@towles/tool 0.0.41 → 0.0.48

package/src/commands/graph.ts

@@ -0,0 +1,970 @@
+import { Flags } from "@oclif/core";
+import { DateTime } from "luxon";
+import * as fs from "node:fs";
+import * as http from "node:http";
+import * as os from "node:os";
+import * as path from "node:path";
+import { fileURLToPath } from "node:url";
+import { x } from "tinyexec";
+import { BaseCommand } from "./base.js";
+
+// Load HTML template from file (resolved relative to this module)
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+const TEMPLATE_PATH = path.join(__dirname, "graph-template.html");
+
+// Types for parsing Claude Code session JSONL files
+interface ContentBlock {
+  type: string;
+  text?: string;
+  id?: string;
+  name?: string;
+  input?: Record<string, unknown>;
+}
+
+interface JournalEntry {
+  type: string;
+  sessionId: string;
+  timestamp: string;
+  message?: {
+    role: "user" | "assistant";
+    model?: string;
+    usage?: {
+      input_tokens?: number;
+      output_tokens?: number;
+      cache_read_input_tokens?: number;
+      cache_creation_input_tokens?: number;
+    };
+    content?: ContentBlock[] | string;
+  };
+  uuid?: string;
+}
+
+interface ToolData {
+  name: string;
+  detail?: string;
+  inputTokens: number;
+  outputTokens: number;
+}
+
+// Bar chart types for stacked bar visualization - aggregated by project
+export interface ProjectBar {
+  project: string;
+  totalTokens: number;
+}
+
+export interface BarChartDay {
+  date: string; // YYYY-MM-DD format
+  projects: ProjectBar[];
+}
+
+export interface BarChartData {
+  days: BarChartDay[];
+}
+
+export interface SessionResult {
+  sessionId: string;
+  path: string;
+  date: string;
+  tokens: number;
+  project: string;
+  mtime: number;
+}
+
+/**
+ * Calculate cutoff timestamp for days filtering.
+ * Returns 0 if days <= 0 (no filtering).
+ */
+export function calculateCutoffMs(days: number): number {
+  return days > 0 ? Date.now() - days * 24 * 60 * 60 * 1000 : 0;
+}
+
+/**
+ * Filter items by mtime against a days cutoff.
+ * Returns all items if days <= 0.
+ */
+export function filterByDays<T extends { mtime: number }>(items: T[], days: number): T[] {
+  const cutoff = calculateCutoffMs(days);
+  if (cutoff === 0) return items;
+  return items.filter((item) => item.mtime >= cutoff);
+}
+
+/**
+ * Parse JSONL file into JournalEntry array.
+ */
+export function parseJsonl(filePath: string): JournalEntry[] {
+  const content = fs.readFileSync(filePath, "utf-8");
+  const entries: JournalEntry[] = [];
+
+  for (const line of content.split("\n")) {
+    if (!line.trim()) continue;
+    try {
+      entries.push(JSON.parse(line) as JournalEntry);
+    } catch {
+      // Skip invalid lines
+    }
+  }
+
+  return entries;
+}
+
+/**
+ * Analyze session entries to get token breakdown by model.
