public static String saxElementToDebugString(String uri, String qName, Attributes attributes) {
// Open start tag
final StringBuilder sb = new StringBuilder("<");
sb.append(qName);
final Set<String> declaredPrefixes = new HashSet<String>();
mapPrefixIfNeeded(declaredPrefixes, uri, qName, sb);
// Attributes if any
for (int i = 0; i < attributes.getLength(); i++) {
mapPrefixIfNeeded(declaredPrefixes, attributes.getURI(i), attributes.getQName(i), sb);
sb.append(' ');
sb.append(attributes.getQName(i));
sb.append("=\"");
sb.append(attributes.getValue(i));
sb.append('\"');
}
// Close start tag
sb.append('>');
// Content
sb.append("[...]");
// Close element with end tag
sb.append("</");
sb.append(qName);
sb.append('>');
return sb.toString();
}
/**
* @param attributes source attributes
* @return new AttributesImpl containing all attributes that were in src attributes and that were
* in the default name space.
*/
public static AttributesImpl getAttributesFromDefaultNamespace(final Attributes attributes) {
final AttributesImpl ret = new AttributesImpl();
final int size = attributes.getLength();
for (int i = 0; i < size; i++) {
final String ns = attributes.getURI(i);
if (!"".equals(ns)) continue;
final String lnam = attributes.getLocalName(i);
final String qnam = attributes.getQName(i);
final String typ = attributes.getType(i);
final String val = attributes.getValue(i);
ret.addAttribute(ns, lnam, qnam, typ, val);
}
return ret;
}
protected void collectExtensionAttributes(Attributes attributes) {
for (int i = 0; i < attributes.getLength(); i++) {
String key = attributes.getURI(i);
if (key.length() == 0
|| key.startsWith("http://www.osgi.org/xmlns/metatype/v")) // $NON-NLS-1$
continue;
Map<String, String> value = extensionAttributes.get(key);
if (value == null) {
value = new HashMap<String, String>();
extensionAttributes.put(key, value);
}
value.put(
getName(attributes.getLocalName(i), attributes.getQName(i)), attributes.getValue(i));
}
}
private Map attributeMap(String tagName, Attributes atts) throws ParserException {
if (null == tagName || null == atts) return null;
Map mapping = null;
try {
mapping = (Map) attributeMaps.get(tagName);
} catch (Exception e) {
throw new ParserException(
"Typecast error, unknown element found in attribute list mappings! " + e.getMessage());
}
if (null == mapping) return null;
Map resultMapping = new HashMap();
for (int i = 0; i < atts.getLength(); i++) {
String xmlName = atts.getQName(i);
String value = atts.getValue(i);
TagMap.AttributeMapping aMap = null;
try {
aMap = (TagMap.AttributeMapping) mapping.get(xmlName);
} catch (Exception e) {
throw new ParserException(
"Typecast error, unknown element found in property mapping! " + e.getMessage());
}
if (null == aMap)
throw new ParserException(
"No attribute mapping specified for attribute: " + xmlName + " in tag: " + tagName);
String propertyName = aMap.getPropertyName();
try {
resultMapping.put(propertyName, aMap.convertValue(value));
} catch (Exception e) {
throw new ParserException(
"Can not convert given value: \""
+ value
+ "\" to specified type: "
+ aMap.getType()
+ " for attribute: "
+ xmlName
+ " in tag: "
+ tagName
+ "! "
+ e.getMessage());
}
}
checkForRequiredAttributes(tagName, resultMapping, mapping);
addDefaultValues(resultMapping, mapping);
return resultMapping;
}
public static AttributesImpl removeAttribute(
Attributes attributes, String uri, String localname) {
final AttributesImpl newAttributes = new AttributesImpl();
for (int i = 0; i < attributes.getLength(); i++) {
final String attributeURI = attributes.getURI(i);
final String attributeValue = attributes.getValue(i);
final String attributeType = attributes.getType(i);
final String attributeQName = attributes.getQName(i);
final String attributeLocalname = attributes.getLocalName(i);
if (!uri.equals(attributeURI) || !localname.equals(attributeLocalname)) {
// Not a matched attribute
newAttributes.addAttribute(
attributeURI, attributeLocalname, attributeQName, attributeType, attributeValue);
}
}
return newAttributes;
}
public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
throws SAXException {
try {
// we output the remaining text
if (!counting) {
