wai-app-file-cgi-3.3.0: Network/Wai/Application/Classic/CGI.hs
{-# LANGUAGE OverloadedStrings, ScopedTypeVariables #-}
module Network.Wai.Application.Classic.CGI (
cgiApp
) where
import qualified Control.Exception as E (SomeException, SomeAsyncException, IOException, try, catch, bracket, fromException, throwIO)
import Control.Concurrent (forkFinally, forkIO, killThread, newEmptyMVar, putMVar, takeMVar)
import Control.Monad (mplus, void)
import Data.ByteString (ByteString)
import Data.CaseInsensitive (foldedCase)
import Data.Char (toUpper)
import Data.List (nub)
import qualified Data.ByteString.Char8 as BS (dropWhile, elem, intercalate, pack, readInt, unpack, tail)
import Data.Conduit
import qualified Data.Conduit.Binary as CB
import qualified Data.Conduit.List as CL
import Network.HTTP.Types
import Network.SockAddr
import Network.Socket (SockAddr (..))
import Network.Wai
import Network.Wai.Conduit
import Network.Wai.Application.Classic.Conduit
import Network.Wai.Application.Classic.Field
import Network.Wai.Application.Classic.FileInfo (decodedPathInfo, isUnsafePath, matchRoute)
import Network.Wai.Application.Classic.Header
import Network.Wai.Application.Classic.Path
import Network.Wai.Application.Classic.Types
import System.Directory (doesFileExist)
import System.Environment
import System.IO
import System.Process
-- $setup
-- >>> :set -XOverloadedStrings
----------------------------------------------------------------
type ENVVARS = [(String,String)]
gatewayInterface :: String
gatewayInterface = "CGI/1.1"
----------------------------------------------------------------
{-|
Handle any methods for CGI.
Paths containing a segment starting with \'.\' are rejected with 400.
CGI processes are reaped by this library. The program to link this
library should not ignore SIGCHLD. Otherwise, the PID of a finished
CGI process may be reused by another process before it is terminated.
-}
cgiApp :: ClassicAppSpec -> CgiAppSpec -> CgiRoute -> Application
cgiApp cspec spec cgii req respond
| isUnsafePath (rawPathInfo req) =
respond $ responseLBS badRequest400 textPlainHeader "Bad Request\r\n"
| not (matchRoute (cgiSrc cgii) (decodedPathInfo req)) = notFound
| otherwise = do
exist <- doesFileExist prog
if exist
then E.bracket setup teardown cgi
else notFound
where
notFound = respond $ responseLBS notFound404 textPlainHeader "Not Found\r\n"
cgiPath@(prog,_,_) =
pathinfoToCGI (cgiSrc cgii)
(cgiDst cgii)
(decodedPathInfo req)
(indexCgi spec)
setup = execProcess cspec cgiPath req
-- An unreaped process keeps its PID, so SIGTERM cannot hit
-- another process. Reaping may block, hence in another thread.
teardown (rhdl,whdl,pid) = do
terminateProcess pid -- SIGTERM
hClose rhdl
hClose whdl
void $ forkIO $ void (waitForProcess pid) `E.catch` ignore
-- The body is written in another thread so that the CGI output
-- can be read concurrently. Otherwise, both pipes may fill up.
cgi (rhdl,whdl,_) = E.bracket (forkWriter whdl) stopWriter $ \_ ->
respond =<< fromCGI rhdl
forkWriter whdl = do
done <- newEmptyMVar
tid <- forkFinally (toCGI whdl req) $ \_ -> do
hClose whdl `E.catch` ignore -- telling EOF
putMVar done ()
return (tid, done)
stopWriter (tid, done) = killThread tid >> takeMVar done
ignore (_ :: E.IOException) = return ()
----------------------------------------------------------------
type TRYPATH = Either E.IOException String
toCGI :: Handle -> Request -> IO ()
toCGI whdl req = runConduit (sourceRequestBody req .| CB.sinkHandle whdl)
fromCGI :: Handle -> IO Response
fromCGI rhdl = do
(src', hs) <- cgiHeader `E.catch` recover
let (st, hdr, hasBody) = case check hs of
Nothing -> (internalServerError500,[],False)
Just (s,h) -> (s,h,True)
let src | hasBody = src'
| otherwise = CL.sourceNull
return $ responseSource st hdr src
where
-- A redirect response may lack Content-Type (RFC 3875 Section 6.2).
