mirror of
https://github.com/CloakHQ/CloakBrowser.git
synced 2026-06-23 11:41:46 +02:00
Compare commits
15
Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
ee346a6a57 | ||
|
|
3e699f554c | ||
|
|
9eb90da012 | ||
|
|
6b8d8b6378 | ||
|
|
252e79b17d | ||
|
|
0ccdc71e47 | ||
|
|
74b1ff64db | ||
|
|
a9a0ba13ba | ||
|
|
b04ad6ec2a | ||
|
|
4459f66593 | ||
|
|
ce8b92ba4f | ||
|
|
4e1027847e | ||
|
|
f164c1c874 | ||
|
|
f5e242a160 | ||
|
|
935beef980 |
@@ -1,62 +0,0 @@
|
||||
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
|
||||
@@ -23,7 +23,7 @@ jobs:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
|
||||
- uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
|
||||
- uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
|
||||
with:
|
||||
node-version: 20
|
||||
- name: Install and build
|
||||
|
||||
@@ -1,148 +0,0 @@
|
||||
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 "ℹ️ No relevant resources found, staying silent"
|
||||
fi
|
||||
@@ -32,7 +32,7 @@ jobs:
|
||||
run: |
|
||||
pip install -e ".[dev]" pytest pytest-asyncio
|
||||
pytest tests/ -v -m "not slow"
|
||||
- uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
|
||||
- uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
|
||||
with:
|
||||
node-version: 22
|
||||
- name: JavaScript tests
|
||||
@@ -71,7 +71,7 @@ jobs:
|
||||
pip install build
|
||||
python -m build
|
||||
- name: Publish to PyPI
|
||||
uses: pypa/gh-action-pypi-publish@ed0c53931b1dc9bd32cbe73a98c7f6766f8a527e # v1
|
||||
uses: pypa/gh-action-pypi-publish@cef221092ed1bacb1cc03d23a2d87d1d172e277b # v1
|
||||
|
||||
publish-npm:
|
||||
needs: [test, validate-version]
|
||||
@@ -81,7 +81,7 @@ jobs:
|
||||
id-token: write # OIDC trusted publishing + provenance — no NPM_TOKEN needed
|
||||
steps:
|
||||
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
|
||||
- uses: actions/setup-node@53b83947a5a98c8d113130e565377fae1a50d02f # v6.3.0
|
||||
- uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
|
||||
with:
|
||||
node-version: 24 # npm 11.11.0 native — no upgrade needed (Node 22.22.2 has broken npm)
|
||||
registry-url: 'https://registry.npmjs.org'
|
||||
@@ -113,7 +113,7 @@ jobs:
|
||||
password: ${{ secrets.DOCKER_PAT }}
|
||||
- name: Build and push
|
||||
id: build
|
||||
uses: docker/build-push-action@d08e5c354a6adb9ed34480a06d141179aa583294 # v7.0.0
|
||||
uses: docker/build-push-action@bcafcacb16a39f128d818304e6c9c0c18556b85f # v7.1.0
|
||||
with:
|
||||
context: .
|
||||
platforms: linux/amd64,linux/arm64
|
||||
|
||||
+21
-1
@@ -6,10 +6,30 @@ Changes are tagged: **[wrapper]** for Python/JS wrapper, **[binary]** for Chromi
|
||||
|
||||
---
|
||||
|
||||
## [Unreleased]
|
||||
|
||||
## [0.3.26] — 2026-04-28
|
||||
|
||||
- **[binary]** Windows x64 upgraded to Chromium 146.0.7680.177.4 — 57 source-level fingerprint patches (up from 33 on 145.0.7632.159.7), now matches Linux. Includes all binary improvements from 0.3.18–0.3.25: native SOCKS5 proxy with UDP ASSOCIATE (QUIC/HTTP3), WebRTC IP spoofing, proxy signal removal, CDP input stealth, storage quota normalization, WebAuthn/AAC/window position patches, WebGL and canvas consistency fixes, expanded GPU model database
|
||||
- **[wrapper]** Auto URL-encode SOCKS5 credentials containing special characters in string URLs (#157)
|
||||
- **[wrapper]** AWS Lambda integration example with cold-start hardening and handler-side retry orchestration (#177, thanks [@AlexTech314](https://github.com/AlexTech314))
|
||||
- **[docker]** Add emoji and extended font packages to resolve Kasada/Akamai canvas fingerprint blocks (#179)
|
||||
- **[docs]** Add Font Setup on Linux section to README (#179)
|
||||
- **[docs]** Add Deployment Integrations section to README (#177)
|
||||
- **[meta]** Bump GitHub Actions dependencies (#178)
|
||||
|
||||
## [0.3.25] — 2026-04-16
|
||||
|
||||
- **[wrapper]** Python: add `launch_context_async()` — async counterpart to `launch_context()`. Returns a BrowserContext with all kwargs forwarded to `browser.new_context()`, enabling `storage_state`, `permissions`, `extra_http_headers`, etc. without a persistent profile folder. Closes #141.
|
||||
- **[wrapper]** JS: `launchContext()` and `launchPersistentContext()` silently dropped unknown options (including `storageState`). New `contextOptions` escape hatch forwards arbitrary options to Playwright's `newContext()`.
|
||||
- **[wrapper]** Fix `humanConfig` TypeScript typing (#151).
|
||||
- **[binary]** New build 146.0.7680.177.3 for Linux x64 + arm64 — 57 source-level fingerprint patches (up from 49): WebAuthn capabilities, AAC audio encoder, and window position spoofing; WebGL and canvas format consistency fixes; SOCKS5 warm connection pool auth fix for credentialed proxies.
|
||||
- **[docs]** Add recommended anti-bot config and SOCKS5 tips to troubleshooting.
|
||||
|
||||
## [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)
|
||||
- **[wrapper]** Add Playwright ElementHandle humanize support — `element_handle.click()`, `.fill()`, `.type()` now use human-like behavior when `humanize=True` (thanks [@evelaa123](https://github.com/evelaa123), #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)
|
||||
|
||||
@@ -9,6 +9,8 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
|
||||
libxcb1 libxext6 libxshmfence1 \
|
||||
libglib2.0-0 libgtk-3-0 libpangocairo-1.0-0 libcairo-gobject2 \
|
||||
libgdk-pixbuf-2.0-0 libxss1 libxtst6 fonts-liberation \
|
||||
fonts-noto-color-emoji fonts-unifont fonts-freefont-ttf \
|
||||
fonts-ipafont-gothic fonts-wqy-zenhei fonts-tlwg-loma-otf \
|
||||
xvfb xdotool \
|
||||
curl ca-certificates \
|
||||
&& curl -fsSL https://deb.nodesource.com/setup_20.x | bash - \
|
||||
|
||||
@@ -128,11 +128,13 @@ Open [http://localhost:8080](http://localhost:8080). Create a profile. Click **L
|
||||
|
||||
---
|
||||
|
||||
## Latest: v0.3.24 (Chromium 146.0.7680.177.2)
|
||||
## Latest: v0.3.26 (Chromium 146.0.7680.177.4)
|
||||
|
||||
- **`launch_context_async()`** — async counterpart to `launch_context()`. Forwards kwargs to `browser.new_context()` for `storage_state`, `permissions`, `extra_http_headers` without a persistent profile folder.
|
||||
- **JS `contextOptions` escape hatch** — forward arbitrary options (including `storageState`) to Playwright's `newContext()` from `launchContext()` / `launchPersistentContext()`.
|
||||
- **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 arm64 now matches Linux x64 on Chromium 146
|
||||
- **57 fingerprint patches** — additional detection-vector coverage (WebAuthn, AAC audio, window position) and WebGL/canvas consistency fixes
|
||||
- **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
|
||||
@@ -316,6 +318,38 @@ page.goto("https://protected-site.com")
|
||||
context.close()
|
||||
```
|
||||
|
||||
Extra kwargs are forwarded to Playwright's `browser.new_context()` — use this for `storage_state`, `permissions`, `extra_http_headers`, etc. without needing a persistent profile folder:
|
||||
|
||||
```python
|
||||
from cloakbrowser import launch_context
|
||||
|
||||
# Restore a saved session (cookies, localStorage) from a JSON file
|
||||
context = launch_context(storage_state="state.json")
|
||||
page = context.new_page()
|
||||
page.goto("https://example.com")
|
||||
# Save state back for next run
|
||||
context.storage_state(path="state.json")
|
||||
context.close()
|
||||
```
|
||||
|
||||
### `launch_context_async()`
|
||||
|
||||
Async counterpart to `launch_context()`. Same signature and kwargs forwarding:
|
||||
|
||||
```python
|
||||
import asyncio
|
||||
from cloakbrowser import launch_context_async
|
||||
|
||||
async def main():
|
||||
ctx = await launch_context_async(storage_state="state.json")
|
||||
page = await ctx.new_page()
|
||||
await page.goto("https://example.com")
|
||||
await ctx.storage_state(path="state.json")
|
||||
await ctx.close()
|
||||
|
||||
asyncio.run(main())
|
||||
```
|
||||
|
||||
### `launch_persistent_context()`
|
||||
|
||||
Same as `launch_context()`, but with a persistent user profile. Cookies, localStorage, and cache persist across sessions.
|
||||
@@ -372,7 +406,7 @@ from cloakbrowser import binary_info, clear_cache, ensure_binary
|
||||
|
||||
# Check binary installation status
|
||||
print(binary_info())
|
||||
# {'version': '146.0.7680.177.2', 'platform': 'linux-x64', 'installed': True, ...}
|
||||
# {'version': '146.0.7680.177.3', 'platform': 'linux-x64', 'installed': True, ...}
|
||||
|
||||
# Force re-download
|
||||
clear_cache()
|
||||
@@ -592,13 +626,39 @@ Supported by the binary but **not set by default** — pass via `args` to custom
|
||||
| `--fingerprint-locale` | Locale (e.g. `en-US`) |
|
||||
| `--fingerprint-storage-quota` | Override storage quota in MB — affects `storage.estimate()`, `storageBuckets`, and legacy webkit APIs. Auto-normalized when `--fingerprint` is set |
|
||||
| `--fingerprint-taskbar-height` | Override taskbar height (binary defaults: Win=48, Mac=95, Linux=0) |
|
||||
| `--fingerprint-fonts-dir` | Path to cross-platform font directory |
|
||||
| `--fingerprint-fonts-dir` | Path to directory containing target-platform fonts (see [Font Setup on Linux](#font-setup-on-linux)) |
|
||||
| `--fingerprint-webrtc-ip` | WebRTC ICE candidate IP replacement. Use `auto` to resolve from proxy exit IP (makes an HTTP call through the proxy), or pass an explicit IP. Auto-injected when `geoip=True` |
|
||||
| `--fingerprint-noise=false` | Disable noise injection (canvas, WebGL, audio, client rects) while keeping the deterministic fingerprint seed active |
|
||||
| `--enable-blink-features=FakeShadowRoot` | Access closed shadow DOM elements |
|
||||
|
||||
> **Note:** All stealth tests were verified with the default fingerprint config above. Changing these flags may affect detection results — test your configuration before using in production.
|
||||
|
||||
### Font Setup on Linux
|
||||
|
||||
**Required for aggressive anti-bot sites (Kasada, Akamai).** These systems render emoji on a hidden canvas and hash the pixel output. Minimal Linux environments (Docker, cloud VMs) often lack emoji and extended fonts, producing hashes that don't match any real browser. Install standard font packages to fix this:
|
||||
|
||||
```bash
|
||||
sudo apt install -y fonts-noto-color-emoji fonts-freefont-ttf fonts-unifont \
|
||||
fonts-ipafont-gothic fonts-wqy-zenhei fonts-tlwg-loma-otf
|
||||
```
|
||||
|
||||
The Docker image (`cloakhq/cloakbrowser`) ships with these pre-installed. If you run the binary directly on a Linux server or in a custom Docker image, install them manually.
|
||||
|
||||
**Optional: Windows fonts for CreepJS font enumeration.** The packages above fix anti-bot canvas checks but won't improve your CreepJS font score. For that, you need actual Windows fonts (Segoe UI, Calibri, Bahnschrift, etc.) from a Windows machine's `C:\Windows\Fonts\` directory — `ttf-mscorefonts-installer` only has old XP-era fonts and isn't enough.
|
||||
|
||||
```bash
|
||||
mkdir -p ~/.local/share/fonts/windows
|
||||
cp /path/to/windows/fonts/*.ttf ~/.local/share/fonts/windows/
|
||||
cp /path/to/windows/fonts/*.TTF ~/.local/share/fonts/windows/
|
||||
fc-cache -f # mandatory for manually copied fonts
|
||||
```
|
||||
|
||||
```python
|
||||
browser = launch(
|
||||
args=["--fingerprint-fonts-dir=/home/user/.local/share/fonts/windows"],
|
||||
)
|
||||
```
|
||||
|
||||
### Examples
|
||||
|
||||
```python
|
||||
@@ -669,15 +729,21 @@ browser = await launch_async(args=["--remote-debugging-port=9242"])
|
||||
| [undetected-chromedriver](https://github.com/ultrafunkamsterdam/undetected-chromedriver) | 12K | Python | [`undetected_chromedriver.py`](examples/integrations/undetected_chromedriver.py) |
|
||||
| [agent-browser](https://github.com/nichochar/agent-browser) | — | Shell | [`agent_browser.sh`](examples/integrations/agent_browser.sh) |
|
||||
|
||||
### Deployment Integrations
|
||||
|
||||
| Platform | Example |
|
||||
|----------|---------|
|
||||
| [AWS Lambda](https://aws.amazon.com/lambda/) | [`aws_lambda/`](examples/integrations/aws_lambda/) — One-shot scrapes in Lambda (container image) |
|
||||
|
||||
## Platforms
|
||||
|
||||
| Platform | Chromium | Patches | Status |
|
||||
|---|---|---|---|
|
||||
| Linux x86_64 | 146 | 49 | ✅ Latest |
|
||||
| Linux arm64 (RPi, Graviton) | 146 | 49 | ✅ Latest |
|
||||
| Linux x86_64 | 146 | 57 | ✅ Latest |
|
||||
| Linux arm64 (RPi, Graviton) | 146 | 57 | ✅ Latest |
|
||||
| macOS arm64 (Apple Silicon) | 145 | 26 | ✅ |
|
||||
| macOS x86_64 (Intel) | 145 | 26 | ✅ |
|
||||
| Windows x86_64 | 145 | 48 | ✅ |
|
||||
| Windows x86_64 | 146 | 57 | ✅ Latest |
|
||||
|
||||
The wrapper auto-downloads the correct binary for your platform.