writer.write(getText());
accumulator.setLength(0);
}
if (!counting) {
writer.write("<" + qName);
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
writer.write(" " + name + "=\"" + value + "\"");
}
}
writer.write(">");
}
if (qName.equals("description")) {
offset = 0;
counting = true;
} else if (qName.equals("patent-document")) {
counting = false;
}
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
}
public static AttributesImpl addOrReplaceAttribute(
Attributes attributes, String uri, String prefix, String localname, String value) {
final AttributesImpl newAttributes = new AttributesImpl();
boolean replaced = false;
for (int i = 0; i < attributes.getLength(); i++) {
final String attributeURI = attributes.getURI(i);
final String attributeValue = attributes.getValue(i);
final String attributeType = attributes.getType(i);
final String attributeQName = attributes.getQName(i);
final String attributeLocalname = attributes.getLocalName(i);
if (uri.equals(attributeURI) && localname.equals(attributeLocalname)) {
// Found existing attribute
replaced = true;
newAttributes.addAttribute(
uri,
localname,
XMLUtils.buildQName(prefix, localname),
ContentHandlerHelper.CDATA,
value);
} else {
// Not a matched attribute
newAttributes.addAttribute(
attributeURI, attributeLocalname, attributeQName, attributeType, attributeValue);
}
}
if (!replaced) {
// Attribute did not exist already so add it
newAttributes.addAttribute(
uri,
localname,
XMLUtils.buildQName(prefix, localname),
ContentHandlerHelper.CDATA,
value);
}
return newAttributes;
}
/**
* This reports the occurrence of an actual element. It will include the element's attributes,
* with the exception of XML vocabulary specific attributes, such as <code>
* xmlns:[namespace prefix]</code> and <code>xsi:schemaLocation</code>.
*
* @param namespaceURI <code>String</code> namespace URI this element is associated with, or an
* empty <code>String</code>
* @param localName <code>String</code> name of element (with no namespace prefix, if one is
* present)
* @param qName <code>String</code> XML 1.0 version of element name: [namespace
* prefix]:[localName]
* @param atts <code>Attributes</code> list for this element
* @throws SAXException when things go wrong
*/
public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
throws SAXException {
if (suppress) return;
Element element = null;
if ((namespaceURI != null) && (!namespaceURI.equals(""))) {
String prefix = "";
// Determine any prefix on the Element
if (!qName.equals(localName)) {
int split = qName.indexOf(":");
prefix = qName.substring(0, split);
}
Namespace elementNamespace = Namespace.getNamespace(prefix, namespaceURI);
element = factory.element(localName, elementNamespace);
} else {
element = factory.element(localName);
}
// Take leftover declared namespaces and add them to this element's
// map of namespaces
if (declaredNamespaces.size() > 0) {
transferNamespaces(element);
}
// Handle attributes
for (int i = 0, len = atts.getLength(); i < len; i++) {
Attribute attribute = null;
String attLocalName = atts.getLocalName(i);
String attQName = atts.getQName(i);
int attType = getAttributeType(atts.getType(i));
// Bypass any xmlns attributes which might appear, as we got
// them already in startPrefixMapping().
// This is sometimes necessary when SAXHandler is used with
// another source than SAXBuilder, as with JDOMResult.
if (attQName.startsWith("xmlns:") || attQName.equals("xmlns")) {
continue;
}
// First clause per http://markmail.org/message/2p245ggcjst27xe6
// patch from Mattias Jiderhamn
if ("".equals(attLocalName) && attQName.indexOf(":") == -1) {
attribute = factory.attribute(attQName, atts.getValue(i), attType);
} else if (!attQName.equals(attLocalName)) {
String attPrefix = attQName.substring(0, attQName.indexOf(":"));
Namespace attNs = Namespace.getNamespace(attPrefix, atts.getURI(i));
attribute = factory.attribute(attLocalName, atts.getValue(i), attType, attNs);
} else {
attribute = factory.attribute(attLocalName, atts.getValue(i), attType);
}
factory.setAttribute(element, attribute);
}
flushCharacters();
if (atRoot) {
document.setRootElement(element); // XXX should we use a factory call?
