The Amazon Web Services API has opened up a
direct door to Amazon's catalog. Developers can now search and browse
Amazon programmatically, and reformat search results in any way they
is an alternative web browser that has been designed from the ground
up to be an application platform for working with the Web. Mozilla's
sidebar offers a way to search several different search engines. For
example, Mozilla comes with Google and dmoz.org installed in the
sidebar, as you see in .
Figure 1. Mozilla sidebar search engine selection
With one search field, you can search these sites individually or mix
the results from both at the same time—all the while keeping
the results separate from the main browser window. It's a new way to
search that removes the problem of paging back and forth between
search results, or trying to keep track of several open windows.
The Mozilla-Search plug-in interface allows you to add your own entry
to the list of search engines available. Although Google and dmoz.org
are there when you install Mozilla, that doesn't mean your search
engine selection can't grow. It just takes some understanding of both
Amazon and Mozilla to get them talking to each other. Once you set up
this hack, you'll have an entry for Amazon in your list of available
Running the Hack
You have two files now, and each needs to go to a specific place to
set up this hack. The plug-in definition file,
amazon.src, needs to be placed in the
searchplugins directory of your Mozilla
installation. On Windows, the default directory is
Files\mozilla.org\Mozilla\searchplugins\. Once the file's
in place, you should have "Amazon" as an option in your list of
The XSL file, amazon_search.xsl, should be
uploaded to a publicly available web server. Once uploaded, find the
URL and make sure the <input
name="f"> value is set to that URL in
With those two files in place, your Amazon sidebar search should be
ready for action! Just bring up the sidebar by hitting F9 in Mozilla.
You should see a list of search engines. Make sure Amazon.com is
checked, type in a search term, and click Search. You should see a
list of results in the sidebar that you can click to bring up in the
main browser window, as in .
Figure 2. Mozilla Amazon sidebar search
The search results stay persistent in your sidebar, so you can click
from result to result without having to navigate back and forth
between pages on the site. And now that Amazon is a search engine
option, you can do other fun things such as search Amazon and Google
at the same time. This allows you to browse, view, and sort the
search results from both sites in one place.
If you really want to go the extra mile to integrate, you can create
an icon to represent Amazon results. Mozilla looks for a 16x16 pixel
image in the searchplugins directory with the
same base name as the search definition file. In this example, a file
named amazon.gif would be displayed in the
sidebar with any Amazon search result.
There are two pieces to setting up the sidebar. One file resides
locally and tells Mozilla how to search Amazon. The other file
resides on a public server and formats the search results for
The local file uses the Mozilla-Search plug-in definition format.
Create a file called amazon.src and add the
# Amazon Search for Mozilla
# Adds Amazon.com as an option to Mozilla's
# sidebar search.
# Creation: May 2003
description="Search for products at Amazon.com"
<input name="BlendedSearch" user>
<input name="dev-t" value="insert developer tag">
<input name="t" value="insert associate tag">
<input name="type" value="lite">
<input name="f" value="http://example.com/amazon_search.xsl">
priceStart = "<b>"
priceEnd = "</b>"
This code creates a new search option for Amazon.com in Mozilla's
sidebar. It defines the URL to use for searching, as well as any
querystring attributes. The <interpret> tag
performs some pattern matching in the resulting HTML page so Mozilla
knows how to format the results for the sidebar. This code tells
Mozilla to start treating the file as search results when it sees
<ul>, and stop when it sees
</ul>. The code uses
<li> to determine when something is an
individual search result.
The action URL in this case performs a search at Amazon's Web
Services, which returns XML. In the querystring, defined by
<input name="f">, a file on your server is
specified to turn that XML into HTML.
The transformation from Amazon's XML to the HTML necessary for the
Mozilla search is handled by an XSL file. Add the following code to a
file called amazon_search.xsl:
<?xml version="1.0" encoding="ISO-8859-1" ?>
<xsl:output indent="yes" media-type="text/html"/>
<font face="verdana" size="2">
</a>[<xsl:value-of select="Catalog"/>] -
This stylesheet produces a simple HTML page of search results. Note
that the XSL has the <ul> and
<li> elements that Mozilla is expecting for
the search defined in amazon.src.