List of usage examples for javax.xml.stream XMLStreamConstants END_DOCUMENT
int END_DOCUMENT
To view the source code for javax.xml.stream XMLStreamConstants END_DOCUMENT.
Click Source Link
From source file:edu.harvard.iq.dvn.ingest.statdataio.impl.plugins.ddi.DDIFileReader.java
private void processDataDscr(XMLStreamReader xmlr, SDIOMetadata smd) throws XMLStreamException { int variableCounter = 0; for (int event = xmlr.next(); event != XMLStreamConstants.END_DOCUMENT; event = xmlr.next()) { if (event == XMLStreamConstants.START_ELEMENT) { if (xmlr.getLocalName().equals("var")) { processVar(xmlr, smd, variableCounter); variableCounter++;/*ww w .ja va2 s . com*/ } else { throw new XMLStreamException( "Unsupported element in the DDI: codeBook/dataDscr/" + xmlr.getLocalName()); } } else if (event == XMLStreamConstants.END_ELEMENT) { if (xmlr.getLocalName().equals("dataDscr")) { dbgLog.info("processed dataDscr section"); return; } else { throw new XMLStreamException( "Mismatched DDI Formatting: </dataDscr> expected, found " + xmlr.getLocalName()); } } } }
From source file:edu.harvard.iq.dvn.ingest.statdataio.impl.plugins.ddi.DDIFileReader.java
private void processVar(XMLStreamReader xmlr, SDIOMetadata smd, int variableNumber) throws XMLStreamException { // Attributes: // ID -- can be ignored (will be reassigned); // name:/* w w w. ja v a 2 s .com*/ String variableName = xmlr.getAttributeValue(null, "name"); if (variableName == null || variableName.equals("")) { throw new XMLStreamException("NULL or empty variable name attribute."); } variableNameList.add(variableName); // interval type: String varIntervalType = xmlr.getAttributeValue(null, "intrvl"); // OK if not specified; defaults to discrete: varIntervalType = (varIntervalType == null ? VAR_INTERVAL_DISCRETE : varIntervalType); // Number of decimal points: String dcmlAttr = xmlr.getAttributeValue(null, "dcml"); Long dcmlPoints = null; if (dcmlAttr != null && !dcmlAttr.equals("")) { try { dcmlPoints = new Long(dcmlAttr); } catch (NumberFormatException nfe) { throw new XMLStreamException("Invalid variable dcml attribute: " + dcmlAttr); } } // weighed variables not supported yet -- L.A. //dv.setWeighted( VAR_WEIGHTED.equals( xmlr.getAttributeValue(null, "wgt") ) ); // default is not-wgtd, so null sets weighted to false Map<String, String> valueLabelPairs = new LinkedHashMap<String, String>(); List<String> missingValues = new ArrayList<String>(); for (int event = xmlr.next(); event != XMLStreamConstants.END_DOCUMENT; event = xmlr.next()) { if (event == XMLStreamConstants.START_ELEMENT) { if (xmlr.getLocalName().equals("location")) { processLocation(xmlr, smd); } else if (xmlr.getLocalName().equals("labl")) { String _labl = processLabl(xmlr, LEVEL_VARIABLE); if (_labl != null && !_labl.equals("")) { variableLabelMap.put(variableName, _labl); } } else if (xmlr.getLocalName().equals("invalrng")) { processInvalrng(xmlr, smd, variableName); } else if (xmlr.getLocalName().equals("varFormat")) { int simpleVariableFormat = processVarFormat(xmlr, smd, variableName); variableTypeList.add(simpleVariableFormat); } else if (xmlr.getLocalName().equals("catgry")) { processCatgry(xmlr, valueLabelPairs, missingValues, variableName); } else if (xmlr.getLocalName().equals("universe")) { // Should not occur in TAB+DDI ingest (?) // ignore. } else if (xmlr.getLocalName().equals("concept")) { // Same deal. } else if (xmlr.getLocalName().equals("notes")) { // Same. } else if (xmlr.getLocalName().equals("sumStat")) { throw new XMLStreamException( "sumStat (Summary Statistics) section found in a variable section; not supported!"); } // the "todos" below are from DDIServiceBean -- L.A. // todo: qstnTxt: wait to handle until we know more of how we will use it // todo: wgt-var : waitng to see example } else if (event == XMLStreamConstants.END_ELEMENT) { if (xmlr.getLocalName().equals("var")) { // Before returning, set rich variable type metadata: if (variableTypeList.get(variableNumber) == 0) { // this is a numeric variable; // It could be discrete or continuous, with the number // of decimal points explicitely specified. unfVariableTypes.put(variableName, 0); // default