|
||||
|
||||
@@ -874,7 +940,47 @@ page.goto("https://heavily-protected-site.com") # passes DataDome, etc.
|
||||
browser.close()
|
||||
```
|
||||
|
||||
This runs a real headed browser rendered on a virtual display — no physical monitor needed. Combined with a residential proxy, this passes even the most aggressive detection services. Datacenter IPs are often flagged by IP reputation regardless of browser fingerprint — a residential proxy makes the difference.
|
||||
This runs a real headed browser rendered on a virtual display — no physical monitor needed. Combine with the recommended config below for maximum stealth.
|
||||
|
||||
---
|
||||
|
||||
### Recommended config for anti-bot sites
|
||||
|
||||
Most blocks come from missing one of these three things, not from browser fingerprint detection:
|
||||
|
||||
```python
|
||||
browser = launch(
|
||||
proxy="http://your-residential-proxy:port", # residential IP — datacenter IPs get blocked by reputation alone
|
||||
geoip=True, # matches timezone + locale to proxy exit IP (without this: UTC + en-US = bot signal)
|
||||
headless=False, # headed mode — some sites detect headless even with C++ patches
|
||||
humanize=True, # human-like mouse, keyboard, scroll behavior
|
||||
)
|
||||
```
|
||||
|
||||
```javascript
|
||||
const browser = await launch({
|
||||
proxy: 'http://your-residential-proxy:port',
|
||||
geoip: true,
|
||||
headless: false,
|
||||
humanize: true,
|
||||
});
|
||||
```
|
||||
|
||||
If your proxy supports SOCKS5, use it for better compatibility — SOCKS5 tunnels raw TCP, avoiding HTTP CONNECT issues that some proxies have with HTTP/2:
|
||||
|
||||
```python
|
||||
browser = launch(proxy="socks5://user:pass@proxy:1080", geoip=True, headless=False, humanize=True)
|
||||
```
|
||||
|
||||
If you're still blocked after this, check the font setup below.
|
||||
|
||||
---
|
||||
|
||||
### Blocked on Kasada / Akamai sites despite correct config?
|
||||
|
||||
On minimal Linux environments, missing font packages cause canvas emoji rendering to produce hashes that anti-bot systems don't recognize. This is the most common cause of blocks on aggressive sites after proxy, geoip, and headed mode are already set up correctly.
|
||||
|
||||
Install the font packages listed in [Font Setup on Linux](#font-setup-on-linux) above.
|
||||
|
||||
---
|
||||
|
||||
@@ -910,7 +1016,7 @@ await ctx.close();
|
||||
ctx = await launchPersistentContext({ userDataDir: './profile' });
|
||||
```
|
||||
|
||||
For stateless/ephemeral use cases, `launch(args=["--disable-http2"])` forces HTTP/1.1 which bypasses the check. Only use this flag for sites that require it — most work fine with HTTP/2.
|
||||
For stateless/ephemeral use cases, `launch(args=["--disable-http2"])` forces HTTP/1.1 which bypasses the check. Only use this flag for sites that require it — most work fine with HTTP/2. If your proxy supports SOCKS5, use `proxy="socks5://user:pass@host:port"` instead — SOCKS5 bypasses HTTP CONNECT entirely.
|
||||
|
||||
---
|
||||
|
||||
@@ -1036,9 +1142,9 @@ A: Yes. Pass `proxy="http://user:pass@host:port"` or `proxy="socks5://user:pass@
|
||||
|
||||
| Feature | Status |
|
||||
|---------|--------|
|
||||
| Linux x64 — Chromium 146 (49 patches) | ✅ Released |
|
||||
| Linux x64 — Chromium 146 (57 patches) | ✅ Released |
|
||||
| macOS arm64/x64 — Chromium 145 (26 patches) | ✅ Released |
|
||||
| Windows x64 — Chromium 145 (33 patches) | ✅ Released |
|
||||
| Windows x64 — Chromium 146 (57 patches) | ✅ Released |
|
||||
| JavaScript/Puppeteer + Playwright support | ✅ Released |
|
||||
| Fingerprint rotation per session | ✅ Released |
|
||||
| Built-in proxy rotation | 📋 Planned |
|
||||
@@ -1060,7 +1166,7 @@ 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-v146.0.7680.177.2
|
||||
git verify-tag chromium-v146.0.7680.177.3
|
||||
|
||||
# Verify GitHub binary attestation (Sigstore)
|
||||
gh attestation verify cloakbrowser-linux-x64.tar.gz --repo CloakHQ/cloakbrowser
|
||||
@@ -1086,3 +1192,5 @@ Issues and PRs welcome. If something isn't working, [open an issue](https://gith
|
||||
- [@evelaa123](https://github.com/evelaa123) — humanize behavior, persistent contexts, Windows fix
|
||||
- [@yahooguntu](https://github.com/yahooguntu) — persistent contexts
|
||||
- [@kitiho](https://github.com/kitiho) — null viewport fix
|
||||
- [@eofreternal](https://github.com/eofreternal) — humanConfig type fix
|
||||
- [@AlexTech314](https://github.com/AlexTech314) — AWS Lambda integration
|
||||
|
||||
+2
-2
@@ -36,7 +36,7 @@ import aiohttp
|
||||
import websockets
|
||||
from aiohttp import web
|
||||
|
||||
from cloakbrowser.browser import build_args, maybe_resolve_geoip, _resolve_webrtc_args
|
||||
from cloakbrowser.browser import build_args, maybe_resolve_geoip, _resolve_webrtc_args, _normalize_socks_string_url
|
||||
from cloakbrowser.download import ensure_binary
|
||||
|
||||
logging.basicConfig(
|
||||
@@ -189,7 +189,7 @@ class ChromePool:
|
||||
if extra_args:
|
||||
fp_extra.extend(extra_args)
|
||||
if proxy:
|
||||
fp_extra.append(f"--proxy-server={proxy}")
|
||||
fp_extra.append(f"--proxy-server={_normalize_socks_string_url(proxy)}")
|
||||
|
||||
# WebRTC IP spoofing: resolve auto, inject geoip exit IP
|
||||
fp_extra = _resolve_webrtc_args(fp_extra, proxy)
|
||||
|
||||
@@ -11,7 +11,7 @@ Usage:
|
||||
browser.close()
|
||||
"""
|
||||
|
||||
from .browser import launch, launch_async, launch_context, launch_persistent_context, launch_persistent_context_async, ProxySettings, build_args, maybe_resolve_geoip
|
||||
from .browser import launch, launch_async, launch_context, launch_context_async, launch_persistent_context, launch_persistent_context_async, ProxySettings, build_args, maybe_resolve_geoip
|
||||
from .config import CHROMIUM_VERSION, get_default_stealth_args
|
||||
from .download import binary_info, check_for_update, clear_cache, ensure_binary
|
||||
from ._version import __version__
|
||||
@@ -32,6 +32,7 @@ __all__ = [
|
||||
"launch",
|
||||
"launch_async",
|
||||
"launch_context",
|
||||
"launch_context_async",
|
||||
"launch_persistent_context",
|
||||
"launch_persistent_context_async",
|
||||
"ensure_binary",
|
||||
|
||||
@@ -1 +1 @@
|
||||
__version__ = "0.3.24"
|
||||
__version__ = "0.3.26"
|
||||
|
||||
+220
-29
@@ -21,6 +21,7 @@ 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
|
||||
from .human.config import HumanConfigOverrides, HumanPreset
|
||||
|
||||
logger = logging.getLogger("cloakbrowser")
|
||||
|
||||
@@ -60,8 +61,8 @@ def launch(
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: str = "default",
|
||||
human_config: dict | None = None,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Launch stealth Chromium browser. Returns a Playwright Browser object.
|
||||
@@ -87,7 +88,7 @@ def launch(
|
||||
Override globally with CLOAKBROWSER_BACKEND env var.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config dict to override preset values.
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed directly to playwright.chromium.launch().
|
||||
|
||||
Returns:
|
||||
@@ -155,8 +156,8 @@ async def launch_async( # noqa: C901
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: str = "default",
|
||||
human_config: dict | None = None,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Async version of launch(). Returns a Playwright Browser object.
|
||||
@@ -172,7 +173,7 @@ async def launch_async( # noqa: C901
|
||||
backend: Playwright backend — 'playwright' (default) or 'patchright'.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config dict to override preset values.
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed directly to playwright.chromium.launch().
|
||||
|
||||
Returns:
|
||||
@@ -249,8 +250,8 @@ def launch_persistent_context(
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: str = "default",
|
||||
human_config: dict | None = None,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Launch stealth browser with a persistent profile and return a BrowserContext.
|
||||
@@ -279,7 +280,7 @@ def launch_persistent_context(
|
||||
backend: Playwright backend — 'playwright' (default) or 'patchright'.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config dict to override preset values.
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed directly to playwright.chromium.launch_persistent_context().
|
||||
|
||||
Returns:
|
||||
@@ -373,8 +374,8 @@ async def launch_persistent_context_async(
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: str = "default",
|
||||
human_config: dict | None = None,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Async version of launch_persistent_context().
|
||||
@@ -400,7 +401,7 @@ async def launch_persistent_context_async(
|
||||
backend: Playwright backend — 'playwright' (default) or 'patchright'.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config dict to override preset values.
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed directly to playwright.chromium.launch_persistent_context().
|
||||
|
||||
Returns:
|
||||
@@ -498,8 +499,8 @@ def launch_context(
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: str = "default",
|
||||
human_config: dict | None = None,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Launch stealth browser and return a BrowserContext with common options pre-set.
|
||||
@@ -523,7 +524,7 @@ def launch_context(
|
||||
backend: Playwright backend — 'playwright' (default) or 'patchright'.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config dict to override preset values.
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed to browser.new_context().
|
||||
|
||||
Returns:
|
||||
@@ -584,6 +585,130 @@ def launch_context(
|
||||
return context
|
||||
|
||||
|
||||
async def launch_context_async(
|
||||
headless: bool = True,
|
||||
proxy: str | ProxySettings | None = None,
|
||||
args: list[str] | None = None,
|
||||
stealth_args: bool = True,
|
||||
user_agent: str | None = None,
|
||||
viewport: dict | None = _VIEWPORT_UNSET,
|
||||
locale: str | None = None,
|
||||
timezone: str | None = None,
|
||||
color_scheme: Literal["light", "dark", "no-preference"] | None = None,
|
||||
geoip: bool = False,
|
||||
backend: str | None = None,
|
||||
humanize: bool = False,
|
||||
human_preset: HumanPreset = "default",
|
||||
human_config: HumanConfigOverrides | None = None,
|
||||
**kwargs: Any,
|
||||
) -> Any:
|
||||
"""Async version of launch_context().
|
||||
|
||||
Launch stealth browser and return a BrowserContext with common options pre-set.
|
||||
All extra kwargs are forwarded to ``browser.new_context()`` — use this for
|
||||
``storage_state``, ``permissions``, ``extra_http_headers``, etc. without needing
|
||||
a persistent profile folder.
|
||||
|
||||
Args:
|
||||
headless: Run in headless mode (default True).
|
||||
proxy: Proxy URL string or Playwright proxy dict (see launch() for details).
|
||||
args: Additional Chromium CLI arguments.
|
||||
stealth_args: Include default stealth fingerprint args (default True).
|
||||
user_agent: Custom user agent string.
|
||||
viewport: Viewport size dict, e.g. {"width": 1920, "height": 1080}.
|
||||
Pass None to disable viewport emulation (use OS window size).
|
||||
locale: Browser locale, e.g. "en-US".
|
||||
timezone: IANA timezone (e.g. 'America/New_York').
|
||||
color_scheme: Color scheme preference — 'light', 'dark', or 'no-preference'.
|
||||
geoip: Auto-detect timezone/locale from proxy IP (default False).
|
||||
backend: Playwright backend — 'playwright' (default) or 'patchright'.
|
||||
humanize: Enable human-like mouse, keyboard, scroll behavior (default False).
|
||||
human_preset: Humanize preset — 'default' or 'careful' (default 'default').
|
||||
human_config: Custom humanize config mapping to override preset values.
|
||||
**kwargs: Passed to browser.new_context() — e.g. storage_state, permissions.
|
||||
|
||||
Returns:
|
||||
Playwright BrowserContext object (async API).
|
||||
Call ``await .close()`` when done — this also closes the underlying browser.
