Compare commits

..
Author SHA1 Message Date
CloakHQ c35afec09a feat: add GitHub Models issue knowledge responder
Auto-responds to new issues with related docs and past issues.
Uses actions/ai-inference with GPT-4.1 (free GitHub Models).
5 search strategies: title, keywords, body terms, labels, error msgs.
No external API keys needed.
2026-04-10 23:22:48 +02:00
CloakHQ c6d3469e4c release: v0.3.24 — SOCKS5 proxy support, arm64 146 upgrade, ElementHandle humanize 2026-04-10 22:42:29 +02:00
CloakHQ cb0b87873e feat: native SOCKS5 proxy support in proxy= parameter
Route SOCKS5/SOCKS5h proxies via --proxy-server Chrome arg instead of
Playwright's proxy dict (which rejects SOCKS5 with credentials).
Handles string URLs, Playwright dicts, IPv6, bypass lists.

SOCKS5 geoip exit IP resolution uses socks-proxy-agent (optional peer
dep). Falls back to DNS if not installed.
2026-04-10 22:20:16 +02:00
lilosandGitHub 2be8cdcc03 feat(humanize): add Playwright ElementHandle support and fix async tests (#133) 2026-04-10 22:18:14 +02:00
CloakHQ be9a98db67 fix(test): correct cloakserve passthrough test for --fingerprint parsing 2026-04-10 21:40:20 +02:00
CloakHQ 9b004bbd85 docs: clarify humanize requires wrapper import over CDP (#126) 2026-04-09 21:08:48 +02:00
27 changed files with 2854 additions and 119 deletions
@@ -0,0 +1,62 @@
name: issue-knowledge-responder
description: First-responder bot that finds related docs and past issues for new GitHub issues
model: openai/gpt-4.1
modelParameters:
maxCompletionTokens: 1500
temperature: 0.3
messages:
- role: system
content: |
You are a first-responder bot for CloakBrowser, an open-source stealth Chromium wrapper.
Your ONLY job: find related resources from the provided context and point the user to them.
You are NOT a maintainer. You do NOT answer questions yourself.
Rules:
- NEVER answer the question yourself. Only point to existing resources.
- NEVER reveal internal patch details, CDP stealth methods, or bypass techniques.
- Keep it short. 5-10 lines max.
- Only include resources that are ACTUALLY relevant, not vaguely similar.
- If the issue is about financial, government, or healthcare targets, output RESPOND: NO
- Maximum 5 related issues. Prefer closed+resolved over open.
- role: user
content: |
## The new issue
{{issue_body}}
## Past issues found by search
{{past_issues}}
## Past issue details (body + comments from top matches)
{{issue_details}}
## README.md (excerpt)
{{readme}}
## js/README.md (excerpt)
{{js_readme}}
## CHANGELOG.md (recent)
{{changelog}}
## Your task
Based on the context above, decide:
**If you found related resources** (doc sections, past issues, or both), output EXACTLY this format:
RESPOND: YES
---
While you wait for the team to respond, I found some resources that might be related:
**From the docs:**
- [description of relevant section] (in `README.md` / `js/README.md`)
**Related issues:**
- #123 — [title] (closed/open) — [one-line summary of resolution or discussion]
Hope this helps! The team will follow up soon.
---
**If you found NOTHING relevant**, output EXACTLY:
RESPOND: NO
@@ -0,0 +1,148 @@
name: Issue Knowledge Responder
on:
issues:
types: [opened]
jobs:
respond:
if: github.event.issue.user.type != 'Bot'
runs-on: ubuntu-latest
timeout-minutes: 5
permissions:
contents: read
issues: write
models: read
steps:
- name: Checkout repository
uses: actions/checkout@v4
with:
fetch-depth: 1
- name: Gather context
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
ISSUE_NUMBER: ${{ github.event.issue.number }}
REPO: ${{ github.repository }}
run: |
mkdir -p /tmp/ctx
# 1. Read the full issue
gh issue view "$ISSUE_NUMBER" --repo "$REPO" --json title,body,labels --jq '
"Title: " + .title + "\n\nBody:\n" + (.body // "No body") + "\n\nLabels: " + ([.labels[].name] | join(", "))
' > /tmp/ctx/issue_body.txt
# 2. Extract keywords from title + body
TITLE=$(gh issue view "$ISSUE_NUMBER" --repo "$REPO" --json title --jq '.title')
BODY_TEXT=$(gh issue view "$ISSUE_NUMBER" --repo "$REPO" --json body --jq '.body // ""' | head -20)
STOPWORDS='^(the|a|an|is|in|on|at|to|for|of|with|and|or|not|bug|issue|error|problem|help|please|how|why|what|does|can|i|my|me|it|this|that|when|from|have|has|been|are|was|were|be|do|did|just|but|if|so|no|yes|all|any|some|like|get|got|use|using|used|try|tried|also|want|need|would|could|should|about|into|after|before|only|very|too|more|most|much|own|same|other|than|then|there|here|each|every|both|few|many|such|these|those|its|new|old|first|last|long|great|little|right|big|high|different|small|large|next|early|young|important|public|bad|same|able)$'
TITLE_KW=$(echo "$TITLE" | tr -cs 'a-zA-Z0-9' ' ' | tr ' ' '\n' | \
grep -v -i -E "$STOPWORDS" | head -5 | tr '\n' ' ')
BODY_KW=$(echo "$BODY_TEXT" | tr -cs 'a-zA-Z0-9' ' ' | tr ' ' '\n' | \
grep -v -i -E "$STOPWORDS" | sort | uniq -c | sort -rn | awk '{print $2}' | head -5 | tr '\n' ' ')
# 3. Search past issues — 5 strategies, deduplicated
# A: full title
gh issue list --repo "$REPO" --state all --search "$TITLE" --limit 10 \
--json number,title,state --jq '.[] | "#\(.number) — \(.title) (\(.state))"' \
2>/dev/null > /tmp/ctx/search_results.txt || true
# B: title keywords
gh issue list --repo "$REPO" --state all --search "$TITLE_KW" --limit 10 \
--json number,title,state --jq '.[] | "#\(.number) — \(.title) (\(.state))"' \
2>/dev/null >> /tmp/ctx/search_results.txt || true
# C: body keywords (most frequent terms from body)
if [ -n "$BODY_KW" ]; then
gh issue list --repo "$REPO" --state all --search "$BODY_KW" --limit 10 \
--json number,title,state --jq '.[] | "#\(.number) — \(.title) (\(.state))"' \
2>/dev/null >> /tmp/ctx/search_results.txt || true
fi
# D: search by labels if the new issue has any
LABELS=$(gh issue view "$ISSUE_NUMBER" --repo "$REPO" --json labels --jq '[.labels[].name] | join(",")')
if [ -n "$LABELS" ]; then
IFS=',' read -ra LABEL_ARR <<< "$LABELS"
for LBL in "${LABEL_ARR[@]}"; do
gh issue list --repo "$REPO" --state all --label "$LBL" --limit 10 \
--json number,title,state --jq '.[] | "#\(.number) — \(.title) (\(.state))"' \
2>/dev/null >> /tmp/ctx/search_results.txt || true
done
fi
# E: search for error messages / code snippets in body (first backtick block or quoted line)
ERROR_MSG=$(echo "$BODY_TEXT" | grep -oP '`[^`]{5,80}`' | head -1 | tr -d '`')
if [ -n "$ERROR_MSG" ]; then
gh issue list --repo "$REPO" --state all --search "$ERROR_MSG" --limit 5 \
--json number,title,state --jq '.[] | "#\(.number) — \(.title) (\(.state))"' \
2>/dev/null >> /tmp/ctx/search_results.txt || true
fi
# Deduplicate, exclude self, keep top 20
sort -u /tmp/ctx/search_results.txt | \
grep -v "^#${ISSUE_NUMBER} " | head -20 > /tmp/ctx/past_issues.txt
# 4. Read top 7 issue details (body + comments)
echo "" > /tmp/ctx/issue_details.txt
for NUM in $(grep -oP '#\K[0-9]+' /tmp/ctx/past_issues.txt | head -7); do
gh issue view "$NUM" --repo "$REPO" --json title,state,body,comments --jq '
"### #'"$NUM"' — " + .title + " (" + .state + ")\n" +
(.body // "")[:500] + "\n\nComments:\n" +
([.comments[:3][] | .body[:300]] | join("\n---\n"))
' >> /tmp/ctx/issue_details.txt 2>/dev/null || true
echo -e "\n---\n" >> /tmp/ctx/issue_details.txt
done
# 5. Read docs
head -200 README.md > /tmp/ctx/readme.txt 2>/dev/null || echo "No README" > /tmp/ctx/readme.txt
head -150 js/README.md > /tmp/ctx/js_readme.txt 2>/dev/null || echo "No JS README" > /tmp/ctx/js_readme.txt
head -100 CHANGELOG.md > /tmp/ctx/changelog.txt 2>/dev/null || echo "No CHANGELOG" > /tmp/ctx/changelog.txt
# Debug: show what we gathered
echo "=== Issue ===" && head -5 /tmp/ctx/issue_body.txt
echo "=== Past issues ===" && cat /tmp/ctx/past_issues.txt
echo "=== Keywords: $KEYWORDS ==="
- name: AI analysis
id: ai
uses: actions/ai-inference@v1
with:
prompt-file: .github/prompts/issue-responder.prompt.yml
file_input: |
issue_body: /tmp/ctx/issue_body.txt
past_issues: /tmp/ctx/past_issues.txt
issue_details: /tmp/ctx/issue_details.txt
readme: /tmp/ctx/readme.txt
js_readme: /tmp/ctx/js_readme.txt
changelog: /tmp/ctx/changelog.txt
- name: Parse and post
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
RESPONSE=$(cat "${{ steps.ai.outputs.response-file }}")
echo "=== AI Response ==="
echo "$RESPONSE"
if echo "$RESPONSE" | grep -q "RESPOND: YES"; then
# Extract content between --- markers
COMMENT=$(echo "$RESPONSE" | sed -n '/^---$/,/^---$/p' | sed '1d;$d')
if [ -z "$COMMENT" ]; then
# Fallback: everything after "RESPOND: YES", strip --- lines
COMMENT=$(echo "$RESPONSE" | sed '1,/RESPOND: YES/d' | sed '/^---$/d')
fi
if [ -n "$COMMENT" ]; then
echo "$COMMENT" > /tmp/comment.txt
gh issue comment "${{ github.event.issue.number }}" \
--repo "${{ github.repository }}" \
--body-file /tmp/comment.txt
echo "✅ Comment posted"
else
echo "⚠️ RESPOND: YES but empty comment body, skipping"
fi
else
echo "️ No relevant resources found, staying silent"
fi
+8
View File
@@ -6,6 +6,14 @@ Changes are tagged: **[wrapper]** for Python/JS wrapper, **[binary]** for Chromi
---
## [0.3.24] — 2026-04-10
- **[wrapper]** Native SOCKS5 proxy support — pass `proxy="socks5://user:pass@host:port"` directly. Credentials handled natively by Chrome. Works across all launch functions, Python + JS.
- **[wrapper]** Add Playwright ElementHandle humanize support — `element_handle.click()`, `.fill()`, `.type()` now use human-like behavior when `humanize=True` (thanks [@lilos](https://github.com/lilos), #133)
- **[binary]** Upgrade Linux arm64 to Chromium 146.0.7680.177.2 (49 patches) — now matches Linux x64
- **[binary]** New build 146.0.7680.177.2 for both Linux platforms: native SOCKS5 proxy with UDP ASSOCIATE (QUIC/HTTP3 over SOCKS5)
- **[docs]** Clarify humanize requires wrapper import over CDP (#126)
## [0.3.23] — 2026-04-09
- **[wrapper]** Add full Puppeteer humanize support — human-like mouse, keyboard, and scroll behavior for `puppeteer-core` users (thanks [@evelaa123](https://github.com/evelaa123), #129)
+17 -7
View File
@@ -128,10 +128,11 @@ Open [http://localhost:8080](http://localhost:8080). Create a profile. Click **L
---
## Latest: v0.3.22 (Chromium 146.0.7680.177.1)
## Latest: v0.3.24 (Chromium 146.0.7680.177.2)
- **Native SOCKS5 proxy** — `proxy="socks5://user:pass@host:port"` works directly in all launch functions, Python + JS. QUIC/HTTP3 tunnels through SOCKS5 via UDP ASSOCIATE.
- **Chromium 146 upgrade** — rebased all patches from 145.0.7632.x to 146.0.7680.177
- **49 fingerprint patches** (Linux x64) — 1 new patch, all existing patches carried forward
- **49 fingerprint patches** Linux arm64 now matches Linux x64 on Chromium 146
- **WebRTC IP spoofing** — `--fingerprint-webrtc-ip=auto` resolves your proxy's exit IP and spoofs WebRTC ICE candidates. Auto-injected when using `geoip=True` (no extra network call)
- **Proxy signal removal** — DNS/connect/SSL timing zeroed, proxy cache headers stripped, Proxy-Connection header leak removed
- **`cloakserve` CDP multiplexer** — rewritten as a multi-connection CDP proxy with per-connection fingerprint seeds
@@ -242,8 +243,9 @@ browser = launch()
# Headed mode (see the browser window)
browser = launch(headless=False)
# With proxy
# With proxy (HTTP or SOCKS5)
browser = launch(proxy="http://user:pass@proxy:8080")
browser = launch(proxy="socks5://user:pass@proxy:1080")
# With proxy dict (bypass, separate auth fields)
browser = launch(proxy={"server": "http://proxy:8080", "bypass": ".google.com", "username": "user", "password": "pass"})
@@ -370,7 +372,7 @@ from cloakbrowser import binary_info, clear_cache, ensure_binary
# Check binary installation status
print(binary_info())
# {'version': '146.0.7680.177.1', 'platform': 'linux-x64', 'installed': True, ...}
# {'version': '146.0.7680.177.2', 'platform': 'linux-x64', 'installed': True, ...}
# Force re-download
clear_cache()
@@ -645,8 +647,16 @@ stealth_args = get_default_stealth_args() # all fingerprint flags
from cloakbrowser import launch_async
browser = await launch_async(args=["--remote-debugging-port=9242"])
# Connect your framework to http://127.0.0.1:9242 — all stealth flags are set
# Note: humanize requires the wrapper (see below)
```
> **Humanize over CDP**: Stealth fingerprint patches work automatically over CDP, but `humanize=True` is a wrapper-level feature. If you connect to CloakBrowser via CDP from a separate script, import the patching functions to add humanization:
>
> ```js
> import { patchBrowser, resolveConfig } from 'cloakbrowser/human';
> patchBrowser(browser, resolveConfig('default'));
> ```
| Framework | Stars | Language | Example |
|-----------|-------|----------|---------|
| [browser-use](https://github.com/browser-use/browser-use) | 70K | Python | [`browser_use_example.py`](examples/integrations/browser_use_example.py) |
@@ -664,7 +674,7 @@ browser = await launch_async(args=["--remote-debugging-port=9242"])
| Platform | Chromium | Patches | Status |
|---|---|---|---|
| Linux x86_64 | 146 | 49 | ✅ Latest |
| Linux arm64 (RPi, Graviton) | 145 | 48 | ✅ |
| Linux arm64 (RPi, Graviton) | 146 | 49 | ✅ Latest |
| macOS arm64 (Apple Silicon) | 145 | 26 | ✅ |
| macOS x86_64 (Intel) | 145 | 26 | ✅ |
| Windows x86_64 | 145 | 48 | ✅ |
@@ -1020,7 +1030,7 @@ A: Camoufox patches Firefox. We patch Chromium. Chromium means native Playwright
A: Possibly. Bot detection is an arms race. Source-level patches are harder to detect than config-level patches, but not impossible. We actively monitor and update when detection evolves.
