feat(network): add networking blocking

Author: j-mendez

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "spider_firewall"
-version = "2.33.17"
+version = "2.33.18"
 authors = [
     "j-mendez <jeff@spider.cloud>"
 ]

src/lib.rs

@@ -12,9 +12,11 @@ pub mod firewall {
     pub static GLOBAL_TRACKING_WEBSITES: OnceLock<&phf::Set<&'static str>> = OnceLock::new();
     /// Gambling websites list.
     pub static GLOBAL_GAMBLING_WEBSITES: OnceLock<&phf::Set<&'static str>> = OnceLock::new();
+    /// Networking XHR|Fetch general list.
+    pub static GLOBAL_NETWORKING_WEBSITES: OnceLock<&phf::Set<&'static str>> = OnceLock::new();
 
     /// Defines a set of websites under a specified category. Available categories
-    /// include "ads", "tracking", "gambling", and a default category for any other strings.
+    /// include "ads", "tracking", "gambling", "networking", and a default category for any other strings.
     ///
     /// # Examples
     ///
@@ -28,6 +30,9 @@ pub mod firewall {
     /// define_firewall!("gambling", "example-gambling.com");
     /// assert!(is_gambling_website_url("example-gambling.com"));
     ///
+    /// define_firewall!("networking", "a.ping.com");
+    /// assert!(is_bad_networking_url("a.ping.com"));
+    ///
     /// define_firewall!("unknown", "example-unknown.com");
     /// assert!(is_bad_website_url("example-unknown.com"));
     /// ```
@@ -59,6 +64,14 @@ pub mod firewall {
                             .expect("Initialization already set.");
                     }
                 },
+                "networking" => {
+                    if $crate::firewall::GLOBAL_NETWORKING_WEBSITES.get().is_none() {
+                        static WEBSITES: phf::Set<&str> = phf::phf_set! { $($site),* };
+                        $crate::firewall::GLOBAL_NETWORKING_WEBSITES
+                            .set(&WEBSITES)
+                            .expect("Initialization already set.");
+                    }
+                },
                 _ => {
                     if $crate::firewall::GLOBAL_BAD_WEBSITES.get().is_none() {
                         static WEBSITES: phf::Set<&str> = phf::phf_set! { $($site),* };
@@ -104,6 +117,13 @@ pub fn is_gambling_website_url(host: &str) -> bool {
         || is_website_in_custom_set(&host, &firewall::GLOBAL_GAMBLING_WEBSITES)
 }
 
+// General networking blocking. At the moment you have to build this list yourself with the macro define_firewall!("networking", "a.ping.com").
+pub fn is_networking_url(host: &str) -> bool {
+    BAD_WEBSITES.contains(&host)
+        || is_website_in_custom_set(&host, &firewall::GLOBAL_BAD_WEBSITES)
+        || is_website_in_custom_set(&host, &firewall::GLOBAL_NETWORKING_WEBSITES)
+}
+
 /// Is the website in one of the custom sets.
 fn is_website_in_custom_set(
     host: &str,
@@ -119,8 +139,7 @@ fn is_website_in_custom_set(
 /// The URL is in the bad list removing the URL http(s):// and paths.
 pub fn is_bad_website_url_clean(host: &str) -> bool {
     if let Some(host) = get_host_from_url(host) {
-        BAD_WEBSITES.contains(&host)
-            || is_website_in_custom_set(host, &firewall::GLOBAL_BAD_WEBSITES)
+        is_bad_website_url(&host)
     } else {
         false
     }
@@ -129,8 +148,7 @@ pub fn is_bad_website_url_clean(host: &str) -> bool {
 /// The URL is in the ads list.
 pub fn is_ad_website_url_clean(host: &str) -> bool {
     if let Some(host) = get_host_from_url(host) {
-        ADS_WEBSITES.contains(&host)
-            || is_website_in_custom_set(host, &firewall::GLOBAL_ADS_WEBSITES)
+        is_ads_website_url(&host)
     } else {
         false
     }
@@ -139,8 +157,16 @@ pub fn is_ad_website_url_clean(host: &str) -> bool {
 /// The URL is in the tracking list.
 pub fn is_tracking_website_url_clean(host: &str) -> bool {
     if let Some(host) = get_host_from_url(host) {
-        TRACKING_WEBSITES.contains(&host)
-            || is_website_in_custom_set(&host, &firewall::GLOBAL_TRACKING_WEBSITES)
+        is_tracking_website_url(&host)
+    } else {
+        false
+    }
+}
+
+/// The URL is in the networking list.
+pub fn is_networking_website_url_clean(host: &str) -> bool {
+    if let Some(host) = get_host_from_url(host) {
+        is_networking_url(&host)
     } else {
         false
     }
@@ -149,8 +175,7 @@ pub fn is_tracking_website_url_clean(host: &str) -> bool {
 /// The URL is in the gambling list.
 pub fn is_gambling_website_url_clean(host: &str) -> bool {
     if let Some(host) = get_host_from_url(host) {
-        GAMBLING_WEBSITES.contains(&host)
-            || is_website_in_custom_set(&host, &firewall::GLOBAL_GAMBLING_WEBSITES)
+        is_gambling_website_url(&host)
     } else {
         false
     }