+ */
+export function analyzeSession(entries: JournalEntry[]): {
+  inputTokens: number;
+  outputTokens: number;
+  opusTokens: number;
+  sonnetTokens: number;
+  haikuTokens: number;
+  cacheHitRate: number;
+  repeatedReads: number;
+  modelEfficiency: number;
+} {
+  let inputTokens = 0;
+  let outputTokens = 0;
+  let opusTokens = 0;
+  let sonnetTokens = 0;
+  let haikuTokens = 0;
+  let cacheRead = 0;
+  let totalInput = 0;
+  const fileReadCounts = new Map<string, number>();
+
+  for (const entry of entries) {
+    // Count file reads for repeatedReads metric
+    if (entry.message?.content && Array.isArray(entry.message.content)) {
+      for (const block of entry.message.content) {
+        if (block.type === "tool_use" && block.name === "Read" && block.input) {
+          const filePath = (block.input as { file_path?: string }).file_path;
+          if (filePath) {
+            fileReadCounts.set(filePath, (fileReadCounts.get(filePath) || 0) + 1);
+          }
+        }
+      }
+    }
+
+    if (!entry.message?.usage) continue;
+    const usage = entry.message.usage;
+    const model = entry.message.model || "";
+    const tokens = (usage.input_tokens || 0) + (usage.output_tokens || 0);
+
+    inputTokens += usage.input_tokens || 0;
+    outputTokens += usage.output_tokens || 0;
+    cacheRead += usage.cache_read_input_tokens || 0;
+    totalInput += usage.input_tokens || 0;
+
+    if (model.includes("opus")) opusTokens += tokens;
+    else if (model.includes("sonnet")) sonnetTokens += tokens;
+    else if (model.includes("haiku")) haikuTokens += tokens;
+  }
+
+  // Count files read more than once
+  let repeatedReads = 0;
+  for (const count of fileReadCounts.values()) {
+    if (count > 1) repeatedReads += count - 1;
+  }
+
+  const totalTokens = opusTokens + sonnetTokens + haikuTokens;
+
+  return {
+    inputTokens,
+    outputTokens,
+    opusTokens,
+    sonnetTokens,
+    haikuTokens,
+    cacheHitRate: totalInput > 0 ? cacheRead / totalInput : 0,
+    repeatedReads,
+    modelEfficiency: totalTokens > 0 ? opusTokens / totalTokens : 0,
+  };
+}
+
+/**
+ * Extract a meaningful label from session entries.
+ */
+export function extractSessionLabel(entries: JournalEntry[], sessionId: string): string {
+  let firstUserText: string | undefined;
+  let firstAssistantText: string | undefined;
+  let gitBranch: string | undefined;
+  let slug: string | undefined;
+
+  for (const entry of entries) {
+    // Extract metadata from any entry
+    if (!gitBranch && (entry as any).gitBranch) {
+      gitBranch = (entry as any).gitBranch;
+    }
+    if (!slug && (entry as any).slug) {
+      slug = (entry as any).slug;
+    }
+
+    if (!entry.message) continue;
+
+    // Look for first user message with actual text (not UUID reference)
+    if (!firstUserText && entry.type === "user" && entry.message.role === "user") {
+      const content = entry.message.content;
+      if (typeof content === "string") {
+        // Check if it's a UUID (skip those) or actual text
+        const isUuid = /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/i.test(
+          content,
+        );
+        if (!isUuid && content.length > 0) {
+          firstUserText = content;
+        }
+      } else if (Array.isArray(content)) {
+        // Look for text blocks in array content
+        for (const block of content) {
+          if (block.type === "text" && block.text && block.text.length > 0) {
+            firstUserText = block.text;
+            break;
+          }
+        }
+      }
+    }
+
+    // Look for first assistant text response
+    if (!firstAssistantText && entry.type === "assistant" && entry.message.role === "assistant") {
+      const content = entry.message.content;
+      if (Array.isArray(content)) {
+        for (const block of content) {
+          if (block.type === "text" && block.text && block.text.length > 0) {
+            firstAssistantText = block.text;
+            break;
+          }
+        }
+      }
+    }
+
+    // Stop early if we have user text
+    if (firstUserText) break;
+  }
+
+  // Priority: user text > assistant text > git branch > slug > short ID
+  let label = firstUserText || firstAssistantText || gitBranch || slug || sessionId.slice(0, 8);
+
+  // Clean up the label
+  label = label
+    .replace(/^\/\S+\s*/, "") // Remove /command prefixes
+    .replace(/<[^>]+>[^<]*<\/[^>]+>/g, "") // Remove XML-style tags with content
+    .replace(/<[^>]+>/g, "") // Remove remaining XML tags
+    .replace(/^\s*Caveat:.*$/m, "") // Remove caveat lines
+    .replace(/\n.*/g, "") // Take only first line
+    // eslint-disable-next-line no-control-regex
+    .replace(/[\x00-\x1F]+/g, " ") // Replace control characters with space
+    .trim();
+
+  // If still empty or too short, use fallback
+  if (label.length < 3) {
+    label = slug || sessionId.slice(0, 8);
+  }
+
+  // Truncate very long labels (will be smart-truncated in UI based on box size)
+  if (label.length > 80) {
+    label = label.slice(0, 77) + "...";
+  }
+
+  return label;
+}
+
+/**
+ * Build bar chart data structure from session results.
