feat: support more lauguage for build in stop words and add remove punct, regex filter (#41412)

relate: https://github.com/milvus-io/milvus/issues/41213

---------

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>

internal/core/thirdparty/tantivy/tantivy-binding/Cargo.lock

@@ -211,6 +211,21 @@ dependencies = [
  "serde",
 ]
 
+[[package]]
+name = "bit-set"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "08807e080ed7f9d5433fa9b275196cfc35414f66a0c79d864dc51a0d825231a3"
+dependencies = [
+ "bit-vec",
+]
+
+[[package]]
+name = "bit-vec"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5e764a1d40d510daf35e07be9eb06e75770908c27d411ee6c92109c9840eaaf7"
+
 [[package]]
 name = "bitflags"
 version = "1.3.2"
@@ -857,6 +872,17 @@ dependencies = [
  "pin-project-lite",
 ]
 
+[[package]]
+name = "fancy-regex"
+version = "0.14.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6e24cb5a94bcae1e5408b0effca5cd7172ea3c5755049c5f3af4cd283a165298"
+dependencies = [
+ "bit-set",
+ "regex-automata",
+ "regex-syntax 0.8.5",
+]
+
 [[package]]
 name = "fastdivide"
 version = "0.4.2"
@@ -3853,6 +3879,7 @@ dependencies = [
  "criterion",
  "either",
  "env_logger",
+ "fancy-regex",
  "futures",
  "icu_segmenter",
  "jieba-rs",
@@ -4568,7 +4595,7 @@ version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
 dependencies = [
- "windows-sys 0.48.0",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]

internal/core/thirdparty/tantivy/tantivy-binding/Cargo.toml

@@ -31,6 +31,7 @@ either = "1.13.0"
 icu_segmenter = "2.0.0-beta2"
 whatlang = "0.16.4"
 lingua = "1.7.1"
+fancy-regex = "0.14.0"
 
 [dev-dependencies]
 rand = "0.3"

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/analyzer.rs

@@ -2,9 +2,7 @@ use serde_json as json;
 use std::collections::HashMap;
 use tantivy::tokenizer::*;
 
-use crate::analyzer::{
-    build_in_analyzer::*, filter::*, tokenizers::get_builder_with_tokenizer, util::*,
-};
+use super::{build_in_analyzer::*, filter::*, tokenizers::get_builder_with_tokenizer};
 use crate::error::Result;
 use crate::error::TantivyBindingError;
 

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/build_in_analyzer.rs

@@ -1,8 +1,8 @@
 use tantivy::tokenizer::*;
 
-use crate::analyzer::tokenizers::*;
-use crate::analyzer::filter::*;
-use crate::analyzer::stop_words;
+use super::tokenizers::*;
+use super::filter::*;
+use super::filter::stop_words;
 
 // default build-in analyzer
 pub(crate) fn standard_analyzer(stop_words: Vec<String>) -> TextAnalyzer {

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/filter.rs

@@ -1,8 +1,9 @@
 use serde_json as json;
 use tantivy::tokenizer::*;
 
-use crate::error::{Result,TantivyBindingError};
-use crate::analyzer::util::*;
+use super::util::*;
+use super::{RegexFilter, RemovePunctFilter};
+use crate::error::{Result, TantivyBindingError};
 
 pub(crate) enum SystemFilter {
     Invalid,
@@ -12,9 +13,11 @@ pub(crate) enum SystemFilter {
     CnCharOnly(CnCharOnlyFilter),
     CnAlphaNumOnly(CnAlphaNumOnlyFilter),
     Length(RemoveLongFilter),
+    RemovePunct(RemovePunctFilter),
     Stop(StopWordFilter),
     Decompounder(SplitCompoundWords),
     Stemmer(Stemmer),
+    Regex(RegexFilter),
 }
 
