private void doProcessIncludes(TagNode html, int depth, FormFlow formFlow)
throws IOException, FormParserException {
if (depth < processIncludesMaxDepth) {
@SuppressWarnings("unchecked")
List<TagNode> includeNodes = html.getElementListByName(Constants.INCLUDE_ELEMENT, true);
for (TagNode includeNode : includeNodes) {
String srcAttribute = includeNode.getAttributeByName("src");
srcAttribute = formFlow.resolveResourcePathIfRelative(srcAttribute);
InputStream resourceAsStream = resourceLoader.getFormResourceAsStream(srcAttribute);
if (resourceAsStream != null) {
TagNode includeHtml = htmlCleaner.clean(resourceAsStream);
TagNode body = includeHtml.findElementByName("body", false);
doProcessIncludes(body, depth + 1, formFlow);
@SuppressWarnings("unchecked")
List<HtmlNode> bodyChildren = body.getChildren();
Collections.reverse(bodyChildren);
TagNode includeParent = includeNode.getParent();
for (HtmlNode bodyChild : bodyChildren) {
includeParent.insertChildAfter(includeNode, bodyChild);
}
includeParent.removeChild(includeNode);
} else {
throw new FormParserException("Include file not found. Path:'" + srcAttribute + "'");
}
}
} else {
throw new FormParserException(
"Exceeded maximum nested "
+ Constants.INCLUDE_ELEMENT
+ " depth of "
+ processIncludesMaxDepth);
}
}
private boolean recuperaDadosPessoais(String html) {
HtmlCleaner cleaner = new HtmlCleaner();
TagNode root = cleaner.clean(html);
String table_name = context.getString(R.string.janusmob_table_dados_aluno_id);
TagNode table = null;
for (TagNode node : root.getElementsByName("table", true)) {
if (table_name.equals(node.getAttributeByName("id"))) {
table = node;
break;
}
}
if (table == null) {
return false;
}
String attribute_class_name = context.getString(R.string.janusmob_attribute_class_name);
String campo_nusp = context.getString(R.string.janusmob_campo_nusp);
String campo_nome = context.getString(R.string.janusmob_campo_nome);
String campo_email = context.getString(R.string.janusmob_campo_email);
String campo_cpf = context.getString(R.string.janusmob_campo_cpf);
Usuario usuario = new Usuario();
for (TagNode node : table.getElementsByName("span", true)) {
if (attribute_class_name.equals(node.getAttributeByName("class"))) {
TagNode parent = node.getParent();
parent.removeChild(node);
if (node.getText().toString().trim().equals(campo_nusp)) {
usuario.setNusp(parent.getText().toString().trim());
} else if (node.getText().toString().trim().equals(campo_nome)) {
usuario.setNome(parent.getText().toString().trim());
} else if (node.getText().toString().trim().equals(campo_email)) {
usuario.setEmail(parent.getText().toString().trim());
} else if (node.getText().toString().trim().equals(campo_cpf)) {
usuario.setCpf(parent.getText().toString().trim());
}
}
}
if (lembrarSenha) {
usuario.setSenha(senha);
} else {
usuario.setSenha(null);
}
dbAdapter.setUsuario(usuario);
return true;
}
@SuppressWarnings("unchecked")
protected void handleFile(File file, int depth, Collection results) {
File f = new File(FilenameUtils.normalize(file.getAbsolutePath()));
logger.debug(f.getAbsoluteFile());
try {
HtmlCleaner cleaner = new HtmlCleaner();
cleaner.setTransformations(ct);
CleanerProperties props = cleaner.getProperties();
props.setAdvancedXmlEscape(false);
// props.setTranslateSpecialEntities(false);
// props.setRecognizeUnicodeChars(false);
TagNode node = cleaner.clean(f);
TagNode tnBody = node.getAllElements(false)[1];
List l = tnBody.getChildren();
if (l != null
&& l.size() > 0) { // This is a hack to remove the <?xml in the beginning of body
tnBody.removeChild(l.get(0));
}
Document myJDom = new JDomSerializer(props, true).createJDom(node);
// Format format = Format.getRawFormat();
Format format = new OutputFormat();
format.setEncoding("iso-8859-1");
XMLWriter outputter = new XMLWriter(format);
OutputStream os = new FileOutputStream(f);
// outputter.output(myJDom,os);
output.setOutputStream(os);
output.write(myJDom);
// sbResult.append(outputter.outputString(myJDom));
results.add(f.getAbsoluteFile());
} catch (IOException e) {
logger.error("", e);
}
}
@SuppressWarnings("unchecked")
protected void handleFile(File file, int depth, Collection results) {
File f = new File(FilenameUtils.normalize(file.getAbsolutePath()));
