feat(chat): resolve embedded file links to full code-host URLs #756
+695
−8
Conversation
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
…d add URL builder for code hosts
WalkthroughAdds generation of absolute code-host file URLs for file references in assistant answers: new buildCodeHostFileUrl helper, convertLLMOutputToPortableMarkdown now accepts optional sources to resolve revisions, and AnswerCard components propagate sources so copied Markdown uses absolute links. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant User as User
participant UI as AnswerCard (UI)
participant Utils as convertLLMOutputToPortableMarkdown
participant Builder as buildCodeHostFileUrl
participant Host as CodeHost
User->>UI: Click "Copy answer"
UI->>Utils: convertLLMOutputToPortableMarkdown(answerText, { sources })
Utils->>Builder: buildCodeHostFileUrl(repo, filePath, revision, start, end)
Builder->>Host: Construct URL for provider (GitHub/GitLab/...)
Host-->>Builder: URL string
Builder-->>Utils: resolved URL(s)
Utils-->>UI: portable Markdown with absolute links
UI-->>User: clipboard contains converted Markdown
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Fix all issues with AI agents
In `@packages/web/src/features/chat/utils.ts`:
- Around line 403-412: Normalize the referenced file path before matching and
URL construction: strip any leading slashes (or run a POSIX normalize) on
fileName before using it to find matchingSource in options?.sources and before
calling buildCodeHostFileUrl so revision lookup doesn’t miss entries when paths
include a leading slash; update the use of fileName in the matching logic and
the url call (references: matchingSource, options?.sources, fileName, repo,
revision, buildCodeHostFileUrl).
- Around line 260-365: The buildCodeHostFileUrl function constructs URLs by
interpolating raw filePath, revision and ownerRepo which can break for spaces or
special chars; update buildCodeHostFileUrl to percent-encode path and revision
components before interpolation (e.g., encode each segment of filePath and
ownerRepo via split('/').map(encodeURIComponent).join('/'), and encode revision
with encodeURIComponent) and ensure query parameter values used for Azure DevOps
(version, path, line, lineEnd) are encoded with encodeURIComponent; keep
line/anchor logic (startLine/endLine) intact but do not percent-encode the
leading “#L” or “:”-style anchors themselves.
Author
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Fix all issues with AI agents
In `@packages/web/src/features/chat/utils.ts`:
- Around line 357-362: The Gerrit URL generation currently always prefixes the
revision with "refs/heads/", which breaks when the revision is a commit SHA or a
tag; update the host.includes('gerrit') branch in the URL builder to inspect
encodedRevision and choose the correct path: if encodedRevision matches a commit
SHA (e.g. 40 hex chars) use `/+/{encodedRevision}/{encodedFilePath}`, if it
represents a tag (e.g. starts with `refs/tags/` or otherwise detected as a tag)
use `/+/refs/tags/{tag}/{encodedFilePath}`, otherwise use
`/+/refs/heads/{encodedRevision}/{encodedFilePath}`; preserve the existing
startLine fragment logic (append `#${startLine}` when startLine is present) and
update any variable names (encodedRevision, encodedFilePath, startLine)
accordingly.
- Around line 330-347: The Azure DevOps branch/file query parameters are being
built using encodePathComponent (which preserves slashes), so values like
encodedFilePath and encodedRevision produce unencoded '/' in query params;
update the code that builds url (the Azure DevOps branch in the
host.includes('dev.azure.com') || host.includes('visualstudio.com') block) to
use full URL encoding for query parameter values (e.g., use encodeURIComponent
on the file path and revision instead of encodePathComponent) so that
encodedFilePath and encodedRevision become safe for query strings (slashes
become %2F) and then rebuild the url variable accordingly, preserving the
existing org/project/repoName assembly and the startLine/endLine logic.
Comment on lines
+330
to
+347
| } else if (host.includes('dev.azure.com') || host.includes('visualstudio.com')) { | ||
| // Azure DevOps Cloud and Server | ||
| const repoParts = ownerRepo.split('/'); | ||
| if (repoParts.length >= 3) { | ||
| const org = encodeURIComponent(repoParts[0]); | ||
| const project = encodeURIComponent(repoParts[1]); | ||
| const repoName = repoParts.slice(2).map(encodeURIComponent).join('/'); | ||
| // For Azure DevOps, encode the path preserving forward slashes | ||
| url = `https://${host}/${org}/${project}/_git/${repoName}?path=/${encodedFilePath}&version=GB${encodedRevision}`; | ||
| if (startLine) { | ||
| url += `&line=${startLine}`; | ||
| if (endLine && startLine !== endLine) { | ||
| url += `&lineEnd=${endLine}`; | ||
| } | ||
| } | ||
| } else { | ||
| return fileName; | ||
| } |
There was a problem hiding this comment.
Query parameter values need full URL encoding, not path encoding.
The encodedFilePath and encodedRevision use encodePathComponent() which preserves / characters. However, for URL query parameter values, slashes must be encoded as %2F. Branch names like feature/my-branch will produce version=GBfeature/my-branch instead of the correct version=GBfeature%2Fmy-branch, which Azure DevOps may misinterpret.
