Quick Scripts

Installing xmgrace on Snow Lepoard

2011-05-16T08:32:00.000-07:00

You want to install grace (xmgrace) on Snow Leopard, but all the information that you find online is obsolete and confusing.

Make sure you have MacPorts.
> sudo port install grace

¿Se escribe con o sin acento?

2011-04-29T06:15:00.000-07:00

Con acento	Sin acento
por qué - why, el porqué - the reason	porque - because
sí - yes, sí mismo - oneself	si - if
sólo - only	solo - alone
tú - you	tu - your

Run script in OS X on network connection

2011-03-22T12:13:00.000-07:00

You want to automatically run a script in OS X every time there is an update in your network connection (e.g. going on- and offline). In particular you may want to run a script to gather information about your environment, guess your current location and execute some tasks accordingly. Also, for some weird reason, you don't want to use an application such as MarcoPolo or Proxi in order to do this.

By far the most reliable way to do this is using crankd that, in a very similar way to launchd, allows you to run scripts in response to events such as network changes, filesystem activity, and application launching. The process to get started is not terribly complicated, but you need to be careful following instructions.

Download and install PyMacAdmin, this includes crankd.
Once installed, open a terminal and run /usr/local/sbin/crankd.py. For some reason on my computer it crashes with an error about an undefined NSNotificationHandler, however you can safely disregard that. If it didn't crash type Ctrl+C to exit the program. The important point is that it would have created an example configuration file in:
~/Library/Preferences/com.googlecode.pymacadmin.crankd.plist.

Open the configuration file in a text editor and modify it so that it looks like the following:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>SystemConfiguration</key>
    <dict>
        <key>State:/Network/Global/IPv4</key>
        <dict>
            <key>command</key>
            <string>/absolute/path/to/your/script</string>
        </dict>
    </dict>
</dict>
</plist>

This will cause your script to be called every time that there is a change in your connection, in particular when going on- and offline. If you want to test and debug your script, you can go back to your terminal and run /usr/local/sbin/crankd.py. On the output you will also get sensible messages about what crankd is doing. Try turning on/off your wireless connection, or plugging and unplugging your ethernet cable.

If everything seems to work, now you probably would like to have your system automatically run and keep alive the crankd daemon. For this we use, as usual, launchd. Create a new file such as ~/Library/LaunchAgents/org.crankd.plist with the following contents:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>KeepAlive</key>
    <true/>
    <key>Label</key>
    <string>org.crankd.plist</string>
    <key>ProgramArguments</key>
    <array>
        <string>/usr/local/sbin/crankd.py</string>
    </array>
    <key>RunAtLoad</key>
    <true/>
</dict>
</plist>

To start the daemon right now, type the following on the terminal:
> launchctl load ~/Library/LaunchAgents/org.crankd.plist
That's it. Turn your connection on/off to check that your script is working properly. The daemon will also be automatically started the next time you restart your computer.

By the way, if you are reading this you probably also want to know how, within your script, to detect whether you have an active connection and, if you do have a connection, to somehow guess your current location. Some good ideas are to check the contents of the file /etc/resolv.conf (the file doesn't exist if you're offline), or to have a look at the output from ifconfig.

Use a perl script and the Graph API to access your own Facebook account

2011-02-28T06:11:00.000-08:00

You want to create a script to read messages and post status updates to your own Facebook account, but you find the official documentation confusing and you aren't sure where to start. Search no more because here you'll find the easiest way to do just this.

The process is actually fairly simple, and recent updates to the Graph API explorer have made it dead easy:

If you haven’t done so, visit your applications page on Facebook to create a new application. The default settings are fine, so you can leave them unchanged.
Visit the Graph API explorer and on the top right corner choose your application. Click on the “Get access token” button, select the permissions you want to give to your application (depending on what you want to do you’ll probably need offline_access, read_stream and publish_stream), then confirm to get your access token. Copy and paste this value into the example code bellow.
That’s it! Now you can go and call any of the allowed methods from the Facebook Graph API. Also make sure to understand the difference between /home, /posts, /feed, and /statuses.

