sniff_test.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  package http_test
     6  
     7  import (
     8  	"bytes"
     9  	"fmt"
    10  	"io"
    11  	"log"
    12  	. "net/http"
    13  	"slices"
    14  	"strconv"
    15  	"strings"
    16  	"testing"
    17  )
    18  
    19  var sniffTests = []struct {
    20  	desc        string
    21  	data        []byte
    22  	contentType string
    23  }{
    24  	// Some nonsense.
    25  	{"Empty", []byte{}, "text/plain; charset=utf-8"},
    26  	{"Binary", []byte{1, 2, 3}, "application/octet-stream"},
    27  
    28  	{"HTML document #1", []byte(`<HtMl><bOdY>blah blah blah</body></html>`), "text/html; charset=utf-8"},
    29  	{"HTML document #2", []byte(`<HTML></HTML>`), "text/html; charset=utf-8"},
    30  	{"HTML document #3 (leading whitespace)", []byte(`   <!DOCTYPE HTML>...`), "text/html; charset=utf-8"},
    31  	{"HTML document #4 (leading CRLF)", []byte("\r\n<html>..."), "text/html; charset=utf-8"},
    32  
    33  	{"Plain text", []byte(`This is not HTML. It has ☃ though.`), "text/plain; charset=utf-8"},
    34  
    35  	{"XML", []byte("\n<?xml!"), "text/xml; charset=utf-8"},
    36  
    37  	// Image types.
    38  	{"Windows icon", []byte("\x00\x00\x01\x00"), "image/x-icon"},
    39  	{"Windows cursor", []byte("\x00\x00\x02\x00"), "image/x-icon"},
    40  	{"BMP image", []byte("BM..."), "image/bmp"},
    41  	{"GIF 87a", []byte(`GIF87a`), "image/gif"},
    42  	{"GIF 89a", []byte(`GIF89a...`), "image/gif"},
    43  	{"WEBP image", []byte("RIFF\x00\x00\x00\x00WEBPVP"), "image/webp"},
    44  	{"PNG image", []byte("\x89PNG\x0D\x0A\x1A\x0A"), "image/png"},
    45  	{"JPEG image", []byte("\xFF\xD8\xFF"), "image/jpeg"},
    46  
    47  	// Audio types.
    48  	{"MIDI audio", []byte("MThd\x00\x00\x00\x06\x00\x01"), "audio/midi"},
    49  	{"MP3 audio/MPEG audio", []byte("ID3\x03\x00\x00\x00\x00\x0f"), "audio/mpeg"},
    50  	{"WAV audio #1", []byte("RIFFb\xb8\x00\x00WAVEfmt \x12\x00\x00\x00\x06"), "audio/wave"},
    51  	{"WAV audio #2", []byte("RIFF,\x00\x00\x00WAVEfmt \x12\x00\x00\x00\x06"), "audio/wave"},
    52  	{"AIFF audio #1", []byte("FORM\x00\x00\x00\x00AIFFCOMM\x00\x00\x00\x12\x00\x01\x00\x00\x57\x55\x00\x10\x40\x0d\xf3\x34"), "audio/aiff"},
    53  
    54  	{"OGG audio", []byte("OggS\x00\x02\x00\x00\x00\x00\x00\x00\x00\x00\x7e\x46\x00\x00\x00\x00\x00\x00\x1f\xf6\xb4\xfc\x01\x1e\x01\x76\x6f\x72"), "application/ogg"},
    55  	{"Must not match OGG", []byte("owow\x00"), "application/octet-stream"},
    56  	{"Must not match OGG", []byte("oooS\x00"), "application/octet-stream"},
    57  	{"Must not match OGG", []byte("oggS\x00"), "application/octet-stream"},
    58  
    59  	// Video types.
    60  	{"MP4 video", []byte("\x00\x00\x00\x18ftypmp42\x00\x00\x00\x00mp42isom<\x06t\xbfmdat"), "video/mp4"},
    61  	{"AVI video #1", []byte("RIFF,O\n\x00AVI LISTÀ"), "video/avi"},
    62  	{"AVI video #2", []byte("RIFF,\n\x00\x00AVI LISTÀ"), "video/avi"},
    63  
    64  	// Font types.