🔧 Suggested fix
} else if (host.includes('dev.azure.com') || host.includes('visualstudio.com')) {
// Azure DevOps Cloud and Server
const repoParts = ownerRepo.split('/');
if (repoParts.length >= 3) {
const org = encodeURIComponent(repoParts[0]);
const project = encodeURIComponent(repoParts[1]);
const repoName = repoParts.slice(2).map(encodeURIComponent).join('/');
- // For Azure DevOps, encode the path preserving forward slashes
- url = `https://${host}/${org}/${project}/_git/${repoName}?path=/${encodedFilePath}&version=GB${encodedRevision}`;
+ // For Azure DevOps query params, use full URI encoding (slashes must be encoded)
+ const pathParam = encodeURIComponent('/' + filePath);
+ const versionParam = encodeURIComponent(revision);
+ url = `https://${host}/${org}/${project}/_git/${repoName}?path=${pathParam}&version=GB${versionParam}`;
if (startLine) {
url += `&line=${startLine}`;🤖 Prompt for AI Agents
In `@packages/web/src/features/chat/utils.ts` around lines 330 - 347, The Azure
DevOps branch/file query parameters are being built using encodePathComponent
(which preserves slashes), so values like encodedFilePath and encodedRevision
produce unencoded '/' in query params; update the code that builds url (the
Azure DevOps branch in the host.includes('dev.azure.com') ||
host.includes('visualstudio.com') block) to use full URL encoding for query
parameter values (e.g., use encodeURIComponent on the file path and revision
instead of encodePathComponent) so that encodedFilePath and encodedRevision
become safe for query strings (slashes become %2F) and then rebuild the url
variable accordingly, preserving the existing org/project/repoName assembly and
the startLine/endLine logic.
Comment on lines
+357
to
+362
| } else if (host.includes('gerrit')) { | ||
| // Gerrit Self-Hosted | ||
| url = `https://${host}/plugins/gitiles/${encodedOwnerRepo}/+/refs/heads/${encodedRevision}/${encodedFilePath}`; | ||
| if (startLine) { | ||
| url += `#${startLine}`; | ||
| } |
There was a problem hiding this comment.
Gerrit URL assumes revision is always a branch.
The URL format hardcodes refs/heads/ which only works for branch names. If revision is a commit SHA (e.g., from a matched source), the generated URL will be invalid. Gerrit/Gitiles expects /+/{sha}/{path} for commits and /+/refs/tags/{tag}/{path} for tags.
Consider detecting the revision type or documenting this limitation.
🔧 Potential enhancement
} else if (host.includes('gerrit')) {
// Gerrit Self-Hosted
- url = `https://${host}/plugins/gitiles/${encodedOwnerRepo}/+/refs/heads/${encodedRevision}/${encodedFilePath}`;
+ // Detect if revision looks like a commit SHA (40 hex chars) vs branch name
+ const isSha = /^[0-9a-f]{40}$/i.test(revision);
+ const refPath = isSha ? revision : `refs/heads/${encodedRevision}`;
+ url = `https://${host}/plugins/gitiles/${encodedOwnerRepo}/+/${isSha ? revision : refPath}/${encodedFilePath}`;
if (startLine) {
url += `#${startLine}`;
}🤖 Prompt for AI Agents
In `@packages/web/src/features/chat/utils.ts` around lines 357 - 362, The Gerrit
URL generation currently always prefixes the revision with "refs/heads/", which
breaks when the revision is a commit SHA or a tag; update the
host.includes('gerrit') branch in the URL builder to inspect encodedRevision and
choose the correct path: if encodedRevision matches a commit SHA (e.g. 40 hex
chars) use `/+/{encodedRevision}/{encodedFilePath}`, if it represents a tag
(e.g. starts with `refs/tags/` or otherwise detected as a tag) use
`/+/refs/tags/{tag}/{encodedFilePath}`, otherwise use
`/+/refs/heads/{encodedRevision}/{encodedFilePath}`; preserve the existing
startLine fragment logic (append `#${startLine}` when startLine is present) and
update any variable names (encodedRevision, encodedFilePath, startLine)
accordingly.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Summary
Convert embedded LLM file citations (e.g.
@file:{github.com/owner/repo::path/to/file.ext:10-20})into portable, clickable URLs that point to a file on the originating code host.
This change ensures links emitted by the model are usable by users (not local paths).
Closes #576
What I changed
packages/web/src/features/chat/utils.tsbuildCodeHostFileUrl()to construct file URLs for GitHub, GitLab, Bitbucket, Azure DevOps, Gitea, Gerrit and generic git hosts.convertLLMOutputToPortableMarkdown()now accepts an optionalsourcesarray so it can use the indexed revision for a file when available, falling back tomain.packages/web/src/features/chat/utils.test.tscovering multiple hosts, line anchors, ranges,.md/.mdxhandling, and branch resolution fromsources.Why
LLM responses embed file references for traceability. Previously those produced local paths (e.g.,
/path/to/file), which are not portable when copying as Markdown. This change converts them into full remote URLs matching the code host's expected blob/src format and preserves line anchors.Notes on platform support
Supported hosts and behaviors:
.../blob/{branch}/{path}#L{n}(adds?plain=1for raw markdown views where appropriate).../-/blob/{branch}/{path}#L{n}.../src/{branch}/{path}#lines-{n}.../_git/{repo}?path=/{path}&version=GB{branch}&line={n}If a file
source(from the LLM retrieval metadata) is provided, the code will prefer therevisionfrom that source to construct the URL instead of defaulting tomain.Summary by CodeRabbit
New Features
Tests
✏️ Tip: You can customize this high-level summary in your review settings.