libgo/go/net/http/cgi/host.go

   1 // Copyright 2011 The Go Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style
   3 // license that can be found in the LICENSE file.
   4
   5 // This file implements the host side of CGI (being the webserver
   6 // parent process).
   7
   8 // Package cgi implements CGI (Common Gateway Interface) as specified
   9 // in RFC 3875.
  10 //
  11 // Note that using CGI means starting a new process to handle each
  12 // request, which is typically less efficient than using a
  13 // long-running server.  This package is intended primarily for
  14 // compatibility with existing systems.
  15 package cgi
  16
  17 import (
  18         "bufio"
  19         "fmt"
  20         "io"
  21         "log"
  22         "net/http"
  23         "os"
  24         "os/exec"
  25         "path/filepath"
  26         "regexp"
  27         "runtime"
  28         "strconv"
  29         "strings"
  30 )
  31
  32 var trailingPort = regexp.MustCompile(`:([0-9]+)$`)
  33
  34 var osDefaultInheritEnv = map[string][]string{
  35         "darwin":  {"DYLD_LIBRARY_PATH"},
  36         "freebsd": {"LD_LIBRARY_PATH"},
  37         "hpux":    {"LD_LIBRARY_PATH", "SHLIB_PATH"},
  38         "irix":    {"LD_LIBRARY_PATH", "LD_LIBRARYN32_PATH", "LD_LIBRARY64_PATH"},
  39         "linux":   {"LD_LIBRARY_PATH"},
  40         "openbsd": {"LD_LIBRARY_PATH"},
  41         "solaris": {"LD_LIBRARY_PATH", "LD_LIBRARY_PATH_32", "LD_LIBRARY_PATH_64"},
  42         "windows": {"SystemRoot", "COMSPEC", "PATHEXT", "WINDIR"},
  43 }
  44
  45 // Handler runs an executable in a subprocess with a CGI environment.
  46 type Handler struct {
  47         Path string // path to the CGI executable
  48         Root string // root URI prefix of handler or empty for "/"
  49
  50         // Dir specifies the CGI executable's working directory.
  51         // If Dir is empty, the base directory of Path is used.
  52         // If Path has no base directory, the current working
  53         // directory is used.
  54         Dir string
  55
  56         Env        []string    // extra environment variables to set, if any, as "key=value"
  57         InheritEnv []string    // environment variables to inherit from host, as "key"
  58         Logger     *log.Logger // optional log for errors or nil to use log.Print
  59         Args       []string    // optional arguments to pass to child process
  60
  61         // PathLocationHandler specifies the root http Handler that
  62         // should handle internal redirects when the CGI process
  63         // returns a Location header value starting with a "/", as
  64         // specified in RFC 3875 § 6.3.2. This will likely be
  65         // http.DefaultServeMux.
  66         //
  67         // If nil, a CGI response with a local URI path is instead sent
  68         // back to the client and not redirected internally.
  69         PathLocationHandler http.Handler
  70 }
  71
  72 // removeLeadingDuplicates remove leading duplicate in environments.
  73 // It's possible to override environment like following.
  74 //    cgi.Handler{
  75 //      ...
  76 //      Env: []string{"SCRIPT_FILENAME=foo.php"},
  77 //    }
  78 func removeLeadingDuplicates(env []string) (ret []string) {
  79         n := len(env)
  80         for i := 0; i < n; i++ {
  81                 e := env[i]
  82                 s := strings.SplitN(e, "=", 2)[0]
  83                 found := false
  84                 for j := i + 1; j < n; j++ {
  85                         if s == strings.SplitN(env[j], "=", 2)[0] {
  86                                 found = true
  87                                 break
  88                         }
  89                 }
  90                 if !found {
  91                         ret = append(ret, e)
  92                 }
  93         }
  94         return
  95 }
  96
  97 func (h *Handler) ServeHTTP(rw http.ResponseWriter, req *http.Request) {
  98         root := h.Root
  99         if root == "" {
 100                 root = "/"
 101         }
 102
 103         if len(req.TransferEncoding) > 0 && req.TransferEncoding[0] == "chunked" {
 104                 rw.WriteHeader(http.StatusBadRequest)
 105                 rw.Write([]byte("Chunked request bodies are not supported by CGI."))
