Additional insertAt options, add support for postprocessors to templated prompts #68
Conversation
Enhance the enrichChatMessages function to include optional global metadata. This allows for injecting global information into the chat message enrichment process. - Update function signature to accept globalMetadata - Merge globalMetadata into the template context
Enhance the functionality of `insertAt` by adding more options to facilitate better template responses. This includes support for inserting at various points within lines, items, and surrounding lines. - Added insertAt options: - "start-of-line" - "end-of-line" - "start-of-item" - "end-of-item" - "new-line-above" - "new-line-below" - "replace-line" - Updated global metadata structure to support additional options - Implemented logic for new insertAt cases
- Enhance `.gitignore` to exclude `docs/Library/Core` - Export utility functions in `editorUtils.ts` for line operations based on text position - Modify `Provider.ts` to handle post-processors in `AbstractProvider`, processing data after response is complete - Update `prompts.ts` to support post-processors in AI prompt templates - Extend `StreamChatOptions` in `types.ts` to include `postProcessors` array - Define `PostProcessorData` type to pass context to post-processors
- Introduced `convertToBulletList` for transforming strings into bullet lists - Added `convertToTaskList` to ensure each line in a string is a markdown task - Implemented `indentOneLevel` to indent lines based on the previous line - Created `removeDuplicateStart` to remove the first line if it's a duplicate These functions enhance the utility of space scripts, making them more versatile.
|
Caution Review failedThe pull request is closed. WalkthroughWalkthroughThis pull request introduces enhancements to task management and text processing capabilities within the application. It adds new AI prompts and space scripts for task decomposition, string conversion to bullet and task lists, and improved text formatting. Additionally, modifications to existing functions increase their accessibility and functionality, particularly in handling AI responses and managing metadata. These changes aim to streamline user interactions and improve task organization. Changes
Possibly related PRs
Tip New featuresWalkthrough comment now includes:
Notes:
Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media? TipsChatThere are 3 ways to chat with CodeRabbit:
Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
CodeRabbit Configuration File (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
Outside diff range, codebase verification and nitpick comments (5)
docs/Library/AICore/Space Script/Remove Duplicate Start.md (1)
11-25: Approve the logic but consider removing debug logs.The function correctly implements the logic to remove duplicate lines at the start of the response. However, consider removing or replacing
console.logstatements with a more appropriate logging mechanism if this script is intended for production use.src/interfaces/Provider.ts (1)
Line range hint
52-131: Approve the integration of new features, suggest improvements for type safety and simplification.The modifications to the
streamChatIntoEditormethod in theAbstractProviderclass effectively integrate the newonResponseCompleteandpostProcessorsfeatures, enhancing the method's functionality.However, consider the following improvements:
- Use more specific types than
anyfor callback parameters to enhance type safety.- Simplify the new logic to improve readability and maintainability, possibly by breaking down the method into smaller, more focused functions.
src/providers/openai.ts (2)
15-15: Consider specifying the type ofdatainonResponseComplete.The addition of
onResponseCompleteas an optional callback inStreamChatOptionsis correctly implemented. However, specifying the type ofdatamore precisely thananycould enhance type safety and code clarity.
Line range hint
56-104: Effective use ofonResponseCompleteinstreamChat.The integration of
onResponseCompletein thestreamChatmethod is well-implemented, with the callback being correctly invoked at the end of the stream and upon the "end" event. It is advisable to consider adding error handling around the callback invocation to manage any potential exceptions that might arise from the callback execution.src/prompts.ts (1)
37-37: Suggestion: Remove commented-out debug code.To maintain a clean codebase, consider removing the commented-out console log unless it's intended for future debugging purposes.
Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Files ignored due to path filters (2)
.gitignoreis excluded by none and included by nonescripts/create-test-space.shis excluded by none and included by none
Files selected for processing (14)
- docs/Library/AICore/AIPrompt/AI Split Task.md (1 hunks)
- docs/Library/AICore/Space Script/AI Query LLM.md (1 hunks)
- docs/Library/AICore/Space Script/AI Search Embeddings.md (1 hunks)
- docs/Library/AICore/Space Script/Convert to bullets.md (1 hunks)
- docs/Library/AICore/Space Script/Convert to task list.md (1 hunks)
- docs/Library/AICore/Space Script/Indent lines.md (1 hunks)
- docs/Library/AICore/Space Script/Remove Duplicate Start.md (1 hunks)
- src/editorUtils.ts (3 hunks)
- src/interfaces/Provider.ts (4 hunks)
- src/prompts.ts (10 hunks)
- src/providers/ollama.ts (1 hunks)
- src/providers/openai.ts (4 hunks)
- src/types.ts (2 hunks)
- src/utils.ts (2 hunks)
Files skipped from review due to trivial changes (2)
- docs/Library/AICore/Space Script/AI Query LLM.md
- docs/Library/AICore/Space Script/AI Search Embeddings.md
Additional context used
Path-based instructions (12)
docs/Library/AICore/Space Script/Remove Duplicate Start.md (1)
Pattern
**/*.md: Review these markdown files. They are written for a markdown note taking tool called Silverbullet, and may contain extra syntax not typically found in markdown.Ensure any documentation makes sense and is good.
