chenjl/canal
1
0
Fork 0
mirror of https://github.com/alibaba/canal.git synced 2026-11-04 00:30:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: preserve commas inside regex grouping when splitting filter patterns (#5588)

Browse Source 
AviaterRegexFilter#init splits the configured pattern on every comma to
separate multiple regular expressions. That blindly breaks regex
quantifiers such as \d{1,2} or non-capturing alternations like
(a,b|c) because the comma inside the grouping construct is treated as
a top-level separator.

Replace the StringUtils.split call with a parse-aware splitter that
tracks the depth of {}, [], () and respects backslash escapes, so only
top-level commas separate patterns. Apply the change to both copies of
AviaterRegexFilter (filter module and connector/core module).

Add AviaterFilterTest#test_regex_with_quantifier_comma covering: a
single-pattern \d{1,2} quantifier, a multi-pattern list combined with
\d{1,2}, and an open-ended {n,} quantifier.

Fixes #5442
This commit is contained in:
Guimu
2026-05-29 10:39:04 +08:00
committed by GitHub
parent cf97b2ae31
commit cf628769d6
3 changed files with 136 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

61
connector/core/src/main/java/com/alibaba/otter/canal/connector/core/filter/AviaterRegexFilter.java
View File

@@ -1,7 +1,6 @@
package com.alibaba.otter.canal.connector.core.filter;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Comparator;
import java.util.HashMap;
import java.util.List;
@@ -19,7 +18,6 @@ import com.googlecode.aviator.Expression;
*/
public class AviaterRegexFilter {
private static final String SPLIT = ",";
private static final String PATTERN_SPLIT = "|";
private static final String FILTER_EXPRESSION = "regex(pattern,target)";
private static final RegexFunction regexFunction = new RegexFunction();
@@ -43,8 +41,9 @@ public class AviaterRegexFilter {
if (StringUtils.isEmpty(pattern)) {
list = new ArrayList<>();
} else {
String[] ss = StringUtils.split(pattern, SPLIT);
list = Arrays.asList(ss);
// fix issue #5442: only split on commas outside regex grouping constructs,
// so quantifiers like \d{1,2} are kept intact.
list = splitPattern(pattern);
}
// 对pattern按照从长到短的排序
@@ -56,6 +55,60 @@ public class AviaterRegexFilter {
this.pattern = StringUtils.join(list, PATTERN_SPLIT);
}
/**
* 按顶层逗号拆分多个正则表达式，跳过位于 {}, [], () 内的逗号，避免破坏类似 \d{1,2} 的量词写法。
*/
private static List<String> splitPattern(String pattern) {
List<String> result = new ArrayList<>();
int braceDepth = 0;
int bracketDepth = 0;
int parenDepth = 0;
int start = 0;
int len = pattern.length();
for (int i = 0; i < len; i++) {
char c = pattern.charAt(i);
// 反斜杠转义：跳过下一个字符，避免把 \{ \[ \( 等误判为分组开始
if (c == '\\' && i + 1 < len) {
i++;
continue;
}
switch (c) {
case '{':
braceDepth++;
break;
case '}':
if (braceDepth > 0) braceDepth--;
break;
case '[':
bracketDepth++;
break;
case ']':
if (bracketDepth > 0) bracketDepth--;
break;
case '(':
parenDepth++;
break;
case ')':
if (parenDepth > 0) parenDepth--;
break;
case ',':
if (braceDepth == 0 && bracketDepth == 0 && parenDepth == 0) {
if (i > start) {
result.add(pattern.substring(start, i));
}
start = i + 1;
}
break;
default:
break;
}
}
if (start < len) {
result.add(pattern.substring(start));
}
return result;
}
public boolean filter(String filtered) {
if (StringUtils.isEmpty(pattern)) {
return defaultEmptyValue;

61
filter/src/main/java/com/alibaba/otter/canal/filter/aviater/AviaterRegexFilter.java
View File