atRoot = false;
} else {
factory.addContent(getCurrentElement(), element);
}
currentElement = element;
}
public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
throws SAXException {
if (qName.equals("patent-document") || qName.equals("fulltext-document")) {
nbNPLRef = 0;
nbPatentRef = 0;
nbAllRef = 0;
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if (name != null) {
if (name.equals("lang")) {
// Global_Language_Code = value.toLowerCase();
}
if (name.equals("doc-number")) {
PatentNumber = "EP" + value;
}
if (name.equals("kind")) {
CodeType = value;
}
if (name.equals("date")) {
PublicDate = value;
}
}
}
CitedPatentNumber = new ArrayList<String>();
accumulatedText = new StringBuffer();
allContent = new StringBuffer();
accumulator.setLength(0);
} else if (qName.equals("description")) {
accumulator.setLength(0);
} else if (qName.equals("ref") || qName.equals("bibl")) {
int length = atts.getLength();
nbAllRef++;
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if (name != null) {
if (name.equals("type") || name.equals("typ")) {
if (value.equals("npl") || value.equals("book") || value.equals("journal")) {
String content = getText();
// we output what has been read so far in the description
// we tokenize the text
// ArrayList<String> tokens =
// StringTokenizer st = new StringTokenizer(content, delimiters, true);
List<String> tokenizations = new ArrayList<String>();
try {
// TBD: pass a language object to the tokenize method call
tokenizations = analyzer.tokenize(content);
} catch (Exception e) {
LOGGER.debug("Tokenization for XML patent document has failed.");
}
// int nbTokens = st.countTokens();
int nbTokens = tokenizations.size();
int j = 0;
// while (st.hasMoreTokens()) {
for (String token : tokenizations) {
// String token = st.nextToken().trim();
if ((token.trim().length() == 0)
|| (token.equals(" "))
|| (token.equals("\t"))
|| (token.equals("\n"))
|| (token.equals("\r"))) {
continue;
}
if ((j > (nbTokens - N) && (N != -1)) || (refFound && (j < N) && (N != -1))) {
try {
accumulatedText.append(token + "\t" + "<other>\n");
allContent.append(token + " ");
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
} else {
try {
accumulatedText.append(token + "\t" + "<ignore>\n");
allContent.append(token + " ");
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
}
j++;
}
accumulator.setLength(0);
npl = true;
ref = true;
} else if (value.equals("patent") || value.equals("pl")) {
String content = getText();
// we output what has been read so far in the description
// we tokenize the text
// ArrayList<String> tokens =
// TextUtilities.segment(content,"[("+TextUtilities.punctuations);
// StringTokenizer st = new StringTokenizer(content, delimiters, true);
List<String> tokenizations = new ArrayList<String>();
try {
// TBD: pass a language object to the tokenize method call
tokenizations = analyzer.tokenize(content);
} catch (Exception e) {
LOGGER.debug("Tokenization for XML patent document has failed.");
}
// int nbTokens = st.countTokens();
int nbTokens = tokenizations.size();
int j = 0;
for (String token : tokenizations) {
// while (st.hasMoreTokens()) {
// String token = st.nextToken().trim();
if ((token.trim().length() == 0)
|| (token.equals(" "))
|| (token.equals("\t"))
|| (token.equals("\n"))
|| (token.equals("\r"))) {
continue;
}
if ((j > (nbTokens - N)) | (refFound & (j < N))) {
try {
accumulatedText.append(token + "\t" + "<other>\n");
allContent.append(token + " ");
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
} else {
try {
accumulatedText.append(token + "\t" + "<ignore>\n");
allContent.append(token + " ");
} catch (Exception e) {
// e.printStackTrace();
throw new GrobidException("An exception occured while running Grobid.", e);
}
}
j++;
}
accumulator.setLength(0);
npl = false;
ref = true;
} else {
System.out.println("Warning: unknown attribute value for ref or bibl: " + value);
ref = false;
npl = false;
}
}
}
}
accumulatorRef.setLength(0);
} else if (qName.equals("claim")) {
accumulator.setLength(0);
} else if (qName.equals("invention-title")) {
accumulator.setLength(0);
} else if (qName.equals("patcit")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if (name != null) {
if (name.equals("ucid")) {
cited_number = value;
// we normally need to normalize a little bit this patent nummer
}
}
}
}
}
/** Receive notification of the start of an element. */
@Override
public void startElement(String uri, String l, String q, Attributes a) {
/*
* 1. Load a class that matches the element name.
