Mercurial > hg > mpdl-group
view software/eXist/mpdl-modules/src/de/mpg/mpiwg/berlin/mpdl/lt/doc/NormDictContentHandler.java @ 16:257f67be5c00
diverse Fehlerbehebungen
author | Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de> |
---|---|
date | Tue, 27 Sep 2011 16:40:57 +0200 |
parents | 5df60f24e997 |
children |
line wrap: on
line source
package de.mpg.mpiwg.berlin.mpdl.lt.doc; import java.util.ArrayList; import org.apache.lucene.analysis.Token; import org.xml.sax.*; import de.mpg.mpiwg.berlin.mpdl.exception.ApplicationException; import de.mpg.mpiwg.berlin.mpdl.lt.analyzer.MpdlNormalizer; import de.mpg.mpiwg.berlin.mpdl.lt.analyzer.MpdlTokenizerAnalyzer; import de.mpg.mpiwg.berlin.mpdl.lt.general.Language; import de.mpg.mpiwg.berlin.mpdl.lt.lex.db.LexHandler; import de.mpg.mpiwg.berlin.mpdl.util.StringUtilEscapeChars; public class NormDictContentHandler implements ContentHandler { private static String COMPLEX_ELEMENT_MARK = new Character('\u2425').toString(); // word delimiting element private static String COMPLEX_ELEMENT_NWD_MARK = new Character('\u2424').toString(); // not word delimiting element private static int COMPLEX_ELEMENT_MARK_SIZE = COMPLEX_ELEMENT_MARK.length(); private static int ELEMENT_TYPE_CHARACTERS = 1; private static int ELEMENT_TYPE_COMPLEX = 2; private String[] normalizeFunctions = {}; // default: without normalize functions private boolean dictMode = false; // default: not in dictionary mode private String xmlnsString = ""; private String language; private String outputXmlFragment = ""; private Element rootElement; private Element currentElement; private ArrayList<Element> elementQueue; public NormDictContentHandler(String[] normalizeFunctions, String language) throws ApplicationException { if (normalizeFunctions == null) { String[] emptyFunctions = {}; this.normalizeFunctions = emptyFunctions; } else { this.normalizeFunctions = normalizeFunctions; } this.language = language; } public void setDictMode(boolean dictMode) { this.dictMode = dictMode; } public String getXmlFragment() { return outputXmlFragment; } public void startDocument() throws SAXException { } public void endDocument() throws SAXException { try { String rootElemToStr = rootElement.toXmlString(); // hack: in echo documents the spaces between sentences should be removed if (rootElemToStr != null && rootElemToStr.startsWith("<echo") && Language.getInstance().isChinese(language)) { rootElemToStr = rootElemToStr.replaceAll("</s>[ \n\t]+<s", "</s><s"); } write(rootElemToStr); write("\n"); } catch (NullPointerException e) { throw new SAXException(e); } } public void characters(char[] c, int start, int length) throws SAXException { char[] cCopy = new char[length]; System.arraycopy(c, start, cCopy, 0, length); String charactersStr = String.valueOf(cCopy); if (charactersStr != null && ! charactersStr.equals("")) { if (currentElement != null) { Element charElement = new Element("characters", ELEMENT_TYPE_CHARACTERS); charElement.value = StringUtilEscapeChars.deresolveXmlEntities(charactersStr); if (currentElement.composites == null) currentElement.composites = new ArrayList<Element>(); currentElement.composites.add(charElement); } } } public void ignorableWhitespace(char[] c, int start, int length) throws SAXException { } public void processingInstruction(String target, String data) throws SAXException { } public void setDocumentLocator(Locator locator) { } public void startPrefixMapping(String prefix, String uri) throws SAXException { xmlnsString += "xmlns:" + prefix + "=\"" + uri + "\" "; if (prefix != null && prefix.equals("")) xmlnsString = "xmlns" + prefix + "=\"" + uri + "\" "; } public void endPrefixMapping(String prefix) throws SAXException { } public void skippedEntity(String