Click here to Skip to main content
14,268,184 members

Desktop Search Application: Part 1

Rate this:
4.89 (91 votes)
Please Sign up or sign in to vote.
4.89 (91 votes)
6 Jul 2006
Building an application that searches your Office documents in tenths of a second.

Sample Image - screenshot1.gif


Let's take the following exercise: Build a C# application that instantly searches your Documents folder. We need the application to:

  • Search the contents of Office documents (Word, Excel, PowerPoint). I.e., XLS, DOC, PPT files.
  • Search the contents of HTML documents.
  • Search the contents of text documents.
  • Search quickly (i.e., give results in less than a second).
  • Allow to open the documents directly from the results list.

Update: This Desktop Search application is now part of the Seekafile Server 1.0 - Open-Source Indexing Server. The server provides automatic indexing on the background, and you can search the index through a Windows Forms client search application. See also the Seekafile Server roadmap.

Task 1: Full-Text Indexing

For the indexing and searching, we will use an excellent search engine called DotLucene. It's a C# port of Java Lucene, maintained by George Aroush. It has many great features:

  • Very good performance.
  • Ranked search results.
  • Search query highlighting in results.
  • Searches structured and unstructured data.
  • Metadata searching (query by date, search custom fields...).
  • Index size approximately 30% of the indexed text.
  • Can also store fully indexed documents.
  • Pure managed .NET in a single assembly.
  • Very friendly licensing (Apache Software License 2.0).
  • Localizable (support for Brazilian, Czech, Chinese, Dutch, English, French, German, Japanese, Korean, and Russian included in the DotLucene National Language Support Pack).
  • Extensible (source code included).

For more details on creating the index and searching, see my previous article: DotLucene: Full-Text Search for Your Intranet or Website using 37 Lines of Code.

Task 2: Parsing Office Documents

As DotLucene can index only plain text, we need to parse the Office documents and extract text from them. Reading their binary structure isn't an easy job. However, on Windows 2000+, we can use the IFilter interface which is a part of the Windows Indexing Service. This is installed by default on all Windows 2000+ systems (no Office installation is required).

The IFilter API is also being used by the Windows Desktop Search (MSN Search Toolbar) and Lookout, so you don't have to be afraid that we will use something obscure to parse the documents. It can also parse other file types if you install the appropriate filter.

Working with the IFilter interface requires a lot of COM interop which is a bit tricky. After tweaking the samples available on the web, I finally had a code that worked correctly in most cases:

public static string Parse(string filename)
  IFilter filter = null;
  try {
    StringBuilder plainTextResult = new StringBuilder();
    filter = loadIFilter(filename); 
    STAT_CHUNK ps = new STAT_CHUNK();
    IFILTER_INIT mFlags = 0;
    uint i = 0;
    filter.Init( mFlags, 0, null, ref i);
    int resultChunk = 0;
    resultChunk = filter.GetChunk(out ps);
    while (resultChunk == 0)
      if (ps.flags == CHUNKSTATE.CHUNK_TEXT)
        uint sizeBuffer = 60000;
        int resultText = 0;
        while (resultText == Constants.FILTER_S_LAST_TEXT || resultText == 0)
          sizeBuffer = 60000;
          System.Text.StringBuilder sbBuffer = 
             new System.Text.StringBuilder((int)sizeBuffer);
          resultText = filter.GetText(ref sizeBuffer, sbBuffer);
          if (sizeBuffer > 0 && sbBuffer.Length > 0)
            string chunk = sbBuffer.ToString(0, (int)sizeBuffer);
      resultChunk = filter.GetChunk(out ps);
    return plainTextResult.ToString();
    if (filter != null)

Assembling the Application

In short:

  • The index can only be built from scratch.
  • We are not returning the sample of the found document.
  • We are skipping the files that can't be parsed successfully.
  • We are loading the associated explorer icon for each document in the results.
  • You can choose the folder to be indexed; by default, it's your Documents folder.
  • The indexed file types are hard-coded (txt, htm/html, doc, xls, ppt).
  • The index is stored in your profile in the Local Settings/Application Data/DesktopSearch folder.
  • Remember that it is possible to search while the indexing is in progress.


Some statistics (Athlon XP 2000+, 1GB RAM, Seagate SATA drive 7200 RPM):

  • Documents indexed: 1185
  • Total size of indexed documents: 120,690,622 bytes
  • Rebuilding the index took: 5 minutes 55 seconds
  • Index size: 4,339,950 bytes
  • Search time (including the rendering on display): from 0.0937 seconds (25 found items) to 0.3125 seconds (213 found items)

To Be Continued...