**Q: Can I use my own proxy?**
A: Yes. Pass `proxy="http://user:pass@host:port"` to `launch()`.
A: Yes. Pass `proxy="http://user:pass@host:port"` or `proxy="socks5://user:pass@host:port"` to `launch()`. Both HTTP and SOCKS5 proxies are supported natively.
## Roadmap
@@ -1050,7 +1060,7 @@ All releases are signed for supply chain verification.
```bash
# Verify GPG signature (binary release tag)
gpg --keyserver keyserver.ubuntu.com --recv-keys C60C0DDC9D0DE2DD
git verify-tag chromium-v146.0.7680.177.1
git verify-tag chromium-v146.0.7680.177.2
# Verify GitHub binary attestation (Sigstore)
gh attestation verify cloakbrowser-linux-x64.tar.gz --repo CloakHQ/cloakbrowser
+1 -1
View File
@@ -1 +1 @@
__version__ = "0.3.23"
__version__ = "0.3.24"
+85 -21
View File
@@ -17,7 +17,7 @@ from __future__ import annotations
import logging
import os
from typing import Any, Literal, TypedDict
from urllib.parse import unquote, urlparse, urlunparse
from urllib.parse import quote, unquote, urlparse, urlunparse
from .config import DEFAULT_VIEWPORT, IGNORE_DEFAULT_ARGS, get_default_stealth_args
from .download import ensure_binary
@@ -105,11 +105,12 @@ def launch(
binary_path = ensure_binary()
timezone, locale, exit_ip = maybe_resolve_geoip(geoip, proxy, timezone, locale)
proxy_kwargs, proxy_extra_args = _resolve_proxy_config(proxy)
args = _resolve_webrtc_args(args, proxy)
if exit_ip and not (args and any(a.startswith("--fingerprint-webrtc-ip") for a in args)):
args = list(args or [])
args.append(f"--fingerprint-webrtc-ip={exit_ip}")
chrome_args = build_args(stealth_args, args, timezone=timezone, locale=locale, headless=headless)
chrome_args = build_args(stealth_args, (args or []) + proxy_extra_args, timezone=timezone, locale=locale, headless=headless)
logger.debug("Launching stealth Chromium (headless=%s, args=%d)", headless, len(chrome_args))
@@ -119,7 +120,7 @@ def launch(
headless=headless,
args=chrome_args,
ignore_default_args=IGNORE_DEFAULT_ARGS,
**_build_proxy_kwargs(proxy),
**proxy_kwargs,
**kwargs,
)
@@ -194,11 +195,12 @@ async def launch_async( # noqa: C901
binary_path = ensure_binary()
timezone, locale, exit_ip = maybe_resolve_geoip(geoip, proxy, timezone, locale)
proxy_kwargs, proxy_extra_args = _resolve_proxy_config(proxy)
args = _resolve_webrtc_args(args, proxy)
if exit_ip and not (args and any(a.startswith("--fingerprint-webrtc-ip") for a in args)):
args = list(args or [])
args.append(f"--fingerprint-webrtc-ip={exit_ip}")
chrome_args = build_args(stealth_args, args, timezone=timezone, locale=locale, headless=headless)
chrome_args = build_args(stealth_args, (args or []) + proxy_extra_args, timezone=timezone, locale=locale, headless=headless)
logger.debug("Launching stealth Chromium async (headless=%s, args=%d)", headless, len(chrome_args))
@@ -208,7 +210,7 @@ async def launch_async( # noqa: C901
headless=headless,
args=chrome_args,
ignore_default_args=IGNORE_DEFAULT_ARGS,
**_build_proxy_kwargs(proxy),
**proxy_kwargs,
**kwargs,
)
@@ -297,11 +299,12 @@ def launch_persistent_context(
binary_path = ensure_binary()
timezone, locale, exit_ip = maybe_resolve_geoip(geoip, proxy, timezone, locale)
proxy_kwargs, proxy_extra_args = _resolve_proxy_config(proxy)
args = _resolve_webrtc_args(args, proxy)
if exit_ip and not (args and any(a.startswith("--fingerprint-webrtc-ip") for a in args)):
args = list(args or [])
args.append(f"--fingerprint-webrtc-ip={exit_ip}")
chrome_args = build_args(stealth_args, args, timezone=timezone, locale=locale, headless=headless)
chrome_args = build_args(stealth_args, (args or []) + proxy_extra_args, timezone=timezone, locale=locale, headless=headless)
logger.debug(
"Launching persistent stealth Chromium (headless=%s, user_data_dir=%s)",
@@ -331,7 +334,7 @@ def launch_persistent_context(
headless=headless,
args=chrome_args,
ignore_default_args=IGNORE_DEFAULT_ARGS,
**_build_proxy_kwargs(proxy),
**proxy_kwargs,
**context_kwargs,
)
@@ -422,11 +425,12 @@ async def launch_persistent_context_async(
binary_path = ensure_binary()
timezone, locale, exit_ip = maybe_resolve_geoip(geoip, proxy, timezone, locale)
proxy_kwargs, proxy_extra_args = _resolve_proxy_config(proxy)
args = _resolve_webrtc_args(args, proxy)
if exit_ip and not (args and any(a.startswith("--fingerprint-webrtc-ip") for a in args)):
args = list(args or [])
args.append(f"--fingerprint-webrtc-ip={exit_ip}")
chrome_args = build_args(stealth_args, args, timezone=timezone, locale=locale, headless=headless)
chrome_args = build_args(stealth_args, (args or []) + proxy_extra_args, timezone=timezone, locale=locale, headless=headless)
logger.debug(
"Launching persistent stealth Chromium async (headless=%s, user_data_dir=%s)",
@@ -456,7 +460,7 @@ async def launch_persistent_context_async(
headless=headless,
args=chrome_args,
ignore_default_args=IGNORE_DEFAULT_ARGS,
**_build_proxy_kwargs(proxy),
**proxy_kwargs,
**context_kwargs,
)
@@ -631,14 +635,42 @@ def _ensure_proxy_scheme(proxy_url: str) -> str:
return proxy_url if "://" in proxy_url else f"http://{proxy_url}"
def _reconstruct_socks_url(proxy: ProxySettings) -> str:
"""Reconstruct a SOCKS5 URL with inline credentials from a Playwright proxy dict."""
server = proxy.get("server", "")
username = proxy.get("username", "")
password = proxy.get("password", "")
if not username:
return server
parsed = urlparse(server)
creds = quote(username, safe="")
if password:
creds += f":{quote(password, safe='')}"
host = parsed.hostname or ""
if ":" in host: # IPv6 literal — re-add brackets
host = f"[{host}]"
netloc = f"{creds}@{host}"
if parsed.port:
netloc += f":{parsed.port}"
return urlunparse((parsed.scheme, netloc, parsed.path, "", "", ""))
def _extract_proxy_url(proxy: str | ProxySettings | None) -> str | None:
"""Extract and normalize proxy URL string from proxy param."""
"""Extract and normalize proxy URL string from proxy param.
For SOCKS5 dicts with separate username/password fields, reconstructs
the full URL with inline credentials so SOCKS5 auth works.
"""
if proxy is None:
return None
raw = proxy.get("server") if isinstance(proxy, dict) else proxy
if not raw:
return None
return _ensure_proxy_scheme(raw)
if isinstance(proxy, dict):
server = proxy.get("server", "")
if not server:
return None
if _is_socks_proxy(proxy):
return _reconstruct_socks_url(proxy)
return _ensure_proxy_scheme(server)
return _ensure_proxy_scheme(proxy)
def maybe_resolve_geoip(
@@ -694,7 +726,7 @@ def _resolve_webrtc_args(
return args
proxy_url = _extract_proxy_url(proxy)
if not proxy_url:
logger.debug("--fingerprint-webrtc-ip=auto but no proxy set — removing flag")
logger.warning("--fingerprint-webrtc-ip=auto requires a proxy; removing flag")
args = list(args)
del args[idx]
return args
@@ -702,7 +734,7 @@ def _resolve_webrtc_args(
from .geoip import _resolve_exit_ip
exit_ip = _resolve_exit_ip(proxy_url)
except Exception:
logger.debug("WebRTC IP resolution failed — removing flag")
logger.warning("Failed to resolve proxy exit IP for WebRTC spoofing; removing --fingerprint-webrtc-ip=auto")
args = list(args)
del args[idx]
return args
@@ -710,6 +742,7 @@ def _resolve_webrtc_args(
args = list(args)
args[idx] = f"--fingerprint-webrtc-ip={exit_ip}"
else:
logger.warning("Could not resolve proxy exit IP for WebRTC spoofing; removing --fingerprint-webrtc-ip=auto")
args = list(args)
del args[idx]
return args
@@ -799,10 +832,41 @@ def _parse_proxy_url(proxy: str) -> dict[str, Any]:
return result
def _build_proxy_kwargs(proxy: str | ProxySettings | None) -> dict[str, Any]:
"""Build proxy kwargs for Playwright launch."""
def _is_socks_proxy(proxy: str | ProxySettings | None) -> bool:
"""Check if the proxy uses SOCKS5 protocol."""
if proxy is None:
return {}
return False
url = proxy.get("server", "") if isinstance(proxy, dict) else proxy
return url.lower().startswith(("socks5://", "socks5h://"))
def _resolve_proxy_config(
proxy: str | ProxySettings | None,
) -> tuple[dict[str, Any], list[str]]:
"""Resolve proxy into Playwright kwargs and Chrome args.
Playwright rejects SOCKS5 proxies with credentials in its proxy dict,
so SOCKS5 is passed via --proxy-server Chrome arg instead.
Returns:
(proxy_kwargs, extra_chrome_args) — one or both will be empty.
"""
if proxy is None:
return {}, []
if _is_socks_proxy(proxy):
# SOCKS5: bypass Playwright, pass directly to Chrome via --proxy-server.
# Chrome handles SOCKS5 auth natively from the URL.
if isinstance(proxy, dict):
url = _reconstruct_socks_url(proxy)
extra_args = [f"--proxy-server={url}"]
if proxy.get("bypass"):
extra_args.append(f"--proxy-bypass-list={proxy['bypass']}")
return {}, extra_args
# String URL — pass as-is (Chrome handles user:pass@ in the URL)
return {}, [f"--proxy-server={proxy}"]
# HTTP/HTTPS: use Playwright's proxy dict as before
if isinstance(proxy, dict):
return {"proxy": proxy}
return {"proxy": _parse_proxy_url(proxy)}
return {"proxy": proxy}, []
return {"proxy": _parse_proxy_url(proxy)}, []
+2 -2
View File
@@ -18,8 +18,8 @@ from ._version import __version__
CHROMIUM_VERSION = "146.0.7680.177.1"
PLATFORM_CHROMIUM_VERSIONS: dict[str, str] = {
"linux-x64": "146.0.7680.177.1",
"linux-arm64": "145.0.7632.159.7",
"linux-x64": "146.0.7680.177.2",
"linux-arm64": "146.0.7680.177.2",
"darwin-arm64": "145.0.7632.109.2",
"darwin-x64": "145.0.7632.109.2",
"windows-x64": "145.0.7632.159.7",
+8 -3
View File
@@ -96,7 +96,7 @@ def resolve_proxy_geo_with_ip(
)
return timezone, locale, ip
except Exception as exc:
logger.debug("GeoIP lookup failed for %s: %s", ip, exc)
logger.warning("GeoIP lookup failed for %s: %s", ip, exc)
return None, None, ip
@@ -132,7 +132,7 @@ def _resolve_proxy_ip(proxy_url: str) -> str | None:
return ip
return None
except Exception as exc:
logger.debug("Failed to resolve proxy hostname: %s", exc)
logger.warning("Failed to resolve proxy hostname: %s", exc)
return None
@@ -165,9 +165,14 @@ def _resolve_exit_ip(proxy_url: str) -> str | None:
ipaddress.ip_address(ip)
logger.debug("Exit IP via %s: %s", url, ip)
return ip
except httpx.UnsupportedProtocol:
logger.warning(
"SOCKS5 proxy requires socksio: pip install cloakbrowser[geoip]"
)
return None
except Exception:
continue
logger.debug("Failed to discover exit IP through proxy")
logger.warning("Failed to discover exit IP through proxy")
return None
+649
View File
@@ -900,6 +900,9 @@ def patch_page(page: Any, cfg: HumanConfig, cursor: _CursorState) -> None:
# --- Patch Frame-level methods (for sub-frames) ---
_patch_frames_sync(page, cfg, cursor, raw_mouse, raw_keyboard, originals)
# --- Patch ElementHandle selectors (query_selector, query_selector_all, wait_for_selector) ---
_patch_page_element_handles_sync(page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session)
# Initialize cursor immediately so it doesn't visibly jump from (0,0)
cursor.x = rand(cfg.initial_cursor_x[0], cfg.initial_cursor_x[1])
cursor.y = rand(cfg.initial_cursor_y[0], cfg.initial_cursor_y[1])
@@ -913,6 +916,273 @@ def patch_page(page: Any, cfg: HumanConfig, cursor: _CursorState) -> None:
_patch_locator_class_sync()
# ============================================================================
# SYNC ElementHandle patching
# ============================================================================
def _is_input_element_handle_sync(el: Any) -> bool:
"""Check if an ElementHandle is an input/textarea/contenteditable (sync)."""
try:
return el.evaluate(
"""(node) => {
const tag = node.tagName ? node.tagName.toLowerCase() : '';
return tag === 'input' || tag === 'textarea'
|| node.getAttribute && node.getAttribute('contenteditable') === 'true';
}"""
)
except Exception:
return False
def _patch_single_element_handle_sync(
el: Any, page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: RawMouse, raw_keyboard: RawKeyboard, originals: Any,
stealth: Any, cdp_session: Any,
) -> None:
"""Patch all interaction methods on a sync Playwright ElementHandle."""
