scrapy-download-handlers-incubator 0.4.0

Incubator for Scrapy download handlers

Overview

This is a collection of semi-official download handlers for Scrapy. See the Scrapy download handler documentation for more information.

They should work and some of them may be later promoted to the official status, but here they are provided as-is and no support or stability promises are given. The documentation, including limitations and unsupported features, is also provided as-is and may be incomplete.

As this code intentionally uses private Scrapy APIs, it specifies a tight dependency on Scrapy. This version of the package only supports Scrapy 2.17.x.

Features overview

The baseline for these handlers is the default Scrapy handler, HTTP11DownloadHandler, which uses Twisted and supports HTTP/1.1. Feature parity with it is an explicit goal but it’s not always possible and not all possible features are implemented in all handlers (which may change in the future). Certain popular features not supported by HTTP11DownloadHandler, like HTTP/2 support, and features unique to some handlers, may or may not be implemented. Please see the sections for individual handlers for more details.

The following table summarizes the most important differences:

Handler	HTTP/2	Proxies	Impersonation	TLS implementation
(HTTP11DownloadHandler)	Not possible	Yes	Not possible	cryptography
AiohttpDownloadHandler	Not possible	Yes	Not possible	Stdlib ssl
CurlCffiDownloadHandler	Yes	Yes	No	libcurl
HttpxDownloadHandler	Yes	Yes	Not possible	Stdlib ssl
NiquestsDownloadHandler	Yes	Yes	Not possible	Stdlib ssl
PyreqwestDownloadHandler	Yes	Not possible	Not possible	rustls

The following basic features are supported by all handlers unless mentioned in their docs:

• Native asyncio integration without requiring a Twisted reactor

• HTTP/1.1 for http and https schemes

• Unified download handler exceptions

• Proxies, including HTTP and HTTPS proxies for HTTP and HTTPS destinations

• Proxy authentication via HttpProxyMiddleware

• IPv6 destinations

• DOWNLOAD_MAXSIZE, DOWNLOAD_WARNSIZE and the respective request meta keys

• DOWNLOAD_TIMEOUT and the respective request meta key

• DOWNLOAD_FAIL_ON_DATALOSS and the "dataloss" flag

• Setting the download_latency request meta

• DOWNLOAD_BIND_ADDRESS

• DOWNLOAD_VERIFY_CERTIFICATES

• headers_received and bytes_received signals

• Not reading the proxy configuration from the environment variables

• Not handling cookies, redirects, compression and other things handled by Scrapy itself

Handlers

AiohttpDownloadHandler

This handler supports HTTP/1.1 and uses the aiohttp library.

Install it with:

pip install scrapy-download-handlers-incubator[aiohttp]

Enable it with:

DOWNLOAD_HANDLERS = {
    "http": "scrapy_download_handlers_incubator.AiohttpDownloadHandler",
    "https": "scrapy_download_handlers_incubator.AiohttpDownloadHandler",
}

Features and limitations

HTTP proxies	Yes
SOCKS proxies	No (not supported by the library)
HTTP/2	No (not supported by the library)
TLS verbose logging	Yes
response.ip_address	Yes
response.certificate	Yes (DER bytes)
Per-request bindaddress	No (not supported by the library)
Proxy certificate verification	Follows DOWNLOAD_VERIFY_CERTIFICATES
TLS implementation	Standard library ssl

Other limitations:

• HTTPS proxies for HTTPS destinations are not supported on Python < 3.11.

Notable features supported by the library but not implemented:

• DNS resolving settings

• Custom DNS resolvers

CurlCffiDownloadHandler

This handler supports HTTP/1.1 and HTTP/2 and uses the curl_cffi library.

Install it with:

pip install scrapy-download-handlers-incubator[curl-cffi]

Enable it with:

DOWNLOAD_HANDLERS = {
    "http": "scrapy_download_handlers_incubator.CurlCffiDownloadHandler",
    "https": "scrapy_download_handlers_incubator.CurlCffiDownloadHandler",
}

Features and limitations

HTTP proxies	Yes
SOCKS proxies	Yes (SOCKS4, SOCKS5)
HTTP/2	Yes
HTTP/3	Yes (but not tested)
TLS verbose logging	No (not supported by the library)
response.ip_address	Yes
response.certificate	No (not supported by the library)
Per-request bindaddress	No (not supported by the library)
Proxy certificate verification	Follows DOWNLOAD_VERIFY_CERTIFICATES
TLS implementation	libcurl