+ * Groups sessions by date and project folder, aggregating tokens per project per day.
+ */
+export function buildBarChartData(
+  sessions: SessionResult[],
+  extractProjectName: (encoded: string) => string,
+): BarChartData {
+  if (sessions.length === 0) {
+    return { days: [] };
+  }
+
+  // Group sessions by date, then by project
+  const byDateProject = new Map<string, Map<string, number>>();
+
+  for (const session of sessions) {
+    const project = extractProjectName(session.project);
+
+    if (!byDateProject.has(session.date)) {
+      byDateProject.set(session.date, new Map());
+    }
+    const projectMap = byDateProject.get(session.date)!;
+    projectMap.set(project, (projectMap.get(project) || 0) + session.tokens);
+  }
+
+  // Build days array sorted chronologically (oldest first for x-axis)
+  const sortedDates = [...byDateProject.keys()].sort();
+  const days: BarChartDay[] = sortedDates.map((date) => {
+    const projectMap = byDateProject.get(date)!;
+    // Sort projects by total tokens descending
+    const projects: ProjectBar[] = [...projectMap.entries()]
+      .map(([project, totalTokens]) => ({ project, totalTokens }))
+      .sort((a, b) => b.totalTokens - a.totalTokens);
+    return { date, projects };
+  });
+
+  return { days };
+}
+
+interface TreemapNode {
+  name: string;
+  value?: number;
+  children?: TreemapNode[];
+  // Metadata for tooltips
+  sessionId?: string;
+  fullSessionId?: string;
+  filePath?: string;
+  startTime?: string;
+  model?: string;
+  inputTokens?: number;
+  outputTokens?: number;
+  ratio?: number;
+  date?: string;
+  project?: string;
+  // Waste metrics
+  repeatedReads?: number;
+  modelEfficiency?: number; // Opus tokens / total tokens
+  // Tool data
+  tools?: ToolData[];
+  toolName?: string; // For coloring by tool type
+}
+
+/**
+ * Generate interactive HTML treemap from session token data
+ */
+export default class Graph extends BaseCommand {
+  static override description = "Generate interactive HTML treemap from session token data";
+
+  static override examples = [
+    {
+      description: "Generate treemap for all recent sessions",
+      command: "<%= config.bin %> <%= command.id %>",
+    },
+    {
+      description: "Generate treemap for a specific session",
+      command: "<%= config.bin %> <%= command.id %> --session abc123",
+    },
+    {
+      description: "Generate and auto-open in browser",
+      command: "<%= config.bin %> <%= command.id %> --open",
+    },
+  ];
+
+  static override flags = {
+    ...BaseCommand.baseFlags,
+    session: Flags.string({
+      char: "s",
+      description: "Session ID to analyze (shows all sessions if not provided)",
+    }),
+    open: Flags.boolean({
+      char: "o",
+      description: "Open treemap in browser after generating",
+      default: true,
+      allowNo: true,
+    }),
+    serve: Flags.boolean({
+      description: "Start local HTTP server to serve treemap (default: true)",
+      default: true,
+      allowNo: true,
+    }),
+    port: Flags.integer({
+      char: "p",
+      description: "Port for local server",
+      default: 8765,
+    }),
+    days: Flags.integer({
+      description: "Filter to sessions from last N days (0=no limit)",
+      default: 7,
+    }),
+  };
+
+  async run(): Promise<void> {
+    const { flags } = await this.parse(Graph);
+
+    const projectsDir = path.join(os.homedir(), ".claude", "projects");
+    if (!fs.existsSync(projectsDir)) {
+      this.error("No Claude projects directory found at ~/.claude/projects/");
+    }
+
+    const sessionId = flags.session;
+    let treemapData: TreemapNode;
+    let barChartData: BarChartData = { days: [] };
+
+    if (!sessionId) {
+      // All sessions mode
