So I thought I had CGI in Ruby figured out, but I'm stumped on the following:
I want use Ruby to dynamically serve an image. The image tag would
look like <img src="file.rb?id=200">, where 200 is the id for an
image.
So here's what I have (I'm trading embarrassment for guidance):
require "cgi"
cgi = CGI.new("html4")
cgi.out('image/png')
File.read('200.png')
Now, I'm pretty sure I'm not reading the file right, but what I seem
to be stuck on first is getting my MIME type right. When hitting the
CGI, I can't get Content-Type to come back as 'image/png'. It's always
'text/html'.
Any tips?
Shouldn't that be
cgi.header("image/png") {
File.read('200.png');
}
Or something similar?
J.
···
On Thu, Apr 21, 2005 at 04:54:24AM +0900, Michael Buffington wrote:
So I thought I had CGI in Ruby figured out, but I'm stumped on the following:
I want use Ruby to dynamically serve an image. The image tag would
look like <img src="file.rb?id=200">, where 200 is the id for an
image.
So here's what I have (I'm trading embarrassment for guidance):
require "cgi"
cgi = CGI.new("html4")
cgi.out('image/png')
File.read('200.png')
So I thought I had CGI in Ruby figured out, but I'm stumped on the
following:
I want use Ruby to dynamically serve an image. The image tag would
look like <img src="file.rb?id=200">, where 200 is the id for an
image.
So here's what I have (I'm trading embarrassment for guidance):
require "cgi"
cgi = CGI.new("html4")
cgi.out('image/png')
File.read('200.png')
Now, I'm pretty sure I'm not reading the file right, but what I seem
to be stuck on first is getting my MIME type right. When hitting the
CGI, I can't get Content-Type to come back as 'image/png'. It's always
'text/html'.
Any tips?
Try the following:
require "cgi"
cgi = CGI.new("html4")
cgi.out('image/png') do
File.read("#{cgi['id']}.png")
end
You have to supply the content in a block passed to out. I also added the
necessary code to read the id and open that file.
Ryan Leavengood
require "cgi"
cgi = CGI.new("html4")
cgi.out('image/png') do
File.read("#{cgi['id']}.png")
end
Still doesn't seem to nail it. Firefox reports the page as
'text/plain'. Hitting the URL directly doesn't pull the image, using
<img src=".."> produces a broken link. In Safari, it downloads
endlessly.
I'm not sure what CGI is doing behind the scenes with cgi.out. From
what I read in the docs, it seems like cgi.out is supposed to write
the header for you, but the header I'm getting is very sparse:
HTTP/1.x 200 OK
Connection: close
Server: lighttpd/1.3.13
I should at least see Content-Type: image/png in the header.
I also haven't ruled out lighttpd in this case, but it's much more
likely I'm doing something wrong, not lighttpd.
···
Date: Wed, 20 Apr 2005 19:20:41 GMT
I have found that you need to write the header manually, despite what
the docs say.
This snippet will save you a lot of hassle whilst developing under
CGI. It should trap errors and give a backtrace in the browser, as
well as indicate if you didn't send the header properly:
---- cgicroak.rb ----
require 'cgi'
class CGI
def self.croak(*args)
cgi = CGI.new(args)
old_stdout = $stdout
buffer = ''
def buffer.write(s)
self << s
end
$stdout = buffer
begin
yield(cgi)
unless (buffer =~ /^Content-type:.*?\r?\n\r?\n/i)
raise RuntimeError, "No header sent"
end
rescue Exception => e
buffer.replace('')
print(cgi.header('type' => 'text/plain'))
puts("Error\n\n")
puts("#{e.class.to_s}: #{e.to_s}\n\n")
puts(e.backtrace.join("\n"))
end
$stdout = old_stdout
$stdout.write(buffer)
end
end
---- index.cgi ----
#!/path/to/your/ruby
require 'croakcgi'
CGI.croak('html4') do |cgi|
# Your requires go here
print(cgi.header)
# Your code
end
--- EOF ---
Paul.
Oops - I referred to the file as both 'croakcgi' and 'cgicroak'.
Choose one or the other!
This is really great - seeing Ruby errors definitely makes things more clear.
So, to test this all out, here's what my "index.cgi" looks like:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "text/html") {"shizzle"}
end
This works as expected. The word shizzle makes it to the browser.
However, when I execute this:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "image/gif") {File.open("image.gif")}
end
I get:
Error NameError: undefined method `length'
I get the same thing with:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "image/gif", "length" => File.size("image.gif"))
{File.open("image.gif")}
end
In either cases, if I use cgi.header instead of cgi.out, I get an error saying:
"No header sent"
Any ideas about what's going on?
···
On 4/20/05, Paul Battley <pbattley@gmail.com> wrote:
I have found that you need to write the header manually, despite what
the docs say.