check hs = (lookup hContentType hs `mplus` location) >> case lookup hStatus hs of
Nothing -> Just (maybe ok200 (const found302) location, hs)
Just l -> toStatus l >>= \s -> Just (s,hs')
where
location = lookup hLocation hs
hs' = filter (\(k,_) -> k /= hStatus) hs
toStatus s = BS.readInt s >>= \(code,msg) -> Just (Status code (BS.dropWhile (== ' ') msg))
emptyHeader = []
recover (e :: E.SomeException)
| isAsync e = E.throwIO e
| otherwise = return (CL.sourceNull, emptyHeader)
isAsync e = case E.fromException e of
Just (_ :: E.SomeAsyncException) -> True
Nothing -> False
cgiHeader = do
(rsrc,hs) <- CB.sourceHandle rhdl $$+ parseHeader
src <- toResponseSource rsrc
return (src,hs)
----------------------------------------------------------------
execProcess :: ClassicAppSpec -> (FilePath, String, String) -> Request -> IO (Handle, Handle, ProcessHandle)
execProcess cspec (prog, scriptName, pathinfo) req = do
let naddr = showSockAddr . remoteHost $ req
epath <- E.try (getEnv "PATH") :: IO TRYPATH
(Just whdl,Just rhdl,_,pid) <- createProcess $ proSpec naddr epath
hSetEncoding rhdl latin1
hSetEncoding whdl latin1
return (rhdl, whdl, pid)
where
proSpec naddr epath = (proc prog []) {
env = Just $ makeEnv req naddr prog scriptName pathinfo (softwareName cspec) epath
, std_in = CreatePipe
, std_out = CreatePipe
, close_fds = True
, create_group = True
}
makeEnv :: Request -> String -> FilePath -> String -> String -> ByteString ->
TRYPATH -> ENVVARS
makeEnv req naddr prog scriptName pathinfo sname epath =
addPath epath . addLen . addType . addPort . addHttps $ baseEnv ++ httpEnv
where
baseEnv = [
("GATEWAY_INTERFACE", gatewayInterface)
, ("SCRIPT_NAME", scriptName)
, ("SCRIPT_FILENAME", prog)
, ("REQUEST_METHOD", pathString . requestMethod $ req)
, ("SERVER_NAME", pathString host)
, ("SERVER_PORT", pathString port)
, ("REMOTE_ADDR", naddr)
, ("SERVER_PROTOCOL", show . httpVersion $ req)
, ("SERVER_SOFTWARE", pathString sname)
, ("PATH_INFO", pathinfo)
, ("QUERY_STRING", query req)
]
headers = requestHeaders req
addLen = addLength "CONTENT_LENGTH" $ requestBodyLength req
addType = addEnv "CONTENT_TYPE" $ lookup hContentType headers
addPort = addEnv "REMOTE_PORT" $ remotePort $ remoteHost req
addHttps ev | isSecure req = ("HTTPS", "on") : ev
| otherwise = ev
httpEnv = httpHeadersToEnv headers
addPath (Left _) ev = ev
addPath (Right path) ev = ("PATH", path) : ev
query = pathString . safeTail . rawQueryString
where
safeTail "" = ""
safeTail bs = BS.tail bs
(host, port) = hostPort req
addEnv :: String -> Maybe ByteString -> ENVVARS -> ENVVARS
addEnv _ Nothing envs = envs
addEnv key (Just val) envs = (key,pathString val) : envs
remotePort :: SockAddr -> Maybe ByteString
remotePort (SockAddrInet p _) = Just $ BS.pack $ show p
remotePort (SockAddrInet6 p _ _ _) = Just $ BS.pack $ show p
remotePort _ = Nothing
{-|
Converting request headers to HTTP_* variables.
Values of the same header are joined.
Headers containing '_' are dropped since they would collide with
ones containing '-' (e.g. X_Forwarded_For vs X-Forwarded-For).
>>> httpHeadersToEnv [("User-Agent","foo"),("Accept","a"),("accept","b"),("Proxy","x"),("Authorization","y"),("Content-Type","z")]
[("HTTP_USER_AGENT","foo"),("HTTP_ACCEPT","a, b")]
>>> httpHeadersToEnv [("X-Forwarded-For","192.0.2.1"),("X_Forwarded_For","192.0.2.2")]
[("HTTP_X_FORWARDED_FOR","192.0.2.1")]
-}
httpHeadersToEnv :: RequestHeaders -> ENVVARS
httpHeadersToEnv hdrs = map toEnv $ nub $ map fst relayed
where
relayed = filter (\(k,_) -> k `notElem` excluded && not (hasUnderscore k)) hdrs
hasUnderscore = BS.elem '_' . foldedCase
-- "Proxy" is excluded to prevent httpoxy.
excluded = [hContentType, hContentLength, hAuthorization, hProxyAuthorization, "proxy"]
toEnv k = (envName k, pathString $ BS.intercalate (sep k) [v | (k',v) <- relayed, k' == k])
sep k | k == hCookie = "; "
| otherwise = ", "
envName k = "HTTP_" ++ map conv (BS.unpack $ foldedCase k)
conv '-' = '_'
conv c = toUpper c
addLength :: String -> RequestBodyLength -> ENVVARS -> ENVVARS
addLength _ ChunkedBody envs = envs
addLength key (KnownLength len) envs = (key, show len) : envs
{-|
>>> pathinfoToCGI "/cgi-bin/" "/User/cgi-bin/" "/cgi-bin/foo" "index.cgi"
("/User/cgi-bin/foo","/cgi-bin/foo","")
>>> pathinfoToCGI "/cgi-bin/" "/User/cgi-bin/" "/cgi-bin/foo/bar" "index.cgi"
("/User/cgi-bin/foo","/cgi-bin/foo","/bar")
>>> pathinfoToCGI "/cgi-bin/" "/User/cgi-bin/" "/cgi-bin/" "index.cgi"
("/User/cgi-bin/index.cgi","/cgi-bin/index.cgi","")
>>> pathinfoToCGI "/cgi-bin/" "/User/cgi-bin/" "/cgi-bin" "index.cgi"
("/User/cgi-bin/index.cgi","/cgi-bin/index.cgi","")
>>> pathinfoToCGI "/cgi-bin/" "/User/cgi-bin/" "/cgi-bin//foo" "index.cgi"
("/User/cgi-bin/index.cgi","/cgi-bin/index.cgi","/foo")
-}
pathinfoToCGI :: Path -> Path -> Path -> Path -> (FilePath, String, String)
pathinfoToCGI src dst path index = (prog, scriptName, pathinfo)
where
path' = path <\> src
(prog',pathinfo') = case breakAtSeparator path' of
("", pinfo) -> (index, pinfo)
x -> x
prog = pathString (dst </> prog')
scriptName = pathString (src </> prog')
pathinfo = pathString pathinfo'