@duckmind/dm-windows-x64 0.62.3 → 0.63.0

package/extensions/greedysearch-dm/src/search/browser-lifecycle.mjs

@@ -1,340 +0,0 @@
-import { execFileSync, execSync } from "node:child_process";
-import {
-  existsSync,
-  mkdirSync,
-  readFileSync,
-  unlinkSync,
-  writeFileSync
-} from "node:fs";
-import { platform } from "node:os";
-import {
-  CHROME_MODE_FILE,
-  GREEDY_ACTIVITY_FILE,
-  GREEDY_LAUNCH_LOCK_FILE,
-  GREEDY_METADATA_FILE,
-  GREEDY_PID_FILE,
-  GREEDY_PORT,
-  GREEDY_PROFILE_DIR
-} from "./constants.mjs";
-import { resolveSystemCmd } from "../utils/system-cmds.mjs";
-import { findListeningProcessPid } from "./port-pid.mjs";
-export const METADATA_FILE = GREEDY_METADATA_FILE;
-const LAUNCH_LOCK_FILE = GREEDY_LAUNCH_LOCK_FILE;
-const LOCK_STALE_MS = 15000;
-const LEGACY_PID_FILE = GREEDY_PID_FILE;
-const LEGACY_MODE_FILE = CHROME_MODE_FILE;
-const LEGACY_ACTIVITY_FILE = GREEDY_ACTIVITY_FILE;
-function processExists(pid) {
-  if (!Number.isInteger(pid) || pid <= 0)
-    return false;
-  try {
-    process.kill(pid, 0);
-    return true;
-  } catch {
-    return false;
-  }
-}
-function getProcessCommandLine(pid) {
-  if (!processExists(pid))
-    return null;
-  try {
-    if (platform() === "win32") {
-      const output = execFileSync(resolveSystemCmd("powershell"), [
-        "-NoProfile",
-        "-NonInteractive",
-        "-Command",
-        `(Get-CimInstance Win32_Process -Filter "ProcessId = ${pid}").CommandLine`
-      ], { encoding: "utf8", windowsHide: true, timeout: 5000 });
-      return output.trim() || null;
-    }
-    const output = execFileSync(resolveSystemCmd("ps"), ["-p", String(pid), "-o", "command="], {
-      encoding: "utf8",
-      timeout: 5000
-    });
-    return output.trim() || null;
-  } catch {
-    return null;
-  }
-}
-export function commandLineMatchesGreedyChrome(cmdLine, tempDir, debugPort = GREEDY_PORT) {
-  if (!cmdLine)
-    return false;
-  const normalize = (value) => String(value || "").replaceAll("\\", "/").toLowerCase();
-  const normalizedCmdLine = normalize(cmdLine);
-  const normalizedTempDir = normalize(tempDir);
-  return normalizedCmdLine.includes(normalizedTempDir) && normalizedCmdLine.includes(`--remote-debugging-port=${debugPort}`) && !normalizedCmdLine.includes("--type=");
-}
-export function verifyBrowserProcess(pid, tempDir, debugPort = GREEDY_PORT) {
-  return commandLineMatchesGreedyChrome(getProcessCommandLine(pid), tempDir, debugPort);
-}
-function getPortPid(port = GREEDY_PORT) {
-  return findListeningProcessPid(port);
-}
-function forceKillProcess(pid, expectedProfile = null, expectedPort = null) {
-  if (expectedProfile && !verifyBrowserProcess(pid, expectedProfile, expectedPort))
-    return false;
-  try {
-    if (platform() === "win32") {
-      execSync(`${resolveSystemCmd("taskkill")} /F /T /PID ${pid}`, {
-        stdio: "ignore"
-      });
-    } else {
-      try {
-        process.kill(-pid, "SIGKILL");
-      } catch {}
-      try {
-        process.kill(pid, "SIGKILL");
-      } catch {}