docs/Library/AICore/Space Script/Convert to task list.md (1)
Pattern
**/*.md: Review these markdown files. They are written for a markdown note taking tool called Silverbullet, and may contain extra syntax not typically found in markdown.Ensure any documentation makes sense and is good.
docs/Library/AICore/Space Script/Convert to bullets.md (1)
Pattern
**/*.md: Review these markdown files. They are written for a markdown note taking tool called Silverbullet, and may contain extra syntax not typically found in markdown.Ensure any documentation makes sense and is good.
docs/Library/AICore/Space Script/Indent lines.md (1)
Pattern
**/*.md: Review these markdown files. They are written for a markdown note taking tool called Silverbullet, and may contain extra syntax not typically found in markdown.Ensure any documentation makes sense and is good.
docs/Library/AICore/AIPrompt/AI Split Task.md (1)
Pattern
**/*.md: Review these markdown files. They are written for a markdown note taking tool called Silverbullet, and may contain extra syntax not typically found in markdown.Ensure any documentation makes sense and is good.
src/editorUtils.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/providers/ollama.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/types.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/interfaces/Provider.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/providers/openai.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/prompts.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.src/utils.ts (1)
Pattern
**/*.ts: Review these typescript files for best practices, readability, and DRY.
Additional comments not posted (11)
docs/Library/AICore/Space Script/Convert to task list.md (1)
11-28: Approve the changes.The function
convertToTaskListis well-implemented, using regular expressions effectively to convert strings into markdown task list items. The logic is clear and the code is clean.docs/Library/AICore/Space Script/Convert to bullets.md (1)
11-34: Approve the changes.The function
convertToBulletListis well-implemented, using the indentation of the previous line effectively to maintain formatting consistency when converting strings into bullet list items. The logic is clear and the code is clean.src/editorUtils.ts (4)
3-3: Well-implemented function for retrieving selected text.The
getSelectedTextfunction is correctly implemented, handling cases where no text is selected and providing a consistent return structure. This function enhances the editor's functionality by allowing other parts of the application to access selected text.
39-39: Simple and effective function for retrieving page length.The
getPageLengthfunction is straightforward and effectively retrieves the length of the text in the current page. This function is essential for features that need to know the size of the content within the editor.
56-62: Well-implemented function for retrieving specific lines of text.The
getLinefunction is correctly implemented, handling cases where the line number is out of range by returning an empty string. This function is essential for features that need to access specific lines of text within the editor.
64-67: Effective function for retrieving lines based on position.The
getLineOfPosfunction effectively combinesgetLineNumberAtPosandgetLineto retrieve the line of text at a specific position. This function provides a useful abstraction for features that need to access text based on its position within the editor.src/types.ts (1)
154-159: Approve the definition ofPostProcessorData.The new type
PostProcessorDatais well-defined and appropriately structured for its intended use in handling post-processing data. The properties included are relevant and useful for the described functionality.src/providers/openai.ts (1)
42-49: Proper integration ofonResponseCompleteinchatWithAI.The update to include
onResponseCompletein the function signature and its subsequent handling within the method are correctly implemented. The logic separation between streaming and non-streaming chats is maintained, ensuring that existing functionalities are not disrupted.src/prompts.ts (2)
Line range hint
6-22: Approved: New imports for enhanced functionality.The addition of
systemandPostProcessorDatasupports the new features introduced in this PR, enhancing system interactions and type safety for postprocessors.
Line range hint
79-313: Review: EnhancedinsertAthandling andpostProcessorsintegration.The expanded functionality for
insertAtoptions and the addition ofpostProcessorssignificantly enhance the flexibility and capabilities of theinsertAiPromptFromTemplatefunction. However, ensure that:
- The
insertAtoptions are thoroughly tested, especially for edge cases.- The
postProcessorsare securely integrated and do not introduce performance bottlenecks.Consider adding more detailed error handling and logging for the
postProcessorsexecution to aid in debugging and maintenance.src/utils.ts (1)
Line range hint
178-248: Review: Integration ofglobalMetadatainenrichChatMessages.The addition of
globalMetadataenhances the function's capability to handle dynamic data. Ensure that:
globalMetadatais properly validated and sanitized to prevent security vulnerabilities.- The function handles cases where
globalMetadatais undefined to avoid runtime errors.Consider adding default values or checks for required metadata properties to improve robustness and error handling.
