What is it?

SVNSearcher is a fulltext search engine for SVN repository. Currently it works on files types:

• Text files
• Adobe Acrobat's files (PDF)
• Microsoft Office's files (DOC, XLS, PPT)
• ZIP archive files (ZIP, JAR)

There is no warranty that all Microsoft Office's files will be correctly indexed, but tests take over 90% effectivity of indexing repository with several tens thousand MS Office's files.

How does it work?

SVNSearcher is composed of two application: SVNIndexer (creates and updates index) and SVNIndexer-webapplication (using index files to searching and provides user interface).

You can use them apart and the only information needed to integrate them is included in configuration file.

Logical architecture of all system is represented below:

• SVN Repository: repository you want to create index.
• Index (Apache Lucene): Index files created by SVNIndexer, using Apache Lucene. These files will be used in SVNIndexer-webapplication.
• SVNIndexer: Java application which creates index files.

Main component is a search engine based on Apache Lucene (http://lucene.apache.org). It affirms indexing and searching from indexed documents, but it requires text data. Lucene isn't able to connect to repository, so this part is realised by other modules used in SVNIndexer.

Indexing files is an incremental process, starts in regular time interval. First usage creates index files from all repository (or repositories). Each next running indexer updates index with updated files from SVN repository (or repositories).

Searching documents is realised in SVNIndexer-webapplication. It's a client of Apache Lucene and it also provides user interface. It uses only index files to search, doesn't connect to repository. Currently usage authorization option needs SVN authorization file to check user's access to found documents.

How to secure it for corporate use?

Security is provided in three ways:

• Authorization to SVNSearcher with digital cerfiticates. There is also possibility to enable authorization with login and password.


• Checking rights to SVN repository on object level: directory and files. It means that user can see only these documents where has read rights. This system is based on SVN authorization file, so all changes are immediately applayed if only SVN server uses the same authorization file (is installed on the same machine). For more information see Authorization configuration [option].


• Single-sign-on for other clients SVN repository (client of command line, Tortoise SVN, web client, search engine).

Can I install SVNSearcher on other machine than SVN repository?

Yes, SVNIndexer is completely independent from SVN server and the only information needed to indexing is access through http protocol to SVN server.
When you installed SVNIndexer on other machine than SVN repository, you can't use directPath property
(to automatically get repositories names) and must set repositories names in file ( repListFile ).

SVNIndexer-webapplication can also works on other machine than SVN repository, but note that in this case SVN authorization file must be copied from server to machine with SVNIndexer-webapplication. This makes that changes in access file on server won't be immediately applied in SVNSearcher.

Requirements

Before you start SVNIndexer you need:

• Java Virtual Machine version 1.5
• Compiled application SVNIndexer with configuration files
• External libraries:

• JavaSVN (http://tmate.org/svn)
• Lucene (http://lucene.apache.org)
• PDFBox (http://www.pdfbox.org)
• POI (http://jakarta.apache.org/poi)
• ICU4J (http://icu.sourceforge.net)
• Library to find file type (currently Jmimemagic)
• Library read ZIP archive (currently TrueZip)

All these libraries you can download with SVNIndexer binaries.

Installation and usage SVNIndexer

First thing you should prepare to install SVNIndexer is setting read, write and execute rights for SVNIndexer directory and also all files and subdirectories in this directory.

Configure SVNIndexer:

• Edit properties file (example configuration file - config.properties attached to application) to set all obligatory properties.
  You can find description about each parameter in example file.
  This file can be modified whenever you want and next indexing will be adjusted to changes.
  Note that this file is properties file also for SVNIndexer-webapplication, so some properties are commons for both applications.

Now you can install SVNIndexer in two selection:

• Installation SVNIndexer for starting from cron (Linux/Unix):
• Edit file addCron.sh to set path to SVNIndexer application directory.
• In addCron.sh you may also set exactly date of starting indexer (for more details see man crontab -S 5),
  or set path to properties file if it's different from default.
  Note that often starting SVNIndexer (for example for each 15 minute) gives similar effect to have indexing "on-line" (after each commit).
  Time needed to connect and check if there is new revision to update, for one repository, takes about 400 ms.
• Execute this script by ./addCron.sh


• Installation SVNIndexer for manually starting (Linux/Unix):
• Set viarable $SVNINDEXER_PATH to directory where you have SVNIndexer application, for example:

  export SVNINDEXER_PATH=/sample/path/SVNIndexer 

• Each one start SVNIndexer needs to give path to the properties file as a parameter.
  If you use default file you can start indexing by:

  $SVNINDEXER_PATH/startJar.sh $SVNINDEXER_PATH/conf.properties 



• Installation SVNIndexer for manually starting (MS Windows):
After configured properties file you can start indexing by running startJar.bat

Requirements

Before you start SVNIndexer-webapplication, you need:

• Java Virtual Machine version 1.5
• Tomcat 5
• Compiled SVNIndexer-webapplication (SVNSearcher-1.2.0.war (or later version) file)

Installation

Copy SVNSearcher-1.2.0.war to your $CATALINA_HOME/webapps and it will be automatically unpacked. Edit file SVNSearcher/WEB-INF/web.xml.
This directory was created after installed by Tomcat.
Set correct patch to conf.properties file (the same file which is used in SVNIndexer) in '<param-value>'.
After it restart your Tomcat.

If you want to SVNSearcher interprets UTF-8 characters, add this line in server.xml (in Connector node) parameter:

		 URIEncoding=”UTF-8”.

Using searcher

There is possibility to use all sytnax avalaible in Apache Lucene, which is similar to google sytnax.
Additionally SVNSearcher enables searching in META-DATA like:

• File content (FileBody)
• Datum of last modification document (Date)
• Author of last modification (Author)
• Comment to commiting file (Comment)
• File type (Type)
• Part of repository tree where start searching (Name)

For detail instruction see Help in SVNSearcher web site.

Authorization configuration [option]

This option check user rights for found documents and display only these where user has read rights

If you are going to use authorization option (you set authorization=true in configuration file)
you have to compile module libAuthorization.so - this module depends on SVN and operating system version. The module is required to check rights for user using SVNSearcher.

You also have to set path to SVN authorization file in access property in proterties file.
If you installed SVNIndexer-webapplication on other machine than SVN server and you can't set path to currently using SVN authorization file, you must copy this file on machine with SVNIndexer-webapplication.
Note that in this case all changes in this file on SVN server should be copied to file using in SVNIndexer-webapplication.

Before you use Makefile to create the module, you have to install and prepare:

• g++ compiler
• apr and apr util library (http://apr.apache.org/download.cgi)
• SVN headers files (http://svn.collab.net/repos/svn/trunk/subversion/include/)
• export viariable $APR_PATH to directory with APR/include
• export viariable $MY_JAVA_PATH to directory with Java VM
• export viariable $SVN_PATH to directory where you copied SVN header files
• export viariable $MY_LIB_PATH to directory with modules: ibsvn_repos-1.so and libsvn_subr-1.so. Default it's /usr/lib

Now you can execute make in Authorization directory

After building libAuthorization.so copy this file to $CATALINA_HOME/shared/lib
and export viarable $LD_LIBRARY_PATH:

			export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CATALINA_HOME/shared/lib.

This variable must be always exported when you start Tomcat.