printFormatList.add(5); formatCategoryTable.put(variableName, "other"); if ((dcmlPoints != null && dcmlPoints > new Long(0)) || (!varIntervalType.equals(VAR_INTERVAL_DISCRETE))) { unfVariableTypes.put(variableName, 1); decimalVariableSet.add(variableNumber); if (dcmlPoints == null || dcmlPoints == new Long(0)) { dcmlPoints = new Long(7); } printFormatNameTable.put(variableName, "F8." + dcmlPoints); } else { unfVariableTypes.put(variableName, 0); } } else { // this is a string variable. // (or Date; whih is stored as a string, for most purposes, // but requires some special treatment. unfVariableTypes.put(variableName, -1); printFormatList.add(1); if (!"date".equals(formatCategoryTable.get(variableName)) && !"time".equals(formatCategoryTable.get(variableName))) { formatCategoryTable.put(variableName, "other"); } } // Variable Label: // (if it's not supplied, we create an empty placeholder) if (variableLabelMap.get(variableName) == null || variableLabelMap.get(variableName).equals("")) { variableLabelMap.put(variableName, ""); } // Value Labels: if (!valueLabelPairs.isEmpty()) { valueLabelTable.put(variableName, valueLabelPairs); dbgLog.warning("valueLabelTable = " + valueLabelTable.toString()); dbgLog.warning("valueLabelTable = " + valueLabelTable); valueVariableMappingTable.put(variableName, variableName); } // Missing Values: if (!missingValues.isEmpty()) { missingValueTable.put(variableName, missingValues); } dbgLog.info("processed variable number " + variableNumber); return; } } } }
From source file:com.liferay.portal.util.LocalizationImpl.java
private void _copyNonExempt(XMLStreamReader xmlStreamReader, XMLStreamWriter xmlStreamWriter, String exemptLanguageId, String defaultLanguageId, boolean cdata) throws XMLStreamException { while (xmlStreamReader.hasNext()) { int event = xmlStreamReader.next(); if (event == XMLStreamConstants.START_ELEMENT) { String languageId = xmlStreamReader.getAttributeValue(null, _LANGUAGE_ID); if (Validator.isNull(languageId)) { languageId = defaultLanguageId; }//from ww w . j a v a 2 s . c o m if (!languageId.equals(exemptLanguageId)) { xmlStreamWriter.writeStartElement(xmlStreamReader.getLocalName()); xmlStreamWriter.writeAttribute(_LANGUAGE_ID, languageId); while (xmlStreamReader.hasNext()) { event = xmlStreamReader.next(); if ((event == XMLStreamConstants.CHARACTERS) || (event == XMLStreamConstants.CDATA)) { String text = xmlStreamReader.getText(); if (cdata) { xmlStreamWriter.writeCData(text); } else { xmlStreamWriter.writeCharacters(xmlStreamReader.getText()); } break; } else if (event == XMLStreamConstants.END_ELEMENT) { break; } } xmlStreamWriter.writeEndElement(); } } else if (event == XMLStreamConstants.END_DOCUMENT) { break; } } }
From source file:com.hp.alm.ali.rest.client.AliRestClient.java
private List<String> getAttributeValues(InputStream xml, String elemName, String attrName) { try {//from ww w . ja va 2s . c o m XMLInputFactory factory = XmlUtils.createBasicInputFactory(); XMLStreamReader parser; parser = factory.createXMLStreamReader(xml); List<String> result = new LinkedList<String>(); while (true) { int event = parser.next(); if (event == XMLStreamConstants.END_DOCUMENT) { parser.close(); break; } if (event == XMLStreamConstants.START_ELEMENT && elemName.equals(parser.getLocalName())) { for (int i = 0; i < parser.getAttributeCount(); i++) { String localName = parser.getAttributeLocalName(i); if (attrName.equals(localName)) { result.add(parser.getAttributeValue(i)); break; } } } } return result; } catch (XMLStreamException e) { throw new RuntimeException(e); } finally { try { xml.close(); } catch (IOException e) { // ignore } } }
From source file:edu.harvard.iq.dvn.ingest.statdataio.impl.plugins.ddi.DDIFileReader.java
