Problem Statement

Desktop Search Engine

• Designing deskptop search engine using LUCENE API in java. Desktop search engine will searches for the given string in the text files and pdf files of the desktop in the specified folders. Input given to search engine was search string , paths to be searched and paths not be searched.The output will be the list of files containing the text along with their path. The ordering of the files done by the ranking
• Input : Directories to be searched , directories not be searched and query string
• Output : Files containing the search string orderby ranking

Input Screen

Output Screen

Motivation

• Searching based on indexing is faster compared to nonindexbased search. So i want to design a desktop application which will search file content based on indexing.
• Providing different search criterias will give flexibilty for users.
• In the current application senario we can index a perticular folder and and its subfolders which will make indexing time shorter.(Avoids unnecessary indexing of all files when you want to search in perticular folder and its subfolders)

Intial Goals and Final achivements

• My initial goal is to index text and pdf files and search the given query string in the indexed files
• I final achievement is i am able to index text and pdf files and able to search the given string in the indexed files basing on the additional parameters such as file type , directories to be searched and the directories not to be searched

Break up of the Effort by module

• Index Module : 4 hours
• Input Module : 6 hours
• Search Module : 6 hours
• Output Module : 4 hours

Solution design

• Architecture Diagram

Solution design cont...

Description of the modules

• Input Module : Takes input from the user it is a swing based UI interface.

• Index Module : Directory entered by the user indexed and the index files kept in the directory specified by the user.

• Search Module : Basing on the inputs given by the user example directories to be searched , directories not be searched , maximum number of results to be displayed and file formats queries will be generated , all the quries of different criteria are combined and the query executed . Results will be sent to output module.

• Output Module : Output of the query will displayed and provides option to open the file.

Solution design cont...

Flow of the application

• First step indexing the files this can be done by entering the directory to be indexed and directory to store the indexed files.

• After completing indexing of files enter the parameters based on which search needs to be performed. These are inputs for the application

• Search module searches the query data and sends the output files list to output module.

• Output of the query will be displayed on the output window

Implementation details cont...

• Programming language : Java
• Indexing API : LUCENE
• packages used :
  lukeall-0.9.2.jar contains lucene
  pdfbox-0.7.3.jar for reading text from pdf
  FontBox-0.1.0-dev.jar internally used by pdfbox
  rt.jar for swing layout
  appframework-1.0.3.jar, swing-worker-1.1.jar : These jars used by netbeans while designing the swing component

Implementation details

• Index Module : I have implemented the code basing on the examples provided over internet for reading the data from pdf files
• Input Module : I have written the actions for the fields and the UI is created using NetBeans
• Search Module : I have implemented the code.
• Output Module : I have implemented the logic and UI for this module.

I have written approximately 700 lines of code

• Following are the test cases i have executed Testcases

Challenges Involved

• Exploring LUCENE is the major challenge involved and good learning also.

• Index Module : How to index non text files using lucene in my case indexing pdf files.
  • I have used pdfbox jar to convert pdf data to text and then indexed pdf files

• Input Module : Design of the input module so that inputs are taken in properorder.
  • I have enabled and disabled buttons to restrict user to enter inputs in different order

• Search Module : How to combine multiple queries.
  • I have used BooleanQuery for combining different quires.

• Output Module : How to open files in linux when filepath and type of file was given.
  • I have used Runtime.getRuntime().exec() method to open the file.

Things i tried and failed

• I have tried to keep auto indexing when ever there is change in file system using JNotifier but i am not able to do that.

Conclusions and future work

Conclusions

• Useful application for searching text in the files and faster because searching is done using indexed files.

Future Work

• Creating more user friendly interface for outputting the results
• Including JDirectory chooser module to select the directory
• Adding process which will notify the changes in the file system.