byterover-cli 0.2.1 → 0.3.0

package/dist/infra/parsers/clean/clean-copilot-service.d.ts

@@ -0,0 +1,255 @@
+/**
+ * Copilot Clean Service
+ * Transforms GitHub Copilot raw parsed data to clean normalized format
+ */
+import type { CleanSession } from '../../../core/domain/entities/parser.js';
+import { Agent } from '../../../core/domain/entities/agent.js';
+import { CleanCopilotProcessResult } from '../../../core/domain/entities/parser.js';
+import { ICleanParserService } from '../../../core/interfaces/parser/i-clean-parser-service.js';
+/**
+ * Copilot Clean Service
+ * Transforms GitHub Copilot raw parsed sessions to clean normalized format
+ */
+export declare class CopilotCleanService implements ICleanParserService {
+    private ide;
+    constructor(ide: Agent);
+    /**
+     * Parse and transform GitHub Copilot raw sessions to clean normalized format
+     *
+     * Reads Copilot raw session files organized by workspace, transforms them using
+     * Claude-compatible format, and normalizes the sessions.
+     * Returns sessions in-memory without file writing.
+     *
+     * @param rawDir - Absolute path to the directory containing raw Copilot session files organized by workspace
+     * @returns Promise resolving to array of clean normalized sessions
+     */
+    parse(rawDir: string): Promise<readonly CleanSession[]>;
+    /**
+     * Process Copilot data by analyzing and transforming valid workspaces
+     *
+     * Performs complete workspace analysis and transformation:
+     * 1. Analyzes all workspaces to identify valid Copilot sessions
+     * 2. Logs workspace validity analysis with session counts
+     * 3. Copies and transforms valid workspaces to Claude-compatible format
+     * Returns statistics about processed workspaces.
+     *
+     * @param rawResultsDir - Absolute path to raw Copilot results directory
+     * @param cleanResultsDir - Absolute path to clean results output directory
+     * @returns Promise with statistics: total workspaces, valid count, invalid count
+     */
+    processCopilotData(rawResultsDir: string, cleanResultsDir: string): Promise<CleanCopilotProcessResult>;
+    /**
+     * Analyze Copilot workspaces to identify valid vs invalid sessions
+     *
+     * Scans workspace directories and validates each by checking for valid Copilot session
+     * files. A workspace is considered valid if it contains at least one session file with
+     * non-empty requests array. Returns a map of workspace information including validity
+     * status, valid/empty session counts, and reason.
+     *
+     * @param copilotResultsDir - Absolute path to directory containing Copilot workspace subdirectories
+     * @returns Promise that resolves to a map of workspace names to WorkspaceInfo objects
+     */
+    private analyzeCopilotWorkspaces;
+    /**
+     * Build session metadata from Copilot session data
+     *
+     * Extracts and formats metadata from Copilot session including message/request counts,
+     * usernames, session ID, and duration. Combines data from transformed messages and
+     * original session metadata.
+     *
+     * @param transformedMessages - Array of transformed message objects
+     * @param copilotSession - Original Copilot session object with metadata
+     * @returns Metadata object with counts, usernames, session ID, and duration
+     */
+    private buildSessionMetadata;
+    /**
+     * Copy and transform valid Copilot workspace to clean results directory
+     *
+     * Processes all JSON files in a valid workspace, transforming valid Copilot sessions
+     * (those with non-empty requests) to Claude-compatible format. Applies full normalization
+     * pipeline and splits multi-content assistant messages. Creates target directory as needed.
+     *
+     * @param workspaceHash - Hash identifier of the workspace directory
+     * @param sourceDir - Absolute path to parent directory containing workspace directories
+     * @param targetDir - Absolute path to clean results output directory
+     * @returns Promise that resolves when all files are processed and copied
+     */
+    private copyValidCopilotWorkspace;
+    /**
+     * Count valid Copilot session files in a workspace directory
+     *
+     * Iterates through provided JSON files and counts how many are valid Copilot sessions
+     * (have non-empty requests array). Used for workspace analysis and validation.
