I have the following simple golang program to download Google's privacy policy. Unfortunately it always crashes with the error unexpected EOF
after reading 6861 bytes, even though the document is much longer. Why?
package main
import "net"
import "fmt"
import "io"
import "os"
func die(msg string, s os.Error) {
fmt.Printf("%s crashed: %v
", msg, s)
os.Exit(1)
}
func main() {
fd, err := net.Dial("tcp", "google.com:80")
if err != nil { die("dial", err) }
req := []byte("GET /intl/en/privacy/ HTTP/1.0
Host: www.google.com
")
_, err = fd.Write(req)
if err != nil { die("dial write", err) }
buf := make([]byte, 1024)
nr := 1
for nr > 0 {
nr, err = io.ReadFull(fd, buf)
if err != nil { die("dial read", err) }
fmt.Printf("read %d
", nr)
}
}
outputs:
read 1024
read 1024
read 1024
read 1024
read 1024
read 1024
dial read crashed: unexpected EOF
Function io.ReadFull(fd, buf)
should be used only when you know that fd
can feed at least len(buf)
bytes.
Instead, try the following:
var buf bytes.Buffer
nr, err := io.Copy(&buf, fd)
if err != nil {
die("dial read", err)
}
If copying to a file, use an os.File
instead of bytes.Buffer
.
The document is not much longer. It's only 412 bytes longer. Sine ReadFull cannot read all 1024 bytes it returns unexpected EOF
.
You can also get the size of the file from the HTTP headers, and read to that length on your socket. Look for Content-Length response header.
You should always know the size of what you are moving with IO, when it's a file and possible, and limit your read to the expected size. This will avoid a lot of issues. pseudocode:
$size = HTTP_RESPONSE['Content-Length'];
while(reading to $size)
{
do something with chunk;
}