 impl SystemFilter {
@@ -29,6 +32,8 @@ impl SystemFilter {
             Self::Stop(filter) => builder.filter(filter).dynamic(),
             Self::Decompounder(filter) => builder.filter(filter).dynamic(),
             Self::Stemmer(filter) => builder.filter(filter).dynamic(),
+            Self::RemovePunct(filter) => builder.filter(filter).dynamic(),
+            Self::Regex(filter) => builder.filter(filter).dynamic(),
             Self::Invalid => builder,
         }
     }
@@ -152,6 +157,7 @@ impl From<&str> for SystemFilter {
             "alphanumonly" => Self::AlphaNumOnly(AlphaNumOnlyFilter),
             "cncharonly" => Self::CnCharOnly(CnCharOnlyFilter),
             "cnalphanumonly" => Self::CnAlphaNumOnly(CnAlphaNumOnlyFilter),
+            "removepunct" => Self::RemovePunct(RemovePunctFilter),
             _ => Self::Invalid,
         }
     }
@@ -174,6 +180,7 @@ impl TryFrom<&json::Map<String, json::Value>> for SystemFilter {
                     "stop" => get_stop_words_filter(params),
                     "decompounder" => get_decompounder_filter(params),
                     "stemmer" => get_stemmer_filter(params),
+                    "regex" => RegexFilter::from_json(params).map(|f| SystemFilter::Regex(f)),
                     other => Err(TantivyBindingError::InternalError(format!(
                         "unsupport filter type: {}",
                         other

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/mod.rs

@@ -0,0 +1,11 @@
+mod filter;
+mod remove_punct_filter;
+mod regex_filter;
+pub(crate) mod stop_words;
+mod util;
+
+use regex_filter::RegexFilter;
+use remove_punct_filter::RemovePunctFilter;
+
+pub(crate) use filter::*;
+pub(crate) use util::*;

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/regex_filter.rs

@@ -0,0 +1,128 @@
+use crate::error::{Result, TantivyBindingError};
+use serde_json as json;
+use fancy_regex as regex;
+use tantivy::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
+
+#[derive(Clone)]
+pub struct RegexFilter {
+    regex: regex::Regex,
+}
+
+impl RegexFilter {
+    /// Creates a `RegexFilter` given regex expression
+    pub fn new(expr: &str) -> Result<RegexFilter> {
+        regex::Regex::new(expr).map_or_else(
+            |e| {
+                Err(TantivyBindingError::InvalidArgument(format!(
+                    "regex expression invalid, expr:{}, err: {}",
+                    expr, e
+                )))
+            },
+            |regex| Ok(RegexFilter { regex }),
+        )
+    }
+
+    pub fn from_json(params: &json::Map<String, json::Value>) -> Result<RegexFilter> {
+        params.get("expr").map_or(
+            Err(TantivyBindingError::InternalError(format!(
+                "must set expr for regex filter"
+            ))),
+            |value| {
+                value.as_str().map_or(
+                    Err(TantivyBindingError::InternalError(format!(
+                        "expr must be string"
+                    ))),
+                    |expr| RegexFilter::new(expr),
+                )
+            },
+        )
+    }
+}
+
+impl TokenFilter for RegexFilter {
+    type Tokenizer<T: Tokenizer> = RegexFilterWrapper<T>;
+
+    fn transform<T: Tokenizer>(self, tokenizer: T) -> RegexFilterWrapper<T> {
+        RegexFilterWrapper {
+            regex: self.regex,
+            inner: tokenizer,
+        }
+    }
+}
+
+#[derive(Clone)]
+pub struct RegexFilterWrapper<T> {
+    regex: regex::Regex,
+    inner: T,
+}
+
+impl<T: Tokenizer> Tokenizer for RegexFilterWrapper<T> {
+    type TokenStream<'a> = RegexFilterStream<T::TokenStream<'a>>;
+
+    fn token_stream<'a>(&'a mut self, text: &'a str) -> Self::TokenStream<'a> {
+        RegexFilterStream {
+            regex: self.regex.clone(),
+            tail: self.inner.token_stream(text),
+        }
+    }
+}
+
+pub struct RegexFilterStream<T> {
+    regex: regex::Regex,
+    tail: T,
+}
+
+impl<T> RegexFilterStream<T> {
+    fn predicate(&self, token: &Token) -> bool {
+        self.regex.is_match(&token.text).map_or(true, |b|b)
+    }
+}
+
+impl<T: TokenStream> TokenStream for RegexFilterStream<T> {
+    fn advance(&mut self) -> bool {
+        while self.tail.advance() {
+            if self.predicate(self.tail.token()) {
+                return true;
+            }
+        }
+        false
+    }
+
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::analyzer::analyzer::create_analyzer;
+
+    #[test]
+    fn test_regex_filter() {
+        let params = r#"{
+            "tokenizer": "standard",
+            "filter": [{
+                "type": "regex",
+                "expr": "^(?!test)"
+            }]
+        }"#;
+
+        let tokenizer = create_analyzer(&params.to_string());
+        assert!(tokenizer.is_ok(), "error: {}", tokenizer.err().unwrap());
+
+        let mut bining = tokenizer.unwrap();
+        let mut stream = bining.token_stream("test milvus");
+
+        let mut results = Vec::<String>::new();
+        while stream.advance() {
+            let token = stream.token();
+            results.push(token.text.clone());
+        }
+
+        print!("test tokens :{:?}\n", results)
+    }
+}