if getattr(el, '_human_patched', False):
return
el._human_patched = True
# Save originals
_orig_click = el.click
_orig_dblclick = el.dblclick
_orig_hover = el.hover
_orig_type = el.type
_orig_fill = el.fill
_orig_press = el.press
_orig_select_option = el.select_option
_orig_check = el.check
_orig_uncheck = el.uncheck
_orig_set_checked = getattr(el, 'set_checked', None)
_orig_tap = el.tap
_orig_focus = el.focus
# Nested selectors
_orig_qs = el.query_selector
_orig_qsa = el.query_selector_all
_orig_wfs = el.wait_for_selector
def _patched_qs(selector: str, **kwargs: Any) -> Any:
child = _orig_qs(selector, **kwargs)
if child is not None:
_patch_single_element_handle_sync(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return child
def _patched_qsa(selector: str, **kwargs: Any) -> Any:
children = _orig_qsa(selector, **kwargs)
for child in children:
_patch_single_element_handle_sync(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return children
def _patched_wfs(selector: str, **kwargs: Any) -> Any:
child = _orig_wfs(selector, **kwargs)
if child is not None:
_patch_single_element_handle_sync(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return child
el.query_selector = _patched_qs
el.query_selector_all = _patched_qsa
el.wait_for_selector = _patched_wfs
# Helper: move cursor to element
def _move_to_element():
if not cursor.initialized:
cursor.x = rand(cfg.initial_cursor_x[0], cfg.initial_cursor_x[1])
cursor.y = rand(cfg.initial_cursor_y[0], cfg.initial_cursor_y[1])
originals.mouse_move(cursor.x, cursor.y)
cursor.initialized = True
box = el.bounding_box()
if not box:
return None
is_inp = _is_input_element_handle_sync(el)
target = click_target(box, is_inp, cfg)
if cfg.idle_between_actions:
human_idle(raw_mouse, rand(cfg.idle_between_duration[0], cfg.idle_between_duration[1]), cursor.x, cursor.y, cfg)
human_move(raw_mouse, cursor.x, cursor.y, target.x, target.y, cfg)
cursor.x = target.x
cursor.y = target.y
return {'box': box, 'is_inp': is_inp}
# --- el.click() ---
def _human_el_click(**kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_click(**kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
# --- el.dblclick() ---
def _human_el_dblclick(**kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_dblclick(**kwargs)
raw_mouse.down(click_count=2)
sleep_ms(rand(30, 60))
raw_mouse.up(click_count=2)
# --- el.hover() ---
def _human_el_hover(**kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_hover(**kwargs)
# Just move, no click
# --- el.type() ---
def _human_el_type(text: str, **kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_type(text, **kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
sleep_ms(rand(100, 250))
human_type(page, raw_keyboard, text, cfg, cdp_session=cdp_session)
# --- el.fill() ---
def _human_el_fill(value: str, **kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_fill(value, **kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
sleep_ms(rand(100, 250))
originals.keyboard_press(_SELECT_ALL)
sleep_ms(rand(30, 80))
originals.keyboard_press("Backspace")
sleep_ms(rand(50, 150))
human_type(page, raw_keyboard, value, cfg, cdp_session=cdp_session)
# --- el.press() ---
def _human_el_press(key: str, **kwargs: Any) -> None:
sleep_ms(rand(20, 60))
originals.keyboard_down(key)
sleep_ms(rand_range(cfg.key_hold))
originals.keyboard_up(key)
# --- el.select_option() ---
def _human_el_select_option(value: Any = None, **kwargs: Any) -> Any:
info = _move_to_element()
if info is None:
return _orig_select_option(value, **kwargs)
human_click(raw_mouse, False, cfg)
sleep_ms(rand(100, 300))
return _orig_select_option(value, **kwargs)
# --- el.check() ---
def _human_el_check(**kwargs: Any) -> None:
try:
if el.is_checked():
return
except Exception:
pass
info = _move_to_element()
if info is None:
return _orig_check(**kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
# --- el.uncheck() ---
def _human_el_uncheck(**kwargs: Any) -> None:
try:
if not el.is_checked():
return
except Exception:
pass
info = _move_to_element()
if info is None:
return _orig_uncheck(**kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
# --- el.set_checked() ---
def _human_el_set_checked(checked: bool, **kwargs: Any) -> None:
try:
current = el.is_checked()
if current == checked:
return
except Exception:
pass
info = _move_to_element()
if info is None and _orig_set_checked:
return _orig_set_checked(checked, **kwargs)
if info:
human_click(raw_mouse, info['is_inp'], cfg)
# --- el.tap() ---
def _human_el_tap(**kwargs: Any) -> None:
info = _move_to_element()
if info is None:
return _orig_tap(**kwargs)
human_click(raw_mouse, info['is_inp'], cfg)
# --- el.focus() ---
# FIX: move cursor humanly but use programmatic focus (no click side-effects).
# Stock Playwright el.focus() never clicks — it just calls element.focus() in JS.
# Clicking would trigger onclick, submit forms, navigate links, etc.
def _human_el_focus() -> None:
_move_to_element() # human-like cursor movement (Bézier)
_orig_focus() # programmatic focus, no click side-effects
el.click = _human_el_click
el.dblclick = _human_el_dblclick
el.hover = _human_el_hover
el.type = _human_el_type
el.fill = _human_el_fill
el.press = _human_el_press
el.select_option = _human_el_select_option
el.check = _human_el_check
el.uncheck = _human_el_uncheck
if _orig_set_checked is not None:
el.set_checked = _human_el_set_checked
el.tap = _human_el_tap
el.focus = _human_el_focus
def _patch_page_element_handles_sync(
page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: RawMouse, raw_keyboard: RawKeyboard, originals: Any,
stealth: Any, cdp_session: Any,
) -> None:
"""Patch page.query_selector, query_selector_all, wait_for_selector to return humanized ElementHandles (sync)."""
_orig_qs = page.query_selector
_orig_qsa = page.query_selector_all
_orig_wfs = page.wait_for_selector
def _patched_qs(selector: str, **kwargs: Any) -> Any:
el = _orig_qs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return el
def _patched_qsa(selector: str, **kwargs: Any) -> Any:
els = _orig_qsa(selector, **kwargs)
for el in els:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return els
def _patched_wfs(selector: str, **kwargs: Any) -> Any:
el = _orig_wfs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return el
page.query_selector = _patched_qs
page.query_selector_all = _patched_qsa
page.wait_for_selector = _patched_wfs
def _patch_frames_sync(
page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: RawMouse, raw_keyboard: RawKeyboard, originals: Any,
@@ -1023,6 +1293,58 @@ def _patch_single_frame_sync(
frame.clear = _frame_clear
frame.drag_and_drop = _frame_drag_and_drop
# --- Patch frame-level ElementHandle selectors ---
stealth_world = getattr(page, '_stealth_world', None)
cdp_session = None
if stealth_world is not None:
try:
cdp_session = stealth_world.get_cdp_session()
except Exception:
pass
_patch_frame_element_handles_sync(
frame, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth_world, cdp_session
)
def _patch_frame_element_handles_sync(
frame: Any, page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: RawMouse, raw_keyboard: RawKeyboard, originals: Any,
stealth: Any, cdp_session: Any,
) -> None:
"""Patch frame.query_selector, query_selector_all, wait_for_selector (sync)."""
_orig_qs = frame.query_selector
_orig_qsa = frame.query_selector_all
_orig_wfs = frame.wait_for_selector
def _patched_qs(selector: str, **kwargs: Any) -> Any:
el = _orig_qs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return el
def _patched_qsa(selector: str, **kwargs: Any) -> Any:
els = _orig_qsa(selector, **kwargs)
for el in els:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return els
def _patched_wfs(selector: str, **kwargs: Any) -> Any:
el = _orig_wfs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session
)
return el
frame.query_selector = _patched_qs
frame.query_selector_all = _patched_qsa
frame.wait_for_selector = _patched_wfs
def _iter_frames(page: Any):
try:
@@ -1108,6 +1430,7 @@ def patch_page_async(page: Any, cfg: HumanConfig, cursor: _CursorState) -> None:
stealth = _AsyncIsolatedWorld(page)
page._stealth_world = stealth
cdp_session_holder: list[Any] = [None] # mutable container for closure
page._cdp_session_holder = cdp_session_holder # expose for frame-level patching
async def _ensure_cdp() -> Any:
if cdp_session_holder[0] is None:
@@ -1265,10 +1588,289 @@ def patch_page_async(page: Any, cfg: HumanConfig, cursor: _CursorState) -> None:
# --- Patch Frame-level methods (for sub-frames) ---
_patch_frames_async(page, cfg, cursor, raw_mouse, raw_keyboard, originals)
# --- Patch ElementHandle selectors (query_selector, query_selector_all, wait_for_selector) ---
_patch_page_element_handles_async(page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder)
# --- Patch async Locator class (class-level, runs once) ---
_patch_locator_class_async()
# ============================================================================
# ASYNC ElementHandle patching
# ============================================================================
async def _async_is_input_element_handle(el: Any) -> bool:
"""Check if an ElementHandle is an input/textarea/contenteditable (async)."""
try:
return await el.evaluate(
"""(node) => {
const tag = node.tagName ? node.tagName.toLowerCase() : '';
return tag === 'input' || tag === 'textarea'
|| node.getAttribute && node.getAttribute('contenteditable') === 'true';
}"""
)
except Exception:
return False
def _patch_single_element_handle_async(
el: Any, page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: AsyncRawMouse, raw_keyboard: AsyncRawKeyboard, originals: Any,
stealth: Any, cdp_session_holder: Any,
) -> None:
"""Patch all interaction methods on an async Playwright ElementHandle."""
if getattr(el, '_human_patched', False):
return
el._human_patched = True
# Save originals
_orig_click = el.click
_orig_dblclick = el.dblclick
_orig_hover = el.hover
_orig_type = el.type
_orig_fill = el.fill
_orig_press = el.press
_orig_select_option = el.select_option
_orig_check = el.check
_orig_uncheck = el.uncheck
_orig_set_checked = getattr(el, 'set_checked', None)
_orig_tap = el.tap
_orig_focus = el.focus
# Nested selectors
_orig_qs = el.query_selector
_orig_qsa = el.query_selector_all
_orig_wfs = el.wait_for_selector
async def _patched_qs(selector: str, **kwargs: Any) -> Any:
child = await _orig_qs(selector, **kwargs)
if child is not None:
_patch_single_element_handle_async(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return child
async def _patched_qsa(selector: str, **kwargs: Any) -> Any:
children = await _orig_qsa(selector, **kwargs)
for child in children:
_patch_single_element_handle_async(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return children
async def _patched_wfs(selector: str, **kwargs: Any) -> Any:
child = await _orig_wfs(selector, **kwargs)
if child is not None:
_patch_single_element_handle_async(
child, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return child
el.query_selector = _patched_qs
el.query_selector_all = _patched_qsa
el.wait_for_selector = _patched_wfs
# Helper: move cursor to element (async)
async def _move_to_element():
if not cursor.initialized:
cursor.x = rand(cfg.initial_cursor_x[0], cfg.initial_cursor_x[1])
cursor.y = rand(cfg.initial_cursor_y[0], cfg.initial_cursor_y[1])
await originals.mouse_move(cursor.x, cursor.y)
cursor.initialized = True
box = await el.bounding_box()
if not box:
return None
is_inp = await _async_is_input_element_handle(el)
target = click_target(box, is_inp, cfg)
if cfg.idle_between_actions:
await async_human_idle(raw_mouse, rand(cfg.idle_between_duration[0], cfg.idle_between_duration[1]), cursor.x, cursor.y, cfg)
await async_human_move(raw_mouse, cursor.x, cursor.y, target.x, target.y, cfg)
cursor.x = target.x
cursor.y = target.y
return {'box': box, 'is_inp': is_inp}
async def _get_cdp():
if cdp_session_holder[0] is None:
try:
cdp_session_holder[0] = await stealth.get_cdp_session()
except Exception:
pass
return cdp_session_holder[0]
# --- el.click() ---
async def _human_el_click(**kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_click(**kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
# --- el.dblclick() ---
async def _human_el_dblclick(**kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_dblclick(**kwargs)
await raw_mouse.down(click_count=2)
await async_sleep_ms(rand(30, 60))
await raw_mouse.up(click_count=2)
# --- el.hover() ---
async def _human_el_hover(**kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_hover(**kwargs)
# --- el.type() ---
async def _human_el_type(text: str, **kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_type(text, **kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
await async_sleep_ms(rand(100, 250))
cdp = await _get_cdp()
await async_human_type(page, raw_keyboard, text, cfg, cdp_session=cdp)
# --- el.fill() ---
async def _human_el_fill(value: str, **kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_fill(value, **kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
await async_sleep_ms(rand(100, 250))
await originals.keyboard_press(_SELECT_ALL)
await async_sleep_ms(rand(30, 80))
await originals.keyboard_press("Backspace")
await async_sleep_ms(rand(50, 150))
cdp = await _get_cdp()
await async_human_type(page, raw_keyboard, value, cfg, cdp_session=cdp)
# --- el.press() ---
async def _human_el_press(key: str, **kwargs: Any) -> None:
await async_sleep_ms(rand(20, 60))
await originals.keyboard_down(key)
await async_sleep_ms(rand_range(cfg.key_hold))
await originals.keyboard_up(key)
# --- el.select_option() ---
async def _human_el_select_option(value: Any = None, **kwargs: Any) -> Any:
info = await _move_to_element()
if info is None:
return await _orig_select_option(value, **kwargs)
await async_human_click(raw_mouse, False, cfg)
await async_sleep_ms(rand(100, 300))
return await _orig_select_option(value, **kwargs)
# --- el.check() ---
async def _human_el_check(**kwargs: Any) -> None:
try:
if await el.is_checked():
return
except Exception:
pass
info = await _move_to_element()
if info is None:
return await _orig_check(**kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
# --- el.uncheck() ---
async def _human_el_uncheck(**kwargs: Any) -> None:
try:
if not await el.is_checked():
return
except Exception:
pass
info = await _move_to_element()
if info is None:
return await _orig_uncheck(**kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
# --- el.set_checked() ---
async def _human_el_set_checked(checked: bool, **kwargs: Any) -> None:
try:
current = await el.is_checked()
if current == checked:
return
except Exception:
pass
info = await _move_to_element()
if info is None and _orig_set_checked:
return await _orig_set_checked(checked, **kwargs)
if info:
await async_human_click(raw_mouse, info['is_inp'], cfg)
# --- el.tap() ---
async def _human_el_tap(**kwargs: Any) -> None:
info = await _move_to_element()
if info is None:
return await _orig_tap(**kwargs)
await async_human_click(raw_mouse, info['is_inp'], cfg)
# --- el.focus() ---
# FIX: move cursor humanly but use programmatic focus (no click side-effects).
# Stock Playwright el.focus() never clicks — it just calls element.focus() in JS.
# Clicking would trigger onclick, submit forms, navigate links, etc.
async def _human_el_focus() -> None:
await _move_to_element() # human-like cursor movement (Bézier)
await _orig_focus() # programmatic focus, no click side-effects
el.click = _human_el_click
el.dblclick = _human_el_dblclick
el.hover = _human_el_hover
el.type = _human_el_type
el.fill = _human_el_fill
el.press = _human_el_press
el.select_option = _human_el_select_option
el.check = _human_el_check
el.uncheck = _human_el_uncheck
if _orig_set_checked is not None:
el.set_checked = _human_el_set_checked
el.tap = _human_el_tap
el.focus = _human_el_focus
def _patch_page_element_handles_async(
page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: AsyncRawMouse, raw_keyboard: AsyncRawKeyboard, originals: Any,
stealth: Any, cdp_session_holder: Any,
) -> None:
"""Patch page.query_selector, query_selector_all, wait_for_selector to return humanized ElementHandles (async)."""
_orig_qs = page.query_selector
_orig_qsa = page.query_selector_all
_orig_wfs = page.wait_for_selector
async def _patched_qs(selector: str, **kwargs: Any) -> Any:
el = await _orig_qs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return el
async def _patched_qsa(selector: str, **kwargs: Any) -> Any:
els = await _orig_qsa(selector, **kwargs)
for el in els:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return els
async def _patched_wfs(selector: str, **kwargs: Any) -> Any:
el = await _orig_wfs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return el
page.query_selector = _patched_qs
page.query_selector_all = _patched_qsa
page.wait_for_selector = _patched_wfs
def _patch_frames_async(
page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: AsyncRawMouse, raw_keyboard: AsyncRawKeyboard, originals: Any,
@@ -1379,6 +1981,53 @@ def _patch_single_frame_async(
frame.clear = _frame_clear
frame.drag_and_drop = _frame_drag_and_drop
# --- Patch frame-level ElementHandle selectors (async) ---
stealth_world = getattr(page, '_stealth_world', None)
cdp_session_holder = getattr(page, '_cdp_session_holder', [None])
_patch_frame_element_handles_async(
frame, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth_world, cdp_session_holder
)
def _patch_frame_element_handles_async(
frame: Any, page: Any, cfg: HumanConfig, cursor: _CursorState,
raw_mouse: AsyncRawMouse, raw_keyboard: AsyncRawKeyboard, originals: Any,
stealth: Any, cdp_session_holder: Any,
) -> None:
"""Patch frame.query_selector, query_selector_all, wait_for_selector (async)."""