 106                 return
 107         }
 108
 109         pathInfo := req.URL.Path
 110         if root != "/" && strings.HasPrefix(pathInfo, root) {
 111                 pathInfo = pathInfo[len(root):]
 112         }
 113
 114         port := "80"
 115         if matches := trailingPort.FindStringSubmatch(req.Host); len(matches) != 0 {
 116                 port = matches[1]
 117         }
 118
 119         env := []string{
 120                 "SERVER_SOFTWARE=go",
 121                 "SERVER_NAME=" + req.Host,
 122                 "SERVER_PROTOCOL=HTTP/1.1",
 123                 "HTTP_HOST=" + req.Host,
 124                 "GATEWAY_INTERFACE=CGI/1.1",
 125                 "REQUEST_METHOD=" + req.Method,
 126                 "QUERY_STRING=" + req.URL.RawQuery,
 127                 "REQUEST_URI=" + req.URL.RequestURI(),
 128                 "PATH_INFO=" + pathInfo,
 129                 "SCRIPT_NAME=" + root,
 130                 "SCRIPT_FILENAME=" + h.Path,
 131                 "REMOTE_ADDR=" + req.RemoteAddr,
 132                 "REMOTE_HOST=" + req.RemoteAddr,
 133                 "SERVER_PORT=" + port,
 134         }
 135
 136         if req.TLS != nil {
 137                 env = append(env, "HTTPS=on")
 138         }
 139
 140         for k, v := range req.Header {
 141                 k = strings.Map(upperCaseAndUnderscore, k)
 142                 joinStr := ", "
 143                 if k == "COOKIE" {
 144                         joinStr = "; "
 145                 }
 146                 env = append(env, "HTTP_"+k+"="+strings.Join(v, joinStr))
 147         }
 148
 149         if req.ContentLength > 0 {
 150                 env = append(env, fmt.Sprintf("CONTENT_LENGTH=%d", req.ContentLength))
 151         }
 152         if ctype := req.Header.Get("Content-Type"); ctype != "" {
 153                 env = append(env, "CONTENT_TYPE="+ctype)
 154         }
 155
 156         if h.Env != nil {
 157                 env = append(env, h.Env...)
 158         }
 159
 160         envPath := os.Getenv("PATH")
 161         if envPath == "" {
 162                 envPath = "/bin:/usr/bin:/usr/ucb:/usr/bsd:/usr/local/bin"
 163         }
 164         env = append(env, "PATH="+envPath)
 165
 166         for _, e := range h.InheritEnv {
 167                 if v := os.Getenv(e); v != "" {
 168                         env = append(env, e+"="+v)
 169                 }
 170         }
 171
 172         for _, e := range osDefaultInheritEnv[runtime.GOOS] {
 173                 if v := os.Getenv(e); v != "" {
 174                         env = append(env, e+"="+v)
 175                 }
 176         }
 177
 178         env = removeLeadingDuplicates(env)
 179
 180         var cwd, path string
 181         if h.Dir != "" {
 182                 path = h.Path
 183                 cwd = h.Dir
 184         } else {
 185                 cwd, path = filepath.Split(h.Path)
 186         }
 187         if cwd == "" {
 188                 cwd = "."
 189         }
 190
 191         internalError := func(err error) {
 192                 rw.WriteHeader(http.StatusInternalServerError)
 193                 h.printf("CGI error: %v", err)
 194         }
 195
 196         cmd := &exec.Cmd{
 197                 Path:   path,
 198                 Args:   append([]string{h.Path}, h.Args...),
 199                 Dir:    cwd,
 200                 Env:    env,
 201                 Stderr: os.Stderr, // for now
 202         }
 203         if req.ContentLength != 0 {
 204                 cmd.Stdin = req.Body
 205         }
 206         stdoutRead, err := cmd.StdoutPipe()
 207         if err != nil {
 208                 internalError(err)
 209                 return
 210         }
 211
 212         err = cmd.Start()
 213         if err != nil {
 214                 internalError(err)
 215                 return
 216         }
 217         if hook := testHookStartProcess; hook != nil {
 218                 hook(cmd.Process)
 219         }
 220         defer cmd.Wait()
 221         defer stdoutRead.Close()
 222
 223         linebody := bufio.NewReaderSize(stdoutRead, 1024)
 224         headers := make(http.Header)
 225         statusCode := 0
 226         headerLines := 0
 227         sawBlankLine := false
 228         for {
 229                 line, isPrefix, err := linebody.ReadLine()
 230                 if isPrefix {
 231                         rw.WriteHeader(http.StatusInternalServerError)
 232                         h.printf("cgi: long header line from subprocess.")