|
||||
|
||||
Example:
|
||||
>>> import asyncio
|
||||
>>> from cloakbrowser import launch_context_async
|
||||
>>>
|
||||
>>> async def main():
|
||||
... # Load saved session (cookies, localStorage)
|
||||
... ctx = await launch_context_async(
|
||||
... headless=True,
|
||||
... storage_state="state.json",
|
||||
... )
|
||||
... page = await ctx.new_page()
|
||||
... await page.goto("https://example.com")
|
||||
... # Save state back
|
||||
... await ctx.storage_state(path="state.json")
|
||||
... await ctx.close()
|
||||
>>>
|
||||
>>> asyncio.run(main())
|
||||
"""
|
||||
timezone = _resolve_timezone(timezone, kwargs)
|
||||
|
||||
# Resolve geoip BEFORE launch_async() to avoid double-resolution and ensure
|
||||
# resolved values flow to binary flags
|
||||
timezone, locale, exit_ip = maybe_resolve_geoip(geoip, proxy, timezone, locale)
|
||||
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}")
|
||||
# --fingerprint-timezone is process-wide (reads CommandLine in renderer),
|
||||
# so it applies to ALL contexts, not just the default one.
|
||||
# locale and timezone are set via binary flags only — no CDP emulation.
|
||||
browser = await launch_async(headless=headless, proxy=proxy, args=args, stealth_args=stealth_args,
|
||||
timezone=timezone, locale=locale, backend=backend)
|
||||
|
||||
context_kwargs: dict[str, Any] = {}
|
||||
if user_agent:
|
||||
context_kwargs["user_agent"] = user_agent
|
||||
if viewport is _VIEWPORT_UNSET:
|
||||
context_kwargs["viewport"] = DEFAULT_VIEWPORT
|
||||
elif viewport is None:
|
||||
context_kwargs["no_viewport"] = True
|
||||
else:
|
||||
context_kwargs["viewport"] = viewport
|
||||
if color_scheme:
|
||||
context_kwargs["color_scheme"] = color_scheme
|
||||
context_kwargs.update(kwargs)
|
||||
|
||||
# Catch BaseException (not just Exception) so that asyncio.CancelledError
|
||||
# triggers browser cleanup — otherwise the underlying Chromium process
|
||||
# leaks when the awaiting task is cancelled.
|
||||
try:
|
||||
context = await browser.new_context(**context_kwargs)
|
||||
except BaseException:
|
||||
try:
|
||||
await browser.close()
|
||||
except BaseException:
|
||||
pass
|
||||
raise
|
||||
|
||||
# Patch close() to also close the browser (and its Playwright instance)
|
||||
_original_ctx_close = context.close
|
||||
|
||||
async def _close_context_with_cleanup() -> None:
|
||||
try:
|
||||
await _original_ctx_close()
|
||||
finally:
|
||||
await browser.close()
|
||||
|
||||
context.close = _close_context_with_cleanup
|
||||
|
||||
# Human-like behavioral patching (async variant)
|
||||
if humanize:
|
||||
from .human import patch_context_async
|
||||
from .human.config import resolve_config
|
||||
cfg = resolve_config(human_preset, human_config)
|
||||
patch_context_async(context, cfg)
|
||||
|
||||
return context
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Backend resolution
|
||||
# ---------------------------------------------------------------------------
|
||||
@@ -635,6 +760,37 @@ def _ensure_proxy_scheme(proxy_url: str) -> str:
|
||||
return proxy_url if "://" in proxy_url else f"http://{proxy_url}"
|
||||
|
||||
|
||||
def _assemble_socks_url(
|
||||
scheme: str,
|
||||
host: str,
|
||||
port: int | None,
|
||||
enc_user: str,
|
||||
enc_pass: str | None,
|
||||
path: str = "",
|
||||
params: str = "",
|
||||
query: str = "",
|
||||
fragment: str = "",
|
||||
) -> str:
|
||||
"""Build a SOCKS URL from already-percent-encoded credentials and host parts.
|
||||
|
||||
``enc_pass is None`` means no password (no colon in userinfo). Empty string
|
||||
means present-but-empty (colon preserved). This mirrors the distinction
|
||||
urlparse makes between ``user@host`` and ``user:@host``.
|
||||
"""
|
||||
if ":" in host: # IPv6 literal — re-add brackets
|
||||
host = f"[{host}]"
|
||||
if enc_pass is not None:
|
||||
userinfo = f"{enc_user}:{enc_pass}@"
|
||||
elif enc_user:
|
||||
userinfo = f"{enc_user}@"
|
||||
else:
|
||||
userinfo = ""
|
||||
netloc = f"{userinfo}{host}"
|
||||
if port is not None:
|
||||
netloc += f":{port}"
|
||||
return urlunparse((scheme, netloc, path, params, query, fragment))
|
||||
|
||||
|
||||
def _reconstruct_socks_url(proxy: ProxySettings) -> str:
|
||||
"""Reconstruct a SOCKS5 URL with inline credentials from a Playwright proxy dict."""
|
||||
server = proxy.get("server", "")
|
||||
@@ -643,16 +799,47 @@ def _reconstruct_socks_url(proxy: ProxySettings) -> str:
|
||||
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, "", "", ""))
|
||||
enc_user = quote(username, safe="")
|
||||
# Dict convention: empty/missing password → no colon.
|
||||
enc_pass = quote(password, safe="") if password else None
|
||||
return _assemble_socks_url(
|
||||
parsed.scheme, parsed.hostname or "", parsed.port,
|
||||
enc_user, enc_pass, parsed.path,
|
||||
)
|
||||
|
||||
|
||||
def _normalize_socks_string_url(url: str) -> str:
|
||||
"""Re-encode credentials in a SOCKS5 URL string so Chromium's parser doesn't
|
||||
truncate them at special chars like '='. Idempotent: pre-encoded input stays
|
||||
the same (decoded then re-encoded).
|
||||
|
||||
On unparseable input (invalid port, broken IPv6 literal, etc.) logs a
|
||||
warning and returns the original string — preserves pre-fix pass-through
|
||||
behavior so Chromium's own error handling kicks in.
|
||||
"""
|
||||
try:
|
||||
parsed = urlparse(url)
|
||||
# Accessing .port raises ValueError on invalid port strings.
|
||||
_ = parsed.port
|
||||
except ValueError as e:
|
||||
logger.warning("Malformed SOCKS5 proxy URL, passing through unchanged: %s", e)
|
||||
return url
|
||||
# Skip only if no credentials at all (username AND password both absent).
|
||||
# urlparse returns None for absent components, "" for present-but-empty.
|
||||
if parsed.username is None and parsed.password is None:
|
||||
return url
|
||||
enc_user = quote(unquote(parsed.username), safe="") if parsed.username else ""
|
||||
# Preserve the colon separator when password component is present, even if
|
||||
# empty, so `user:@host` stays `user:@host`.
|
||||
if parsed.password is not None:
|
||||
enc_pass = quote(unquote(parsed.password), safe="") if parsed.password else ""
|
||||
else:
|
||||
enc_pass = None
|
||||
return _assemble_socks_url(
|
||||
parsed.scheme, parsed.hostname or "", parsed.port,
|
||||
enc_user, enc_pass,
|
||||
parsed.path, parsed.params, parsed.query, parsed.fragment,
|
||||
)
|
||||
|
||||
|
||||
def _extract_proxy_url(proxy: str | ProxySettings | None) -> str | None:
|
||||
@@ -801,11 +988,14 @@ def build_args(
|
||||
|
||||
|
||||
def _parse_proxy_url(proxy: str) -> dict[str, Any]:
|
||||
"""Parse proxy URL, extracting credentials into separate Playwright fields.
|
||||
"""Parse HTTP(S) proxy URL, extracting credentials into separate Playwright fields.
|
||||
|
||||
Handles: http://user:pass@host:port -> {server: "http://host:port", username: "user", password: "pass"}
|
||||
Also handles: no credentials, URL-encoded special chars, socks5://, missing port,
|
||||
Also handles: no credentials, URL-encoded special chars, missing port,
|
||||
and bare proxy strings without a scheme (e.g. 'user:pass@host:port' -> treated as http).
|
||||
|
||||
SOCKS5 URLs are NOT handled here — they take a dedicated path via
|
||||
``_normalize_socks_string_url`` in ``_resolve_proxy_config``.
|
||||
"""
|
||||
# Bare format: "user:pass@host:port" — urlparse needs a scheme to extract credentials.
|
||||
normalized = proxy
|
||||
@@ -863,8 +1053,9 @@ def _resolve_proxy_config(
|
||||
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}"]
|
||||
# String URL — re-encode creds to work around Chromium parser truncating
|
||||
# passwords at '=' and other special chars (#157).
|
||||
return {}, [f"--proxy-server={_normalize_socks_string_url(proxy)}"]
|
||||
|
||||
# HTTP/HTTPS: use Playwright's proxy dict as before
|
||||
if isinstance(proxy, dict):
|
||||
|
||||
@@ -15,14 +15,14 @@ from ._version import __version__
|
||||
# CHROMIUM_VERSION is the latest across all platforms (for display/reference).
|
||||
# Use get_chromium_version() for the current platform's actual version.
|
||||
# ---------------------------------------------------------------------------
|
||||
CHROMIUM_VERSION = "146.0.7680.177.1"
|
||||
CHROMIUM_VERSION = "146.0.7680.177.3"
|
||||
|
||||
PLATFORM_CHROMIUM_VERSIONS: dict[str, str] = {
|
||||
"linux-x64": "146.0.7680.177.2",
|
||||
"linux-arm64": "146.0.7680.177.2",
|
||||
"linux-x64": "146.0.7680.177.3",
|
||||
"linux-arm64": "146.0.7680.177.3",
|
||||
"darwin-arm64": "145.0.7632.109.2",
|
||||
"darwin-x64": "145.0.7632.109.2",
|
||||
"windows-x64": "145.0.7632.159.7",
|
||||
"windows-x64": "146.0.7680.177.4",
|
||||
}
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
@@ -10,7 +10,7 @@ import math
|
||||
import random
|
||||
import time
|
||||
from dataclasses import dataclass, field
|
||||
from typing import Literal, Tuple
|
||||
from typing import Literal, Tuple, TypedDict
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Type alias
|
||||
@@ -20,6 +20,50 @@ Range = Tuple[float, float]
|
||||
HumanPreset = Literal["default", "careful"]
|
||||
|
||||
|
||||
class HumanConfigOverrides(TypedDict, total=False):
|
||||
typing_delay: float
|
||||
typing_delay_spread: float
|
||||
typing_pause_chance: float
|
||||
typing_pause_range: Range
|
||||
shift_down_delay: Range
|
||||
shift_up_delay: Range
|
||||
key_hold: Range
|
||||
field_switch_delay: Range
|
||||
mistype_chance: float
|
||||
mistype_delay_notice: Range
|
||||
mistype_delay_correct: Range
|
||||
mouse_steps_divisor: float
|
||||
mouse_min_steps: int
|
||||
mouse_max_steps: int
|
||||
mouse_wobble_max: float
|
||||
mouse_overshoot_chance: float
|
||||
mouse_overshoot_px: Range
|
||||
mouse_burst_size: Range
|
||||
mouse_burst_pause: Range
|
||||
click_aim_delay_input: Range
|
||||
click_aim_delay_button: Range
|
||||
click_hold_input: Range
|
||||
click_hold_button: Range
|
||||
click_input_x_range: Range
|
||||
idle_drift_px: float
|
||||
idle_pause_range: Range
|
||||
scroll_delta_base: Range
|
||||
scroll_delta_variance: float
|
||||
scroll_pause_fast: Range
|
||||
scroll_pause_slow: Range
|
||||
scroll_accel_steps: Range
|
||||
scroll_decel_steps: Range
|
||||
scroll_overshoot_chance: float
|
||||
scroll_overshoot_px: Range
|
||||
scroll_settle_delay: Range
|
||||
scroll_target_zone: Range
|
||||
scroll_pre_move_delay: Range
|
||||
initial_cursor_x: Range
|
||||
initial_cursor_y: Range
|
||||
idle_between_actions: bool
|
||||
idle_between_duration: Range
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Configuration dataclass
|
||||
# ---------------------------------------------------------------------------
|
||||
@@ -130,13 +174,13 @@ _PRESETS: dict[str, HumanConfig] = {
|
||||
|
||||
def resolve_config(
|
||||
preset: HumanPreset = "default",
|
||||
overrides: dict | None = None,
|
||||
overrides: HumanConfigOverrides | None = None,
|
||||
) -> HumanConfig:
|
||||
"""Resolve a preset name + optional overrides into a full HumanConfig.
|
||||
|
||||
Args:
|
||||
preset: 'default' or 'careful'.
|
||||
overrides: Dict of field names to override values.
|
||||
overrides: Typed mapping of HumanConfig field names to override values.
|
||||
|
||||
Returns:
|
||||
A new HumanConfig instance.
