Managing a list

Yesterday I was reading about how to implement the MatrixSSL lib in Monkey, well it's pretty easy because we have been thinking and coding lots of idea that came up in the last two months, one of this things it a better API for plug-ins, which it's the easy way to implement a new layer for networking within Monkey. Well it was that while a big doubt it came to my mind, "how can I manage the poll of Ssl_t types per connection?" the answer was easy, use a thread key that should be created in "core_prctx" which it's one of the part of our API for plug-ins, this should contain a list, well as a curious mind I was wondering if we should implement this with macros in C or something like that when freyes came with an idea that he has hear about a lib that does exactly that but edsiper answer that it was in the linux kernel, well, both are right uthash it's one project and the other it's a tiny header file in the kernel, of course, tiny it's better and this blog post it's about this curious implementation.

Everything start by defining something that we can call our list, something that for our example it will use the Monkey prefix:

struct mk_list {
  struct mk_list *prev, *next;
};

Yes as you can see it's tiny and you only can go forward or backward but never store something there, the magic it comes when you only store the pointer of your current position and how to go forward or backward, funny right? well it's more easy with code I think.

First we need to define at least to functions, one to initialize the first node and other to add nodes, here it's the code:

static inline void mk_list_init( struct mk_list *list ) {
  list->next = list;
  list->prev = list;
}

static inline void __mk_list_add( struct mk_list *new, struct mk_list *prev, struct mk_list *next ) {
  next->prev = new;
  new->next = next;
  new->prev = prev;
  prev->next = new;
}

static inline void mk_list_add( struct mk_list *new, struct mk_list *head ) {
  __mk_list_add( new, head->prev, head );
}

With these two function we can work with our list, I just build a tiny example with a struct that will store an incremental value in a bucle.

struct ssl_connection {
  int connection;
  struct mk_list cons;
};

void main( void ) {
  int i;
  struct mk_list *lista = (struct mk_list *) malloc(sizeof(struct mk_list *));
  struct mk_list *curr;

  struct ssl_connection *con = NULL;
  mk_list_init( lista );

  for( i = 0; i < 20; i++ ) {
    con = (struct ssl_connection *) malloc(sizeof(struct ssl_connection *));
    con->connection = (i * 10);
    mk_list_add( &con->cons, lista );
  }

  mk_list_foreach( curr, lista ) {
    con = mk_list_entry( curr, struct ssl_connection, cons );
    printf( "Connection: %d\n", con->connection );
  }
}

Well I initialize the mk_list and ssl_connection struct for 20 rounds I multiply the value of i for 10 and store in a ssl_connection struct, but here it comes the magic, our function mk_list only accept mk_list struct, you can check that I just store in the "lista" the pointer of cons inside of our ssl_connection, and then jump again to the begin of the for and allocate new memory for our next ssl_connection. What the hell, how you suppose to get the data? well there's two define's that I haven't mention yet

#define mk_list_foreach(curr, head) for( curr = (head)->next; curr != (head); curr = curr->next )

#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)

#define container_of(ptr, type, member) ({                      \
      const typeof( ((type *)0)->member ) *__mptr = (ptr);      \
      (type *)( (char *)__mptr - offsetof(type,member) );})

#define mk_list_entry( ptr, type, member ) container_of( ptr, type, member )

Well three actually the mk_listforeach it's a define that edsiper wanted and freyes and I didn't have any argument against :P.

The container_of it's the big magic it's explained by Greg K-H (http://www.kroah.com/log/linux/container_of.html) in a simplify way this what it does it's look for our ssl_connection struct in the space of memory pointer in our current position in the list using the mk_list member of ssl_connection list as a reference. this will cause that by pointer arithmetic we can access the complet struct of that type with that member inside, sweet right?

Well what's the heavy point complicating with this? well there's a lot of problems with the memory when you manage a double linked list, especially because you must always create the "add" function for your specific type of struct, well with this you just create a member of that struct with the type mk_list and you can manage as a double-linked list =).

Well that's isn't the only super! you can always look that the linux kernel code asking for new ideas on how to manage this kind of list, this it's a tiny example of how to use it =)

Note for me: Can I use line numbers with the BEGIN_SRC macro inside of org-mode?

Simple and easy captcha for Django

Last night around 4 am I was asking about how to implement a captcha within Django, I started to search in the web and found two projects, both do a great job but were too complicated to implement since you must install them and join to your project as another app, which wasn't what I was looking for.

I was looking for a function to display a captcha just calling a function, this is because I want it to use in other place of the project without any hard work. Well, after look into some graphic library I found python-gd which comes with a very practical example (which I used to create the random list of fonts), now with the library just need it to finish the idea.

def captcha( request ):
    fonts = [
        '/usr/share/fonts/truetype/ttf-liberation/LiberationMono-Bold.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationMono-BoldItalic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationMono-Italic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationMono-Regular.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSans-Bold.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSans-BoldItalic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSans-Italic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSans-Regular.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSerif-Bold.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSerif-BoldItalic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSerif-Italic.ttf',
        '/usr/share/fonts/truetype/ttf-liberation/LiberationSerif-Regular.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSans.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSans-Bold.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSansMono.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSansMono-Bold.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif.ttf',
        '/usr/share/fonts/truetype/ttf-dejavu/DejaVuSerif-Bold.ttf',
        ]

    fontpath = '.'
    fontlist = dict()
    i = 0
    for f in fonts:
        if os.path.exists(f):
            fontpath = fontpath + ':' + os.path.dirname(f)
            fontlist[i] = f
            i = i + 1