+     *
+     * @param workspacePath - Absolute path to the workspace directory
+     * @param jsonFiles - Array of JSON filename strings to validate
+     * @returns Promise that resolves to the count of valid Copilot session files
+     */
+    private countValidCopilotSessions;
+    /**
+     * Create assistant message from Copilot request response
+     *
+     * Extracts response data from Copilot request and builds an assistant message:
+     * 1. Converts response text to thinking block
+     * 2. Extracts tool invocations with arguments and results
+     * Returns null if no content could be extracted.
+     *
+     * @param req - Copilot request object with response and result data
+     * @returns Assistant message object with content array, or null if no valid content
+     */
+    private createAssistantMessageFromRequest;
+    /**
+     * Create user message from Copilot request
+     *
+     * Extracts user message text from request and builds a user message object.
+     * Includes attachments from variableData if present (file references).
+     * Returns null if no message text found.
+     *
+     * @param req - Copilot request object with message and variableData
+     * @returns User message object with text content and optional attachments, or null
+     */
+    private createUserMessageFromRequest;
+    /**
+     * Extract file attachments from Copilot variableData
+     *
+     * Scans the variableData object for file variables and extracts their names.
+     * File variables are identified by having kind='file'. Returns an array of filenames
+     * representing files attached to a Copilot request/message.
+     *
+     * @param variableData - Object containing variables array with file references
+     * @returns Array of file attachment names extracted from variables
+     */
+    private extractCopilotAttachments;
+    /**
+     * Extract text content from Copilot response items
+     *
+     * Filters response items to find main text responses (items without kind or with kind='unknown')
+     * and concatenates their values into a single text string. Skips prepareToolInvocation items
+     * which are used for tool setup rather than content. Returns trimmed concatenated text.
+     *
+     * @param responseItems - Array of response item objects with kind and value properties
+     * @returns Concatenated text content from non-tool response items
+     */
+    private extractCopilotResponseText;
+    /**
+     * Extract tool invocations and their results from Copilot response
+     *
+     * Processes toolInvocationSerialized response items and matches them with detailed tool calls
+     * from toolCallRounds and results from toolCallResults. Constructs tool_use content blocks
+     * with parsed arguments (converting from string JSON when needed) and associated output/results.
+     * Matches sequentially: the nth toolInvocationSerialized corresponds to the nth toolCall
+     * in the collected toolCalls array.
+     *
+     * @param responseItems - Array of response item objects, filtering for toolInvocationSerialized kind
+     * @param toolCallRounds - Optional array of tool call rounds, each containing toolCalls array
+     * @param toolCallResults - Optional map of tool results keyed by tool call ID
+     * @returns Array of tool_use content blocks with input arguments and output results
+     */
+    private extractCopilotToolInvocations;
+    /**
+     * Extract valid workspace path from object's baseUri property
+     *
+     * Checks if an object contains a baseUri with a valid path property. Path must be
+     * a string starting with '/' and must not have been seen before (checked against seenPaths set).
+     * Adds valid paths to seenPaths to prevent duplicates. Used for workspace path extraction
+     * from nested Copilot session data structures.
+     *
+     * @param obj - Object that may contain baseUri.path structure
+     * @param seenPaths - Set of previously seen paths to avoid duplicates
+     * @returns Valid path string if found and not previously seen, null otherwise
+     */
+    private extractPathFromBaseUri;
+    /**
+     * Extract readable text from deeply nested VS Code/Copilot object structures
+     *
+     * Recursively traverses nested object trees commonly found in Copilot tool results,
+     * extracting all text values from nodes. Handles multiple text source patterns:
+     * 1. Direct text property on nodes
+     * 2. String values in value property
+     * 3. Nested structures in children array
+     * 4. Nested structures in node property
+     * Returns concatenated text from all found sources, trimmed.