|
||||
|
||||
@@ -0,0 +1,79 @@
|
||||
# CloakBrowser on AWS Lambda — derived from the official CloakHQ image.
|
||||
#
|
||||
# `FROM cloakhq/cloakbrowser:<tag>` is an official distribution channel under
|
||||
# the CloakBrowser Binary License — pulling it isn't redistribution. We just
|
||||
# layer Lambda glue on top: the Lambda Runtime Interface Client (awslambdaric),
|
||||
# the Lambda Runtime Interface Emulator (for local `docker run` testing), the
|
||||
# dual-mode entrypoint, and the handler module.
|
||||
#
|
||||
# This directory is self-contained — copy/clone it anywhere and build from
|
||||
# inside it. No files outside this directory are referenced.
|
||||
#
|
||||
# ─── Lambda invocation (default CMD) ──────────────────────────────────────────
|
||||
# # From inside this directory:
|
||||
# docker buildx build --platform linux/arm64 -t cloakbrowser-lambda:arm64 --load .
|
||||
#
|
||||
# # Or from a parent dir, pointing at this directory as the build context:
|
||||
# docker buildx build --platform linux/arm64 \
|
||||
# -f path/to/aws_lambda/Dockerfile -t cloakbrowser-lambda:arm64 --load \
|
||||
# path/to/aws_lambda
|
||||
#
|
||||
# docker run --rm -p 9000:8080 cloakbrowser-lambda:arm64
|
||||
# curl -XPOST http://localhost:9000/2015-03-31/functions/function/invocations \
|
||||
# -d '{"url":"https://example.com"}'
|
||||
#
|
||||
# ─── Same as the canonical CloakHQ image (CMD overridden) ─────────────────────
|
||||
# docker run --rm -it cloakbrowser-lambda:arm64 python # REPL
|
||||
# docker run --rm cloakbrowser-lambda:arm64 python examples/basic.py # examples
|
||||
# docker run --rm -p 9222:9222 cloakbrowser-lambda:arm64 cloakserve --port=9222 # CDP server
|
||||
# docker run --rm cloakbrowser-lambda:arm64 cloaktest # stealth tests
|
||||
# docker run --rm -it cloakbrowser-lambda:arm64 node # JS wrapper
|
||||
# docker run --rm -it cloakbrowser-lambda:arm64 bash # shell
|
||||
#
|
||||
# Pin a specific tag (e.g. cloakhq/cloakbrowser:0.3.25) for reproducible builds;
|
||||
# `latest` floats with CloakHQ's release cadence.
|
||||
|
||||
FROM cloakhq/cloakbrowser:latest
|
||||
|
||||
# ─── Lambda Runtime Interface Client ──────────────────────────────────────────
|
||||
RUN pip install --no-cache-dir awslambdaric
|
||||
|
||||
# ─── Lambda Runtime Interface Emulator (local `docker run` testing) ───────────
|
||||
# Bundled into the image so users can hit the standard local-invoke endpoint
|
||||
# without mounting the RIE separately. TARGETARCH is provided by buildx.
|
||||
ARG TARGETARCH
|
||||
ADD https://github.com/aws/aws-lambda-runtime-interface-emulator/releases/latest/download/aws-lambda-rie-${TARGETARCH} \
|
||||
/usr/local/bin/aws-lambda-rie
|
||||
RUN chmod +x /usr/local/bin/aws-lambda-rie
|
||||
|
||||
# ─── Lambda glue ──────────────────────────────────────────────────────────────
|
||||
# Dual-mode entrypoint replaces the canonical bin/docker-entrypoint.sh: same
|
||||
# Xvfb startup, plus routing for `module.func` CMDs through awslambdaric.
|
||||
COPY lambda-entrypoint.sh /entrypoint.sh
|
||||
RUN chmod +x /entrypoint.sh
|
||||
|
||||
# Handler sits at /app (already on Python's import path in the canonical image,
|
||||
# WORKDIR=/app), imports cloakbrowser as a normal library.
|
||||
COPY lambda_handler.py /app/lambda_handler.py
|
||||
|
||||
# ─── Lambda non-root readability fix ──────────────────────────────────────────
|
||||
# The canonical image bakes the Chromium binary at /root/.cloakbrowser/ (root's
|
||||
# HOME at build time). Lambda runs the container as a non-root user that can't
|
||||
# read /root by default (mode 750). Make the whole binary tree world-readable
|
||||
# and traversable. Also restore the .welcome_shown marker the canonical image
|
||||
# rm's (Lambda's read-only runtime FS can't recreate it, so the welcome would
|
||||
# print to CloudWatch on every cold start otherwise).
|
||||
RUN touch /root/.cloakbrowser/.welcome_shown \
|
||||
&& chmod -R o+rX /root /root/.cloakbrowser
|
||||
|
||||
# ─── Lambda runtime env ───────────────────────────────────────────────────────
|
||||
# HOME=/tmp gives Chromium a writable scratch dir (Lambda only allows writes
|
||||
# under /tmp). CLOAKBROWSER_CACHE_DIR points at the baked binary location since
|
||||
# HOME=/tmp would otherwise make get_cache_dir() resolve to /tmp/.cloakbrowser
|
||||
# (empty). Auto-update is disabled because the runtime FS is read-only.
|
||||
ENV HOME=/tmp \
|
||||
CLOAKBROWSER_CACHE_DIR=/root/.cloakbrowser \
|
||||
CLOAKBROWSER_AUTO_UPDATE=false
|
||||
|
||||
ENTRYPOINT ["/entrypoint.sh"]
|
||||
CMD ["lambda_handler.handler"]
|
||||
@@ -0,0 +1,181 @@
|
||||
# CloakBrowser on AWS Lambda
|
||||
|
||||
Run stealth Chromium one-shot scrapes inside an AWS Lambda function (container image package type). The image derives directly from the official CloakHQ Docker Hub image (`cloakhq/cloakbrowser`) and adds Lambda runtime support on top — Lambda is an additional invocation surface, not a replacement. Every other surface from the canonical image (`python`, `cloakserve`, `cloaktest`, `node`, `bash`, examples) keeps working.
|
||||
|
||||
This document covers what the image is, how to build and locally test it, and the event/response contract. **It does not prescribe a deployment method** — push the resulting image to ECR and create the Lambda function however you prefer (AWS CLI, CDK, Terraform, SAM, console, etc.). Configuration tips for whichever tool you use are at the bottom.
|
||||
|
||||
## Files in this directory
|
||||
|
||||
| File | Purpose |
|
||||
|---|---|
|
||||
| `Dockerfile` | `FROM cloakhq/cloakbrowser` plus a thin Lambda layer. Self-contained — no files outside this directory are referenced. |
|
||||
| `lambda-entrypoint.sh` | Dual-mode entrypoint. Starts Xvfb, then routes `module.func` CMDs through `awslambdaric` (via the bundled `aws-lambda-rie` locally, or the AWS Runtime API in production), and execs everything else (`python`, `cloakserve`, `cloaktest`, `node`, `bash`) directly. |
|
||||
| `lambda_handler.py` | Default handler. Takes `{url, ...}`, returns `{title, url, html, screenshot_b64?}`. Always headed via Xvfb. |
|
||||
| `INSTRUCTIONS.md` | This file. |
|
||||
|
||||
The Lambda layer is ~30 lines on top of the official image — no apt list, no Node install, no JS-wrapper build, no Chromium download. The canonical CloakHQ image owns those.
|
||||
|
||||
This directory is **standalone**: copy or clone it anywhere (its own repo, a subdirectory of an existing project, a CI artifact bundle) and the build still works. It depends only on the upstream `cloakhq/cloakbrowser` image on Docker Hub and the `aws-lambda-rie` binary on GitHub Releases — both fetched at build time.
|
||||
|
||||
## Build
|
||||
|
||||
From inside this directory:
|
||||
|
||||
```bash
|
||||
docker buildx build --platform linux/arm64 -t cloakbrowser-lambda:arm64 --load .
|
||||
```
|
||||
|
||||
Or from anywhere, pointing at this directory as the build context:
|
||||
|
||||
```bash
|
||||
docker buildx build --platform linux/arm64 \
|
||||
-f path/to/aws_lambda/Dockerfile \
|
||||
-t cloakbrowser-lambda:arm64 --load \
|
||||
path/to/aws_lambda
|
||||
```
|
||||
|
||||
The build pulls `cloakhq/cloakbrowser:latest` from Docker Hub and adds the Lambda layer on top. Pin a specific tag (e.g. `cloakhq/cloakbrowser:0.3.25`) in the `FROM` line for reproducible builds; `latest` floats with the upstream release cadence.
|
||||
|
||||
For x86_64, switch `--platform linux/amd64` (slower on Apple Silicon under emulation).
|
||||
|
||||
## Local smoke test (no AWS account needed)
|
||||
|
||||
> **What's the RIE?** Lambda container images can't be run with a plain `docker run` — they expect to talk to AWS's Runtime API (the HTTP service Lambda exposes inside its sandbox to deliver events and collect responses). AWS publishes a small binary called the **Runtime Interface Emulator** that stands up a fake Runtime API on localhost so you can test the container exactly the way Lambda will invoke it, without deploying. We bake the RIE into the image, and the dual-mode entrypoint uses it automatically when `AWS_LAMBDA_RUNTIME_API` isn't set (i.e. you're not running in real Lambda).
|
||||
|
||||
The image bakes in `aws-lambda-rie`, so the standard Lambda local-invoke endpoint works without mounting anything:
|
||||
|
||||
```bash
|
||||
docker run --rm -p 9000:8080 cloakbrowser-lambda:arm64
|
||||
|
||||
# In another shell:
|
||||
curl -sS -XPOST "http://localhost:9000/2015-03-31/functions/function/invocations" \
|
||||
-d '{"url":"https://example.com"}'
|
||||
```
|
||||
|
||||
Other invocation surfaces stay intact (these match the canonical CloakHQ image):
|
||||
|
||||
```bash
|
||||
docker run --rm -it cloakbrowser-lambda:arm64 python # REPL
|
||||
docker run --rm cloakbrowser-lambda:arm64 python examples/basic.py # examples
|
||||
docker run --rm -p 9222:9222 cloakbrowser-lambda:arm64 cloakserve --port=9222 # CDP server
|
||||
docker run --rm cloakbrowser-lambda:arm64 cloaktest # stealth tests
|
||||
docker run --rm -it cloakbrowser-lambda:arm64 node # JS wrapper
|
||||
```
|
||||
|
||||
## Event schema
|
||||
|
||||
Only `url` is required. Everything else is optional.
|
||||
|
||||
### Launch options (forwarded to `cloakbrowser.launch_context_async`)
|
||||
|
||||
| Field | Type | Default |
|
||||
|---|---|---|
|
||||
| `url` | str | required |
|
||||
| `proxy` | str / dict | none — `http://user:pass@host:port` or a Playwright proxy dict |
|
||||
| `humanize` | bool | `false` — enable human-like mouse / keyboard / scroll |
|
||||
| `human_preset` | str | `"default"` or `"careful"` |
|
||||
| `geoip` | bool | `false` — auto timezone+locale from proxy IP |
|
||||
| `timezone` | str | none — IANA tz, e.g. `"America/New_York"` |
|
||||
| `locale` | str | none — BCP-47, e.g. `"en-US"` |
|
||||
| `viewport` | `{width,height}` | `1920x947` (cloakbrowser default) |
|
||||
| `user_agent` | str | none |
|
||||
| `extra_args` | `list[str]` | `[]` — extra Chromium CLI flags |
|
||||
|
||||
### Navigation
|
||||
|
||||
| Field | Type | Default |
|
||||
|---|---|---|
|
||||
| `wait_until` | str | `"domcontentloaded"` — `load` / `domcontentloaded` / `networkidle` / `commit` |
|
||||
| `goto_timeout_ms` | int | `30000` |
|
||||
|
||||
### Post-navigation waits
|
||||
|
||||
`smart_wait` is the default when no other wait is specified. It polls `document.documentElement.outerHTML.length` and returns when the size hasn't changed for `dom_stable_ms`. Robust for at-scale scraping because it ignores network activity (analytics beacons, long-poll, websockets) that doesn't mutate the DOM — `wait_until: "networkidle"` is unreliable on modern SPAs for exactly this reason.
|
||||
|
||||
| Field | Type | Default |
|
||||
|---|---|---|
|
||||
| `smart_wait` | bool | `true` if no other wait is set |
|
||||
| `dom_stable_ms` | int | `1500` |
|
||||
| `max_settle_ms` | int | `15000` |
|
||||
| `wait_for_load_state` | str | none — `load` / `domcontentloaded` / `networkidle` |
|
||||
| `wait_for_load_state_timeout_ms` | int | `30000` |
|
||||
| `wait_for_selector` | str | none — CSS or XPath |
|
||||
| `wait_for_selector_state` | str | `"visible"` — also `attached` / `detached` / `hidden` |
|
||||
| `wait_for_selector_timeout_ms` | int | `30000` |
|
||||
| `wait_for_function` | str | none — JS expression returning truthy when ready |
|
||||
| `wait_for_function_timeout_ms` | int | `30000` |
|
||||
| `wait_ms` | int | none — fixed pause |
|
||||
|
||||
### Capture
|
||||
|
||||
| Field | Type | Default |
|
||||
|---|---|---|
|
||||
| `screenshot` | bool | `true` |
|
||||
| `full_page_screenshot` | bool | `false` |
|
||||
|
||||
### Retry orchestration
|
||||
|
||||
The handler retries transient navigation failures inline within the same Lambda invocation. Two layers, both built-in:
|
||||
|
||||
- **Launch retries** — 3 attempts with 0.3 s + 0.6 s backoff. Recovers Xvfb / Chromium spawn races at cold start. Fast and cheap; not configurable.
|
||||
- **Strategy retries** — default 1 attempt, configurable via the `retries` event field. Recovers specific post-launch error classes by relaunching with adjusted Chromium args / page-load budgets.
|
||||
|
||||
| Field | Type | Default |
|
||||
|---|---|---|
|
||||
| `retries` | int | `1` — number of strategy-retry attempts after the first failure. Set to `0` to disable retry entirely. |
|
||||
|
||||
Strategies (priority order — first match wins):
|
||||
|
||||
| Error pattern | Strategy applied |
|
||||
|---|---|
|
||||
| `ERR_CERT_*` (any cert error) | `extra_args: ["--ignore-certificate-errors"]`, `goto_timeout_ms: 60000` |
|
||||
| `Timeout … exceeded` | `goto_timeout_ms: 90000`, `max_settle_ms: 25000` |
|
||||
| `ERR_CONNECTION_TIMED_OUT` | same as `Timeout … exceeded` |
|
||||
|
||||
Errors that are **not retried** (no anonymous scraper can recover): `ERR_NAME_NOT_RESOLVED`, `ERR_SSL_PROTOCOL_ERROR`, `ERR_CONNECTION_REFUSED`, `ERR_HTTP_RESPONSE_CODE_FAILURE`. These bail immediately.
|
||||
|
||||
On final failure, the raised `RuntimeError`'s message includes a `retry_history` block listing every attempt (strategy applied + error seen). Successful invocations return the standard response shape unchanged — no surprise fields when retries didn't fire.
|
||||
|
||||
### Response
|
||||
|
||||
```json
|
||||
{
|
||||
"title": "...",
|
||||
"url": "https://example.com/",
|
||||
"html": "<!DOCTYPE html>...",
|
||||
"screenshot_b64": "<base64 PNG>"
|
||||
}
|
||||
```
|
||||
|
||||
## Lambda-specific Chromium hardening (baked in, do not remove)
|
||||
|
||||
Two flags are forced on every launch by `lambda_handler.py`:
|
||||
|
||||
- `--disable-dev-shm-usage` — Lambda's `/dev/shm` is ~64 MB; Chromium's renderer crashes mid-paint without this.
|
||||
- `--no-zygote` — Lambda's restricted process model can't fork from Chromium's zygote process; without this the browser launches but child renderers fail to spawn and the first `page.new_page()` raises `TargetClosedError`.
|
||||
|
||||
## Function configuration recommendations
|
||||
|
||||
Whatever tool you use to create the Lambda function (CLI, CDK, Terraform, SAM, console), apply these settings:
|
||||
|
||||
| Setting | Value | Why |
|
||||
|---|---|---|
|
||||
| Package type | Image | Required — this is a container image, not a zip. |
|
||||
| Architecture | `arm64` | Roughly 20% cheaper than x86_64. Native build on Apple Silicon. Match the architecture you built for. |
|
||||
| Memory | 3008 MB | Memory in Lambda is tied to vCPU. Below ~1769 MB Chromium starts noticeably slower. |
|
||||
| Timeout | 120–180 s | Single-attempt scrapes complete in 3–15 s warm; under retry, a `Timeout`-class first failure (30 s default) plus a longer-budget retry (90 s) plus cleanup can total ~120-130 s. 180 s leaves headroom; below 120 s the function will time out before the retry completes. Cold-start init adds 5-10 s on top. |
|
||||
| Ephemeral storage (`/tmp`) | 1024 MB | Chromium profile dirs and screenshots can fill the 512 MB default. |
|
||||
| Networking | Default (no VPC) | Binary is baked in, no network needed at cold start. Add VPC + NAT only if your proxy egress requires it. |
|
||||
| Execution role | `AWSLambdaBasicExecutionRole` | Just CloudWatch Logs. Add more permissions only if your handler needs them. |
|
||||
|
||||
## Cold start
|
||||
|
||||
First invocation in a new container takes ~80–90 s (image extraction, Chromium binary mmap, JS engine warmup, no DNS/TLS caches). Subsequent warm invocations on the same container are 3–15 s.
