Merge pull request #10736 from ClickHouse/extract-groups-range-check

Added range check for extractGroups function

src/Functions/extractAllGroups.cpp

@@ -18,6 +18,7 @@ namespace DB
 namespace ErrorCodes
 {
     extern const int ARGUMENT_OUT_OF_BOUND;
+    extern const int BAD_ARGUMENTS;
 }
 
 
@@ -64,8 +65,15 @@ public:
 
         const auto regexp = Regexps::get<false, false>(needle);
         const auto & re2 = regexp->getRE2();
+
+        if (!re2)
+            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
         const size_t groups_count = re2->NumberOfCapturingGroups();
 
+        if (!groups_count)
+            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
         // Including 0-group, which is the whole regexp.
         PODArrayWithStackMemory<re2_st::StringPiece, 128> matched_groups(groups_count + 1);
 

src/Functions/extractGroups.cpp

@@ -18,6 +18,7 @@ namespace DB
 namespace ErrorCodes
 {
     extern const int ARGUMENT_OUT_OF_BOUND;
+    extern const int BAD_ARGUMENTS;
 }
 
 
@@ -64,8 +65,15 @@ public:
 
         const auto regexp = Regexps::get<false, false>(needle);
         const auto & re2 = regexp->getRE2();
+
+        if (!re2)
+            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
         const size_t groups_count = re2->NumberOfCapturingGroups();
 
+        if (!groups_count)
+            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
         // Including 0-group, which is the whole regexp.
         PODArrayWithStackMemory<re2_st::StringPiece, 128> matched_groups(groups_count + 1);
 

tests/queries/0_stateless/01246_extractAllGroups.reference

 0 groups, zero matches
-[[],[]]
 1 group, multiple matches, String and FixedString
 [['hello'],['world']]
 [['hello'],['world']]

tests/queries/0_stateless/01246_extractAllGroups.sql

@@ -4,10 +4,10 @@ SELECT extractAllGroups('hello');  --{serverError 42} not enough arguments
 SELECT extractAllGroups('hello', 123);  --{serverError 43} invalid argument type
 SELECT extractAllGroups(123, 'world');  --{serverError 43}  invalid argument type
 SELECT extractAllGroups('hello world', '((('); --{serverError 427}  invalid re
-SELECT extractAllGroups('hello world', materialize('\\w+')); --{serverError 44} non-cons needle
+SELECT extractAllGroups('hello world', materialize('\\w+')); --{serverError 44} non-const needle
 
 SELECT '0 groups, zero matches';
-SELECT extractAllGroups('hello world', '\\w+');
+SELECT extractAllGroups('hello world', '\\w+'); -- { serverError 36 }
 
 SELECT '1 group, multiple matches, String and FixedString';
 SELECT extractAllGroups('hello world', '(\\w+)');

tests/queries/0_stateless/01273_extractGroups.reference

 0 groups, zero matches
-[]
 1 group, multiple matches, String and FixedString
 ['hello','world']
 ['hello','world']

tests/queries/0_stateless/01273_extractGroups.sql

@@ -4,10 +4,10 @@ SELECT extractGroups('hello');  --{serverError 42} not enough arguments
 SELECT extractGroups('hello', 123);  --{serverError 43} invalid argument type
 SELECT extractGroups(123, 'world');  --{serverError 43}  invalid argument type
 SELECT extractGroups('hello world', '((('); --{serverError 427}  invalid re
-SELECT extractGroups('hello world', materialize('\\w+')); --{serverError 44} non-cons needle
+SELECT extractGroups('hello world', materialize('\\w+')); --{serverError 44} non-const needle
 
 SELECT '0 groups, zero matches';
-SELECT extractGroups('hello world', '\\w+');
+SELECT extractGroups('hello world', '\\w+'); -- { serverError 36 }
 
 SELECT '1 group, multiple matches, String and FixedString';
 SELECT extractGroups('hello world', '(\\w+) (\\w+)');

tests/queries/0_stateless/01275_extract_groups_check.sql

+SELECT extractGroups('hello', ''); -- { serverError 69 }
+SELECT extractAllGroups('hello', ''); -- { serverError 69 }
+
+SELECT extractGroups('hello', ' '); -- { serverError 36 }
+SELECT extractAllGroups('hello', ' '); -- { serverError 36 }
+
+SELECT extractGroups('hello', '\0'); -- { serverError 36 }
+SELECT extractAllGroups('hello', '\0'); -- { serverError 36 }
+
+SELECT extractGroups('hello', 'world'); -- { serverError 36 }
+SELECT extractAllGroups('hello', 'world'); -- { serverError 36 }
+
+SELECT extractGroups('hello', 'hello|world'); -- { serverError 36 }
+SELECT extractAllGroups('hello', 'hello|world'); -- { serverError 36 }