Arctus Blog

Getting Microsoft Office SharePoint Server 2007 to search inside PDF files

adrian@arctus.co.uk (Adrian O'Connor) — Thu, 06 Dec 2007 11:40:00 -0500

It is possible to get MOSS 2007 to search inside your PDF documents. This solution has not been tested (by us) on WSS3, so we’d be interested to hear back from anybody who tries.

Acrobat reader

Previously, searching PDF documents with Index Server required an IFilter plugin from Adobe. The most recent version of the IFilter is 6.0, and it is now several years old.

Adobe have taken a different approach now, and bundled the IFilter as part of the PDF reader application. This means that you need to download and install Adobe Acrobat Reader on your SharePoint server.

http://www.adobe.com/products/reader/

SharePoint configuration

There is very little configuration to be done inside SharePoint itself, aside from configuring the search service to index files with the extension .pdf.

You do this in SharePoint 3.0 Central Administration (see the Microsoft Office Server folder in the Start Menu on the server).

You should open the search configuration screen in the Shared Services section. Click File Types, and in the File Types list click Add New. You should enter pdf and click OK.

Registry Changes

There are two registry keys that you need to change. These will register the Adobe PDF IFilter with the Office Search service.

The values that need to be changed are:

HKEY_LOCAL_MACHINE
SOFTWARE\
  Microsoft\
    Office Server\
      12.0\
        Search\
          Setup\
            ContentIndexCommon\
              Filters\
                Extension\
                  .pdf

HKEY_LOCAL_MACHINE
SOFTWARE\
  Microsoft\
    Shared Tools\
      Web Server Extensions\
        12.0\
          Search\
            Setup\
              ContentIndexCommon\
                Filters\
                  Extension\
                    .pdf

Both values should be changed to:

{E8978DA6-047F-4E3D-9C78-CDBE46041603}

Add the Adobe Program Directory to the System Path

So that the search service can find the DLL which provides the IFilter service, you must add the Adobe Acrobat Reader program directory to the system path.

Right click My Computer > Properties Click Advanced > Environment Variables

In the lower part of the window, scroll down to find the variable Path. Double click on it, and at the end of the ‘Variable value’ add ;C:\Program Files\Adobe\Reader 8.0\Reader

Note the semi-colon – that separates this path from the one before it. Note also that if you installed to a non-default location that you must change this value accordingly.

Restart the Office Search service

To register your changes, you should restart the Office Search service. Open a command shell (Start -> Run -> cmd [press enter]) and type the following commands:

sc stop osearch [press enter]
sc start osearch [press enter]

Finally

If you already have PDF documents in SharePoint that you want to be included in the search, you must Reset all crawled content in Search Settings and then, in Content sources and crawl settings, start a new Full Crawl. These screens are in the SharePoint Central Configuration application’s Shared Services that we used in step 2.

MySQL DDL Commands

adrian@arctus.co.uk (Adrian O'Connor) — Fri, 09 Nov 2007 08:09:00 -0500

At Arctus we like to automate the management of our database systems. We especially like to write tools that can compare and upgrade schemas. To do this, we need to access the Data Definition Language (DDL) that created the objects in the first place. In Oracle there is a dedicated DDL package for this very job and in SQL Server we have the SYS tables (and the new INFORMATION_SCHEMA views).

MySQL does it a slightly different way, and the following commands are extremely useful:

SHOW TABLES;	Lists all tables by name
SHOW TABLE STATUS;	Lists all tables with additional information – date created, number of rows etc.
SHOW CREATE TABLE table_name	Shows the DDL used to create the table
SHOW COLUMNS FROM table_name	Lists the columns for a given table in a results set, with type information and other metadata

Finally, like SQL Server, there is a way of renaming a table:

RENAME TABLE current_name TO new_name;

If (like us) you occasionally have to deal with MySQL but it is not your primary database, we hope you find this little list of commands useful.

Apple pricing, UK vs. Stateside

adrian@arctus.co.uk (Adrian O'Connor) — Fri, 02 Nov 2007 12:32:00 -0400

In a recent post (adobe creative suite 3 pricing is seriously unfair to uk buyers) we noted that Adobe CS 3 Web Premium was significantly more expensive for UK buyers than those in the States.

Historically, this used to also be true for apple equipment. Today, however, things seem better. We’ve just done a full price comparison and on average, UK prices are 10% more than US prices.

