I’ve been looking around for a site search engine which would allow people to find stuff on my sites. The basic features I was looking for are:

• It should not take up huge resources on my servers, however, it should be relatively fast in delivering results
• I don’t require immediate indexing of new content, a one day delay is acceptable
• It should be possible to style the results so they appear as being part of my site. My sites depend on PHP and SSI for various “add-on” content (menus, etc) and it should be possible to include these on the search result page
• A big plus is if I can host it on my servers. I don’t like depending on other services
• I could live with some ads as long as they are unobtrusive
• Easy to install and configure across my sites
• It should be possible to analyze the statistics of search terms

There are a bunch of services that provide a site search engine (just google for that term). These services spider your site once in a while and store the indexed data on their servers. The result page is displayed on their servers and free accounts are always filled with ads and have limitations in the number of indexed pages allowed. A cool thing about these services is that they usually provide a page where you can view search statistics, ie. you can easily see what terms your users are searching for. The one service that seems to lead this type of services is FreeFind.

A drawback of using these services is the pain of integrating the results into the look and feel of your sites. Even though some of them do provide templates, you have to upload a copy of the non-CSS template to their servers which doesn’t provide enough flexibility. Obviously, these don’t work with my PHP and SSI bits and pieces.

The next type of search engines I considered are the ones that you install on your servers. mnoGoSearch looks like it could do the job very well, but it requires compilation on the servers. This was unacceptable for me as I can’t do this for some of my sites.

So, I started looking for another way to integrate a search engine on my sites and Google came to the rescue. I first bumped into this script which uses the Google Web API. This works nicely, but I didn’t like the fact that they have access to my Google API key and also require a link to be placed close to the search box (I don’t mind linking, but just couldn’t get it to look nice enough).

Finally, I found a great PHP wrapper for Google API. This script comes with detailed installation instructions and is extremely easy to integrate with the look and feel of a site. As long as you have a Google API key (which is free and easy to obtain) and your site is indexed by Google, this script will work like a charm and fulfills all of my requirements.

If you would like to see what the results look like, you can check it out here.

Until recently I had well over 1000 bookmarks which were centrally located in my online bookmark manager (Linkagogo). I had loads of items I’ve bookmarked and never accessed again, so during the last year I’ve been using Furl, Yahoo! MyWeb and del.icio.us for storing pages and tagging content. I also relied on recently introduced search histories provided by Google and Yahoo. In the past few months the variety of places I used to store information caused a small chaos and often frustration, I was getting slower in finding the information I knew I once accessed so it was time for reorganizing the process.

Having been used all of the mentioned services, I settled for the following:

• I’m now using bookmarks only for pages which I access on a regular basis
• I continue to use Linkagogo to synchronize bookmarks across browsers
• I use Furl for storing interesting pages and tagging them with keywords
• I do most of my online searching through Google, with the search history enabled

Of course, this process is complemented with numerous RSS feeds I track through my feed reader.

Bookmarks

I now have only around 150 bookmarks. They consists strictly of links to the sites I visit on a regular basis (such as online services, forums, my websites, etc). The hierarchy is at most 2 levels deep, is very intuitive and I’m now taking full advantage of Opera’s nicknames.

Saving pages

I have moved all other previously bookmarked items to Furl, which is now the core component of my web usage. I save every interesting page I stumble upon and tag it if needed. The search is fast and flexible (can search through pages’ full content, keywords only, etc). Furl allows me to export bookmarks if I ever want to switch to another service, provides means to download the complete content of “my personal web file” (that’s quite cool) and provides an RSS feed of my public links. I’ve added Furl as an Opera search item (URL=http://www.furl.net/search?enc=UTF-8&search=true&showRead=all&keyword=%s), so my saved pages are just an “f whatever” away. Yes, I love shortcuts .

In case you wonder why I didn’t go for Yahoo’s MyWeb - the main reason is that it doesn’t have the exporting flexibility of Furl. Jeremy Zawodny mentions that it does offer export through an RSS feed, but we all know that that’s not what export is.

Search History

I quite liked Yahoo’s search history, but some of the things that bothered me about it is that I would often find it turned off (hence missing some of the searches). I sometimes find it difficult to navigate through their pages and they don’t always render correctly in Opera. On the other hand, Google’s search history just works and it does so unobtrusively. I rarely search through the search history, but it is good to know that all of my searches are there in case the need arises.