-    }
-    return true;
-  } catch {
-    return false;
-  }
-}
-export function readMetadata() {
-  try {
-    if (existsSync(METADATA_FILE)) {
-      const raw = readFileSync(METADATA_FILE, "utf8");
-      const parsed = JSON.parse(raw);
-      if (parsed && typeof parsed.tempDir === "string" && typeof parsed.debugPort === "number") {
-        return {
-          browserPid: Number.isInteger(parsed.browserPid) ? parsed.browserPid : undefined,
-          debugPort: parsed.debugPort,
-          tempDir: parsed.tempDir,
-          clientPids: Array.isArray(parsed.clientPids) ? parsed.clientPids.filter((p) => Number.isInteger(p) && p > 0) : [],
-          sessionMode: parsed.sessionMode === "visible" ? "visible" : "headless",
-          lastActivity: Number.isFinite(parsed.lastActivity) ? parsed.lastActivity : 0,
-          launchedAt: Number.isFinite(parsed.launchedAt) ? parsed.launchedAt : 0
-        };
-      }
-    }
-  } catch {}
-  try {
-    const browserPid = existsSync(LEGACY_PID_FILE) ? Number.parseInt(readFileSync(LEGACY_PID_FILE, "utf8").trim(), 10) || undefined : undefined;
-    const sessionMode = existsSync(LEGACY_MODE_FILE) ? readFileSync(LEGACY_MODE_FILE, "utf8").trim() === "visible" ? "visible" : "headless" : "headless";
-    const lastActivity = existsSync(LEGACY_ACTIVITY_FILE) ? Number.parseInt(readFileSync(LEGACY_ACTIVITY_FILE, "utf8").trim(), 10) || 0 : 0;
-    return {
-      browserPid,
-      debugPort: GREEDY_PORT,
-      tempDir: GREEDY_PROFILE_DIR,
-      clientPids: browserPid ? [browserPid] : [],
-      sessionMode,
-      lastActivity,
-      launchedAt: 0
-    };
-  } catch {
-    return null;
-  }
-}
-export function writeMetadata(metadata) {
-  try {
-    writeFileSync(METADATA_FILE, JSON.stringify({
-      browserPid: metadata.browserPid,
-      debugPort: metadata.debugPort,
-      tempDir: metadata.tempDir,
-      clientPids: [...new Set(metadata.clientPids.filter((p) => p > 0))],
-      sessionMode: metadata.sessionMode,
-      lastActivity: metadata.lastActivity,
-      launchedAt: metadata.launchedAt
-    }, null, 2), "utf8");
-  } catch {}
-  try {
-    if (metadata.browserPid)
-      writeFileSync(LEGACY_PID_FILE, String(metadata.browserPid), "utf8");
-  } catch {}
-  try {
-    writeFileSync(LEGACY_MODE_FILE, metadata.sessionMode, "utf8");
-  } catch {}
-  try {
-    writeFileSync(LEGACY_ACTIVITY_FILE, String(metadata.lastActivity), "utf8");
-  } catch {}
-}
-export function clearMetadata() {
-  try {
-    unlinkSync(METADATA_FILE);
-  } catch {}
-  try {
-    unlinkSync(LEGACY_PID_FILE);
-  } catch {}
-  try {
-    unlinkSync(LEGACY_MODE_FILE);
-  } catch {}
-  try {
-    unlinkSync(LEGACY_ACTIVITY_FILE);
-  } catch {}
-}
-export function registerClient(metadata) {
-  if (!metadata)
-    return metadata;
-  const updated = {
-    ...metadata,
-    clientPids: [
-      ...new Set([...metadata.clientPids, process.pid].filter((p) => processExists(p) || p === process.pid))
-    ]
-  };
-  writeMetadata(updated);
-  return updated;
-}