Prices

Range/Model	GB ex	GB inc	US ($)	US (£)	US (£, inc)	Difference
Mac mini
1.83GHz: Combo Drive	£339.57	£399.00	$599.00	£299.50	£351.91	13.4%
2.0GHz: Super Drive	£424.68	£499.00	$799.00	£399.50	£469.41	6.3%
MacBook
13-inch: White 2.0GHz	£594.89	£699.00	$1,099.00	£549.50	£645.66	8.3%
13-inch: White 2.2GHz	£705.53	£829.00	$1,299.00	£649.50	£763.16	8.6%
13-inch: Black 2.2GHz	£807.66	£949.00	$1,499.00	£749.50	£880.66	7.8%
Imac
20-inch: 2.0GHz	£680.00	£799.00	$1,199.00	£599.50	£704.41	13.4%
20-inch: 2.4GHz	£807.66	£949.00	$1,499.00	£749.50	£880.66	7.8%
24-inch: 2.4GHz	£977.87	£1,149.00	$1,799.00	£899.50	£1,056.91	8.7%
24-inch: 2.8GHz	£1,241.70	£1,459.00	$2,299.00	£1,149.50	£1,350.66	8.0%
MacBook Pro
15-inch: 2.2GHz	£1,105.53	£1,299.00	$1,999.00	£999.50	£1,174.41	10.6%
15-inch: 2.4GHz	£1,360.85	£1,599.00	$2,499.00	£1,249.50	£1,468.16	8.9%
17-inch: 2.4GHz	£1,531.06	£1,799.00	$2,799.00	£1,399.50	£1,644.41	9.4%
Mac Pro
Base model	£1,445.96	£1,699.00	$2,499.00	£1,249.50	£1,468.16	15.7%

Our Thoughts

The current Mac range represents value for money and offers something unique when compared to other PCs.

The weak point in the range is probably the Mac Book Pro, which has a lower screen resolution than comparable HP business notebooks, and if you are going to run Windows or Linux HP might actually serve you better (see for example their 8150p notebook, sporting a 1680×1050 WSXGA+ display and genrally higher specification than the Mac Book).

The iMac is currently Apple’s strongest PC, ofering a truly impressive build with reasonable specification.

Problem storing large amounts of data in a MySQL database

adrian@arctus.co.uk (Adrian O'Connor) — Thu, 01 Nov 2007 16:24:00 -0400

MySQL isn’t a database server that we use an awful lot at Arctus, but because it is so hugely popular it must be our number 1 priority when it comes to supporting it for our Rails applications.

We were concerned when two of our current applications, a web mail client and a document management system, sometimes failed to save data to the database for no apparent reason. We suspected that it was caused by too much data (because shorter messages/documents worked), but the error message we were getting gave absolutely no clues – “MySQL server has gone away” is all it offered.

We first suspected that our TEXT field wasn’t able to hold the amount of data we were passing in, but LONGTEXT didn’t work either. We quickly tracked the problem down to a run-time setting called max_allowed_packet. This is an absolute limit on any individual command’s length, and we had exceeded the build in default.

We located my.ini, added max_allowed_packed=16M and restarted the server. Now everything is working fine.

We know that this is quite a trivial problem to overcome, but we are documenting it here because the current top few Google hits aren’t as useful as they might be for that particular error.

See http://dev.mysql.com/doc/refman/5.0/en/packet-too-large.html for more details.

Adobe Creative Suite 3 pricing is seriously unfair to UK buyers

adrian@arctus.co.uk (Adrian O'Connor) — Tue, 30 Oct 2007 12:50:00 -0400

Interested in buying the latest version of Adobe Creative Suite 3? If you are in the UK you will pay 50% more than if you are in America. That’s before you add V.A.T.

In America, you pay $1600 for the FULL version of CS3 Web Premium.

In the UK, you pay £1200 for the FULL version. That’s $2400. If you pay VAT, the final price is £1400, or $2800.

This massive gulf is completely unjustified. We’ll write to Adobe and ask them why it is so. If we get a response, we’ll post it here.

As an aside, I’ll give Inkscape an honourable mention – it is shaping up to be a rather fine replacement for Illustrator. In fact, we now use Inkscape for nearly all of our basic vector work. Best of all, it is Free and Open. Visit http://www.inkscape.org to download it!

With the GIMP team now focused on bringing significant UI improvements, maybe the web developer’s need for Adobe tools will diminish significantly in the not-too-distant future.

DocBox Beta 1

adrian@arctus.co.uk (Adrian O'Connor) — Thu, 18 Oct 2007 10:05:00 -0400

We have released beta 1 of DocBox, our tag based document management system.

This is a very minor release, and fixes the problem we had with FastCGI and Rails using 100% of the CPU. Functionality of the application remains unchanged.

The installer can be downloaded at the DocBox product page.

FastCGI problem

The problem is that FastCGI continuously spawns ruby processes, which use 100% of the CPU. Mod_fastcgi produces very little useful logging information, and although we could see that Ruby was being invoked, we had no real idea as to where the problem lay.

Initially, we suspected that a conflicting version of Rails was the culprit. Our first release of DocBox coincided with the release of Rails 1.2.4, and it just so happened that the first system to show this problem was the first system on which we had installed Ruby’s OneClick package and Rails 1.2.4. Since then we’ve seen the problem happen elsewhere and we can’t find a common link.

I’ll write a detailed analysis of our installer soon, but the major component is an almost full copy of win32 ruby (minus the docs and gem cache). We updated Ruby and all of the gems, repackaged the installer and still the same problem affected at least three of our test systems.

