".into()],
- };
+ let walker = UrlFilteredWalker::new(Box::new(FakeWalker(None)), vec!["chase.com".into()]);
assert!(matches!(
walker.walk_focused_window().unwrap(),
TreeWalkResult::Found(_)
diff --git a/crates/screenpipe-a11y/src/url_filter.rs b/crates/screenpipe-a11y/src/url_filter.rs
index b58ee732b..c881b6fed 100644
--- a/crates/screenpipe-a11y/src/url_filter.rs
+++ b/crates/screenpipe-a11y/src/url_filter.rs
@@ -3,6 +3,7 @@
//! Lives here because the dependency direction is screen → a11y: an ignored
//! URL must produce neither frames nor accessibility snapshots.
+use std::borrow::Cow;
use url::Url;
/// Check if a URL should be filtered out for privacy.
@@ -21,28 +22,49 @@ pub fn is_url_blocked(url: &str, blocked_patterns: &[String]) -> bool {
return false;
}
+ let normalized_patterns = normalize_blocked_patterns(blocked_patterns);
+ is_url_blocked_by_normalized_patterns(url, &normalized_patterns)
+}
+
+/// Lowercase ignored URL patterns once before entering repeated match loops.
+pub(crate) fn normalize_blocked_patterns(blocked_patterns: &[String]) -> Vec<String> {
+ blocked_patterns
+ .iter()
+ .map(|blocked| blocked.to_lowercase())
+ .collect()
+}
+
+/// Check if a URL should be filtered using already-lowercased patterns.
+pub(crate) fn is_url_blocked_by_normalized_patterns(
+ url: &str,
+ blocked_patterns_lower: &[String],
+) -> bool {
+ if blocked_patterns_lower.is_empty() {
+ return false;
+ }
+
// Normalize so bare hosts ("wellsfargo.com") parse too.
let url_to_parse = if !url.starts_with("http://") && !url.starts_with("https://") {
- format!("https://{}", url)
+ Cow::Owned(format!("https://{}", url))
} else {
- url.to_string()
+ Cow::Borrowed(url)
};
- if let Ok(parsed) = Url::parse(&url_to_parse) {
+ if let Ok(parsed) = Url::parse(url_to_parse.as_ref()) {
if let Some(host) = parsed.host_str() {
let host_lower = host.to_lowercase();
- return blocked_patterns
+ return blocked_patterns_lower
.iter()
- .any(|blocked| host_matches_pattern(&host_lower, &blocked.to_lowercase()));
+ .any(|blocked| host_matches_pattern(&host_lower, blocked));
}
}
// Fallback to simple contains check if URL parsing fails.
// Less precise, but ensures we don't miss obvious matches.
let url_lower = url.to_lowercase();
- blocked_patterns
+ blocked_patterns_lower
.iter()
- .any(|blocked| url_lower.contains(&blocked.to_lowercase()))
+ .any(|blocked| url_lower.contains(blocked))
}
/// Domain-boundary match of one lowercased host against one lowercased pattern.
@@ -53,7 +75,10 @@ fn host_matches_pattern(host_lower: &str, blocked: &str) -> bool {
}
// Subdomain match: host ends with ".blocked"
- if host_lower.ends_with(&format!(".{}", blocked)) {
+ if host_lower.len() > blocked.len()
+ && host_lower.ends_with(blocked)
+ && host_lower.as_bytes()[host_lower.len() - blocked.len() - 1] == b'.'
+ {
return true;
}
@@ -137,4 +162,23 @@ mod tests {
assert!(is_url_blocked("https://www.bankofamerica.com", &b));
assert!(!is_url_blocked("https://google.com", &b));
}
+
+ #[test]
+ fn test_normalized_patterns_match_public_api() {
+ let b = blocked(&["WellsFargo.com", "CHASE"]);
+ let normalized = normalize_blocked_patterns(&b);
+
+ assert!(is_url_blocked_by_normalized_patterns(
+ "https://www.wellsfargo.com/login",
+ &normalized
+ ));
+ assert!(is_url_blocked_by_normalized_patterns(
+ "https://online.chase.co.uk",
+ &normalized
+ ));
+ assert!(!is_url_blocked_by_normalized_patterns(
+ "https://purchase.com",
+ &normalized
+ ));
+ }
}