DBZ-8170 Error caused by database name containing regular keywords by jw-itq · Pull Request #5800 · debezium/debezium

jw-itq · 2024-08-28T05:51:24Z

https://issues.redhat.com/browse/DBZ-8170

Error caused by database name containing regular keywords, such as database name: test$user

github-actions · 2024-08-28T05:51:35Z

Welcome as a new contributor to Debezium, @jw-itq. Reviewers, please add missing author name(s) and alias name(s) to the COPYRIGHT.txt and Aliases.txt respectively.

Naros · 2024-08-29T03:30:18Z

Hi @jw-itq would it be possible to add a test not only for the use case you've described with test$user, but also checking what this change will do should the configuration supply the regular expression contain escaped syntax values, such as test\$user, does this cause such a configuration to no longer work or be interpreted incorrectly?

jw-itq · 2024-08-29T06:07:32Z

Hi @jw-itq would it be possible to add a test not only for the use case you've described with test$user, but also checking what this change will do should the configuration supply the regular expression contain escaped syntax values, such as test\$user, does this cause such a configuration to no longer work or be interpreted incorrectly?

thanks, I will attempt to add a test case. Yes, this could potentially cause test$user not to be interpreted correctly. However, on the JDBC URL, you cannot use $ but you can use \$. If the Matcher.quoteReplacement is used, test$user will still be effective.

Naros · 2024-08-29T06:11:16Z

Hi @jw-itq would it be possible to add a test not only for the use case you've described with test$user, but also checking what this change will do should the configuration supply the regular expression contain escaped syntax values, such as test\$user, does this cause such a configuration to no longer work or be interpreted incorrectly?

thanks, I will attempt to add a test case. Yes, this could potentially cause test$user not to be interpreted correctly. However, on the JDBC URL, you cannot use $ but you can use $. If the Matcher.quoteReplacement is used, test$user will still be effective.

But you're also adjusting the Predicates class which is used for many other configuration options such as the include and exclude list configurations that we've always communicated that special regex characters should be escaped if they're part of the database, schema, or table name. So we need to consider all those variants with this change imo.

jw-itq · 2024-08-29T06:17:03Z

Hi @jw-itq would it be possible to add a test not only for the use case you've described with test$user, but also checking what this change will do should the configuration supply the regular expression contain escaped syntax values, such as test\$user, does this cause such a configuration to no longer work or be interpreted incorrectly?

thanks, I will attempt to add a test case. Yes, this could potentially cause test$user not to be interpreted correctly. However, on the JDBC URL, you cannot use $ but you can use $. If the Matcher.quoteReplacement is used, test$user will still be effective.

But you're also adjusting the Predicates class which is used for many other configuration options such as the include and exclude list configurations that we've always communicated that special regex characters should be escaped if they're part of the database, schema, or table name. So we need to consider all those variants with this change imo.

thanks, the Predicate indeed requires more thought, and I will reconsider the changes regarding the Predicate. Or do you have any better suggestions?

Ritabrata1080

Please squash the commits to have a linear history.

Naros · 2026-01-06T13:59:12Z

❌ Developer Certificate of Origin (DCO) check failed.

Hi @jw-itq, please sign off all commits with:

git commit -s

If pull request commits are not signed off, the pull request cannot be merged. For more information about why this is required, please see our blog about contribution requirement changes.

jw-itq · 2026-01-06T15:44:18Z

❌ Developer Certificate of Origin (DCO) check failed.

Hi @jw-itq, please sign off all commits with:
git commit -s
If pull request commits are not signed off, the pull request cannot be merged. For more information about why this is required, please see our blog about contribution requirement changes.

ok, thanks. I'll handle it right away.

jw-itq · 2026-01-06T15:44:49Z

Please squash the commits to have a linear history.

ok!

jw-itq · 2026-01-07T14:04:55Z

@Naros @Ritabrata1080 hi,commits squashed and DCO sign-off added.

Ritabrata1080 · 2026-01-10T07:55:28Z

debezium-core/src/main/java/io/debezium/function/Predicates.java


    public static <T, U> BiPredicate<T, U> includes(String regexPatterns, BiFunction<T, U, String> conversion) {
-        Set<Pattern> patterns = Strings.setOfRegex(regexPatterns, Pattern.CASE_INSENSITIVE);
+        Set<Pattern> patterns = Strings.setOfRegex(regexPatterns != null ? Matcher.quoteReplacement(regexPatterns) : null, Pattern.CASE_INSENSITIVE);


What happens if patterns set tries to set regex of null value ? Please check for potential NPE crash across patch.