+      const sessions = this.findRecentSessions(projectsDir, 500, flags.days);
+      if (sessions.length === 0) {
+        this.error("No sessions found");
+      }
+
+      const daysMsg = flags.days > 0 ? ` (last ${flags.days} days)` : "";
+      this.log(`📊 Generating treemap for ${sessions.length} sessions${daysMsg}...`);
+      treemapData = this.buildAllSessionsTreemap(sessions);
+      barChartData = buildBarChartData(sessions, this.extractProjectName.bind(this));
+    } else {
+      // Single session mode
+      const sessionPath = this.findSessionPath(projectsDir, sessionId);
+      if (!sessionPath) {
+        this.error(`Session ${sessionId} not found`);
+      }
+
+      this.log(`📊 Generating treemap for session ${sessionId}...`);
+      const entries = this.parseJsonl(sessionPath);
+      treemapData = this.buildSessionTreemap(sessionId, entries);
+      // Bar chart not meaningful for single session, leave empty
+    }
+
+    // Generate HTML
+    const html = this.generateTreemapHtml(treemapData, barChartData);
+
+    // Write output file
+    const reportsDir = path.join(os.homedir(), ".claude", "reports");
+    if (!fs.existsSync(reportsDir)) {
+      fs.mkdirSync(reportsDir, { recursive: true });
+    }
+
+    const timestamp = DateTime.now().toFormat("yyyy-MM-dd'T'HH-mmZZZ");
+    const daysLabel = flags.days > 0 ? `${flags.days}d` : "all";
+    const filename = sessionId
+      ? `treemap-${sessionId.slice(0, 8)}-${timestamp}.html`
+      : `treemap-${daysLabel}-${timestamp}.html`;
+    const outputPath = path.join(reportsDir, filename);
+
+    fs.writeFileSync(outputPath, html);
+    this.log(`✓ Saved to ${outputPath}`);
+
+    if (flags.serve) {
+      // Start local HTTP server
+      const server = http.createServer((req, res) => {
+        // Serve the generated HTML file
+        if (req.url === "/" || req.url === `/${filename}`) {
+          res.writeHead(200, { "Content-Type": "text/html" });
+          res.end(html);
+        } else {
+          res.writeHead(404);
+          res.end("Not found");
+        }
+      });
+
+      // Try to start server, fallback to next port if in use
+      const startPort = flags.port;
+      const maxAttempts = 10;
+
+      const tryPort = (port: number): Promise<number> => {
+        return new Promise((resolve, reject) => {
+          const onError = (err: NodeJS.ErrnoException) => {
+            server.removeListener("listening", onListening);
+            if (err.code === "EADDRINUSE" && port < startPort + maxAttempts - 1) {
+              resolve(tryPort(port + 1));
+            } else {
+              reject(err);
+            }
+          };
+
+          const onListening = () => {
+            server.removeListener("error", onError);
+            resolve(port);
+          };
+
+          server.once("error", onError);
+          server.once("listening", onListening);
+          server.listen(port);
+        });
+      };
+
+      const tryListen = (): Promise<number> => tryPort(startPort);
+
+      const actualPort = await tryListen();
+      const url = `http://localhost:${actualPort}/`;
+      if (actualPort !== startPort) {
+        this.log(`\n⚠️  Port ${startPort} in use, using ${actualPort}`);
+      }
+      this.log(`🌐 Server running at ${url}`);
+      this.log("   Press Ctrl+C to stop\n");
+
+      if (flags.open) {
+        const openCmd = process.platform === "darwin" ? "open" : "xdg-open";
+        x(openCmd, [url]);
+      }
+
+      // Keep server running until Ctrl+C
+      await new Promise<void>((resolve) => {
+        process.on("SIGINT", () => {
+          this.log("\n👋 Stopping server...");
+          server.close();
+          resolve();
+        });
+      });
+    } else if (flags.open) {
+      this.log("\n📈 Opening treemap...");