_orig_qs = frame.query_selector
_orig_qsa = frame.query_selector_all
_orig_wfs = frame.wait_for_selector
async def _patched_qs(selector: str, **kwargs: Any) -> Any:
el = await _orig_qs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return el
async def _patched_qsa(selector: str, **kwargs: Any) -> Any:
els = await _orig_qsa(selector, **kwargs)
for el in els:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return els
async def _patched_wfs(selector: str, **kwargs: Any) -> Any:
el = await _orig_wfs(selector, **kwargs)
if el is not None:
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, cdp_session_holder
)
return el
frame.query_selector = _patched_qs
frame.query_selector_all = _patched_qsa
frame.wait_for_selector = _patched_wfs
def patch_context_async(context: Any, cfg: HumanConfig) -> None:
cursor = _CursorState()
+5 -2
View File
@@ -63,10 +63,13 @@ await browser.close();
```javascript
import { launch, launchContext, launchPersistentContext } from 'cloakbrowser';
// With proxy
// With proxy (HTTP or SOCKS5)
const browser = await launch({
proxy: 'http://user:pass@proxy:8080',
});
const browser = await launch({
proxy: 'socks5://user:pass@proxy:1080',
});
// With proxy object (bypass, separate auth fields)
const browser = await launch({
@@ -211,7 +214,7 @@ const page = await browser.newPage();
## Requirements
- Node.js >= 18
- Node.js >= 20
- One of: `playwright-core` >= 1.40 or `puppeteer-core` >= 21
## Troubleshooting
+57 -9
View File
@@ -1,31 +1,36 @@
{
"name": "cloakbrowser",
"version": "0.3.9",
"version": "0.3.23",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "cloakbrowser",
"version": "0.3.9",
"version": "0.3.23",
"license": "MIT",
"dependencies": {
"tar": "^7.0.0"
},
"bin": {
"cloakbrowser": "dist/cli.js"
},
"devDependencies": {
"@types/node": "^20.10.0",
"mmdb-lib": "^3.0.2",
"playwright-core": "^1.40.0",
"puppeteer-core": "^21.0.0",
"socks-proxy-agent": "^10.0.0",
"typescript": "^5.3.0",
"vitest": "^1.0.0"
},
"engines": {
"node": ">=18.0.0"
"node": ">=20.0.0"
},
"peerDependencies": {
"mmdb-lib": ">=2.0.0",
"playwright-core": ">=1.40.0",
"puppeteer-core": ">=21.0.0"
"puppeteer-core": ">=21.0.0",
"socks-proxy-agent": ">=8.0.0"
},
"peerDependenciesMeta": {
"mmdb-lib": {
@@ -36,6 +41,9 @@
},
"puppeteer-core": {
"optional": true
},
"socks-proxy-agent": {
"optional": true
}
}
},
@@ -2003,6 +2011,21 @@
"node": ">= 14"
}
},
"node_modules/pac-proxy-agent/node_modules/socks-proxy-agent": {
"version": "8.0.5",
"resolved": "https://registry.npmjs.org/socks-proxy-agent/-/socks-proxy-agent-8.0.5.tgz",
"integrity": "sha512-HehCEsotFqbPW9sJ8WVYB6UbmIMv7kUUORIF2Nncq4VQvBfNBLibW9YZR5dlYCSUhwcD628pRllm7n+E+YTzJw==",
"dev": true,
"license": "MIT",
"dependencies": {
"agent-base": "^7.1.2",
"debug": "^4.3.4",
"socks": "^2.8.3"
},
"engines": {
"node": ">= 14"
}
},
"node_modules/pac-resolver": {
"version": "7.0.1",
"resolved": "https://registry.npmjs.org/pac-resolver/-/pac-resolver-7.0.1.tgz",
@@ -2164,6 +2187,21 @@
"node": ">= 14"
}
},
"node_modules/proxy-agent/node_modules/socks-proxy-agent": {
"version": "8.0.5",
"resolved": "https://registry.npmjs.org/socks-proxy-agent/-/socks-proxy-agent-8.0.5.tgz",
"integrity": "sha512-HehCEsotFqbPW9sJ8WVYB6UbmIMv7kUUORIF2Nncq4VQvBfNBLibW9YZR5dlYCSUhwcD628pRllm7n+E+YTzJw==",
"dev": true,
"license": "MIT",
"dependencies": {
"agent-base": "^7.1.2",
"debug": "^4.3.4",
"socks": "^2.8.3"
},
"engines": {
"node": ">= 14"
}
},
"node_modules/proxy-from-env": {
"version": "1.1.0",
"resolved": "https://registry.npmjs.org/proxy-from-env/-/proxy-from-env-1.1.0.tgz",
@@ -2332,18 +2370,28 @@
}
},
"node_modules/socks-proxy-agent": {
"version": "8.0.5",
"resolved": "https://registry.npmjs.org/socks-proxy-agent/-/socks-proxy-agent-8.0.5.tgz",
"integrity": "sha512-HehCEsotFqbPW9sJ8WVYB6UbmIMv7kUUORIF2Nncq4VQvBfNBLibW9YZR5dlYCSUhwcD628pRllm7n+E+YTzJw==",
"version": "10.0.0",
"resolved": "https://registry.npmjs.org/socks-proxy-agent/-/socks-proxy-agent-10.0.0.tgz",
"integrity": "sha512-pyp2YR3mNxAMu0mGLtzs4g7O3uT4/9sQOLAKcViAkaS9fJWkud7nmaf6ZREFqQEi24IPkBcjfHjXhPTUWjo3uA==",
"dev": true,
"license": "MIT",
"dependencies": {
"agent-base": "^7.1.2",
"agent-base": "9.0.0",
"debug": "^4.3.4",
"socks": "^2.8.3"
},
"engines": {
"node": ">= 14"
"node": ">= 20"
}
},
"node_modules/socks-proxy-agent/node_modules/agent-base": {
"version": "9.0.0",
"resolved": "https://registry.npmjs.org/agent-base/-/agent-base-9.0.0.tgz",
"integrity": "sha512-TQf59BsZnytt8GdJKLPfUZ54g/iaUL2OWDSFCCvMOhsHduDQxO8xC4PNeyIkVcA5KwL2phPSv0douC0fgWzmnA==",
"dev": true,
"license": "MIT",
"engines": {
"node": ">= 20"
}
},
"node_modules/source-map": {
+8 -3
View File
@@ -1,6 +1,6 @@
{
"name": "cloakbrowser",
"version": "0.3.23",
"version": "0.3.24",
"description": "Stealth Chromium that passes every bot detection test. Drop-in Playwright/Puppeteer replacement with source-level fingerprint patches.",
"type": "module",
"main": "dist/index.js",
@@ -55,12 +55,13 @@
},
"homepage": "https://github.com/CloakHQ/cloakbrowser#javascript--nodejs",
"engines": {
"node": ">=18.0.0"
"node": ">=20.0.0"
},
"peerDependencies": {
"mmdb-lib": ">=2.0.0",
"playwright-core": ">=1.40.0",
"puppeteer-core": ">=21.0.0"
"puppeteer-core": ">=21.0.0",
"socks-proxy-agent": ">=10.0.0"
},
"peerDependenciesMeta": {
"playwright-core": {
@@ -71,6 +72,9 @@
},
"mmdb-lib": {
"optional": true
},
"socks-proxy-agent": {
"optional": true
}
},
"dependencies": {
@@ -79,6 +83,7 @@
"devDependencies": {
"@types/node": "^20.10.0",
"mmdb-lib": "^3.0.2",
"socks-proxy-agent": "^10.0.0",
"playwright-core": "^1.40.0",
"puppeteer-core": "^21.0.0",
"typescript": "^5.3.0",
+2 -2
View File
@@ -30,8 +30,8 @@ export { WRAPPER_VERSION };
export const CHROMIUM_VERSION = "146.0.7680.177.1";
export const PLATFORM_CHROMIUM_VERSIONS: Record<string, string> = {
"linux-x64": "146.0.7680.177.1",
"linux-arm64": "145.0.7632.159.7",
"linux-x64": "146.0.7680.177.2",
"linux-arm64": "146.0.7680.177.2",
"darwin-arm64": "145.0.7632.109.2",
"darwin-x64": "145.0.7632.109.2",
"windows-x64": "145.0.7632.159.7",
+60 -8
View File
@@ -15,7 +15,7 @@ import dns from "node:dns/promises";
import net from "node:net";
import { getCacheDir } from "./config.js";
import type { LaunchOptions } from "./types.js";
import { ensureProxyScheme } from "./proxy.js";
import { ensureProxyScheme, isSocksProxy, reconstructSocksUrl, type ProxyDict } from "./proxy.js";
// P3TERX mirror of MaxMind GeoLite2-City — no license key needed
const GEOIP_DB_URL =
@@ -129,9 +129,44 @@ const IP_ECHO_URLS = [
];
async function resolveExitIp(proxyUrl: string): Promise<string | null> {
// Node.js fetch doesn't support proxy natively — use a CONNECT tunnel via http
// For simplicity, use a direct HTTP request to a plain-text IP echo service
// through the proxy using Node's http module
const isSocks = isSocksProxy(proxyUrl);
// SOCKS5: tunnel through the SOCKS5 proxy via socks-proxy-agent
if (isSocks) {
let SocksProxyAgent: typeof import("socks-proxy-agent").SocksProxyAgent;
try {
({ SocksProxyAgent } = await import("socks-proxy-agent"));
} catch {
console.warn("[cloakbrowser] socks-proxy-agent not installed — cannot resolve exit IP through SOCKS5 proxy. Install it: npm install socks-proxy-agent");
return null;
}
const { default: https } = await import("node:https");
const agent = new SocksProxyAgent(proxyUrl);
for (const echoUrl of IP_ECHO_URLS) {
try {
const ip = await new Promise<string | null>((resolve) => {
const req = https.request(echoUrl, { agent, timeout: 10_000 }, (res) => {
let data = "";
res.on("data", (chunk: Buffer) => (data += chunk.toString()));
res.on("end", () => {
const ip = data.trim();
resolve(net.isIP(ip) ? ip : null);
});
});
req.on("error", () => resolve(null));
req.on("timeout", () => { req.destroy(); resolve(null); });
req.end();
});
if (ip) return ip;
} catch {
continue;
}
}
return null;
}
// HTTP/HTTPS: use a CONNECT tunnel via http
try {
const { default: http } = await import("node:http");
const { default: https } = await import("node:https");
@@ -264,6 +299,22 @@ function maybeTriggerUpdate(dbPath: string): void {
downloadGeoipDb(dbPath).catch(() => {});
}
/**
* Extract a usable proxy URL from LaunchOptions.proxy.
* For SOCKS5 dicts with separate credentials, reconstructs the full URL
* with inline credentials so SOCKS5 auth works.
*/
function extractProxyUrl(proxy: string | ProxyDict | undefined): string | null {
if (!proxy) return null;
if (typeof proxy === "string") return ensureProxyScheme(proxy);
const p = proxy as ProxyDict;
if (!p.server) return null;
if (p.username && isSocksProxy(p)) {
return reconstructSocksUrl(p);
}
return ensureProxyScheme(p.server);
}
/**
* Auto-fill timezone/locale from proxy IP when geoip is enabled.
* Also returns exitIp as a free bonus (reused for WebRTC spoofing).
@@ -273,9 +324,8 @@ export async function maybeResolveGeoip(
): Promise<{ timezone?: string; locale?: string; exitIp?: string }> {
if (!options.geoip || !options.proxy) return { timezone: options.timezone, locale: options.locale };
let proxyUrl = typeof options.proxy === "string" ? options.proxy : options.proxy.server;
const proxyUrl = extractProxyUrl(options.proxy);
if (!proxyUrl) return { timezone: options.timezone, locale: options.locale };
proxyUrl = ensureProxyScheme(proxyUrl);
// When both tz/locale are explicit, still resolve exit IP for WebRTC
if (options.timezone && options.locale) {
@@ -304,13 +354,13 @@ export async function resolveWebrtcArgs(
const idx = args.findIndex(a => a === "--fingerprint-webrtc-ip=auto");
if (idx === -1) return args;
let proxyUrl = typeof options.proxy === "string" ? options.proxy : options.proxy?.server;
const proxyUrl = extractProxyUrl(options.proxy);
if (!proxyUrl) {
console.warn("[cloakbrowser] --fingerprint-webrtc-ip=auto requires a proxy; removing flag");
const result = [...args];
result.splice(idx, 1);
return result;
}
proxyUrl = ensureProxyScheme(proxyUrl);
try {
const ip = await resolveExitIp(proxyUrl);
@@ -318,10 +368,12 @@ export async function resolveWebrtcArgs(
if (ip) {
result[idx] = `--fingerprint-webrtc-ip=${ip}`;
} else {
console.warn("[cloakbrowser] Could not resolve proxy exit IP for WebRTC spoofing; removing --fingerprint-webrtc-ip=auto");
result.splice(idx, 1);
}
return result;
} catch {
console.warn("[cloakbrowser] Failed to resolve proxy exit IP for WebRTC spoofing; removing --fingerprint-webrtc-ip=auto");
const result = [...args];
result.splice(idx, 1);
return result;
+366
View File
@@ -0,0 +1,366 @@
/**
* ElementHandle humanization for Playwright.
*
* Mirrors Puppeteer's ElementHandle patching architecture.
* Patches page.$(), page.$$(), page.waitForSelector() to return humanized handles,
* and patches all interaction methods on each ElementHandle instance.