The following example code shows how to read posts from your news feed and post a new message to your own wall:

#!/usr/bin/perl

use strict;
use warnings;
use open qw(:std :utf8);
use LWP::Simple;
use YAML::Tiny;
use JSON;
use URI;
use utf8;

my $access_token = 'YOUR_ACCESS_TOKEN';

# Fetch your News Feed from Facebook
my $resp = graph_api('me/home', { access_token => $access_token });
for my $post (@{ $resp->{data} }) {
  # do something with each $post
  print Dump($post);
}

# Publish a new message to your own wall
graph_api('me/feed', {
  access_token => $access_token,
  message      => 'Hello World! I’m posting Facebook updates from a script!',
  link         => 'http://qscripts.blogspot.com/2011/02/post-to-your-own-facebook-account-from.html',
  picture      => 'http://navarroj.com/stuff/share-icon-128x128.png',
  name         => 'Post to your own Facebook account from a script',
  caption      => 'qscripts.blogspot.com',
  description  => 'You want to create a script to read messages and post status updates to your own '
                . 'Facebook account, but you find the official documentation confusing and '
                . 'you aren’t sure where to start. Search no more because here you’ll find '
                . 'the easiest way to do just this.',
  method       => 'post'
});

exit 0;

sub graph_api {
  my $uri = new URI('https://graph.facebook.com/' . shift);
  $uri->query_form(shift);
  my $resp = get("$uri");
  return defined $resp ? decode_json($resp) : undef;
}

Note that, to be able to make secure https requests, you'll also need to have installed either Net::SSL or LWP::Protocol::https. Finally, if everything works, you should be able to see something like this posted on your wall.

You’re welcome.

An Explosm RSS feed that doesn't suck

2011-02-24T08:33:00.000-08:00

I am a huge fan of Explosm.net and their awesome comics, but their RSS feed sucks so I made my own.

Feel free to subscribe to my Explosm RSS feed.

It basically features comic/forum content directly into the feed, and no “distracting you” comic if your reader happens to fetch the feed before the actual content is published at the site.

Enjoy!

launchd: Start and keep alive a script in OS X

2011-02-04T09:06:00.000-08:00

You want to start a script, and make sure it always keeps running in the background.

First write a file org.myscript.plist containing:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
        <key>KeepAlive</key>
        <true/>
        <key>Label</key>
        <string>org.myscript</string>
        <key>ProgramArguments</key>
        <array>
                <string>/path/to/my/script.sh</string>
        </array>
</dict>
</plist>

Copy or move your file into ~/Library/LaunchAgents/ and then run

> launchctl load org.myscript

and use something like

> launchctl list | grep myscript

if you want to very that it is indeed running.

The script will continue to run forever, restarted if necessary, even after you login/logout of your user session.

Perl: Growl notifications for new feed/rss items

2011-02-04T08:04:00.000-08:00

You want to monitor a web feed and show Growl notifications on your desktop when new items are published. Further, you want to be able to click on a notification to open the relevant link on your default web browser. Also, for some rather bizarre reason, you don't want to use RSS Menu.

In my case the reason was that I don't want to track the “read/unread” status of individual items. I found it distracting, because I was always going to click on “Mark all read” after each batch of new items arrived. This perl script gives items in the feed a one off chance to grab your attention and make you click on it. Otherwise the item is gone into oblivion and you never get to hear about it again.

Using Cocoa::EventLoop and Cocoa::Growl it was actually rather simple. And this script can also work as a skeleton for any program where you want to post Growl notifications and do something on the click callback.

#!/usr/bin/perl -w

use strict;
use Cocoa::EventLoop;
use Cocoa::Growl ':all';
use LWP::Simple;
use XML::Simple;

die "Growl is not running.\n" unless growl_running();

my $feed  = 'http://tex.stackexchange.com/feeds';
my $local = 'tex.se.xml';
my $last  = '';

growl_register(
  app           => 'FeedGrowl',
  icon          => 'http://tex.stackexchange.com/favicon.ico',
  notifications => ['New feed item'],
);

my $timer = Cocoa::EventLoop->timer(
  interval => 300, # every five minutes
  cb       => \&refresh_feed
);

Cocoa::EventLoop->run;

exit 0;

sub refresh_feed {
  mirror($feed, $local);
  my $data = (XMLin($local, keyattr => [], ForceArray => ['entry']))->{'entry'};
  return unless @$data;
  
  my $i = 0;
  for my $item (@$data) {
    last if $item->{'updated'} le $last;
    $i++;
    notify_new_item($item->{'title'}->{'content'}, $item->{'link'}->{'href'});
    last if $i == 5; # show at most 5 new items
  }
  
  $last = $data->[0]->{'updated'};
}

sub notify_new_item {
  my ($item, $link) = @_;
  my $open_link = eval "sub { system('open', '$link') }";
  growl_notify(
    name        => 'New feed item',
    title       => 'New item on TeX.se',
    description => $item,
    on_click    => $open_link
  );  
}

Using screen to leave uninterrupted processes

2011-01-29T09:07:00.000-08:00

You want to leave a process running (often on a remote computer), and you want the process to keep running even if you close the terminal session (often running on a local computer).