* 2. If no class found, assume the element maps to a String.
* 3. Otherwise, construct a new object of the class with element attributes.
*/
_logger.fine(
S.fine(_logger)
? "Consider element " + l + "\n uri " + uri + "\n q " + q
: null);
ElementInfo info = new ElementInfo();
// Record java packages defined on this element as xmlns
for (int i = 0; i < a.getLength(); ++i) {
_logger.fine(
S.fine(_logger)
? " attr "
+ a.getQName(i)
+ "="
+ a.getValue(i)
+ "\n "
+ a.getQName(i)
+ ":"
+ a.getURI(i)
: null);
if (a.getQName(i).startsWith("xmlns:") && a.getValue(i).startsWith("java://")) {
info.pkgs.put(a.getQName(i).substring(6), a.getValue(i).substring(7));
}
}
// Resolve the package name of this element, which could be empty (default package)
int colon = q.indexOf(':');
if (colon > 0) {
String xmlns = q.substring(0, colon);
// is it defined right here?
info.jpkg = info.pkgs.get(xmlns);
// find a matching namespace from ancesters
if (info.jpkg == null && !_stack.isEmpty()) {
for (int i = _stack.size() - 1; i >= 0; --i) {
info.jpkg = _stack.get(i).pkgs.get(xmlns);
if (info.jpkg != null) {
break;
}
}
}
} else if (isPrimitiveType(q)) {
info.jpkg = "java.lang";
} else if (!_stack.isEmpty()) {
info.jpkg = _stack.get(_stack.size() - 1).jpkg;
} else {
info.jpkg = _jpkg;
}
_logger.fine("to create element with package = " + info.jpkg);
try {
info.name =
(info.jpkg != null) ? info.jpkg + '.' + Strings.toCamelCase(l) : Strings.toCamelCase(l);
try {
if (info.name.endsWith("...")) {
// Array construction
info.type = Class.forName(info.name.substring(0, info.name.length() - 3));
info.data = new ArrayList<Object>();
} else {
// Non-array construction
int size = a.getLength();
TypedValueGroup arguments = new TypedValueGroup();
for (int i = 0; i < size; ++i) {
if (!a.getQName(i).startsWith("xmlns:") && !a.getQName(i).equals("xmlns")) {
arguments.add(guessUntypedValue(a.getQName(i), a.getValue(i)));
}
}
arguments.complete();
_logger.fine(S.fine(_logger) ? "arguments=" + arguments : null);
if (arguments.size() > 0) {
if (arguments.size() == 1 && "java.lang".equals(info.jpkg)) {
info.inst.put(
"@as",
Strings.toCamelCase(
arguments.get(0).name, '-', false)); // respect original spelling
info.data = arguments.get(0).get(0).data;
info.type = arguments.get(0).get(0).type;
} else {
Exception last = null;
Object[] args = new Object[arguments.size()];
while (arguments.load(args, 0)) {
try {
_logger.fine(
S.fine(_logger)
? "to create " + info.name + " with args: " + args.length + args(args)
: null);
info.data = _factory.create(info.name, args);
info.type = info.data.getClass();
break;
} catch (InvocationTargetException x) {
throw x;
} catch (Exception x) {
last = x;
_logger.fine(
"failure in creating " + info.name + ": probing for other constructors");
}
}
if (info.data == null) {
throw last;
}
}
} else {
_logger.fine("Create " + info.name + " with the default constructor");
info.data = _factory.create(info.name);
info.type = info.data.getClass();
}
}
} catch (ClassNotFoundException x) {
// no class by the element name is found, assumed String
if (!_lenient) {
throw new BeanAssemblyException("No class associated with element " + q);
} else {
_logger.log(Level.WARNING, "can't find class " + info.name, x);
}
}
_stack.add(info);
// _logger.fine(">>ElementInfo: " + info.type.getName() + " in " + info);
// all other exceptions indicate mismatches between the beans and the XML schema
} catch (Exception x) {
if (!_lenient) {
throw new BeanAssemblyException("Failed to assemble bean from element " + q, x);
} else {
_logger.log(Level.SEVERE, "can't create object for this element", x);
