Implement REGEXMATCH function (#15)

Co-authored-by: Cursor Agent <cursoragent@cursor.com>
This commit is contained in:
Pijus Kamandulis
2026-05-30 21:31:45 +03:00
committed by GitHub
parent c3726a6633
commit 05e8cd2842
8 changed files with 1160 additions and 922 deletions
@@ -162,6 +162,8 @@ func (r rowContext) selectItem_SelectItemTypeFunctionCall(functionCall parsers.F
return r.strings_EndsWith(functionCall.Arguments)
case parsers.FunctionCallStartsWith:
return r.strings_StartsWith(functionCall.Arguments)
case parsers.FunctionCallRegexMatch:
return r.strings_RegexMatch(functionCall.Arguments)
case parsers.FunctionCallConcat:
return r.strings_Concat(functionCall.Arguments)
case parsers.FunctionCallIndexOf:
@@ -2,6 +2,7 @@ package memoryexecutor
import (
"fmt"
"regexp"
"strings"
"github.com/pikami/cosmium/internal/logger"
@@ -75,6 +76,46 @@ func (r rowContext) strings_StartsWith(arguments []interface{}) bool {
return strings.HasPrefix(str1, str2)
}
func (r rowContext) strings_RegexMatch(arguments []interface{}) bool {
value, valueOk := r.parseString(arguments[0])
pattern, patternOk := r.parseString(arguments[1])
if !valueOk || !patternOk {
return false
}
modifiers, ok := r.getStringFlag(arguments)
if !ok {
return false
}
regexPattern := pattern
if strings.Contains(modifiers, "x") {
regexPattern = stripRegexIgnoredWhitespace(regexPattern)
}
var flags strings.Builder
if strings.Contains(modifiers, "i") {
flags.WriteByte('i')
}
if strings.Contains(modifiers, "m") {
flags.WriteByte('m')
}
if strings.Contains(modifiers, "s") {
flags.WriteByte('s')
}
if flags.Len() > 0 {
regexPattern = "(?" + flags.String() + ")" + regexPattern
}
matched, err := regexp.MatchString(regexPattern, value)
if err != nil {
logger.Errorf("strings_RegexMatch - invalid pattern %q: %v", pattern, err)
return false
}
return matched
}
func (r rowContext) strings_Concat(arguments []interface{}) string {
result := ""
@@ -318,6 +359,20 @@ func (r rowContext) getBoolFlag(arguments []interface{}) bool {
return ignoreCase
}
func (r rowContext) getStringFlag(arguments []interface{}) (string, bool) {
if len(arguments) <= 2 || arguments[2] == nil {
return "", true
}
flagItem := arguments[2].(parsers.SelectItem)
if value, ok := r.resolveSelectItem(flagItem).(string); ok {
return value, true
}
logger.ErrorLn("getStringFlag - got parameters of wrong type")
return "", false
}
func (r rowContext) parseString(argument interface{}) (value string, ok bool) {
exItem := argument.(parsers.SelectItem)
ex := r.resolveSelectItem(exItem)
@@ -329,6 +384,41 @@ func (r rowContext) parseString(argument interface{}) (value string, ok bool) {
return "", false
}
func stripRegexIgnoredWhitespace(pattern string) string {
var result strings.Builder
inCharClass := false
escaped := false
for _, r := range pattern {
if escaped {
result.WriteRune(r)
escaped = false
continue
}
if r == '\\' {
result.WriteRune(r)
escaped = true
continue
}
switch r {
case '[':
inCharClass = true
case ']':
inCharClass = false
}
if !inCharClass && (r == ' ' || r == '\t' || r == '\n' || r == '\r' || r == '\f') {
continue
}
result.WriteRune(r)
}
return result.String()
}
func convertToString(value interface{}) string {
switch v := value.(type) {
case string:
@@ -231,6 +231,42 @@ func Test_Execute_StringFunctions(t *testing.T) {
)
})
t.Run("Should execute function REGEXMATCH()", func(t *testing.T) {
testQueryExecute(
t,
parsers.SelectStmt{
SelectItems: []parsers.SelectItem{
{
Path: []string{"c", "id"},
Type: parsers.SelectItemTypeField,
},
{
Alias: "regexMatch",
Type: parsers.SelectItemTypeFunctionCall,
Value: parsers.FunctionCall{
Type: parsers.FunctionCallRegexMatch,
Arguments: []interface{}{
parsers.SelectItem{
Path: []string{"c", "str"},
Type: parsers.SelectItemTypeField,
},
testutils.SelectItem_Constant_String("COOL WORLD"),
testutils.SelectItem_Constant_String("i"),
},
},
},
},
Table: parsers.Table{SelectItem: testutils.SelectItem_Path("c")},
},
mockData,
[]memoryexecutor.RowType{
map[string]interface{}{"id": "123", "regexMatch": false},
map[string]interface{}{"id": "456", "regexMatch": false},
map[string]interface{}{"id": "789", "regexMatch": true},
},
)
})
t.Run("Should execute function INDEX_OF()", func(t *testing.T) {
testQueryExecute(
t,