Gopher client tests (URLs)

These tests check how well your Gopher client handles gopher
URLs per RFC 4266[1] and RFC 3986[2]. (Obviously, if your
client doesn't support Gopher URLs, these tests are moot.)

Copy and paste each URL in to your client. If your client
does not properly handle a URL, the result page will report
what the server both received and expected.

------------------------------------------------------------
                     Path info tests
------------------------------------------------------------

gopher://asciz.com/1/client-tests/pathinfo/foo/%20%252525%23text
gopher://asciz.com/1/client-tests/pathinfo/foo%2F%20%252525%23text
gopher://asciz.com/1/client-tests/pathinfo/foo/%20%252525%23text#fragment
gopher://asciz.com/1/client-tests/pathinfo/foo%2F%20%252525%23text#fragment

(You may be wondering why these URLs contain "%252525". I
use the same test endpoints for both URL tests (this page)
and selector tests (see below). At least one Gopher client
unescapes percent-encoded bytes _in a selector_ (contrary
to RFC 1436, which says "[t]he selector string ... should
never be modified by the client"). The Gopher server that
I'm running also unescapes percent-encoded bytes in the
path portion of a request; a "%25" in a selector would be
unescaped to "%" by the aforementioned unfortunate client,
and the server would attempt to unescape "%", resulting in
a "bad selector" error. So I use "%2525" in the selectors,
which must be percent-encoded as "%252525" in URLs, to allow
this test script to provide some useful information about
what went wrong with the request.)

(The same unfortunate client also strips off anything
following a "#" in a selector, which is further damage to
the selector.)

------------------------------------------------------------
                    Query string tests
------------------------------------------------------------

gopher://asciz.com/1/client-tests/query1/?query%20string%25%23text
gopher://asciz.com/1/client-tests/query1/%3Fquery%20string%25%23text
gopher://asciz.com/1/client-tests/query2/?query%20string%25?%23text
gopher://asciz.com/1/client-tests/query2/%3Fquery%20string%25%3F%23text

------------------------------------------------------------
                   Search string tests
------------------------------------------------------------

gopher://asciz.com/1/client-tests/search/%09search%20string%25/?%23text
gopher://asciz.com/1/client-tests/search/%09search%20string%25%2F%3F%23text

------------------------------------------------------------
             These tests as Gopher selectors
------------------------------------------------------------

gopher://asciz.com/1/client-tests/menu

------------------------------------------------------------
              Proper handling of Gopher URLs
------------------------------------------------------------

According to RFC 4266, a Gopher URL takes the form:

   gopher://<host>:<port>/<gopher-path>

where <gopher-path> is one of:

   <gophertype><selector>
   <gophertype><selector>%09<search>

Per RFC 4266 and RFC 3986, the URL must be percent-encoded
(but note that the <TAB> byte that separates the <selector>
from the <search> is already percent-encoded in the above
description and should not be encoded a second time).

To properly handle a Gopher URL, a client must decode
percent-encoded bytes in the <gopher-path> (e.g., %2F
decodes to /) and then send the decoded bytes to the Gopher
server. It also must NOT handle most "special" characters,
such as ? or /, differently from any other character; the
Gopher protocol does not reserve these and treats them as
regular characters.

On the other hand, RFC 3986 specifies that a number sign
("#") is used to delimit a fragment and that:

   the fragment identifier is separated from the rest of
   the URI prior to a dereference, and thus the identifying
   information within the fragment itself is dereferenced
   solely by the user agent, regardless of the URI scheme.

(from RFC 3986 section 3.5.)

Therefore if a number sign is part of a selector, it must be
percent-encoded in a Gopher URL, and a client must decode
it and send it as a number sign to the Gopher server. On
the other hand, if the number sign is used to delimit a
fragment, it must NOT be percent-encoded in a Gopher URL,
and a client must NOT send the number sign nor the fragment
identifier to the Gopher server.

Note: This script uses "path info" and/or a "query string"
(a la CGI) in the various tests above to identify what is
being tested. These are purely server-side constructs as
the Gopher protocol itself has no concept of them; however,
a client must send the full selector, unmodified, to the
server which allows the server to handle them as expected.

------------------------------------------------------------
                        References
------------------------------------------------------------

[1] gopher://asciz.com/0/rfc/rfc4266.txt
[2] gopher://asciz.com/0/rfc/rfc3986.txt

You are viewing proxied material from asciz.com. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.