[SLING-1761] JcrPropertyResource sets incorrect content length for strings containing non-ascii character

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: JCR Resource 2.0.6
• Fix Version/s: JCR Resource 2.0.8
• Component/s: JCR
• Assignee: Carsten Ziegeler
• Reporter: Julian Sedding
• Created: 11/Sep/10 08:41
• Updated: 28/Jan/11 09:05
• Resolved: 30/Dec/10 12:51

Description

JcrPropertyResource sets the content length of the property in its metadata. To do so, it uses javax.jcr.Property#getLength() to determine the content length.

The documentation for javax.jcr.Property#getLength() states "[...] Returns the length in bytes if the value is a PropertyType.BINARY, otherwise it returns the number of characters needed to display the value in its string form. [...]".

The documentation in ResourceMetadata is not explicit, but from its usage in StreamRendererServlet I conclude that ResourceMetadata.getContentLength() is intended for use in the Content-Length HTTP header. If my assumptions are correct, the content length indicates the number of bytes in the string, while javax.jcr.Property#getLength() returns the number of characters.

The effect of this can be observed by the following steps:

• create a string property "/utf8string" with value "Bär"
• access this property using a browser (e.g. http://localhost:8888/utf8string), so that the property gets rendered by the StreamRendererServlet
  => the string is rendered incorrectly (due to a missing Content-Type header)
  => the string is cut off (due to the incorrectly set Content-Length header)