dsrosario / clearsilver Goto Github PK

What steps will reproduce the problem?
1. write cgi with python and use neo_cgi.so
2. use firefox and request cgi with post method
3. check out the hdf

What is the expected output? What do you see instead?
we expected parsed information( key/value) from the post,but nothing  

What version of the product are you using? On what operating system?
clearsilver-0.10.5 linux  

Please provide any additional information below.

cgi/cgi.c  line: 643 

from :
    if (type && !strcmp(type, "application/x-www-form-urlencoded" )

change to :
    if (type && !strncmp(type, "application/x-www-form-urlencoded",33))


firefox send post like this :
    Content-Type: application/x-www-form-urlencoded; charset=UTF-8

What steps will reproduce the problem?
1. gcc -I/usr/pkg/include/ClearSilver -o test test.c -L/usr/pkg/lib  -lz
-lneo_cgi -lneo_cs -lneo_utl
2. put toto.hdf and toto.cs in the same directory than the test binary
3. ./test

What is the expected output? What do you see instead?
" <p>Ici je vais ajouter Mercurial à <a
href="http://brokk.etoilebsd.net/">brokk</a>.</p>"
instead I got : 
" <p>Ici je vais ajouter Mercurial à<a
href="http://brokk.etoilebsd.net/">brokk</a>.</p>"

a space is missing before '<a href'

What version of the product are you using? On what operating system?

clearsilver 0.10.5 from pkgsrc, on 
5.0_STABLE NetBSD 5.0_STABLE (XEN3_DOMU_PF) #0: Wed Oct 21 14:59:59

Please provide any additional information below.

the workaround is to add Config.WhiteSpaceStrip=0 to the hdf file

I tested the samething on FreeBSD with clearsilver 0.10.5 from ports I
haven't the problem

I can reproduce this with every à and ù not with é or è 
All the encoding are in utf8

$ cd /usr/local/lib/python2.4/site-packages/site-packages$ python -c "import 
neo_cgi"
Traceback (most recent call last):
  File "<string>", line 1, in ?
ImportError: ./neo_cgi.so: undefined symbol: crc32

I am sure you assumed it would come from zlib.h ... not on my system!

I imagine this has to do with Config.CompressionEnabled.

Linux 3.2.0-29-generic #46-Ubuntu SMP Fri Jul 27 17:03:23 UTC 2012 x86_64 
x86_64 x86_64 GNU/Linux

With data1 = <title>:

<?cs escape: "html" ?><?cs alt:data1 ?>nonsense<?cs /alt ?><?cs /escape ?>

doesn't escape title.

Memory leak in include_parse(). Added missing deallocations of arg1 members. 
Removed redundant initialization (arg1.argexpr = NULL).

What steps will reproduce the problem?
1. cgi_cookie_set(cgi, "cgi_cookie_test", "hellocgi", "/", "localhost", "Sat. 
06 Oct 2013 08:30:00", 1, 0)
2. http://localhost/cgi-bin/hellocs
3. Set-Cookie: cgi_cookie_test=hellocgi; path=/; expires=Sat. 06 Oct 2013 
08:30:00; domain=localhost Content-Type: text/html 
HTTP_COOKIE 
SESS29af1facda0a866a687d5055f2fade2c=EaXgcoenUyy9_AvsvrFvniUv6PjfS2n6aS3iunlvLu0
; pla3412=123456; pla3412_salt=lt260tdkeqc

What is the expected output? What do you see instead?
there is cgi_cookie_set in HTTP_COOKIE, but nope.

What version of the product are you using? On what operating system?
clearsilver-0.10.5
Linux 3.5.2-gentoo x86_64 Intel(R) Core(TM) i7-2630QM CPU @ 2.00GHz 
GenuineIntel GNU/Linux

Please provide any additional information below.

Hi,

I like ClearSilver an i use it for my websides.
I tryed to compile CS for python under windows but the README in foder
python is a bit to old. I even cant get some downloads. In trunk i foun a
file called README_VS.TXT. This sounds good, but i can't find the nessesary
patch file.

The only way I found is, to use cygwin. But i allways hav to use cygwin
when I would like to run a python-script using CS. And cygwin ist slow.

I would be greate if there would be a way compiling and unsing under
windows in combination with python (2.6) for windows

Thx

I would dearly love the ability to use delimiters other than <?cs ?>
I would like to use either <? ?> or (like eRuby) <% %>
The reason being that this means I can use editors that grok these types of 
delimiter as nothing groks <?cs
Am I right in saying that I only need to hack find_open_delim() in cs_parse and 
nowhere else?
I use Clearsilver with the Ruby API on Gentoo. I've hacked the Ruby API 
somewhat :)
I should have a fairly up-to-date version of my marauding on github soon.

The package clearsilver fails to compile with the new hardened compiler
flags dpkg-buildflag outputs [0].
The problematic flag is: -Werror=format-security
See the ubuntu buildlog:
https://launchpadlibrarian.net/85252523/buildlog_ubuntu-precise-i386.clearsilver
_0.10.5-1.2_FAILEDTOBUILD.txt.gz

Snippet:
 neo_cgi.c: In function 'p_cgi_error':
 neo_cgi.c:181:3: error: format not a string literal and no format arguments [-Werror=format-security]
 cc1: some warnings being treated as errors


The problem bould be solved with:

--- a/python/neo_cgi.c
+++ b/python/neo_cgi.c
@@ -178,7 +178,7 @@
   if (!PyArg_ParseTuple(args, "s:error(str)", &s))
     return NULL;

-  cgi_error (cgi, s);
+  cgi_error (cgi, "%s", s);
   rv = Py_None;
   Py_INCREF(rv);
   return rv;


Please, apply this patch as soon as possible.


Best regards,
Leo Iannacone



[0] http://lists.debian.org/debian-devel-announce/2011/09/msg00001.html


-- System Information:
Debian Release: wheezy/sid
  APT prefers oneiric
  APT policy: (500, 'oneiric')
Architecture: i386 (i686)

Kernel: Linux 3.0.0-12-generic (SMP w/2 CPU cores)
Locale: LANG=it_IT.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

patch:

=====
static NEOERR * _copy_nodes (HDF *dest, HDF *src)
{
  NEOERR *err = STATUS_OK;
  HDF *dt, *st;
  HDF_ATTR *attr_copy;

+  if (src == NULL)
+    return STATUS_OK;
  st = src->child;
  while (st != NULL)
  {
    err = _copy_attr(&attr_copy, st->attr);
=====

Running cs regression tests
Failed Regression Test: test_crc.cs
  See test_crc.cs.out and test_crc.cs.err
make: *** [test] Error 1

test_crc.cs.err
4c4
< -2128917020

---
> 2166050276

test_crc.cs.out
Parsing test_crc.cs
419156592
1357503972
2166050276


test_crc.cs.gold

Parsing test_crc.cs
419156592
1357503972
-2128917020

This is caused by n in struct _arg being a long int, which is 32 bits on
32bit systems, and 64 bits on 64bit systems.  Making n just an int causes
it to pass, and making it a long will change the gold but should make it
pass on both as well.  The consequences of making either of those changes
will have to be investigated.