Our final resolution was to remove Apache and use Mongrel as a win32 service. This works well for small-scale installations but won’t scale without some extra work.

Hopefully we’ll resolve the FastCGI problem one day, but we’ll also look at other options.

In the meantime, please let us know what you think to DocBox!

Introducting DocBox: Our open source document management system

adrian@arctus.co.uk (Adrian O'Connor) — Tue, 09 Oct 2007 08:32:00 -0400

We would like to (quietly) announce that we have just released the first beta of our new document management system. This system is currently called DocBox.

What is DocBox?

DocBox is a tag-based document management system, written in Ruby on Rails. It’s like del.icio.us, but for your documents. Because your documents are private, you run DocBox on your own server or workstation, not on our website.

We’ve started using it internally alongside our existing folders-based .NET solution (WebShare). We’ll still continue to develop and sell WebShare, but we’re really enjoying the speed and agility of DocBox.

It features a clean, modern interface, and a java-based bridge between the web site and your desktop.

While many features are yet to be added, and many bugs need to be ironed out, we’ve decided that it is time to let you see how it’s shaping up.

To download a copy, please visit the DocBox product page

Cost

We are going to release DocBox under the terms of the GPL, version 3. This effectively makes it free, but commercial customers will be asked to pay for downloading the software. For this they will receive support and service agreements. If they do not wish to pay, they may still use the software but will receive no quality assurance or support.

Initially, we anticpate charging no more than £500 for a workgroup license, and £80 for a small-office/home-office version. Non-profit orginsations, schools and personal use will be free.

One-click Installer

One of the exciting things about DocBox is that it installs everything from a single executable, and runs as a stand-alone service on your Microsoft Windows server or workstation. Many companies who have evaluated Rails for their own products will be keen to know that a zero configuration installer for Rails is possible, and with DocBox we prove it is so. Get in touch if you’d like us to help you package your Rails application!

The Future

We are currently working on: Templates, Security & Permissions, Improved Tag Browsing for release sometime towards the middle of November.

Beyond that, we plan to add a full web-service api for automation and integration, workflow and advanced notifications.

Finding an unused network port for server applications with Ruby

adrian@arctus.co.uk (Adrian O'Connor) — Thu, 04 Oct 2007 05:05:00 -0400

When delpoying web applications that run their own web-servers, it is quite likely that the port you want (normally port 80) will be in use. At the very least, it is only good manners to check first.

Here’s a quick Ruby script that will loop through an array of port numbers and return the first one that is free:

require 'socket'

def find_unused_port

  ports = [80, 8000, 8080, 8800, 8088, 8001, 8002, 8003, 8004, 8005, 8006, 8007, 8008]
  port = 0

  ports.each do |p|

    begin

      port = p
      TCPServer.new('localhost', port) rescue port = 0
      if (port > 0) then break; end

    end

  end

  port

end

puts find_unused_port.to_s

It is Ruby because my own applications include a packaged version of the Ruby interpreter. This affords me a great deal of power, being able to execute .rb scripts as part of the install process. You can do similar things in other languages, or, if you don’t mind processing large amounts of text, using the command netstat -a -n, and parsing all that are “LISTENING”.

Debugging in NUnit with Visual Studio 2005

adrian@arctus.co.uk (Adrian O'Connor) — Mon, 01 Oct 2007 12:09:00 -0400

You already know that the debugger is one of the best tools in the entire Visual Studio package. You might not be aware that it is possible to debug your source code when it is running in NUnit (and other code-hosting applications).

To debug your project, follow these simple instructions:

Make sure that Visual Studio 2005 and NUnit are both loaded and that you have opened your project in each.

In Visual Studio, select “Attach to Process” on the Debug menu. Because NUnit is running, you will see nunit.exe in the list of processes. Double click on it.

That’s literally it! Now you can set break-points in your test code (and any code that your test code references), and step through it just as you would if it were running in Visual Studio.

There is one thing to note: when you are compiling and testing in regular and small steps, you might find that things get ‘out of sync’. The first you know about it is likely that Visual Studio tells you that it “can’t find the code for the current location” (sorry, the exact error message slips my memory right now, and I can’t recreate it).

When this happens, close NUnit. If you have the options available, I also suggest running a ‘Clean Solution’ and ‘Rebuild Solution’ before re-loading NUnit. That always fixes the problem, at least for a while.

Subversion Hosting at CodeSpaces.com

adrian@arctus.co.uk (Adrian O'Connor) — Thu, 27 Sep 2007 05:03:00 -0400

If you are interested in trying out Subversion for your team, or are looking for a reliable and cost-effective Subversion host, I recommend you try CodeSpaces.com.

Apart from Subversion hosting, CodeSpaces is also a Trac-like project management system, built on top of the marvellous Ext-js toolkit. If nothing else, it’s a great example of what Ext-js can do.

A single user, single repository account is free, and provides an ideal way to try before you buy. The cost of a multi-user plan seems most reasonable.

Arctus are not affiliated with CodeSpaces in any way, we just like their service.