 233                         return
 234                 }
 235                 if err == io.EOF {
 236                         break
 237                 }
 238                 if err != nil {
 239                         rw.WriteHeader(http.StatusInternalServerError)
 240                         h.printf("cgi: error reading headers: %v", err)
 241                         return
 242                 }
 243                 if len(line) == 0 {
 244                         sawBlankLine = true
 245                         break
 246                 }
 247                 headerLines++
 248                 parts := strings.SplitN(string(line), ":", 2)
 249                 if len(parts) < 2 {
 250                         h.printf("cgi: bogus header line: %s", string(line))
 251                         continue
 252                 }
 253                 header, val := parts[0], parts[1]
 254                 header = strings.TrimSpace(header)
 255                 val = strings.TrimSpace(val)
 256                 switch {
 257                 case header == "Status":
 258                         if len(val) < 3 {
 259                                 h.printf("cgi: bogus status (short): %q", val)
 260                                 return
 261                         }
 262                         code, err := strconv.Atoi(val[0:3])
 263                         if err != nil {
 264                                 h.printf("cgi: bogus status: %q", val)
 265                                 h.printf("cgi: line was %q", line)
 266                                 return
 267                         }
 268                         statusCode = code
 269                 default:
 270                         headers.Add(header, val)
 271                 }
 272         }
 273         if headerLines == 0 || !sawBlankLine {
 274                 rw.WriteHeader(http.StatusInternalServerError)
 275                 h.printf("cgi: no headers")
 276                 return
 277         }
 278
 279         if loc := headers.Get("Location"); loc != "" {
 280                 if strings.HasPrefix(loc, "/") && h.PathLocationHandler != nil {
 281                         h.handleInternalRedirect(rw, req, loc)
 282                         return
 283                 }
 284                 if statusCode == 0 {
 285                         statusCode = http.StatusFound
 286                 }
 287         }
 288
 289         if statusCode == 0 && headers.Get("Content-Type") == "" {
 290                 rw.WriteHeader(http.StatusInternalServerError)
 291                 h.printf("cgi: missing required Content-Type in headers")
 292                 return
 293         }
 294
 295         if statusCode == 0 {
 296                 statusCode = http.StatusOK
 297         }
 298
 299         // Copy headers to rw's headers, after we've decided not to
 300         // go into handleInternalRedirect, which won't want its rw
 301         // headers to have been touched.
 302         for k, vv := range headers {
 303                 for _, v := range vv {
 304                         rw.Header().Add(k, v)
 305                 }
 306         }
 307
 308         rw.WriteHeader(statusCode)
 309
 310         _, err = io.Copy(rw, linebody)
 311         if err != nil {
 312                 h.printf("cgi: copy error: %v", err)
 313                 // And kill the child CGI process so we don't hang on
 314                 // the deferred cmd.Wait above if the error was just
 315                 // the client (rw) going away. If it was a read error
 316                 // (because the child died itself), then the extra
 317                 // kill of an already-dead process is harmless (the PID
 318                 // won't be reused until the Wait above).
 319                 cmd.Process.Kill()
 320         }
 321 }
 322
 323 func (h *Handler) printf(format string, v ...interface{}) {
 324         if h.Logger != nil {
 325                 h.Logger.Printf(format, v...)
 326         } else {
 327                 log.Printf(format, v...)
 328         }
 329 }
 330
 331 func (h *Handler) handleInternalRedirect(rw http.ResponseWriter, req *http.Request, path string) {
 332         url, err := req.URL.Parse(path)
 333         if err != nil {
 334                 rw.WriteHeader(http.StatusInternalServerError)
 335                 h.printf("cgi: error resolving local URI path %q: %v", path, err)
 336                 return
 337         }
 338         // TODO: RFC 3875 isn't clear if only GET is supported, but it
 339         // suggests so: "Note that any message-body attached to the
 340         // request (such as for a POST request) may not be available
 341         // to the resource that is the target of the redirect."  We
 342         // should do some tests against Apache to see how it handles
 343         // POST, HEAD, etc. Does the internal redirect get the same
 344         // method or just GET? What about incoming headers?
 345         // (e.g. Cookies) Which headers, if any, are copied into the
 346         // second request?
 347         newReq := &http.Request{
 348                 Method:     "GET",
 349                 URL:        url,
 350                 Proto:      "HTTP/1.1",
 351                 ProtoMajor: 1,
 352                 ProtoMinor: 1,
 353                 Header:     make(http.Header),
 354                 Host:       url.Host,
 355                 RemoteAddr: req.RemoteAddr,
 356                 TLS:        req.TLS,
 357         }
 358         h.PathLocationHandler.ServeHTTP(rw, newReq)
 359 }
 360
 361 func upperCaseAndUnderscore(r rune) rune {
 362         switch {
 363         case r >= 'a' && r <= 'z':
 364                 return r - ('a' - 'A')
 365         case r == '-':
 366                 return '_'
 367         case r == '=':
 368                 // Maybe not part of the CGI 'spec' but would mess up
 369                 // the environment in any case, as Go represents the
 370                 // environment as a slice of "key=value" strings.
 371                 return '_'
 372         }
 373         // TODO: other transformations in spec or practice?
 374         return r
 375 }
 376
 377 var testHookStartProcess func(*os.Process) // nil except for some tests