use built-in hashset instead of handcrafted boomfilter

Signed-off-by: yuguorui <[email protected]>
yuguorui · Dec 30, 2024 · fb54dfb · fb54dfb
1 parent baea3ca
commit fb54dfb
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 54 deletions.
diff --git a/src/rules.rs b/src/rules.rs
@@ -13,7 +13,7 @@ use tokio::{
 };
 use url::Url;
 
-use crate::utils::{to_io_err, BoomHashSet, ToV6Net, ToV6SockAddr};
+use crate::utils::{to_io_err, ToV6Net, ToV6SockAddr};
 use crate::SETTINGS;
 
 use fast_socks5::client as socks_client;
@@ -32,7 +32,7 @@ pub struct Outbound {
 pub struct Condition {
     pub maxmind_regions: Vec<String>,
     pub dst_ip_range: IpRange<Ipv6Net>,
-    pub domains: Option<BoomHashSet<String>>,
+    pub domains: Option<std::collections::HashSet<String>>,
 }
 
 impl Condition {
@@ -46,15 +46,15 @@ impl Condition {
 
     pub fn match_domain(&self, name: &str) -> bool {
         if let Some(ref domains) = self.domains {
-            if domains.get(&name.to_owned()) {
+            if domains.get(&name.to_owned()).is_some() {
                 return true;
             }
 
             /* Check the DOMAIN-SUFFIX rules */
             let domain = format!(".{}{}", name, RULE_DOMAIN_SUFFIX_TAG);
             let indices = domain.match_indices(".");
             for index in indices {
-                if domains.get(&domain[index.0 + 1..].to_string()) {
+                if domains.get(&domain[index.0 + 1..].to_string()).is_some() {
                     return true;
                 }
             }

diff --git a/src/settings.rs b/src/settings.rs
@@ -9,7 +9,7 @@ use std::net::{Ipv4Addr, Ipv6Addr};
 
 use crate::rules::{RouteTable, RULE_DOMAIN_SUFFIX_TAG};
 
-use crate::utils::{vec_to_array, BoomHashSet, ToV6Net};
+use crate::utils::{vec_to_array, ToV6Net};
 
 const DIRECT_OUTBOUND_NAME: &str = "DIRECT";
 const DROP_OUTBOUND_NAME: &str = "DROP";
@@ -148,10 +148,11 @@ fn parse_intercept_mode(s: &mut Config) -> Result<InterceptMode, ConfigError> {
                 .get("mode")
                 .expect("mode field not found.")
                 .clone()
-                .into_str()?.to_lowercase();
+                .into_str()?
+                .to_lowercase();
             match mode.as_str() {
                 "manual" => return Ok(InterceptMode::MANUAL),
-                "auto"|"tproxy"|"redirect" => {
+                "auto" | "tproxy" | "redirect" => {
                     let capture_local_traffic = table
                         .get("local-traffic")
                         .and_then(|v| {
@@ -230,7 +231,7 @@ fn parse_intercept_mode(s: &mut Config) -> Result<InterceptMode, ConfigError> {
                             )
                         })
                         .unwrap_or(DEFAULT_IPRULE_TABLE);
-                    
+
                     if mode.as_str() != "redirect" {
                         return Ok(InterceptMode::TPROXY {
                             local_traffic: capture_local_traffic,
@@ -249,7 +250,6 @@ fn parse_intercept_mode(s: &mut Config) -> Result<InterceptMode, ConfigError> {
                             proxy_chain,
                         });
                     }
-
                 }
                 _ => Err(ConfigError::Message(
                     "either `auto/tproxy`, `redirect` or `manual` is expected.".to_owned(),
@@ -415,7 +415,7 @@ fn parse_route_rules(s: &mut Config, route: &mut RouteTable) -> Result<(), Confi
 
     for (i, v) in domain_sets.iter().enumerate() {
         let cond = route.rules.get_mut(i).unwrap();
-        cond.domains = Some(BoomHashSet::new(v.to_owned().into_iter().collect_vec()));
+        cond.domains = Some(v.clone());
     }
 
     Ok(())

diff --git a/src/utils.rs b/src/utils.rs
@@ -88,50 +88,6 @@ pub struct BoomHashSet<K: Hash> {
     keys: Vec<K>,
 }
 
-impl<K> BoomHashSet<K>
-where
-    K: Clone + Hash + Debug + PartialEq,
-{
-    fn create_set(mut keys: Vec<K>, mphf: Mphf<K>) -> BoomHashSet<K> {
-        // reorder the keys and values according to the Mphf
-        for i in 0..keys.len() {
-            loop {
-                let kmer_slot = mphf.hash(&keys[i]) as usize;
-                if i == kmer_slot {
-                    break;
-                }
-                keys.swap(i, kmer_slot);
-            }
-        }
-        BoomHashSet {
-            mphf: mphf,
-            keys: keys,
-        }
-    }
-
-    /// Create a new hash map from the parallel array `keys` and `values`
-    pub fn new(keys: Vec<K>) -> BoomHashSet<K> {
-        let mphf = Mphf::new(1.7, &keys);
-        Self::create_set(keys, mphf)
-    }
-
-    /// Get the value associated with `key`, if available, otherwise return None
-    pub fn get(&self, kmer: &K) -> bool {
-        let maybe_pos = self.mphf.try_hash(kmer);
-        match maybe_pos {
-            Some(pos) => {
-                let hashed_kmer = &self.keys[pos as usize];
-                if *kmer == hashed_kmer.clone() {
-                    true
-                } else {
-                    false
-                }
-            }
-            None => false,
-        }
-    }
-}
-
 pub async fn transfer_tcp(in_sock: &mut TcpStream, rt_context: RouteContext) -> Result<()> {
     let mut out_sock = match SETTINGS
         .read()