Naros · 2026-01-13T16:11:35Z

debezium-core/src/main/java/io/debezium/function/Predicates.java

     */
    public static <T> Predicate<T> includes(String regexPatterns, Function<T, String> conversion) {
-        Set<Pattern> patterns = Strings.setOfRegex(regexPatterns, Pattern.CASE_INSENSITIVE);
+        Set<Pattern> patterns = Strings.setOfRegex(regexPatterns != null ? Matcher.quoteReplacement(regexPatterns) : null, Pattern.CASE_INSENSITIVE);


We've always expected that if a provided include/exclude list configuration property contains regex characters that may not be interpreted as literal, they should be manually escaped in the configuration. This change here would specifically create breaking behavior in environments that already pre-escape special characters.

I'm personally not convinced we should change this behavior, as using $ and, to a lesser degree, ? in tablenames is quite common among certain database vendors. I am concerned about the potential impact on a longstanding expectation when handling these strings.

When it comes to the JDBC URL, we can certainly be significantly more lenient there.
@jpechane any thoughts?

I agree, we should not break backward compatibility in this regard.

jw-itq · 2026-01-14T14:57:34Z

@Naros @Ritabrata1080 @jpechane Thank you for your review. I will revert the changes made to Predicates.java

jpechane · 2026-01-15T09:37:29Z

@jw-itq LGTM. Could you please rebase the PR on the latest main and add a test case if possible to make sure the results are expected?

jw-itq · 2026-01-15T14:58:46Z

@jw-itq LGTM. Could you please rebase the PR on the latest main and add a test case if possible to make sure the results are expected?

@jpechane ok.I see that the latest version has been updated. Do we not need to proceed with the current PR?

jpechane · 2026-01-16T12:14:28Z

@jw-itq Please take a lok at the PR changes. I belive it is broken now.

jw-itq · 2026-01-16T13:35:52Z

@jw-itq Please take a lok at the PR changes. I belive it is broken now.

@jpechane hi. the JdbcConnection.java fix using Matcher.quoteReplacement() is already present in the main branch (commit fe5758e or earlier). After rebasing, my PR now only contains contributor information updates. Should I close this PR since the actual fix is already merged, or would you like me to keep it for the contributor credits?

jpechane · 2026-01-22T15:40:23Z

@jw-itq Please keep the credits, just it is necessary to rebase on the latest main and remove the extra lines ike <<<<<

jw-itq · 2026-01-28T03:43:54Z

@jw-itq Please keep the credits, just it is necessary to rebase on the latest main and remove the extra lines ike <<<<<

@jpechane Thanks! Done.

…uch as database name: test Signed-off-by: Shiwanming <[email protected]>

Ritabrata1080 reviewed Sep 6, 2024

View reviewed changes

jw-itq force-pushed the DBZ-8170 branch from e9df98b to 7c1d4d9 Compare January 7, 2026 13:25

Ritabrata1080 reviewed Jan 10, 2026

View reviewed changes

jw-itq force-pushed the DBZ-8170 branch 3 times, most recently from 80bc397 to e85dd76 Compare January 12, 2026 03:22

Naros reviewed Jan 13, 2026

View reviewed changes

jw-itq force-pushed the DBZ-8170 branch 2 times, most recently from b05dab0 to 79aa9bb Compare January 14, 2026 14:51

jw-itq force-pushed the DBZ-8170 branch from 79aa9bb to 7ecf17e Compare January 15, 2026 14:53

DBZ-8170 Error caused by database name containing regular keywords, s…

bacba72

…uch as database name: test Signed-off-by: Shiwanming <[email protected]>

jw-itq force-pushed the DBZ-8170 branch from 7ecf17e to bacba72 Compare January 28, 2026 03:58

jpechane merged commit 5adf12e into debezium:main Jan 30, 2026
29 checks passed

Uh oh!

Conversation

jw-itq commented Aug 28, 2024

Uh oh!

github-actions bot commented Aug 28, 2024

Uh oh!

Naros commented Aug 29, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jw-itq commented Aug 29, 2024

Uh oh!

Naros commented Aug 29, 2024

Uh oh!

jw-itq commented Aug 29, 2024

Uh oh!

Ritabrata1080 left a comment

Choose a reason for hiding this comment

Uh oh!

Naros commented Jan 6, 2026

Uh oh!

jw-itq commented Jan 6, 2026

Uh oh!

jw-itq commented Jan 6, 2026

Uh oh!

jw-itq commented Jan 7, 2026

Uh oh!

Ritabrata1080 Jan 10, 2026

Choose a reason for hiding this comment

Uh oh!

Naros Jan 13, 2026

Choose a reason for hiding this comment

Uh oh!

jpechane Jan 14, 2026

Choose a reason for hiding this comment

Uh oh!

jw-itq commented Jan 14, 2026

Uh oh!

jpechane commented Jan 15, 2026

Uh oh!

jw-itq commented Jan 15, 2026

Uh oh!

jpechane commented Jan 16, 2026

Uh oh!

jw-itq commented Jan 16, 2026

Uh oh!

jpechane commented Jan 22, 2026

Uh oh!

jw-itq commented Jan 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Naros commented Aug 29, 2024 •

edited

Loading

jw-itq commented Jan 28, 2026 •

edited

Loading