*
* Playwright ElementHandle methods patched:
* click, dblclick, hover, type, fill, press, selectOption,
* check, uncheck, setChecked, tap, focus
* + $, $$, waitForSelector (nested elements are also patched)
*
* Stealth-aware:
* - Uses CDP DOM.describeNode when available to check element type
* (no main-world JS execution)
* - Falls back to el.evaluate() only when CDP is unavailable
*/
import type { Page, Frame, ElementHandle, CDPSession } from 'playwright-core';
import type { HumanConfig } from './config.js';
import { rand, randRange, sleep } from './config.js';
import { RawMouse, RawKeyboard, humanMove, humanClick, clickTarget, humanIdle } from './mouse.js';
import { humanType } from './keyboard.js';
// --- Platform-aware select-all shortcut ---
const SELECT_ALL = process.platform === 'darwin' ? 'Meta+a' : 'Control+a';
// ============================================================================
// Stealth ElementHandle input check — uses CDP DOM.describeNode
// ============================================================================
async function isInputElementHandle(
stealth: any, // StealthEval from index.ts
el: ElementHandle,
): Promise<boolean> {
// Try CDP DOM.describeNode first (no main-world JS execution)
if (stealth) {
try {
const cdp: CDPSession = await stealth.getCdpSession();
// Playwright exposes the JSHandle's internal preview via _objectId or similar
// We need the remote object ID. Try to get it via internal API.
const impl = (el as any)._impl ?? (el as any)._object ?? el;
const guid = (impl as any)._guid;
// Use el.evaluate as a reliable fallback within stealth context
// Playwright doesn't expose remoteObject directly like Puppeteer
} catch { /* fallthrough */ }
}
// Fallback: el.evaluate (works reliably in Playwright)
try {
return await el.evaluate((node: any) => {
const tag = node.tagName?.toLowerCase();
return tag === 'input' || tag === 'textarea'
|| node.getAttribute?.('contenteditable') === 'true';
});
} catch {
return false;
}
}
// ============================================================================
// CursorState type (matches index.ts)
// ============================================================================
interface CursorState {
x: number;
y: number;
initialized: boolean;
}
// ============================================================================
// Patch a single Playwright ElementHandle
// ============================================================================
export function patchSingleElementHandle(
el: ElementHandle,
page: Page,
cfg: HumanConfig,
cursor: CursorState,
raw: RawMouse,
rawKb: RawKeyboard,
originals: any,
stealth: any,
): void {
if ((el as any)._humanPatched) return;
(el as any)._humanPatched = true;
// Save originals
const origElClick = el.click.bind(el);
const origElDblclick = el.dblclick.bind(el);
const origElHover = el.hover.bind(el);
const origElType = el.type.bind(el);
const origElFill = el.fill.bind(el);
const origElPress = el.press.bind(el);
const origElSelectOption = el.selectOption.bind(el);
const origElCheck = el.check.bind(el);
const origElUncheck = el.uncheck.bind(el);
const origElSetChecked = (el as any).setChecked?.bind(el);
const origElTap = el.tap.bind(el);
const origElFocus = el.focus.bind(el);
// Nested selectors
const origEl$ = el.$.bind(el);
const origEl$$ = el.$$.bind(el);
const origElWaitForSelector = el.waitForSelector.bind(el);
// --- Nested elements are also patched ---
(el as any).$ = async (selector: string) => {
const child = await origEl$(selector);
if (child) patchSingleElementHandle(child, page, cfg, cursor, raw, rawKb, originals, stealth);
return child;
};
(el as any).$$ = async (selector: string) => {
const children = await origEl$$(selector);
for (const child of children) {
patchSingleElementHandle(child, page, cfg, cursor, raw, rawKb, originals, stealth);
}
return children;
};
(el as any).waitForSelector = async (selector: string, options?: any) => {
const child = await origElWaitForSelector(selector, options);
if (child) patchSingleElementHandle(child, page, cfg, cursor, raw, rawKb, originals, stealth);
return child;
};
// --- Helper: get bounding box and move cursor to element ---
const moveToElement = async () => {
// Ensure cursor is initialized
const ensureCursorInit = (page as any)._ensureCursorInit;
if (ensureCursorInit) await ensureCursorInit();
const box = await el.boundingBox();
if (!box) return null;
const isInp = await isInputElementHandle(stealth, el);
const target = clickTarget(box, isInp, cfg);
if (cfg.idle_between_actions) {
await humanIdle(raw, rand(cfg.idle_between_duration[0], cfg.idle_between_duration[1]), cursor.x, cursor.y, cfg);
}
await humanMove(raw, cursor.x, cursor.y, target.x, target.y, cfg);
cursor.x = target.x;
cursor.y = target.y;
return { box, isInp };
};
// --- el.click() ---
(el as any).click = async (options?: any) => {
const info = await moveToElement();
if (!info) return origElClick(options);
await humanClick(raw, info.isInp, cfg);
};
// --- el.dblclick() ---
(el as any).dblclick = async (options?: any) => {
const info = await moveToElement();
if (!info) return origElDblclick(options);
await raw.down({ clickCount: 2 });
await sleep(rand(30, 60));
await raw.up({ clickCount: 2 });
};
// --- el.hover() ---
(el as any).hover = async (options?: any) => {
const info = await moveToElement();
if (!info) return origElHover(options);
// Just move — no click
};
// --- el.type() ---
(el as any).type = async (text: string, options?: any) => {
const info = await moveToElement();
if (!info) return origElType(text, options);
await humanClick(raw, info.isInp, cfg);
await sleep(rand(100, 250));
let cdpSession: CDPSession | null = null;
try { cdpSession = await stealth?.getCdpSession(); } catch {}
await humanType(page, rawKb, text, cfg, cdpSession);
};
// --- el.fill() ---
(el as any).fill = async (value: string, options?: any) => {
const info = await moveToElement();
if (!info) return origElFill(value, options);
await humanClick(raw, info.isInp, cfg);
await sleep(rand(100, 250));
// Clear existing content
await originals.keyboardPress(SELECT_ALL);
await sleep(rand(30, 80));
await originals.keyboardPress('Backspace');
await sleep(rand(50, 150));
let cdpSession: CDPSession | null = null;
try { cdpSession = await stealth?.getCdpSession(); } catch {}
await humanType(page, rawKb, value, cfg, cdpSession);
};
// --- el.press() ---
(el as any).press = async (key: string, options?: any) => {
await sleep(rand(20, 60));
await originals.keyboardDown(key);
await sleep(randRange(cfg.key_hold));
await originals.keyboardUp(key);
};
// --- el.selectOption() ---
(el as any).selectOption = async (values: any, options?: any) => {
const info = await moveToElement();
if (!info) return origElSelectOption(values, options);
await humanClick(raw, false, cfg);
await sleep(rand(100, 300));
return origElSelectOption(values, options);
};
// --- el.check() ---
(el as any).check = async (options?: any) => {
try {
const checked = await el.isChecked();
if (checked) return; // Already checked
} catch {}
const info = await moveToElement();
if (!info) return origElCheck(options);
await humanClick(raw, info.isInp, cfg);
};
// --- el.uncheck() ---
(el as any).uncheck = async (options?: any) => {
try {
const checked = await el.isChecked();
if (!checked) return; // Already unchecked
} catch {}
const info = await moveToElement();
if (!info) return origElUncheck(options);
await humanClick(raw, info.isInp, cfg);
};
// --- el.setChecked() ---
if (origElSetChecked) {
(el as any).setChecked = async (checked: boolean, options?: any) => {
try {
const current = await el.isChecked();
if (current === checked) return;
} catch {}
const info = await moveToElement();
if (!info) return origElSetChecked(checked, options);
await humanClick(raw, info.isInp, cfg);
};
}
// --- el.tap() ---
(el as any).tap = async (options?: any) => {
const info = await moveToElement();
if (!info) return origElTap(options);
await humanClick(raw, info.isInp, cfg);
};
// --- el.focus() ---
// Move cursor humanly but use programmatic focus (no click side-effects).
// Stock Playwright el.focus() never clicks — clicking would trigger onclick,
// submit forms, navigate links, etc.
(el as any).focus = async () => {
await moveToElement(); // human-like Bézier cursor movement
await origElFocus(); // programmatic focus, no click
};
}
// ============================================================================
// Page-level ElementHandle patching
// ============================================================================
export function patchPageElementHandles(
page: Page,
cfg: HumanConfig,
cursor: CursorState,
raw: RawMouse,
rawKb: RawKeyboard,
originals: any,
stealth: any,
): void {
// Patch page.$() — only if the method exists
if (typeof page.$ === 'function') {
const orig$ = page.$.bind(page);
(page as any).$ = async (selector: string) => {
const el = await orig$(selector);
if (el) patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
return el;
};
}
// Patch page.$$()
if (typeof page.$$ === 'function') {
const orig$$ = page.$$.bind(page);
(page as any).$$ = async (selector: string) => {
const els = await orig$$(selector);
for (const el of els) {
patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
}
return els;
};
}
// Patch page.waitForSelector()
if (typeof page.waitForSelector === 'function') {
const origWaitForSelector = page.waitForSelector.bind(page);
(page as any).waitForSelector = async (selector: string, options?: any) => {
const el = await origWaitForSelector(selector, options);
if (el) patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
return el;
};
}
}
// ============================================================================
// Frame-level ElementHandle patching
// ============================================================================
export function patchFrameElementHandles(
frame: Frame,
page: Page,
cfg: HumanConfig,
cursor: CursorState,
raw: RawMouse,
rawKb: RawKeyboard,
originals: any,
stealth: any,
): void {
// Patch frame.$() — only if the method exists
if (typeof frame.$ === 'function') {
const origFrame$ = frame.$.bind(frame);
(frame as any).$ = async (selector: string) => {
const el = await origFrame$(selector);
if (el) patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
return el;
};
}
// Patch frame.$$()
if (typeof frame.$$ === 'function') {
const origFrame$$ = frame.$$.bind(frame);
(frame as any).$$ = async (selector: string) => {
const els = await origFrame$$(selector);
for (const el of els) {
patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
}
return els;
};
}
// Patch frame.waitForSelector()
if (typeof frame.waitForSelector === 'function') {
const origFrameWaitForSelector = frame.waitForSelector.bind(frame);
(frame as any).waitForSelector = async (selector: string, options?: any) => {
const el = await origFrameWaitForSelector(selector, options);
if (el) patchSingleElementHandle(el, page, cfg, cursor, raw, rawKb, originals, stealth);
return el;
};
}
}
+15
View File
@@ -12,6 +12,14 @@
* Patches all interaction methods:
* click, dblclick, hover, type, fill, check, uncheck, selectOption,
* press, pressSequentially, tap, dragTo, clear + Frame-level equivalents.
*
* ELEMENTHANDLE-LEVEL:
* click, dblclick, hover, type, fill, press, selectOption,
* check, uncheck, setChecked, tap, focus
* + $, $$, waitForSelector (nested elements are also patched)
*
* page.$(), page.$$(), page.waitForSelector() and Frame equivalents
* return patched ElementHandles automatically.
*/
import type { Browser, BrowserContext, Page, Frame, CDPSession } from 'playwright-core';
@@ -19,11 +27,13 @@ import { HumanConfig, resolveConfig, rand, randRange, sleep } from './config.js'
import { RawMouse, RawKeyboard, humanMove, humanClick, clickTarget, humanIdle } from './mouse.js';
import { humanType } from './keyboard.js';
import { scrollToElement } from './scroll.js';
import { patchPageElementHandles, patchFrameElementHandles, patchSingleElementHandle } from './elementhandle.js';
export { HumanConfig, resolveConfig } from './config.js';
export { humanMove, humanClick, clickTarget, humanIdle } from './mouse.js';
export { humanType } from './keyboard.js';
export { scrollToElement } from './scroll.js';
export { patchSingleElementHandle } from './elementhandle.js';
// --- Platform-aware select-all shortcut (macOS uses Meta, others use Control) ---
const SELECT_ALL = process.platform === 'darwin' ? 'Meta+a' : 'Control+a';
@@ -488,6 +498,9 @@ function patchPage(page: Page, cfg: HumanConfig, cursor: CursorState): void {
// --- Patch Frame-level methods (for sub-frames) ---
patchFrames(page, cfg, cursor, raw, rawKb, originals, stealth);
// --- Patch ElementHandle selectors (page.$, page.$$, page.waitForSelector) ---
patchPageElementHandles(page, cfg, cursor, raw, rawKb, originals, stealth);
}
@@ -511,6 +524,8 @@ function patchFrames(
): void {
for (const frame of iterFrames(page)) {
patchSingleFrame(frame, page, cfg, originals, stealth);
// Patch frame-level ElementHandle selectors ($, $$, waitForSelector)
patchFrameElementHandles(frame, page, cfg, cursor, raw, rawKb, originals, stealth);
}
}
+7 -9
View File
@@ -8,7 +8,7 @@ import type { LaunchOptions, LaunchContextOptions, LaunchPersistentContextOption
import { DEFAULT_VIEWPORT, IGNORE_DEFAULT_ARGS } from "./config.js";
import { buildArgs } from "./args.js";
import { ensureBinary } from "./download.js";
import { parseProxyUrl } from "./proxy.js";
import { resolveProxyConfig } from "./proxy.js";
import { maybeResolveGeoip, resolveWebrtcArgs } from "./geoip.js";
/** @internal Accept both timezone and timezoneId — either works, no warning. Exported for testing. */
@@ -39,20 +39,19 @@ export async function launch(options: LaunchOptions = {}): Promise<Browser> {
const binaryPath = process.env.CLOAKBROWSER_BINARY_PATH || (await ensureBinary());
const { exitIp, ...resolved } = await maybeResolveGeoip(options);
const { proxyOption, proxyArgs } = resolveProxyConfig(options.proxy);
let resolvedArgs = await resolveWebrtcArgs(options);
if (exitIp && !(resolvedArgs ?? []).some(a => a.startsWith("--fingerprint-webrtc-ip"))) {
resolvedArgs = [...(resolvedArgs ?? []), `--fingerprint-webrtc-ip=${exitIp}`];
}
const args = buildArgs({ ...options, ...resolved, args: resolvedArgs });
const args = buildArgs({ ...options, ...resolved, args: [...(resolvedArgs ?? []), ...proxyArgs] });
const browser = await chromium.launch({
executablePath: binaryPath,
headless: options.headless ?? true,
args,
ignoreDefaultArgs: IGNORE_DEFAULT_ARGS,
...(options.proxy
? { proxy: typeof options.proxy === "string" ? parseProxyUrl(options.proxy) : options.proxy }
: {}),
...(proxyOption ? { proxy: proxyOption } : {}),
...options.launchOptions,
});
@@ -164,11 +163,12 @@ export async function launchPersistentContext(
const binaryPath = process.env.CLOAKBROWSER_BINARY_PATH || (await ensureBinary());
const { exitIp, ...resolved } = await maybeResolveGeoip(options);
const { proxyOption, proxyArgs } = resolveProxyConfig(options.proxy);
let resolvedArgs = await resolveWebrtcArgs(options);
if (exitIp && !(resolvedArgs ?? []).some(a => a.startsWith("--fingerprint-webrtc-ip"))) {
resolvedArgs = [...(resolvedArgs ?? []), `--fingerprint-webrtc-ip=${exitIp}`];
}
const args = buildArgs({ ...options, ...resolved, args: resolvedArgs });
const args = buildArgs({ ...options, ...resolved, args: [...(resolvedArgs ?? []), ...proxyArgs] });
// locale and timezone are set via binary flags (--lang, --fingerprint-timezone)
// — NOT via Playwright context kwargs which use detectable CDP emulation.
@@ -177,9 +177,7 @@ export async function launchPersistentContext(
headless: options.headless ?? true,
args,
ignoreDefaultArgs: IGNORE_DEFAULT_ARGS,
...(options.proxy
? { proxy: typeof options.proxy === "string" ? parseProxyUrl(options.proxy) : options.proxy }
: {}),
...(proxyOption ? { proxy: proxyOption } : {}),
...(options.userAgent ? { userAgent: options.userAgent } : {}),
viewport: options.viewport === undefined ? DEFAULT_VIEWPORT : options.viewport,
...(options.colorScheme ? { colorScheme: options.colorScheme } : {}),
+59
View File
@@ -23,6 +23,65 @@ export function ensureProxyScheme(proxyUrl: string): string {
* Also handles: no credentials, URL-encoded special chars, socks5://, missing port,
* and bare proxy strings without a scheme (e.g. "user:pass@host:port" -> treated as http).
*/
/** Proxy dict shape accepted by Playwright/Puppeteer wrappers. */
export type ProxyDict = { server: string; bypass?: string; username?: string; password?: string };
/** Result of resolveProxyConfig — either Playwright dict OR Chrome arg, never both. */
export interface ProxyConfig {
/** Playwright proxy option (for HTTP proxies). */
proxyOption?: ParsedProxy;
/** Chrome CLI args (for SOCKS5 proxies, e.g. ["--proxy-server=socks5://..."]). */
proxyArgs: string[];
}
/**
* Check if a proxy uses the SOCKS5 protocol.
*/
export function isSocksProxy(proxy: string | ProxyDict | undefined | null): boolean {
if (!proxy) return false;
const url = typeof proxy === "string" ? proxy : proxy.server;
return /^socks5h?:\/\//i.test(url);
}
/**
* Reconstruct a SOCKS5 URL with inline credentials from a proxy dict.
