Source file src/pkg/net/http/fs.go

     1	// Copyright 2009 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	// HTTP file system request handler
     6	
     7	package http
     8	
     9	import (
    10		"errors"
    11		"fmt"
    12		"io"
    13		"mime"
    14		"mime/multipart"
    15		"net/textproto"
    16		"os"
    17		"path"
    18		"path/filepath"
    19		"strconv"
    20		"strings"
    21		"time"
    22	)
    23	
    24	// A Dir implements http.FileSystem using the native file
    25	// system restricted to a specific directory tree.
    26	//
    27	// An empty Dir is treated as ".".
    28	type Dir string
    29	
    30	func (d Dir) Open(name string) (File, error) {
    31		if filepath.Separator != '/' && strings.IndexRune(name, filepath.Separator) >= 0 ||
    32			strings.Contains(name, "\x00") {
    33			return nil, errors.New("http: invalid character in file path")
    34		}
    35		dir := string(d)
    36		if dir == "" {
    37			dir = "."
    38		}
    39		f, err := os.Open(filepath.Join(dir, filepath.FromSlash(path.Clean("/"+name))))
    40		if err != nil {
    41			return nil, err
    42		}
    43		return f, nil
    44	}
    45	
    46	// A FileSystem implements access to a collection of named files.
    47	// The elements in a file path are separated by slash ('/', U+002F)
    48	// characters, regardless of host operating system convention.
    49	type FileSystem interface {
    50		Open(name string) (File, error)
    51	}
    52	
    53	// A File is returned by a FileSystem's Open method and can be
    54	// served by the FileServer implementation.
    55	type File interface {
    56		Close() error
    57		Stat() (os.FileInfo, error)
    58		Readdir(count int) ([]os.FileInfo, error)
    59		Read([]byte) (int, error)
    60		Seek(offset int64, whence int) (int64, error)
    61	}
    62	
    63	func dirList(w ResponseWriter, f File) {
    64		w.Header().Set("Content-Type", "text/html; charset=utf-8")
    65		fmt.Fprintf(w, "<pre>\n")
    66		for {
    67			dirs, err := f.Readdir(100)
    68			if err != nil || len(dirs) == 0 {
    69				break
    70			}
    71			for _, d := range dirs {
    72				name := d.Name()
    73				if d.IsDir() {
    74					name += "/"
    75				}
    76				// TODO htmlescape
    77				fmt.Fprintf(w, "<a href=\"%s\">%s</a>\n", name, name)
    78			}
    79		}
    80		fmt.Fprintf(w, "</pre>\n")
    81	}
    82	
    83	// ServeContent replies to the request using the content in the
    84	// provided ReadSeeker.  The main benefit of ServeContent over io.Copy
    85	// is that it handles Range requests properly, sets the MIME type, and
    86	// handles If-Modified-Since requests.
    87	//
    88	// If the response's Content-Type header is not set, ServeContent
    89	// first tries to deduce the type from name's file extension and,
    90	// if that fails, falls back to reading the first block of the content
    91	// and passing it to DetectContentType.
    92	// The name is otherwise unused; in particular it can be empty and is
    93	// never sent in the response.
    94	//
    95	// If modtime is not the zero time, ServeContent includes it in a
    96	// Last-Modified header in the response.  If the request includes an
    97	// If-Modified-Since header, ServeContent uses modtime to decide
    98	// whether the content needs to be sent at all.
    99	//
   100	// The content's Seek method must work: ServeContent uses
   101	// a seek to the end of the content to determine its size.
   102	//
   103	// If the caller has set w's ETag header, ServeContent uses it to
   104	// handle requests using If-Range and If-None-Match.
   105	//
   106	// Note that *os.File implements the io.ReadSeeker interface.
   107	func ServeContent(w ResponseWriter, req *Request, name string, modtime time.Time, content io.ReadSeeker) {
   108		size, err := content.Seek(0, os.SEEK_END)
   109		if err != nil {
   110			Error(w, "seeker can't seek", StatusInternalServerError)
   111			return
   112		}
   113		_, err = content.Seek(0, os.SEEK_SET)
   114		if err != nil {
   115			Error(w, "seeker can't seek", StatusInternalServerError)
   116			return
   117		}
   118		serveContent(w, req, name, modtime, size, content)
   119	}
   120	
   121	// if name is empty, filename is unknown. (used for mime type, before sniffing)
   122	// if modtime.IsZero(), modtime is unknown.
   123	// content must be seeked to the beginning of the file.
   124	func serveContent(w ResponseWriter, r *Request, name string, modtime time.Time, size int64, content io.ReadSeeker) {
   125		if checkLastModified(w, r, modtime) {
   126			return
   127		}
   128		rangeReq, done := checkETag(w, r)
   129		if done {
   130			return
   131		}
   132	
   133		code := StatusOK
   134	
   135		// If Content-Type isn't set, use the file's extension to find it.