}
}
}
public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
throws SAXException {
if (qName.equals("PAGE")) {
int length = atts.getLength();
currentPage++;
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {;
} else if (name.equals("number")) {
} else if (name.equals("width")) {
} else if (name.equals("height")) {
}
}
}
/*
* if (block != null) { blabla.append("\n");
* tokenizations.add("\n"); block.setText(blabla.toString());
* block.setNbTokens(nbTokens); doc.addBlock(block); } Block block0
* = new Block(); block0.setText("@PAGE\n"); block0.setNbTokens(0);
* doc.addBlock(block0);
*/
/*
* block = new Block(); blabla = new StringBuffer(); nbTokens = 0;
* //blabla.append("\n@block\n"); tokenizations.add("\n");
*/
} else if (qName.equals("BLOCK")) {
block = new Block();
blabla = new StringBuffer();
nbTokens = 0;
block.setPage(currentPage);
// blabla.append("\n@block\n");
} else if (qName.equals("IMAGE")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("href")) {
// if (images == null)
// images = new ArrayList<String>();
images.add(value);
} else if (name.equals("x")) {
double x = Double.parseDouble(value);
if (x != currentX) {
currentX = x;
}
} else if (name.equals("y")) {
double y = Double.parseDouble(value);
if (y != currentY) {
currentY = y;
}
} else if (name.equals("width")) {
double width = Double.parseDouble(value);
if (width != currentWidth) {
currentWidth = width;
}
} else if (name.equals("height")) {
double height = Double.parseDouble(value);
if (height != currentHeight) {
currentHeight = height;
}
}
}
}
} else if (qName.equals("TEXT")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {
} else if (name.equals("x")) {
} else if (name.equals("y")) {
} else if (name.equals("width")) {
} else if (name.equals("height")) {
}
}
}
} else if (qName.equals("TOKEN")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("id")) {;
} else if (name.equals("font-name")) {
if (!value.equals(currentFont)) {
currentFont = value;
blabla.append(" ");
}
} else if (name.equals("font-size")) {
double fontSize = Double.parseDouble(value);
if (fontSize != currentFontSize) {
currentFontSize = fontSize;
blabla.append(" ");
}
} else if (name.equals("bold")) {
if (value.equals("yes")) {
currentBold = true;
} else {
currentBold = false;
}
} else if (name.equals("italic")) {
if (value.equals("yes")) {
currentItalic = true;
} else {
currentItalic = false;
}
} else if (name.equals("font-color")) {
if (!value.equals(colorFont)) {
colorFont = value;
}
} else if (name.equals("rotation")) {
if (value.equals("0")) currentRotation = false;
else currentRotation = true;
} else if (name.equals("x")) {
double x = Double.parseDouble(value);
if (x != currentX) {
currentX = x;
}
} else if (name.equals("y")) {
double y = Double.parseDouble(value);
if (y != currentY) {
currentY = y;
}
} else if (name.equals("base")) {
double base = Double.parseDouble(value);
} else if (name.equals("width")) {
double width = Double.parseDouble(value);
if (width != currentWidth) {
currentWidth = width;
}
} else if (name.equals("height")) {
double height = Double.parseDouble(value);
if (height != currentHeight) {
currentHeight = height;
}
}
}
}
} else if (qName.equals("xi:include")) {
int length = atts.getLength();
// Process each attribute
for (int i = 0; i < length; i++) {
// Get names and values for each attribute
String name = atts.getQName(i);
String value = atts.getValue(i);
if ((name != null) && (value != null)) {
if (name.equals("href")) {
// if (images == null)
// images = new ArrayList<String>();
images.add(value);
}
}
}
}
// accumulator.setLength(0);
}