-export function unregisterClient(metadata) {
-  if (!metadata)
-    return metadata;
-  const updated = {
-    ...metadata,
-    clientPids: metadata.clientPids.filter((p) => p !== process.pid).filter((p) => processExists(p))
-  };
-  writeMetadata(updated);
-  return updated;
-}
-export function touchActivity(metadata) {
-  const ts = Date.now();
-  try {
-    if (metadata) {
-      writeMetadata({ ...metadata, lastActivity: ts });
-    } else {
-      writeFileSync(LEGACY_ACTIVITY_FILE, String(ts), "utf8");
-    }
-  } catch {}
-}
-const IDLE_TIMEOUT_MINUTES = Number.parseInt(process.env.GREEDY_SEARCH_IDLE_TIMEOUT_MINUTES || "5", 10) || 5;
-export function isIdleExpired(metadata) {
-  if (IDLE_TIMEOUT_MINUTES <= 0)
-    return false;
-  if (!metadata || !metadata.lastActivity)
-    return false;
-  const idleMs = Date.now() - metadata.lastActivity;
-  return idleMs / 60000 >= IDLE_TIMEOUT_MINUTES;
-}
-let _staleCleanupPerformed = false;
-export function cleanupStaleSessions() {
-  if (process.env.GREEDY_SEARCH_RESEARCH_CHILD)
-    return;
-  if (_staleCleanupPerformed)
-    return;
-  _staleCleanupPerformed = true;
-  const metadata = readMetadata();
-  if (!metadata)
-    return;
-  if (metadata.browserPid) {
-    const alive = processExists(metadata.browserPid);
-    if (!alive) {
-      clearMetadata();
-      return;
-    }
-    const verified = verifyBrowserProcess(metadata.browserPid, metadata.tempDir, metadata.debugPort);
-    if (!verified) {
-      clearMetadata();
-    } else {
-      writeMetadata({
-        ...metadata,
-        clientPids: metadata.clientPids.filter((p) => processExists(p))
-      });
-    }
-  }
-  const portPid = getPortPid();
-  if (portPid && portPid !== metadata.browserPid) {
-    const verified = verifyBrowserProcess(portPid, GREEDY_PROFILE_DIR, GREEDY_PORT);
-    if (verified) {
-      writeMetadata({
-        browserPid: portPid,
-        debugPort: GREEDY_PORT,
-        tempDir: GREEDY_PROFILE_DIR,
-        clientPids: [portPid],
-        sessionMode: metadata.sessionMode,
-        lastActivity: Date.now(),
-        launchedAt: Date.now()
-      });
-    } else {
-      process.stderr.write(`[greedysearch] Refusing to kill unverified listener ${portPid} on port ${GREEDY_PORT}.
-`);
-    }
-  }
-}
-export function acquireLaunchLock() {
-  mkdirSync(GREEDY_PROFILE_DIR, { recursive: true });
-  try {
-    const payload = JSON.stringify({ pid: process.pid, ts: Date.now() });
-    writeFileSync(LAUNCH_LOCK_FILE, payload, { encoding: "utf8", flag: "wx" });
-    return {
-      acquired: true,
-      release: () => {
-        try {
-          const raw = readFileSync(LAUNCH_LOCK_FILE, "utf8");
-          const parsed = JSON.parse(raw);
-          if (parsed.pid === process.pid) {
-            unlinkSync(LAUNCH_LOCK_FILE);
-          }
-        } catch {}
-      }
-    };
-  } catch (e) {
-    if (e?.code !== "EEXIST") {
-      return { acquired: false, release: () => {} };
-    }
-  }
-  try {
-    const raw = readFileSync(LAUNCH_LOCK_FILE, "utf8");