+     *
+     * @param obj - Nested object structure to traverse (may be object, array, or primitive)
+     * @returns Concatenated text content extracted from all nested text sources
+     */
+    private extractTextFromNestedObject;
+    /**
+     * Extract human-readable text content from Copilot tool result objects
+     *
+     * Handles multiple result format patterns commonly returned by Copilot tools:
+     * 1. Array-based content: Simple string extraction from value properties (e.g., run_in_terminal)
+     * 2. Array-based content: Fallback to nested object extraction (e.g., copilot_readFile)
+     * 3. Object-based content: Traverses nested structures for text values
+     * Prefers simple extraction if content found, falls back to complex nested traversal.
+     * Used to extract readable output from tool invocation results.
+     *
+     * @param result - Tool result object with content property (array, object, or nested structure)
+     * @returns Extracted text content from the result, empty string if no content found
+     */
+    private extractToolResultContent;
+    /**
+     * Extract workspace path from Copilot session data structures
+     *
+     * Recursively searches through all properties and nested objects in a Copilot session
+     * to find workspace paths. Looks for baseUri.path patterns which contain absolute paths
+     * starting with '/'. Returns the first valid path found. Uses a seenPaths set internally
+     * to avoid returning duplicate paths if multiple workspace paths exist.
+     *
+     * @param copilotSession - Copilot session object containing nested data structures with baseUri patterns
+     * @returns First valid workspace path found (absolute path starting with '/'), or null if not found
+     */
+    private extractWorkspacePathFromCopilot;
+    /**
+     * Validate if a Copilot session file contains valid session data
+     *
+     * Reads and parses a JSON file, checking if it contains a non-empty requests array.
+     * A session is considered valid if requests field exists and contains at least one request.
+     * Used during workspace analysis to identify which session files have meaningful data.
+     * Returns false silently if file cannot be read or parsed.
+     *
+     * @param filePath - Absolute path to JSON session file
+     * @returns Promise resolving to true if file contains valid requests, false otherwise
+     */
+    private isValidCopilotSession;
+    /**
+     * Normalize workspace paths from Copilot session to standard array format
+     *
+     * Ensures workspace paths are always returned as a string array. Handles multiple input formats:
+     * 1. Already-array workspace paths (monorepo with multiple paths)
+     * 2. String workspace paths (single workspace)
+     * 3. Extracted workspace paths via deep extraction if not found in top-level workspacePath property
+     * Returns empty array if no workspace paths found. Provides fallback extraction mechanism for
+     * complex Copilot session structures where workspace info is nested.
+     *
+     * @param copilotSession - Copilot session object which may have workspacePath property or nested paths
+     * @returns Array of workspace paths (may be empty if no paths found)
+     */
+    private normalizeWorkspacePaths;
+    /**
+     * Split all messages in a session that have multiple content blocks
+     *
+     * Copilot assistant messages often contain multiple content blocks (thinking + tool_use)
+     * in a single message. This function splits such messages into separate messages, one per
+     * content block, to create a cleaner, more normalized message sequence.
+     *
+     * Processing logic:
+     * - User messages: Preserved as-is (no splitting)
+     * - Assistant messages with single content block: Preserved as-is
+     * - Assistant messages with multiple content blocks: Split into separate messages,
+     *   one per content block, each with the same timestamp
+     *
+     * After splitting, recalculates turn_id for all messages to maintain proper sequential
+     * numbering (1, 2, 3, ...). This ensures the message sequence remains valid after the
+     * splitting transformation.
+     *
+     * @param session - Copilot session object with messages array to be split
+     * @returns New session object with split messages and recalculated turn_ids
+     */
+    private splitAllCopilotContent;
+    /**
+     * Transform Copilot session to match Claude format
+     * Uses the requests array which contains detailed conversation structure
+     */
+    private transformCopilotToClaudeFormat;
+}