name) throws SAXException { } public void startElement(String uri, String localName, String name, Attributes attrs) throws SAXException { if (elementQueue == null) elementQueue = new ArrayList<Element>(); Element newElement = new Element(name); // element of type: complex if (currentElement != null) { if (currentElement.composites == null) currentElement.composites = new ArrayList<Element>(); if (currentElement.lang != null) newElement.lang = currentElement.lang; // language is inherited to childs currentElement.composites.add(newElement); } currentElement = newElement; int attrSize = attrs.getLength(); String attrString = ""; for (int i=0; i<attrSize; i++) { String attrQName = attrs.getQName(i); String attrValue = attrs.getValue(i); attrValue = StringUtilEscapeChars.forXML(attrValue); attrString = attrString + " " + attrQName + "=\"" + attrValue + "\""; if (attrQName != null && (attrQName.toLowerCase().equals("xml:lang") || attrQName.toLowerCase().equals("lang"))) currentElement.lang = attrValue; // if xml:lang is set, it is set to the new element and overwrites values inherited by the father } currentElement.attrString = attrString; if (! xmlnsString.equals("")) { currentElement.xmlnsString = xmlnsString; } xmlnsString = ""; elementQueue.add(currentElement); // only the first element is the root element if(rootElement == null) rootElement = currentElement; } public void endElement(String uri, String localName, String name) throws SAXException { if (elementQueue != null && elementQueue.size() > 0) { int lastIndex = elementQueue.size() - 1; elementQueue.remove(lastIndex); } if (elementQueue != null && elementQueue.size() > 0) { int lastIndex = elementQueue.size() - 1; currentElement = elementQueue.get(lastIndex); } else { currentElement = null; } } private void write(String outStr) throws SAXException { outputXmlFragment += outStr; } private class Element { private int type; private String name; private String xmlnsString; private String attrString; private String value; private String lang; // normally value of attribute xml:lang or the inherited xml:lang value of the father node private ArrayList<Element> composites; private Element(String name) { this.type = ELEMENT_TYPE_COMPLEX; this.name = name; } private Element(String name, int type) { this.type = type; this.name = name; } private boolean isComplex() { boolean isComplex = false; if (type == ELEMENT_TYPE_COMPLEX) isComplex = true; return isComplex; } /** * feel free to add/remove some element names; element content must be empty * @return true if element is a word delimiter element else false */ private boolean isWordDelimiterElement() { boolean isWordDelimiterElement = true; // "note" causes problems: word after the note is not recognized // "emph" causes problems: e.g. "Natur<emph>ereignis</emph> enthüllte" is replaced by "Natur<emph><w>ereignis</w></emph>enthüllte" if (name.equals("lb") || name.equals("br") || name.equals("cb") || name.equals("figure") || name.equals("image") || name.equals("handwritten") || name.equals("anchor")) isWordDelimiterElement = false; return isWordDelimiterElement; } private String toXmlString() throws SAXException { String retString = ""; String elemLanguage = language; // default value for the document/page if (lang != null) elemLanguage = lang; // value of the element if available // write this element if (! isComplex()) { retString += value; } else { String xmlNsString = this.xmlnsString; if (xmlNsString == null || xmlNsString.equals("")) { retString = retString + "<" + name + attrString + ">"; } else { retString = retString + "<" + name + " " + xmlNsString + attrString + ">"; } if (composites != null) { String compositesCharsWithMarks = ""; ArrayList<Element> complexElements = new