fix(caddyfile): Prevent parser to panic when no token were added by empty {block}#7543
Merged
francislavoie merged 2 commits intocaddyserver:masterfrom Mar 3, 2026
Merged
Conversation
In #7206, some code was changed to make the parser ignore `{block}` tokens when there is no body to replace it with. As such, no tokens were appended to `tokensCopy` after this change when this condition was fulfilled. Unfortunately, the parser code assumed in line 510 that tokens were added on every iteration of `importedTokens`. This caused the program to panic with an "index out of range" runtime error. As a fix, and since this code is supposed to be checking the last index of the slice, we're now checking the length of `tokensCopy` instead of solely relying on the loop's range index variable. Signed-off-by: prettysunflower <me@prettysunflower.moe>
|
|
prettysunflower
changed the title
Member
|
Interesting. Could you add the reproduce case from the issue as a testcase to prove that it fixes it? Thanks! |
Signed-off-by: prettysunflower <me@prettysunflower.moe>
Contributor
Author
|
Sure, I've added a test case (slightly modified from the issue), it panics without this fix, and passes with the fix. |
Member
|
Perfect, thanks! @ZingyAwesome if you'd like to build from this branch to play around with it to confirm that would be great :) |
francislavoie
approved these changes
Mar 3, 2026
|
It's working now, thank you for such a fast fix 🎉 |
4 tasks
prettysunflower
added a commit
to prettysunflower/caddy
that referenced
this pull request
Mar 10, 2026
Resolve issue caddyserver#7557 So, here is the situation: - Pull request caddyserver#7206 included some changes to the doImport's function of Caddyfile's parser. What it does is that if there is no token within a block that follows the import, and the import contains `{block}`, then the `{block}` token is discarded. - This implementation had a few flaws: - Issue caddyserver#7518 noticed that in cases that `{block}` was not imported, a runtime error was raised due to the assumption that tokens were always added to `tokensCopy` on every iteration of `importedTokens`. This was fixed by pull request caddyserver#7543. - Issue caddyserver#7557 notices that {block} can be ignored when imported from a certain file. There, it's again an issue with how the import works. When `import snippets` is called, this import instruction doesn't contains any nested blocks. And when the argument replacer that is the `importedTokens` loop is called and finds `{block}`, it uses the block from the file's import (which in this case is nothing), `{block}` is erased, and unavailable when the import directive is called for the imported snippet. These changes fix the issues raised in caddyserver#7557 and tries to stabilize the situation by adding a `Disposable` field to `Token`. It works like this: - When a `{block}` is encountered in the argument replacer loop, instead of ignoring it and not copying it, the token is now copied (instead of being erased), but with the `Disposable` field toggled on. - When subsequent imports happens, the "disposable" `{block}` token is available for the processing of the new import, and can be used to call snippets from other files. - When all tokens are processed within a server block, disposable tokens are removed from the keys and segments of the blocks. This way, `{block}` tokens are there for subsequent inputs, and are removed if unused. Tests added in pull requests caddyserver#7206 and caddyserver#7543 passes with this new implementation, confirming that unused `{block}` are accepted if nothing is passed to `import`, as well as the other usual tests. A new test was also added based on issue caddyserver#7557 reporting, and also passes. ... hopefully this commit description is clear? Signed-off-by: prettysunflower <me@prettysunflower.moe>
prettysunflower
added a commit
to prettysunflower/caddy
that referenced
this pull request
Mar 10, 2026
Resolve issue caddyserver#7557 So, here is the situation: - Pull request caddyserver#7206 included some changes to the doImport's function of Caddyfile's parser. What it does is that if there is no token within a block that follows the import, and the import contains `{block}`, then the `{block}` token is discarded. - After this pull request: - Issue caddyserver#7518 noticed that in cases that `{block}` was not imported, a runtime error was raised due to the assumption that tokens were always added to `tokensCopy` on every iteration of `importedTokens`. This was fixed by pull request caddyserver#7543. - Issue caddyserver#7557 notices that {block} can be ignored when imported from a certain file. There, it's again an issue with how the import works. When `import snippets` is called, this import instruction doesn't contains any nested blocks. And when the argument replacer that is the `importedTokens` loop is called and finds `{block}`, it uses the block from the file's import (which in this case is nothing), `{block}` is erased, and unavailable when the import directive is called for the imported snippet. The changed in this commit addresses the second issue by checking before replacing `{block}` if we're currently in a snippet definition, and appending the `{block}` token to `tokensCopy` if we are. With this changes, when importing those snippets, the `{block}` token will be available to be replaced by the nested blocks in `tokensToAdd` if needed, or erased if there are no nested blocks and `tokensToAdd` is empty. Tests added in pull requests caddyserver#7206 and caddyserver#7543 passes with this new implementation, confirming that unused `{block}` are accepted if nothing is passed to `import`, as well as the other usual tests. A new test was also added based on issue caddyserver#7557 reporting, and also passes. Signed-off-by: prettysunflower <me@prettysunflower.moe>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Nyallo~
In #7206, some code was changed to make the parser ignore
{block}tokens when there is no body to replace it with. As such, no tokens were appended totokensCopyafter this change when this condition was fulfilled.Unfortunately, the parser code assumed in line 510 that tokens were added to
tokensCopyon every iteration ofimportedTokens. This caused the program to panic with an "index out of range" runtime error.As a fix, and since this code is supposed to be checking the last index of the slice, we're now checking the length of
tokensCopyinstead of solely relying on the loop's range index variable (i).This should fix #7518?
Assistance Disclosure
No AI/LLM was used.