Merge pull request #495 from weibocom/bkdrsubh

增加hash方法bkdrsubhat

sharding/src/hash/bkdrsub.rs

@@ -1,14 +1,24 @@
 /// <pre>用于支持key中部分字符串做hashkey，且hash算法类似bkdr的hash算法；
-/// key格式：abc#123_456，hashkey则是‘#’之后、‘_’之前的内容；
-/// 格式注意：'#'需要存在，否则hashkey为空；'_'可能不存在，如果'_'不存在，则'#'之后的全部是hashkey</pre>
+/// hashkey是‘#’之后、$delimiter之前的内容；
+/// 例如key：abc#123_456
+///   如果$delimiter是'_'，则hashkey是123；
+///   如果$delimiter是'^'，则hashkey是123_456；
+/// 格式注意：'#'需要存在，否则hashkey为空；$delimiter可能不存在，如果$delimiter不存在，则'#'之后的全部是hashkey</pre>
 #[derive(Clone, Default, Debug)]
-pub struct Bkdrsub;
+pub struct BkdrsubDelimiter {
+    delimiter: u8,
+}
+impl BkdrsubDelimiter {
+    pub fn from(delimiter: u8) -> Self {
+        Self { delimiter }
+    }
+}
 
-impl super::Hash for Bkdrsub {
+impl super::Hash for BkdrsubDelimiter {
     fn hash<S: super::HashKey>(&self, key: &S) -> i64 {
         const SEED: i32 = 131; // 31 131 1313 13131 131313 etc..
         const START_CHAR_VAL: u8 = '#' as u8;
-        const END_CHAR_VAL: u8 = '_' as u8;
+        let end_char_val: u8 = self.delimiter;
 
         let mut hash = 0_i32;
         let mut found_start_char = false;
@@ -17,7 +27,7 @@ impl super::Hash for Bkdrsub {
             let c = key.at(i);
             if found_start_char {
                 // hashkey 计算
-                if c != END_CHAR_VAL {
+                if c != end_char_val {
                     hash = hash.wrapping_mul(SEED).wrapping_add(c as i32);
                     continue;
                 }

sharding/src/hash/mod.rs

@@ -25,7 +25,7 @@ pub use rawsuffix::RawSuffix;
 
 pub mod crc;
 
-use self::{bkdrsub::Bkdrsub, crc64::Crc64, fnv1::Fnv1F32, fnv1::Fnv1aF64};
+use self::{bkdrsub::BkdrsubDelimiter, crc64::Crc64, fnv1::Fnv1F32, fnv1::Fnv1aF64};
 use enum_dispatch::enum_dispatch;
 
 // 占位hash，主要用于兼容服务框架，供mq等业务使用
@@ -69,7 +69,7 @@ pub enum Hasher {
     Padding(Padding),
     Raw(Raw), // redis raw, long型字符串直接用数字作为hash
     Bkdr(Bkdr),
-    Bkdrsub(Bkdrsub),
+    BkdrsubDelimiter(BkdrsubDelimiter),
     BkdrAbsCrc32(BkdrAbsCrc32), // 混合三种hash：先bkdr，再abs，最后进行crc32计算
     Crc32(Crc32),
     Crc32Short(Crc32Short),         // mc short crc32
@@ -120,7 +120,8 @@ impl Hasher {
             return match alg_parts[0] {
                 HASH_PADDING => Self::Padding(Default::default()),
                 "bkdr" => Self::Bkdr(Default::default()),
-                "bkdrsub" => Self::Bkdrsub(Default::default()),
+                "bkdrsub" => Self::BkdrsubDelimiter(BkdrsubDelimiter::from('_' as u8)),
+                "bkdrsubhat" => Self::BkdrsubDelimiter(BkdrsubDelimiter::from('^' as u8)),
                 "bkdrabscrc32" => Self::BkdrAbsCrc32(Default::default()),
                 "raw" => Self::Raw(Raw::from(Default::default())),
                 "crc32" => Self::Crc32(Default::default()),

tests/src/bkdrsub.rs

@@ -20,9 +20,24 @@ fn bkdrsub_one() {
     let dist = Distribute::from("modrange-8640", &servers);
     let dist_idx = dist.index(hash1);
 
-    println!("key:{}, hash:{}, idx:{}", key1, hash1, dist_idx);
+    println!("bkdrsub key:{}, hash:{}, idx:{}", key1, hash1, dist_idx);
+    assert_eq!(dist_idx, 905)
 }
+#[test]
+fn bkdrsubhat_one() {
+    let hasher = Hasher::from("bkdrsubhat");
+
+    let key1 = "otdn#1042015:carSubBrand^e4ab74c125e9e95edad691ffe9820118";
+    let hash1 = hasher.hash(&key1.as_bytes());
 
+    let shards = 1080;
+    let servers = vec!["padding".to_string(); shards];
+    let dist = Distribute::from("modrange-8640", &servers);
+    let dist_idx = dist.index(hash1);
+
+    println!("bkdrsubhat key:{}, hash:{}, idx:{}", key1, hash1, dist_idx);
+    assert_eq!(dist_idx, 905)
+}
 // TODO 临时批量文件的hash、dist校验测试，按需打开
 #[test]
 fn bkdrsub_dist() {

tests/src/hash_test.rs

@@ -70,12 +70,26 @@ mod hash_test {
 
         let key1 = "abc#12345678901234567";
         let hash1 = hasher.hash(&key1.as_bytes());
-        println!("key:{}, hash:{}", key1, hash1);
+        println!("bkdrsub key:{}, hash:{}", key1, hash1);
         assert_eq!(hash1, 1108486745);
 
         let key2 = "abc#12345678901234567_123456";
         let hash2 = hasher.hash(&key2.as_bytes());
-        println!("key:{}, hash:{}", key2, hash2);
+        println!("bkdrsub key:{}, hash:{}", key2, hash2);
+        assert_eq!(hash2, 1108486745);
+    }
+    #[test]
+    fn bkdrsubhat() {
+        let hasher = Hasher::from("bkdrsubhat");
+
+        let key1 = "abc#12345678901234567";
+        let hash1 = hasher.hash(&key1.as_bytes());
+        println!("bkdrsubhat key:{}, hash:{}", key1, hash1);
+        assert_eq!(hash1, 1108486745);
+
+        let key2 = "abc#12345678901234567^123456";
+        let hash2 = hasher.hash(&key2.as_bytes());
+        println!("bkdrsubhat key:{}, hash:{}", key2, hash2);
         assert_eq!(hash2, 1108486745);
     }