Merge pull request #9082 from TCeason/ISSUE-9077

fix(query): vector_const like bug fix

Author: bohutang

Cargo.lock

@@ -32,7 +32,6 @@ ctor = "0.1.26"
 hex = "0.4.3"
 itertools = "0.10.5"
 match-template = "0.0.1"
-memchr = "2.5.0"
 num-traits = "0.2.15"
 once_cell = "1.15.0"
 ordered-float = { version = "3.1.0", features = [

src/query/functions-v2/Cargo.toml

@@ -33,7 +33,6 @@ use common_expression::FunctionDomain;
 use common_expression::FunctionProperty;
 use common_expression::FunctionRegistry;
 use common_expression::ValueRef;
-use memchr::memmem;
 use regex::bytes::Regex;
 
 use crate::scalars::string_multi_args::regexp;
@@ -337,26 +336,7 @@ fn register_like(registry: &mut FunctionRegistry) {
                     let ends_with = &pat[1..];
                     Ok(str.ends_with(ends_with))
                 }
-                PatternType::PatternStr => {
-                    let pattern_str = simdutf8::basic::from_utf8(pat)
-                        .expect("Unable to convert the LIKE pattern to string: {}");
-                    let mut sub_strings: Vec<&str> = pattern_str
-                        .split(|c: char| c == '%' || c == '_' || c == '\\')
-                        .collect();
-                    sub_strings.retain(|&substring| !substring.is_empty());
-                    if std::intrinsics::unlikely(sub_strings.is_empty()) {
-                        Ok(like(str, pat))
-                    } else {
-                        let sub_string = sub_strings[0].as_bytes();
-                        if sub_strings.len() == 1 {
-                            Ok(search_sub_str(str, sub_string).is_some())
-                        } else if memmem::find(str, sub_string).is_none() {
-                            Ok(false)
-                        } else {
-                            Ok(like(str, pat))
-                        }
-                    }
-                }
+                PatternType::PatternStr => Ok(like(str, pat)),
             }
         }),
     );
@@ -553,15 +533,6 @@ pub fn check_pattern_type(pattern: &[u8], is_pruning: bool) -> PatternType {
     }
 }
 
-#[inline]
-fn search_sub_str(str: &[u8], substr: &[u8]) -> Option<usize> {
-    if substr.len() <= str.len() {
-        str.windows(substr.len()).position(|w| w == substr)
-    } else {
-        None
-    }
-}
-
 #[inline]
 fn decode_one(data: &[u8]) -> Option<(u8, usize)> {
     if data.is_empty() {

src/query/functions-v2/src/scalars/comparison.rs

@@ -35,7 +35,6 @@ h3ron = "0.15.1"
 hex = "0.4.3"
 itertools = "0.10.5"
 md-5 = "0.10.5"
-memchr = "2.5.0"
 naive-cityhash = "0.2.0"
 num = "0.4.0"
 num-format = "0.4.0"

src/query/functions/Cargo.toml

@@ -13,7 +13,6 @@
 // limitations under the License.
 
 use common_datavalues::prelude::*;
-use memchr::memmem;
 
 use super::comparison::StringSearchCreator;
 use super::utils::StringSearchImpl;
@@ -58,36 +57,7 @@ impl StringSearchImpl for StringSearchLike {
                 BooleanColumn::from_iterator(lhs.scalar_iter().map(|x| op(x.ends_with(ends_with))))
             }
             PatternType::PatternStr => {
-                let pattern = simdutf8::basic::from_utf8(rhs)
-                    .expect("Unable to convert the LIKE pattern to string: {}");
-                let mut sub_strings: Vec<&str> = pattern
-                    .split(|c: char| c == '%' || c == '_' || c == '\\')
-                    .collect();
-                sub_strings.retain(|&substring| !substring.is_empty());
-                if std::intrinsics::unlikely(sub_strings.is_empty()) {
-                    BooleanColumn::from_iterator(lhs.scalar_iter().map(|x| op(like(x, rhs))))
-                } else {
-                    let sub_string = sub_strings[0].as_bytes();
-                    // This impl like position function
-                    if sub_strings.len() == 1 {
-                        BooleanColumn::from_iterator(lhs.scalar_iter().map(|x| {
-                            let contain = search_sub_str(x, sub_string);
-                            if contain.is_none() {
-                                op(false)
-                            } else {
-                                op(true)
-                            }
-                        }))
-                    } else {
-                        BooleanColumn::from_iterator(lhs.scalar_iter().map(|x| {
-                            if memmem::find(x, sub_string).is_none() {
-                                op(false)
-                            } else {
-                                op(like(x, rhs))
-                            }
-                        }))
-                    }
-                }
+                BooleanColumn::from_iterator(lhs.scalar_iter().map(|x| op(like(x, rhs))))
             }
         }
     }
@@ -169,15 +139,6 @@ pub fn check_pattern_type(pattern: &[u8], is_pruning: bool) -> PatternType {
     }
 }
 
-#[inline]
-fn search_sub_str(str: &[u8], substr: &[u8]) -> Option<usize> {
-    if substr.len() <= str.len() {
-        str.windows(substr.len()).position(|w| w == substr)
-    } else {
-        None
-    }
-}
-
 #[inline]
 fn decode_one(data: &[u8]) -> Option<(u8, usize)> {
     if data.is_empty() {

src/query/functions/src/scalars/comparisons/comparison_like.rs

@@ -1227,5 +1227,17 @@ SELECT parse_json('"cd"') not regexp '.*';
 ----
 0
 
+statement ok
+drop table if exists t;
+
+statement ok
+create table t(id String);
 
+statement ok
+insert into t values('IRxxSIPD');
 
+statement query T
+select id from t where id not like '%_SIP';
+
+----
+IRxxSIPD