stupid gxhash

Lips7 · Jul 18, 2024 · b4df2bf · b4df2bf
1 parent debe14e
commit b4df2bf
Show file tree

Hide file tree

Showing 4 changed files with 64 additions and 23 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/matcher_rs/Cargo.toml b/matcher_rs/Cargo.toml
@@ -23,12 +23,12 @@ name = "matcher_rs"
 crate-type = ["rlib"]
 
 [dependencies]
+ahash = "0.8.11"
 # aho-corasick = "1.1.3"
 aho-corasick-unsafe = { version = "0.0.4", git = "https://github.com/Lips7/aho-corasick" }
 bitflags = { version = "2.6.0", features = ["serde"] }
 daachorse = "1.0.0"
 fancy-regex = "0.13.0"
-gxhash = "3.4.1"
 id-set = "0.2.2"
 lazy_static = "1.5.0"
 nohash-hasher = "0.2.0"

diff --git a/matcher_rs/src/process/process_matcher.rs b/matcher_rs/src/process/process_matcher.rs
@@ -2,6 +2,9 @@ use std::borrow::Cow;
 use std::fmt::Display;
 use std::sync::Arc;
 
+#[cfg(any(feature = "runtime_build", feature = "dfa"))]
+use ahash::AHashMap;
+use ahash::HashMapExt;
 use aho_corasick_unsafe::AhoCorasick;
 #[cfg(any(feature = "runtime_build", feature = "dfa"))]
 use aho_corasick_unsafe::{AhoCorasickBuilder, AhoCorasickKind, MatchKind as AhoCorasickMatchKind};
@@ -13,9 +16,6 @@ use daachorse::{
     CharwiseDoubleArrayAhoCorasick, CharwiseDoubleArrayAhoCorasickBuilder,
     MatchKind as DoubleArrayAhoCorasickMatchKind,
 };
-#[cfg(any(feature = "runtime_build", feature = "dfa"))]
-use gxhash::HashMap as GxHashMap;
-use gxhash::HashMapExt;
 use id_set::IdSet;
 use lazy_static::lazy_static;
 use nohash_hasher::{IntMap, IsEnabled};
@@ -405,7 +405,7 @@ pub fn get_process_matcher(
 
     #[cfg(feature = "runtime_build")]
     {
-        let mut process_dict = GxHashMap::default();
+        let mut process_dict = AHashMap::default();
 
         match process_type_bit {
             ProcessType::None => {}
@@ -531,7 +531,7 @@ pub fn get_process_matcher(
             ProcessType::Delete => {
                 #[cfg(feature = "dfa")]
                 {
-                    let mut process_dict = GxHashMap::default();
+                    let mut process_dict = AHashMap::default();
                     process_dict.extend(TEXT_DELETE.trim().lines().map(|pair_str| (pair_str, "")));
                     process_dict.extend(WHITE_SPACE.iter().map(|&c| (c, "")));
                     process_dict.retain(|&key, &mut value| key != value);

diff --git a/matcher_rs/src/simple_matcher.rs b/matcher_rs/src/simple_matcher.rs
@@ -1,8 +1,8 @@
 use std::iter;
 use std::{borrow::Cow, collections::HashMap};
 
+use ahash::AHashMap;
 use aho_corasick_unsafe::{AhoCorasick, AhoCorasickBuilder, AhoCorasickKind};
-use gxhash::HashMap as GxHashMap;
 use id_set::IdSet;
 use nohash_hasher::IntMap;
 use sonic_rs::{Deserialize, Serialize};
@@ -211,15 +211,15 @@ impl SimpleMatcher {
 
         let mut ac_dedup_word_id = 0;
         let mut ac_dedup_word_list = Vec::new();
-        let mut ac_dedup_word_id_map = GxHashMap::default();
+        let mut ac_dedup_word_id_map = AHashMap::default();
 
         for (&process_type, simple_word_map) in process_type_word_map {
             let word_process_type = process_type - ProcessType::Delete;
             process_type_list.push(process_type);
 
             for (&simple_word_id, simple_word) in simple_word_map {
-                let mut ac_split_word_and_counter = GxHashMap::default();
-                let mut ac_split_word_not_counter = GxHashMap::default();
+                let mut ac_split_word_and_counter = AHashMap::default();
+                let mut ac_split_word_not_counter = AHashMap::default();
 
                 let mut start = 0;
                 let mut is_and = false;
@@ -403,7 +403,7 @@ impl<'a> TextMatcherTrait<'a, SimpleResult<'a>> for SimpleMatcher {
         &'a self,
         processed_text_process_type_set: &[(Cow<'a, str>, IdSet)],
     ) -> bool {
-        let mut word_id_split_bit_map = GxHashMap::default();
+        let mut word_id_split_bit_map = AHashMap::default();
         let mut word_id_set = IdSet::default();
         let mut not_word_id_set = IdSet::default();
 
@@ -535,7 +535,7 @@ impl<'a> TextMatcherTrait<'a, SimpleResult<'a>> for SimpleMatcher {
         &'a self,
         processed_text_process_type_set: &[(Cow<'a, str>, IdSet)],
     ) -> Vec<SimpleResult<'a>> {
-        let mut word_id_split_bit_map = GxHashMap::default();
+        let mut word_id_split_bit_map = AHashMap::default();
         let mut not_word_id_set = IdSet::default();
 
         let processed_times = processed_text_process_type_set.len();