private void processInvalrng(XMLStreamReader xmlr, SDIOMetadata smd, String variableName) throws XMLStreamException { for (int event = xmlr.next(); event != XMLStreamConstants.END_DOCUMENT; event = xmlr.next()) { if (event == XMLStreamConstants.START_ELEMENT) { if (xmlr.getLocalName().equals("item")) { // Ranges -- not supported? (TODO) //VariableRange range = new VariableRange(); //dv.getInvalidRanges().add(range); //range.setDataVariable(dv); //range.setBeginValue( xmlr.getAttributeValue(null, "VALUE") ); //range.setBeginValueType(varService.findVariableRangeTypeByName( variableRangeTypeList, DB_VAR_RANGE_TYPE_POINT ) ); String invalidValue = xmlr.getAttributeValue(null, "VALUE"); // STORE (?) // TODO: // Figure out what to do with these. // Aren't the same values specified as "MISSING" in the // categories? What's the difference? -- otherwise this // is kind of redundant. //} else if (xmlr.getLocalName().equals("range")) { // ... } } else { throw new XMLStreamException( "Unsupported DDI Element: codeBook/dataDscr/var/invalrng/" + xmlr.getLocalName()); }/*from w w w .ja va 2s. c o m*/ } else if (event == XMLStreamConstants.END_ELEMENT) { if (xmlr.getLocalName().equals("invalrng")) { return; } else if (xmlr.getLocalName().equals("item")) { // continue; } else { throw new XMLStreamException( "Mismatched DDI Formatting: </invalrng> expected, found " + xmlr.getLocalName()); } } } }
From source file:edu.harvard.iq.dvn.ingest.statdataio.impl.plugins.ddi.DDIFileReader.java
private void processCatgry(XMLStreamReader xmlr, Map<String, String> valueLabelPairs, List<String> missingValues, String variableName) throws XMLStreamException { boolean isMissing = "Y".equals(xmlr.getAttributeValue(null, "missing")); // (default is N, so null sets missing to false) // STORE -- (TODO NOW) //cat.setDataVariable(dv); //dv.getCategories().add(cat); String varValue = null;/* w ww . jav a 2s .c om*/ String valueLabel = null; for (int event = xmlr.next(); event != XMLStreamConstants.END_DOCUMENT; event = xmlr.next()) { if (event == XMLStreamConstants.START_ELEMENT) { if (xmlr.getLocalName().equals("labl")) { valueLabel = processLabl(xmlr, LEVEL_CATEGORY); } else if (xmlr.getLocalName().equals("catValu")) { varValue = parseText(xmlr, false); } else if (xmlr.getLocalName().equals("catStat")) { // category statistics should not be present in the // TAB file + DDI card ingest: throw new XMLStreamException( "catStat (Category Statistics) section found in a variable section; not supported!"); } } else if (event == XMLStreamConstants.END_ELEMENT) { if (xmlr.getLocalName().equals("catgry")) { if (varValue != null && !varValue.equals("")) { if (valueLabel != null && !valueLabel.equals("")) { dbgLog.fine("DDI Reader: storing label " + valueLabel + " for value " + varValue); valueLabelPairs.put(varValue, valueLabel); } if (isMissing) { missingValues.add(varValue); } } return; } else if (xmlr.getLocalName().equals("catValu")) { // continue; } else if (xmlr.getLocalName().equals("labl")) { // continue; } else { throw new XMLStreamException( "Mismatched DDI Formatting: </catgry> expected, found " + xmlr.getLocalName()); } } } }
From source file:edu.harvard.iq.safe.lockss.impl.LOCKSSDaemonStatusTableXmlStreamParser.java
/** * * @param stream/*w w w . j a v a 2s . c om*/ * @param encoding */ @Override public void read(InputStream stream, String encoding) { // logger.setLevel(Level.FINE); // 1. create Input factory XMLInputFactory xmlif = XMLInputFactory.newInstance(); xmlif.setProperty("javax.xml.stream.isCoalescing", java.lang.Boolean.TRUE); xmlif.setProperty("javax.xml.stream.isNamespaceAware", java.lang.Boolean.TRUE); long startTime = System.currentTimeMillis(); int noAUs = 0; String aus = null; String currentTableId = null; String currentTableTitle = null; String currentTableKey = null; boolean hasErrorsColumn = false; String siAuId = null; XMLStreamReader xmlr = null; try { // create reader xmlr = xmlif.createXMLStreamReader(new BufferedInputStream(stream), encoding); String curElement = ""; boolean isLastTagnameTable = false; String targetTagName = "row"; String cellTagName = "columnname"; boolean withinSummaryinfo = false; boolean withinColumndescriptor = false; boolean withinRow = false; boolean withinCell = false; boolean withinReference = false; boolean isCrawlStatusActive = false; boolean isCrawlStatusColumn = false; int valueTagCounter = 0; String currentColumnName = null; String currentCellValue = null; String currentCellKey = null; SummaryInfo si = null; List<String> rowData = null; Map<String, String> rowDataH = null; w1: while (xmlr.hasNext()) { int eventType = xmlr.next(); switch (eventType) { case XMLStreamConstants.START_ELEMENT: curElement = xmlr.getLocalName(); // note: getName() -> // QName logger.log(Level.FINE, "--------- start tag = <{0}> ---------", curElement); // check the table name first if (curElement.equals("table")) { isLastTagnameTable = true; } else if (curElement.equals("error")) { isTargetPageValid = false; break w1; } if (isLastTagnameTable) { if (curElement.equals("name")) { currentTableId = xmlr.getElementText(); logger.log(Level.FINE, "########### table Id = [{0}] ###########", currentTableId); tableId = currentTableId; if (belongsInclusionTableList.contains(currentTableId)) { logger.log(Level.FINE, "!!!!! Table ({0}) belongs to the target list !!!!!", currentTableId); } else { logger.log(Level.FINE, "XXXXXXXXXXX Table ({0}) does not belong to the target list XXXXXXXXXXX", currentTableId); break w1; } } else if (curElement.equals("key")) { currentTableKey = xmlr.getElementText(); logger.log(Level.FINE, "---------- table key = ({0}) ----------", currentTableKey); tableKey = currentTableKey; } else if (curElement.equals("title")) { currentTableTitle = xmlr.getElementText(); logger.log(Level.FINE, "+++++++++ table Title = ({0}) +++++++++", currentTableTitle); if (currentTableId.equals("PeerRepair")) { if (currentTableTitle.startsWith("Repair candidates for AU: ")) { currentTableTitle = currentTableTitle.replaceFirst("Repair candidates for AU: ", ""); logger.log(Level.FINE, "save this modified table-Title as auName={0}", currentTableTitle); this.tableTitle = currentTableTitle; } else { logger.log(Level.WARNING, "The table-Title does not start with the expected token={0}", currentTableTitle); } } isLastTagnameTable = false; } } if (curElement.equals("columndescriptor")) { withinColumndescriptor = true; } else if (curElement.equals("row")) { withinRow = true; rowCounter++; logger.log(Level.FINE, "================== {0}-th row starts here ==================", rowCounter); // set-up the table storage //if (rowCounter == 1) { // 1st row rowData = new ArrayList<String>(); rowDataH = new LinkedHashMap<String, String>(); //} } else if (curElement.equals("cell")) { logger.log(Level.FINE, "entering a cell"); withinCell = true; } else if (curElement.equals("reference")) { withinReference = true; logger.log(Level.FINE, "within reference on"); } else if (curElement.equals("summaryinfo")) { withinSummaryinfo = true; si = new SummaryInfo(); } else if (curElement.equals("value")) { logger.log(Level.FINE, "entering a value"); valueTagCounter++; } //---- columndescriptor tag --------------------------------------------------- if (withinColumndescriptor) { if (curElement.equals("name")) { String nameText = xmlr.getElementText(); logger.log(Level.FINE, "\tcolumndescriptor: name = {0}", nameText); columndescriptorList.add(nameText); } else if (curElement.equals("title")) { String titleText = xmlr.getElementText(); logger.log(Level.FINE, "\tcolumndescriptor: title = {0}", titleText); } else if (curElement.equals("type")) { String typeText = xmlr.getElementText(); logger.log(Level.FINE, "\tcolumndescriptor: type = {0}", typeText); getTypeList().add(typeText); } } //---- cell tag ---------------------------------------------------------------- if (withinCell) { logger.log(Level.FINE, "parsing withinCell"); if (curElement.equals("columnname")) { String columnname = xmlr.getElementText(); logger.log(Level.FINE, "\t\tcolumnname = {0}", columnname); currentColumnName = columnname; if (columnname.equals("crawl_status")) { isCrawlStatusColumn = true; } else { isCrawlStatusColumn = false; } if (columnname.equals("Errors")) { hasErrorsColumn = true; } } else { // value tag block: either value-tag WO a child element // or with a child element /* * <value><reference>...