-    const parsed = JSON.parse(raw);
-    const lockAge = Date.now() - (parsed.ts || 0);
-    const lockPidAlive = processExists(parsed.pid);
-    if (!lockPidAlive || lockAge > LOCK_STALE_MS) {
-      try {
-        unlinkSync(LAUNCH_LOCK_FILE);
-      } catch {}
-      try {
-        const payload = JSON.stringify({ pid: process.pid, ts: Date.now() });
-        writeFileSync(LAUNCH_LOCK_FILE, payload, {
-          encoding: "utf8",
-          flag: "wx"
-        });
-        return {
-          acquired: true,
-          release: () => {
-            try {
-              unlinkSync(LAUNCH_LOCK_FILE);
-            } catch {}
-          }
-        };
-      } catch {
-        return { acquired: false, release: () => {} };
-      }
-    }
-  } catch {}
-  return { acquired: false, release: () => {} };
-}
-export function checkRunning() {
-  const metadata = readMetadata();
-  if (!metadata || !metadata.browserPid) {
-    const portPid = getPortPid();
-    if (portPid && verifyBrowserProcess(portPid, GREEDY_PROFILE_DIR, GREEDY_PORT)) {
-      return { running: true, pid: portPid, metadata };
-    }
-    return { running: false, pid: null, metadata: null };
-  }
-  const alive = processExists(metadata.browserPid);
-  if (!alive) {
-    return { running: false, pid: null, metadata };
-  }
-  const verified = verifyBrowserProcess(metadata.browserPid, metadata.tempDir, metadata.debugPort);
-  if (!verified) {
-    const portPid = getPortPid();
-    if (portPid && verifyBrowserProcess(portPid, GREEDY_PROFILE_DIR, GREEDY_PORT)) {
-      return { running: true, pid: portPid, metadata };
-    }
-    return { running: false, pid: null, metadata };
-  }
-  return { running: true, pid: metadata.browserPid, metadata };
-}
-export function killChrome() {
-  const { running, pid, metadata } = checkRunning();
-  if (!running || !pid) {
-    clearMetadata();
-    return false;
-  }
-  const killed = forceKillProcess(pid, metadata?.tempDir || GREEDY_PROFILE_DIR, metadata?.debugPort || GREEDY_PORT);
-  clearMetadata();
-  return killed;
-}
-export async function ensureCleanSession() {
-  cleanupStaleSessions();
-  return checkRunning();
-}

package/extensions/greedysearch-dm/src/search/challenge-detect.mjs

@@ -1,156 +0,0 @@
-import { cdp } from "../../extractors/common.mjs";
-const DEFAULT_TIMEOUT_MS = Number.parseInt(process.env.GREEDY_SEARCH_CHALLENGE_WAIT_MS || "300000", 10);
-const POLL_INTERVAL_MS = 3000;
-const ENGINE_SIGNALS = {
-  chatgpt: {
-    name: "chatgpt",
-    isCleared: async (tab) => {
-      const probe = await cdp([
-        "eval",
-        tab,
-        `(() => {
-					const title = document.title;
-					const onChatGPT = location.hostname === "chatgpt.com";
-					const hasProseMirror = !!document.querySelector("div.ProseMirror");
-					const hasTurnstileInput =
-						!!document.querySelector("input[name=\\"cf-turnstile-response\\"]") ||
-						!!document.querySelector("iframe[id^=\\"cf-chl-widget-\\"]");
-					// Body innerText is empty while on the Turnstile page.
