ObjectClarity.com

I’ve set this up before but couldn’t remember just how I did it. Instead of having to figure it out all over again next time, I thought I’d write it somewhere.

Setting the time in CPanel’s Standard Cron UI is very simple so I won’t bother with that but the command for triggering a web page has a few things that messed me up.

For those who don’t care to read much, here’s the command:

 wget -O - -q -t 1 'http://www.mydomain.net/doit.php?id=12345' >/dev/null 2>&1 

Or, if you’re using a .htaccess username and password, like this:

wget -O - -q -t 1 'http://username:password@www.mydomain.net/doit.php?id=12345' >/dev/null 2>&1

Of course you’ll need to insert your own url but you can otherwise cut and paste the line above and start running your php script as a cpanel cron job. But what does it all mean?

“wget” is a linux command line utility to fetch the contents of a web page. In our case we don’t care about the output of the page, only that it gets triggered – or ‘looked’ at.

“-O -” says to discard the output (aka, the page contents), we’re not interested to keep these!

“-q” put wget into quiet mode, I don’t want to know about errors here either.

“-t 1″ tells wget to only try once, if the page doesn’t work for some reason, give up.

The url should be wrapped in single quotes! This is an easy one to miss because in a simple case, it’s not required but if you have url parameters it’s a must otherwise the url gets truncated and the request won’t work as you planned it to.

“>/dev/null 2>&1″ This tells cron *not* to send out a notification that the script was called. The details of this part gets a bit hairy but I’ll do my best to explain. As the script I’m triggering already sends an email when it completes, I don’t really want the one that is sent out automatically each time the cron runs. As well, the subject for the email is the complete cron command. I don’t know about you but if your scheduled cron command includes anything even sort-of private, I’d prefer not to blast it out in email every day! Consider if you included a URL to a script behind an htaccess password, the username and password would be right there in the subject of your email every day!

The first part, “>/dev/null” redirects the output (standard out) of the command to /dev/null (the Linux equivalent of a black hole!). The second part, “2>&1″ redirects ‘standard error’ to ‘standard out’. The short story is that there is no more output from your command and therefore there’s no email to be sent out!

This morning I find myself in need of a file that used to exist in an SVN project. The file was deleted several months ago and I don’t have a copy anywhere else. Finding it in my SVN repository turned out to be quite simple.

Run the following command to dump the complete activity log from the project.

> svn log --verbose > myLog.txt

Then simply open the ‘myLog.txt’ file with your favourite editor and hunt for the last mention of the file you need. I’m looking for “register.module.php”, so for me that looks like this:

------------------------------------------------------------------------
r428 | mvoorberg | 2010-11-15 22:06:33 -0700 (Mon, 15 Nov 2010) | 2 lines
Changed paths:
   M /faq.php
   M /login.action.php
   D /register.module.php

Use the following command to restore the file to the local filesystem at which point you can copy it out or do whatever you like with it. Notice the version number (427) is one less than the version that included the Delete action above.

> svn up -r 427 register.module.php

Once I copied the file out, I removed it again by simply running:

> svn up

Don’t forget to remove the ‘myLog.txt’ file when you’re done with it.

Before rebuilding this website, I wanted to make sure I was able to keep the old site in a subfolder to help keep my search ranking and not break any external links to my site. This is clearly a job for modRewrite so I had a look around and goolged things like “Joomla to subdirectory” and the like. Funny enough I couldn’t find any posts about it and no-one had posted the required .htaccess entries to make it happen. As I’d never used SEF urls with my Joomla site and I wasn’t going to use Joomla for the new website, it was easy to find a pattern that all my urls would match. I could simply send anything that contained the string “option=com” to my “archived” subfolder.

Here’s the entries I added to the .htaccess in the root of my website:

# Make sure this module is turned on
RewriteEngine on

# Match any Joomla Urls
RewriteCond %{QUERY_STRING} ^option=com(.*)$

#Only forward Get and Head requests to the archived site
RewriteCond %{THE_REQUEST} ^(GET|HEAD)

# Don't match urls that already contain the archived subfolder
RewriteCond %{REQUEST_URI} !^/archived/$

# Send a 301(Permanent Move) to the archived folder
#  and don't process any more rules(L)
RewriteRule ^index.php$ /archived/index.php [L,R=301]

Follow-up:
As it turns out you can’t set the database permissions for specific tables when running in a typical shared LAMP environment. I was hoping for SELECT only and INSERT on some Joomla session tables. I noticed that I was still getting some new spam user registrations on the old site so I’ve disabled POST requests with the following entries added to the .htaccess in the /archive folder of the old Joomla site:

# Make sure this module is turned on
RewriteEngine on

# Disallow POST requests to the archived site by sending a 404 error
RewriteCond %{THE_REQUEST} !^(GET|HEAD)
RewriteRule ^ / [R=404]

I was reading through someone else’s C# code the other day when I saw something that wasn’t immediately clear. It went something like this:

Object o;
for (int i=0; i < myCollection.count; i++) {
    o = myCollection[i] as SomeObjectType;  // Huh?
    if (o != null) {
        // Do something with 'o'...
        break;
    }
}

While the cast seems clear enough, there are a few things to be aware of. If the above cast fails, it assigns a null value to ‘o’. If the null value is not handled immediately, you may not realize the problem until some time later when you run into the null reference.