First start a screen session using
> screen -S [name]where [name] is any meaningful identifier for you to remember.

Now you can start your processes, and close the terminal window if you want to.

Later, to recover your session use
> screen -d -r [name]so that you can get your session back and monitor the processes that you left running.

If you forget the name of your screen session, you can also always type
> screen -lsto see a list of all the currently active “screens”.

Convert PS to PDF to JPG/PNG/etc on the command line on OSX

2011-01-28T10:04:00.000-08:00

> ps2pdf figure.ps
> sips -s format jpeg figue.pdf --out figure.jpg

Perl: die if child process is interrupted

2011-01-27T04:12:00.000-08:00

You are using system to run a child process, and you would like your program to terminate if the process happens to be interrupted, e.g. by pressing Ctrl+C.

die "Child process terminated\n" if system($cmd) & 127 == 2;

In general system($cmd) & 127 is the signal number, if any, that caused the child process to terminate.

Create a new rails project and add to svn (with adequate ignores)

2010-06-09T06:26:00.000-07:00

You want to create a new rails project and add it to your svn repository, but ignoring log and temp files.

What you actually have to do is first create the project with rails, then slowly add folders (e.g. using “svn add -N myapp/log”) and set their appropriate ignore properties (e.g. using “svn propset svn:ignore "*.log" myapp/log”). Finally you can add all the remaining unversioned files and folders before making your initial commit.

For convenience, I've put all these commands into a svn-rails shell script which can then be run as

> svn-rails myapp

to do all the required stuff.

Embed YouTube video as valid XHTML 1.0

2010-06-03T08:26:00.000-07:00

You want to embed a YouTube video in your XHTML 1.0 document, but the code that YouTube gives you does not validate. Instead, use something like

<object type="application/x-shockwave-flash" width="480" height="385"
data="http://www.youtube.com/v/_ZG8HBuDjgc"><param name="movie"
value="http://www.youtube.com/v/_ZG8HBuDjgc" /><param name="wmode"
value="transparent" /></object>

Take screenshots on OS X

2010-05-25T09:20:00.000-07:00

To take screenshots (screen captures) on Apple Mac OS X use one of the shortcuts:

Command-Shift-3		Capture the entire screen to a file on the desktop
Command-Shift-4		Capture a selection to a file on the desktop
Control-Command-Shift-3		Capture the entire screen to the clipboard
Control-Command-Shift-4		Capture a selection to the clipboard

Parse Prolog-like terms in Perl

2010-05-11T06:08:00.000-07:00

Use Parse::RecDescent to parse a Prolog-like term from a string. The given example parser will read a term (and its subterms) and return a lisp-like array reference of the form: [atom, terms*]; where atom is a string scalar and each argument (if any) is itself a term given again as a lisp-like array reference. A reverse, term_str, is also provided as an example on how to recursively process such term structures.

use Parse::RecDescent;

my $parser = Parse::RecDescent->new(q(
  terms  : <leftop: term ',' term>
  term   : atom '(' terms ')' <commit>
           { $return = $item{terms}; unshift @$return, $item{atom}; }
         | atomic
           { $return = [$item{atomic}] }
  atomic : atom | num
  atom   : /[a-zA-Z_]+/
  num    : /\d+/
));

sub term_str {
  my @arg = @{ shift @_ };
  my $fun = shift @arg;
  # here $fun and @arg hold their expected values 
  return $fun unless @arg; # nulary term
  for my $a (@arg) { $a = term_str($a) } # process arguments recursively
  return $fun . "[" . join (",", @arg) . "]";
}

print term_str($parser->term('ls(s(p),q,d(x,p))'));
# outputs: ls[s[p],q,d[x,p]]

Share a folder between a host and a virtual guest OS

2010-05-07T07:12:00.000-07:00

Using VirtualBox, you want to share a folder between the host OS and a virtual guest OS. The following instructions are for a Linux guest OS. For other systems the process should be similar just RTFM (Help → Contents...) for the details.