| ```space-script | ||
| silverbullet.registerFunction({ name: "indentOneLevel" }, async (data) => { | ||
| const { response, lineBefore, lineCurrent } = data; | ||
| console.log(data); | ||
|
|
||
| // Function to determine the indentation of a line | ||
| const getIndentation = (line) => line.match(/^\s*/)[0]; | ||
|
|
||
| // Determine the maximum indentation of lineBefore and lineCurrent | ||
| const maxIndentation = getIndentation(lineBefore).length > getIndentation(lineCurrent).length | ||
| ? getIndentation(lineBefore) | ||
| : getIndentation(lineCurrent); | ||
|
|
||
| // Define additional indentation level | ||
| const additionalIndentation = ' '; | ||
|
|
||
| // Compute new indentation | ||
| const newIndentation = maxIndentation + additionalIndentation; | ||
|
|
||
| // Apply new indentation to all lines in the response | ||
| const indentedLines = response.split('\n').map(line => `${newIndentation}${line.trim()}`).join('\n'); | ||
|
|
||
| console.log("indentedLines:", indentedLines); | ||
|
|
||
| return indentedLines; | ||
| }); |
There was a problem hiding this comment.
Well-implemented function with clear documentation.
The indentOneLevel function is well-documented and logically structured. It effectively uses regular expressions and string manipulation to indent lines based on the maximum indentation of two reference lines.
Suggestion for Improvement:
Consider adding error handling for potential edge cases, such as null or undefined inputs, to enhance robustness.
| @@ -17,7 +17,7 @@ async function getSelectedText() { | |||
| }; | |||
| } | |||
|
|
|||
| async function getSelectedTextOrNote() { | |||
| export async function getSelectedTextOrNote() { | |||
There was a problem hiding this comment.
Logical extension of text selection functionality.
The getSelectedTextOrNote function effectively extends the getSelectedText function by adding the capability to determine if the entire note is selected. This is useful for features that need to differentiate between partial and full note selections.
Suggestion for Improvement:
Consider simplifying the condition for isWholeNote to enhance readability and maintainability.
| export function getLineNumberAtPos(text: string, pos: number): number { | ||
| const lines = text.split("\n"); | ||
| let currentPos = 0; | ||
| for (let i = 0; i < lines.length; i++) { | ||
| if (currentPos <= pos && pos < currentPos + lines[i].length + 1) { | ||
| return i; | ||
| } | ||
| currentPos += lines[i].length + 1; // +1 for the newline character | ||
| } | ||
| return -1; | ||
| } |
There was a problem hiding this comment.
Efficient function for calculating line numbers.
The getLineNumberAtPos function is well-implemented and efficiently calculates the line number corresponding to a given position in the text. This function is crucial for features that need to manipulate text based on line numbers.
Suggestion for Improvement:
Consider optimizing the loop to reduce the number of iterations for large texts, potentially using a binary search approach if the lines are of relatively uniform length.
| export function getLineBefore(text: string, pos: number): string { | ||
| const lineNumber = getLineNumberAtPos(text, pos); | ||
| return getLine(text, lineNumber - 1); | ||
| } | ||
|
|
||
| export function getLineAfter(text: string, pos: number): string { | ||
| const lineNumber = getLineNumberAtPos(text, pos); | ||
| return getLine(text, lineNumber + 1); | ||
| } |
There was a problem hiding this comment.
Useful functions for navigating text based on position.
The getLineBefore and getLineAfter functions effectively leverage getLineNumberAtPos and getLine to retrieve the lines before and after a specific position. These functions are crucial for features that need to navigate text based on its position within the editor.
Suggestion for Improvement:
Consider adding error handling for cases where the position is at the start or end of the text to prevent potential out-of-range errors.
| { messages, stream, onDataReceived, onResponseComplete }: StreamChatOptions, | ||
| ): Promise<any> { | ||
| return await this.openaiProvider.chatWithAI({ | ||
| messages, | ||
| stream, | ||
| onDataReceived, | ||
| onResponseComplete, |
There was a problem hiding this comment.
Approve the addition of onResponseComplete, suggest specifying return type.
The integration of onResponseComplete in the chatWithAI method is correctly implemented, enhancing the method's functionality by allowing additional callback handling.
However, consider specifying a more detailed return type than Promise<any> to improve type safety and clarity.
| onResponseComplete?: (data: any) => void; | ||
| postProcessors?: string[]; |
There was a problem hiding this comment.
Approve additions, suggest more specific type for onResponseComplete.
The additions of onResponseComplete and postProcessors to StreamChatOptions are well-aligned with the PR objectives and enhance the functionality significantly.
However, consider using a more specific type than any for the onResponseComplete parameter to improve type safety and clarity.
Initial support for the use-cases described in #59
insertAt now allows these values:
"cursor",
"page-start",
"page-end",
"start-of-line",
"end-of-line",
"start-of-item",
"end-of-item",
"new-line-above",
"new-line-below",
"replace-line",
Templated Prompts also now have a new
postProcessorsoption. It's a list of space-script functions. After the template prompt is executed and finished streaming, the full response is sent to each post processor function. That post processor function can then manipulate the response data if needed and return the modified version.As a demo, I'm using it to ensure the response from the llm is a list of markdown tasks and also indented as subtasks:
Summary by CodeRabbit
New Features
aiSplitTodocommand.Improvements
Bug Fixes
Documentation