*/
export function reconstructSocksUrl(proxy: ProxyDict): string {
const url = new URL(proxy.server);
if (proxy.username) {
url.username = encodeURIComponent(proxy.username);
if (proxy.password) url.password = encodeURIComponent(proxy.password);
}
return url.href.replace(/\/$/, "");
}
/**
* Resolve proxy into Playwright option and/or Chrome args.
*
* Playwright rejects SOCKS5 proxies with credentials in its proxy dict,
* so SOCKS5 is passed via --proxy-server Chrome arg instead.
*/
export function resolveProxyConfig(proxy: string | ProxyDict | undefined): ProxyConfig {
if (!proxy) return { proxyArgs: [] };
if (isSocksProxy(proxy)) {
// SOCKS5: bypass Playwright, pass directly to Chrome via --proxy-server.
if (typeof proxy === "string") {
return { proxyArgs: [`--proxy-server=${proxy}`] };
}
const socksUrl = reconstructSocksUrl(proxy);
const args = [`--proxy-server=${socksUrl}`];
if (proxy.bypass) args.push(`--proxy-bypass-list=${proxy.bypass}`);
return { proxyArgs: args };
}
// HTTP/HTTPS: use Playwright's proxy dict
if (typeof proxy === "string") {
return { proxyOption: parseProxyUrl(proxy), proxyArgs: [] };
}
return { proxyOption: proxy as ParsedProxy, proxyArgs: [] };
}
export function parseProxyUrl(proxy: string): ParsedProxy {
let url: URL;
// Bare format: "user:pass@host:port" — new URL() throws without a scheme.
+9 -7
View File
@@ -9,7 +9,7 @@ import type { LaunchOptions } from "./types.js";
import { IGNORE_DEFAULT_ARGS } from "./config.js";
import { buildArgs } from "./args.js";
import { ensureBinary } from "./download.js";
import { parseProxyUrl } from "./proxy.js";
import { isSocksProxy, parseProxyUrl, resolveProxyConfig } from "./proxy.js";
import { maybeResolveGeoip, resolveWebrtcArgs } from "./geoip.js";
/**
@@ -39,25 +39,27 @@ export async function launch(options: LaunchOptions = {}): Promise<Browser> {
const args = buildArgs({ ...options, ...resolved, args: resolvedArgs });
// Puppeteer handles proxy via CLI args, not a separate option.
// Chromium's --proxy-server does NOT support inline credentials,
// so we strip them and use page.authenticate() instead.
// SOCKS5: Chrome supports inline credentials natively (RFC 1929 auth).
// HTTP: Chrome does NOT support inline credentials — strip them and
// use page.authenticate() for Proxy-Authorization headers instead.
let proxyAuth: { username: string; password: string } | undefined;
if (options.proxy) {
if (typeof options.proxy === "string") {
if (isSocksProxy(options.proxy)) {
// SOCKS5: pass full URL with credentials to Chrome directly
const { proxyArgs } = resolveProxyConfig(options.proxy);
args.push(...proxyArgs);
} else if (typeof options.proxy === "string") {
const { server, username, password } = parseProxyUrl(options.proxy);
args.push(`--proxy-server=${server}`);
if (username) {
proxyAuth = { username, password: password ?? "" };
}
} else {
// Strip any inline credentials from the server URL — Chromium's
// --proxy-server doesn't support them; use page.authenticate() instead.
const parsed = parseProxyUrl(options.proxy.server);
args.push(`--proxy-server=${parsed.server}`);
if (options.proxy.bypass) {
args.push(`--proxy-bypass-list=${options.proxy.bypass}`);
}
// Explicit username/password fields take precedence over inline creds
const username = options.proxy.username ?? parsed.username;
const password = options.proxy.password ?? parsed.password;
if (username) {
+344
View File
@@ -1,6 +1,7 @@
import { describe, it, expect, vi } from "vitest";
import { resolveConfig, rand, randRange, sleep } from "../src/human/config.js";
import { humanMove, humanClick, clickTarget, humanIdle } from "../src/human/mouse.js";
import { patchPageElementHandles } from "../src/human/elementhandle.js";
// =========================================================================
// Config resolution
@@ -689,6 +690,349 @@ describe("humanType non-ASCII", () => {
// =========================================================================
// ElementHandle patching (Playwright)
// =========================================================================
function buildMockElementHandle(overrides: Record<string, any> = {}): any {
const el: any = {
click: vi.fn(async () => {}),
dblclick: vi.fn(async () => {}),
hover: vi.fn(async () => {}),
type: vi.fn(async () => {}),
fill: vi.fn(async () => {}),
press: vi.fn(async () => {}),
selectOption: vi.fn(async () => {}),
check: vi.fn(async () => {}),
uncheck: vi.fn(async () => {}),
setChecked: vi.fn(async () => {}),
tap: vi.fn(async () => {}),
focus: vi.fn(async () => {}),
boundingBox: overrides.boundingBox ?? vi.fn(async () => ({ x: 100, y: 100, width: 200, height: 30 })),
evaluate: overrides.evaluate ?? vi.fn(async () => false),
isChecked: overrides.isChecked ?? vi.fn(async () => false),
$: vi.fn(async () => null),
$$: vi.fn(async () => []),
waitForSelector: vi.fn(async () => null),
_humanPatched: false,
};
return el;
}
describe("patchSingleElementHandle", () => {
it("marks element as patched", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 100, y: 100, initialized: true };
const raw = {
move: vi.fn(async () => {}),
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
wheel: vi.fn(async () => {}),
};
const rawKb = {
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
type: vi.fn(async () => {}),
insertText: vi.fn(async () => {}),
};
const originals = {
keyboardPress: vi.fn(async () => {}),
keyboardDown: vi.fn(async () => {}),
keyboardUp: vi.fn(async () => {}),
};
const el = buildMockElementHandle();
const page = buildMockPage();
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
expect(el._humanPatched).toBe(true);
});
it("el.click calls mouse.move and mouse.down/up (humanized path)", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default", { idle_between_actions: false });
const cursor = { x: 50, y: 50, initialized: true };
let moveCount = 0;
let downCalled = false;
let upCalled = false;
const raw = {
move: vi.fn(async () => { moveCount++; }),
down: vi.fn(async () => { downCalled = true; }),
up: vi.fn(async () => { upCalled = true; }),
wheel: vi.fn(async () => {}),
};
const rawKb = {
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
type: vi.fn(async () => {}),
insertText: vi.fn(async () => {}),
};
const originals = {
keyboardPress: vi.fn(async () => {}),
keyboardDown: vi.fn(async () => {}),
keyboardUp: vi.fn(async () => {}),
};
const el = buildMockElementHandle();
const page = buildMockPage();
(page as any)._ensureCursorInit = vi.fn(async () => {});
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
await el.click();
expect(moveCount).toBeGreaterThan(0);
expect(downCalled).toBe(true);
expect(upCalled).toBe(true);
}, 30000);
it("el.hover calls mouse.move but NOT down/up", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default", { idle_between_actions: false });
const cursor = { x: 50, y: 50, initialized: true };
let downCalled = false;
const raw = {
move: vi.fn(async () => {}),
down: vi.fn(async () => { downCalled = true; }),
up: vi.fn(async () => {}),
wheel: vi.fn(async () => {}),
};
const rawKb = {
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
type: vi.fn(async () => {}),
insertText: vi.fn(async () => {}),
};
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el = buildMockElementHandle();
const page = buildMockPage();
(page as any)._ensureCursorInit = vi.fn(async () => {});
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
await el.hover();
expect(raw.move).toHaveBeenCalled();
expect(downCalled).toBe(false);
}, 30000);
it("el.type triggers mouse move + click + keyboard events", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default", { idle_between_actions: false, mistype_chance: 0 });
const cursor = { x: 50, y: 50, initialized: true };
const raw = {
move: vi.fn(async () => {}),
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
wheel: vi.fn(async () => {}),
};
const rawKb = {
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
type: vi.fn(async () => {}),
insertText: vi.fn(async () => {}),
};
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el = buildMockElementHandle({ evaluate: vi.fn(async () => true) }); // isInput = true
const page = buildMockPage();
(page as any)._ensureCursorInit = vi.fn(async () => {});
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
await el.type("abc");
expect(raw.move).toHaveBeenCalled();
expect(raw.down).toHaveBeenCalled(); // click to focus
expect(rawKb.down).toHaveBeenCalled(); // keyboard typing
}, 30000);
it("el.fill calls selectAll + backspace + type", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default", { idle_between_actions: false, mistype_chance: 0 });
const cursor = { x: 50, y: 50, initialized: true };
const pressedKeys: string[] = [];
const raw = {
move: vi.fn(async () => {}),
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
wheel: vi.fn(async () => {}),
};
const rawKb = {
down: vi.fn(async () => {}),
up: vi.fn(async () => {}),
type: vi.fn(async () => {}),
insertText: vi.fn(async () => {}),
};
const originals = {
keyboardPress: vi.fn(async (key: string) => { pressedKeys.push(key); }),
keyboardDown: vi.fn(async () => {}),
keyboardUp: vi.fn(async () => {}),
};
const el = buildMockElementHandle({ evaluate: vi.fn(async () => true) });
const page = buildMockPage();
(page as any)._ensureCursorInit = vi.fn(async () => {});
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
await el.fill("newtext");
const expected = process.platform === "darwin" ? "Meta+a" : "Control+a";
expect(pressedKeys).toContain(expected);
expect(pressedKeys).toContain("Backspace");
}, 30000);
it("no double patching", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el = buildMockElementHandle();
const page = buildMockPage();
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
const firstClick = el.click;
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
expect(el.click).toBe(firstClick);
});
it("nested $() returns patched child handle", async () => {
const { patchSingleElementHandle } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const child = buildMockElementHandle();
const el = buildMockElementHandle();
el.$ = vi.fn(async () => child);
const page = buildMockPage();
patchSingleElementHandle(el, page as any, cfg, cursor as any, raw, rawKb, originals, null);
const result = await el.$("span");
expect(result._humanPatched).toBe(true);
});
});
describe("patchPageElementHandles", () => {
it("page.$() returns patched ElementHandle", async () => {
const { patchPageElementHandles } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el = buildMockElementHandle();
const page = buildMockPage();
(page as any).$ = vi.fn(async () => el);
(page as any).$$ = vi.fn(async () => [el]);
(page as any).waitForSelector = vi.fn(async () => el);
patchPageElementHandles(page as any, cfg, cursor as any, raw, rawKb, originals, null);
const result = await (page as any).$("#test");
expect(result._humanPatched).toBe(true);
});
it("page.$$() returns all patched handles", async () => {
const { patchPageElementHandles } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el1 = buildMockElementHandle();
const el2 = buildMockElementHandle();
const page = buildMockPage();
(page as any).$ = vi.fn(async () => null);
(page as any).$$ = vi.fn(async () => [el1, el2]);
(page as any).waitForSelector = vi.fn(async () => null);
patchPageElementHandles(page as any, cfg, cursor as any, raw, rawKb, originals, null);
const results = await (page as any).$$("div");
expect(results[0]._humanPatched).toBe(true);
expect(results[1]._humanPatched).toBe(true);
});
it("page.waitForSelector() returns patched handle", async () => {
const { patchPageElementHandles } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const el = buildMockElementHandle();
const page = buildMockPage();
(page as any).$ = vi.fn(async () => null);
(page as any).$$ = vi.fn(async () => []);
(page as any).waitForSelector = vi.fn(async () => el);
patchPageElementHandles(page as any, cfg, cursor as any, raw, rawKb, originals, null);
const result = await (page as any).waitForSelector("#test");
expect(result._humanPatched).toBe(true);
});
it("page.$() returns null when no element found (no crash)", async () => {
const { patchPageElementHandles } = await import("../src/human/elementhandle.js");
const cfg = resolveConfig("default");
const cursor = { x: 0, y: 0, initialized: false };
const raw = { move: vi.fn(async () => {}), down: vi.fn(async () => {}), up: vi.fn(async () => {}), wheel: vi.fn(async () => {}) };
const rawKb = { down: vi.fn(async () => {}), up: vi.fn(async () => {}), type: vi.fn(async () => {}), insertText: vi.fn(async () => {}) };
const originals = { keyboardPress: vi.fn(async () => {}), keyboardDown: vi.fn(async () => {}), keyboardUp: vi.fn(async () => {}) };
const page = buildMockPage();
(page as any).$ = vi.fn(async () => null);
(page as any).$$ = vi.fn(async () => []);
(page as any).waitForSelector = vi.fn(async () => null);
patchPageElementHandles(page as any, cfg, cursor as any, raw, rawKb, originals, null);
const result = await (page as any).$("#nonexistent");
expect(result).toBeNull();
});
});
describe("patchPage integrates ElementHandle patching", () => {
it("patchPage patches page.$ automatically", async () => {
const { patchPage } = await import("../src/human/index.js");
const el = buildMockElementHandle();
const page = buildMockPage();
(page as any).$ = vi.fn(async () => el);
(page as any).$$ = vi.fn(async () => []);
(page as any).waitForSelector = vi.fn(async () => null);
const cfg = resolveConfig("default");
const cursor = { x: 100, y: 100, initialized: true };
patchPage(page as any, cfg, cursor as any);
const result = await (page as any).$("#test");
expect(result._humanPatched).toBe(true);
});
});
function buildMockFrame(): any {
return {
click: vi.fn(async () => {}),
+90 -1
View File
@@ -1,5 +1,5 @@
import { describe, it, expect } from "vitest";
import { parseProxyUrl } from "../src/proxy.js";
import { parseProxyUrl, isSocksProxy, resolveProxyConfig } from "../src/proxy.js";
import type { LaunchOptions } from "../src/types.js";
describe("parseProxyUrl", () => {
@@ -115,3 +115,92 @@ describe("bare proxy format (user:pass@host:port)", () => {
expect(parseProxyUrl("proxy:8080")).toEqual({ server: "proxy:8080" });
});
});
describe("isSocksProxy", () => {
it("detects socks5 string", () => {
expect(isSocksProxy("socks5://user:pass@host:1080")).toBe(true);
});
it("detects socks5h string", () => {
expect(isSocksProxy("socks5h://host:1080")).toBe(true);
});
it("case insensitive", () => {
expect(isSocksProxy("SOCKS5://host:1080")).toBe(true);
});
it("rejects http", () => {
expect(isSocksProxy("http://host:8080")).toBe(false);
});
it("detects socks5 dict", () => {
expect(isSocksProxy({ server: "socks5://host:1080" })).toBe(true);
});
it("rejects http dict", () => {
expect(isSocksProxy({ server: "http://host:8080" })).toBe(false);
});
it("returns false for undefined", () => {
expect(isSocksProxy(undefined)).toBe(false);
});
});
describe("resolveProxyConfig", () => {
it("returns empty for undefined", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig(undefined);
expect(proxyOption).toBeUndefined();
expect(proxyArgs).toEqual([]);
});
it("returns playwright dict for http string", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig("http://user:pass@proxy:8080");
expect(proxyOption).toEqual({ server: "http://proxy:8080", username: "user", password: "pass" });
expect(proxyArgs).toEqual([]);
});
it("returns playwright dict for http dict", () => {
const proxy = { server: "http://proxy:8080", bypass: ".example.com" };
const { proxyOption, proxyArgs } = resolveProxyConfig(proxy);
expect(proxyOption).toEqual(proxy);
expect(proxyArgs).toEqual([]);
});
it("returns chrome arg for socks5 string", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig("socks5://user:pass@host:1080");
expect(proxyOption).toBeUndefined();
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:pass@host:1080"]);
});
it("returns chrome arg for socks5 no auth", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig("socks5://host:1080");
expect(proxyOption).toBeUndefined();
expect(proxyArgs).toEqual(["--proxy-server=socks5://host:1080"]);
});
it("returns chrome arg for socks5h string", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig("socks5h://user:pass@host:1080");
expect(proxyOption).toBeUndefined();
expect(proxyArgs).toEqual(["--proxy-server=socks5h://user:pass@host:1080"]);
});
it("reconstructs URL from socks5 dict with auth", () => {
const { proxyOption, proxyArgs } = resolveProxyConfig({
server: "socks5://host:1080",
username: "user",
password: "p@ss",
});
expect(proxyOption).toBeUndefined();
expect(proxyArgs.length).toBe(1);
expect(proxyArgs[0]).toContain("--proxy-server=socks5://user:p%40ss@host:1080");
});
it("includes bypass for socks5 dict", () => {
const { proxyArgs } = resolveProxyConfig({
server: "socks5://host:1080",
bypass: ".example.com",
});
expect(proxyArgs).toContain("--proxy-server=socks5://host:1080");
expect(proxyArgs).toContain("--proxy-bypass-list=.example.com");
});
});
+25
View File
@@ -113,4 +113,29 @@ describe("puppeteer launch", () => {
expect(callArgs.args).toContain("--disable-gpu");
expect(callArgs.args).toContain("--no-first-run");
});
it("keeps SOCKS5 credentials in --proxy-server URL", async () => {
const { launch } = await import("../src/puppeteer.js");
const browser = await launch({ proxy: "socks5://user:pass@proxy:1080" });
const callArgs = vi.mocked(puppeteerMock.default.launch).mock.calls[0][0];
expect(callArgs.args).toContain("--proxy-server=socks5://user:pass@proxy:1080");
// Should NOT set up page.authenticate for SOCKS5
const page = await browser.newPage();
expect(page.authenticate).not.toHaveBeenCalled();
});
it("reconstructs SOCKS5 dict with auth into --proxy-server URL", async () => {
const { launch } = await import("../src/puppeteer.js");
const browser = await launch({
proxy: { server: "socks5://proxy:1080", username: "user", password: "p@ss" },
});
const callArgs = vi.mocked(puppeteerMock.default.launch).mock.calls[0][0];
expect(callArgs.args).toContain("--proxy-server=socks5://user:p%40ss@proxy:1080");
const page = await browser.newPage();
expect(page.authenticate).not.toHaveBeenCalled();
});
});
+1 -1
View File
@@ -54,7 +54,7 @@ dependencies = [
]
[project.optional-dependencies]
geoip = ["geoip2>=4.0"]
geoip = ["geoip2>=4.0", "socksio>=1.0"] # socksio: SOCKS5 transport for httpx
patchright = ["patchright>=1.40"]
serve = ["aiohttp>=3.9", "websockets>=12.0"]
dev = ["pytest>=7.0", "pytest-asyncio>=0.23"]
+4 -2
View File
@@ -105,8 +105,10 @@ class TestParseCliArgs:
def test_passthrough_args(self):
args = ["--no-sandbox", "--disable-gpu", "--fingerprint=999"]
_, passthrough = parse_cli_args(args)
assert passthrough == args
config, passthrough = parse_cli_args(args)
# --fingerprint=999 is consumed into config["default_seed"], not passed through
assert passthrough == ["--no-sandbox", "--disable-gpu"]
assert config["default_seed"] == "999"
def test_port_not_in_passthrough(self):
_, passthrough = parse_cli_args(["--port=9222", "--no-sandbox"])
+61
View File
@@ -279,6 +279,67 @@ if __name__ == "__main__":
check("keyboard.type", kb_ms > 500, f"{kb_ms} ms")
time.sleep(1)
# ============================================================
# SCENARIO 7: ElementHandle — query_selector interactions
# ============================================================
step("ElementHandle — query_selector click, type, fill, hover")
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(2)
inject(page)
time.sleep(1)
print(" Watch: get element via query_selector, cursor moves smoothly")
el = page.query_selector('#searchInput')
assert el is not None, "query_selector returned None"
assert getattr(el, '_human_patched', False), "ElementHandle not patched!"