+      const openCmd = process.platform === "darwin" ? "open" : "xdg-open";
+      await x(openCmd, [outputPath]);
+    }
+  }
+
+  private generateTreemapHtml(data: TreemapNode, barChartData: BarChartData): string {
+    const width = 1200;
+    const height = 800;
+
+    // Read template from file and replace placeholders
+    // Use function replacement to avoid special $& $' $` patterns in data being interpreted
+    const template = fs.readFileSync(TEMPLATE_PATH, "utf-8");
+    return template
+      .replace(/\{\{WIDTH\}\}/g, String(width))
+      .replace(/\{\{HEIGHT\}\}/g, String(height))
+      .replace(/\{\{DATA\}\}/g, () => JSON.stringify(data))
+      .replace(/\{\{BAR_CHART_DATA\}\}/g, () => JSON.stringify(barChartData));
+  }
+
+  private findRecentSessions(projectsDir: string, limit: number, days: number): SessionResult[] {
+    const sessions: Array<{
+      sessionId: string;
+      path: string;
+      date: string;
+      tokens: number;
+      project: string;
+      mtime: number;
+    }> = [];
+
+    const cutoffMs = calculateCutoffMs(days);
+
+    const projectDirs = fs.readdirSync(projectsDir);
+    for (const project of projectDirs) {
+      const projectPath = path.join(projectsDir, project);
+      if (!fs.statSync(projectPath).isDirectory()) continue;
+
+      const files = fs.readdirSync(projectPath).filter((f) => f.endsWith(".jsonl"));
+      for (const file of files) {
+        const filePath = path.join(projectPath, file);
+        const stat = fs.statSync(filePath);
+
+        // Filter by days if cutoff is set
+        if (cutoffMs > 0 && stat.mtimeMs < cutoffMs) continue;
+
+        const sessionId = file.replace(".jsonl", "");
+
+        // Quick token count from file
+        const tokens = this.quickTokenCount(filePath);
+
+        sessions.push({
+          sessionId,
+          path: filePath,
+          date: stat.mtime.toLocaleDateString("en-CA"), // YYYY-MM-DD in local timezone
+          tokens,
+          project,
+          mtime: stat.mtimeMs,
+        });
+      }
+    }
+
+    // Sort by modification time, most recent first
+    sessions.sort((a, b) => b.mtime - a.mtime);
+    return sessions.slice(0, limit);
+  }
+
+  private quickTokenCount(filePath: string): number {
+    try {
+      const content = fs.readFileSync(filePath, "utf-8");
+      let total = 0;
+      for (const line of content.split("\n")) {
+        if (!line.trim()) continue;
+        try {
+          const entry = JSON.parse(line) as JournalEntry;
+          if (entry.message?.usage) {
+            total +=
+              (entry.message.usage.input_tokens || 0) + (entry.message.usage.output_tokens || 0);
+          }
+        } catch {
+          // Skip invalid lines
+        }
+      }
+      return total;
+    } catch {
+      return 0;
+    }
+  }
+
+  private findSessionPath(projectsDir: string, sessionId: string): string | undefined {
+    const projectDirs = fs.readdirSync(projectsDir);
+    for (const project of projectDirs) {
+      const projectPath = path.join(projectsDir, project);
+      if (!fs.statSync(projectPath).isDirectory()) continue;
+
+      const jsonlPath = path.join(projectPath, `${sessionId}.jsonl`);
+      if (fs.existsSync(jsonlPath)) {
+        return jsonlPath;
+      }
+    }
+    return undefined;
+  }
+
+  private parseJsonl(filePath: string): JournalEntry[] {
+    return parseJsonl(filePath);
+  }
+
+  private buildSessionTreemap(sessionId: string, entries: JournalEntry[]): TreemapNode {
+    return {
+      name: `Session ${sessionId.slice(0, 8)}`,
+      children: this.buildTurnNodes(sessionId, entries),
+    };
+  }
+
+  /**
+   * Build turn-level nodes from session entries.
+   * Used by both single-session and all-sessions views.
