@appkit/llamacpp-cli 1.12.0 → 1.13.0

package/src/commands/logs.ts

@@ -1,345 +0,0 @@
-import chalk from 'chalk';
-import { spawn } from 'child_process';
-import * as readline from 'readline';
-import * as fs from 'fs';
-import { stateManager } from '../lib/state-manager';
-import { fileExists } from '../utils/file-utils';
-import { execCommand } from '../utils/process-utils';
-import { logParser } from '../utils/log-parser';
-import {
-  getFileSize,
-  formatFileSize,
-  rotateLogFile,
-  clearLogFile,
-  getArchivedLogInfo,
-  deleteArchivedLogs,
-} from '../utils/log-utils';
-
-interface LogsOptions {
-  follow?: boolean;
-  lines?: number;
-  errors?: boolean;
-  verbose?: boolean;
-  http?: boolean;
-  stdout?: boolean;
-  filter?: string;
-  clear?: boolean;
-  rotate?: boolean;
-  clearArchived?: boolean;
-  clearAll?: boolean;
-  includeHealth?: boolean;
-}
-
-export async function logsCommand(identifier: string, options: LogsOptions): Promise<void> {
-  // Find server
-  const server = await stateManager.findServer(identifier);
-  if (!server) {
-    throw new Error(`Server not found: ${identifier}\n\nUse: llamacpp ps`);
-  }
-
-  // Determine log file (default to stderr where verbose logs go)
-  const logPath = options.stdout ? server.stdoutPath : server.stderrPath;
-  const logType = options.stdout ? 'stdout' : 'stderr';
-
-  // Handle --clear-archived option (deletes only archived logs)
-  if (options.clearArchived) {
-    const archivedInfo = await deleteArchivedLogs(server.id);
-
-    if (archivedInfo.count === 0) {
-      console.log(chalk.yellow(`⚠️  No archived logs found for ${server.modelName}`));
-      console.log(chalk.dim(`   Archived logs are created via --rotate or automatic rotation`));
-      return;
-    }
-
-    console.log(chalk.green(`✅ Deleted archived logs for ${server.modelName}`));
-    console.log(chalk.dim(`   Files deleted: ${archivedInfo.count}`));
-    console.log(chalk.dim(`   Space freed: ${formatFileSize(archivedInfo.totalSize)}`));
-    console.log(chalk.dim(`   Current logs preserved`));
-    return;
-  }
-
-  // Handle --clear-all option (clears both current and archived logs)
-  if (options.clearAll) {
-    let totalFreed = 0;
-    let currentSize = 0;
-    let archivedSize = 0;
-
-    // Clear current stderr
-    if (await fileExists(server.stderrPath)) {
-      currentSize += await getFileSize(server.stderrPath);
-      await clearLogFile(server.stderrPath);
-    }
-
-    // Clear current stdout
-    if (await fileExists(server.stdoutPath)) {
-      currentSize += await getFileSize(server.stdoutPath);
-      await clearLogFile(server.stdoutPath);
-    }
-
-    // Delete all archived logs
-    const archivedInfo = await deleteArchivedLogs(server.id);
-    archivedSize = archivedInfo.totalSize;
-
-    totalFreed = currentSize + archivedSize;
-
-    console.log(chalk.green(`✅ Cleared all logs for ${server.modelName}`));
-    if (currentSize > 0) {
-      console.log(chalk.dim(`   Current logs: ${formatFileSize(currentSize)}`));
-    }
-    if (archivedSize > 0) {
-      console.log(chalk.dim(`   Archived logs: ${formatFileSize(archivedSize)} (${archivedInfo.count} file${archivedInfo.count > 1 ? 's' : ''})`));
-    }
-    console.log(chalk.dim(`   Total freed: ${formatFileSize(totalFreed)}`));
-    return;
-  }
-
-  // Handle --clear option
-  if (options.clear) {
-    if (!(await fileExists(logPath))) {
-      console.log(chalk.yellow(`⚠️  No ${logType} found for ${server.modelName}`));
-      console.log(chalk.dim(`   Log file does not exist: ${logPath}`));
-      return;
-    }
-
-    const sizeBefore = await getFileSize(logPath);
-    await clearLogFile(logPath);
-
-    console.log(chalk.green(`✅ Cleared ${logType} for ${server.modelName}`));
-    console.log(chalk.dim(`   Freed: ${formatFileSize(sizeBefore)}`));
-    console.log(chalk.dim(`   ${logPath}`));
-    return;
-  }
-
-  // Handle --rotate option
-  if (options.rotate) {