-					const bodyText = (document.body && document.body.innerText) || "";
-					return JSON.stringify({
-						title,
-						url: location.href,
-						hasProseMirror,
-						hasTurnstileInput,
-						bodyLen: bodyText.length,
-						onChatGPT,
-					});
-				})()`
-      ]).catch(() => null);
-      if (!probe)
-        return false;
-      let info;
-      try {
-        info = JSON.parse(probe);
-      } catch {
-        return false;
-      }
-      if (!info.onChatGPT)
-        return false;
-      if (info.title && /περιμένετε|please wait|just a moment|verifying|checking/i.test(info.title)) {
-        return false;
-      }
-      if (info.hasTurnstileInput)
-        return false;
-      return info.hasProseMirror || info.bodyLen > 50;
-    }
-  },
-  bing: {
-    name: "bing",
-    isCleared: async (tab) => {
-      const probe = await cdp([
-        "eval",
-        tab,
-        `(() => {
-					const url = location.href;
-					const title = document.title;
-					const onCopilot = /copilot\\.microsoft\\.com/.test(location.hostname);
-					const onChallenge =
-						/challenge|turnstile|cdn-cgi\\/challenge/i.test(url) ||
-						/verify|human|robot/i.test(title);
-					const hasTextarea =
-						!!document.querySelector("textarea") ||
-						!!document.querySelector("div[contenteditable=\\"true\\"]");
-					const hasTurnstileInput =
-						!!document.querySelector("iframe[id^=\\"cf-chl-widget-\\"]") ||
-						!!document.querySelector("input[name=\\"cf-turnstile-response\\"]");
-					const bodyText = (document.body && document.body.innerText) || "";
-					return JSON.stringify({
-						url,
-						title,
-						onCopilot,
-						onChallenge,
-						hasTextarea,
-						hasTurnstileInput,
-						bodyLen: bodyText.length,
-					});
-				})()`
-      ]).catch(() => null);
-      if (!probe)
-        return false;
-      let info;
-      try {
-        info = JSON.parse(probe);
-      } catch {
-        return false;
-      }
-      if (!info.onCopilot)
-        return false;
-      if (info.onChallenge)
-        return false;
-      if (info.hasTurnstileInput)
-        return false;
-      return info.hasTextarea || info.bodyLen > 50;
-    }
-  }
-};
-async function pollForCfClearanceCookie(tab) {
-  const probe = await cdp([
-    "eval",
-    tab,
-    `(() => {
-			const cookies = document.cookie || "";
-			return JSON.stringify({
-				hasCfClearance: /(?:^|;\\s*)cf_clearance=/.test(cookies),
-				hasCfBm: /(?:^|;\\s*)__cf_bm=/.test(cookies),
-				cookiesLength: cookies.length,
-			});
-		})()`
-  ]).catch(() => null);
-  if (!probe)
-    return false;
-  try {
-    const info = JSON.parse(probe);
-    return info.hasCfClearance || info.hasCfBm;
-  } catch {
-    return false;
-  }
-}
-export async function waitForChallengeCleared({
-  tab,
-  engine,
-  timeoutMs = DEFAULT_TIMEOUT_MS,
-  intervalMs = POLL_INTERVAL_MS,
-  signal: externalSignal,
-  log = () => {}
-}) {
-  const def = ENGINE_SIGNALS[engine];
-  const start = Date.now();
-  let lastState = null;
-  while (Date.now() - start < timeoutMs) {
-    if (externalSignal?.aborted) {
-      return { cleared: false, reason: "aborted" };
-    }
-    const elapsed = Math.floor((Date.now() - start) / 1000);
-    let cleared = false;
-    if (def) {
-      cleared = await def.isCleared(tab).catch(() => false);
-    } else {
-      cleared = await pollForCfClearanceCookie(tab).catch(() => false);
-    }
-    if (cleared) {
-      log(`[greedysearch] ✅ ${engine} challenge cleared after ${elapsed}s — auto-resuming extraction.`);
-      return { cleared: true, signal: def ? "dom-marker" : "cookie" };
-    }
-    if (elapsed > 0 && elapsed % 30 === 0 && lastState !== elapsed) {
-      lastState = elapsed;
-      log(`[greedysearch] ⏳ Waiting for ${engine} challenge to clear (${elapsed}s/${Math.floor(timeoutMs / 1000)}s)...`);
-    }
-    await new Promise((r) => setTimeout(r, intervalMs));
-  }
-  return {
-    cleared: false,
-    reason: `Challenge not cleared within ${Math.floor(timeoutMs / 1000)}s`
-  };
-}
-export const CHALLENGE_ENGINES = Object.keys(ENGINE_SIGNALS);