@@ -1,7 +1,6 @@
package com.alibaba.otter.canal.filter.aviater;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Comparator;
import java.util.HashMap;
import java.util.List;
@@ -21,7 +20,6 @@ import com.googlecode.aviator.Expression;
*/
public class AviaterRegexFilter implements CanalEventFilter<String> {
private static final String SPLIT = ",";
private static final String PATTERN_SPLIT = "|";
private static final String FILTER_EXPRESSION = "regex(pattern,target)";
private static final RegexFunction regexFunction = new RegexFunction();
@@ -45,8 +43,9 @@ public class AviaterRegexFilter implements CanalEventFilter<String> {
if (StringUtils.isEmpty(pattern)) {
list = new ArrayList<>();
} else {
String[] ss = StringUtils.split(pattern, SPLIT);
list = Arrays.asList(ss);
// fix issue #5442: only split on commas outside regex grouping constructs,
// so quantifiers like \d{1,2} are kept intact.
list = splitPattern(pattern);
}
// 对pattern按照从长到短的排序
@@ -58,6 +57,60 @@ public class AviaterRegexFilter implements CanalEventFilter<String> {
this.pattern = StringUtils.join(list, PATTERN_SPLIT);
}
/**
* 按顶层逗号拆分多个正则表达式，跳过位于 {}, [], () 内的逗号，避免破坏类似 \d{1,2} 的量词写法。
*/
private static List<String> splitPattern(String pattern) {
List<String> result = new ArrayList<>();
int braceDepth = 0;
int bracketDepth = 0;
int parenDepth = 0;
int start = 0;
int len = pattern.length();
for (int i = 0; i < len; i++) {
char c = pattern.charAt(i);
// 反斜杠转义：跳过下一个字符，避免把 \{ \[ \( 等误判为分组开始
if (c == '\\' && i + 1 < len) {
i++;
continue;
}
switch (c) {
case '{':
braceDepth++;
break;
case '}':
if (braceDepth > 0) braceDepth--;
break;
case '[':
bracketDepth++;
break;
case ']':
if (bracketDepth > 0) bracketDepth--;
break;
case '(':
parenDepth++;
break;
case ')':
if (parenDepth > 0) parenDepth--;
break;
case ',':
if (braceDepth == 0 && bracketDepth == 0 && parenDepth == 0) {
if (i > start) {
result.add(pattern.substring(start, i));
}
start = i + 1;
}
break;
default:
break;
}
}
if (start < len) {
result.add(pattern.substring(start));
}
return result;
}
public boolean filter(String filtered) throws CanalFilterException {
if (StringUtils.isEmpty(pattern)) {
return defaultEmptyValue;

22
filter/src/test/java/com/alibaba/otter/canal/filter/AviaterFilterTest.java
View File

@@ -77,6 +77,28 @@ public class AviaterFilterTest {
Assert.assertEquals(true, result);
}
@Test
public void test_regex_with_quantifier_comma() {
// fix issue #5442: comma inside {n,m} quantifier should not split the pattern
AviaterRegexFilter filter = new AviaterRegexFilter("test\\.order_\\d{1,2}");
Assert.assertTrue(filter.filter("test.order_5"));
Assert.assertTrue(filter.filter("test.order_42"));
Assert.assertFalse(filter.filter("test.order_123"));
Assert.assertFalse(filter.filter("test.order_"));
// multiple patterns combined with quantifier should still split correctly
AviaterRegexFilter filter2 = new AviaterRegexFilter("test\\.order_\\d{1,2},test\\.user_\\d+");
Assert.assertTrue(filter2.filter("test.order_7"));
Assert.assertTrue(filter2.filter("test.user_99"));
Assert.assertFalse(filter2.filter("test.order_123"));
// open-ended quantifier {n,}
AviaterRegexFilter filter3 = new AviaterRegexFilter("test\\.t_\\d{2,}");
Assert.assertTrue(filter3.filter("test.t_12"));
Assert.assertTrue(filter3.filter("test.t_123"));
Assert.assertFalse(filter3.filter("test.t_1"));
}
@Test
public void testDisordered() {
AviaterRegexFilter filter = new AviaterRegexFilter("u\\..*,uvw\\..*,uv\\..*,a\\.x,a\\.xyz,a\\.xy,abc\\.x,abc\\.xyz,abc\\.xy,ab\\.x,ab\\.xyz,ab\\.xy");