This snippet will save you a lot of hassle whilst developing under
CGI. It should trap errors and give a backtrace in the browser, as
well as indicate if you didn't send the header properly:
---- cgicroak.rb ----
require 'cgi'
class CGI
def self.croak(*args)
cgi = CGI.new(args)
old_stdout = $stdout
buffer = ''
def buffer.write(s)
self << s
end
$stdout = buffer
begin
yield(cgi)
unless (buffer =~ /^Content-type:.*?\r?\n\r?\n/i)
raise RuntimeError, "No header sent"
end
rescue Exception => e
buffer.replace('')
print(cgi.header('type' => 'text/plain'))
puts("Error\n\n")
puts("#{e.class.to_s}: #{e.to_s}\n\n")
puts(e.backtrace.join("\n"))
end
$stdout = old_stdout
$stdout.write(buffer)
end
end
---- index.cgi ----
#!/path/to/your/ruby
require 'croakcgi'
CGI.croak('html4') do |cgi|
# Your requires go here
print(cgi.header)
# Your code
end
--- EOF ---
Paul.
Looks I've figure most of it. This sort of works:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
print(cgi.header("type" => "image/gif", "length" =>
File.size("image.gif"), "status" => "304"))
File.read("agent.gif")
end
Problem is, the server reports that it can't display the image due to
errors. Any ideas what I'm doing wrong when I actually read the image
file in?
···
On 4/20/05, Michael Buffington <michael.buffington@gmail.com> wrote:
This is really great - seeing Ruby errors definitely makes things more clear.
So, to test this all out, here's what my "index.cgi" looks like:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "text/html") {"shizzle"}
end
This works as expected. The word shizzle makes it to the browser.
However, when I execute this:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "image/gif") {File.open("image.gif")}
end
I get:
Error NameError: undefined method `length'
I get the same thing with:
#! /usr/bin/ruby
require 'cgicroak'
CGI.croak('html4') do |cgi|
cgi.out("type" => "image/gif", "length" => File.size("image.gif"))
{File.open("image.gif")}
end
In either cases, if I use cgi.header instead of cgi.out, I get an error saying:
"No header sent"
Any ideas about what's going on?
On 4/20/05, Paul Battley <pbattley@gmail.com> wrote:
> I have found that you need to write the header manually, despite what
> the docs say.
>
> This snippet will save you a lot of hassle whilst developing under
> CGI. It should trap errors and give a backtrace in the browser, as
> well as indicate if you didn't send the header properly:
>
> ---- cgicroak.rb ----
>
> require 'cgi'
>
> class CGI
> def self.croak(*args)
> cgi = CGI.new(args)
> old_stdout = $stdout
> buffer = ''
> def buffer.write(s)
> self << s
> end
> $stdout = buffer
> begin
> yield(cgi)
> unless (buffer =~ /^Content-type:.*?\r?\n\r?\n/i)
> raise RuntimeError, "No header sent"
> end
> rescue Exception => e
> buffer.replace('')
> print(cgi.header('type' => 'text/plain'))
> puts("Error\n\n")
> puts("#{e.class.to_s}: #{e.to_s}\n\n")
> puts(e.backtrace.join("\n"))
> end
> $stdout = old_stdout
> $stdout.write(buffer)
> end
> end
>
> ---- index.cgi ----
>
> #!/path/to/your/ruby
>
> require 'croakcgi'
>
> CGI.croak('html4') do |cgi|
>
> # Your requires go here
>
> print(cgi.header)
>
> # Your code
>
> end
>
> --- EOF ---
>
> Paul.
>
>
You are reading the data, but not actually sending it to the client.
- File.read("agent.gif")
+ print(File.read("agent.gif"))
I think that's the problem. You can also get rid of the 304 status;
it's not appropriate:
http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
Finally, you are sending the length of a different file in the code
sample you gave.
You are reading the data, but not actually sending it to the client.
- File.read("agent.gif")
+ print(File.read("agent.gif"))
Adding print produces the same result - can't display the image due to errors.
I think that's the problem. You can also get rid of the 304 status;
it's not appropriate:
HTTP/1.1: Status Code Definitions
Fair enough - I added because when accessing the image directly, the
returned status was 304.
Interestingly, when I remove the 304, I get a broken image rather than
an error message.
Finally, you are sending the length of a different file in the code
sample you gave.
Woops. Email editing faux pas. The code is correct.
···
On 4/20/05, Paul Battley <pbattley@gmail.com> wrote:
I just ran up a similar example using a jpeg I had to hand (yes, it
really was called 'foo.jpeg'!).
require 'cgicroak'
CGI.croak do |cgi|
print(cgi.header('type' => 'image/jpeg'))
print(File.read('foo.jpeg'))
end
It works, and displays the image, so the code seems all right. Do
check things like file permissions, though - it could be that it is
failing to read the image from disk. And, of course, check the image
itself.