logger.debug(f.getAbsoluteFile());
try {
HtmlCleaner cleaner = new HtmlCleaner();
cleaner.setTransformations(ct);
CleanerProperties props = cleaner.getProperties();
// props.setAdvancedXmlEscape(false);
props.setUseEmptyElementTags(false);
// props.setTranslateSpecialEntities(false);
// props.setRecognizeUnicodeChars(false);
TagNode node = cleaner.clean(f);
TagNode tnBody = node.getAllElements(false)[1];
List l = tnBody.getChildren();
if (l != null
&& l.size() > 0) { // This is a hack to remove the <?xml in the beginning of body
tnBody.removeChild(l.get(0));
}
for (int i = 1; i <= anzElements; i++) {
String tag = config.getString("substitute[" + i + "]/@tag");
String att = config.getString("substitute[" + i + "]/@att");
String from = config.getString("substitute[" + i + "]/from");
String to = config.getString("substitute[" + i + "]/to");
to = subSpecial(to);
TagNode[] imgs = node.getElementsByName(tag, true);
for (TagNode tn : imgs) {
String srcAtt = tn.getAttributeByName(att);
int index = srcAtt.indexOf(from);
if (index >= 0) {
tn.addAttribute(att, to);
}
}
}
BrowserCompactXmlSerializer serializer = new BrowserCompactXmlSerializer(props);
// PrettyXmlSerializer serializer = new PrettyXmlSerializer(props);
String s = serializer.getXmlAsString(node, "ISO-8859-1");
Writer fw = null;
try {
fw = new FileWriter(f);
fw.write(s);
} catch (IOException e) {
logger.error("", e);
} finally {
if (fw != null)
try {
fw.close();
} catch (IOException e) {
}
}
results.add(f.getAbsoluteFile());
} catch (IOException e) {
logger.error("", e);
}
}
/**
* Remove this node from the tree.
*
* @return True if element is removed (if it is not root node).
*/
public boolean removeFromTree() {
return parent != null ? parent.removeChild(this) : false;
}
public static Boolean updateArtists(StaticDataStore db) {
ArrayList<ArrayList<String>> artists = new ArrayList<ArrayList<String>>();
int numArtists;
HtmlCleaner pageParser = new HtmlCleaner();
CleanerProperties props = pageParser.getProperties();
props.setAllowHtmlInsideAttributes(true);
props.setAllowMultiWordAttributes(true);
props.setRecognizeUnicodeChars(true);
props.setOmitComments(true);
try {
String url =
"http://www.archive.org/browse.php?field=/metadata/bandWithMP3s&collection=etree";
HttpParams params = new BasicHttpParams();
int timeout = (int) (15 * DateUtils.SECOND_IN_MILLIS);
HttpConnectionParams.setConnectionTimeout(params, timeout);
HttpConnectionParams.setSoTimeout(params, timeout);
HttpClient client = new DefaultHttpClient(params);
HttpGet request = new HttpGet(url);
HttpResponse response = client.execute(request);
StatusLine status = response.getStatusLine();
if (status.getStatusCode() == HttpStatus.SC_OK) {
ResponseHandler<String> responseHandler = new BasicResponseHandler();
TagNode node = pageParser.clean(responseHandler.handleResponse(response));
client.getConnectionManager().shutdown();
// XPATH to get the nodes that we Want.
Object[] artistsNodes = node.evaluateXPath("//tr[@valign='top']//li");
numArtists = artistsNodes.length;
for (int i = 0; i < numArtists; i++) {
// Cast the artistNode as a TagNode.
TagNode artist = ((TagNode) artistsNodes[i]);
// Grab the first child node, which is the link to the artist's page.
// The inner HTML of this node will be the title.
TagNode artistTitleSubNode = artist.getChildTags()[0];
// Remove the child node, so that the inner HTML of the artistNode
// only contains the number of shows that the artist has.
artist.removeChild(artistTitleSubNode);
String artistTitle = pageParser.getInnerHtml(artistTitleSubNode);
if (artistTitle != null) {
ArrayList<String> artistPair = new ArrayList<String>();
artistPair.add(
artistTitle
.replace("'", "'")
.replace(">", ">")
.replace("<", "<")
.replace(""", "\"")
.replace("&", "&"));
artistPair.add(pageParser.getInnerHtml(artist).trim());
/*
* VibeVault.db.addArtist(artistTitle, pageParser
* .getInnerHtml(artist).trim());
*/
artists.add(artistPair);
}
}
if (artists.size() > 0) {
db.insertArtistBulk(artists);
String s = DateFormat.format("yyyy-MM-dd", new GregorianCalendar().getTime()).toString();
db.updatePref("artistUpdate", s);
} else {
}
} else {
client.getConnectionManager().shutdown();
}
} catch (Exception e) {
e.printStackTrace();
}
return true;
}