First you have to install the “Guest Additions”:

Start the virtual machine, make its window active and select the menu option “Devices → Install Guest Additions...”.
This should mount a virtual CD with the additions from where you can install them.
If installation didn't start automatically, try on a terminal
```
> cd /cdrom/
> ./autorun.sh
```
Upon successful installation, reboot the virtual machine.

Now you can share a folder:

Crate a folder on the host OS, e.g. host/shared/folder, and put some files in there.
Run the virtual machine with the guest OS.
On the menu of the virtual machine select “Devices → Shared Folders...”.
Add a new shared folder with the path to your host/shared/folder and an appropriate name like host-shared.

On the guest OS, create a corresponding folder and mount the new device there

> mkdir guest/mount/point
> sudo mount -t vboxsf host-shared guest/mount/point

If everything works you should be able to cd to the newly mounted folder and see the files from your host.

You might want the shared folder to be automatically mounted when booting the virtual machine. If so, edit the /etc/fstab file (you need privileges) and add the following line to the end

host-shared /guest/mount/point vboxsf defaults 0 0

CPAN as non-root user

2010-05-07T06:14:00.000-07:00

Forget about trying to configure cpan as a non-root user. The easiest solution is to use cpanm to install your packages and forget about it.

So go and click that link. Now.

Ok, for some rather bizarre reason, you insist on configuring cpan as non-root and you're still reading this. So, let's do it.

First, to set up the PERL5LIB environment variable, add the following to your ~/.bash_profile or equivalent:

PERL5LIB=~/perl5lib/lib/:$PERL5LIB
PERL5LIB=~/perl5lib/share/perl/5.8.8:$PERL5LIB
PERL5LIB=~/perl5lib/share/perl/5.8:$PERL5LIB
PERL5LIB=~/perl5lib/lib/perl/5.8.8:$PERL5LIB
PERL5LIB=~/perl5lib/lib/perl/5.8:$PERL5LIB
export PERL5LIB

Reboot or source the file. Then, specially if you are having troubles, you might want to make sure that you are starting with a clean user instalation.

> cd ~
> rm -rf .cpan
> rm -rf perl5lib

Create a folder to hold the modules, and start CPAN

> mkdir perl5lib
> cpan

Manual configuration should start, follow the instructions on screen, until at some point the script asks you for any extra arguments for Makefile.PL. You should then supply

PREFIX=~/perl5lib

When the configuration finishes, make sure CPAN tells you it has writen the updated configuration file. If everything seems fine, type quit.

To install a package now type

cpan "Some::Module"

If things keep failing, you should probably consider doing a manual package install.

Note: if for some reason your script will run in an environment where PERL5LIB is not (or cannot) be set, then your script should also include

use lib '~/perl5lib/lib/';
use lib '~/perl5lib/share/perl/5.8.8';
use lib '~/perl5lib/share/perl/5.8';
use lib '~/perl5lib/lib/perl/5.8.8';
use lib '~/perl5lib/lib/perl/5.8';

But then, if you need this, you might also need to expand those ~'s into their actual absolute paths.

Stay away from the utf8 mess in Perl

2010-05-06T07:10:00.000-07:00

You are writing a Perl script that uses and manipulates data and/or text files with utf8 characters. Perl gives you sometimes weird error messages, funny characters show up in your output, and by searching out there you only get confusing and contradictory advice.

If you think that everything should be utf8 and you don't care about anybody using any other kind of encoding, simply include the following pragmas:

use open qw/:std :utf8/; # all your input/output is utf8
use utf8;                # your program file is utf8

For most cases this should be enough, and you should be able to forget about this stupid encoding business.

Anyway, if somehow you manage to get a $string that you know is encoded as utf8 but perl doesn't seem to realize this, yo should call

utf8::decode($string);

so that perl stores the string in its internal utf8 representation.

Post to Twitter from Perl using a single access token

2010-05-06T05:46:00.000-07:00

You want to post status updates to your own Twitter account from a Perl script. Because of the new Twitter API you need to authenticate using OAuth, but you don't want (nor need, nor care) to do the “OAuth token request” or whatever dance. It's my own account, I just want to post!

Go to dev.twitter.com and register a new application. From this you will obtain a consumer key and a consumer secret.
Once your app is registered, click on “My access token” on the details page of your application. From this you will obtain an access token and an access token secret.
Go and code in Perl using Net::Twitter::Lite.