|
||||
|
||||
For latency-sensitive use cases: provision concurrency, schedule a CloudWatch/EventBridge warmer ping, or accept the cold tail.
|
||||
|
||||
If you see empty/missing dynamic content on cold-start invocations, raise `max_settle_ms` in the event payload (e.g. `25000`) — the default `15000` is tuned for warm runs.
|
||||
|
||||
## License
|
||||
|
||||
The patched Chromium binary inside the upstream `cloakhq/cloakbrowser` image is governed by the **CloakBrowser Binary License** (published at https://github.com/CloakHQ/CloakBrowser/blob/main/BINARY-LICENSE.md). Internal organizational use (private ECR, your own scraping pipelines, your own business) is free. Exposing this Lambda as a paid API to third-party customers — i.e. browser-as-a-service — requires an OEM/SaaS license from CloakHQ (`cloakhq@pm.me`). Do not push the resulting image to a public registry; that would be redistribution and is prohibited.
|
||||
@@ -0,0 +1,52 @@
|
||||
#!/bin/sh
|
||||
# Dual-mode entrypoint for the CloakBrowser Lambda image.
|
||||
#
|
||||
# 1. Always start Xvfb on :99 (same as the canonical bin/docker-entrypoint.sh)
|
||||
# so headed Chromium works no matter how the container is invoked.
|
||||
# 2. Detect whether the CMD looks like a Lambda handler (a single
|
||||
# `module.func`-shaped argument). If yes, route through the Lambda runtime
|
||||
# client (using the bundled aws-lambda-rie locally, or talking to the real
|
||||
# Lambda Runtime API when AWS_LAMBDA_RUNTIME_API is set in production).
|
||||
# 3. Otherwise exec the CMD directly — preserving the canonical Dockerfile's
|
||||
# interaction surface (`python`, `cloakserve`, `cloaktest`, `node`, `bash`,
|
||||
# `python examples/basic.py`, etc.).
|
||||
set -e
|
||||
|
||||
mkdir -p /tmp/.X11-unix
|
||||
chmod 1777 /tmp/.X11-unix 2>/dev/null || true
|
||||
|
||||
# Clean any stale Xvfb state. If a previous Xvfb died and left its lock file
|
||||
# behind (we observed this in cold-start storms), a new Xvfb refuses to start
|
||||
# with "Server is already active for display 99". Removing both files makes
|
||||
# Xvfb start cleanly every time.
|
||||
rm -f /tmp/.X99-lock /tmp/.X11-unix/X99
|
||||
|
||||
Xvfb :99 -screen 0 1920x1080x24 -nolisten tcp >/tmp/Xvfb.log 2>&1 &
|
||||
|
||||
# Wait for the X11 socket to appear AND for Xvfb to be ready to serve. The
|
||||
# socket file appears at bind(), but listen() and the first accept() come
|
||||
# slightly later — under cold-start CPU contention this gap matters.
|
||||
i=0
|
||||
while [ ! -e /tmp/.X11-unix/X99 ] && [ "$i" -lt 200 ]; do
|
||||
i=$((i + 1))
|
||||
sleep 0.05
|
||||
done
|
||||
# Small buffer after the socket appears so Xvfb has a moment to call listen()
|
||||
# and start accepting clients. Cheap insurance against the bind/listen gap.
|
||||
sleep 0.2
|
||||
|
||||
# Lambda handler shape: exactly one arg, dotted identifier (no spaces, no slashes,
|
||||
# no leading dot). `python`, `cloakserve`, `cloaktest`, `bash`, `node` all fail
|
||||
# this test and pass through to plain exec.
|
||||
if [ $# -eq 1 ] && \
|
||||
echo "$1" | grep -qE '^[a-zA-Z_][a-zA-Z0-9_]*(\.[a-zA-Z_][a-zA-Z0-9_]*)+$'; then
|
||||
if [ -z "${AWS_LAMBDA_RUNTIME_API}" ]; then
|
||||
# Local invocation via bundled RIE.
|
||||
exec /usr/local/bin/aws-lambda-rie /usr/local/bin/python -m awslambdaric "$@"
|
||||
else
|
||||
# Real Lambda — runtime API endpoint already provided by the platform.
|
||||
exec /usr/local/bin/python -m awslambdaric "$@"
|
||||
fi
|
||||
fi
|
||||
|
||||
exec "$@"
|
||||
@@ -0,0 +1,336 @@
|
||||
"""AWS Lambda handler for one-off stealth-browser invocations.
|
||||
|
||||
Always runs **headed** via the Xvfb display started by `lambda-entrypoint.sh`.
|
||||
|
||||
Event schema (all fields except `url` are optional):
|
||||
|
||||
Launch options (passed to cloakbrowser.launch_context_async):
|
||||
url str required, the page to scrape
|
||||
proxy str|dict http://user:pass@host:port or Playwright proxy dict
|
||||
humanize bool False — enable human-like mouse/keyboard/scroll
|
||||
human_preset str "default" | "careful"
|
||||
geoip bool False — auto timezone+locale from proxy IP
|
||||
timezone str IANA tz, e.g. "America/New_York"
|
||||
locale str BCP-47, e.g. "en-US"
|
||||
viewport {width,height} defaults to 1920x947 (cloakbrowser DEFAULT_VIEWPORT)
|
||||
user_agent str custom UA (rare — cloakbrowser sets one already)
|
||||
extra_args list[str] additional Chromium CLI flags
|
||||
|
||||
Navigation options (passed to page.goto):
|
||||
wait_until str "load"|"domcontentloaded"|"networkidle"|"commit"
|
||||
default "domcontentloaded"
|
||||
goto_timeout_ms int 30000
|
||||
|
||||
Post-navigation waits (run in this order if specified):
|
||||
smart_wait bool ON by default if no other wait is set.
|
||||
Polls document.outerHTML.length and bails when it
|
||||
hasn't changed for `dom_stable_ms`. Handles lazy
|
||||
hydration, async chunks, and lazy images, and is
|
||||
immune to analytics beacons / long-poll that keep
|
||||
the network busy without mutating the DOM.
|
||||
dom_stable_ms int 1500 — how long DOM must be quiet
|
||||
max_settle_ms int 15000 — hard cap on smart_wait
|
||||
wait_for_load_state str "load"|"domcontentloaded"|"networkidle"
|
||||
wait_for_load_state_timeout_ms int 30000
|
||||
wait_for_selector str CSS or XPath selector
|
||||
wait_for_selector_state str "attached"|"detached"|"visible"|"hidden", default "visible"
|
||||
wait_for_selector_timeout_ms int 30000
|
||||
wait_for_function str JS expression that returns truthy when ready
|
||||
wait_for_function_timeout_ms int 30000
|
||||
wait_ms int fixed pause in ms (page.wait_for_timeout)
|
||||
|
||||
Capture options:
|
||||
screenshot bool True
|
||||
full_page_screenshot bool False — capture entire scrollable page
|
||||
|
||||
Retry orchestration:
|
||||
retries int default 1. Number of retry attempts after the first
|
||||
failure. Set to 0 to disable retries entirely (the
|
||||
handler will fail fast on the first error).
|
||||
Retried errors:
|
||||
ERR_CERT_* -> retry with --ignore-certificate-errors
|
||||
Timeout exceeded -> retry with goto_timeout_ms=90000, max_settle_ms=25000
|
||||
ERR_CONNECTION_TIMED_OUT -> same as Timeout
|
||||
Not retried (unrecoverable): ERR_NAME_NOT_RESOLVED,
|
||||
ERR_SSL_PROTOCOL_ERROR, generic ERR_CONNECTION_REFUSED.
|
||||
On final failure, the error message includes a
|
||||
retry_history block with strategy + error per attempt.
|
||||
|
||||
Returns:
|
||||
{"title": ..., "url": ..., "html": ..., "screenshot_b64"?: ...}
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import base64
|
||||
import json
|
||||
import logging
|
||||
import subprocess
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
from cloakbrowser import launch_context_async
|
||||
|
||||
logger = logging.getLogger("cloakbrowser.lambda")
|
||||
logger.setLevel(logging.INFO)
|
||||
|
||||
|
||||
def _diag_snapshot() -> str:
|
||||
"""Capture Xvfb status, Xvfb log, X11 socket state, and env for error reports."""
|
||||
import os
|
||||
parts = []
|
||||
try:
|
||||
r = subprocess.run(["pgrep", "-fa", "Xvfb"], capture_output=True, text=True)
|
||||
parts.append(f"pgrep Xvfb: rc={r.returncode} stdout={r.stdout.strip()!r}")
|
||||
except Exception as e:
|
||||
parts.append(f"pgrep failed: {e}")
|
||||
try:
|
||||
r = subprocess.run(["ls", "-la", "/tmp/.X11-unix"], capture_output=True, text=True)
|
||||
parts.append(f"ls /tmp/.X11-unix:\n{r.stdout}{r.stderr}")
|
||||
except Exception as e:
|
||||
parts.append(f"ls /tmp/.X11-unix failed: {e}")
|
||||
try:
|
||||
log = Path("/tmp/Xvfb.log").read_text()
|
||||
parts.append(f"/tmp/Xvfb.log:\n{log}")
|
||||
except Exception as e:
|
||||
parts.append(f"Xvfb log unreadable: {e}")
|
||||
parts.append(f"env: DISPLAY={os.environ.get('DISPLAY')!r} HOME={os.environ.get('HOME')!r}")
|
||||
return "\n".join(parts)
|
||||
|
||||
|
||||
def handler(event: dict, context: Any) -> dict:
|
||||
return asyncio.run(_run(event))
|
||||
|
||||
|
||||
def _build_launch_kwargs(event: dict) -> dict:
|
||||
"""Translate the event dict into kwargs for launch_context_async.
|
||||
|
||||
Only includes keys explicitly set in the event so cloakbrowser's defaults
|
||||
(DEFAULT_VIEWPORT etc.) kick in when fields are absent — passing
|
||||
viewport=None would *disable* viewport emulation, which we don't want.
|
||||
"""
|
||||
kwargs: dict = {
|
||||
"headless": False, # always headed via Xvfb
|
||||
"args": [
|
||||
# Lambda /dev/shm is ~64 MB — Chromium crashes mid-render without this.
|
||||
"--disable-dev-shm-usage",
|
||||
# Lambda's restricted process model can't fork from Chromium's zygote
|
||||
# — without this, child renderer processes fail to spawn.
|
||||
"--no-zygote",
|
||||
*event.get("extra_args", []),
|
||||
],
|
||||
}
|
||||
for key in ("proxy", "humanize", "human_preset", "geoip",
|
||||
"timezone", "locale", "viewport", "user_agent"):
|
||||
if key in event:
|
||||
kwargs[key] = event[key]
|
||||
return kwargs
|
||||
|
||||
|
||||
async def _smart_wait(page, dom_stable_ms: int = 1500, max_settle_ms: int = 15000) -> None:
|
||||
"""Wait until the document HTML hasn't changed for `dom_stable_ms`.
|
||||
|
||||
Generic stopping condition for at-scale scraping when you can't tune
|
||||
selectors per site. More robust than `networkidle` because it ignores
|
||||
network activity that doesn't mutate the DOM (analytics beacons,
|
||||
long-poll, websockets, web vitals streams).
|
||||
"""
|
||||
js = f"""
|
||||
(() => {{
|
||||
if (!window.__cb_settle) {{
|
||||
window.__cb_settle = {{ len: -1, since: Date.now() }};
|
||||
}}
|
||||
const cur = document.documentElement.outerHTML.length;
|
||||
const s = window.__cb_settle;
|
||||
if (cur !== s.len) {{
|
||||
s.len = cur;
|
||||
s.since = Date.now();
|
||||
return false;
|
||||
}}
|
||||
return (Date.now() - s.since) >= {int(dom_stable_ms)};
|
||||
}})()
|
||||
"""
|
||||
try:
|
||||
await page.wait_for_function(js, timeout=max_settle_ms, polling=200)
|
||||
except Exception:
|
||||
# Hit max_settle_ms cap — return what we have rather than fail the whole invoke
|
||||
logger.warning("smart_wait hit max_settle_ms=%d cap", max_settle_ms)
|
||||
|
||||
|
||||
_EXPLICIT_WAIT_KEYS = (
|
||||
"wait_for_load_state", "wait_for_selector", "wait_for_function", "wait_ms",
|
||||
)
|
||||
|
||||
|
||||
async def _post_nav_waits(page, event: dict) -> None:
|
||||
"""Run waits in priority order. smart_wait is the default unless the
|
||||
caller asked for a more specific stopping condition."""
|
||||
explicit = any(k in event for k in _EXPLICIT_WAIT_KEYS)
|
||||
if event.get("smart_wait", not explicit):
|
||||
await _smart_wait(
|
||||
page,
|
||||
dom_stable_ms=event.get("dom_stable_ms", 1500),
|
||||
max_settle_ms=event.get("max_settle_ms", 15000),
|
||||
)
|
||||
if "wait_for_load_state" in event:
|
||||
await page.wait_for_load_state(
|
||||
event["wait_for_load_state"],
|
||||
timeout=event.get("wait_for_load_state_timeout_ms", 30000),
|
||||
)
|
||||
if "wait_for_selector" in event:
|
||||
await page.wait_for_selector(
|
||||
event["wait_for_selector"],
|
||||
state=event.get("wait_for_selector_state", "visible"),
|
||||
timeout=event.get("wait_for_selector_timeout_ms", 30000),
|
||||
)
|
||||
if "wait_for_function" in event:
|
||||
await page.wait_for_function(
|
||||
event["wait_for_function"],
|
||||
timeout=event.get("wait_for_function_timeout_ms", 30000),
|
||||
)
|
||||
if "wait_ms" in event:
|
||||
await page.wait_for_timeout(event["wait_ms"])
|
||||
|
||||
|
||||
async def _launch_with_retry(event: dict, attempts: int = 3, backoff_s: float = 0.3):
|
||||
"""Retry launch_context_async up to `attempts` times with linear backoff.