ArrayList<Element>(); for (int i=0; i<composites.size(); i++) { Element composite = composites.get(i); if (! composite.isComplex()) { if (composite.value != null && ! composite.value.equals("")) { String compositeValueStr = composite.value; compositeValueStr = compositeValueStr.replaceAll("\n", ""); // remove all newlines, they are no separators for words. compositeValueStr = compositeValueStr.replaceAll("[ \t]+", " "); // if there are many Blanks/Tabs make them to one compositesCharsWithMarks = compositesCharsWithMarks + compositeValueStr; } } else { if (! composite.isWordDelimiterElement()) { compositesCharsWithMarks = compositesCharsWithMarks + COMPLEX_ELEMENT_NWD_MARK; // add a special mark symbol at the position of the "not word delimiter element" (e.g. <lb>) } else { compositesCharsWithMarks = compositesCharsWithMarks + COMPLEX_ELEMENT_MARK; // add a special mark symbol at the position of the "word delimiter element" (e.g. <var>) } complexElements.add(composite); } } compositesCharsWithMarks = compositesCharsWithMarks.replaceAll(COMPLEX_ELEMENT_NWD_MARK + " +", COMPLEX_ELEMENT_NWD_MARK); // remove Blanks after the non word breaking mark (e.g. "praebi<lb/> ta" is changed to "praebi<lb/>ta") String compositesCharsWithMarksNormalized = normalizeWords(compositesCharsWithMarks, elemLanguage); compositesCharsWithMarksNormalized = compositesCharsWithMarksNormalized.replaceAll(COMPLEX_ELEMENT_NWD_MARK, COMPLEX_ELEMENT_MARK); // mark symbols are unified to COMPLEX_ELEMENT_MARK to replace them by the element values if (complexElements.size() > 0) { for (int i=0; i<complexElements.size(); i++) { int indexComplexElemCompositesCharsWithMarks = compositesCharsWithMarksNormalized.indexOf(COMPLEX_ELEMENT_MARK); Element complexElem = complexElements.get(i); String complexElementStr = complexElem.toXmlString(); String firstPiece = ""; if (indexComplexElemCompositesCharsWithMarks > 0) { firstPiece = compositesCharsWithMarksNormalized.substring(0, indexComplexElemCompositesCharsWithMarks); compositesCharsWithMarksNormalized = compositesCharsWithMarksNormalized.substring(indexComplexElemCompositesCharsWithMarks); } retString = retString + firstPiece + complexElementStr; compositesCharsWithMarksNormalized = compositesCharsWithMarksNormalized.substring(COMPLEX_ELEMENT_MARK_SIZE); } retString = retString + compositesCharsWithMarksNormalized; // last one must also be added } else { retString = retString + compositesCharsWithMarksNormalized; // last one must also be added } } retString = retString + "</" + name + ">"; } return retString; } private String normalizeWords(String charactersStrDeresolved, String language) throws SAXException { String charactersStr = StringUtilEscapeChars.resolveXmlEntities(charactersStrDeresolved); String retStr = ""; try { MpdlNormalizer mpdlNormalizer = new MpdlNormalizer(normalizeFunctions, language); if (dictMode) { mpdlNormalizer.setNormMode(MpdlNormalizer.DICTIONARY); } else { mpdlNormalizer.setNormMode(MpdlNormalizer.DISPLAY); } MpdlTokenizerAnalyzer tokenAnalyzer = new MpdlTokenizerAnalyzer(mpdlNormalizer, language); ArrayList<Token> wordTokens = tokenAnalyzer.getToken(charactersStr); int endPos = 0; for (int i=0; i < wordTokens.size(); i++) { Token wordToken = wordTokens.get(i); int startPos = wordToken.startOffset(); String beforeStr = charactersStr.substring(endPos, startPos); endPos = wordToken.endOffset(); String displayWordStr = charactersStr.substring(startPos, endPos); String normalizedWord = displayWordStr; if (! dictMode) { normalizedWord = normalize(mpdlNormalizer, displayWordStr); // normalizer in DISPLAY mode normalizedWord = StringUtilEscapeChars.deresolveXmlEntities(normalizedWord); } else { normalizedWord = getLexWord(mpdlNormalizer, displayWordStr); // normalizer