<value>xxxx</value> * <value>xxxx</value> */ if ((curElement.equals("value")) && (!withinReference)) { logger.log(Level.FINE, "entering el:value/WO-REF block"); if (!hasReferenceTag.contains(currentColumnName)) { logger.log(Level.FINE, "No child reference tag is expected for this value tag"); logger.log(Level.FINEST, "xmlr.getEventType():pre-parsing={0}", xmlr.getEventType()); String cellValue = xmlr.getElementText(); // note: the above parsing action moves the // cursor to the end-tag, i.e., </value> // therefore, the end-element-switch-block below // cannot catch this </value> tag logger.log(Level.FINE, "\t\t\t[No ref: value] {0} = {1}", new Object[] { currentColumnName, cellValue }); currentCellValue = cellValue; logger.log(Level.FINEST, "xmlr.getEventType():post-parsing={0}", xmlr.getEventType()); // store this value // rowData logger.log(Level.FINE, "current column name={0}", currentColumnName); logger.log(Level.FINE, "valueTagCounter={0}", valueTagCounter); if (currentColumnName.endsWith("Damaged")) { if (valueTagCounter <= 1) { // 2nd value tag is footnot for this column // ignore this value rowData.add(cellValue); rowDataH.put(currentColumnName, currentCellValue); } } else { rowData.add(cellValue); rowDataH.put(currentColumnName, currentCellValue); } } else { // previously this block was unthinkable, but // it was found that there are columns that // temporarily have a <reference> tag in // crawl_status_table; these columns are // included in hasReferenceTag by default; // thus, for such unstable columns, // when they hava a <reference tag, // data are caputred in another within- // reference block; however, when these // columns no longer have <reference> tag, // text data would be left uncaptured unless // some follow-up processing takes place here logger.log(Level.FINE, "May have to capture data: column={0}", currentColumnName); if (mayHaveReferenceTag.contains(currentColumnName) && !isCrawlStatusActive) { // because the crawling is not active, // it is safely assume that the maybe columns have no reference tag // 2011-10-24 the above assumption was found wrong // a crawling cell does not say active but // subsequent columns have a reference logger.log(Level.FINE, "a text or a reference tag : try to parse it as a text"); String cellValue = null; try { cellValue = xmlr.getElementText(); } catch (javax.xml.stream.XMLStreamException ex) { continue; } finally { } logger.log(Level.FINE, "\t\t\t[value WO-ref(crawling_NOT_active case)={0}]", currentColumnName + " = " + cellValue); currentCellValue = cellValue; // store this value // rowData logger.log(Level.FINE, "\t\t\tcurrent columnName={0}", currentColumnName); rowData.add(cellValue); rowDataH.put(currentColumnName, currentCellValue); } else { logger.log(Level.FINE, "WO-Ref: no processing items now:{0}", curElement); } } } else if (withinReference) { // reference tag exists logger.log(Level.FINE, "WR:curElement={0}", curElement); if (curElement.equals("key")) { String cellKey = xmlr.getElementText(); logger.log(Level.FINE, "\t\tcurrentCellKey is set to={0}", cellKey); currentCellKey = cellKey; } else if (curElement.equals("value")) { String cellValue = xmlr.getElementText(); logger.log(Level.FINE, "\t\twr: {0} = {1}", new Object[] { currentColumnName, cellValue }); // exception cases follow: if (currentColumnName.equals("AuName")) { logger.log(Level.FINE, "\t\tAuName is replaced with the key[=AuId]= {0}", currentCellKey); // rowData // This block is for ArchivalUnitStatusTable // add the key as a new datum (auId) // ahead of its value rowData.add(currentCellKey); rowDataH.put("AuId", currentCellKey); currentCellValue = cellValue; } else if (currentColumnName.equals("auId")) { // This block is for V3PollerTable logger.log(Level.FINE, "\t\tnew value for auId(V3PollerTable)={0}", currentCellKey); // deprecated after 2012-02-02: use key as data // currentCellValue = currentCellKey; // add auName as a new column ahead of auId rowData.add(cellValue); rowDataH.put("auName", cellValue); logger.log(Level.FINE, "\t\tauName(V3PollerTable)={0}", cellValue); currentCellValue = currentCellKey; } else if (currentColumnName.equals("pollId")) { // this block is for V3PollerTable logger.log(Level.FINE, "\t\tFull string (key) is used={0}", currentCellKey); // The key has the complete string whereas // the value is its truncated copy currentCellValue = currentCellKey; } else if (currentColumnName.equals("au")) { logger.log(Level.FINE, "\t\tauId is used instead for au(crawl_status_table)={0}", currentCellKey); // 