|
||||
|
||||
Lambda cold-start storms occasionally race Xvfb readiness or hit transient
|
||||
Chromium spawn failures — both surface as "Target page, context or browser
|
||||
has been closed" at launch. The failure is fast (~0.5s) so retries are
|
||||
cheap, and a retry on a now-warm container almost always succeeds.
|
||||
|
||||
Pairs with the lock-cleanup + socket-poll in lambda-entrypoint.sh: the
|
||||
entrypoint catches the common case at container init; this catches the
|
||||
residual race when the first invocation hits before Xvfb is fully ready.
|
||||
"""
|
||||
last_err: Exception | None = None
|
||||
for i in range(attempts):
|
||||
try:
|
||||
return await launch_context_async(**_build_launch_kwargs(event))
|
||||
except Exception as e:
|
||||
last_err = e
|
||||
logger.warning("launch attempt %d/%d failed: %s",
|
||||
i + 1, attempts, str(e)[:200])
|
||||
if i + 1 < attempts:
|
||||
await asyncio.sleep(backoff_s * (i + 1)) # 0.3s, 0.6s
|
||||
raise last_err # type: ignore[misc]
|
||||
|
||||
|
||||
def _classify_error(err: Exception) -> dict | None:
|
||||
"""Map a Playwright error to a retry-strategy override dict, or None
|
||||
if the error is unrecoverable.
|
||||
|
||||
Match on str(e) because Playwright errors carry their codes inside the
|
||||
message (Error.__str__ includes ERR_CERT_AUTHORITY_INVALID etc.); there
|
||||
is no stable structured `.error_code` attribute to rely on.
|
||||
|
||||
Strategies (priority order — first match wins):
|
||||
ERR_CERT_* -> --ignore-certificate-errors + 60s goto budget
|
||||
Timeout exceeded -> 90s goto budget + 25s smart_wait cap
|
||||
ERR_CONNECTION_TIMED_OUT -> same as Timeout
|
||||
Returns None for unrecoverable site issues (DNS, SSL, refused, HTTP 4xx/5xx).
|
||||
"""
|
||||
msg = str(err)
|
||||
if "ERR_CERT" in msg:
|
||||
return {
|
||||
"extra_args": ["--ignore-certificate-errors"],
|
||||
"goto_timeout_ms": 60000,
|
||||
}
|
||||
if ("Timeout" in msg and "exceeded" in msg) or "ERR_CONNECTION_TIMED_OUT" in msg:
|
||||
return {
|
||||
"goto_timeout_ms": 90000,
|
||||
"max_settle_ms": 25000,
|
||||
}
|
||||
return None
|
||||
|
||||
|
||||
async def _attempt_scrape(url: str, event: dict) -> dict:
|
||||
"""One self-contained scrape attempt: launch, navigate, wait, capture, close.
|
||||
|
||||
Extracted from `_run` so the retry loop can call it repeatedly with an
|
||||
overridden event dict. Each attempt relaunches the browser — uniform
|
||||
behavior across strategies (the cert-bypass strategy *requires* a relaunch
|
||||
because `--ignore-certificate-errors` is a Chromium CLI arg, not a per-
|
||||
context switch), and the ~3-5s relaunch cost is fine on the slow path.
|
||||
"""
|
||||
ctx = await _launch_with_retry(event)
|
||||
try:
|
||||
page = await ctx.new_page()
|
||||
await page.goto(
|
||||
url,
|
||||
wait_until=event.get("wait_until", "domcontentloaded"),
|
||||
timeout=event.get("goto_timeout_ms", 30000),
|
||||
)
|
||||
|
||||
await _post_nav_waits(page, event)
|
||||
|
||||
result: dict = {
|
||||
"title": await page.title(),
|
||||
"url": page.url,
|
||||
"html": await page.content(),
|
||||
}
|
||||
|
||||
if event.get("screenshot", True):
|
||||
png = await page.screenshot(
|
||||
full_page=event.get("full_page_screenshot", False),
|
||||
)
|
||||
result["screenshot_b64"] = base64.b64encode(png).decode()
|
||||
|
||||
return result
|
||||
finally:
|
||||
try:
|
||||
await ctx.close()
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
|
||||
def _raise_with_history(err: Exception, history: list[dict]) -> None:
|
||||
"""Surface a final failure with a retry_history block embedded in the
|
||||
error message, so callers see what was tried before bailing."""
|
||||
diag = _diag_snapshot()
|
||||
if history:
|
||||
diag = "retry_history: " + json.dumps(history, default=str) + "\n\n" + diag
|
||||
logger.error("scrape failed (after %d retries): %s\nDIAG:\n%s",
|
||||
len(history), err, diag)
|
||||
raise RuntimeError(f"scrape failed: {err}\n--- DIAG ---\n{diag}") from err
|
||||
|
||||
|
||||
async def _run(event: dict) -> dict:
|
||||
"""Top-level scrape with strategy-based retry orchestration.
|
||||
|
||||
First attempt uses the event verbatim. If it fails with a classifiable
|
||||
error (cert / timeout), retry with that strategy's overrides merged into
|
||||
the event. `retries` bounds the number of strategy retries (default 1;
|
||||
set to 0 to disable retry entirely).
|
||||
"""
|
||||
url = event["url"]
|
||||
retries_left = max(0, int(event.get("retries", 1)))
|
||||
history: list[dict] = []
|
||||
current_event = event
|
||||
|
||||
while True:
|
||||
try:
|
||||
return await _attempt_scrape(url, current_event)
|
||||
except Exception as e:
|
||||
if retries_left <= 0:
|
||||
_raise_with_history(e, history)
|
||||
strategy = _classify_error(e)
|
||||
if strategy is None:
|
||||
_raise_with_history(e, history)
|
||||
history.append({
|
||||
"attempt": len(history) + 1,
|
||||
"error": str(e)[:300],
|
||||
"strategy": strategy,
|
||||
})
|
||||
logger.warning("attempt %d failed (%s); retrying with strategy=%s",
|
||||
len(history), str(e)[:120], strategy)
|
||||
merged_args = list(current_event.get("extra_args", [])) + list(strategy.get("extra_args", []))
|
||||
current_event = {**current_event, **strategy, "extra_args": merged_args}
|
||||
retries_left -= 1
|
||||
# No backoff: strategy overrides change goto budget directly;
|
||||
# the prior failure was either fast (cert reject) or already
|
||||
# waited its full timeout. Container is warm.
|
||||
|
||||
|
||||
+1
-1
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "cloakbrowser",
|
||||
"version": "0.3.24",
|
||||
"version": "0.3.26",
|
||||
"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",
|
||||
|
||||
+4
-4
@@ -27,14 +27,14 @@ export { WRAPPER_VERSION };
|
||||
// CHROMIUM_VERSION is the latest across all platforms (for display/reference).
|
||||
// Use getChromiumVersion() for the current platform's actual version.
|
||||
// ---------------------------------------------------------------------------
|
||||
export const CHROMIUM_VERSION = "146.0.7680.177.1";
|
||||
export const CHROMIUM_VERSION = "146.0.7680.177.3";
|
||||
|
||||
export const PLATFORM_CHROMIUM_VERSIONS: Record<string, string> = {
|
||||
"linux-x64": "146.0.7680.177.2",
|
||||
"linux-arm64": "146.0.7680.177.2",
|
||||
"linux-x64": "146.0.7680.177.3",
|
||||
"linux-arm64": "146.0.7680.177.3",
|
||||
"darwin-arm64": "145.0.7632.109.2",
|
||||
"darwin-x64": "145.0.7632.109.2",
|
||||
"windows-x64": "145.0.7632.159.7",
|
||||
"windows-x64": "146.0.7680.177.4",
|
||||
};
|
||||
|
||||
// ---------------------------------------------------------------------------
|
||||
|
||||
+36
-7
@@ -3,7 +3,7 @@
|
||||
* Mirrors Python cloakbrowser/browser.py.
|
||||
*/
|
||||
|
||||
import type { Browser, BrowserContext } from "playwright-core";
|
||||
import type { Browser, BrowserContext, BrowserContextOptions } from "playwright-core";
|
||||
import type { LaunchOptions, LaunchContextOptions, LaunchPersistentContextOptions } from "./types.js";
|
||||
import { DEFAULT_VIEWPORT, IGNORE_DEFAULT_ARGS } from "./config.js";
|
||||
import { buildArgs } from "./args.js";
|
||||
@@ -21,6 +21,29 @@ export function resolveTimezone<T extends { timezone?: string; timezoneId?: stri
|
||||
return options;
|
||||
}
|
||||
|
||||
/**
|
||||
* Strip `locale` and `timezoneId` from user-provided contextOptions — both route
|
||||
* through detectable CDP emulation. The wrapper's top-level `locale`/`timezone`
|
||||
* fields use binary flags instead (undetectable). Warn so users notice.
|
||||
*/
|
||||
function filterStealthCtxOptions(ctx?: BrowserContextOptions): Partial<BrowserContextOptions> {
|
||||
if (!ctx) return {};
|
||||
const { locale, timezoneId, ...rest } = ctx;
|
||||
if (locale !== undefined) {
|
||||
console.warn(
|
||||
"[cloakbrowser] contextOptions.locale ignored — use top-level `locale` " +
|
||||
"instead (routes through binary flag, avoids detectable CDP emulation)."
|
||||
);
|
||||
}
|
||||
if (timezoneId !== undefined) {
|
||||
console.warn(
|
||||
"[cloakbrowser] contextOptions.timezoneId ignored — use top-level `timezone` " +
|
||||
"instead (routes through binary flag, avoids detectable CDP emulation)."
|
||||
);
|
||||
}
|
||||
return rest;
|
||||
}
|
||||
|
||||
/**
|
||||
* Launch stealth Chromium browser via Playwright.
|
||||
*
|
||||
@@ -60,8 +83,8 @@ export async function launch(options: LaunchOptions = {}): Promise<Browser> {
|
||||
const { patchBrowser } = await import('./human/index.js');
|
||||
const { resolveConfig } = await import('./human/config.js');
|
||||
const cfg = resolveConfig(
|
||||
(options.humanPreset as any) ?? 'default',
|
||||
options.humanConfig as any,
|
||||
options.humanPreset ?? 'default',
|
||||
options.humanConfig,
|
||||
);
|
||||
patchBrowser(browser, cfg);
|
||||
}
|
||||
@@ -104,6 +127,9 @@ export async function launchContext(
|
||||
let context: BrowserContext;
|
||||
try {
|
||||
context = await browser.newContext({
|
||||
// contextOptions first — explicit wrapper fields below override it.
|
||||
// filterStealthCtxOptions strips locale/timezoneId to prevent CDP detection.
|
||||
...filterStealthCtxOptions(options.contextOptions),
|
||||
...(options.userAgent ? { userAgent: options.userAgent } : {}),
|
||||
viewport: options.viewport === undefined ? DEFAULT_VIEWPORT : options.viewport,
|
||||
...(options.colorScheme ? { colorScheme: options.colorScheme } : {}),
|
||||
@@ -125,8 +151,8 @@ export async function launchContext(
|
||||
const { patchContext } = await import('./human/index.js');
|
||||
const { resolveConfig } = await import('./human/config.js');
|
||||
const cfg = resolveConfig(
|
||||
(options.humanPreset as any) ?? 'default',
|
||||
options.humanConfig as any,
|
||||
options.humanPreset ?? 'default',
|
||||
options.humanConfig,
|
||||
);
|
||||
patchContext(context, cfg);
|
||||
}
|
||||
@@ -178,6 +204,9 @@ export async function launchPersistentContext(
|
||||
args,
|
||||
ignoreDefaultArgs: IGNORE_DEFAULT_ARGS,
|
||||
...(proxyOption ? { proxy: proxyOption } : {}),
|
||||
// contextOptions before explicit wrapper fields so explicit wins.
|
||||
// filterStealthCtxOptions strips locale/timezoneId to prevent CDP detection.
|
||||
...filterStealthCtxOptions(options.contextOptions),
|
||||
...(options.userAgent ? { userAgent: options.userAgent } : {}),
|
||||
viewport: options.viewport === undefined ? DEFAULT_VIEWPORT : options.viewport,
|
||||
...(options.colorScheme ? { colorScheme: options.colorScheme } : {}),
|
||||
@@ -189,8 +218,8 @@ export async function launchPersistentContext(
|
||||
const { patchContext } = await import('./human/index.js');
|
||||
const { resolveConfig } = await import('./human/config.js');
|
||||
const cfg = resolveConfig(
|
||||
(options.humanPreset as any) ?? 'default',
|
||||
options.humanConfig as any,
|
||||
options.humanPreset ?? 'default',
|
||||
options.humanConfig,
|
||||
);
|
||||
patchContext(context, cfg);
|
||||
}
|
||||
|
||||
+91
-1
@@ -43,6 +43,40 @@ export function isSocksProxy(proxy: string | ProxyDict | undefined | null): bool
|
||||
return /^socks5h?:\/\//i.test(url);
|
||||
}
|
||||
|
||||
/**
|
||||
* Build a SOCKS URL from already-percent-encoded credentials and a host suffix.
|
||||
*
|
||||
* `encPass === null` means no password (no colon in userinfo). Empty string
|
||||
* means present-but-empty (colon preserved).
|
||||
*/
|
||||
function assembleSocksUrl(
|
||||
scheme: string,
|
||||
encUser: string,
|
||||
encPass: string | null,
|
||||
hostAndRest: string,
|
||||
): string {
|
||||
let userinfo: string;
|
||||
if (encPass !== null) {
|
||||
userinfo = `${encUser}:${encPass}@`;
|
||||
} else if (encUser) {
|
||||
userinfo = `${encUser}@`;
|
||||
} else {
|
||||
userinfo = "";
|
||||
}
|
||||
return `${scheme}://${userinfo}${hostAndRest}`;
|
||||
}
|
||||
|
||||
/**
|
||||
* Lenient percent-decode that handles malformed escapes gracefully, matching
|
||||
* Python's ``urllib.parse.unquote``: valid ``%XX`` sequences are decoded,
|
||||
* bare ``%`` not followed by two hex digits is left as a literal ``%``.