-    if (!(await fileExists(logPath))) {
-      console.log(chalk.yellow(`⚠️  No ${logType} found for ${server.modelName}`));
-      console.log(chalk.dim(`   Log file does not exist: ${logPath}`));
-      return;
-    }
-
-    try {
-      const archivedPath = await rotateLogFile(logPath);
-      const size = await getFileSize(archivedPath);
-
-      console.log(chalk.green(`✅ Rotated ${logType} for ${server.modelName}`));
-      console.log(chalk.dim(`   Archived: ${formatFileSize(size)}`));
-      console.log(chalk.dim(`   → ${archivedPath}`));
-    } catch (error) {
-      throw new Error(`Failed to rotate log: ${(error as Error).message}`);
-    }
-    return;
-  }
-
-  // Check if log file exists
-  if (!(await fileExists(logPath))) {
-    console.log(chalk.yellow(`⚠️  No ${logType} found for ${server.modelName}`));
-    console.log(chalk.dim(`   Log file does not exist: ${logPath}`));
-    return;
-  }
-
-  // Determine filter pattern and mode
-  let filterPattern: string | null = null;
-  let filterDesc = '';
-  let useCompactMode = false;
-
-  // Whether to include health check requests (filtered by default)
-  const includeHealth = options.includeHealth ?? false;
-
-  if (options.verbose) {
-    // Show everything (no filter)
-    filterDesc = ' (all messages)';
-  } else if (options.errors) {
-    // Show only errors
-    filterPattern = 'error|Error|ERROR|failed|Failed|FAILED';
-    filterDesc = ' (errors only)';
-  } else if (options.http) {
-    // Full HTTP JSON logs
-    filterPattern = 'log_server_r';
-    filterDesc = ' (HTTP JSON)';
-  } else if (options.filter) {
-    // Custom filter
-    filterPattern = options.filter;
-    filterDesc = ` (filter: ${options.filter})`;
-  } else {
-    // Default: Compact one-liner format
-    filterPattern = 'log_server_r';
-    filterDesc = ' (compact)';
-    useCompactMode = true;
-  }
-
-  console.log(chalk.blue(`📋 Logs for ${server.modelName} (${logType}${filterDesc})`));
-  console.log(chalk.dim(`   ${logPath}`));
-
-  // Show log size information
-  const currentSize = await getFileSize(logPath);
-  const archivedInfo = await getArchivedLogInfo(server.id);
-
-  if (archivedInfo.count > 0) {
-    console.log(chalk.dim(`   Current: ${formatFileSize(currentSize)} | Archived: ${formatFileSize(archivedInfo.totalSize)} (${archivedInfo.count} file${archivedInfo.count > 1 ? 's' : ''})`));
-  } else {
-    console.log(chalk.dim(`   Current: ${formatFileSize(currentSize)}`));
-  }
-
-  // Show subtle note if verbose logging is not enabled
-  if (!server.verbose && !options.verbose && !options.errors && !options.http && !options.filter) {
-    console.log(chalk.dim(`   verbosity is disabled`));
-  }
-  console.log();
-
-  if (options.follow) {
-    // Follow logs in real-time with optional filtering
-    if (useCompactMode) {
-      // Compact mode with follow: parse lines in real-time
-      const tailProcess = spawn('tail', ['-f', logPath]);
-      const rl = readline.createInterface({
-        input: tailProcess.stdout,
-        crlfDelay: Infinity,
-      });
-
-      rl.on('line', (line) => {
-        if (line.includes('log_server_r')) {
-          // Skip health check requests unless --include-health is set
-          if (!includeHealth && logParser.isHealthCheckRequest(line)) {
-            return;
-          }
-          logParser.processLine(line, (compactLine) => {
-            // Double-check the parsed line for health checks (in case buffered)
-            if (!includeHealth && logParser.isHealthCheckRequest(compactLine)) {
-              return;
-            }
-            console.log(compactLine);
-          });
-        }
-      });
-
-      // Handle Ctrl+C gracefully
-      process.on('SIGINT', () => {
-        tailProcess.kill();
-        rl.close();
-        console.log();
-        process.exit(0);
-      });
-
-      tailProcess.on('exit', () => {
-        process.exit(0);
-      });
-    } else if (filterPattern) {
