fac.useSchemeSpecificRules("mailto", true); // XXX broken
fac.useSchemeSpecificRules("file", true);
fac.useSchemeSpecificRules("data", true); // XXX broken
// XXX javascript?
// fac.setQueryCharacterRestrictions(false);
IRI iri;
boolean data = false;
try {
CharSequencePair pair = splitScheme(literal);
if (pair == null) {
// no scheme or scheme is private
iri = fac.construct(trimHtmlSpaces(literal.toString()));
} else {
CharSequence scheme = pair.getHead();
CharSequence tail = pair.getTail();
if (isWellKnown(scheme)) {
iri = fac.construct(trimHtmlSpaces(literal.toString()));
} else if ("javascript".contentEquals(scheme)) {
// StringBuilder sb = new StringBuilder(2 +
// literal.length());
// sb.append("x-").append(literal);
// iri = fac.construct(sb.toString());
iri = null; // Don't bother user with generic IRI syntax
Reader reader = new BufferedReader(
new Utf8PercentDecodingReader(new StringReader(
"function(event){" + tail.toString() + "}")));
// XXX CharSequenceReader
reader.mark(1);
int c = reader.read();
if (c != 0xFEFF) {
reader.reset();
}
try {
Context context = ContextFactory.getGlobal().enterContext();
context.setOptimizationLevel(0);
context.setLanguageVersion(Context.VERSION_1_6);
// -1 for lineno arg prevents Rhino from appending
// "(unnamed script#1)" to all error messages
context.compileReader(reader, null, -1, null);
} finally {
Context.exit();
}
} else if ("data".contentEquals(scheme)) {
data = true;
iri = fac.construct(trimHtmlSpaces(literal.toString()));
} else if (isHttpAlias(scheme)) {
StringBuilder sb = new StringBuilder(5 + tail.length());
sb.append("http:").append(tail);
iri = fac.construct(trimHtmlTrailingSpaces(sb.toString()));
} else {
StringBuilder sb = new StringBuilder(2 + literal.length());
sb.append("x-").append(literal);
iri = fac.construct(trimHtmlTrailingSpaces(sb.toString()));
}
}
} catch (IRIException e) {
Violation v = e.getViolation();
/*
* Violation codes that are not "known" codes get assigned the
* dummy value so that handling of them will fall through to
* the default case.
*/
KnownViolationCode vc = KnownViolationCode.valueOf("ZZZ_DUMMY_DEFAULT");
try {
/*
* If this violation code is one of the "known" Jena IRI
* violation codes we want to handle specifically, then use it
* as-is.
*/
vc = KnownViolationCode.valueOf(v.codeName());
} catch (Exception ex) { }
switch (vc) {
case HAS_PASSWORD:
if (WARN) {
throw newDatatypeException(
underbarStringToSentence(v.component())
+ " component contains a password.",
WARN);
} else {
return;
}
case NON_INITIAL_DOT_SEGMENT:
if (WARN) {
throw newDatatypeException(
"Path component contains a segment \u201C/../\u201D not at the beginning of a relative reference, or it contains a \u201C/./\u201D. These should be removed.",
WARN);
} else {
return;
}
case PORT_SHOULD_NOT_BE_WELL_KNOWN:
return; // not a useful warning case
case COMPATIBILITY_CHARACTER:
if (WARN) {
throw newDatatypeException(
underbarStringToSentence(v.codeName()) + " in "
+ toAsciiLowerCase(v.component())
+ " component.", WARN);
} else {
return;
}
case DNS_LABEL_DASH_START_OR_END:
throw newDatatypeException("Host component contains a DNS name with a \u201C-\u201D (dash) character at the beginning or end.");
case DOUBLE_WHITESPACE:
case WHITESPACE:
throw newDatatypeException("Whitespace in "
+ toAsciiLowerCase(v.component()) + " component. "
+ "Use \u201C%20\u201D in place of spaces.");
case EMPTY_SCHEME:
throw newDatatypeException("Scheme component is empty.");
case ILLEGAL_PERCENT_ENCODING:
throw newDatatypeException(underbarStringToSentence(v.component())
+ " component contains a percent sign that is not followed by two hexadecimal digits.");
case IP_V4_HAS_FOUR_COMPONENTS:
throw newDatatypeException("Host component is entirely numeric but does not have four components like an IPv4 address.");
case IP_V4_OCTET_RANGE:
throw newDatatypeException("Host component contains a number not in the range 0-255, or a number with a leading zero.");
case IP_V6_OR_FUTURE_ADDRESS_SYNTAX:
throw newDatatypeException("Host component contains an IPv6 (or IPvFuture) syntax violation.");
case NOT_DNS_NAME:
throw newDatatypeException("Host component did not meet the restrictions on DNS names.");
case REQUIRED_COMPONENT_MISSING:
throw newDatatypeException("A component that is required by the scheme is missing.");
case SCHEME_MUST_START_WITH_LETTER:
throw newDatatypeException("Scheme component must start with a letter.");
case UNREGISTERED_NONIETF_SCHEME_TREE:
throw newDatatypeException("Scheme component has a \u201C-\u201D (dash) character, but does not start with \u201Cx-\u201D, and the prefix is not known as the prefix of an alternative tree for URI schemes.");
case CONTROL_CHARACTER:
case ILLEGAL_CHARACTER:
case UNDEFINED_UNICODE_CHARACTER:
case UNICODE_WHITESPACE:
throw newDatatypeException(underbarStringToSentence(v.codeName())
+ " in "
+ toAsciiLowerCase(v.component())
+ " component.");
default:
throw newDatatypeException(v.codeName() + " in "
+ toAsciiLowerCase(v.component()) + " component.");
}
} catch (IOException e) {
throw newDatatypeException(e.getMessage());
} catch (RhinoException e) {
throw newDatatypeException(e.getMessage());
}
if (isAbsolute()) {
if (iri != null && !iri.isAbsolute()) {
throw newDatatypeException("Not an absolute IRI.");
}
}
if (iri != null) {
if ("".equals(iri.toString())) {
throw newDatatypeException("Must be non-empty.");
}
if (data) {
try {
DataUri dataUri = new DataUri(iri);