|
||||
*/
|
||||
function lenientDecodeURIComponent(s: string): string {
|
||||
return s.replace(/%([0-9A-Fa-f]{2})|%/g, (match, hex) =>
|
||||
hex ? String.fromCharCode(parseInt(hex, 16)) : "%",
|
||||
);
|
||||
}
|
||||
|
||||
/**
|
||||
* Reconstruct a SOCKS5 URL with inline credentials from a proxy dict.
|
||||
*/
|
||||
@@ -55,6 +89,60 @@ export function reconstructSocksUrl(proxy: ProxyDict): string {
|
||||
return url.href.replace(/\/$/, "");
|
||||
}
|
||||
|
||||
/**
|
||||
* Re-encode credentials in a SOCKS5 URL string so Chromium's parser doesn't
|
||||
* truncate them at special chars like '='. Idempotent: pre-encoded input stays
|
||||
* the same (decoded then re-encoded).
|
||||
*
|
||||
* Parsing is done manually rather than via `new URL` + setters, because WHATWG
|
||||
* URL's username/password setters re-encode `%` on assignment, causing
|
||||
* double-encoding when we round-trip decode-then-encode.
|
||||
*
|
||||
* On any unexpected failure, logs a warning and returns the original string
|
||||
* so Chromium's own error handling can surface the real problem.
|
||||
*/
|
||||
export function normalizeSocksStringUrl(urlStr: string): string {
|
||||
// Split userinfo from host at the LAST '@' (RFC 3986), so a raw '@' inside
|
||||
// a password like `socks5://user:p@ss@host:1080` parses correctly. Matches
|
||||
// Python urlparse's rpartition('@') behavior.
|
||||
const schemeMatch = urlStr.match(/^([a-z][a-z0-9+\-.]*):\/\/(.*)$/i);
|
||||
if (!schemeMatch) return urlStr;
|
||||
const [, scheme, rest] = schemeMatch;
|
||||
const hostStart = rest.search(/[/?#]/);
|
||||
const authority = hostStart === -1 ? rest : rest.slice(0, hostStart);
|
||||
const suffix = hostStart === -1 ? "" : rest.slice(hostStart);
|
||||
const atIdx = authority.lastIndexOf("@");
|
||||
if (atIdx === -1) return urlStr; // no creds
|
||||
const userinfo = authority.slice(0, atIdx);
|
||||
const hostPart = authority.slice(atIdx + 1);
|
||||
// Validate port (matches Python's urlparse().port ValueError guard).
|
||||
// Extract port after last ':' — but skip IPv6 brackets (e.g. [::1]:1080).
|
||||
const bracketEnd = hostPart.lastIndexOf("]");
|
||||
const portColonIdx = hostPart.indexOf(":", Math.max(bracketEnd, 0));
|
||||
if (portColonIdx !== -1) {
|
||||
const portStr = hostPart.slice(portColonIdx + 1);
|
||||
if (portStr && !/^\d+$/.test(portStr)) {
|
||||
console.warn(`[cloakbrowser] Malformed SOCKS5 proxy URL, passing through unchanged: invalid port`);
|
||||
return urlStr;
|
||||
}
|
||||
}
|
||||
const hostAndRest = hostPart + suffix;
|
||||
const colonIdx = userinfo.indexOf(":");
|
||||
const rawUserEnc = colonIdx === -1 ? userinfo : userinfo.slice(0, colonIdx);
|
||||
const hasPassword = colonIdx !== -1;
|
||||
const rawPassEnc = hasPassword ? userinfo.slice(colonIdx + 1) : "";
|
||||
try {
|
||||
const encUser = rawUserEnc ? encodeURIComponent(lenientDecodeURIComponent(rawUserEnc)) : "";
|
||||
const encPass = hasPassword
|
||||
? (rawPassEnc ? encodeURIComponent(lenientDecodeURIComponent(rawPassEnc)) : "")
|
||||
: null;
|
||||
return assembleSocksUrl(scheme, encUser, encPass, hostAndRest);
|
||||
} catch (e) {
|
||||
console.warn(`[cloakbrowser] Could not normalize SOCKS5 proxy URL, passing through unchanged: ${(e as Error).message}`);
|
||||
return urlStr;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Resolve proxy into Playwright option and/or Chrome args.
|
||||
*
|
||||
@@ -67,7 +155,9 @@ export function resolveProxyConfig(proxy: string | ProxyDict | undefined): Proxy
|
||||
if (isSocksProxy(proxy)) {
|
||||
// SOCKS5: bypass Playwright, pass directly to Chrome via --proxy-server.
|
||||
if (typeof proxy === "string") {
|
||||
return { proxyArgs: [`--proxy-server=${proxy}`] };
|
||||
// Re-encode creds to work around Chromium parser truncating passwords
|
||||
// at '=' and other special chars (#157).
|
||||
return { proxyArgs: [`--proxy-server=${normalizeSocksStringUrl(proxy)}`] };
|
||||
}
|
||||
const socksUrl = reconstructSocksUrl(proxy);
|
||||
const args = [`--proxy-server=${socksUrl}`];
|
||||
|
||||
+2
-2
@@ -94,8 +94,8 @@ export async function launch(options: LaunchOptions = {}): Promise<Browser> {
|
||||
const { patchBrowser } = await import('./human-puppeteer/index.js');
|
||||
const { resolveConfig } = await import('./human/config.js');
|
||||
const cfg = resolveConfig(
|
||||
(options.humanPreset as any) ?? 'default',
|
||||
options.humanConfig as any,
|
||||
options.humanPreset ?? 'default',
|
||||
options.humanConfig,
|
||||
);
|
||||
patchBrowser(browser, cfg);
|
||||
}
|
||||
|
||||
+14
-2
@@ -2,6 +2,9 @@
|
||||
* Shared types for cloakbrowser launch wrappers.
|
||||
*/
|
||||
|
||||
import type { BrowserContextOptions } from "playwright-core";
|
||||
import type { HumanConfig, HumanPreset } from "./human/config.js";
|
||||
|
||||
export interface LaunchOptions {
|
||||
/** Run in headless mode (default: true). */
|
||||
headless?: boolean;
|
||||
@@ -27,9 +30,9 @@ export interface LaunchOptions {
|
||||
/** Enable human-like mouse, keyboard, and scroll behavior. */
|
||||
humanize?: boolean;
|
||||
/** Human behavior preset: 'default' or 'careful'. */
|
||||
humanPreset?: 'default' | 'careful';
|
||||
humanPreset?: HumanPreset;
|
||||
/** Override individual human behavior parameters. */
|
||||
humanConfig?: Record<string, unknown>;
|
||||
humanConfig?: Partial<HumanConfig>;
|
||||
}
|
||||
|
||||
export interface LaunchContextOptions extends LaunchOptions {
|
||||
@@ -43,6 +46,15 @@ export interface LaunchContextOptions extends LaunchOptions {
|
||||
timezoneId?: string;
|
||||
/** Color scheme preference — 'light', 'dark', or 'no-preference'. */
|
||||
colorScheme?: "light" | "dark" | "no-preference";
|
||||
/**
|
||||
* Extra options forwarded directly to Playwright's `browser.newContext()` —
|
||||
* e.g. `storageState`, `permissions`, `geolocation`, `extraHTTPHeaders`,
|
||||
* `httpCredentials`. Use this for context-level options not surfaced as
|
||||
* top-level fields. `locale` and `timezoneId` are stripped here to avoid
|
||||
* detectable CDP emulation — use the top-level `locale` and `timezone`
|
||||
* wrapper fields instead (they route through undetectable binary flags).
|
||||
*/
|
||||
contextOptions?: BrowserContextOptions;
|
||||
}
|
||||
|
||||
export interface LaunchPersistentContextOptions extends LaunchContextOptions {
|
||||
|
||||
+116
-7
@@ -4,14 +4,20 @@ import { DEFAULT_VIEWPORT, getChromiumVersion } from "../src/config.js";
|
||||
|
||||
describe("binaryInfo", () => {
|
||||
it("returns correct structure", () => {
|
||||
const info = binaryInfo();
|
||||
const orig = process.env.CLOAKBROWSER_CACHE_DIR;
|
||||
process.env.CLOAKBROWSER_CACHE_DIR = `/tmp/cloakbrowser-test-${Date.now()}`;
|
||||
try {
|
||||
const info = binaryInfo();
|
||||
|
||||
expect(info.version).toBe(getChromiumVersion());
|
||||
expect(info.platform).toMatch(/^(linux|darwin|windows)-(x64|arm64)$/);
|
||||
expect(info.binaryPath).toBeTruthy();
|
||||
expect(typeof info.installed).toBe("boolean");
|
||||
expect(info.cacheDir).toContain("cloakbrowser");
|
||||
expect(info.downloadUrl).toContain(".tar.gz");
|
||||
expect(info.version).toBe(getChromiumVersion());
|
||||
expect(info.platform).toMatch(/^(linux|darwin|windows)-(x64|arm64)$/);
|
||||
expect(info.binaryPath).toBeTruthy();
|
||||
expect(typeof info.installed).toBe("boolean");
|
||||
expect(info.cacheDir).toContain("cloakbrowser");
|
||||
} finally {
|
||||
if (orig) process.env.CLOAKBROWSER_CACHE_DIR = orig;
|
||||
else delete process.env.CLOAKBROWSER_CACHE_DIR;
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
@@ -130,6 +136,60 @@ describe("launchContext (unit)", () => {
|
||||
// Browser also closed
|
||||
expect(mockBrowser.close).toHaveBeenCalledOnce();
|
||||
});
|
||||
|
||||
it("forwards contextOptions to newContext (storageState, etc.)", async () => {
|
||||
const { launchContext } = await import("../src/playwright.js");
|
||||
await launchContext({
|
||||
contextOptions: {
|
||||
storageState: "state.json",
|
||||
permissions: ["geolocation"],
|
||||
},
|
||||
});
|
||||
|
||||
const ctxArgs = mockBrowser.newContext.mock.calls[0][0];
|
||||
expect(ctxArgs.storageState).toBe("state.json");
|
||||
expect(ctxArgs.permissions).toEqual(["geolocation"]);
|
||||
});
|
||||
|
||||
it("explicit top-level fields win over contextOptions on collision", async () => {
|
||||
const { launchContext } = await import("../src/playwright.js");
|
||||
await launchContext({
|
||||
userAgent: "Explicit/1.0",
|
||||
viewport: { width: 1280, height: 720 },
|
||||
colorScheme: "dark",
|
||||
contextOptions: {
|
||||
userAgent: "ShouldBeOverridden/9.9",
|
||||
viewport: { width: 9999, height: 9999 },
|
||||
colorScheme: "light",
|
||||
},
|
||||
});
|
||||
|
||||
const ctxArgs = mockBrowser.newContext.mock.calls[0][0];
|
||||
expect(ctxArgs.userAgent).toBe("Explicit/1.0");
|
||||
expect(ctxArgs.viewport).toEqual({ width: 1280, height: 720 });
|
||||
expect(ctxArgs.colorScheme).toBe("dark");
|
||||
});
|
||||
|
||||
it("strips locale and timezoneId from contextOptions (stealth-sensitive)", async () => {
|
||||
const warnSpy = vi.spyOn(console, "warn").mockImplementation(() => {});
|
||||
const { launchContext } = await import("../src/playwright.js");
|
||||
await launchContext({
|
||||
contextOptions: {
|
||||
storageState: "state.json",
|
||||
locale: "de-DE",
|
||||
timezoneId: "Europe/Berlin",
|
||||
},
|
||||
});
|
||||
|
||||
const ctxArgs = mockBrowser.newContext.mock.calls[0][0];
|
||||
// Stealth-sensitive keys stripped — they would reintroduce detectable CDP emulation.
|
||||
expect(ctxArgs.locale).toBeUndefined();
|
||||
expect(ctxArgs.timezoneId).toBeUndefined();
|
||||
// Benign keys preserved
|
||||
expect(ctxArgs.storageState).toBe("state.json");
|
||||
// Warning was logged for both stripped keys
|
||||
expect(warnSpy).toHaveBeenCalledTimes(2);
|
||||
});
|
||||
});
|
||||
|
||||
describe("launchPersistentContext (unit)", () => {
|
||||
@@ -207,4 +267,53 @@ describe("launchPersistentContext (unit)", () => {
|
||||
expect(args.userAgent).toBe("Custom/1.0");
|
||||
expect(args.colorScheme).toBe("dark");
|
||||
});
|
||||
|
||||
it("forwards contextOptions to launchPersistentContext", async () => {
|
||||
const { launchPersistentContext } = await import("../src/playwright.js");
|
||||
await launchPersistentContext({
|
||||
userDataDir: "/tmp/profile",
|
||||
contextOptions: {
|
||||
permissions: ["geolocation"],
|
||||
extraHTTPHeaders: { "X-Custom": "1" },
|
||||
},
|
||||
});
|
||||
|
||||
const args = mockChromium.launchPersistentContext.mock.calls[0][1];
|
||||
expect(args.permissions).toEqual(["geolocation"]);
|
||||
expect(args.extraHTTPHeaders).toEqual({ "X-Custom": "1" });
|
||||
});
|
||||
|
||||
it("explicit top-level fields win over contextOptions in persistent context", async () => {
|
||||
const { launchPersistentContext } = await import("../src/playwright.js");
|
||||
await launchPersistentContext({
|
||||
userDataDir: "/tmp/profile",
|
||||
userAgent: "Explicit/1.0",
|
||||
viewport: { width: 1280, height: 720 },
|
||||
contextOptions: {
|
||||
userAgent: "ShouldBeOverridden/9.9",
|
||||
viewport: { width: 9999, height: 9999 },
|
||||
},
|
||||
});
|
||||
|
||||
const args = mockChromium.launchPersistentContext.mock.calls[0][1];
|
||||
expect(args.userAgent).toBe("Explicit/1.0");
|
||||
expect(args.viewport).toEqual({ width: 1280, height: 720 });
|
||||
});
|
||||
|
||||
it("strips locale and timezoneId from contextOptions (persistent context)", async () => {
|
||||
const warnSpy = vi.spyOn(console, "warn").mockImplementation(() => {});
|
||||
const { launchPersistentContext } = await import("../src/playwright.js");
|
||||
await launchPersistentContext({
|
||||
userDataDir: "/tmp/profile",
|
||||
contextOptions: {
|
||||
locale: "de-DE",
|
||||
timezoneId: "Europe/Berlin",
|
||||
},
|
||||
});
|
||||
|
||||
const args = mockChromium.launchPersistentContext.mock.calls[0][1];
|
||||
expect(args.locale).toBeUndefined();
|
||||
expect(args.timezoneId).toBeUndefined();
|
||||
expect(warnSpy).toHaveBeenCalledTimes(2);
|
||||
});
|
||||
});
|
||||
|
||||
+61
-2
@@ -191,8 +191,7 @@ describe("resolveProxyConfig", () => {
|
||||
password: "p@ss",
|
||||
});
|
||||
expect(proxyOption).toBeUndefined();
|
||||
expect(proxyArgs.length).toBe(1);
|
||||
expect(proxyArgs[0]).toContain("--proxy-server=socks5://user:p%40ss@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:p%40ss@host:1080"]);
|
||||
});
|
||||
|
||||
it("includes bypass for socks5 dict", () => {
|
||||
@@ -203,4 +202,64 @@ describe("resolveProxyConfig", () => {
|
||||
expect(proxyArgs).toContain("--proxy-server=socks5://host:1080");
|
||||
expect(proxyArgs).toContain("--proxy-bypass-list=.example.com");
|
||||
});
|
||||
|
||||
// Chromium's --proxy-server parser truncates passwords at '=' (#157).
|
||||
// Wrapper must auto URL-encode before passing to Chrome.
|
||||
it("encodes '=' in socks5 string password", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:pass=123@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:pass%3D123@host:1080"]);
|
||||
});
|
||||
|
||||
it("encoding is idempotent for already-encoded socks5 string", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:pass%3D123@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:pass%3D123@host:1080"]);
|
||||
});
|
||||
|
||||
it("leaves socks5 string without creds unchanged", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://host:1080"]);
|
||||
});
|
||||
|
||||
it("encodes password even with empty username (password-only userinfo)", () => {
|
||||
// Regression: empty-username bypass would skip encoding, leaving the
|
||||
// Chromium truncation bug alive for this userinfo shape.