    65  	// {"MS.FontObject", []byte("\x00\x00")},
    66  	{"TTF sample  I", []byte("\x00\x01\x00\x00\x00\x17\x01\x00\x00\x04\x01\x60\x4f"), "font/ttf"},
    67  	{"TTF sample II", []byte("\x00\x01\x00\x00\x00\x0e\x00\x80\x00\x03\x00\x60\x46"), "font/ttf"},
    68  
    69  	{"OTTO sample  I", []byte("\x4f\x54\x54\x4f\x00\x0e\x00\x80\x00\x03\x00\x60\x42\x41\x53\x45"), "font/otf"},
    70  
    71  	{"woff sample  I", []byte("\x77\x4f\x46\x46\x00\x01\x00\x00\x00\x00\x30\x54\x00\x0d\x00\x00"), "font/woff"},
    72  	{"woff2 sample", []byte("\x77\x4f\x46\x32\x00\x01\x00\x00\x00"), "font/woff2"},
    73  	{"wasm sample", []byte("\x00\x61\x73\x6d\x01\x00"), "application/wasm"},
    74  
    75  	// Archive types
    76  	{"RAR v1.5-v4.0", []byte("Rar!\x1A\x07\x00"), "application/x-rar-compressed"},
    77  	{"RAR v5+", []byte("Rar!\x1A\x07\x01\x00"), "application/x-rar-compressed"},
    78  	{"Incorrect RAR v1.5-v4.0", []byte("Rar \x1A\x07\x00"), "application/octet-stream"},
    79  	{"Incorrect RAR v5+", []byte("Rar \x1A\x07\x01\x00"), "application/octet-stream"},
    80  }
    81  
    82  func TestDetectContentType(t *testing.T) {
    83  	for _, tt := range sniffTests {
    84  		ct := DetectContentType(tt.data)
    85  		if ct != tt.contentType {
    86  			t.Errorf("%v: DetectContentType = %q, want %q", tt.desc, ct, tt.contentType)
    87  		}
    88  	}
    89  }
    90  
    91  func TestServerContentTypeSniff(t *testing.T) { run(t, testServerContentTypeSniff) }
    92  func testServerContentTypeSniff(t *testing.T, mode testMode) {
    93  	cst := newClientServerTest(t, mode, HandlerFunc(func(w ResponseWriter, r *Request) {
    94  		i, _ := strconv.Atoi(r.FormValue("i"))
    95  		tt := sniffTests[i]
    96  		n, err := w.Write(tt.data)
    97  		if n != len(tt.data) || err != nil {
    98  			log.Fatalf("%v: Write(%q) = %v, %v want %d, nil", tt.desc, tt.data, n, err, len(tt.data))
    99  		}
   100  	}))
   101  	defer cst.close()
   102  
   103  	for i, tt := range sniffTests {
   104  		resp, err := cst.c.Get(cst.ts.URL + "/?i=" + strconv.Itoa(i))
   105  		if err != nil {
   106  			t.Errorf("%v: %v", tt.desc, err)
   107  			continue
   108  		}
   109  		// DetectContentType is defined to return
   110  		// text/plain; charset=utf-8 for an empty body,
   111  		// but as of Go 1.10 the HTTP server has been changed
   112  		// to return no content-type at all for an empty body.
   113  		// Adjust the expectation here.
   114  		wantContentType := tt.contentType
   115  		if len(tt.data) == 0 {
   116  			wantContentType = ""
   117  		}
   118  		if ct := resp.Header.Get("Content-Type"); ct != wantContentType {
   119  			t.Errorf("%v: Content-Type = %q, want %q", tt.desc, ct, wantContentType)
   120  		}
   121  		data, err := io.ReadAll(resp.Body)
   122  		if err != nil {
   123  			t.Errorf("%v: reading body: %v", tt.desc, err)
   124  		} else if !bytes.Equal(data, tt.data) {
   125  			t.Errorf("%v: data is %q, want %q", tt.desc, data, tt.data)
   126  		}
   127  		resp.Body.Close()
   128  	}
   129  }
   130  
   131  // Issue 5953: shouldn't sniff if the handler set a Content-Type header,
   132  // even if it's the empty string.
   133  func TestServerIssue5953(t *testing.T) { run(t, testServerIssue5953) }
   134  func testServerIssue5953(t *testing.T, mode testMode) {
   135  	cst := newClientServerTest(t, mode, HandlerFunc(func(w ResponseWriter, r *Request) {
   136  		w.Header()["Content-Type"] = []string{""}
   137  		fmt.Fprintf(w, "<html><head></head><body>hi</body></html>")
   138  	}))
   139  
   140  	resp, err := cst.c.Get(cst.ts.URL)
   141  	if err != nil {
   142  		t.Fatal(err)
   143  	}
   144  
   145  	got := resp.Header["Content-Type"]
   146  	want := []string{""}
   147  	if !slices.Equal(got, want) {
   148  		t.Errorf("Content-Type = %q; want %q", got, want)
   149  	}
   150  	resp.Body.Close()
   151  }
   152  
   153  type byteAtATimeReader struct {
   154  	buf []byte
   155  }
   156  
   157  func (b *byteAtATimeReader) Read(p []byte) (n int, err error) {
   158  	if len(p) < 1 {
   159  		return 0, nil
   160  	}
   161  	if len(b.buf) == 0 {
   162  		return 0, io.EOF
   163  	}
   164  	p[0] = b.buf[0]
   165  	b.buf = b.buf[1:]
   166  	return 1, nil
   167  }
   168  
   169  func TestContentTypeWithVariousSources(t *testing.T) { run(t, testContentTypeWithVariousSources) }
   170  func testContentTypeWithVariousSources(t *testing.T, mode testMode) {
   171  	const (
   172  		input    = "\n<html>\n\t<head>\n"
   173  		expected = "text/html; charset=utf-8"
   174  	)
   175  
   176  	for _, test := range []struct {
   177  		name    string
   178  		handler func(ResponseWriter, *Request)
   179  	}{{
   180  		name: "write",
   181  		handler: func(w ResponseWriter, r *Request) {
   182  			// Write the whole input at once.
