@gswangg/duncan-cc 0.1.0

package/src/parser.ts

@@ -0,0 +1,257 @@
+/**
+ * CC Session JSONL Parser
+ * 
+ * Parses Claude Code session files into structured entries.
+ * Separates transcript messages from metadata entries.
+ * 
+ * Equivalent to CC's mu() + G26() separation logic.
+ */
+
+// ============================================================================
+// Types
+// ============================================================================
+
+export interface CCMessage {
+  uuid: string;
+  parentUuid: string | null;
+  session_id?: string;
+  type: "user" | "assistant" | "system" | "progress" | "attachment";
+  timestamp: string;
+  isSidechain?: boolean;
+  isMeta?: boolean;
+  isVisibleInTranscriptOnly?: boolean;
+  isCompactSummary?: boolean;
+  isApiErrorMessage?: boolean;
+  apiError?: string;
+  requestId?: string;
+  parent_tool_use_id?: string | null;
+  parentToolUseID?: string | null;
+  toolUseID?: string | null;
+  toolUseResult?: string;
+  sourceToolAssistantUUID?: string;
+  imagePasteIds?: string[];
+  permissionMode?: string;
+  origin?: { kind: string };
+  gitBranch?: string;
+  teamName?: string;
+  cwd?: string;
+  subtype?: string;
+  content?: string;
+  compactMetadata?: {
+    preservedSegment?: {
+      headUuid: string;
+      tailUuid: string;
+      anchorUuid: string;
+    };
+  };
+  attachment?: any;
+  message: {
+    role: string;
+    content: string | any[];
+    model?: string;
+    usage?: any;
+    id?: string;
+    type?: string;
+    stop_reason?: string | null;
+    stop_sequence?: string | null;
+  };
+  // Additional fields we pass through
+  [key: string]: any;
+}
+
+export interface SummaryEntry {
+  type: "summary";
+  leafUuid: string;
+  summary: string;
+}
+
+export interface ContentReplacementEntry {
+  type: "content-replacement";
+  sessionId?: string;
+  agentId?: string;
+  replacements: Array<{
+    kind: string;
+    toolUseId: string;
+    replacement: string;
+  }>;
+}
+
+export interface MetadataEntry {
+  type: string;
+  sessionId?: string;
+  [key: string]: any;
+}
+
+export interface ParsedSession {
+  messages: Map<string, CCMessage>;
+  summaries: Map<string, string>;          // leafUuid → summary text
+  customTitles: Map<string, string>;       // sessionId → title
+  tags: Map<string, string>;              // sessionId → tag
+  agentNames: Map<string, string>;        // sessionId → name
+  agentColors: Map<string, string>;       // sessionId → color
+  agentSettings: Map<string, string>;     // sessionId → setting
+  modes: Map<string, string>;            // sessionId → mode
+  contentReplacements: Map<string, ContentReplacementEntry["replacements"]>; // sessionId → replacements
+  contextCollapseCommits: any[];
+  contextCollapseSnapshot: any | null;
+}
+
+// ============================================================================
+// Entry type checks — mirrors CC's mi(), of(), Ns6()
+// ============================================================================
+
+/** Transcript message check — CC's mi() */
+export function isTranscriptMessage(entry: any): entry is CCMessage {
+  return (
+    entry.type === "user" ||
+    entry.type === "assistant" ||
+    entry.type === "attachment" ||
+    entry.type === "system" ||
+    entry.type === "progress"
+  );
+}
+
+/** Compact boundary check — CC's of() */
+export function isCompactBoundary(entry: any): boolean {
+  return entry?.type === "system" && entry.subtype === "compact_boundary";
+}
+
+/** Ephemeral progress types — CC's Ns6() */
+const EPHEMERAL_PROGRESS_TYPES = new Set([
+  "bash_progress",
+  "powershell_progress",
+  "mcp_progress",
+]);
+
+export function isEphemeralProgress(type: string): boolean {
+  return typeof type === "string" && EPHEMERAL_PROGRESS_TYPES.has(type);
+}
+
+/** API error message check — CC's Lt1() */
+const INTERNAL_ERROR_MODEL = "internal_error";
+
+export function isApiErrorMessage(entry: any): boolean {
+  return (
+    entry.type === "assistant" &&
+    entry.isApiErrorMessage === true &&
+    entry.message?.model === INTERNAL_ERROR_MODEL
+  );
+}
+
+/** Local command system message check — CC's gp1() */
+export function isLocalCommand(entry: any): boolean {
+  return entry.type === "system" && entry.subtype === "local_command";
+}
+