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/remove_punct_filter.rs

@@ -0,0 +1,80 @@
+use tantivy::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
+
+pub struct RemovePunctFilter;
+
+pub struct RemovePunctFilterStream<T> {
+    regex: regex::Regex,
+    tail: T,
+}
+
+impl TokenFilter for RemovePunctFilter {
+    type Tokenizer<T: Tokenizer> = RemovePunctFilterWrapper<T>;
+
+    fn transform<T: Tokenizer>(self, tokenizer: T) -> RemovePunctFilterWrapper<T> {
+        RemovePunctFilterWrapper(tokenizer)
+    }
+}
+#[derive(Clone)]
+pub struct RemovePunctFilterWrapper<T>(T);
+
+impl<T: Tokenizer> Tokenizer for RemovePunctFilterWrapper<T> {
+    type TokenStream<'a> = RemovePunctFilterStream<T::TokenStream<'a>>;
+
+    fn token_stream<'a>(&'a mut self, text: &'a str) -> Self::TokenStream<'a> {
+        RemovePunctFilterStream {
+            regex: regex::Regex::new(r"[\p{Punct}\s]+").unwrap(),
+            tail: self.0.token_stream(text),
+        }
+    }
+}
+
+impl<T: TokenStream> TokenStream for RemovePunctFilterStream<T> {
+    fn advance(&mut self) -> bool {
+        while self.tail.advance() {
+            if !self.regex.is_match(&self.tail.token().text) {
+                return true;
+            }
+        }
+
+        false
+    }
+
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::analyzer::analyzer::create_analyzer;
+
+    #[test]
+    #[cfg(feature = "lindera-ipadic")]
+    fn test_remove_punct_filter() {
+        let params = r#"{
+            "tokenizer": {
+                "type": "lindera",
+                "dict_kind": "ipadic"
+            },
+            "filter": ["removepunct"]
+        }"#;
+
+        let tokenizer = create_analyzer(&params.to_string());
+        assert!(tokenizer.is_ok(), "error: {}", tokenizer.err().unwrap());
+
+        let mut bining = tokenizer.unwrap();
+        let mut stream = bining.token_stream("ミルヴァスの日本語テスト、句読点テスト");
+
+        let mut results = Vec::<String>::new();
+        while stream.advance() {
+            let token = stream.token();
+            results.push(token.text.clone());
+        }
+
+        print!("test tokens :{:?}\n", results)
+    }
+}

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/util.rs

@@ -1,7 +1,7 @@
 use serde_json as json;
 
-use crate::error::{Result,TantivyBindingError};
-use crate::analyzer::stop_words;
+use super::stop_words;
+use crate::error::{Result, TantivyBindingError};
 
 pub(crate) fn get_string_list(value: &json::Value, label: &str) -> Result<Vec<String>> {
     if !value.is_array() {
@@ -29,14 +29,14 @@ pub(crate) fn get_stop_words_list(str_list: Vec<String>) -> Vec<String> {
     let mut stop_words = Vec::new();
     for str in str_list {
         if str.len() > 0 && str.chars().nth(0).unwrap() == '_' {
-            match str.as_str() {
-                "_english_" => {
-                    for word in stop_words::ENGLISH {
+            match stop_words::fetch_language_stop_words(str.as_str()) {
+                Some(words) => {
+                    for word in words {
                         stop_words.push(word.to_string());
                     }
                     continue;
                 }
-                _other => {}
+                None => {}
             }
         }
         stop_words.push(str);

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/mod.rs

@@ -1,10 +1,8 @@
 mod analyzer;
-mod stop_words;
 mod build_in_analyzer;
 mod filter;
-mod util;
 
 pub mod tokenizers;
 pub use self::analyzer::{create_analyzer, create_analyzer_by_json};
 
-pub(crate) use self::build_in_analyzer::standard_analyzer;
+pub(crate) use self::build_in_analyzer::standard_analyzer;

internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/stop_words.rs

@@ -1,5 +0,0 @@
-pub const ENGLISH: &[&str] = &[
-    "a", "an", "and", "are", "as", "at", "be", "but", "by", "for", "if", "in",
-    "into", "is", "it", "no", "not", "of", "on", "or", "such", "that", "the",
-    "their", "then", "there", "these", "they", "this", "to", "was", "will", "with",
-];