Personal blog of Chris Ergatides

This example will show you how easy it is using sed to find a particular line in a file and replace all or part of the next line. In this example, using htaccess we want to deny access to all clients except a particular dynamic IP (for your home connection for instance).

Your .htaccess file might look like this:

order deny,allow
deny from all

# allow from example.dyndns.org
allow from 8.8.4.4

And your sed code to update the allowed IP address might look like this:

sed -i -r "/# allow from example.dyndns.org/I{\
n; s/([0-9]{1,3}.){3}[0-9]{1,3}/8.8.8.8/\
}" .htaccess

Let’s look at it piece by piece.

• /# allow from example.dyndns.org/I will search the file for this string. The I flag at the end makes the search case insensitive.
• n; will tell sed to continue processing on the next line.
• s/([0-9]{1,3}\.){3}[0-9]{1,3}/8.8.8.8/ will regex search for an IP and replace the match with your updated IP (in this case 8.8.8.8).
• .htaccess is of course the input file to process.

Note that in .htaccess files, comments must be on a line of their own. End of line or inline comments are not permitted. In cases where you do not need to replace a part of the next line but rather all of it, our replacement becomes s/.*/8.8.8.8/

In 2003, I wrote a custom Amazon-like shopping cart script. Back then, my web scraping skills weren’t, ahem, the best they could have been. For the site’s currency conversions, I had a cron job pull down 3 sets of converstions, calculated the average for each currency pair, then stored the result locally for my use.

At the time, I did this hoping that in a worse-case-scenario, a maximum of 2 sources would die in any weekend. Luckily, the sources lasted for the 3 years whilst I was working for that company (and apparently for another two after that). Now, I hear, they are manually updating a static text file… once in a while.

Today, I have no such needs but as always I am constantly looking for new ways to do things, should the need arise. I also needed a simple currency converter for doing small calculations like how much to top up a prepaid credit card by to purchase something from eBay which is listed in another currency.

Full credits to rafacas at commandliners for the inspiration and as he points out, note that the script uses Google Finance’s page.

#!/bin/bash

# 2009 - Rafa Casado - http://bit.ly/bYZwex
# 2010 - Chris Ergatides - http://bit.ly/9XjAUz

if [ $# -lt 2 ]; then
    echo -e "\nUsage: $0 currency1 currency2 amount"
    echo "Default: $0 GBP EUR 1"
    echo "Example 1: $0 USD GBP"
    echo "Example 2: $0 GBP USD 42"
fi

toUpper() {
    echo $@ | tr "[:lower:]" "[:upper:]"
}

if [ -n "$1" ]; then FROM=$(toUpper "$1"); else FROM=GBP; fi
if [ -n "$2" ]; then TO=$(toUpper "$2"); else TO=EUR; fi
if [ $TO == $FROM ]; then echo 'Nothing to do!'; exit 2; fi
if [ -n "$3" ]; then A=$3; else A=1; fi

CONVERTER="http://www.google.com/finance/converter?a=$A&from=$FROM&to=$TO"

RESULT=`wget -nv -O - "$CONVERTER" 2>&1 | \
sed -n -e 's/.*<span class=bld>\(.*\)<\/span>.*/\1/p'`

echo -e "\nResult: $A $FROM = $RESULT\n"

So, let’s say you’ve named the script currency.sh and made it executable with chmod +x, what does the output look like?

#Called without any parameters:
[chris@r500 ~/bin]$ ./currency.sh

Usage: /home/chris/bin/currency.sh currency1 currency2 amount
Default: /home/chris/bin/currency.sh GBP EUR 1
Example 1: /home/chris/bin/currency.sh USD GBP
Example 2: /home/chris/bin/currency.sh GBP USD 42

Result: 1 GBP = 1.1387 EUR

#Called with example 1's parameters:
[chris@r500 ~/bin]$ ./currency.sh usd gbp

Result: 1 USD = 0.6475 GBP

#Called with example 2's parameters
[chris@r500 ~/bin]$ ./currency.sh gbp usd 42

Result: 42 GBP = 64.8606 USD