-      // Use tail piped to grep for filtering
-      const grepProcess = spawn('sh', ['-c', `tail -f "${logPath}" | grep --line-buffered -E "${filterPattern}"`], {
-        stdio: 'inherit',
-      });
-
-      // Handle Ctrl+C gracefully
-      process.on('SIGINT', () => {
-        grepProcess.kill();
-        console.log();
-        process.exit(0);
-      });
-
-      grepProcess.on('exit', () => {
-        process.exit(0);
-      });
-    } else {
-      // No filter, just tail
-      const tail = spawn('tail', ['-f', logPath], {
-        stdio: 'inherit',
-      });
-
-      process.on('SIGINT', () => {
-        tail.kill();
-        console.log();
-        process.exit(0);
-      });
-
-      tail.on('exit', () => {
-        process.exit(0);
-      });
-    }
-  } else {
-    // Show last N lines with optional filtering
-    const lines = options.lines || 50;
-
-    if (useCompactMode) {
-      // Compact mode: read file and parse
-      try {
-        // Use large multiplier to account for verbose debug output between requests
-        // Add || true to prevent grep from failing when no matches found
-        const command = `tail -n ${lines * 100} "${logPath}" | grep -E "log_server_r" || true`;
-        const output = await execCommand(command);
-        const logLines = output.split('\n').filter((l) => l.trim());
-
-        if (logLines.length === 0) {
-          console.log(chalk.dim('No HTTP request logs in compact format.'));
-          console.log(chalk.dim('The server may be starting up, or only simple GET requests have been made.'));
-          console.log(chalk.dim('\nTip: Use --http to see raw HTTP logs, or --verbose for all server logs.'));
-          return;
-        }
-
-        const compactLines: string[] = [];
-        for (const line of logLines) {
-          // Skip health check requests unless --include-health is set
-          if (!includeHealth && logParser.isHealthCheckRequest(line)) {
-            continue;
-          }
-          logParser.processLine(line, (compactLine) => {
-            // Double-check the parsed line for health checks (in case buffered)
-            if (!includeHealth && logParser.isHealthCheckRequest(compactLine)) {
-              return;
-            }
-            compactLines.push(compactLine);
-          });
-        }
-
-        // Flush any remaining buffered logs (handles simple format)
-        logParser.flush((compactLine) => {
-          // Filter health checks from flushed lines too
-          if (!includeHealth && logParser.isHealthCheckRequest(compactLine)) {
-            return;
-          }
-          compactLines.push(compactLine);
-        });
-
-        // Check if we got any parsed output
-        if (compactLines.length === 0) {
-          console.log(chalk.dim('HTTP request logs found, but could not parse in compact format.'));
-          console.log(chalk.dim('This usually happens with simple GET requests (health checks, slots, etc.).'));
-          console.log(chalk.dim('\nTip: Use --http to see raw HTTP logs instead.'));
-          return;
-        }
-
-        // Show only the last N compact lines
-        const limitedLines = compactLines.slice(-lines);
-        limitedLines.forEach((line) => console.log(line));
-      } catch (error) {
-        throw new Error(`Failed to read logs: ${(error as Error).message}`);
-      }
-    } else {
-      // Regular filtering
-      try {
-        let command: string;
-
-        if (filterPattern) {
-          // Use tail piped to grep
-          // Add || true to prevent grep from failing when no matches found
-          command = `tail -n ${lines} "${logPath}" | grep -E "${filterPattern}" || true`;
-        } else {
-          // No filter
-          command = `tail -n ${lines} "${logPath}"`;
-        }
-
-        const output = await execCommand(command);
-
-        if (filterPattern && output.trim() === '') {
-          console.log(chalk.dim(`No logs matching pattern: ${filterPattern}`));