In the next part of the article, we will extend this simple application with the following features:

  • The indexing will be handled by a separate application which will update the index continuously on the background.
  • The results will contain a sample of the document with highlighted query words.
  • We will index and search the file name and the last modified date.

Resources and Acknowledgements

Search Engine:

Seekafile Server - Open-Source Indexing Server

Office Documents Parsing:



This article has no explicit license attached to it but may contain usage terms in the article text or the download files themselves. If in doubt please contact the author via the discussion board below.

A list of licenses authors might use can be found here


About the Author

Dan Letecky
Czech Republic Czech Republic

Comments and Discussions

QuestionUnable to filter Msoffice Documents Pin
Member 974777518-Mar-13 2:05
memberMember 974777518-Mar-13 2:05 
QuestionUnble to filter office Documents Pin
Member 974777518-Mar-13 0:09
memberMember 974777518-Mar-13 0:09 
Questionis it possible to use IFilters to extract multilanguage text content? Pin
Sweetdin18-Mar-10 20:19
memberSweetdin18-Mar-10 20:19 
GeneralThe IFilter doesn't parse Word docs correctly in office 2003 but work perfectly in office2007 Pin
8482977mn15-Mar-10 23:39
member8482977mn15-Mar-10 23:39 
Generalfinding just the filename Pin
emmmatty116-Aug-09 22:36
memberemmmatty116-Aug-09 22:36 
GeneralRe: finding just the filename Pin
8482977mn16-Mar-10 0:22
member8482977mn16-Mar-10 0:22 
GeneralQuestion Please Pin
hom.rcp25-Feb-09 4:30
memberhom.rcp25-Feb-09 4:30 
Generalwaoh! Pin
chenli051324-Nov-08 20:28
memberchenli051324-Nov-08 20:28 
GeneralHighlighter or fragmenter Pin
mbowles2019-Oct-07 4:31
membermbowles2019-Oct-07 4:31 
GeneralRe: Highlighter or fragmenter Pin
Dan Letecky9-Oct-07 21:48
memberDan Letecky9-Oct-07 21:48 
GeneralRe: Highlighter or fragmenter Pin
mbowles20110-Oct-07 3:15
membermbowles20110-Oct-07 3:15 
GeneralRe: Highlighter or fragmenter Pin
Dan Letecky10-Oct-07 3:32
memberDan Letecky10-Oct-07 3:32 
GeneralI look forward to Part 2 Pin
DrTom21-Jan-07 11:00
memberDrTom21-Jan-07 11:00 
GeneralRe: I look forward to Part 2 Pin
Dan Letecky2-May-07 4:59
memberDan Letecky2-May-07 4:59 
GeneralRe: I look forward to Part 2 Pin
mbowles20125-Sep-07 16:43
membermbowles20125-Sep-07 16:43 
QuestionHow to search in pdf files? Pin
Ajayarora14-Oct-06 0:56
memberAjayarora14-Oct-06 0:56 
AnswerRe: How to search in pdf files? Pin
Dan Letecky2-May-07 5:02
memberDan Letecky2-May-07 5:02 
GeneralVB.Net Version Pin
Jonathan M Patrick8-Sep-06 18:10
memberJonathan M Patrick8-Sep-06 18:10 
QuestionPart 2? Pin
james_carter18-Jul-06 2:45
memberjames_carter18-Jul-06 2:45 
AnswerRe: Part 2? Pin
Dan Letecky2-May-07 4:55
memberDan Letecky2-May-07 4:55 
GeneralRe: Part 2? Pin
james_carter2-May-07 5:01
memberjames_carter2-May-07 5:01 
QuestionIS Dotlucene Free? Pin
visva6-Apr-06 17:45
membervisva6-Apr-06 17:45 
AnswerRe: IS Dotlucene Free? Pin
Dan Letecky6-Apr-06 21:57
memberDan Letecky6-Apr-06 21:57 
GeneralTitle and score questions Pin
mbowles20120-Mar-06 6:22
membermbowles20120-Mar-06 6:22 
GeneralNeed Information Pin
Vijay Kumar Raja Grandhi3-Mar-06 3:15
memberVijay Kumar Raja Grandhi3-Mar-06 3:15 

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Praise Praise    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.

Posted 16 Jun 2005


253 bookmarked