This is an automated email from the ASF dual-hosted git repository.
xuyang pushed a commit to branch branch-1.2-lts
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/branch-1.2-lts by this push:
new 3e14c984f1c [branch-1.2](bug) fix unexpected be core in string search
function (#31411)
3e14c984f1c is described below
commit 3e14c984f1c0fac89aae2f90654bb87b0cd51843
Author: xy720 <22125576+xy...@users.noreply.github.com>
AuthorDate: Mon Feb 26 20:18:23 2024 +0800
[branch-1.2](bug) fix unexpected be core in string search function (#31411)
cherry-pick #31312
---
.../functions/functions_multi_string_position.cpp | 15 +++++++------
.../functions/functions_multi_string_search.cpp | 17 +++++++-------
.../apache/doris/analysis/FunctionCallExpr.java | 4 ++++
.../java/org/apache/doris/catalog/FunctionSet.java | 6 +++++
.../test_multi_string_position.groovy | 12 ++++++++++
.../test_multi_string_search.groovy | 26 +++++++++++++++++++++-
6 files changed, 64 insertions(+), 16 deletions(-)
diff --git a/be/src/vec/functions/functions_multi_string_position.cpp
b/be/src/vec/functions/functions_multi_string_position.cpp
index 69e84ec4b3f..27f29776eaf 100644
--- a/be/src/vec/functions/functions_multi_string_position.cpp
+++ b/be/src/vec/functions/functions_multi_string_position.cpp
@@ -55,26 +55,22 @@ public:
Status execute_impl(FunctionContext* context, Block& block, const
ColumnNumbers& arguments,
size_t result, size_t input_rows_count) override {
auto haystack_column = block.get_by_position(arguments[0]).column;
- auto haystack_ptr = haystack_column;
-
auto needles_column = block.get_by_position(arguments[1]).column;
- auto needles_ptr = needles_column;
bool haystack_nullable = false;
bool needles_nullable = false;
if (haystack_column->is_nullable()) {
- haystack_ptr =
check_and_get_column<ColumnNullable>(haystack_column.get())
- ->get_nested_column_ptr();
haystack_nullable = true;
}
if (needles_column->is_nullable()) {
- needles_ptr =
check_and_get_column<ColumnNullable>(needles_column.get())
- ->get_nested_column_ptr();
needles_nullable = true;
}
+ auto haystack_ptr = remove_nullable(haystack_column);
+ auto needles_ptr = remove_nullable(needles_column);
+
const ColumnString* col_haystack_vector =
check_and_get_column<ColumnString>(&*haystack_ptr);
const ColumnConst* col_haystack_const =
@@ -85,6 +81,11 @@ public:
const ColumnConst* col_needles_const =
check_and_get_column_const<ColumnArray>(needles_ptr.get());
+ if (!col_needles_const && !col_needles_vector)
+ return Status::InvalidArgument(
+ "function '{}' encountered unsupported needles column,
found {}", name,
+ needles_column->get_name());
+
if (col_haystack_const && col_needles_vector) {
return Status::InvalidArgument(
"function '{}' doesn't support search with non-constant
needles "
diff --git a/be/src/vec/functions/functions_multi_string_search.cpp
b/be/src/vec/functions/functions_multi_string_search.cpp
index 7a3d842cfbe..04c73375cad 100644
--- a/be/src/vec/functions/functions_multi_string_search.cpp
+++ b/be/src/vec/functions/functions_multi_string_search.cpp
@@ -58,26 +58,22 @@ public:
Status execute_impl(FunctionContext* context, Block& block, const
ColumnNumbers& arguments,