-          console.log(chalk.dim('\nTip: Try --verbose to see all logs, or adjust your filter pattern.'));
-          return;
-        }
-
-        console.log(output);
-      } catch (error) {
-        throw new Error(`Failed to read logs: ${(error as Error).message}`);
-      }
-    }
-  }
-}

package/src/commands/monitor.ts

@@ -1,110 +0,0 @@
-import chalk from 'chalk';
-import blessed from 'blessed';
-import { stateManager } from '../lib/state-manager.js';
-import { createMonitorUI } from '../tui/MonitorApp.js';
-import { createMultiServerMonitorUI } from '../tui/MultiServerMonitorApp.js';
-
-export async function monitorCommand(identifier?: string): Promise<void> {
-  // Initialize state manager
-  await stateManager.initialize();
-
-  // Get all servers
-  const allServers = await stateManager.getAllServers();
-  if (allServers.length === 0) {
-    throw new Error(
-      `No servers configured.\n\n` +
-        `Create a server first: llamacpp server create <model>`
-    );
-  }
-
-  // Create blessed screen
-  const screen = blessed.screen({
-    smartCSR: true,
-    title: 'llama.cpp Server Monitor',
-  });
-
-  // Determine which UI to launch
-  if (identifier) {
-    // User specified a server - single server mode
-    const server = await stateManager.findServer(identifier);
-    if (!server) {
-      screen.destroy();
-      throw new Error(
-        `Server not found: ${identifier}\n\n` +
-          `Use: llamacpp ps (to list servers)\n` +
-          `Or create a new server: llamacpp server create <model>`
-      );
-    }
-
-    // Update server status to reflect actual launchctl state
-    const { statusChecker } = await import('../lib/status-checker.js');
-    const status = await statusChecker.checkServer(server);
-    server.status = status.isRunning ? 'running' : 'stopped';
-    server.pid = status.pid || undefined;
-
-    // Check if server is running
-    if (server.status !== 'running') {
-      screen.destroy();
-      throw new Error(
-        `Server ${server.modelName} is not running.\n\n` +
-          `Start it first: llamacpp server start ${server.id}`
-      );
-    }
-
-    // Launch single-server TUI
-    await createMonitorUI(screen, server);
-  } else if (allServers.length === 1) {
-    // Only one server - single server mode
-    const server = allServers[0];
-
-    // Update server status to reflect actual launchctl state
-    const { statusChecker } = await import('../lib/status-checker.js');
-    const status = await statusChecker.checkServer(server);
-    server.status = status.isRunning ? 'running' : 'stopped';
-    server.pid = status.pid || undefined;
-
-    // Check if server is running
-    if (server.status !== 'running') {
-      screen.destroy();
-      throw new Error(
-        `Server ${server.modelName} is not running.\n\n` +
-          `Start it first: llamacpp server start ${server.id}`
-      );
-    }
-
-    // Launch single-server TUI
-    await createMonitorUI(screen, server);
-  } else {
-    // Multiple servers - multi-server mode
-    // Update all server statuses to reflect actual launchctl state
-    const { statusChecker } = await import('../lib/status-checker.js');
-    for (const server of allServers) {
-      const status = await statusChecker.checkServer(server);
-      server.status = status.isRunning ? 'running' : 'stopped';
-      server.pid = status.pid || undefined;
-    }
-
-    // Filter to only running servers for monitoring
-    const runningServers = allServers.filter(s => s.status === 'running');
-
-    if (runningServers.length === 0) {
-      screen.destroy();
-      throw new Error(
-        `No servers are currently running.\n\n` +
-          `Start a server first:\n` +
-          allServers
-            .map((s) => `  llamacpp server start ${s.id}  # ${s.modelName}`)
-            .join('\n')
-      );
-    }
-
-    // Launch multi-server TUI (pass all servers so we can see stopped ones too)