t0 = time.time()
el.click()
eh_click_ms = int((time.time() - t0) * 1000)
check("ElementHandle click", eh_click_ms > 100, f"{eh_click_ms} ms")
time.sleep(0.5)
print(" Watch: ElementHandle type — characters appear one by one")
t0 = time.time()
el.type('ElementHandle typing')
eh_type_ms = int((time.time() - t0) * 1000)
val = page.locator('#searchInput').input_value()
check("ElementHandle type", val == 'ElementHandle typing' and eh_type_ms > 1500, f"{eh_type_ms} ms, value='{val}'")
time.sleep(0.5)
print(" Watch: ElementHandle fill — clears then types")
t0 = time.time()
el.fill('Filled via EH')
eh_fill_ms = int((time.time() - t0) * 1000)
val = page.locator('#searchInput').input_value()
check("ElementHandle fill", val == 'Filled via EH' and eh_fill_ms > 1000, f"{eh_fill_ms} ms, value='{val}'")
time.sleep(0.5)
print(" Watch: ElementHandle hover — cursor moves without clicking")
btn_el = page.query_selector('button[type="submit"]')
t0 = time.time()
btn_el.hover()
eh_hover_ms = int((time.time() - t0) * 1000)
check("ElementHandle hover", eh_hover_ms > 50, f"{eh_hover_ms} ms")
time.sleep(0.5)
print(" Watch: query_selector_all returns patched handles")
page.goto('https://the-internet.herokuapp.com/checkboxes', wait_until='domcontentloaded')
time.sleep(2)
inject(page)
time.sleep(1)
els = page.query_selector_all('input[type="checkbox"]')
all_patched = all(getattr(e, '_human_patched', False) for e in els)
check("query_selector_all all patched", all_patched and len(els) >= 2, f"{len(els)} elements, all_patched={all_patched}")
if els:
print(" Watch: click checkbox via ElementHandle")
t0 = time.time()
els[0].click()
cb_click_ms = int((time.time() - t0) * 1000)
check("ElementHandle checkbox click", cb_click_ms > 100, f"{cb_click_ms} ms")
time.sleep(1)
# ============================================================
# SUMMARY
# ============================================================
+634 -26
View File
@@ -12,7 +12,7 @@ Can also run directly: python tests/test_humanize_unit.py
import math
import time
import sys
import asyncio
import pytest
@@ -535,7 +535,7 @@ class TestNonAsciiKeyboardAsync:
class TestBrowserFill:
def test_fill_clears_existing(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
@@ -550,7 +550,7 @@ class TestBrowserFill:
def test_fill_timing_humanized(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
@@ -562,7 +562,7 @@ class TestBrowserFill:
def test_clear_empties_field(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
@@ -579,7 +579,7 @@ class TestBrowserFill:
class TestBrowserPatching:
def test_page_has_original(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
assert hasattr(page, '_original')
assert hasattr(page, '_human_cfg')
@@ -587,7 +587,7 @@ class TestBrowserPatching:
def test_locator_methods_patched(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
from playwright.sync_api._generated import Locator
methods = ['fill', 'click', 'type', 'dblclick', 'hover', 'check', 'uncheck',
@@ -608,7 +608,7 @@ class TestBrowserPatching:
def test_page_human_cfg_persists(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True)
browser = launch(headless=False, humanize=True)
page = browser.new_page()
assert page._human_cfg is not None
assert hasattr(page._human_cfg, 'idle_between_actions')
@@ -618,7 +618,7 @@ class TestBrowserPatching:
@pytest.mark.slow
class TestBrowserBotDetection:
PROXY = ''
PROXY = None
def test_behavioral_checks_pass(self):
from cloakbrowser import launch
@@ -644,7 +644,7 @@ class TestBrowserBotDetection:
def test_form_timing(self):
from cloakbrowser import launch
browser = launch(headless=True, humanize=True, proxy=self.PROXY, geoip=True)
browser = launch(headless=False, humanize=True, proxy=self.PROXY, geoip=True)
page = browser.new_page()
page.goto('https://deviceandbrowserinfo.com/are_you_a_bot_interactions',
wait_until='domcontentloaded')
@@ -661,31 +661,639 @@ class TestBrowserBotDetection:
@pytest.mark.slow
class TestAsyncEndToEnd:
def test_async_launch_click_fill(self):
@pytest.mark.asyncio
async def test_async_launch_click_fill(self):
"""launch_async(humanize=True) — async page.click and page.fill work end-to-end."""
import asyncio
from cloakbrowser import launch_async
browser = await launch_async(headless=False, humanize=True)
page = await browser.new_page()
assert hasattr(page, '_original'), "async page not patched"
assert hasattr(page, '_human_cfg'), "async page missing _human_cfg"
async def _run():
browser = await launch_async(headless=True, humanize=True)
page = await browser.new_page()
assert hasattr(page, '_original'), "async page not patched"
assert hasattr(page, '_human_cfg'), "async page missing _human_cfg"
await page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
await asyncio.sleep(1)
await page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
await asyncio.sleep(1)
t0 = time.time()
await page.locator('#searchInput').fill('async test')
elapsed_ms = int((time.time() - t0) * 1000)
assert elapsed_ms > 500, f"async fill too fast: {elapsed_ms}ms"
t0 = time.time()
await page.locator('#searchInput').fill('async test')
elapsed_ms = int((time.time() - t0) * 1000)
assert elapsed_ms > 500, f"async fill too fast: {elapsed_ms}ms"
val = await page.locator('#searchInput').input_value()
assert val == 'async test', f"async fill wrong value: {val}"
val = await page.locator('#searchInput').input_value()
assert val == 'async test', f"async fill wrong value: {val}"
await browser.close()
await browser.close()
asyncio.run(_run())
# =========================================================================
# 12. ElementHandle patching — SYNC
# =========================================================================
class TestElementHandlePatchingSync:
"""Test that ElementHandle objects returned by query_selector etc. are humanized."""
def test_patch_single_element_handle_marks_patched(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
cursor.initialized = True
cursor.x = 100
cursor.y = 100
page = MagicMock()
page._original = MagicMock()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=True) # is_input
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
raw_mouse = MagicMock()
raw_keyboard = MagicMock()
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, page._original, None, None
)
assert el._human_patched is True
def test_element_handle_click_calls_human_move(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", {"idle_between_actions": False})
cursor = _CursorState()
cursor.initialized = True
cursor.x = 100
cursor.y = 100
page = MagicMock()
page._original = MagicMock()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 200, "y": 200, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
raw_mouse = MagicMock()
raw_mouse.move = MagicMock()
raw_mouse.down = MagicMock()
raw_mouse.up = MagicMock()
raw_mouse.wheel = MagicMock()
raw_keyboard = MagicMock()
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, page._original, None, None
)
# Call the patched click
el.click()
# Should call raw_mouse.move (Bezier path) and then down/up
assert raw_mouse.move.called
assert raw_mouse.down.called
assert raw_mouse.up.called
def test_element_handle_hover_moves_cursor_without_click(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", {"idle_between_actions": False})
cursor = _CursorState()
cursor.initialized = True
cursor.x = 50
cursor.y = 50
page = MagicMock()
page._original = MagicMock()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 200, "y": 200, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
raw_mouse = MagicMock()
raw_mouse.move = MagicMock()
raw_mouse.down = MagicMock()
raw_mouse.up = MagicMock()
raw_mouse.wheel = MagicMock()
raw_keyboard = MagicMock()
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, page._original, None, None
)
el.hover()
# Move should be called, but NOT down/up (hover, not click)
assert raw_mouse.move.called
assert not raw_mouse.down.called
def test_element_handle_type_calls_human_type(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", {"idle_between_actions": False, "mistype_chance": 0})
cursor = _CursorState()
cursor.initialized = True
cursor.x = 50
cursor.y = 50
page = MagicMock()
originals = MagicMock()
page._original = originals
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 200, "y": 200, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=True) # is input
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
raw_mouse = MagicMock()
raw_mouse.move = MagicMock()
raw_mouse.down = MagicMock()
raw_mouse.up = MagicMock()
raw_mouse.wheel = MagicMock()
raw_keyboard = MagicMock()
raw_keyboard.down = MagicMock()
raw_keyboard.up = MagicMock()
raw_keyboard.insert_text = MagicMock()
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, None, None
)
el.type("hello")
# Mouse moved + clicked (to focus), then keyboard used
assert raw_mouse.move.called
assert raw_mouse.down.called # click to focus the input
# Keyboard events should have fired (down/up for ASCII chars)
assert raw_keyboard.down.called or raw_keyboard.insert_text.called
def test_element_handle_fill_clears_and_types(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock, call
cfg = resolve_config("default", {"idle_between_actions": False, "mistype_chance": 0})
cursor = _CursorState()
cursor.initialized = True
cursor.x = 50
cursor.y = 50
page = MagicMock()
originals = MagicMock()
page._original = originals
pressed_keys = []
originals.keyboard_press = MagicMock(side_effect=lambda k: pressed_keys.append(k))
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 200, "y": 200, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=True)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
raw_mouse = MagicMock()
raw_mouse.move = MagicMock()
raw_mouse.down = MagicMock()
raw_mouse.up = MagicMock()
raw_mouse.wheel = MagicMock()
raw_keyboard = MagicMock()
raw_keyboard.down = MagicMock()
raw_keyboard.up = MagicMock()
raw_keyboard.insert_text = MagicMock()
_patch_single_element_handle_sync(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, None, None
)
el.fill("replaced")
# Should have pressed Select-All and Backspace to clear
import sys
expected_select = "Meta+a" if sys.platform == "darwin" else "Control+a"
assert expected_select in pressed_keys
assert "Backspace" in pressed_keys
def test_element_handle_no_double_patching(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
page = MagicMock()
page._original = MagicMock()
el = MagicMock()
el._human_patched = False
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
_patch_single_element_handle_sync(
el, page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
# Save patched click
first_click = el.click
# Try to patch again
_patch_single_element_handle_sync(
el, page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
# Should be the same — no double wrap
assert el.click is first_click
def test_nested_query_selector_returns_patched_handle(self):
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
page = MagicMock()
page._original = MagicMock()
child = MagicMock()
child._human_patched = False
child.bounding_box = MagicMock(return_value={"x": 10, "y": 10, "width": 50, "height": 30})
child.evaluate = MagicMock(return_value=False)
child.is_checked = MagicMock(return_value=False)
child.query_selector = MagicMock(return_value=None)
child.query_selector_all = MagicMock(return_value=[])
child.wait_for_selector = MagicMock(return_value=None)
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=child)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
_patch_single_element_handle_sync(
el, page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
result = el.query_selector("span")
assert result._human_patched is True
def test_page_query_selector_patched(self):
from cloakbrowser.human import _patch_page_element_handles_sync, _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
page = MagicMock()
page._original = MagicMock()
page.query_selector = MagicMock(return_value=el)
page.query_selector_all = MagicMock(return_value=[el])
page.wait_for_selector = MagicMock(return_value=el)
_patch_page_element_handles_sync(
page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
result = page.query_selector("#test")
assert result._human_patched is True
def test_page_query_selector_all_patches_all(self):
from cloakbrowser.human import _patch_page_element_handles_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
def make_el():
e = MagicMock()
e._human_patched = False
e.bounding_box = MagicMock(return_value={"x": 10, "y": 10, "width": 50, "height": 30})
e.evaluate = MagicMock(return_value=False)
e.is_checked = MagicMock(return_value=False)
e.query_selector = MagicMock(return_value=None)
e.query_selector_all = MagicMock(return_value=[])
e.wait_for_selector = MagicMock(return_value=None)
return e
el1, el2, el3 = make_el(), make_el(), make_el()
page = MagicMock()
page._original = MagicMock()
page.query_selector = MagicMock(return_value=None)
page.query_selector_all = MagicMock(return_value=[el1, el2, el3])
page.wait_for_selector = MagicMock(return_value=None)
_patch_page_element_handles_sync(
page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
results = page.query_selector_all("div")
for r in results:
assert r._human_patched is True
def test_wait_for_selector_patched(self):
from cloakbrowser.human import _patch_page_element_handles_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
page = MagicMock()
page._original = MagicMock()
page.query_selector = MagicMock(return_value=None)
page.query_selector_all = MagicMock(return_value=[])
page.wait_for_selector = MagicMock(return_value=el)
_patch_page_element_handles_sync(
page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
result = page.wait_for_selector("#test")
assert result._human_patched is True
def test_element_handle_all_methods_patched(self):
"""Verify all expected interaction methods are replaced."""
