HtmlQuill

Convert HTML or a URL to Markdown.

Installation

pip install htmlquill

Optional Playwright backend:

pip install "htmlquill[browser]"
playwright install chromium

CLI usage

# Auto-save using the first Markdown heading
htmlquill convert https://example.com/article

# Manual output path
htmlquill convert https://example.com/article -o article.md

# Preview generated filename without saving
htmlquill convert https://example.com/article --filename-only

# Print Markdown content without saving
htmlquill convert https://example.com/article --stdout

# Save generated filename to a target directory
htmlquill convert https://example.com/article --output-dir notes

# Limit generated filename stem length
htmlquill convert https://example.com/article --filename-max-length 60

# Inspect effective config
htmlquill config show https://example.com

# Initialize config and inspect paths
htmlquill config init
htmlquill config path

# Run diagnostics
htmlquill doctor

# Count generated Markdown structure
htmlquill analyse example.md

# Preview Markdown in the terminal
htmlquill preview example.md

htmlquill SOURCE is retained as shorthand for htmlquill convert SOURCE; it now follows the same auto-save behavior unless --stdout is used.

Command overview

• htmlquill convert SOURCE [options]
• htmlquill config path|show|init|validate
• htmlquill auth path|show|init
• htmlquill doctor [--url URL] [--fetch] [--json] [--strict]
• htmlquill analyse SOURCE (alias: htmlquill analyze SOURCE)
• htmlquill preview SOURCE

Convert options

Browser mode details

• auto (default): tries requests first; on HTTP 403 or detected challenge page, falls back to system Chromium, then Playwright.
• requests: plain HTTP via requests.
• chromium: uses system Chromium via subprocess.
• playwright: uses Playwright Chromium (optional dependency).

Configuration files

htmlquill resolves config file paths in this order:

1. --config PATH
2. HTMLQUILL_CONFIG
3. $XDG_CONFIG_HOME/htmlquill/config.toml
4. ~/.config/htmlquill/config.toml

Example config.toml:

version = 1

[defaults]
adapter = "html"
browser = "auto"
timeout = 30.0
fail_on_challenge = true
fallback_on_challenge = true

[paths]
auth_file = "~/.config/htmlquill/auth.json"

[challenge]
markers = [
  "Performing security verification",
  "verifies you are not a bot",
  "You've been blocked by network security",
  "blocked by network security",
  "If you think you've been blocked by mistake, file a ticket",
]

[sites."medium.com"]
browser = "chromium"
timeout = 60.0
auth = "medium"

Authentication

HtmlQuill supports browser-state auth profiles through auth.json. Use this when a site works in an already-authenticated browser session and you want HtmlQuill to reuse that state.

Auth file resolution order:

1. --auth-file PATH
2. HTMLQUILL_AUTH
3. [paths].auth_file from config
4. $XDG_CONFIG_HOME/htmlquill/auth.json or ~/.config/htmlquill/auth.json

Example auth.json:

{
  "version": 1,
  "profiles": {
    "medium": {
      "kind": "browser_state",
      "playwright_storage_state": "~/.config/htmlquill/auth/medium.storage-state.json",
      "chromium_user_data_dir": "~/.config/htmlquill/chromium/medium"
    }
  }
}

Security notes:

• Do not commit auth files, storage-state files, or browser profile directories.
• Recommended permissions: chmod 600 ~/.config/htmlquill/auth.json.
• Recommended browser profile directory permissions: chmod 700 ~/.config/htmlquill/chromium/medium.

Reddit

HtmlQuill no longer ships a Reddit API/OAuth adapter. Reddit URLs are processed through the normal HTML fetch path, the same as other URLs. If Reddit returns a network-security or login interstitial, use a browser-based fetch profile, retry later, or export/save the page manually. htmlquill auth login reddit is intentionally not available.

Library usage

from htmlquill import html_to_markdown, url_to_markdown

markdown = html_to_markdown("<h1>Hello</h1><p>World</p>")

markdown = url_to_markdown("https://example.com")

# New optional controls (all optional)
markdown = url_to_markdown(
    "https://example.com",
    browser="requests",
    config=True,
    auth=False,
)

Development

pip install -e ".[dev]"
pytest -q
ruff check .