-    await createMultiServerMonitorUI(screen, allServers);
-  }
-
-  // Render the screen
-  screen.render();
-
-  // Note: TUI functions handle their own key events and exit directly
-  // The process will stay alive until user presses Q/Ctrl+C
-}

package/src/commands/ps.ts

@@ -1,84 +0,0 @@
-import chalk from 'chalk';
-import Table from 'cli-table3';
-import { stateManager } from '../lib/state-manager.js';
-import { statusChecker } from '../lib/status-checker.js';
-import { formatUptime, formatBytes } from '../utils/format-utils.js';
-import { getProcessMemory } from '../utils/process-utils.js';
-import { ServerConfig } from '../types/server-config.js';
-
-const STATUS_CONFIG = {
-  running: { text: '✅ RUNNING', color: chalk.green },
-  crashed: { text: '❌ CRASHED', color: chalk.red },
-  stopped: { text: '⚠️  STOPPED', color: chalk.yellow },
-} as const;
-
-async function getServerMemory(server: ServerConfig): Promise<string> {
-  if (server.status !== 'running' || !server.pid) {
-    return '-';
-  }
-
-  const cpuMemoryBytes = await getProcessMemory(server.pid);
-  if (cpuMemoryBytes === null) {
-    return '-';
-  }
-
-  const metalMemoryBytes = server.metalMemoryMB ? server.metalMemoryMB * 1024 * 1024 : 0;
-  return formatBytes(cpuMemoryBytes + metalMemoryBytes);
-}
-
-export async function psCommand(): Promise<void> {
-  const servers = await stateManager.getAllServers();
-
-  if (servers.length === 0) {
-    console.log(chalk.yellow('No servers configured.'));
-    console.log(chalk.dim('\nCreate a server: llamacpp server create <model-filename>'));
-    return;
-  }
-
-  console.log(chalk.dim('Checking server statuses...\n'));
-  const serversWithStatus = await statusChecker.updateAllServerStatuses();
-
-  const table = new Table({
-    head: ['SERVER ID', 'MODEL', 'PORT', 'STATUS', 'PID', 'MEMORY', 'UPTIME'],
-  });
-
-  const counts = { running: 0, stopped: 0, crashed: 0 };
-
-  for (const server of serversWithStatus) {
-    const status = server.status || 'stopped';
-    const config = STATUS_CONFIG[status] || STATUS_CONFIG.stopped;
-    counts[status]++;
-
-    const uptime = server.status === 'running' && server.lastStarted
-      ? formatUptime(server.lastStarted)
-      : '-';
-
-    const memoryText = await getServerMemory(server);
-
-    table.push([
-      server.id,
-      server.modelName,
-      server.port.toString(),
-      config.color(config.text),
-      server.pid?.toString() || '-',
-      memoryText,
-      uptime,
-    ]);
-  }
-
-  console.log(table.toString());
-
-  const summary = [
-    chalk.green(`${counts.running} running`),
-    chalk.yellow(`${counts.stopped} stopped`),
-  ];
-  if (counts.crashed > 0) {
-    summary.push(chalk.red(`${counts.crashed} crashed`));
-  }
-
-  console.log(chalk.dim(`\nTotal: ${servers.length} servers (${summary.join(', ')})`));
-
-  if (counts.crashed > 0) {
-    console.log(chalk.red('\n⚠️  Some servers have crashed. Check logs with: llamacpp server logs <id> --errors'));
-  }
-}

package/src/commands/pull.ts

@@ -1,44 +0,0 @@
-import chalk from 'chalk';
-import { modelDownloader } from '../lib/model-downloader';
-import { ensureModelsDirectory } from '../lib/models-dir-setup';
-
-interface PullOptions {
-  file?: string;
-}
-
-export async function pullCommand(identifier: string, options: PullOptions): Promise<void> {
-  // Parse repository identifier
-  const parsed = modelDownloader.parseHFIdentifier(identifier);
-
-  // Determine filename - from --file flag or from identifier path
-  let filename = options.file || parsed.file;
-
-  if (!filename) {
-    throw new Error(
-      'Please specify a file to download:\n\n' +
-      'Option 1: llamacpp pull owner/repo/filename.gguf\n' +
-      'Option 2: llamacpp pull owner/repo --file filename.gguf'
-    );
-  }
-
-  // Ensure filename ends with .gguf
-  if (!filename.toLowerCase().endsWith('.gguf')) {
-    filename += '.gguf';
-  }
-
-  // Ensure models directory exists (prompts user if needed)