   136		ctype := w.Header().Get("Content-Type")
   137		if ctype == "" {
   138			ctype = mime.TypeByExtension(filepath.Ext(name))
   139			if ctype == "" {
   140				// read a chunk to decide between utf-8 text and binary
   141				var buf [1024]byte
   142				n, _ := io.ReadFull(content, buf[:])
   143				b := buf[:n]
   144				ctype = DetectContentType(b)
   145				_, err := content.Seek(0, os.SEEK_SET) // rewind to output whole file
   146				if err != nil {
   147					Error(w, "seeker can't seek", StatusInternalServerError)
   148					return
   149				}
   150			}
   151			w.Header().Set("Content-Type", ctype)
   152		}
   153	
   154		// handle Content-Range header.
   155		sendSize := size
   156		var sendContent io.Reader = content
   157		if size >= 0 {
   158			ranges, err := parseRange(rangeReq, size)
   159			if err != nil {
   160				Error(w, err.Error(), StatusRequestedRangeNotSatisfiable)
   161				return
   162			}
   163			if sumRangesSize(ranges) >= size {
   164				// The total number of bytes in all the ranges
   165				// is larger than the size of the file by
   166				// itself, so this is probably an attack, or a
   167				// dumb client.  Ignore the range request.
   168				ranges = nil
   169			}
   170			switch {
   171			case len(ranges) == 1:
   172				// RFC 2616, Section 14.16:
   173				// "When an HTTP message includes the content of a single
   174				// range (for example, a response to a request for a
   175				// single range, or to a request for a set of ranges
   176				// that overlap without any holes), this content is
   177				// transmitted with a Content-Range header, and a
   178				// Content-Length header showing the number of bytes
   179				// actually transferred.
   180				// ...
   181				// A response to a request for a single range MUST NOT
   182				// be sent using the multipart/byteranges media type."
   183				ra := ranges[0]
   184				if _, err := content.Seek(ra.start, os.SEEK_SET); err != nil {
   185					Error(w, err.Error(), StatusRequestedRangeNotSatisfiable)
   186					return
   187				}
   188				sendSize = ra.length
   189				code = StatusPartialContent
   190				w.Header().Set("Content-Range", ra.contentRange(size))
   191			case len(ranges) > 1:
   192				for _, ra := range ranges {
   193					if ra.start > size {
   194						Error(w, err.Error(), StatusRequestedRangeNotSatisfiable)
   195						return
   196					}
   197				}
   198				sendSize = rangesMIMESize(ranges, ctype, size)
   199				code = StatusPartialContent
   200	
   201				pr, pw := io.Pipe()
   202				mw := multipart.NewWriter(pw)
   203				w.Header().Set("Content-Type", "multipart/byteranges; boundary="+mw.Boundary())
   204				sendContent = pr
   205				defer pr.Close() // cause writing goroutine to fail and exit if CopyN doesn't finish.
   206				go func() {
   207					for _, ra := range ranges {
   208						part, err := mw.CreatePart(ra.mimeHeader(ctype, size))
   209						if err != nil {
   210							pw.CloseWithError(err)
   211							return
   212						}
   213						if _, err := content.Seek(ra.start, os.SEEK_SET); err != nil {
   214							pw.CloseWithError(err)
   215							return
   216						}
   217						if _, err := io.CopyN(part, content, ra.length); err != nil {
   218							pw.CloseWithError(err)
   219							return
   220						}
   221					}
   222					mw.Close()
   223					pw.Close()
   224				}()
   225			}
   226	
   227			w.Header().Set("Accept-Ranges", "bytes")
   228			if w.Header().Get("Content-Encoding") == "" {
   229				w.Header().Set("Content-Length", strconv.FormatInt(sendSize, 10))
   230			}
   231		}
   232	
   233		w.WriteHeader(code)
   234	
   235		if r.Method != "HEAD" {
   236			io.CopyN(w, sendContent, sendSize)
   237		}
   238	}
   239	
   240	// modtime is the modification time of the resource to be served, or IsZero().
   241	// return value is whether this request is now complete.
   242	func checkLastModified(w ResponseWriter, r *Request, modtime time.Time) bool {
   243		if modtime.IsZero() {
   244			return false
   245		}
   246	
   247		// The Date-Modified header truncates sub-second precision, so
   248		// use mtime < t+1s instead of mtime <= t to check for unmodified.
   249		if t, err := time.Parse(TimeFormat, r.Header.Get("If-Modified-Since")); err == nil && modtime.Before(t.Add(1*time.Second)) {
   250			h := w.Header()
   251			delete(h, "Content-Type")
   252			delete(h, "Content-Length")
   253			w.WriteHeader(StatusNotModified)
   254			return true
   255		}
   256		w.Header().Set("Last-Modified", modtime.UTC().Format(TimeFormat))
   257		return false
   258	}
   259	
   260	// checkETag implements If-None-Match and If-Range checks.