from cloakbrowser.human import _patch_single_element_handle_sync, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock
cfg = resolve_config("default", None)
cursor = _CursorState()
page = MagicMock()
page._original = MagicMock()
el = MagicMock()
el._human_patched = False
el.bounding_box = MagicMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = MagicMock(return_value=False)
el.is_checked = MagicMock(return_value=False)
el.query_selector = MagicMock(return_value=None)
el.query_selector_all = MagicMock(return_value=[])
el.wait_for_selector = MagicMock(return_value=None)
el.set_checked = MagicMock() # ensure it exists
_patch_single_element_handle_sync(
el, page, cfg, cursor, MagicMock(), MagicMock(), page._original, None, None
)
expected_methods = ['click', 'dblclick', 'hover', 'type', 'fill', 'press',
'select_option', 'check', 'uncheck', 'set_checked',
'tap', 'focus', 'query_selector', 'query_selector_all',
'wait_for_selector']
for method in expected_methods:
fn = getattr(el, method)
assert not isinstance(fn, MagicMock), f"el.{method} was not patched"
# =========================================================================
# 13. ElementHandle patching — ASYNC
# =========================================================================
class TestElementHandlePatchingAsync:
@pytest.mark.asyncio
async def test_async_element_handle_click(self):
from cloakbrowser.human import _patch_single_element_handle_async, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock, AsyncMock
cfg = resolve_config("default", {"idle_between_actions": False})
cursor = _CursorState()
cursor.initialized = True
cursor.x = 100
cursor.y = 100
page = MagicMock()
originals = MagicMock()
originals.mouse_move = AsyncMock()
page._original = originals
el = MagicMock()
el._human_patched = False
el.bounding_box = AsyncMock(return_value={"x": 200, "y": 200, "width": 100, "height": 30})
el.evaluate = AsyncMock(return_value=False)
el.is_checked = AsyncMock(return_value=False)
el.query_selector = AsyncMock(return_value=None)
el.query_selector_all = AsyncMock(return_value=[])
el.wait_for_selector = AsyncMock(return_value=None)
raw_mouse = MagicMock()
raw_mouse.move = AsyncMock()
raw_mouse.down = AsyncMock()
raw_mouse.up = AsyncMock()
raw_mouse.wheel = AsyncMock()
raw_keyboard = MagicMock()
raw_keyboard.down = AsyncMock()
raw_keyboard.up = AsyncMock()
raw_keyboard.insert_text = AsyncMock()
stealth = MagicMock()
stealth.get_cdp_session = AsyncMock(return_value=None)
_patch_single_element_handle_async(
el, page, cfg, cursor, raw_mouse, raw_keyboard, originals, stealth, [None]
)
await el.click()
assert raw_mouse.move.called
assert raw_mouse.down.called
assert raw_mouse.up.called
@pytest.mark.asyncio
async def test_async_page_query_selector_patched(self):
from cloakbrowser.human import _patch_page_element_handles_async, _CursorState
from cloakbrowser.human.config import resolve_config
from unittest.mock import MagicMock, AsyncMock
cfg = resolve_config("default", None)
cursor = _CursorState()
el = MagicMock()
el._human_patched = False
el.bounding_box = AsyncMock(return_value={"x": 50, "y": 50, "width": 100, "height": 30})
el.evaluate = AsyncMock(return_value=False)
el.is_checked = AsyncMock(return_value=False)
el.query_selector = AsyncMock(return_value=None)
el.query_selector_all = AsyncMock(return_value=[])
el.wait_for_selector = AsyncMock(return_value=None)
page = MagicMock()
page._original = MagicMock()
page.query_selector = AsyncMock(return_value=el)
page.query_selector_all = AsyncMock(return_value=[el])
page.wait_for_selector = AsyncMock(return_value=el)
stealth = MagicMock()
stealth.get_cdp_session = AsyncMock(return_value=None)
_patch_page_element_handles_async(
page, cfg, cursor, MagicMock(), MagicMock(), page._original, stealth, [None]
)
result = await page.query_selector("#test")
assert result._human_patched is True
# =========================================================================
# 14. SLOW: Browser ElementHandle end-to-end
# =========================================================================
@pytest.mark.slow
class TestBrowserElementHandle:
def test_query_selector_click_humanized(self):
"""page.query_selector() returns a patched handle — el.click() uses human curves."""
from cloakbrowser import launch
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
el = page.query_selector('#searchInput')
assert el is not None
assert getattr(el, '_human_patched', False), "ElementHandle not patched"
t0 = time.time()
el.click()
click_ms = int((time.time() - t0) * 1000)
assert click_ms > 100, f"ElementHandle click too fast: {click_ms}ms (not humanized)"
browser.close()
def test_query_selector_type_humanized(self):
"""el.type() should type character-by-character with human timing."""
from cloakbrowser import launch
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
el = page.query_selector('#searchInput')
assert el is not None
t0 = time.time()
el.type('ElementHandle test')
type_ms = int((time.time() - t0) * 1000)
assert type_ms > 1000, f"ElementHandle type too fast: {type_ms}ms"
val = page.locator('#searchInput').input_value()
assert val == 'ElementHandle test'
browser.close()
def test_query_selector_fill_humanized(self):
"""el.fill() should clear + type with human timing."""
from cloakbrowser import launch
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
el = page.query_selector('#searchInput')
el.type('initial')
time.sleep(0.3)
t0 = time.time()
el.fill('replaced')
fill_ms = int((time.time() - t0) * 1000)
assert fill_ms > 500, f"ElementHandle fill too fast: {fill_ms}ms"
val = page.locator('#searchInput').input_value()
assert val == 'replaced'
browser.close()
def test_query_selector_all_returns_patched(self):
"""page.query_selector_all() returns all handles patched."""
from cloakbrowser import launch
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://the-internet.herokuapp.com/checkboxes', wait_until='domcontentloaded')
time.sleep(1)
els = page.query_selector_all('input[type="checkbox"]')
assert len(els) >= 2
for el in els:
assert getattr(el, '_human_patched', False), "ElementHandle not patched"
browser.close()
def test_query_selector_hover_humanized(self):
"""el.hover() should move cursor with human Bezier curve."""
from cloakbrowser import launch
browser = launch(headless=False, humanize=True)
page = browser.new_page()
page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
time.sleep(1)
el = page.query_selector('#searchInput')
t0 = time.time()
el.hover()
hover_ms = int((time.time() - t0) * 1000)
assert hover_ms > 50, f"ElementHandle hover too fast: {hover_ms}ms"
browser.close()
@pytest.mark.slow
class TestAsyncElementHandle:
@pytest.mark.asyncio
async def test_async_query_selector_click(self):
from cloakbrowser import launch_async
browser = await launch_async(headless=False, humanize=True)
page = await browser.new_page()
await page.goto('https://www.wikipedia.org', wait_until='domcontentloaded')
await asyncio.sleep(1)
el = await page.query_selector('#searchInput')
assert el is not None
assert getattr(el, '_human_patched', False), "Async ElementHandle not patched"
t0 = time.time()
await el.click()
click_ms = int((time.time() - t0) * 1000)
assert click_ms > 100, f"Async ElementHandle click too fast: {click_ms}ms"
await browser.close()
# =========================================================================
+127 -15
View File
@@ -2,7 +2,12 @@
from unittest.mock import patch
from cloakbrowser.browser import _build_proxy_kwargs, maybe_resolve_geoip, _parse_proxy_url
from cloakbrowser.browser import (
_is_socks_proxy,
_parse_proxy_url,
_resolve_proxy_config,
maybe_resolve_geoip,
)
class TestParseProxyUrl:
@@ -38,23 +43,30 @@ class TestParseProxyUrl:
class TestBuildProxyKwargs:
"""Tests for _resolve_proxy_config (formerly _build_proxy_kwargs) HTTP path."""
def test_none(self):
assert _build_proxy_kwargs(None) == {}
kwargs, args = _resolve_proxy_config(None)
assert kwargs == {}
assert args == []
def test_simple_proxy(self):
result = _build_proxy_kwargs("http://proxy:8080")
assert result == {"proxy": {"server": "http://proxy:8080"}}
kwargs, args = _resolve_proxy_config("http://proxy:8080")
assert kwargs == {"proxy": {"server": "http://proxy:8080"}}
assert args == []
def test_proxy_with_auth(self):
result = _build_proxy_kwargs("http://user:pass@proxy:8080")
assert result == {
kwargs, args = _resolve_proxy_config("http://user:pass@proxy:8080")
assert kwargs == {
"proxy": {"server": "http://proxy:8080", "username": "user", "password": "pass"}
}
assert args == []
def test_proxy_dict_passthrough(self):
proxy_dict = {"server": "http://proxy:8080", "bypass": ".google.com,localhost"}
result = _build_proxy_kwargs(proxy_dict)
assert result == {"proxy": proxy_dict}
kwargs, args = _resolve_proxy_config(proxy_dict)
assert kwargs == {"proxy": proxy_dict}
assert args == []
def test_proxy_dict_with_auth(self):
proxy_dict = {
@@ -63,8 +75,9 @@ class TestBuildProxyKwargs:
"password": "pass",
"bypass": ".example.com",
}
result = _build_proxy_kwargs(proxy_dict)
assert result == {"proxy": proxy_dict}
kwargs, args = _resolve_proxy_config(proxy_dict)
assert kwargs == {"proxy": proxy_dict}
assert args == []
class TestMaybeResolveGeoip:
@@ -117,6 +130,27 @@ class TestMaybeResolveGeoip:
mock_geo.assert_called_once_with("http://proxy:8080")
assert tz == "America/New_York"
@patch("cloakbrowser.geoip.resolve_proxy_geo_with_ip", return_value=("Europe/Berlin", "de-DE", "5.6.7.8"))
def test_geoip_socks5_dict_reconstructs_credentials(self, mock_geo):
proxy_dict = {"server": "socks5://proxy:1080", "username": "user", "password": "pass"}
tz, locale, ip = maybe_resolve_geoip(True, proxy_dict, None, None)
mock_geo.assert_called_once_with("socks5://user:pass@proxy:1080")
assert tz == "Europe/Berlin"
assert locale == "de-DE"
@patch("cloakbrowser.geoip.resolve_proxy_geo_with_ip", return_value=("Europe/Berlin", "de-DE", "5.6.7.8"))
def test_geoip_socks5_dict_no_auth_uses_server(self, mock_geo):
proxy_dict = {"server": "socks5://proxy:1080"}
tz, locale, ip = maybe_resolve_geoip(True, proxy_dict, None, None)
mock_geo.assert_called_once_with("socks5://proxy:1080")
@patch("cloakbrowser.geoip.resolve_proxy_geo_with_ip", return_value=("Europe/London", "en-GB", "1.1.1.1"))
def test_geoip_http_dict_does_not_inline_creds(self, mock_geo):
# HTTP dict: credentials stay separate, only server URL passed
proxy_dict = {"server": "http://proxy:8080", "username": "user", "password": "pass"}
tz, locale, ip = maybe_resolve_geoip(True, proxy_dict, None, None)
mock_geo.assert_called_once_with("http://proxy:8080")
class TestBareProxyFormat:
"""_parse_proxy_url must handle bare 'user:pass@host:port' strings (no scheme)."""
@@ -149,8 +183,86 @@ class TestBareProxyFormat:
r = _parse_proxy_url("proxy:8080")
assert r == {"server": "proxy:8080"}
def test_build_proxy_kwargs_bare(self):
r = _build_proxy_kwargs("user:pass@proxy:8080")
assert r["proxy"]["username"] == "user"
assert r["proxy"]["password"] == "pass"
assert "user" not in r["proxy"]["server"]
def test_resolve_proxy_config_bare(self):
kwargs, args = _resolve_proxy_config("user:pass@proxy:8080")
assert kwargs["proxy"]["username"] == "user"
assert kwargs["proxy"]["password"] == "pass"
assert "user" not in kwargs["proxy"]["server"]
class TestIsSocksProxy:
def test_socks5_string(self):
assert _is_socks_proxy("socks5://user:pass@host:1080") is True
def test_socks5h_string(self):
assert _is_socks_proxy("socks5h://host:1080") is True
def test_socks5_uppercase(self):
assert _is_socks_proxy("SOCKS5://host:1080") is True
def test_http_string(self):
assert _is_socks_proxy("http://host:8080") is False
def test_dict_socks5(self):
assert _is_socks_proxy({"server": "socks5://host:1080"}) is True
def test_dict_http(self):
assert _is_socks_proxy({"server": "http://host:8080"}) is False
def test_none(self):
assert _is_socks_proxy(None) is False
class TestResolveProxyConfig:
def test_none(self):
kwargs, args = _resolve_proxy_config(None)
assert kwargs == {}
assert args == []
def test_http_string_returns_playwright_dict(self):
kwargs, args = _resolve_proxy_config("http://user:pass@proxy:8080")
assert "proxy" in kwargs
assert kwargs["proxy"]["server"] == "http://proxy:8080"
assert kwargs["proxy"]["username"] == "user"
assert args == []
def test_http_dict_passthrough(self):
proxy = {"server": "http://proxy:8080", "bypass": ".example.com"}
kwargs, args = _resolve_proxy_config(proxy)
assert kwargs == {"proxy": proxy}
assert args == []
def test_socks5_string_returns_chrome_arg(self):
kwargs, args = _resolve_proxy_config("socks5://user:pass@host:1080")
assert kwargs == {}
assert args == ["--proxy-server=socks5://user:pass@host:1080"]
def test_socks5_no_auth_returns_chrome_arg(self):
kwargs, args = _resolve_proxy_config("socks5://host:1080")
assert kwargs == {}
assert args == ["--proxy-server=socks5://host:1080"]
def test_socks5h_returns_chrome_arg(self):
kwargs, args = _resolve_proxy_config("socks5h://user:pass@host:1080")
assert kwargs == {}
assert args == ["--proxy-server=socks5h://user:pass@host:1080"]
def test_socks5_dict_reconstructs_url(self):
proxy = {"server": "socks5://host:1080", "username": "user", "password": "p@ss"}
kwargs, args = _resolve_proxy_config(proxy)
assert kwargs == {}
assert len(args) == 1
assert args[0].startswith("--proxy-server=socks5://user:p%40ss@host:1080")
def test_socks5_dict_ipv6_preserves_brackets(self):
proxy = {"server": "socks5://[::1]:1080", "username": "user", "password": "pass"}
kwargs, args = _resolve_proxy_config(proxy)
assert kwargs == {}
assert "[::1]" in args[0]
def test_socks5_dict_with_bypass(self):
proxy = {"server": "socks5://host:1080", "bypass": ".example.com"}
kwargs, args = _resolve_proxy_config(proxy)
assert kwargs == {}
assert "--proxy-server=socks5://host:1080" in args
assert "--proxy-bypass-list=.example.com" in args