+   */
+  private buildTurnNodes(
+    sessionId: string,
+    entries: JournalEntry[],
+    filePath?: string,
+  ): TreemapNode[] {
+    const children: TreemapNode[] = [];
+    let turnNumber = 0;
+
+    for (const entry of entries) {
+      if (entry.type !== "user" && entry.type !== "assistant") continue;
+      if (!entry.message) continue;
+
+      const role = entry.message.role;
+      const usage = entry.message.usage;
+      const model = entry.message.model;
+
+      if (role === "user") {
+        turnNumber++;
+      }
+
+      if (!usage) continue;
+
+      const inputTokens = usage.input_tokens || 0;
+      const outputTokens = usage.output_tokens || 0;
+      const totalTokens = inputTokens + outputTokens;
+
+      if (totalTokens === 0) continue;
+
+      const ratio = outputTokens > 0 ? inputTokens / outputTokens : inputTokens > 0 ? 999 : 0;
+
+      // Extract individual tool calls from content blocks
+      const tools = this.extractToolData(entry.message.content, inputTokens, outputTokens);
+
+      // Create individual tool children nodes
+      const toolChildren: TreemapNode[] = tools.map((tool) => ({
+        name: tool.detail ? `${tool.name}: ${tool.detail}` : tool.name,
+        value: tool.inputTokens + tool.outputTokens,
+        inputTokens: tool.inputTokens,
+        outputTokens: tool.outputTokens,
+        ratio: tool.outputTokens > 0 ? tool.inputTokens / tool.outputTokens : 0,
+        toolName: tool.name,
+      }));
+
+      // Format turn name based on tools used
+      let turnName: string;
+      let primaryToolName: string | undefined;
+      if (role === "user") {
+        turnName = `Turn ${turnNumber}: User`;
+      } else if (tools.length === 1) {
+        // Single tool: show tool name and detail
+        const t = tools[0];
+        turnName = t.detail ? `${t.name}: ${t.detail}` : t.name;
+        primaryToolName = t.name;
+      } else if (tools.length > 1) {
+        // Multiple tools: list unique tool names, primary is most common
+        const uniqueNames = [...new Set(tools.map((t) => t.name))];
+        turnName = uniqueNames.slice(0, 3).join(", ") + (uniqueNames.length > 3 ? "..." : "");
+        primaryToolName = tools[0].name; // Use first tool as primary
+      } else {
+        turnName = `Turn ${turnNumber}: Response`;
+        primaryToolName = "Response";
+      }
+      children.push({
+        name: turnName,
+        value: toolChildren.length > 0 ? undefined : totalTokens, // Let children sum if present
+        children: toolChildren.length > 0 ? toolChildren : undefined,
+        sessionId: sessionId.slice(0, 8),
+        fullSessionId: sessionId,
+        filePath,
+        toolName: primaryToolName,
+        model: this.getModelName(model),
+        inputTokens,
+        outputTokens,
+        ratio,
+        tools: tools.length > 0 ? tools : undefined,
+      });
+    }
+
+    return children;
+  }
+
+  /**
+   * Extract individual tool calls from message content blocks.
+   * Returns each tool call with its detail (file path, command, etc.).
+   */
+  private extractToolData(
+    content: ContentBlock[] | string | undefined,
+    turnInputTokens: number,
+    turnOutputTokens: number,
+  ): ToolData[] {
+    if (!content || typeof content === "string") return [];
+
+    // Collect individual tool_use blocks
+    const toolBlocks: Array<{ name: string; detail?: string }> = [];
+    for (const block of content) {
+      if (block.type === "tool_use" && block.name) {
+        const detail = this.extractToolDetail(block.name, block.input);
+        toolBlocks.push({ name: block.name, detail });
+      }
+    }
+
+    if (toolBlocks.length === 0) return [];
+
+    // Distribute tokens proportionally across individual calls
+    const tokensPerCall = {
+      input: Math.round(turnInputTokens / toolBlocks.length),
+      output: Math.round(turnOutputTokens / toolBlocks.length),
+    };
+
+    return toolBlocks.map((tool) => ({
+      name: tool.name,
+      detail: tool.detail,
+      inputTokens: tokensPerCall.input,
+      outputTokens: tokensPerCall.output,
+    }));
+  }
+
+  /**
+   * Extract a meaningful detail string from tool input.