+// ============================================================================
+// JSONL Parser — mirrors CC's mu()
+// ============================================================================
+
+export function parseJsonl(content: string | Buffer): any[] {
+  const text = typeof content === "string" ? content : content.toString("utf-8");
+  const results: any[] = [];
+  let pos = 0;
+  const len = text.length;
+
+  while (pos < len) {
+    let end = text.indexOf("\n", pos);
+    if (end === -1) end = len;
+    const line = text.substring(pos, end).trim();
+    pos = end + 1;
+    if (!line) continue;
+    try {
+      results.push(JSON.parse(line));
+    } catch {
+      // skip unparseable lines
+    }
+  }
+  return results;
+}
+
+// ============================================================================
+// Session Parser — mirrors CC's G26() separation logic
+// ============================================================================
+
+export function parseSession(content: string | Buffer): ParsedSession {
+  const entries = parseJsonl(content);
+  
+  const messages = new Map<string, CCMessage>();
+  const summaries = new Map<string, string>();
+  const customTitles = new Map<string, string>();
+  const tags = new Map<string, string>();
+  const agentNames = new Map<string, string>();
+  const agentColors = new Map<string, string>();
+  const agentSettings = new Map<string, string>();
+  const modes = new Map<string, string>();
+  const contentReplacements = new Map<string, ContentReplacementEntry["replacements"]>();
+  const contextCollapseCommits: any[] = [];
+  let contextCollapseSnapshot: any | null = null;
+
+  for (const entry of entries) {
+    if (isTranscriptMessage(entry)) {
+      // Skip ephemeral progress messages
+      if (
+        entry.type === "progress" &&
+        entry.data &&
+        typeof entry.data === "object" &&
+        "type" in entry.data &&
+        isEphemeralProgress(entry.data.type as string)
+      ) {
+        continue;
+      }
+
+      // Strip normalizedMessages from progress data (save memory)
+      if (
+        entry.type === "progress" &&
+        entry.data &&
+        typeof entry.data === "object" &&
+        "normalizedMessages" in entry.data &&
+        Array.isArray(entry.data.normalizedMessages) &&
+        entry.data.normalizedMessages.length > 0
+      ) {
+        entry.data.normalizedMessages = [];
+      }
+
+      messages.set(entry.uuid, entry as CCMessage);
+    } else if (entry.type === "summary" && entry.leafUuid) {
+      summaries.set(entry.leafUuid, entry.summary);
+    } else if (entry.type === "custom-title" && entry.sessionId) {
+      customTitles.set(entry.sessionId, entry.customTitle);
+    } else if (entry.type === "tag" && entry.sessionId) {
+      tags.set(entry.sessionId, entry.tag);
+    } else if (entry.type === "agent-name" && entry.sessionId) {
+      agentNames.set(entry.sessionId, entry.agentName);
+    } else if (entry.type === "agent-color" && entry.sessionId) {
+      agentColors.set(entry.sessionId, entry.agentColor);
+    } else if (entry.type === "agent-setting" && entry.sessionId) {
+      agentSettings.set(entry.sessionId, entry.agentSetting);
+    } else if (entry.type === "mode" && entry.sessionId) {
+      modes.set(entry.sessionId, entry.mode);
+    } else if (entry.type === "content-replacement") {
+      const key = entry.agentId || entry.sessionId;
+      if (key) {
+        const existing = contentReplacements.get(key) ?? [];
+        existing.push(...entry.replacements);
+        contentReplacements.set(key, existing);
+      }
+    } else if (entry.type === "marble-origami-commit") {
+      contextCollapseCommits.push(entry);
+    } else if (entry.type === "marble-origami-snapshot") {
+      contextCollapseSnapshot = entry;
+    }
+  }
+
+  return {
+    messages,
+    summaries,
+    customTitles,
+    tags,
+    agentNames,
+    agentColors,
+    agentSettings,
+    modes,
+    contentReplacements,
+    contextCollapseCommits,
+    contextCollapseSnapshot,
+  };
+}

package/src/pipeline.ts

@@ -0,0 +1,274 @@
+/**
+ * Full Pipeline Integration
+ * 
+ * Wires all layers together:
+ * parse → relink → walk → slice → strip → normalize → 
+ * content-replace → microcompact → userContext → API format
+ * 
+ * Produces the final messages array ready for an API call.