Object o;
    // As-casting - Returns null if the cast fails!
    o = myCollection[i] as SomeObjectType;

    // Prefix casting - throws an exception if the cast fails!
    o = (SomeObjectType)myCollection[i];

There may be a performance benefit to the As-cast but I’m much more comfortable knowing that my cast exceptions were handled properly using the Java-esque Prefix cast.

You’ll need to install a few packages before this works:

    sudo apt-get install netpbm
    sudo apt-get install dcraw
    sudo apt-get install libimage-exiftool-perl

Once you’ve got these installed, put the following in a file in ~/.gnome2/nautilus-scripts/ and restart Nautilus to add it to the right-click context menu.

#!/bin/bash
files=$#
count=1
message=`echo "Converting $files RAW files to Jpeg"`

(while [ $# -gt 0 ]; do

	# Strip off the file extension, including the "."
	upperExt=`echo $1 | sed 's/.*\(\..*\)/\1/' | tr '[a-z]' '[A-Z]'`

	if [ -f "$1" ]
	then
		# Get the file name without the extension
		trimmed=`echo $1 | sed 's/\(.*\)\..*/\1/'`

		if [ $upperExt = ".PPM" -o $upperExt = ".CR2" -o $upperExt = ".NEF" -o $upperExt = ".OTHER_RAW_EXTENSION" ]
		then
			if [ $upperExt = ".PPM" ]
			then
				# Convert the PPM image to a Jpeg
				pnmtojpeg "$1" > "$trimmed.jpg"
			else
				# Convert the RAW image to a Jpeg
				dcraw -c -w -h -b 1.0 "$1" | pnmtojpeg > "$trimmed.jpg"
			fi
			# Copy EXIF data to the new Jpeg image
			exiftool -overwrite_original -TagsFromFile "$1" "$trimmed.jpg" >/dev/null

			# Set the Jpeg's file timestamp to match the EXIF date
			dcraw -z "$trimmed.jpg"
		fi
	fi
	# Output the zenity progress bar
	sav=`echo "(($count / $files) * 100)" | bc -l`
	echo $sav
	count=`expr $count + 1`
shift
done) | zenity --progress --auto-close --auto-kill --text "$message"

You’ll have to make the file executable by using:

    chmod +x raw2jpeg.sh

Update:
I modified the script to convert PPM files created by UFRaw as well. Also now using pnmtojpeg instead of ppmtojpeg.

I use NetworkManager to connect to my wireless network and wanted a way to map the drives automagically when I connect.  However, as I connect to a number of wireless access points I didn’t want to send my credentials trying to find my home server out each time I connect at the coffee shop.

Enter the NetworkManager dispatcher.  Scripts in /etc/NetworkManager/dispatcher.d/  are executed when the network manager connection is made.  I created the following file as “99mountmydrives” and chmodded it with execute permissions.

See the script below:

#!/bin/sh

IF=$1
STATUS=$2
USER=myusername
SERVERIP=192.168.1.200
SERVERMAC=00:aa:bb:cc:dd:ee

echo "running 99mountmydrives as `whoami`;" >> /tmp/dispatcher-log
echo "int=$1 and status=$2" >> /tmp/dispatcher-log

wait_for_process() {
    PNAME=$1
    PID=`/usr/bin/pgrep $PNAME`
    while [ -z "$PID" ]; do
        echo "waiting." >> /tmp/dispatcher-log
          sleep 3;
          PID=`/usr/bin/pgrep $PNAME`
    done
}

mount_drives() {
    echo "waiting for nm-applet" >> /tmp/dispatcher-log
    wait_for_process nm-applet
    echo "mounting drives" >> /tmp/dispatcher-log
    /home/mvoorberg/automountmyth.sh
}

if [ "$IF" = "wlan0" ] && [ "$STATUS" = "up" ]; then
    #WIFI Subnet at home

    test=`ping -c 3 $SERVERIP >/dev/null;arp -a $SERVERIP`
    echo "$test" >> /tmp/dispatcher-log

    if expr "$test" : ".*$SERVERMAC" 2> /dev/null ; then
        mount_drives
        exit $?
    fi
fi

While working on a simple test page, I wanted to use getElementById to get an Image object from JavaScript.  I quickly wrote up the code and tried it.  Firefox was fine but it failed on my IE7 with “Microsoft JScript runtime error: Object doesn’t support this property or method”.  It took some digging to see what the problem was.

First thing I checked was that I used both element Name and Id and that they were the same.  I *always* do that given that IE will use whichever one matches the desired string.  After reading a few blog posts about common problems but with no good ideas I went back to my code and reviewed it carefully.  It turns out that I used the same name for my element and my local JavaScript variable.  That by itself isn’t a problem but what made it an issue for me was that I failed to initialize my variable with the “var” keyword.

See the code below for the example:

<html>
<head>
<script type="text/javascript">
    function getMyImg() {
        var myFirstImg = document.getElementById("myFirstImg");
        mySecondImg = document.getElementById("mySecondImg");
        // ^---This one will fail on IE7(probably IE5, IE6, IE8, IE9, IE...)
        if (myFirstImg.src == mySecondImg.src) {
            alert('src=' + mySecondImg.src);
        }
    }
</script>
</head>
<body>
    <a href="#" onclick="getMyImg(); return false;">Do Something</a>
    <img id="myFirstImg" src="sample.png" />
    <img id="mySecondImg" src="sample.png" />
</body>
</html>