    os.environ["GDFONTPATH"] = fontpath

    im = gd.image((150, 80))

    captcha_str = str( random.randint( 0, 900000))

    bg_color = im.colorAllocate(( random.randint( 200, 250 ), random.randint( 200, 250 ), random.randint( 200, 250 ) ))
    im.filledRectangle((0,0),(199,99), bg_color)

    i = 15
    for c in captcha_str:
        font_color = im.colorAllocate( (random.randint( 0, 100 ), random.randint(0,100),random.randint(0,100)) )
        font = fontlist[ random.randint( 0, len(fontlist) - 1)]
        im.string_ttf( font, 30.0, random.uniform(-0.5,0.5), (i,50), c, font_color)
        i = i + 20

    request.session['captcha_str'] = captcha_str

    str_img = cStringIO.StringIO()
    im.writePng( str_img )
    str_img.seek(0)

    response = HttpResponse()
    response['Content-Type'] = 'image/png'
    response.write( str_img.read() )
    return response

# apt-get install ttf-dejavu-core ttf-liberation

url('^captcha.png', views.captcha),

<img src="htt://host/captcha.png" alt="Captcha"/>

Xapian with init script

I wrote an init script for xapian-tcpsrv that runs on CentOS ( don't ask why CentOS ) well after a few test I can say that it works just fine even with HeartBeat.

In case that you don't know xapian is a search engine that you can use with many languages, if you need a search engine just check out this great project.

#!/bin/bash
#
# chkconfig: - 91 35
# description: Starts and stops xapian-tcpsrv.
# processname: xapian-tcpsrv
# pidfile: /var/run/xapian-tcpsrv.pid

# Source function library.
. /etc/init.d/functions

PORT="12345"
DATABASE_PATH="<xapian db path>"
USER="<user to run xapian-tcpsrv>"

CONFIG="/etc/sysconfig/xapian"

[ -r "$CONFIG" ] && source "$CONFIG"

xapiantcpsrv=/usr/bin/xapian-tcpsrv
xapianargs="--port $PORT --quiet $DATABASE_PATH"
pidfile=${PIDFILE-/var/run/xapian-tcpsrv.pid}
lockfile=${LOCKFILE-/var/lock/subsys/xapian-tcpsrv}
RETVAL=0

start() {
        echo -n $"Starting xapian-tcpsrv: "
        xapian-check $DATABASE_PATH > /dev/null 2>&1
        [ "$?" -eq 0 ] && daemon --user=$USER "$xapiantcpsrv $xapianargs </dev/null >/dev/null 2>&1 &" || failure $"xapiantcpsrv start"
        RETVAL=$?
        [ $RETVAL = 0 ] && touch ${lockfile}
}

stop() {
        echo -n $"Stopping xapain:"
        killproc -d 10 $xapiantcpsrv
        RETVAL=$?
        echo
        [ $RETVAL = 0 ] && rm -f ${lockfile} ${pidfile}
}

case "$1" in
  start)
        start
        ;;
  stop)
        stop
        ;;
  status)
        status $xapiantcpsrv
        RETVAL=$?
        ;;
  restart)
        stop
        start
        ;;
  condrestart)
        if [ -f ${pidfile} ] ; then
                stop
                start
        fi
        ;;
  *)
        echo $"Usage: $0 {start|stop|restart|status}"
        exit 1
esac

##
## ldap_stress.py
## Login : <zeus@lunareja.example.org>
## Started on  Sat Sep 12 17:07:19 2009 Jonathan Gonzalez V.
## $Id$
##
## Copyright (C) 2009 Jonathan Gonzalez V. <jonathan@blueplanet.cl>
## This program is free software: you can redistribute it and/or modify
## it under the terms of the GNU Affero General Public License as
## published by the Free Software Foundation, either version 3 of the
## License, or (at your option) any later version.
##
## This program is distributed in the hope that it will be useful,
## but WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
## GNU Affero General Public License for more details.
##
## You should have received a copy of the GNU Affero General Public License
## along with this program.  If not, see <http://www.gnu.org/licenses/>.
##

import threading
import time
import ldap

uri="ldaps://<host>:<port>"
binddn="<binddn>"
password="<password>"
basedn="<basedn>"
ldap_filter="(&(objectClass=posixAccount)(|(uid=unixuser)(personID=1234)))"

class TestConnections( threading.Thread ):
    def __init__ (self, number):
        threading.Thread.__init__(self)
        self.number = number

    def run(self):
        time.sleep(5)
        for i in range(1, 2):
            try:
                l = ldap.initialize( uri, 1)
            except:
                print "Can't Connect ?"

            ldap.set_option( ldap.OPT_X_TLS_REQUIRE_CERT, 0 )
            ldap.set_option( ldap.OPT_NETWORK_TIMEOUT, 10 )

            try:
                l.bind_s( binddn, password )
            except ldap.SERVER_DOWN:
                print "Server down!"
            except:
                print "Otro Error!"

            try:
                r = l.search_s( basedn, ldap.SCOPE_SUBTREE, ldap_filter )
            except:
                print "Can't get results"

if __name__ == '__main__':
    for x in range (1, 20):
        test = TestConnections (x)
        test.start ()

Blogging again!

After trying OrgMode to blog I decided to keep using Muse, the decision was pretty simple, I want to use something easy and fast, OrgMode it's easy but it doesn't have the journal method to export/publish text so, that's the why… besides, this work perfectly to me!! :D

My old text browser

A few days ago I started to get angry with my browser at that time FireFox, everything started because using my laptop with GNU Emacs and FireFox was a real pain, so I took the decision of use a text browser inside of my GNU Emacs.