size_t result, size_t input_rows_count) override {
auto haystack_column = block.get_by_position(arguments[0]).column;
- auto haystack_ptr = haystack_column;
-
auto needles_column = block.get_by_position(arguments[1]).column;
- auto needles_ptr = needles_column;
bool haystack_nullable = false;
bool needles_nullable = false;
if (haystack_column->is_nullable()) {
- haystack_ptr =
check_and_get_column<ColumnNullable>(haystack_column.get())
- ->get_nested_column_ptr();
haystack_nullable = true;
}
if (needles_column->is_nullable()) {
- needles_ptr =
check_and_get_column<ColumnNullable>(needles_column.get())
- ->get_nested_column_ptr();
needles_nullable = true;
}
+ auto haystack_ptr = remove_nullable(haystack_column);
+ auto needles_ptr = remove_nullable(needles_column);
+
const ColumnString* col_haystack_vector =
check_and_get_column<ColumnString>(&*haystack_ptr);
const ColumnConst* col_haystack_const =
@@ -88,6 +84,11 @@ public:
const ColumnConst* col_needles_const =
check_and_get_column_const<ColumnArray>(needles_ptr.get());
+ if (!col_needles_const && !col_needles_vector)
+ return Status::InvalidArgument(
+ "function '{}' encountered unsupported needles column,
found {}", name,
+ needles_column->get_name());
+
if (col_haystack_const && col_needles_vector)
return Status::InvalidArgument(
"function '{}' doesn't support search with non-constant
needles "
@@ -323,4 +324,4 @@ void
register_function_multi_string_search(SimpleFunctionFactory& factory) {
factory.register_function<FunctionMultiMatchAny>();
}
-} // namespace doris::vectorized
\ No newline at end of file
+} // namespace doris::vectorized
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/analysis/FunctionCallExpr.java
b/fe/fe-core/src/main/java/org/apache/doris/analysis/FunctionCallExpr.java
index 45ae5751025..568d6e32de0 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/analysis/FunctionCallExpr.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/analysis/FunctionCallExpr.java
@@ -78,6 +78,10 @@ public class FunctionCallExpr extends Expr {
.add("round").add("round_bankers").add("ceil").add("floor")
.add("truncate").add("dround").add("dceil").add("dfloor").build();
+ public static final ImmutableSet<String> STRING_SEARCH_FUNCTION_SET = new
ImmutableSortedSet.Builder(
+ String.CASE_INSENSITIVE_ORDER)
+ .add("multi_search_all_positions").add("multi_match_any").build();
+
static {
java.util.function.BiFunction<ArrayList<Expr>, Type, Type> sumRule =
(children, returnType) -> {
Preconditions.checkArgument(children != null && children.size() >
0);
diff --git a/fe/fe-core/src/main/java/org/apache/doris/catalog/FunctionSet.java
b/fe/fe-core/src/main/java/org/apache/doris/catalog/FunctionSet.java
index af5600078eb..af27685bce4 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/catalog/FunctionSet.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/catalog/FunctionSet.java
@@ -1290,6 +1290,12 @@ public class FunctionSet<T> {
return false;
}
}
+ if
(FunctionCallExpr.STRING_SEARCH_FUNCTION_SET.contains(desc.functionName())) {
+ if (descArgTypes[1].isStringType() &&
candicateArgTypes[1].isArrayType()) {
+ // The needles arg of search functions should not be allowed
to cast from string.
