fix(amazonq): Previous and subsequent cells are used as context for completion in a Jupyter notebook #7086
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
|
justinmk3
reviewed
Apr 17, 2025
justinmk3
approved these changes
Apr 17, 2025
Contributor
justinmk3
left a comment
justinmk3
left a comment
There was a problem hiding this comment.
This is helpful, thanks! Tests for this function are here:
You can run that test file by opening it in vscode then choose Extension tests (current file) (amazon) in the run/debug menu:
brdskggs
force-pushed
the
multi-cell-context
branch
from
8ca009c to
5f29d4a
Compare
Contributor
Author
|
Will consolidate the function doing the suffix and prefix context fetching. |
Contributor
Author
|
Both code cells and Markdown cells are used for context now; Markdown cells have lines prefixed with a language-specific comment based on the language of the cell the completion is triggered from. |
brdskggs
changed the title
brdskggs
force-pushed
the
multi-cell-context
branch
3 times, most recently
from
5b7ca22 to
76e096a
Compare
justinmk3
reviewed
May 5, 2025
packages/amazonq/test/unit/codewhisperer/util/editorContext.test.ts
Outdated
Show resolved
Hide resolved
justinmk3
reviewed
May 5, 2025
packages/amazonq/test/unit/codewhisperer/util/editorContext.test.ts
Outdated
Show resolved
Hide resolved
justinmk3
reviewed
May 5, 2025
justinmk3
reviewed
May 5, 2025
justinmk3
reviewed
May 5, 2025
brdskggs
force-pushed
the
multi-cell-context
branch
from
6fc5a9f to
db49f5d
Compare
justinmk3
reviewed
May 7, 2025
Contributor
|
lint failure |
brdskggs
force-pushed
the
multi-cell-context
branch
from
130c19d to
2650145
Compare
justinmk3
reviewed
May 7, 2025
|
|
||
| const languageCommentChars: Record<string, string> = { | ||
| python: '# ', | ||
| java: '// ', |
Contributor
There was a problem hiding this comment.
I just noticed this. It's not sustainable for us to introduce yet another language-specific handling just for notebooks. Why isn't the notebook logic using the existing logic used by other Q functionality ?
Contributor
Author
There was a problem hiding this comment.
Good question! I'm new to this codebase, so I'll have to research what existing logic you are referring to -- any quick pointers of where to start looking are appreciated.
Contributor
Author
There was a problem hiding this comment.
Never mind, Q to the rescue -- figured out the changes I needed to make to runtimeLanguageContext.
brdskggs
force-pushed
the
multi-cell-context
branch
from
2650145 to
a6625ea
Compare
…mpletion in a Jupyter notebook
…mpletion in a Jupyter notebook
Renaming `extractCellsSliceContext` to `getNotebookCellsSliceContext` Co-authored-by: Justin M. Keyes <[email protected]>
brdskggs
force-pushed
the
multi-cell-context
branch
from
a6625ea to
dbfe86b
Compare
leigaol
added a commit
to aws/language-server-runtimes
that referenced
this pull request
Aug 12, 2025
## Problem Support of inline completion in JL Notebook is gone since the language server migration. aws/aws-toolkit-vscode#7086 ## Solution Let the IDE provide the file context. There is a technical limitations of the language server that it cannot find or construct the `vscode.NotebookDocument`, `vscode.NotebookCell`. The `workspace.getTextDocument` does not work with JL Notebook as it gets a URI in the format of "`vscode-notebook-cell:/../A1.ipynb#W2sZmlsZQ%3D%3D`". The language server does not have the cell number, cell state, cursor position, etc of the current cell. <!--- REMINDER: - Read CONTRIBUTING.md first. - Add test coverage for your changes. - Link to related issues/commits. - Testing: how did you test your changes? - Screenshots if applicable --> ## License By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
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.
Problem
VS Code treats each cell in a notebook as a separate editor. As a result, when building the left- and right-contexts for the completion from the current editor, we were limited to just the current cell, which might be very small and/or reference variables and functions defined in other cells. That meant that completions never used the context of other cells when making suggestions, and were often very generic. #7031
Solution
The
extractContextForCodeWhispererfunction now checks if it is being called in a cell in a Jupyter notebook. If so, it collects the surrounding cells to use as context, respecting the maximum context length. During this process, Markdown cells have each line prefixed with a language-specific comment character.feature/xbranches will not be squash-merged at release time.