Example code:

use Net::Twitter::Lite;

my $nt = Net::Twitter::Lite->new(
  consumer_key        => 'your consumer key',
  consumer_secret     => 'your consumer secret',
  access_token        => 'your access token',
  access_token_secret => 'your access token secret'
);

my $result = eval { $nt->update('Hello, World! http://goo.gl/zAbwc') };

warn "$@\n" if $@;

Unescape tab and newline escapes

2010-04-28T08:43:00.000-07:00

You want to use regular expressions in perl to turn the literal "\t" and "\n" character sequences into the actual tab and newline characters.

The trivial solution doesn't work if you also want to allow the use of "\\" to escape the "escape" sequence (e.g. so that "\\node" expands to "\node", but "\\\n" does expand to a backslash followed by a newline).

What you actually need, is something like

  s/\\(.)/\\$1]/g;
  s/\\n]/\n/g;
  s/\\t]/\t/g;
  s/\\(.)]/$1/g;

Small discussion: this first encapsulates every escape sequence "\X" into "\X]" so that the escaped "X" can not interfere with the rest of the string. Then you can evaluate all the escapes you want, e.g. turn "\n]" into a newline. Finally all remaining "\X]" are expanded back to "X".

Add every unversioned file to version control

2010-04-23T10:18:00.000-07:00

You want to recursively seek every unversioned file in a directory, and add them to your svn repository.

First, to list the unversioned files, you can use

> svn st | grep '^?'

and to add all of them to your repository

> svn st | awk '($1 == "?") { print "svn add", $2 }' | sh

As usual, you might want to check for sanity and run first without the last sh pipe.

Check if a file contains special (non-ascii) characters

2010-04-15T05:39:00.000-07:00

To check if a file contains any special (non "standard" ascii) characters, use Perl and POSIX character classes. Depending on how permissive you want to be use one of:


# allow only whitespace and standard ascii print characters (i.e. [\x20-\x7E])
> perl -nwe 'print if /[^[:space:][:print:]]/' file.txt

# also allow control characters (i.e. [\x00-\x7F])
> perl -nwe 'print if /[^[:ascii:]]/' file.txt

Locate which path/file is loaded for a Perl module

2010-03-09T05:13:00.001-08:00

You want to figure out which is the source file being loaded for a given Perl module. Copy the code bellow in a file whichpm, make it executable, and then use, e.g.,

> whichpm Date::Manip
/Library/Perl/5.10.0/Date/Manip.pm

Source of whichpm:

#!/usr/bin/perl -w
$M = shift @ARGV;
die "Usage:\n  whichpm Some::Module\n" unless defined $M;
eval("use $M; 1") or die "Can't locate module $M in \@INC\n";
$M =~ s|::|/|g;
print $INC{"$M.pm"}, "\n";

Timestamps using ISO8601 strings

2010-03-03T09:53:00.000-08:00

You want to generate timestamps, be able to store them, retrieve them and do math with them; but you don't really care about parsing (i.e. you're not reading user input, just manipulating timestamps that you created yourself). Use perl's DateTime module, and the functions bellow to serialize and unserialize.

use DateTime;

sub now {
  return DateTime->now(time_zone => 'UTC');
}

sub date_str {
  return $_[0]->iso8601();
}

sub str_date {
  return undef unless $_[0] =~ m/^(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)$/;
  return DateTime->new(
           year       => $1,
           month      => $2,
           day        => $3,
           hour       => $4,
           minute     => $5,
           second     => $6,
           time_zone  => 'UTC'
         );
}

Change executable property of files under svn

2010-02-26T14:04:00.000-08:00

You can not ‘chmod’ files on svn, as changes in properties are not propagated to the svn server. To set or unset the executable propery on a versioned file use:

> svn propset svn:executable on FILE
> svn propdel svn:executable FILE

Use an ssh tunnel for web browsing

2010-02-11T11:31:00.000-08:00

You want to surf the web as if, e.g., you were at your office's network but from your computer at home. If you can access the work computer through ssh at, say, user@example.com, then set up a tunnel using:

> ssh -ND 9999 user@example.com

Then configure, e.g., Firefox as follows: Preferences → Advanced → Network → Settings. Select “Manual proxy configuration”, enter “localhost” as “SOCKS Host” and “9999” as its corresponding “Port”.

Accept the new settings and browse to, e.g., www.ip-adress.com to check that it's working.

