Revision 1411
Added by Jing Tao almost 22 years ago
| src/edu/ucsb/nceas/metacat/DBSAXHandler.java | ||
|---|---|---|
| 71 | 71 |
private boolean endDocument = false; |
| 72 | 72 |
protected int serverCode = 1; |
| 73 | 73 |
protected Hashtable namespaces = new Hashtable(); |
| 74 |
protected boolean hitTextNode = false; // a flag to hit text node |
|
| 75 |
// a buffer to keep all text nodes for same element |
|
| 76 |
// it is for element was splited |
|
| 77 |
protected StringBuffer textBuffer = new StringBuffer(); |
|
| 78 |
protected Stack textBufferStack = new Stack(); |
|
| 74 | 79 |
|
| 75 | 80 |
protected static final int MAXDATACHARS = 4000; |
| 76 | 81 |
protected static final long INDEXDELAY = 10000; |
| ... | ... | |
| 194 | 199 |
MetaCatUtil.debugMessage("Start ELEMENT(localName) " + localName, 50);
|
| 195 | 200 |
MetaCatUtil.debugMessage("Start ELEMENT(uri) " + uri, 50);
|
| 196 | 201 |
|
| 197 |
|
|
| 198 | 202 |
DBSAXNode parentNode = null; |
| 199 | 203 |
DBSAXNode currentNode = null; |
| 200 | 204 |
|
| ... | ... | |
| 204 | 208 |
} catch (EmptyStackException e) {
|
| 205 | 209 |
parentNode = null; |
| 206 | 210 |
} |
| 211 |
|
|
| 212 |
// If hit a text node, we need write this text for current's parent node |
|
| 213 |
// This will happend if the element is mixted |
|
| 214 |
if (hitTextNode && parentNode != null) |
|
| 215 |
{
|
|
| 216 |
// write the textbuffer into db for parent node. |
|
| 217 |
writeTextForDBSAXNode(textBuffer, parentNode); |
|
| 218 |
// rest hitTextNode |
|
| 219 |
hitTextNode =false; |
|
| 220 |
// reset textbuffer |
|
| 221 |
textBuffer = null; |
|
| 222 |
textBuffer = new StringBuffer(); |
|
| 223 |
|
|
| 224 |
} |
|
| 207 | 225 |
|
| 208 | 226 |
// Document representation that points to the root document node |
| 209 | 227 |
if (atFirstElement) |
| ... | ... | |
| 449 | 467 |
|
| 450 | 468 |
|
| 451 | 469 |
/** SAX Handler that is called for each XML text node */ |
| 452 |
public void characters(char[] cbuf, int start, int len) throws SAXException {
|
|
| 470 |
public void characters(char[] cbuf, int start, int len) throws SAXException |
|
| 471 |
{
|
|
| 453 | 472 |
MetaCatUtil.debugMessage("CHARACTERS", 50);
|
| 454 |
DBSAXNode currentNode = (DBSAXNode)nodeStack.peek(); |
|
| 455 |
String data = null; |
|
| 456 |
int leftover = len; |
|
| 457 |
int offset = start; |
|
| 458 |
boolean moredata = true; |
|
| 473 |
// buffer all text nodes for same element. This is for text was splited |
|
| 474 |
// into different nodes |
|
| 475 |
textBuffer.append(new String(cbuf, start,len)); |
|
| 476 |
// set hittextnode true |
|
| 477 |
hitTextNode = true; |
|
| 478 |
} |
|
| 459 | 479 |
|
| 460 |
// This loop deals with the case where there are more characters |
|
| 461 |
// than can fit in a single database text field (limit is |
|
| 462 |
// MAXDATACHARS). If the text to be inserted exceeds MAXDATACHARS, |
|
| 463 |
// write a series of nodes that are MAXDATACHARS long, and then the |
|
| 464 |
// final node contains the remainder |
|
| 465 |
while (moredata) {
|
|
| 466 |
if (leftover > MAXDATACHARS) {
|
|
| 467 |
data = new String(cbuf, offset, MAXDATACHARS); |
|
| 468 |
leftover -= MAXDATACHARS; |
|
| 469 |
offset += MAXDATACHARS; |
|
| 470 |
} else {
|
|
| 471 |
data = new String(cbuf, offset, leftover); |
|
| 472 |
moredata = false; |
|
| 473 |
} |
|
| 474 |
|
|
| 475 |
// Write the content of the node to the database |
