Convert Curl to Python Requests

requests: simplest API, sync only, dependency you already have. Use for scripts and most code. httpx: similar API to requests but with HTTP/2, sync + async support, type hints. Drop-in upgrade if you want HTTP/2. aiohttp: async only, requires asyncio. Use when concurrent requests matter (scraping, fanout to many APIs). The aiohttp output wraps your call in async def main() + asyncio.run(main()) boilerplate.

When the body parses as JSON, the tool emits a Python literal dict/list (not a string) and passes it as json=data. requests/httpx/aiohttp all auto-set Content-Type: application/json and serialize. If you had an explicit Content-Type header in your cURL, it’s preserved – the library generally honors the explicit one. For non-JSON bodies (form-encoded, plain text), uses data= with the body as a Python double-quoted string.

cURL: -u admin:secret becomes auth=('admin', 'secret') on requests/httpx (built-in tuple form). On aiohttp: auth=aiohttp.BasicAuth('admin', 'secret'). The password preserves colons: -u 'user:pa:ss' → ('user', 'pa:ss') (split on first colon only).

cURL: -F '[email protected]' → files={'file': ('photo.jpg', open('photo.jpg', 'rb'))} with a comment that the path is relative to Python cwd at runtime. Non-file fields (-F 'caption=hello') go into a separate form_data dict passed as data=form_data. Don’t forget to close() the file handle or use a with block in production code (the tool emits the minimal version for brevity).

cURL: curl -G 'https://api.com/search' -d 'q=hi' -d 'lim=10' → emits params={'q': 'hi', 'lim': '10'} and passes as params=params to requests.get(). No body. Method stays GET. The -d values are URL-decoded for the dict (so -d 'q=hello%20world' becomes 'q': 'hello world').

Yes. Backslash-newline (and backslash-CR-LF on Windows) is collapsed to a single space before tokenisation. So curl 'https://x.com' \
-H 'A: b' \
-d 'data' works as expected.

In POSIX shell (bash, sh, zsh), single-quoted strings are 100% literal – no backslash escapes, no variable expansion. Double-quoted strings honor \", \\, \$, \`. Chrome DevTools “Copy as cURL” uses single quotes for everything, so this bug hit every real-world command with backslashes (regex patterns, Windows paths, etc.).

Not converted: --cert, --key, --cacert (requests’ cert= takes a path or tuple – add manually with the file path), --proxy (would become proxies={'http': '...', 'https': '...'} – easy to add manually), -o/--output (use with open('file', 'wb') as f: f.write(response.content)), --resolve (DNS override), HTTP/2-specific flags. NTLM/Digest auth not handled (would need requests-ntlm package).

Yes. cURL input + generated Python stay in your browser. No network requests are made and no data is logged. Bearer tokens, basic auth credentials, and cookies in your cURL never leave your device.