"Readline" for any command line interface

2010-02-09T05:09:00.000-08:00

To add "readline" functionality (e.g. keyword input editing, and command history) to any command simply use
> rlwrap command

Passwordless ssh

2009-09-28T12:21:00.000-07:00

On the local computer: if don't have them already, generate your private/public pair of keys
$ ssh-keygen -t rsaAccept all the default options and when asked for a passphrase just press enter.

This should create the two files

~/.ssh/id_rsa     - your private key
~/.ssh/id_rsa.pub - your public key

On the remote computer: if it doesn't exist already, create the ~/.ssh directory

$ mkdir ~/.ssh
$ chmod 700 ~/.ssh

Finally append the contents of the local file ~/.ssh/id_rsa.pub to the remote file ~/.ssh/authorized_keys. For example, on the local computer:
scp ~/.ssh/id_rsa.pub user@remote.example.com:.ssh/my_new_keyAnd then on the remote computer:

cat ~/.ssh/my_new_key >> ~/.ssh/authorized_keys
rm ~/.ssh/my_new_key

Remove Firefox/Iceweasel lock

2009-09-03T03:52:00.000-07:00

If Firefox/Iceweasel is "already running, but is not responding", and you are sure it's really not running (e.g. after a clean boot), remove its lock with

rm -i ~/.mozilla/firefox/*/.parentlock ~/.mozilla/firefox/*/lock

Chop "n" lines from the beginning/end of a file

2009-02-26T03:52:00.000-08:00

You need some sort of "inverse" for the standard behavior of the head and tail commands, i.e., print everything but the first/last n lines of a file.

To skip the first n = 10 lines of a file
$ awk '(NR > 10)' file

To print everything but the last n = 10 lines of a file
$ head -n -10 file

Make "less" show utf-8 characters

2009-02-18T09:57:00.000-08:00

To make the less unix command properly display utf-8 characters, you might want to add the following to your .bash_profile
export LESSCHARSET=utf-8

Create TeXShop project

2009-01-28T06:12:00.000-08:00

Quick hack to turn a directory into a TeXShop project (i.e. make all .tex files point to a single project root).


ls -1 *.tex | awk '{ print "/bin/echo -n \"main.tex\" > " $1 "shop" }' | sh

to check for sanity you might want to run first without the sh pipe at the end.

Add line numbers to a file

2009-01-07T14:05:00.000-08:00


awk '{print NR "\t" $0 }' input.txt

Shell output redirection

2008-11-27T02:44:00.000-08:00

stdout to file
command > output.txt

stderr to file
command 2> output.txt

stdout to stderr
command 1>&2

stderr to stdout
command 2>&1

stdout and stderr to file
command &> output.txt

Produce valid dates in RFC3339

2008-11-23T03:50:00.000-08:00

In perl, to produce valid dates according to the RFC3339 standard, e.g. as required by the Atom specification, use UnixDate from Date::Manip. UnixDate is magic, it parses and translates strings containing dates in pretty much any conceivable format.

use Date::Manip;

# normalize all dates to UTC
Date_Init("ConvTZ=UTC");

sub valid_date {
  return UnixDate($_[0], "%OZ");
}

# then you use it as
$date = valid_date("now"); # 2008-11-23T12:22:06Z
$date = valid_date("Sat, 07 Sep 2002 00:00:01 GMT"); # 2002-09-07T00:00:01Z

Testing whether an optinal argument is supplied

2008-09-25T08:36:00.000-07:00

In LaTeX, you want to test whether a macro was called with or without an optional argument.

\def\test@non{\@empty}
\newcommand{\mycommand}[1][\@empty]{%
  \def\test@opt{#1}%
  \ifx\test@opt\test@non
    % \mycommand was used without optional argument
  \else
    % \mycommand was used with optional argument #1
  \fi
}

Installing a Perl module from CPAN

2008-09-15T02:21:00.000-07:00

The easiest way to install CPAN modules in perl is using cpanm which, furthermore, allows you to install without requiring any root access.

In OS X, look into your ~/.bash_profile and remove any references that you might have to the PERL5LIB environment variable. Then run the following commands on a terminal:

> curl -L http://cpanmin.us | perl - -l ~/Library/perl5 App::cpanminus local::lib
> perl -I ~/Library/perl5/lib/perl5 -Mlocal::lib=~/Library/perl5 >> ~/.bash_profile
> eval `perl -I ~/Library/perl5/lib/perl5 -Mlocal::lib=~/Library/perl5`

If everything went right, now you can install packages using:

> cpanm Some::Package

Some more details and an explanation can be found on this question about using CPAN as non-root user on Stack Overflow.