|
| 476 |
currentNode.writeChildNodeToDB("TEXT", null, data, docid);
|
|
| 477 |
} |
|
| 478 |
} |
|
| 479 |
|
|
| 480 | 480 |
/** |
| 481 | 481 |
* SAX Handler that is called for each XML text node that is |
| 482 | 482 |
* Ignorable white space |
| ... | ... | |
| 530 | 530 |
public void endElement(String uri, String localName, |
| 531 | 531 |
String qName) throws SAXException {
|
| 532 | 532 |
MetaCatUtil.debugMessage("End ELEMENT " + qName, 50);
|
| 533 |
|
|
| 534 |
// write buffered text nodes into db (so no splited) |
|
| 535 |
DBSAXNode currentNode = (DBSAXNode)nodeStack.peek(); |
|
| 536 |
|
|
| 537 |
// If before the end element, the parser hit text nodes and store them |
|
| 538 |
// into the buffer, write the buffer to data base. The reason we put |
|
| 539 |
// write database here is for xerces some time split text node |
|
| 540 |
if (hitTextNode) |
|
| 541 |
{
|
|
| 542 |
MetaCatUtil.debugMessage("Write text into DB in End Element", 50);
|
|
| 543 |
writeTextForDBSAXNode(textBuffer, currentNode); |
|
| 544 |
}//if |
|
| 545 |
|
|
| 546 |
//set hitText false |
|
| 547 |
hitTextNode = false; |
|
| 548 |
// reset textbuff |
|
| 549 |
textBuffer = null; |
|
| 550 |
textBuffer = new StringBuffer(); |
|
| 533 | 551 |
|
| 534 | 552 |
// Get the node from the stack |
| 535 |
DBSAXNode currentNode = (DBSAXNode)nodeStack.pop();
|
|
| 553 |
currentNode = (DBSAXNode)nodeStack.pop(); |
|
| 536 | 554 |
} |
| 537 | 555 |
|
| 538 | 556 |
// |
| ... | ... | |
| 705 | 723 |
public boolean processingDTD() {
|
| 706 | 724 |
return processingDTD; |
| 707 | 725 |
} |
| 726 |
|
|
| 727 |
/* Method to write a text buffer for DBSAXNode*/ |
|
| 728 |
protected void writeTextForDBSAXNode(StringBuffer strBuffer, DBSAXNode node) |
|
| 729 |
throws SAXException |
|
| 730 |
{
|
|
| 731 |
// Check parameter |
|
| 732 |
if ( strBuffer == null || node == null) |
|
| 733 |
{
|
|
| 734 |
return; |
|
| 735 |
} |
|
| 736 |
boolean moredata = true; |
|
| 737 |
String data = null; |
|
| 738 |
int bufferSize = strBuffer.length(); |
|
| 739 |
int start = 0; |
|
| 740 |
|
|
| 741 |
// if there are some cotent in buffer, write it |
|
| 742 |
if (bufferSize > 0) |
|
| 743 |
{
|
|
| 744 |
MetaCatUtil.debugMessage("Write text into DB", 50);
|
|
| 745 |
// This loop deals with the case where there are more characters |
|
| 746 |
// than can fit in a single database text field (limit is |
|
| 747 |
// MAXDATACHARS). If the text to be inserted exceeds MAXDATACHARS, |
|
| 748 |
// write a series of nodes that are MAXDATACHARS long, and then the |
|
| 749 |
// final node contains the remainder |
|
| 750 |
while (moredata) |
|
| 751 |
{
|
|
| 752 |
bufferSize = strBuffer.length(); |
|
| 753 |
if (bufferSize > MAXDATACHARS) |
|
| 754 |
{
|
|
| 755 |
data = strBuffer.substring(start, MAXDATACHARS); |
|
| 756 |
// cut the stringbuffer part that already written into db |
|
| 757 |
strBuffer = strBuffer.delete(start, MAXDATACHARS); |
|
| 758 |
} |
|
| 759 |
else |
|
| 760 |
{
|
|
| 761 |
data = strBuffer.substring(start, bufferSize); |
|
| 762 |
moredata = false; |
|
| 763 |
} |
|
| 764 |
|
|
| 765 |
// Write the content of the node to the database |
|
| 766 |
node.writeChildNodeToDB("TEXT", null, data, docid);
|
|
| 767 |
}//while |
|
| 768 |
}//if |
|
| 769 |
} |
|
| 708 | 770 |
} |
Also available in: Unified diff
Add the code the handle text node was splitted