-  const modelsDir = await ensureModelsDirectory();
-
-  // Download the model
-  try {
-    const modelPath = await modelDownloader.downloadModel(parsed.repo, filename, undefined, modelsDir);
-
-    console.log();
-    console.log(chalk.dim(`Create server: llamacpp server create ${filename}`));
-  } catch (error) {
-    if ((error as Error).message.includes('interrupted')) {
-      console.log(chalk.dim('\nDownload was interrupted. Run the same command again to retry.'));
-    }
-    throw error;
-  }
-}

package/src/commands/rm.ts

@@ -1,107 +0,0 @@
-import chalk from 'chalk';
-import * as readline from 'readline';
-import * as fs from 'fs/promises';
-import { modelScanner } from '../lib/model-scanner';
-import { stateManager } from '../lib/state-manager';
-import { launchctlManager } from '../lib/launchctl-manager';
-
-export async function rmCommand(modelIdentifier: string): Promise<void> {
-  await stateManager.initialize();
-
-  // 1. Resolve model path
-  const modelPath = await modelScanner.resolveModelPath(modelIdentifier);
-  if (!modelPath) {
-    throw new Error(`Model not found: ${modelIdentifier}\n\nRun: llamacpp ls`);
-  }
-
-  // 2. Check if any servers are using this model
-  const allServers = await stateManager.getAllServers();
-  const serversUsingModel = allServers.filter((s) => s.modelPath === modelPath);
-
-  // 3. Confirm deletion
-  console.log(chalk.yellow(`⚠️  Delete model file: ${modelPath}`));
-
-  if (serversUsingModel.length > 0) {
-    console.log(chalk.yellow(`\n   This model has ${serversUsingModel.length} server(s) configured:`));
-    for (const server of serversUsingModel) {
-      const statusColor = server.status === 'running' ? chalk.green : chalk.dim;
-      console.log(chalk.yellow(`   - ${server.id} (${statusColor(server.status)})`));
-    }
-    console.log(chalk.yellow(`\n   These servers will be removed before deleting the model.`));
-  }
-
-  console.log();
-
-  const confirmed = await confirmDeletion();
-  if (!confirmed) {
-    console.log(chalk.dim('Cancelled'));
-    return;
-  }
-
-  console.log();
-
-  // 4. Delete all servers using this model
-  if (serversUsingModel.length > 0) {
-    console.log(chalk.blue(`🗑️  Removing ${serversUsingModel.length} server(s)...\n`));
-
-    for (const server of serversUsingModel) {
-      console.log(chalk.dim(`  Removing server: ${server.id}`));
-
-      // Unload service (stops and removes from launchd)
-      try {
-        await launchctlManager.unloadService(server.plistPath);
-        if (server.status === 'running') {
-          await launchctlManager.waitForServiceStop(server.label, 5000);
-        }
-      } catch (error) {
-        console.log(chalk.yellow(`    ⚠️  Failed to unload service gracefully`));
-      }
-
-      // Delete plist
-      await launchctlManager.deletePlist(server.plistPath);
-
-      // Delete server config
-      await stateManager.deleteServerConfig(server.id);
-
-      console.log(chalk.dim(`    ✓ Server removed`));
-    }
-
-    console.log();
-  }
-
-  // 5. Delete model file
-  console.log(chalk.blue(`🗑️  Deleting model file...`));
-
-  try {
-    await fs.unlink(modelPath);
-  } catch (error) {
-    throw new Error(`Failed to delete model file: ${(error as Error).message}`);
-  }
-
-  // Success
-  console.log();
-  console.log(chalk.green('✅ Model deleted successfully'));
-
-  if (serversUsingModel.length > 0) {
-    console.log(chalk.dim(`   Removed ${serversUsingModel.length} server(s)`));
-  }
-
-  console.log(chalk.dim(`   Deleted: ${modelPath}`));
-}
-
-/**
- * Prompt user for confirmation
- */
-function confirmDeletion(): Promise<boolean> {
-  return new Promise((resolve) => {
-    const rl = readline.createInterface({
-      input: process.stdin,
-      output: process.stdout,
-    });
-
-    rl.question(chalk.yellow("   Type 'yes' to confirm: "), (answer) => {
-      rl.close();
-      resolve(answer.toLowerCase() === 'yes');
-    });
-  });
-}