[SLING-2143] SlingPostServlet ImportOperation :content parameter overrides _charset_ parameter with system default encoding

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Servlets Post 2.1.0
• Fix Version/s: Servlets Post 2.1.2
• Component/s: Servlets
• Environment: OS: Windows 2k8, Windows 7; Web Servers: Tomcat 6; Sling 6
• Assignee: Eric Norman
• Reporter: Stephen Sanchez
• Created: 14/Jul/11 16:58
• Updated: 20/Sep/19 11:34
• Resolved: 27/Jul/11 05:43

Description

The ImportOperation on the SlingPostServlet (2.1.0 to Trunk) does not support encoding specified in the form charset parameter on a POST request.

REPRODUCTION STEPS:
Create a POST request using the SlingPostServlet and ImportOperation:

curl -F":operation=import" -F"charset=UTF-8" -F":contentType=json" -F":replace=true" -F":replaceProperties=true" -F":content=

{'latin':'øµå', 'chinese':'玄牛'}

" http://admin:admin@localhost:8080

PROPOSED SOLUTION:

In ImportOperation.java, line 137 (as of 2.1.0 tag):

contentStream = new ByteArrayInputStream(content.getBytes());

This line will take the :content parameter on a request, properly encoded (ex. UTF-8) and get the bytes using system-level encoding. This causes all unicode characters in the content to be encoded, on windows, with the wrong encoding.

The simple fix, which resolves the issue and allows UTF-8 encoding across all operating systems:

contentStream = new ByteArrayInputStream(content.getBytes("UTF-8"));

My proposed fix, is to support the form parameter charset, since :content is a parameter on the form in the post request:

RequestParameter encodingParam = request.getRequestParameter("charset");
byte[] contentBytes;
if (encodingParam != null && encodingParam.getString() != null) {
contentBytes = content.getBytes(encodingParam.getString());
} else {
contentBytes = content.getBytes();
}
contentStream = new ByteArrayInputStream(contentBytes);