+ */
+
+import { readFileSync } from "node:fs";
+import { parseSession, type ParsedSession } from "./parser.js";
+import { buildRawChain, sliceFromBoundary, stripInternalFields, getCompactionWindows, type CompactionWindow } from "./tree.js";
+import { normalizeMessages } from "./normalize.js";
+import { applyContentReplacements, microcompact } from "./content-replacements.js";
+import { injectUserContext, buildSystemPromptString, extractToolNames, type SystemPromptOptions } from "./system-prompt.js";
+import type { CCMessage } from "./parser.js";
+
+// ============================================================================
+// API Format Conversion
+// ============================================================================
+
+interface ApiMessage {
+  role: "user" | "assistant";
+  content: string | any[];
+}
+
+/**
+ * Convert a CC message to API format — strip everything except role + content.
+ * Converts internal messages to API format ({role, content} only).
+ */
+function toApiMessage(msg: CCMessage): ApiMessage {
+  return {
+    role: msg.type === "assistant" ? "assistant" : "user",
+    content: Array.isArray(msg.message.content)
+      ? [...msg.message.content]
+      : msg.message.content,
+  };
+}
+
+/**
+ * Convert an array of CC messages to API format.
+ */
+export function toApiMessages(messages: CCMessage[]): ApiMessage[] {
+  return messages.map(toApiMessage);
+}
+
+// ============================================================================
+// Pipeline Options
+// ============================================================================
+
+export interface PipelineOptions {
+  /** Working directory the session was run from */
+  cwd?: string;
+  /** Apply content replacements (default: true) */
+  applyReplacements?: boolean;
+  /** Apply microcompact (default: true) */
+  applyMicrocompact?: boolean;
+  /** Microcompact gap threshold in minutes (default: 30) */
+  microcompactGapMinutes?: number;
+  /** Microcompact: number of recent turns to keep (default: 2) */
+  microcompactKeepTurns?: number;
+  /** Inject userContext (CLAUDE.md + date) (default: true) */
+  injectContext?: boolean;
+  /** Skip system prompt building (default: false) */
+  skipSystemPrompt?: boolean;
+  /** CC project directory (~/.claude/projects/<hash>/) for memory loading */
+  projectDir?: string | null;
+}
+
+// ============================================================================
+// Pipeline Result
+// ============================================================================
+
+export interface PipelineResult {
+  /** Messages ready for the API (role + content only) */
+  messages: ApiMessage[];
+  /** System prompt string */
+  systemPrompt: string;
+  /** Model info extracted from session */
+  modelInfo?: { provider: string; modelId: string };
+  /** Number of messages before normalization */
+  rawMessageCount: number;
+  /** Session CWD (extracted from messages) */
+  sessionCwd: string;
+}
+
+// ============================================================================
+// Full Pipeline
+// ============================================================================
+
+/**
+ * Run the full pipeline on a session file.
+ * Returns API-ready messages + system prompt.
+ */
+export function processSessionFile(sessionFile: string, opts: PipelineOptions = {}): PipelineResult {
+  const content = readFileSync(sessionFile, "utf-8");
+  return processSessionContent(content, sessionFile, opts);
+}
+
+/**
+ * Run the full pipeline on session content (string).
+ */
+export function processSessionContent(
+  content: string,
+  sessionFile?: string,
+  opts: PipelineOptions = {},
+): PipelineResult {
+  const parsed = parseSession(content);
+  return processSession(parsed, sessionFile, opts);
+}
+
+/**
+ * Run the full pipeline on a parsed session.