+   */
+  private extractToolDetail(toolName: string, input?: Record<string, unknown>): string | undefined {
+    if (!input) return undefined;
+
+    switch (toolName) {
+      case "Read":
+        return this.truncateDetail(input.file_path as string);
+      case "Write":
+      case "Edit":
+        return this.truncateDetail(input.file_path as string);
+      case "Bash":
+        return this.truncateDetail(input.command as string, 50);
+      case "Glob":
+        return this.truncateDetail(input.pattern as string, 50);
+      case "Grep":
+        return this.truncateDetail(input.pattern as string, 50);
+      case "Task":
+        return this.truncateDetail(input.description as string, 50);
+      case "WebFetch":
+        return this.truncateDetail(input.url as string, 40);
+      default:
+        return undefined;
+    }
+  }
+
+  /**
+   * Sanitize string by replacing control characters (newlines, tabs, etc.) with spaces.
+   */
+  private sanitizeString(str: string): string {
+    // Replace all control characters (ASCII 0-31) with space, collapse multiple spaces
+    // eslint-disable-next-line no-control-regex
+    return str.replace(/[\x00-\x1F]+/g, " ").trim();
+  }
+
+  /**
+   * Truncate a string and extract just the filename for paths.
+   */
+  private truncateDetail(str: string | undefined, maxLen = 30): string | undefined {
+    if (!str) return undefined;
+    // Sanitize control characters first
+    const sanitized = this.sanitizeString(str);
+    // For file paths, show just the filename
+    if (sanitized.includes("/")) {
+      const parts = sanitized.split("/");
+      const filename = parts[parts.length - 1];
+      return filename.length > maxLen ? filename.slice(0, maxLen - 3) + "..." : filename;
+    }
+    return sanitized.length > maxLen ? sanitized.slice(0, maxLen - 3) + "..." : sanitized;
+  }
+
+  /**
+   * Aggregate tool usage across all entries in a session.
+   * Returns combined tool data for session-level tooltips (aggregated by name).
+   */
+  private aggregateSessionTools(entries: JournalEntry[]): ToolData[] {
+    const toolAgg = new Map<string, { count: number; inputTokens: number; outputTokens: number }>();
+
+    for (const entry of entries) {
+      if (!entry.message?.content || typeof entry.message.content === "string") continue;
+      if (!entry.message.usage) continue;
+
+      const inputTokens = entry.message.usage.input_tokens || 0;
+      const outputTokens = entry.message.usage.output_tokens || 0;
+      const turnTools = this.extractToolData(entry.message.content, inputTokens, outputTokens);
+
+      for (const tool of turnTools) {
+        const existing = toolAgg.get(tool.name);
+        if (existing) {
+          existing.count += 1;
+          existing.inputTokens += tool.inputTokens;
+          existing.outputTokens += tool.outputTokens;
+        } else {
+          toolAgg.set(tool.name, {
+            count: 1,
+            inputTokens: tool.inputTokens,
+            outputTokens: tool.outputTokens,
+          });
+        }
+      }
+    }
+
+    // Convert to array and sort by token usage
+    const tools: ToolData[] = [...toolAgg.entries()].map(([name, data]) => ({
+      name,
+      detail: `${data.count}x`,
+      inputTokens: data.inputTokens,
+      outputTokens: data.outputTokens,
+    }));
+    tools.sort((a, b) => b.inputTokens + b.outputTokens - (a.inputTokens + a.outputTokens));
+
+    return tools;
+  }
+
+  private buildAllSessionsTreemap(
+    sessions: Array<{
+      sessionId: string;
+      path: string;
+      date: string;
+      tokens: number;
+      project: string;
+    }>,
+  ): TreemapNode {
+    // Group sessions by project, then by date
+    const byProject = new Map<string, typeof sessions>();
+    for (const session of sessions) {
+      const projectName = this.extractProjectName(session.project);
+      if (!byProject.has(projectName)) {
+        byProject.set(projectName, []);
+      }
+      byProject.get(projectName)!.push(session);
+    }
+
+    // Sort projects by total tokens
+    const projectTotals = [...byProject.entries()].map(([name, sess]) => ({
+      name,
+      sessions: sess,
+      total: sess.reduce((sum, s) => sum + s.tokens, 0),
+    }));
+    projectTotals.sort((a, b) => b.total - a.total);
+
+    const projectChildren: TreemapNode[] = [];