in DICTIONARY mode } String beforeStrDeresolved = StringUtilEscapeChars.deresolveXmlEntities(beforeStr); retStr = retStr + beforeStrDeresolved + normalizedWord; } String lastAfterStr = charactersStr.substring(endPos); String lastAfterStrDeresolved = StringUtilEscapeChars.deresolveXmlEntities(lastAfterStr); retStr = retStr + lastAfterStrDeresolved; } catch (ApplicationException e) { throw new SAXException(e); } return retStr; } /** * if word contains "not word delimiting symbol" (e.g. for line break) it is replaced * by a "Blank" so that the Lex normalizer could handle it. Other cases see below. * The Lex normalizer then e.g. gets "præbi- ta" and normalize it to "praebi- ta". * @param mpdlNormalizer Lex normalizer * @param word * @return normalized word * @throws ApplicationException */ private String normalize(MpdlNormalizer mpdlNormalizer, String word) throws ApplicationException { if (word.trim().isEmpty()) return word; String cleanedWord = word; // starting nwd mark and more than one nwd mark are removed before normalization; after normalization they are added again boolean startsWithNWDMark = cleanedWord.startsWith(COMPLEX_ELEMENT_NWD_MARK); if (startsWithNWDMark) cleanedWord = cleanedWord.substring(1); int countNWDMarks = cleanedWord.length() - cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK, "").length(); if (countNWDMarks > 1) cleanedWord = cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK + "+", COMPLEX_ELEMENT_NWD_MARK); // boolean notHyphenPlusNWD = cleanedWord.matches(".*[^-]+" + COMPLEX_ELEMENT_NWD_MARK + "+.*"); // e.g. "praebi ta" // if (notHyphenPlusNWD) // cleanedWord = cleanedWord.replaceAll("([^-]+)" + COMPLEX_ELEMENT_NWD_MARK + "+", "$1-" + COMPLEX_ELEMENT_NWD_MARK); // e.g. "praebi ta" is replaced by "praebi- ta" String inputWord = cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK, " "); String normalizedWordStr = mpdlNormalizer.normalize(inputWord); normalizedWordStr = normalizedWordStr.replaceAll(" ", COMPLEX_ELEMENT_NWD_MARK); // if (notHyphenPlusNWD) // normalizedWordStr = normalizedWordStr.replaceAll("-" + COMPLEX_ELEMENT_NWD_MARK, COMPLEX_ELEMENT_NWD_MARK); // e.g. "praebi- ta" is replaced by "praebi ta" if (countNWDMarks > 1) { String nwdStr = ""; for (int i=0; i<countNWDMarks; i++) nwdStr += COMPLEX_ELEMENT_NWD_MARK; normalizedWordStr = normalizedWordStr.replaceAll(COMPLEX_ELEMENT_NWD_MARK, nwdStr); } if (startsWithNWDMark) normalizedWordStr = COMPLEX_ELEMENT_NWD_MARK + normalizedWordStr; return normalizedWordStr; } private String getLexWord(MpdlNormalizer mpdlNormalizer, String displayWord) throws ApplicationException { String lexWord = null; String wordForm = removeSpecialSymbols(displayWord); wordForm = wordForm.toLowerCase(); wordForm = normalize(mpdlNormalizer, wordForm); // delivers lex entries by help of the morphology component (lex entry of the stem of the normalized word form) LexHandler lexHandler = LexHandler.getInstance(); String lang = mpdlNormalizer.getLanguage(); ArrayList<String> lexEntryKeys = lexHandler.getLexEntryKeys(wordForm, lang, false); String displayWordDeresolved = StringUtilEscapeChars.deresolveXmlEntities(displayWord); if (lexEntryKeys != null) { String lexForms = ""; for (int j=0; j<lexEntryKeys.size(); j++) { String lexEntryKey = lexEntryKeys.get(j); lexForms = lexForms + lexEntryKey + " "; } lexForms = lexForms.substring(0, lexForms.length() - 1); lexWord = "<w lang=\"" + lang + "\"" + " form=\"" + wordForm + "\"" + " lexForms=\"" + lexForms + "\">" + displayWordDeresolved + "</w>"; } else { lexWord = displayWordDeresolved; } return lexWord; } private String removeSpecialSymbols(String inputStr) { String retStr = inputStr.replaceAll(" |\n|\t|-|\u2424|\u2425", ""); return retStr; } } }