+ */
+export function processSession(
+  parsed: ParsedSession,
+  sessionFile?: string,
+  opts: PipelineOptions = {},
+): PipelineResult {
+  // 1. Build raw chain (relink + tree walk)
+  const chain = buildRawChain(parsed);
+  if (chain.length === 0) {
+    return {
+      messages: [],
+      systemPrompt: "",
+      rawMessageCount: 0,
+      sessionCwd: opts.cwd ?? process.cwd(),
+    };
+  }
+
+  // Extract CWD from session messages
+  const sessionCwd = opts.cwd ?? extractCwd(chain) ?? process.cwd();
+
+  // Extract model info
+  const modelInfo = extractModelInfo(chain);
+
+  // 2. Slice from last boundary
+  let messages = sliceFromBoundary(chain);
+
+  // 3. Strip internal fields
+  messages = stripInternalFields(messages);
+
+  // 4. Normalize (filter, convert, merge, post-transform)
+  messages = normalizeMessages(messages);
+
+  // 5. Content replacements
+  if (opts.applyReplacements !== false) {
+    messages = applyContentReplacements(messages, parsed, sessionFile);
+  }
+
+  // 6. Microcompact
+  if (opts.applyMicrocompact !== false) {
+    messages = microcompact(
+      messages,
+      opts.microcompactGapMinutes ?? 30,
+      opts.microcompactKeepTurns ?? 2,
+    );
+  }
+
+  // 7. Inject userContext
+  if (opts.injectContext !== false) {
+    messages = injectUserContext(messages, sessionCwd);
+  }
+
+  // 8. Build system prompt (full parity with CC's U2)
+  const toolNames = extractToolNames(messages);
+  const systemPrompt = opts.skipSystemPrompt
+    ? ""
+    : buildSystemPromptString({
+        cwd: sessionCwd,
+        modelId: modelInfo?.modelId,
+        toolNames,
+        projectDir: opts.projectDir ?? null,
+      });
+
+  // 9. Convert to API format
+  const apiMessages = toApiMessages(messages);
+
+  return {
+    messages: apiMessages,
+    systemPrompt,
+    modelInfo,
+    rawMessageCount: chain.length,
+    sessionCwd,
+  };
+}
+
+// ============================================================================
+// Compaction Window Pipeline
+// ============================================================================
+
+export interface WindowPipelineResult extends PipelineResult {
+  windowIndex: number;
+}
+
+/**
+ * Process a session into compaction windows, each independently queryable.
+ */
+export function processSessionWindows(
+  sessionFile: string,
+  opts: PipelineOptions = {},
+): WindowPipelineResult[] {
+  const content = readFileSync(sessionFile, "utf-8");
+  const parsed = parseSession(content);
+  const chain = buildRawChain(parsed);
+
+  if (chain.length === 0) return [];
+
+  const windows = getCompactionWindows(chain);
+  const sessionCwd = opts.cwd ?? extractCwd(chain) ?? process.cwd();
+
+  return windows.map((window) => {
+    let messages = stripInternalFields(window.messages);
+    messages = normalizeMessages(messages);
+
+    if (opts.applyReplacements !== false) {
+      messages = applyContentReplacements(messages, parsed, sessionFile);
+    }
+
+    if (opts.applyMicrocompact !== false) {
+      messages = microcompact(
+        messages,
+        opts.microcompactGapMinutes ?? 30,
+        opts.microcompactKeepTurns ?? 2,
+      );
+    }
+
+    if (opts.injectContext !== false) {
+      messages = injectUserContext(messages, sessionCwd);
+    }
+
+    const modelInfo = window.modelInfo;
+    const systemPrompt = opts.skipSystemPrompt
+      ? ""
+      : buildSystemPromptString({
+          cwd: sessionCwd,
+          modelId: modelInfo?.modelId,
+          modelName: modelInfo?.modelId,
+        });
+
+    return {
+      windowIndex: window.windowIndex,
+      messages: toApiMessages(messages),
+      systemPrompt,
+      modelInfo,
+      rawMessageCount: window.messages.length,
+      sessionCwd,
+    };
+  });
+}
+
+// ============================================================================
+// Helpers
+// ============================================================================
+
+function extractCwd(chain: CCMessage[]): string | undefined {
+  // Try to find cwd from messages (most messages have a cwd field)
+  for (const msg of chain) {
+    if (msg.cwd) return msg.cwd;
+  }
+  return undefined;
+}
+
+function extractModelInfo(chain: CCMessage[]): { provider: string; modelId: string } | undefined {
+  // Find the last assistant message with a model
+  for (let i = chain.length - 1; i >= 0; i--) {
+    const msg = chain[i];
+    if (msg.type === "assistant" && msg.message.model) {
+      return { provider: "anthropic", modelId: msg.message.model };
+    }
+  }
+  return undefined;
+}