in src/org/apache/xml/utils/URI.java [365:590]
private void initialize(URI p_base, String p_uriSpec)
throws MalformedURIException
{
if (p_base == null
&& (p_uriSpec == null || p_uriSpec.trim().length() == 0))
{
throw new MalformedURIException(
XMLMessages.createXMLMessage(XMLErrorResources.ER_CANNOT_INIT_URI_EMPTY_PARMS, null)); //"Cannot initialize URI with empty parameters.");
}
// just make a copy of the base if spec is empty
if (p_uriSpec == null || p_uriSpec.trim().length() == 0)
{
initialize(p_base);
return;
}
String uriSpec = p_uriSpec.trim();
int uriSpecLen = uriSpec.length();
int index = 0;
// check for scheme
int colonIndex = uriSpec.indexOf(':');
if (colonIndex < 0)
{
if (p_base == null)
{
throw new MalformedURIException(XMLMessages.createXMLMessage(XMLErrorResources.ER_NO_SCHEME_IN_URI, new Object[]{uriSpec})); //"No scheme found in URI: "+uriSpec);
}
}
else
{
initializeScheme(uriSpec);
uriSpec = uriSpec.substring(colonIndex+1);
// This is a fix for XALANJ-2059.
if(m_scheme != null && p_base != null)
{
// a) If <uriSpec> starts with a slash (/), it means <uriSpec> is absolute
// and p_base can be ignored.
// For example,
// uriSpec = file:/myDIR/myXSLFile.xsl
// p_base = file:/myWork/
//
// Here, uriSpec has absolute path after scheme file and :
// Hence p_base can be ignored.
//
// b) Similarily, according to RFC 2396, uri is resolved for <uriSpec> relative to <p_base>
// if scheme in <uriSpec> is same as scheme in <p_base>, else p_base can be ignored.
//
// c) if <p_base> is not hierarchical, it can be ignored.
//
if(uriSpec.startsWith("/") || !m_scheme.equals(p_base.m_scheme) || !p_base.getSchemeSpecificPart().startsWith("/"))
{
p_base = null;
}
}
// Fix for XALANJ-2059
uriSpecLen = uriSpec.length();
}
// two slashes means generic URI syntax, so we get the authority
if (uriSpec.startsWith("//"))
{
index += 2;
int startPos = index;
// get authority - everything up to path, query or fragment
char testChar = '\0';
while (index < uriSpecLen)
{
testChar = uriSpec.charAt(index);
if (testChar == '/' || testChar == '?' || testChar == '#')
{
break;
}
index++;
}
// if we found authority, parse it out, otherwise we set the
// host to empty string
if (index > startPos)
{
initializeAuthority(uriSpec.substring(startPos, index));
}
else
{
m_host = "";
}
}
initializePath(uriSpec.substring(index));
// Resolve relative URI to base URI - see RFC 2396 Section 5.2
// In some cases, it might make more sense to throw an exception
// (when scheme is specified is the string spec and the base URI
// is also specified, for example), but we're just following the
// RFC specifications
if (p_base != null)
{
// check to see if this is the current doc - RFC 2396 5.2 #2
// note that this is slightly different from the RFC spec in that
// we don't include the check for query string being null
// - this handles cases where the urispec is just a query
// string or a fragment (e.g. "?y" or "#s") -
// see <http://www.ics.uci.edu/~fielding/url/test1.html> which
// identified this as a bug in the RFC
if (m_path.length() == 0 && m_scheme == null && m_host == null)
{
m_scheme = p_base.getScheme();
m_userinfo = p_base.getUserinfo();
m_host = p_base.getHost();
m_port = p_base.getPort();
m_path = p_base.getPath();
if (m_queryString == null)
{
m_queryString = p_base.getQueryString();
}
return;
}
// check for scheme - RFC 2396 5.2 #3
// if we found a scheme, it means absolute URI, so we're done
if (m_scheme == null)
{
m_scheme = p_base.getScheme();
}
// check for authority - RFC 2396 5.2 #4
// if we found a host, then we've got a network path, so we're done
if (m_host == null)
{
m_userinfo = p_base.getUserinfo();
m_host = p_base.getHost();
m_port = p_base.getPort();
}
else
{
return;
}
// check for absolute path - RFC 2396 5.2 #5
if (m_path.length() > 0 && m_path.startsWith("/"))
{
return;
}
// if we get to this point, we need to resolve relative path
// RFC 2396 5.2 #6
String path = new String();
String basePath = p_base.getPath();
// 6a - get all but the last segment of the base URI path
if (basePath != null)
{
int lastSlash = basePath.lastIndexOf('/');
if (lastSlash != -1)
{
path = basePath.substring(0, lastSlash + 1);
}
}
// 6b - append the relative URI path
path = path.concat(m_path);
// 6c - remove all "./" where "." is a complete path segment
index = -1;
while ((index = path.indexOf("/./")) != -1)
{
path = path.substring(0, index + 1).concat(path.substring(index + 3));
}
// 6d - remove "." if path ends with "." as a complete path segment
if (path.endsWith("/."))
{
path = path.substring(0, path.length() - 1);
}
// 6e - remove all "<segment>/../" where "<segment>" is a complete
// path segment not equal to ".."
index = -1;
int segIndex = -1;
String tempString = null;
while ((index = path.indexOf("/../")) > 0)
{
tempString = path.substring(0, path.indexOf("/../"));
segIndex = tempString.lastIndexOf('/');
if (segIndex != -1)
{
if (!tempString.substring(segIndex++).equals(".."))
{
path = path.substring(0, segIndex).concat(path.substring(index
+ 4));
}
}
}
// 6f - remove ending "<segment>/.." where "<segment>" is a
// complete path segment
if (path.endsWith("/.."))
{
tempString = path.substring(0, path.length() - 3);
segIndex = tempString.lastIndexOf('/');
if (segIndex != -1)
{
path = path.substring(0, segIndex + 1);
}
}
m_path = path;
}
}