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://:pass=123@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://:pass%3D123@host:1080"]);
|
||||
});
|
||||
|
||||
it("handles literal '%' in password without throwing (malformed escape)", () => {
|
||||
// JS's decodeURIComponent throws on '%sure' (% not followed by 2 hex digits).
|
||||
// Must fall back to treating '%' as literal and percent-encoding it.
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:100%sure@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:100%25sure@host:1080"]);
|
||||
});
|
||||
|
||||
it("passes malformed SOCKS5 URLs through unchanged (no throw)", () => {
|
||||
// Broken IPv6 bracket — wrapper must not throw;
|
||||
// Chromium will surface its own error.
|
||||
const { proxyArgs: a1 } = resolveProxyConfig("socks5://user:pass@[::1");
|
||||
expect(a1).toEqual(["--proxy-server=socks5://user:pass@[::1"]);
|
||||
});
|
||||
|
||||
it("passes non-numeric port through unchanged", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:pass@host:abc");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:pass@host:abc"]);
|
||||
});
|
||||
|
||||
it("encodes special chars in IPv6 SOCKS5 string password", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:pass=eq@[::1]:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:pass%3Deq@[::1]:1080"]);
|
||||
});
|
||||
|
||||
// Regression #157: userinfo must be split at the LAST '@' (RFC 3986),
|
||||
// not the first, so raw '@' in a password parses correctly.
|
||||
it("encodes raw '@' in socks5 string password (last-@ split)", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:p@ss@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:p%40ss@host:1080"]);
|
||||
});
|
||||
|
||||
it("handles multiple raw '@' in password (splits at last)", () => {
|
||||
const { proxyArgs } = resolveProxyConfig("socks5://user:a@b@c@host:1080");
|
||||
expect(proxyArgs).toEqual(["--proxy-server=socks5://user:a%40b%40c@host:1080"]);
|
||||
});
|
||||
});
|
||||
|
||||
@@ -320,8 +320,14 @@ describe("download fallback", () => {
|
||||
|
||||
describe("effective version", () => {
|
||||
it("returns platform version when no marker exists", () => {
|
||||
// Default behavior — no marker file in test environment
|
||||
expect(getEffectiveVersion()).toBe(getChromiumVersion());
|
||||
const orig = process.env.CLOAKBROWSER_CACHE_DIR;
|
||||
process.env.CLOAKBROWSER_CACHE_DIR = `/tmp/cloakbrowser-test-${Date.now()}`;
|
||||
try {
|
||||
expect(getEffectiveVersion()).toBe(getChromiumVersion());
|
||||
} finally {
|
||||
if (orig) process.env.CLOAKBROWSER_CACHE_DIR = orig;
|
||||
else delete process.env.CLOAKBROWSER_CACHE_DIR;
|
||||
}
|
||||
});
|
||||
});
|
||||
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
"""Unit tests for launch_context() — context kwargs, viewport defaults, close cleanup."""
|
||||
|
||||
from unittest.mock import MagicMock, call, patch
|
||||
from unittest.mock import AsyncMock, MagicMock, call, patch
|
||||
|
||||
import pytest
|
||||
|
||||
@@ -207,3 +207,125 @@ def test_kwargs_passthrough(mock_launch, _mock_bin):
|
||||
# Verify kwarg did NOT leak to launch()
|
||||
launch_kwargs = mock_launch.call_args[1]
|
||||
assert "record_video_dir" not in launch_kwargs
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Async: launch_context_async()
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
|
||||
def _make_mock_async_browser():
|
||||
"""Create a mock async browser whose new_context() returns a mock context."""
|
||||
browser = AsyncMock()
|
||||
context = AsyncMock()
|
||||
browser.new_context.return_value = context
|
||||
return browser, context
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_storage_state_forwarded(mock_launch_async, _mock_bin):
|
||||
"""storage_state kwarg forwarded to browser.new_context() in async path.
|
||||
|
||||
This is the motivating use case from issue #141.
|
||||
"""
|
||||
browser, context = _make_mock_async_browser()
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
await launch_context_async(storage_state="state.json")
|
||||
|
||||
ctx_kwargs = browser.new_context.call_args
|
||||
assert ctx_kwargs[1]["storage_state"] == "state.json"
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_default_viewport(mock_launch_async, _mock_bin):
|
||||
"""DEFAULT_VIEWPORT applied when no viewport given (async)."""
|
||||
browser, context = _make_mock_async_browser()
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
await launch_context_async()
|
||||
|
||||
ctx_kwargs = browser.new_context.call_args
|
||||
assert ctx_kwargs[1]["viewport"] == DEFAULT_VIEWPORT
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_locale_flows_to_binary_not_cdp(mock_launch_async, _mock_bin):
|
||||
"""locale flows to launch_async() for --lang flag, NOT to new_context() CDP."""
|
||||
browser, context = _make_mock_async_browser()
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
await launch_context_async(locale="de-DE", timezone="Europe/Berlin")
|
||||
|
||||
# Binary flags
|
||||
assert mock_launch_async.call_args[1]["locale"] == "de-DE"
|
||||
assert mock_launch_async.call_args[1]["timezone"] == "Europe/Berlin"
|
||||
# Not in context — would trigger detectable CDP emulation
|
||||
ctx_kwargs = browser.new_context.call_args
|
||||
assert "locale" not in ctx_kwargs[1]
|
||||
assert "timezone_id" not in ctx_kwargs[1]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_close_closes_browser(mock_launch_async, _mock_bin):
|
||||
"""await ctx.close() also closes the underlying browser."""
|
||||
browser, context = _make_mock_async_browser()
|
||||
original_ctx_close = context.close
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
ctx = await launch_context_async()
|
||||
|
||||
await ctx.close()
|
||||
original_ctx_close.assert_called_once()
|
||||
browser.close.assert_called_once()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_error_closes_browser(mock_launch_async, _mock_bin):
|
||||
"""If new_context() raises in async path, browser is still closed."""
|
||||
browser = AsyncMock()
|
||||
browser.new_context.side_effect = RuntimeError("context creation failed")
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
with pytest.raises(RuntimeError, match="context creation failed"):
|
||||
await launch_context_async()
|
||||
|
||||
browser.close.assert_called_once()
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@patch("cloakbrowser.browser.ensure_binary", return_value="/fake/chrome")
|
||||
@patch("cloakbrowser.browser.launch_async")
|
||||
async def test_async_cancellation_closes_browser(mock_launch_async, _mock_bin):
|
||||
"""asyncio.CancelledError during new_context() still closes browser.
|
||||
|
||||
CancelledError derives from BaseException (not Exception) in Python 3.8+,
|
||||
so the cleanup must catch BaseException to prevent browser process leaks
|
||||
when the awaiting task is cancelled.
|
||||
"""
|
||||
import asyncio
|
||||
|
||||
browser = AsyncMock()
|
||||
browser.new_context.side_effect = asyncio.CancelledError()
|
||||
mock_launch_async.return_value = browser
|
||||
|
||||
from cloakbrowser.browser import launch_context_async
|
||||
with pytest.raises(asyncio.CancelledError):
|
||||
await launch_context_async()
|
||||
|
||||
browser.close.assert_called_once()
|
||||
|
||||
@@ -266,3 +266,75 @@ class TestResolveProxyConfig:
|
||||
assert kwargs == {}
|
||||
assert "--proxy-server=socks5://host:1080" in args
|
||||
assert "--proxy-bypass-list=.example.com" in args
|
||||
|
||||
def test_socks5_string_encodes_equals_in_password(self):
|
||||
# Chromium's --proxy-server parser truncates passwords at '=' (#157).
|
||||
# Wrapper must auto URL-encode before passing to Chrome.
|
||||
_, args = _resolve_proxy_config("socks5://user:pass=123@host:1080")
|
||||
assert args == ["--proxy-server=socks5://user:pass%3D123@host:1080"]
|
||||
|
||||
def test_socks5_string_encodes_at_in_password(self):
|
||||
_, args = _resolve_proxy_config("socks5://user:p@ss@host:1080")
|
||||
# Note: parsing "user:p@ss@host" — urlparse takes everything up to LAST @
|
||||
# as userinfo, so password = "p@ss".
|
||||
assert args == ["--proxy-server=socks5://user:p%40ss@host:1080"]
|
||||
|
||||
def test_socks5_string_encoding_idempotent(self):
|
||||
# Already-encoded input should remain encoded (not double-encoded).
|
||||
_, args = _resolve_proxy_config("socks5://user:pass%3D123@host:1080")
|
||||
assert args == ["--proxy-server=socks5://user:pass%3D123@host:1080"]
|
||||
|
||||
def test_socks5_string_no_creds_unchanged(self):
|
||||
_, args = _resolve_proxy_config("socks5://host:1080")
|
||||
assert args == ["--proxy-server=socks5://host:1080"]
|
||||
|
||||
def test_socks5_string_password_only_still_encoded(self):
|
||||
# Empty username with password: fix must still re-encode the password
|
||||
# (regression test for empty-username bypass).
|
||||
_, args = _resolve_proxy_config("socks5://:pass=123@host:1080")
|
||||
assert args == ["--proxy-server=socks5://:pass%3D123@host:1080"]
|
||||
|
||||
def test_socks5_string_empty_password_preserves_colon(self):
|
||||
# `user:@host` (empty password) must NOT collapse to `user@host` —
|
||||
# semantics differ between the two forms.
|
||||
_, args = _resolve_proxy_config("socks5://user:@host:1080")
|
||||
assert args == ["--proxy-server=socks5://user:@host:1080"]
|
||||
|
||||
def test_socks5_string_literal_percent_in_password(self):
|
||||
# Literal '%' not followed by 2 hex digits must be encoded as '%25'
|
||||
# so Chrome decodes it back to '%'. Must not crash.
|
||||
_, args = _resolve_proxy_config("socks5://user:100%sure@host:1080")
|
||||
assert args == ["--proxy-server=socks5://user:100%25sure@host:1080"]
|
||||
|
||||
def test_socks5_string_malformed_port_passes_through(self, caplog):
|
||||
# Invalid port (non-numeric) raises in urlparse.port. Wrapper should
|
||||
# log a warning and pass original through to Chromium.
|
||||
import logging
|
||||
with caplog.at_level(logging.WARNING, logger="cloakbrowser"):
|
||||
_, args = _resolve_proxy_config("socks5://user:pass@host:abc")
|
||||
assert args == ["--proxy-server=socks5://user:pass@host:abc"]
|
||||
assert any("Malformed SOCKS5" in r.message for r in caplog.records)
|
||||
|
||||
def test_socks5_string_malformed_ipv6_passes_through(self, caplog):
|
||||
# Broken IPv6 bracket — must not crash, and must reach Chromium
|
||||
# verbatim so its own error surfaces instead of a silent rewrite.
|
||||
import logging
|
||||
with caplog.at_level(logging.WARNING, logger="cloakbrowser"):
|
||||
_, args = _resolve_proxy_config("socks5://user:pass@[::1")
|
||||
assert args == ["--proxy-server=socks5://user:pass@[::1"]
|
||||
|
||||
def test_socks5_string_preserves_path_and_query(self):
|
||||
# Nonstandard for SOCKS5, but don't silently drop user-supplied suffixes.
|
||||
# Matches JS behavior.
|
||||
_, args = _resolve_proxy_config("socks5://user:pass@host:1080/p?x=1#f")
|
||||
assert args[0] == "--proxy-server=socks5://user:pass@host:1080/p?x=1#f"
|
||||
|
||||
def test_socks5_string_ipv6_with_special_char_password(self):
|
||||
# IPv6 host + special char in password — both must be handled.
|
||||
_, args = _resolve_proxy_config("socks5://user:pass=eq@[::1]:1080")
|
||||
assert args[0] == "--proxy-server=socks5://user:pass%3Deq@[::1]:1080"
|
||||
|
||||
def test_socks5_string_port_zero_preserved(self):
|
||||
# Port 0 is an unusual but valid URL component; don't silently strip it.
|
||||
_, args = _resolve_proxy_config("socks5://user:pass=1@host:0")
|
||||
assert args[0] == "--proxy-server=socks5://user:pass%3D1@host:0"
|
||||
|
||||
Reference in New Issue
Block a user