2012-02-02: add auName ahead of au rowData.add(cellValue); rowDataH.put("auName", cellValue); logger.log(Level.FINE, "\t\tauName={0}", cellValue); // rowData // This block is for crawl_status_table // save the key(auId) instead of value currentCellValue = currentCellKey; } else if (currentColumnName.equals("Peers")) { logger.log(Level.FINE, "\t\tURL (key) is used={0}", currentCellKey); currentCellValue = DaemonStatusDataUtil.escapeHtml(currentCellKey); logger.log(Level.FINE, "\t\tAfter encoding ={0}", currentCellValue); } else { if (isCrawlStatusColumn) { // if the craw status column is // "active", some later columns // may have a reference tag // so turn on the switch if (cellValue.equals("Active") || (cellValue.equals("Pending"))) { isCrawlStatusActive = true; } else { isCrawlStatusActive = false; } } // the default processing currentCellValue = cellValue; } // store currentCellValue logger.log(Level.FINE, "currentCellValue={0}", currentCellValue); // rowData rowData.add(currentCellValue); rowDataH.put(currentColumnName, currentCellValue); } // Within ref tag: key and valu processing } // value with text or value with ref tag } // columnname or value } // within cell // ---- summaryinfo tag -------------------------------------------------------- if (withinSummaryinfo) { logger.log(Level.FINE, "============================ Within SummaryInfo ============================ "); if (curElement.equals("title")) { String text = xmlr.getElementText(); si.setTitle(text); logger.log(Level.FINE, "\tsi:titile={0}", si.getTitle()); } else if (curElement.equals("type")) { String text = xmlr.getElementText(); si.setType(Integer.parseInt(text)); logger.log(Level.FINE, "\tsi:type={0}", si.getType()); } else if (curElement.equals("key")) { if (withinReference && si.getTitle().equals("Volume")) { String text = xmlr.getElementText(); logger.log(Level.FINE, "\tsi:key contents(Volume case)={0}", text); siAuId = text; // si.setValue(text); logger.log(Level.FINE, "\tsi:value(Volume case)={0}", siAuId); } } else if (curElement.equals("value")) { if (withinReference) { if (hasRefTitileTagsSI.contains(si.getTitle())) { if (si.getTitle().equals("Volume")) { // 2012-02-02 use the au name String text = xmlr.getElementText(); si.setValue(text); logger.log(Level.FINE, "\tsi:value(Volume case)={0}", si.getValue()); } else { String text = xmlr.getElementText(); si.setValue(text); logger.log(Level.FINE, "\tsi:value={0}", si.getValue()); } } } else { // note: 2012-02-07 // daemon 1.59.2 uses the new layout for AU page // this layout includes a summaryinfo tag // that now contains a reference tag String text = null; try { text = xmlr.getElementText(); if (!hasRefTitileTagsSI.contains(si.getTitle())) { si.setValue(text); logger.log(Level.FINE, "\tsi:value={0}", si.getValue()); } } catch (javax.xml.stream.XMLStreamException ex) { logger.log(Level.WARNING, "encounter a reference tag rather than text"); continue; } finally { } } } /* * aus = xmlr.getElementText(); * out.println("found token=[" + aus + "]"); if * (currentTableId.equals("ArchivalUnitStatusTable")) { * m = pau.matcher(aus); if (m.find()) { * out.println("How many AUs=" + m.group(1)); noAUs = * Integer.parseInt(m.group(1)); } else { * out.println("not found within[" + aus + "]"); } } */ } break; case XMLStreamConstants.CHARACTERS: break; case XMLStreamConstants.ATTRIBUTE: break; case XMLStreamConstants.END_ELEMENT: if (xmlr.getLocalName().equals("columndescriptor")) { withinColumndescriptor = false; logger.log(Level.FINE, "leaving columndescriptor"); } else if (xmlr.getLocalName().equals("row")) { if (withinRow) { logger.log(Level.FINE, "========= end of the target row element"); withinRow = false; } if (!isCrawlStatusActive) { tabularData.add(rowData); tableData.add(rowDataH); } else { rowIgnored++; rowCounter--; } rowData = null; rowDataH = null; isCrawlStatusActive = false; } else if (xmlr.getLocalName().equals("cell")) { // rowDataH.add(cellDatum); cellCounter++; withinCell = false; currentColumnName = null; currentCellValue = null; currentCellKey = null; isCrawlStatusColumn = false; valueTagCounter = 0; logger.log(Level.FINE, "leaving