   261	// The ETag must have been previously set in the ResponseWriter's headers.
   262	//
   263	// The return value is the effective request "Range" header to use and
   264	// whether this request is now considered done.
   265	func checkETag(w ResponseWriter, r *Request) (rangeReq string, done bool) {
   266		etag := w.Header().get("Etag")
   267		rangeReq = r.Header.get("Range")
   268	
   269		// Invalidate the range request if the entity doesn't match the one
   270		// the client was expecting.
   271		// "If-Range: version" means "ignore the Range: header unless version matches the
   272		// current file."
   273		// We only support ETag versions.
   274		// The caller must have set the ETag on the response already.
   275		if ir := r.Header.get("If-Range"); ir != "" && ir != etag {
   276			// TODO(bradfitz): handle If-Range requests with Last-Modified
   277			// times instead of ETags? I'd rather not, at least for
   278			// now. That seems like a bug/compromise in the RFC 2616, and
   279			// I've never heard of anybody caring about that (yet).
   280			rangeReq = ""
   281		}
   282	
   283		if inm := r.Header.get("If-None-Match"); inm != "" {
   284			// Must know ETag.
   285			if etag == "" {
   286				return rangeReq, false
   287			}
   288	
   289			// TODO(bradfitz): non-GET/HEAD requests require more work:
   290			// sending a different status code on matches, and
   291			// also can't use weak cache validators (those with a "W/
   292			// prefix).  But most users of ServeContent will be using
   293			// it on GET or HEAD, so only support those for now.
   294			if r.Method != "GET" && r.Method != "HEAD" {
   295				return rangeReq, false
   296			}
   297	
   298			// TODO(bradfitz): deal with comma-separated or multiple-valued
   299			// list of If-None-match values.  For now just handle the common
   300			// case of a single item.
   301			if inm == etag || inm == "*" {
   302				h := w.Header()
   303				delete(h, "Content-Type")
   304				delete(h, "Content-Length")
   305				w.WriteHeader(StatusNotModified)
   306				return "", true
   307			}
   308		}
   309		return rangeReq, false
   310	}
   311	
   312	// name is '/'-separated, not filepath.Separator.
   313	func serveFile(w ResponseWriter, r *Request, fs FileSystem, name string, redirect bool) {
   314		const indexPage = "/index.html"
   315	
   316		// redirect .../index.html to .../
   317		// can't use Redirect() because that would make the path absolute,
   318		// which would be a problem running under StripPrefix
   319		if strings.HasSuffix(r.URL.Path, indexPage) {
   320			localRedirect(w, r, "./")
   321			return
   322		}
   323	
   324		f, err := fs.Open(name)
   325		if err != nil {
   326			// TODO expose actual error?
   327			NotFound(w, r)
   328			return
   329		}
   330		defer f.Close()
   331	
   332		d, err1 := f.Stat()
   333		if err1 != nil {
   334			// TODO expose actual error?
   335			NotFound(w, r)
   336			return
   337		}
   338	
   339		if redirect {
   340			// redirect to canonical path: / at end of directory url
   341			// r.URL.Path always begins with /
   342			url := r.URL.Path
   343			if d.IsDir() {
   344				if url[len(url)-1] != '/' {
   345					localRedirect(w, r, path.Base(url)+"/")
   346					return
   347				}
   348			} else {
   349				if url[len(url)-1] == '/' {
   350					localRedirect(w, r, "../"+path.Base(url))
   351					return
   352				}
   353			}
   354		}
   355	
   356		// use contents of index.html for directory, if present
   357		if d.IsDir() {
   358			index := name + indexPage
   359			ff, err := fs.Open(index)
   360			if err == nil {
   361				defer ff.Close()
   362				dd, err := ff.Stat()
   363				if err == nil {
   364					name = index
   365					d = dd
   366					f = ff
   367				}
   368			}
   369		}
   370	
   371		// Still a directory? (we didn't find an index.html file)
   372		if d.IsDir() {
   373			if checkLastModified(w, r, d.ModTime()) {
   374				return
   375			}
   376			dirList(w, f)
   377			return
   378		}
   379	
   380		// serverContent will check modification time
   381		serveContent(w, r, d.Name(), d.ModTime(), d.Size(), f)
   382	}
   383	
   384	// localRedirect gives a Moved Permanently response.
   385	// It does not convert relative paths to absolute paths like Redirect does.