Hopefully your package is now installed and you can enjoy life.

If for some reason the installation above didn't work and you are still reading, then you might want to try installing your module using either CPANPLUS, by typing:

> cpanp -i Some::Module

or the more traditional cpan:

> cpan Some::Module

If even that fails, try to look at the produced output and install any dependencies or outdated packages first. Then retry.

Ok, you haven't succeeded and you are still reading. You probably need to do a manual install. You may also need a manual install if you get the “Sorry, PREFIX is not supported” error when trying to use cpan to install a package which has been built using Module::Build. Go figure.

So, lookup the download url for the .tar.gz package on CPAN and then type

> wget http://search.cpan.org/url/to/Some-Module.tar.gz
> tar -zxvf Some-Module.tar.gz
> cd Some-Module/
> ls

If you see a Build.PL file then do

> perl Build.PL
% if dependencies or module updates are needed do those first!
> ./Build
> ./Build test
> ./Build install --install_base ~/perl5lib

otherwise do

> perl Makefile.PL
% if dependencies or module updates are needed do those first!
> make
> make test
> make install PREFIX=~/perl5lib

The --install_base or PREFIX options are only needed if you are running a non-root installation, and it should point to the proper location in your home directory. Otherwise you need to sudo that last command to do a root install.

Life is hard.

C++ stringify

2008-08-30T13:21:00.000-07:00

Quick stringify hack, turn anything (printable) into a string.

#include <sstream>
#include <string>

template<typename T>
inline std::string stringify(const T& x) {
  std::ostringstream o;
  o << x;
  return o.str();
}

Discussion and a more robust version at the C++ FAQ Lite.

Reservoir Sampling

2008-08-19T05:27:00.000-07:00

Obtain a random sample of n elements evenly distributed from an input stream of unknown length.

my $n = shift @ARGV;
my $i = 0;
my @reservoir;

while (<>) {
  $i++;
  push @reservoir, [$i, $_];
  last unless $i < $n;
}

while (<>) {
  $i++;
  $reservoir[int(rand($n))] = [$i, $_] if int(rand($i)) < $n;
}

@reservoir = sort { $a->[0] <=> $b->[0] } @reservoir;

for my $x (@reservoir) { print $x->[1]; }

Useful perl implementation: reservoir.
Discussion and proof: Reservoir Sampling.

yendo

2008-08-15T02:10:00.000-07:00

Gerundio del verbo ir: yendo.
Están mal: ~~llendo~~, ~~iendo~~.

Generating Random Trees and Connected Graphs

2008-05-16T04:58:00.000-07:00

To generate a random (undirected) Connected Graph you first generate a spanning tree and then add additional edges at random.

To generate the random tree use the following algorithm

dst := random permutation of all nodes;
src.push(dst.pop()); % picks the root
while (!dst.empty()) {
    a := random element from src;
    b := dst.pop();
    add the edge (a, b)
    src.push(b);
}

Proof of correctness (all trees are possible and equally likely): Alexey S. Rodionov and Hyunseung Choo, On Generating Random Network Structures: Trees, ICCS 2003, LNCS 2658, pp. 879-887, 2003.

To generate the random permutation see Fisher-Yates shuffle, e.g. Recipe 4.17 in the perl cookbook.

Create pdf of single picture

2008-03-13T05:08:00.000-07:00

Use the package preview to generate .pdf files containing single pictures. Works particularly well with tikz. Use the following wrapper latex code, then run pdflatex on it.

\documentclass{article}

\usepackage{tikz}
\usepackage[active,pdftex,tightpage]{preview}
\PreviewEnvironment[{[]}]{tikzpicture}

\begin{document}

\begin{tikzpicture}[options]
...
\end{tikzpicture}

\end{document}

Find all hyphenated words

2007-09-05T04:47:00.000-07:00

Finds all hyphenated words within a set of text files.

grep -oh --perl '\b\w+(-\w+)+\b' *.tex | sort -uf

Find possible variations of hyphenated words to check for consistency

grep -i --color -C 1 --perl 'semi[\s-]*decision' *.tex

This will find, e.g.: semidecision, semi decision and semi-decision.