cell"); } else if (xmlr.getLocalName().equals("columnname")) { logger.log(Level.FINE, "leaving columnname"); } else if (xmlr.getLocalName().equals("reference")) { withinReference = false; } else if (xmlr.getLocalName().equals("summaryinfo")) { logger.log(Level.FINE, "si={0}", si.toString()); summaryInfoList.add(si); si = null; withinSummaryinfo = false; } else if (xmlr.getLocalName().equals("value")) { logger.log(Level.FINE, "leaving value"); } else { logger.log(Level.FINE, "--------- end tag = <{0}> ---------", curElement); } break; case XMLStreamConstants.END_DOCUMENT: logger.log(Level.FINE, "Total of {0} row occurrences", rowCounter); } // end: switch } // end:while } catch (XMLStreamException ex) { logger.log(Level.WARNING, "XMLStreamException occurs", ex); this.isTargetPageValid = false; } catch (RuntimeException re) { logger.log(Level.WARNING, "some RuntimeException occurs", re); this.isTargetPageValid = false; } catch (Exception e) { logger.log(Level.WARNING, "some Exception occurs", e); this.isTargetPageValid = false; } finally { // 5. close reader/IO if (xmlr != null) { try { xmlr.close(); } catch (XMLStreamException ex) { logger.log(Level.WARNING, "XMLStreamException occurs during close()", ex); } } if (!this.isTargetPageValid) { logger.log(Level.WARNING, "This parsing session may not be complete due to some exception reported earlier"); } } // end of try if (currentTableId.equals("V3PollerDetailTable")) { summaryInfoList.add(new SummaryInfo("auId", 4, siAuId)); summaryInfoMap = new LinkedHashMap<String, String>(); for (SummaryInfo si : summaryInfoList) { summaryInfoMap.put(si.getTitle(), si.getValue()); } } // parsing summary logger.log(Level.FINE, "###################### parsing summary ######################"); logger.log(Level.FINE, "currentTableId={0}", currentTableId); logger.log(Level.FINE, "currentTableTitle={0}", currentTableTitle); logger.log(Level.FINE, "currentTableKey={0}", currentTableKey); logger.log(Level.FINE, "columndescriptorList={0}", columndescriptorList); logger.log(Level.FINE, "# of columndescriptors={0}", columndescriptorList.size()); logger.log(Level.FINE, "typeList={0}", typeList); logger.log(Level.FINE, "# of rows counted={0}", rowCounter); logger.log(Level.FINE, "# of rows excluded[active ones are excluded]={0}", rowIgnored); logger.log(Level.FINE, "summaryInfoList:size={0}", summaryInfoList.size()); logger.log(Level.FINE, "summaryInfoList={0}", summaryInfoList); logger.log(Level.FINE, "table: cell counts = {0}", cellCounter); logger.log(Level.FINE, "tableData[map]=\n{0}", tableData); logger.log(Level.FINE, "tabularData[list]=\n{0}", tabularData); /* * if (currentTableId.equals("ArchivalUnitStatusTable")) { if * (rowCounter == noAUs) { out.println("au counting is OK=" + * rowCounter); } else { err.println("au counting disagreement"); throw * new RuntimeException("parsing error is suspected"); } } */ logger.log(Level.FINE, " completed in {0} ms\n\n", (System.currentTimeMillis() - startTime)); if (!columndescriptorList.isEmpty()) { int noCols = columndescriptorList.size(); if (currentTableId.equals("V3PollerTable") && !hasErrorsColumn) { noCols--; } int noCellsExpd = rowCounter * noCols; if (noCols > 0) { // this table has a table logger.log(Level.FINE, "checking parsing results: table dimmensions"); if (noCellsExpd == cellCounter) { logger.log(Level.FINE, "table dimensions and cell-count are consistent"); } else { int diff = noCellsExpd - cellCounter; logger.log(Level.FINE, "The table has {0} incomplete cells", diff); hasIncompleteRows = true; setIncompleteRowList(); logger.log(Level.FINE, "incomplete rows: {0}", incompleteRows); } } } }
From source file:edu.harvard.iq.dvn.ingest.statdataio.impl.plugins.ddi.DDIFileReader.java
