[JSPWIKI-411] LuceneSearchProvider dies with corrupted files

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.6.4
• Fix Version/s: 2.8.1
• Component/s: Plugins

Environment

LuceneSearchProvider and ReadableAttachmentProvider (

• Assignee: Harry Metske
• Reporter: Kurt Stein
• Created: 04/Nov/08 07:37
• Updated: 10/Sep/11 23:11
• Resolved: 04/Nov/08 18:41

Description

If someone uploads a corrupted files(i.e. 0-Byte or non readable pdf-content) the lucene index crashes. Even with a new full reindex the lucene indexer chrashes every time he touches the corrupted file.

I have analyzed the code. In the method doFullLuceneReindex() there is a Try- Catch block that catches the exception coming form the call getAttachmentContent(att.getName(), WikiProvider.LATEST_VERSION ).
But thus the indexer stops at this points and no other attachment will be added to the lucene index.

for( Iterator iterator = allAttachments.iterator(); iterator.hasNext(); )

{ Attachment att = (Attachment) iterator.next(); String text = getAttachmentContent( att.getName(), WikiProvider.LATEST_VERSION ); luceneIndexPage( att, text, writer ); }

I would fixed it like this

for( Iterator iterator = allAttachments.iterator(); iterator.hasNext(); )
{
Attachment att = (Attachment) iterator.next();
try

{ String text = getAttachmentContent( att.getName(), WikiProvider.LATEST_VERSION ); luceneIndexPage( att, text, writer ); }

catch (Throwable t)

{ //goto next Attachment iterator log.error("Error on indexing file " + att.getFileName(), t); }

}