smuggler.py

import hashlib
import hmac
import json
import os

import klein
import treq

from twisted.web.client import UNKNOWN_LENGTH

# TODO: Move Security and Informational Headers So they apply globally


__version__ = "1.0"


SECRET_KEY = os.environ["SMUGGLER_KEY"]

USER_AGENT = os.environ.get(
    "SMUGGLER_USER_AGENT",
    "Smuggler/{}".format(__version__),
)

TIMEOUT = int(os.environ.get("SMUGGLER_TIMEOUT", 10))

MAX_IMAGE_SIZE = int(os.environ.get("SMUGGLER_SIZE_LIMIT", 5242880))

with open(
    os.environ.get(
        "SMUGGLER_MIMETYPES",
        os.path.join(os.path.dirname(__file__), "mimetypes.json"),
        )) as fp:
    MIMETYPES = json.load(fp)


def getHeader(headers, name, default=None):
    return headers.getRawHeaders(name, [default])[0]


def processResponse(sourceResponse, request):
    # Ensure that we know what size the image is
    if sourceResponse.length == UNKNOWN_LENGTH:
        request.setResponseCode(413)
        return "cannot determine size of source image"

    # Ensure that the image isn't too large for us
    if sourceResponse.length > MAX_IMAGE_SIZE:
        request.setResponseCode(413)
        return "source image exceeds max allowed size"

    contentType = getHeader(sourceResponse.headers, "Content-Type")
    etag = getHeader(sourceResponse.headers, "ETag")
    lastModified = getHeader(sourceResponse.headers, "Last-Modified")
    expires = getHeader(sourceResponse.headers, "Expires")
    cacheControl = getHeader(sourceResponse.headers, "Cache-Control")

    # Verify Content Type
    contentTypePrefix = contentType.split(";")[0].strip()
    if contentTypePrefix not in MIMETYPES:
        request.setResponseCode(415)
        return "non-Image content-type returned '%s'" % (contentTypePrefix,)

    request.setHeader("Content-Type", contentType)
    #request.setHeader("Content-Length", sourceResponse.length)

    if etag is not None:
        request.setETag(etag)

    if lastModified is not None:
        request.setLastModified(lastModified)

    if expires is not None:
        request.setHeader("Expires", expires)

    if cacheControl is not None:
        request.setHeader("Cache-Control", cacheControl)

    # Security based headers
    request.setHeader("X-Content-Type-Options", "nosniff")
    request.setHeader("X-Frame-Options", "deny")

    # Set our own headers
    request.setHeader("Server", USER_AGENT)

    return treq.content(sourceResponse)


@klein.route("/<hmacRequest>/<path:url>")
def proxyImage(request, hmacRequest, url):
    hmacDigest = hmac.new(
        SECRET_KEY, url,
        digestmod=hashlib.sha224,
    ).hexdigest()

    # if hmacDigest != hmacRequest:  # TODO: Constant Time
    #     request.setResponseCode(404)
    #     return "checksum mismatch"

    d = treq.get(
        url.encode("utf8"),
        headers={
            "Accept": "images/*",
            "User-Agent": USER_AGENT,
        },
        persistent=False,
        timeout=TIMEOUT,
    )
    d.addCallback(processResponse, request)

    return d


resource = klein.resource