   386	func localRedirect(w ResponseWriter, r *Request, newPath string) {
   387		if q := r.URL.RawQuery; q != "" {
   388			newPath += "?" + q
   389		}
   390		w.Header().Set("Location", newPath)
   391		w.WriteHeader(StatusMovedPermanently)
   392	}
   393	
   394	// ServeFile replies to the request with the contents of the named file or directory.
   395	func ServeFile(w ResponseWriter, r *Request, name string) {
   396		dir, file := filepath.Split(name)
   397		serveFile(w, r, Dir(dir), file, false)
   398	}
   399	
   400	type fileHandler struct {
   401		root FileSystem
   402	}
   403	
   404	// FileServer returns a handler that serves HTTP requests
   405	// with the contents of the file system rooted at root.
   406	//
   407	// To use the operating system's file system implementation,
   408	// use http.Dir:
   409	//
   410	//     http.Handle("/", http.FileServer(http.Dir("/tmp")))
   411	func FileServer(root FileSystem) Handler {
   412		return &fileHandler{root}
   413	}
   414	
   415	func (f *fileHandler) ServeHTTP(w ResponseWriter, r *Request) {
   416		upath := r.URL.Path
   417		if !strings.HasPrefix(upath, "/") {
   418			upath = "/" + upath
   419			r.URL.Path = upath
   420		}
   421		serveFile(w, r, f.root, path.Clean(upath), true)
   422	}
   423	
   424	// httpRange specifies the byte range to be sent to the client.
   425	type httpRange struct {
   426		start, length int64
   427	}
   428	
   429	func (r httpRange) contentRange(size int64) string {
   430		return fmt.Sprintf("bytes %d-%d/%d", r.start, r.start+r.length-1, size)
   431	}
   432	
   433	func (r httpRange) mimeHeader(contentType string, size int64) textproto.MIMEHeader {
   434		return textproto.MIMEHeader{
   435			"Content-Range": {r.contentRange(size)},
   436			"Content-Type":  {contentType},
   437		}
   438	}
   439	
   440	// parseRange parses a Range header string as per RFC 2616.
   441	func parseRange(s string, size int64) ([]httpRange, error) {
   442		if s == "" {
   443			return nil, nil // header not present
   444		}
   445		const b = "bytes="
   446		if !strings.HasPrefix(s, b) {
   447			return nil, errors.New("invalid range")
   448		}
   449		var ranges []httpRange
   450		for _, ra := range strings.Split(s[len(b):], ",") {
   451			ra = strings.TrimSpace(ra)
   452			if ra == "" {
   453				continue
   454			}
   455			i := strings.Index(ra, "-")
   456			if i < 0 {
   457				return nil, errors.New("invalid range")
   458			}
   459			start, end := strings.TrimSpace(ra[:i]), strings.TrimSpace(ra[i+1:])
   460			var r httpRange
   461			if start == "" {
   462				// If no start is specified, end specifies the
   463				// range start relative to the end of the file.
   464				i, err := strconv.ParseInt(end, 10, 64)
   465				if err != nil {
   466					return nil, errors.New("invalid range")
   467				}
   468				if i > size {
   469					i = size
   470				}
   471				r.start = size - i
   472				r.length = size - r.start
   473			} else {
   474				i, err := strconv.ParseInt(start, 10, 64)
   475				if err != nil || i > size || i < 0 {
   476					return nil, errors.New("invalid range")
   477				}
   478				r.start = i
   479				if end == "" {
   480					// If no end is specified, range extends to end of the file.
   481					r.length = size - r.start
   482				} else {
   483					i, err := strconv.ParseInt(end, 10, 64)
   484					if err != nil || r.start > i {
   485						return nil, errors.New("invalid range")
   486					}
   487					if i >= size {
   488						i = size - 1
   489					}
   490					r.length = i - r.start + 1
   491				}
   492			}
   493			ranges = append(ranges, r)
   494		}
   495		return ranges, nil
   496	}
   497	
   498	// countingWriter counts how many bytes have been written to it.
   499	type countingWriter int64
   500	
   501	func (w *countingWriter) Write(p []byte) (n int, err error) {
   502		*w += countingWriter(len(p))
   503		return len(p), nil
   504	}
   505	
   506	// rangesMIMESize returns the nunber of bytes it takes to encode the
   507	// provided ranges as a multipart response.
   508	func rangesMIMESize(ranges []httpRange, contentType string, contentSize int64) (encSize int64) {
   509		var w countingWriter
   510		mw := multipart.NewWriter(&w)
   511		for _, ra := range ranges {
   512			mw.CreatePart(ra.mimeHeader(contentType, contentSize))
   513			encSize += ra.length
   514		}
   515		mw.Close()
   516		encSize += int64(w)
   517		return
   518	}
   519	
   520	func sumRangesSize(ranges []httpRange) (size int64) {
   521		for _, ra := range ranges {
   522			size += ra.length
   523		}
   524		return
   525	}

View as plain text