@Test
public void test() throws GalimatiasParseException {
final URLCanonicalizer canon = new DecodeUnreservedCanonicalizer();
for (final String[] pair :
new String[][] {
new String[] {
"http://%41%5A%61%7A%30%39%2D%2E%5F%[email protected]/", "http://[email protected]/"
},
new String[] {
"http://:%41%5A%61%7A%30%39%2D%2E%5F%[email protected]/", "http://:[email protected]/"
},
new String[] {
"http://example.com/%41%5A%61%7A%30%39%2D%2E%5F%7E", "http://example.com/AZaz09-._~"
},
new String[] {
"http://example.com/?%41%5A%61%7A%30%39%2D%2E%5F%7E", "http://example.com/?AZaz09-._~"
},
new String[] {
"http://example.com/#%41%5A%61%7A%30%39%2D%2E%5F%7E", "http://example.com/#AZaz09-._~"
}
}) {
assertThat(canon.canonicalize(URL.parse(pair[0])).toString())
.isEqualTo(URL.parse(pair[1]).toString());
}
}
@Theory
public void idempotence(
final @TestURL.TestURLs(dataset = TestURL.DATASETS.WHATWG) TestURL testURL)
throws GalimatiasParseException {
assumeNotNull(testURL.parsedURL);
final URLCanonicalizer canon = new DecodeUnreservedCanonicalizer();
final URL roundOne = canon.canonicalize(testURL.parsedURL);
final URL roundTwo = canon.canonicalize(roundOne);
assertThat(roundOne).isEqualTo(roundTwo);
final URL reparse = URL.parse(roundTwo.toString());
assertThat(reparse).isEqualTo(roundTwo);
}
/** @see org.xml.sax.EntityResolver#resolveEntity(java.lang.String, java.lang.String) */
@Override
public InputSource resolveEntity(String publicId, String systemId)
throws SAXException, IOException {
if (requestsLeft > -1) {
if (requestsLeft == 0) {
throw new IOException("Number of permitted HTTP requests exceeded.");
} else {
requestsLeft--;
}
}
HttpGet m = null;
try {
URL url;
try {
url = URL.parse(systemId);
} catch (GalimatiasParseException e) {
IOException ioe = (IOException) new IOException(e.getMessage()).initCause(e);
SAXParseException spe =
new SAXParseException(e.getMessage(), publicId, systemId, -1, -1, ioe);
if (errorHandler != null) {
errorHandler.fatalError(spe);
}
throw spe;
}
String scheme = url.scheme();
if (!("http".equals(scheme) || "https".equals(scheme))) {
String msg = "Unsupported URI scheme: \u201C" + scheme + "\u201D.";
SAXParseException spe =
new SAXParseException(msg, publicId, systemId, -1, -1, new IOException(msg));
if (errorHandler != null) {
errorHandler.fatalError(spe);
}
throw spe;
}
systemId = url.toString();
try {
m = new HttpGet(systemId);
} catch (IllegalArgumentException e) {
SAXParseException spe =
new SAXParseException(
e.getMessage(),
publicId,
systemId,
-1,
-1,
(IOException) new IOException(e.getMessage()).initCause(e));
if (errorHandler != null) {
errorHandler.fatalError(spe);
}
throw spe;
}
m.setHeader("User-Agent", userAgent);
m.setHeader("Accept", buildAccept());
m.setHeader("Accept-Encoding", "gzip");
log4j.info(systemId);
HttpResponse response = client.execute(m);
int statusCode = response.getStatusLine().getStatusCode();
if (statusCode != 200) {
String msg =
"HTTP resource not retrievable. The HTTP status from the remote server was: "
+ statusCode
+ ".";
SAXParseException spe =
new SAXParseException(
msg, publicId, m.getURI().toString(), -1, -1, new IOException(msg));
if (errorHandler != null) {
errorHandler.fatalError(spe);
}
throw spe;
}
HttpEntity entity = response.getEntity();
long len = entity.getContentLength();
if (sizeLimit > -1 && len > sizeLimit) {
SAXParseException spe =
new SAXParseException(
"Resource size exceeds limit.",
publicId,
m.getURI().toString(),
-1,
-1,
new StreamBoundException("Resource size exceeds limit."));