   183  			n, err := w.Write([]byte(input))
   184  			if int(n) != len(input) || err != nil {
   185  				t.Errorf("w.Write(%q) = %v, %v want %d, nil", input, n, err, len(input))
   186  			}
   187  		},
   188  	}, {
   189  		name: "write one byte at a time",
   190  		handler: func(w ResponseWriter, r *Request) {
   191  			// Write the input one byte at a time.
   192  			buf := []byte(input)
   193  			for i := range buf {
   194  				n, err := w.Write(buf[i : i+1])
   195  				if n != 1 || err != nil {
   196  					t.Errorf("w.Write(%q) = %v, %v want 1, nil", input, n, err)
   197  				}
   198  			}
   199  		},
   200  	}, {
   201  		name: "copy from Reader",
   202  		handler: func(w ResponseWriter, r *Request) {
   203  			// Use io.Copy from a plain Reader.
   204  			type readerOnly struct{ io.Reader }
   205  			buf := bytes.NewBuffer([]byte(input))
   206  			n, err := io.Copy(w, readerOnly{buf})
   207  			if int(n) != len(input) || err != nil {
   208  				t.Errorf("io.Copy(w, %q) = %v, %v want %d, nil", input, n, err, len(input))
   209  			}
   210  		},
   211  	}, {
   212  		name: "copy from bytes.Buffer",
   213  		handler: func(w ResponseWriter, r *Request) {
   214  			// Use io.Copy from a bytes.Buffer to trigger ReadFrom.
   215  			buf := bytes.NewBuffer([]byte(input))
   216  			n, err := io.Copy(w, buf)
   217  			if int(n) != len(input) || err != nil {
   218  				t.Errorf("io.Copy(w, %q) = %v, %v want %d, nil", input, n, err, len(input))
   219  			}
   220  		},
   221  	}, {
   222  		name: "copy one byte at a time",
   223  		handler: func(w ResponseWriter, r *Request) {
   224  			// Use io.Copy from a Reader that returns one byte at a time.
   225  			n, err := io.Copy(w, &byteAtATimeReader{[]byte(input)})
   226  			if int(n) != len(input) || err != nil {
   227  				t.Errorf("io.Copy(w, %q) = %v, %v want %d, nil", input, n, err, len(input))
   228  			}
   229  		},
   230  	}} {
   231  		t.Run(test.name, func(t *testing.T) {
   232  			cst := newClientServerTest(t, mode, HandlerFunc(test.handler))
   233  
   234  			resp, err := cst.c.Get(cst.ts.URL)
   235  			if err != nil {
   236  				t.Fatalf("Get: %v", err)
   237  			}
   238  			if ct := resp.Header.Get("Content-Type"); ct != expected {
   239  				t.Errorf("Content-Type = %q, want %q", ct, expected)
   240  			}
   241  			if want, got := resp.Header.Get("Content-Length"), fmt.Sprint(len(input)); want != got {
   242  				t.Errorf("Content-Length = %q, want %q", want, got)
   243  			}
   244  			data, err := io.ReadAll(resp.Body)
   245  			if err != nil {
   246  				t.Errorf("reading body: %v", err)
   247  			} else if !bytes.Equal(data, []byte(input)) {
   248  				t.Errorf("data is %q, want %q", data, input)
   249  			}
   250  			resp.Body.Close()
   251  
   252  		})
   253  
   254  	}
   255  }
   256  
   257  func TestSniffWriteSize(t *testing.T) { run(t, testSniffWriteSize) }
   258  func testSniffWriteSize(t *testing.T, mode testMode) {
   259  	cst := newClientServerTest(t, mode, HandlerFunc(func(w ResponseWriter, r *Request) {
   260  		size, _ := strconv.Atoi(r.FormValue("size"))
   261  		written, err := io.WriteString(w, strings.Repeat("a", size))
   262  		if err != nil {
   263  			t.Errorf("write of %d bytes: %v", size, err)
   264  			return
   265  		}
   266  		if written != size {
   267  			t.Errorf("write of %d bytes wrote %d bytes", size, written)
   268  		}
   269  	}))
   270  	for _, size := range []int{0, 1, 200, 600, 999, 1000, 1023, 1024, 512 << 10, 1 << 20} {
   271  		res, err := cst.c.Get(fmt.Sprintf("%s/?size=%d", cst.ts.URL, size))
   272  		if err != nil {
   273  			t.Fatalf("size %d: %v", size, err)
   274  		}
   275  		if _, err := io.Copy(io.Discard, res.Body); err != nil {
   276  			t.Fatalf("size %d: io.Copy of body = %v", size, err)
   277  		}
   278  		if err := res.Body.Close(); err != nil {
   279  			t.Fatalf("size %d: body Close = %v", size, err)
   280  		}
   281  	}
   282  }
   283
View as plain text