+
+    for (const { name: projectName, sessions: projectSessions } of projectTotals) {
+      // Group by date within project
+      const byDate = new Map<string, typeof sessions>();
+      for (const session of projectSessions) {
+        if (!byDate.has(session.date)) {
+          byDate.set(session.date, []);
+        }
+        byDate.get(session.date)!.push(session);
+      }
+
+      // Sort dates (most recent first)
+      const sortedDates = [...byDate.keys()].sort().reverse();
+
+      const dateChildren: TreemapNode[] = [];
+
+      for (const date of sortedDates) {
+        const dateSessions = byDate.get(date)!;
+
+        const sessionChildren: TreemapNode[] = [];
+
+        for (const session of dateSessions) {
+          const entries = this.parseJsonl(session.path);
+          const analysis = this.analyzeSession(entries);
+          const label = this.extractSessionLabel(entries, session.sessionId);
+          const tools = this.aggregateSessionTools(entries);
+          const startTime = entries[0]?.timestamp
+            ? new Date(entries[0].timestamp).toLocaleTimeString()
+            : undefined;
+
+          // Build turn-level children for drill-down
+          const turnChildren = this.buildTurnNodes(session.sessionId, entries, session.path);
+
+          sessionChildren.push({
+            name: label,
+            // If we have turn children, let them sum; otherwise use session total
+            value: turnChildren.length > 0 ? undefined : session.tokens,
+            children: turnChildren.length > 0 ? turnChildren : undefined,
+            sessionId: session.sessionId.slice(0, 8),
+            fullSessionId: session.sessionId,
+            filePath: session.path,
+            startTime,
+            model: this.getPrimaryModel(analysis),
+            inputTokens: analysis.inputTokens,
+            outputTokens: analysis.outputTokens,
+            ratio: analysis.outputTokens > 0 ? analysis.inputTokens / analysis.outputTokens : 0,
+            date: session.date,
+            project: projectName,
+            repeatedReads: analysis.repeatedReads,
+            modelEfficiency: analysis.modelEfficiency,
+            tools: tools.length > 0 ? tools : undefined,
+          });
+        }
+
+        dateChildren.push({
+          name: date,
+          children: sessionChildren,
+          date,
+        });
+      }
+
+      projectChildren.push({
+        name: projectName,
+        children: dateChildren,
+        project: projectName,
+      });
+    }
+
+    return {
+      name: "All Sessions",
+      children: projectChildren,
+    };
+  }
+
+  private extractProjectName(encodedProject: string): string {
+    // Directory names encode paths: -home-ctowles-code-p-towles-tool
+    const parts = encodedProject.split("-").filter(Boolean);
+    const pathMarkers = new Set(["code", "projects", "src", "p", "repos", "git", "workspace"]);
+
+    // Find LAST index of a path marker
+    let lastMarkerIdx = -1;
+    for (let i = 0; i < parts.length; i++) {
+      if (pathMarkers.has(parts[i].toLowerCase())) {
+        lastMarkerIdx = i;
+      }
+    }
+
+    // Take everything after the last marker
+    const projectParts = lastMarkerIdx >= 0 ? parts.slice(lastMarkerIdx + 1) : parts.slice(-2);
+
+    if (projectParts.length === 0) {
+      return parts[parts.length - 1] || encodedProject.slice(0, 20);
+    }
+    return projectParts.join("-");
+  }
+
+  private extractSessionLabel(entries: JournalEntry[], sessionId: string): string {
+    return extractSessionLabel(entries, sessionId);
+  }
+
+  private analyzeSession(entries: JournalEntry[]): ReturnType<typeof analyzeSession> {
+    return analyzeSession(entries);
+  }
+
+  private getPrimaryModel(analysis: ReturnType<typeof this.analyzeSession>): string {
+    const { opusTokens, sonnetTokens, haikuTokens } = analysis;
+    if (opusTokens >= sonnetTokens && opusTokens >= haikuTokens) return "Opus";
+    if (sonnetTokens >= haikuTokens) return "Sonnet";
+    return "Haiku";
+  }
+
+  private getModelName(model?: string): string {
+    if (!model) return "unknown";
+    if (model.includes("opus")) return "Opus";
+    if (model.includes("sonnet")) return "Sonnet";
+    if (model.includes("haiku")) return "Haiku";
+    return model.split("-")[0] || "unknown";
+  }
+}