if (errorHandler != null) {
errorHandler.fatalError(spe);
}
throw spe;
}
TypedInputSource is;
org.apache.http.Header ct = response.getFirstHeader("Content-Type");
String contentType = null;
final String baseUri = m.getURI().toString();
if (ct != null) {
contentType = ct.getValue();
}
is = contentTypeParser.buildTypedInputSource(baseUri, publicId, contentType);
Header cl = response.getFirstHeader("Content-Language");
if (cl != null) {
is.setLanguage(cl.getValue().trim());
}
Header xuac = response.getFirstHeader("X-UA-Compatible");
if (xuac != null) {
String val = xuac.getValue().trim();
if (!"ie=edge".equalsIgnoreCase(val)) {
SAXParseException spe =
new SAXParseException(
"X-UA-Compatible HTTP header must have the value \u201CIE=edge\u201D,"
+ " was \u201C"
+ val
+ "\u201D.",
publicId,
systemId,
-1,
-1);
errorHandler.error(spe);
}
}
Header csp = response.getFirstHeader("Content-Security-Policy");
if (csp != null) {
try {
ContentSecurityPolicy.THE_INSTANCE.checkValid(csp.getValue().trim());
} catch (DatatypeException e) {
SAXParseException spe =
new SAXParseException(
"Content-Security-Policy HTTP header: " + e.getMessage(),
publicId,
systemId,
-1,
-1);
Html5DatatypeException ex5 = (Html5DatatypeException) e;
if (ex5.isWarning()) {
errorHandler.warning(spe);
} else {
errorHandler.error(spe);
}
}
}
final HttpGet meth = m;
InputStream stream = entity.getContent();
if (sizeLimit > -1) {
stream = new BoundedInputStream(stream, sizeLimit, baseUri);
}
Header ce = response.getFirstHeader("Content-Encoding");
if (ce != null) {
String val = ce.getValue().trim();
if ("gzip".equalsIgnoreCase(val) || "x-gzip".equalsIgnoreCase(val)) {
stream = new GZIPInputStream(stream);
if (sizeLimit > -1) {
stream = new BoundedInputStream(stream, sizeLimit, baseUri);
}
}
}
is.setByteStream(
new ObservableInputStream(
stream,
new StreamObserver() {
private final Logger log4j =
Logger.getLogger("nu.validator.xml.PrudentEntityResolver.StreamObserver");
private boolean released = false;
@Override
public void closeCalled() {
log4j.debug("closeCalled");
if (!released) {
log4j.debug("closeCalled, not yet released");
released = true;
try {
meth.releaseConnection();
} catch (Exception e) {
log4j.debug("closeCalled, releaseConnection", e);
}
}
}
@Override
public void exceptionOccurred(Exception ex) throws IOException {
if (!released) {
released = true;
try {
meth.abort();
} catch (Exception e) {
log4j.debug("exceptionOccurred, abort", e);
} finally {
try {
meth.releaseConnection();
} catch (Exception e) {
log4j.debug("exceptionOccurred, releaseConnection", e);
}
}
}
if (ex instanceof SystemIdIOException) {
throw (SystemIdIOException) ex;
} else if (ex instanceof IOException) {
IOException ioe = (IOException) ex;
throw new SystemIdIOException(baseUri, ioe.getMessage(), ioe);
} else if (ex instanceof RuntimeException) {
throw (RuntimeException) ex;
} else {
throw new RuntimeException("API contract violation. Wrong exception type.", ex);
}
}
@Override
public void finalizerCalled() {
if (!released) {
released = true;
try {
meth.abort();
} catch (Exception e) {
log4j.debug("finalizerCalled, abort", e);
} finally {
try {
meth.releaseConnection();
} catch (Exception e) {
log4j.debug("finalizerCalled, releaseConnection", e);
}
}
}
}
}));
return is;
} catch (IOException | RuntimeException | SAXException e) {
if (m != null) {
try {
m.abort();
} catch (Exception ex) {
log4j.debug("abort", ex);
} finally {
try {
m.releaseConnection();
} catch (Exception ex) {
log4j.debug("releaseConnection", ex);
}
}
}
throw e;
}
}
