zendriver_interception/
url_pattern.rs

1//! CDP-style URL pattern matching.
2//!
3//! Compiles a CDP wildcard pattern (`*` matches any sequence, `?` matches a
4//! single character) into a [`regex::Regex`]. All other regex metacharacters
5//! in the input are escaped so the user-facing surface stays the simple CDP
6//! syntax used by `Fetch.RequestPattern.urlPattern`.
7
8use regex::Regex;
9
10use crate::error::InterceptionError;
11
12/// A compiled CDP URL pattern.
13///
14/// Constructed via [`UrlPattern::new`] from a string using CDP wildcard
15/// syntax. Use [`UrlPattern::matches`] to test URLs against it, and
16/// [`UrlPattern::pattern_str`] to recover the original pattern (e.g. when
17/// forwarding to `Fetch.enable { patterns }`).
18#[derive(Debug, Clone)]
19pub struct UrlPattern {
20    pattern: String,
21    regex: Regex,
22}
23
24impl UrlPattern {
25    /// Compile a CDP-style URL pattern.
26    ///
27    /// `*` matches any sequence of characters (including empty), `?` matches
28    /// exactly one character, and every other regex metacharacter is escaped
29    /// so the user sees the simple wildcard surface.
30    ///
31    /// Returns [`InterceptionError::InvalidPattern`] if the resulting regex
32    /// fails to compile.
33    #[allow(clippy::result_large_err)] // InterceptionError shape fixed by error.rs; boxing is a cross-cutting decision tracked separately.
34    pub fn new(pattern: impl Into<String>) -> Result<Self, InterceptionError> {
35        let pattern = pattern.into();
36        let regex_src = compile_to_regex(&pattern);
37        let regex = Regex::new(&regex_src)
38            .map_err(|e| InterceptionError::InvalidPattern(format!("{pattern}: {e}")))?;
39        Ok(Self { pattern, regex })
40    }
41
42    /// Test whether `url` matches this pattern.
43    pub fn matches(&self, url: &str) -> bool {
44        self.regex.is_match(url)
45    }
46
47    /// The original pattern string passed to [`UrlPattern::new`].
48    ///
49    /// Useful when forwarding the pattern to CDP via
50    /// `Fetch.enable { patterns: [{ urlPattern: ... }] }`.
51    pub fn pattern_str(&self) -> &str {
52        &self.pattern
53    }
54}
55
56/// Convert a CDP wildcard pattern into an anchored regex source.
57///
58/// `*` → `.*`, `?` → `.`, all other regex metacharacters are escaped via
59/// [`regex::escape`]. The result is anchored at both ends so `matches`
60/// behaves as a full-URL match.
61fn compile_to_regex(pattern: &str) -> String {
62    let mut out = String::with_capacity(pattern.len() + 4);
63    out.push('^');
64    for ch in pattern.chars() {
65        match ch {
66            '*' => out.push_str(".*"),
67            '?' => out.push('.'),
68            other => out.push_str(&regex::escape(&other.to_string())),
69        }
70    }
71    out.push('$');
72    out
73}
74
75#[cfg(test)]
76#[allow(clippy::panic, clippy::unwrap_used)]
77mod tests {
78    use super::*;
79
80    #[test]
81    fn wildcard_star_matches_all() {
82        let p = UrlPattern::new("*").unwrap();
83        assert!(p.matches("https://example.com/"));
84        assert!(p.matches("http://foo.bar/baz?qux=1"));
85        assert!(p.matches(""));
86        assert_eq!(p.pattern_str(), "*");
87    }
88
89    #[test]
90    fn subdomain_wildcard_matches() {
91        let p = UrlPattern::new("*.example.com/*").unwrap();
92        assert!(p.matches("https://api.example.com/foo"));
93        assert!(p.matches("ws://cdn.example.com/socket"));
94        assert!(!p.matches("https://example.org/foo"));
95    }
96
97    #[test]
98    fn invalid_pattern_errors() {
99        // Our wildcard expansion escapes every regex metachar in the input
100        // so syntax-level failures (`[`, `\`, etc.) cannot happen via user
101        // input. The deterministic failure path is `regex`'s default
102        // 10 MiB compiled-size limit: enough `*` wildcards expand into
103        // enough `.*` repetition to trip it.
104        let huge = "*".repeat(50_000);
105        let err = UrlPattern::new(huge).expect_err("expected size-limit failure");
106        match err {
107            InterceptionError::InvalidPattern(msg) => {
108                assert!(
109                    msg.contains("exceeds size limit") || msg.contains("size"),
110                    "expected size-limit message, got: {msg}",
111                );
112            }
113            other => panic!("unexpected error variant: {other:?}"),
114        }
115    }
116}
zendriver_interception/url_pattern.rs

zendriver_interception/
url_pattern.rs