Notable features supported by the library but not implemented:

• Impersonation

• Advanced libcurl tunables

Settings

• CURL_CFFI_HTTP_VERSION (str, default: "v1", corresponding to “Enforce HTTP/1.1”): The HTTP version to use. The value is passed directly to the library so the possible values are set by curl_cffi.requests.utils.normalize_http_version() and the meanings of the underlying constants can be seen in libcurl docs (CURLOPT_HTTP_VERSION). Set this to "v2tls" or "v2" to enable HTTP/2 for HTTPS requests or for all requests respectively. Set this to "v3" to enable HTTP/3.

HttpxDownloadHandler

This is an updated copy of the official scrapy.core.downloader.handlers._httpx.HttpxDownloadHandler handler. It supports HTTP/1.1 and HTTP/2 and uses the httpx2 library or the httpx library.

Install it with:

# this will install httpx2
pip install scrapy-download-handlers-incubator[httpx2]
# this will install httpx
pip install scrapy-download-handlers-incubator[httpx]

Enable it with:

DOWNLOAD_HANDLERS = {
    "http": "scrapy_download_handlers_incubator.HttpxDownloadHandler",
    "https": "scrapy_download_handlers_incubator.HttpxDownloadHandler",
}

Features and limitations

HTTP proxies	Yes
SOCKS proxies	Yes (SOCKS5; requires httpx[socks])
HTTP/2	Yes (requires httpx[http2])
HTTP/3	No (not supported by the library)
TLS verbose logging	Yes
response.ip_address	Yes
response.certificate	Yes (DER bytes)
Per-request bindaddress	No (not supported by the library)
Proxy certificate verification	Follows DOWNLOAD_VERIFY_CERTIFICATES
TLS implementation	Standard library ssl

Other limitations:

• The handler creates a separate connection pool for each proxy URL (due to limitations of httpx) which may lead to higher resource usage when using proxy rotation.

Notable features supported by the library but not implemented:

• Alternative transports

• Limiting the number of per-proxy connection pool to save resources

Settings

• HTTPX_HTTP2_ENABLED (bool, default: False): Whether to enable HTTP/2.

NiquestsDownloadHandler

This handler supports HTTP/1.1 and HTTP/2 and uses the niquests library.

Install it with:

pip install scrapy-download-handlers-incubator[niquests]

Enable it with:

DOWNLOAD_HANDLERS = {
    "http": "scrapy_download_handlers_incubator.NiquestsDownloadHandler",
    "https": "scrapy_download_handlers_incubator.NiquestsDownloadHandler",
}

Features and limitations

HTTP proxies	Yes
SOCKS proxies	Yes (SOCKS4, SOCKS5; requires niquests[socks])
HTTP/2	Yes
HTTP/3	No (not implemented)
TLS verbose logging	Yes
response.ip_address	Yes
response.certificate	Yes (DER bytes)
Per-request bindaddress	No (not supported by the library)
Proxy certificate verification	Follows DOWNLOAD_VERIFY_CERTIFICATES

Notable features supported by the library but not implemented:

• Custom DNS resolvers

• HTTP/2 tunables

Settings

• NIQUESTS_HTTP2_ENABLED (bool, default: False): Whether to enable HTTP/2.

PyreqwestDownloadHandler

This handler supports HTTP/1.1 and HTTP/2 and uses the pyreqwest library.

Install it with:

pip install scrapy-download-handlers-incubator[pyreqwest]

Enable it with:

DOWNLOAD_HANDLERS = {
    "http": "scrapy_download_handlers_incubator.PyreqwestDownloadHandler",
    "https": "scrapy_download_handlers_incubator.PyreqwestDownloadHandler",
}

Features and limitations

Proxies	No (not supported by the library)
HTTP/2	Yes
HTTP/3	No (not supported by the library)
TLS verbose logging	No (not supported by the library)
response.ip_address	No (not supported by the library)
response.certificate	No (not supported by the library)
Per-request bindaddress	No (not supported by the library)
TLS implementation	Rust rustls

Notable features supported by the library but not implemented:

• HTTP/2 tunables

Settings

• PYREQWEST_HTTP2_ENABLED (bool, default: False): Whether to enable HTTP/2.