Install LaTeX Package with MPM

2007-08-27T04:54:00.000-07:00

sudo mpm --install=<package>
sudo texhash

BibTeX Names

2007-08-23T10:10:00.000-07:00

Syntax for writing BibTeX author names:

"First von Last"

"von Last, First"

"von Last, Jr, First"

Recursive directory listing

2007-07-27T02:49:00.000-07:00

And get one line of output per file use find

find . -print

Excel trendline equations

2007-07-24T07:32:00.000-07:00

When you add a trendline to a chart, Excel provides an option to display the trendline equation in the chart. This tip describes how to create formulas that generate the trendline coefficients. You can then use these formulas to calculate predicted y values for give values of x.

These equations assume that your sheet has two named ranges: x and y.

Linear Trendline

Equation: y = m * x + b
m: =SLOPE(y,x)
b: =INTERCEPT(y,x)

Logarithmic Trendline

Equation: y = (c * LN(x)) - b
c: =INDEX(LINEST(y,LN(x)),1)
b: =INDEX(LINEST(y,LN(x)),1,2)

Power Trendline

Equation: y=c*x^b
c: =EXP(INDEX(LINEST(LN(y),LN(x),,),1,2))
b: =INDEX(LINEST(LN(y),LN(x),,),1)

Exponential Trendline

Equation: y = c *e ^(b * x)
c: =EXP(INDEX(LINEST(LN(y),x),1,2))
b: =INDEX(LINEST(LN(y),x),1)

2nd Order Polynomial Trendline

Equation: y = (c2 * x^2) + (c1 * x ^1) + b
c2: =INDEX(LINEST(y,x^{1,2}),1)
C1: =INDEX(LINEST(y,x^{1,2}),1,2)
b = =INDEX(LINEST(y,x^{1,2}),1,3)

3rd Order Polynomial Trendline

Equation: y = (c3 * x^3) + (c2 * x^2) + (c1 * x^1) + b
c3: =INDEX(LINEST(y,x^{1,2,3}),1)
c2: =INDEX(LINEST(y,x^{1,2,3}),1,2)
C1: =INDEX(LINEST(y,x^{1,2,3}),1,3)
b: =INDEX(LINEST(y,x^{1,2,3}),1,4)

Higher Order Polynomial Trendline

Notice the pattern in the two preceding sets of formulas

[source]

Emacs shortcuts

2007-07-24T01:23:00.000-07:00

Save: Ctrl-X Ctrl-S
Quit: Ctrl-X Ctrl-C

Select: Ctrl-space
Copy: M-W
Cut: Ctrl-W
Paste: Ctrl-Y

Go to line: M-x goto-line

Word count a pdf file

2007-07-19T06:23:00.000-07:00

ps2ascii myfile.pdf | wc -w

Command line tar and untar

2007-07-02T06:05:00.000-07:00

Untar

tar xvfz something.tar.gz

Tar

tar cfz blah.tar.gz blah/

Trim strings

2007-06-12T03:34:00.000-07:00

Remove white-space at beginning and end of strings


sub trim($)
{
 my $string = shift;
 $string =~ s/^\s+//;
 $string =~ s/\s+$//;
 return $string;
}

Escape characters in URLs

2007-06-02T12:32:00.000-07:00

sub escape_uri {
  my $s = shift;
  $s =~s/([^a-zA-Z0-9_\-. ])/uc sprintf("%%%02x",ord($1))/eg;
  $s =~ tr/ /+/;
  return $s;
}

Or even better

use URI::Escape;
my $uri = uri_escape($string)

To unescape use

$string =~ s/%([0-9A-Fa-f]{2})/chr(hex($1))/eg;

Shell control commands

2007-05-31T04:16:00.000-07:00

#!/bin/bash

for VAR in ITEMA ITEMB ITEMC ITEMD; do
  # do something with $VAR
done

for VAR in {1..5}; do
  # do something with $VAR
done

while [ condition ]; do
  # do something
done

if [ "$str1" == "$str2" ]; then
  # do something
elif [ other_condition ]; then
  # do another thing
else
  # do something else
fi

Parse command line options

2007-05-31T03:58:00.000-07:00

Use the getopts shell utility


while getopts d:s o
do case "$o" in
 d) value_of_d="$OPTARG";;
 s) saw_an_s="TRUE";;
 esac
done
shift $OPTIND-1

# the remaining arguments are in $*

getopts optstring name
optstring contains a letter for each option that the script understands, follow a letter with ":" if the option receives an argument.

Timing out the execution of a command

2007-05-31T03:33:00.000-07:00

Use the ulimit shell command.


#!/bin/sh

ulimit -t 10
# your command here