private String getElementText(XMLStreamReader xmlr) throws XMLStreamException { if (xmlr.getEventType() != XMLStreamConstants.START_ELEMENT) { throw new XMLStreamException("parser must be on START_ELEMENT to read next text", xmlr.getLocation()); }/*from w w w .ja v a 2 s. co m*/ int eventType = xmlr.next(); StringBuffer content = new StringBuffer(); while (eventType != XMLStreamConstants.END_ELEMENT) { if (eventType == XMLStreamConstants.CHARACTERS || eventType == XMLStreamConstants.CDATA || eventType == XMLStreamConstants.SPACE /* || eventType == XMLStreamConstants.ENTITY_REFERENCE*/) { content.append(xmlr.getText()); } else if (eventType == XMLStreamConstants.PROCESSING_INSTRUCTION || eventType == XMLStreamConstants.COMMENT || eventType == XMLStreamConstants.ENTITY_REFERENCE) { // skipping } else if (eventType == XMLStreamConstants.END_DOCUMENT) { throw new XMLStreamException("unexpected end of document when reading element text content"); } else if (eventType == XMLStreamConstants.START_ELEMENT) { throw new XMLStreamException("element text content may not contain START_ELEMENT", xmlr.getLocation()); } else { throw new XMLStreamException("Unexpected event type " + eventType, xmlr.getLocation()); } eventType = xmlr.next(); } return content.toString(); }
From source file:ca.uhn.fhir.parser.XmlParser.java
private void encodeXhtml(XhtmlDt theDt, XMLStreamWriter theEventWriter) throws XMLStreamException { if (theDt == null || theDt.getValue() == null) { return;//from w ww . j av a 2 s .co m } boolean firstElement = true; for (XMLEvent event : theDt.getValue()) { switch (event.getEventType()) { case XMLStreamConstants.ATTRIBUTE: Attribute attr = (Attribute) event; if (isBlank(attr.getName().getPrefix())) { if (isBlank(attr.getName().getNamespaceURI())) { theEventWriter.writeAttribute(attr.getName().getLocalPart(), attr.getValue()); } else { theEventWriter.writeAttribute(attr.getName().getNamespaceURI(), attr.getName().getLocalPart(), attr.getValue()); } } else { theEventWriter.writeAttribute(attr.getName().getPrefix(), attr.getName().getNamespaceURI(), attr.getName().getLocalPart(), attr.getValue()); } break; case XMLStreamConstants.CDATA: theEventWriter.writeCData(((Characters) event).getData()); break; case XMLStreamConstants.CHARACTERS: case XMLStreamConstants.SPACE: String data = ((Characters) event).getData(); theEventWriter.writeCharacters(data); break; case XMLStreamConstants.COMMENT: theEventWriter.writeComment(((Comment) event).getText()); break; case XMLStreamConstants.END_ELEMENT: theEventWriter.writeEndElement(); break; case XMLStreamConstants.ENTITY_REFERENCE: EntityReference er = (EntityReference) event; theEventWriter.writeEntityRef(er.getName()); break; case XMLStreamConstants.NAMESPACE: Namespace ns = (Namespace) event; theEventWriter.writeNamespace(ns.getPrefix(), ns.getNamespaceURI()); break; case XMLStreamConstants.START_ELEMENT: StartElement se = event.asStartElement(); if (firstElement) { if (StringUtils.isBlank(se.getName().getPrefix())) { String namespaceURI = se.getName().getNamespaceURI(); if (StringUtils.isBlank(namespaceURI)) { namespaceURI = "http://www.w3.org/1999/xhtml"; } theEventWriter.writeStartElement(se.getName().getLocalPart()); theEventWriter.writeDefaultNamespace(namespaceURI); } else { String prefix = se.getName().getPrefix(); String namespaceURI = se.getName().getNamespaceURI(); theEventWriter.writeStartElement(prefix, se.getName().getLocalPart(), namespaceURI); theEventWriter.writeNamespace(prefix, namespaceURI); } firstElement = false; } else { if (isBlank(se.getName().getPrefix())) { if (isBlank(se.getName().getNamespaceURI())) { theEventWriter.writeStartElement(se.getName().getLocalPart()); } else { if (StringUtils.isBlank(se.getName().getPrefix())) { theEventWriter.writeStartElement(se.getName().getLocalPart()); // theEventWriter.writeDefaultNamespace(se.getName().getNamespaceURI()); } else { theEventWriter.writeStartElement(se.getName().getNamespaceURI(), se.getName().getLocalPart()); } } } else { theEventWriter.writeStartElement(se.getName().getPrefix(), se.getName().getLocalPart(), se.getName().getNamespaceURI()); } for (Iterator<?> attrIter = se.getAttributes(); attrIter.hasNext();) { Attribute next = (Attribute) attrIter.next(); theEventWriter.writeAttribute(next.getName().getLocalPart(), next.getValue()); } } break; case XMLStreamConstants.DTD: case XMLStreamConstants.END_DOCUMENT: case XMLStreamConstants.ENTITY_DECLARATION: case XMLStreamConstants.NOTATION_DECLARATION: case XMLStreamConstants.PROCESSING_INSTRUCTION: case XMLStreamConstants.START_DOCUMENT: break; } } }