/**
* For given content the list with shrink position information is searched and on the
* corresponding positions the correct repositioning information is calculated and generated.
*/
public void addRepositioningInfo(String content, int pos, int extractedPos) {
int contentLength = content.length();
// wrong way (without correction and analysing)
// reposInfo.addPositionInfo(pos, contentLength, extractedPos, contentLength);
RepositioningInfo.PositionInfo pi = null;
long startPos = pos;
long correction = 0;
long substituteStart;
long remainingLen;
long offsetInExtracted;
for (int i = 0; i < ampCodingInfo.size(); ++i) {
pi = ampCodingInfo.get(i);
substituteStart = pi.getOriginalPosition();
if (substituteStart >= startPos) {
if (substituteStart > pos + contentLength + correction) {
break; // outside the current text
} // if
// should create two repositioning information records
remainingLen = substituteStart - (startPos + correction);
offsetInExtracted = startPos - pos;
if (remainingLen > 0) {
reposInfo.addPositionInfo(
startPos + correction, remainingLen, extractedPos + offsetInExtracted, remainingLen);
} // if
// record for shrank text
reposInfo.addPositionInfo(
substituteStart,
pi.getOriginalLength(),
extractedPos + offsetInExtracted + remainingLen,
pi.getCurrentLength());
startPos = startPos + remainingLen + pi.getCurrentLength();
correction += pi.getOriginalLength() - pi.getCurrentLength();
} // if
} // for
// there is some text remaining for repositioning
offsetInExtracted = startPos - pos;
remainingLen = contentLength - offsetInExtracted;
if (remainingLen > 0) {
reposInfo.addPositionInfo(
startPos + correction, remainingLen, extractedPos + offsetInExtracted, remainingLen);
} // if
} // addRepositioningInfo
/** This method is called when all characters between specific tags have been read completely */
public void charactersAction(char[] text, int start, int length) throws SAXException {
// correction of real offset. Didn't affect on other data.
super.characters(text, start, length);
// create a string object based on the reported text
String content = new String(text, start, length);
StringBuffer contentBuffer = new StringBuffer("");
int tmpDocContentSize = tmpDocContent.length();
boolean incrementStartIndex = false;
boolean addExtraSpace = true;
if (Gate.getUserConfig().get(GateConstants.DOCUMENT_ADD_SPACE_ON_UNPACK_FEATURE_NAME) != null) {
addExtraSpace =
Gate.getUserConfig()
.getBoolean(GateConstants.DOCUMENT_ADD_SPACE_ON_UNPACK_FEATURE_NAME)
.booleanValue();
}
// If the first char of the text just read "text[0]" is NOT whitespace AND
// the last char of the tmpDocContent[SIZE-1] is NOT whitespace then
// concatenation "tmpDocContent + content" will result into a new different
// word... and we want to avoid that, because the tokenizer, gazetter and
// Jape work on the raw text and concatenating tokens might be not good.
if (tmpDocContentSize != 0
&& content.length() != 0
&& !Character.isWhitespace(content.charAt(0))
&& !Character.isWhitespace(tmpDocContent.charAt(tmpDocContentSize - 1))) {
// If we are here it means that a concatenation between the last
// token in the tmpDocContent and the content(which doesn't start
// with a white space) will be performed. In order to prevent this,
// we will add a " " space char in order to assure that the 2 tokens
// stay apart. Howerver we will except from this rule the most known
// internal entities like &, <, >, etc
if (( // Testing the length against 1 makes it more likely that
// an internal entity was called. characters() gets called for
// each entity separately.
(content.length() == 1)
&& (content.charAt(0) == '&'
|| content.charAt(0) == '<'
|| content.charAt(0) == '>'
|| content.charAt(0) == '"'
|| content.charAt(0) == '\''))
|| (tmpDocContent.charAt(tmpDocContentSize - 1) == '&'
|| tmpDocContent.charAt(tmpDocContentSize - 1) == '<'
|| tmpDocContent.charAt(tmpDocContentSize - 1) == '>'
|| tmpDocContent.charAt(tmpDocContentSize - 1) == '"'
|| tmpDocContent.charAt(tmpDocContentSize - 1)
== '\'')) { // do nothing. The content will be appended
} else if (!addExtraSpace) {
} else {
// In all other cases append " "
contentBuffer.append(" ");
incrementStartIndex = true;
} // End if
} // End if
// put the repositioning information
if (reposInfo != null) {
if (!(start == 0 && length == 1 && text.length <= 2)) {
// normal piece of text
reposInfo.addPositionInfo(
getRealOffset(),
content.length(),
tmpDocContent.length() + contentBuffer.length(),
content.length());
if (DEBUG) {
Out.println("Info: " + getRealOffset() + ", " + content.length());
Out.println("Start: " + start + " len" + length);
} // DEBUG
} else {
// unicode char or &xxx; coding
// Reported from the parser offset is 0
// The real offset should be found in the ampCodingInfo structure.
long lastPosition = 0;
RepositioningInfo.PositionInfo pi;
if (reposInfo.size() > 0) {
pi = reposInfo.get(reposInfo.size() - 1);
lastPosition = pi.getOriginalPosition();
} // if
for (int i = 0; i < ampCodingInfo.size(); ++i) {
pi = ampCodingInfo.get(i);
if (pi.getOriginalPosition() > lastPosition) {
// found
reposInfo.addPositionInfo(
pi.getOriginalPosition(),
pi.getOriginalLength(),
tmpDocContent.length() + contentBuffer.length(),
content.length());
break;
} // if
} // for
} // if
} // if
// update the document content
contentBuffer.append(content);
// calculate the End index for all the elements of the stack
// the expression is : End index = Current doc length + text length
Long end = new Long(tmpDocContent.length() + contentBuffer.length());
CustomObject obj = null;
// Iterate through stack to modify the End index of the existing elements
Iterator<CustomObject> anIterator = stack.iterator();
while (anIterator.hasNext()) {
// get the object and move to the next one
obj = anIterator.next();
if (incrementStartIndex && obj.getStart().equals(obj.getEnd())) {
obj.setStart(new Long(obj.getStart().longValue() + 1));
} // End if
// sets its End index
obj.setEnd(end);
} // End while
tmpDocContent.append(contentBuffer.toString());
} // characters();