Hi
I find where a problem with NLS in html-docs in Web Content portlet.1.0
- the InputStreamReader's encoding take the value from ( for some
reasone ) is incorrected httpMethod.getResponseCharSet(). Look at this:
org.apache.portals.applications.webcontent.portlet.WebContentPortlet {
// Create the httpMethod object
doWebContent(...,RenderRequest request,..) {
...
HttpClient httpClient = getHttpClient(request);
httpMethod = getHttpMethod(httpClient, getURLSource(sourceAttr,
sourceParams, request, response), sourceParams, method, request);
...
}
// Use a httpMethod object for getting response.
doHttpWebContent(..., HttpMethod httpMethod,...) {
...
String encoding = ((HttpMethodBase)httpMethod).getResponseCharSet();
if (encoding == null)
// Vladimir Figurov comment - NOT ONCE NO RUN HERE -
encoding is ISO-8859-1
encoding = getContentCharSet(bis);
Reader htmlReader = new InputStreamReader(bis, encoding);
...
}
}
As I see, the Author is try read the first the source's HttpServer
encoding, and only if is NULL, try getted encode value from html-doc's
head...
But I think, that the must read from html-doc's in first - since the on
ONE HttServer instance may be a perfectly hosted a different html-docs
with different encoding... and only if html.doc's not including a <meta
.. encoding> tag - then read a Server encoding parameter.
--
Vladimir