+ return false;
+ }
+ }
// If set `roundPreciseDecimalV2Value`, only use decimalv3 as target
type to execute round function
if (ConnectContext.get() != null
&&
ConnectContext.get().getSessionVariable().roundPreciseDecimalV2Value
diff --git
a/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_position.groovy
b/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_position.groovy
index 28cc08efb88..86d207b9d74 100644
---
a/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_position.groovy
+++
b/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_position.groovy
@@ -28,4 +28,16 @@ suite("test_multi_string_position") {
qt_select "select
multi_search_all_positions('jnckhtjqwycyihuejibqmddrdxe', ['tajzx',
'vuddoylclxatcjvinusdwt', 'spxkhxvzsljkmnzpeubszjnhqczavgtqopxn',
'ckhtjqwycyi', 'xlbfzdxspldoes', 'u', 'czosfebeznt', 'gzhabdsuyreisxvyfrfrkq',
'yihuejibqmd', 'jqwycyihuejibqm', 'cfbvprgzx', 'hxu', 'vxbhrfpzacgd', 'afoaij',
'htjqwycyihu', 'httzbskqd'])"
qt_select "select
multi_search_all_positions('dzejajvpoojdkqbnayahygidyrjmb',
['khwxxvtnqhobbvwgwkpusjlhlzifiuclycml', 'nzvuhtwdaivo', 'dkqbnayahygidyr',
'jajvpoo', 'j', 'wdtbvwmeqgyvetu', 'kqbn', 'idyrjmb',
'tsnxuxevsxrxpgpfdgrkhwqpkse', '', 'efsdgzuefhdzkmquxu',
'zejajvpoojdkqbnayahyg', 'ugwfuighbygrxyctop', 'fcbxzbdugc', 'dxmzzrcplob',
'ejaj', 'wmmupyxrylvawsyfccluiiene', 'ohzmsqhpzbafvbzqwzftbvftei'])"
qt_select "select multi_search_all_positions('ffaujlverosspbzaqefjzql',
['lvero', 'erossp', 'f', 'ujlverosspbz', 'btfimgklzzxlbkbuqyrmnud', 'osspb',
'muqexvtjuaar', 'f', 'bzaq', 'lprihswhwkdhqciqhfaowarn', 'ffaujlve',
'uhbbjrqjb', 'jlver', 'umucyhbbu', 'pjthtzmgxhvpbdphesnnztuu',
'xfqhfdfsbbazactpastzvzqudgk', 'lvovjfoatc', 'z', 'givejzhoqsd', ''])"
+
+ try {
+ sql "select multi_search_all_positions('ffaujlverosspbzaqefjzql',
'lvero, erossp, f, ujlverosspbz, btfimgklzzxlbkbuqyrmnud, osspb, muqexvtjuaar,
f, bzaq, lprihswhwkdhqciqhfaowarn, ffaujlve, uhbbjrqjb, jlver, umucyhbbu,
pjthtzmgxhvpbdphesnnztuu, xfqhfdfsbbazactpastzvzqudgk, lvovjfoatc, z,
givejzhoqsd')"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_search_all_positions"))
+ }
+
+ try {
+ sql "select multi_search_all_positions('ffaujlverosspbzaqefjzql',
'[lvero, erossp, f, ujlverosspbz, btfimgklzzxlbkbuqyrmnud, osspb, muqexvtjuaar,
f, bzaq, lprihswhwkdhqciqhfaowarn, ffaujlve, uhbbjrqjb, jlver, umucyhbbu,
pjthtzmgxhvpbdphesnnztuu, xfqhfdfsbbazactpastzvzqudgk, lvovjfoatc, z,
givejzhoqsd]')"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_search_all_positions"))
+ }
}
diff --git
a/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_search.groovy
b/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_search.groovy
index 5a3229ce361..3a0a9b061fe 100644
---
a/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_search.groovy
+++
b/regression-test/suites/query_p0/sql_functions/search_functions/test_multi_string_search.groovy
@@ -71,4 +71,28 @@ suite("test_multi_string_search") {
qt_select "select multi_match_any('ldrzgttlqaphekkkdukgngl',
['gttlqaphekkkdukgn', 'ekkkd', 'gttlqaphe', 'qaphek', 'h', 'kdu', 'he', 'phek',
'', 'drzgttlqaphekkkd'])"
qt_select "select multi_match_any('ololo', ['ololo', 'ololo', 'ololo'])"
qt_select "select multi_match_any('khljxzxlpcrxpkrfybbfk', ['k'])"
-}
\ No newline at end of file
+
+ try {
+ sql "select multi_match_any(content, 'hello') from ${table_name} order
by col1"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_match_any"))
+ }
+
+ try {
+ sql "select multi_match_any(content, 'hello, !, world, Hello, World')
from ${table_name} order by col1"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_match_any"))
+ }
+
+ try {
+ sql "select multi_match_any(content, '[hello]') from ${table_name}
order by col1"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_match_any"))
+ }
+
+ try {
+ sql "select multi_match_any(content, '[hello, !, world, Hello,
World]') from ${table_name} order by col1"
+ } catch (Exception ex) {
+ assert("${ex}".contains("errCode = 2, detailMessage = No matching
function with signature: multi_match_any"))
+ }
+}
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
