package eu.transkribus.core.io.formats;
import org.apache.commons.io.FileUtils;
import org.apache.commons.io.FilenameUtils;
import org.apache.commons.lang3.StringUtils;
import org.eclipse.persistence.tools.file.FileUtil;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.w3c.dom.*;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
import com.google.common.io.Files;
import com.itextpdf.text.Image;
import eu.transkribus.core.io.LocalDocReader;
import eu.transkribus.core.model.beans.pagecontent.PcGtsType;
import eu.transkribus.core.util.JaxbUtils;
import eu.transkribus.core.util.PageXmlUtils;
import eu.transkribus.core.util.XmlUtils;
import eu.transkribus.interfaces.types.util.ImageUtils;
import java.awt.Dimension;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.FileFilter;
import java.io.FileInputStream;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.lang.reflect.Array;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.Comparator;
import javax.imageio.ImageIO;
import javax.xml.bind.JAXBException;
import javax.xml.namespace.NamespaceContext;
import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerConfigurationException;
import javax.xml.transform.TransformerException;
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;
import javax.xml.xpath.XPath;
import javax.xml.xpath.XPathConstants;
import javax.xml.xpath.XPathExpressionException;
import javax.xml.xpath.XPathFactory;
public class IngestHTRIntoAbbyyXML {
private static final Logger logger = LoggerFactory.getLogger(IngestHTRIntoAbbyyXML.class);
private static boolean createFinalPageXml = false;
public static void main(String[] args) throws IOException {
if (createFinalPageXml){
String imgDir = args[2];
File imgFileDir = new File(imgDir);
File convertedFileDir = new File(imgDir +"/convertedAbbyy/" );
if(!convertedFileDir.exists()){
logger.error("'convertedAbbyy' directoriy does not exist: " + args[2]);
return;
}
File[] ocrFiles = convertedFileDir.listFiles();
File[] imgFiles = imgFileDir.listFiles(new FileFilter() {
@Override
public boolean accept(File pathname) {
return pathname.isFile();
}
});
String resultDir = imgFileDir.getAbsolutePath()+"/convertedPage/";
new File(resultDir).mkdirs();
for (File abbyyXml : ocrFiles){
if (abbyyXml != null){
BufferedImage img = null;
String imgFn = abbyyXml != null ? imgDir + File.separator + StringUtils.substringBefore(abbyyXml.getName(),".xml") + ".jpg" : "";
logger.debug("imgFn " + imgFn);
File imgFile = new File(imgFn);
if (imgFile.exists()){
try
{
img = ImageIO.read(imgFile.getAbsoluteFile());
File pageOutFile = new File(resultDir+abbyyXml.getName());
if (pageOutFile.exists()){
continue;
}
createPageXml(pageOutFile, true, abbyyXml, true, true, false, imgFile.getName(), new Dimension(img.getWidth(), img.getHeight()));
}
catch (IOException e)
{
e.printStackTrace();
}
}
}
}
}
else{
//Y:\HTR\Wiener_Adressbuch\done\htr_in Y:\HTR\Wiener_Adressbuch\done\ocr
if(args.length != 2){
usage();
return;
}
final String htrDir, ocrDir;
htrDir = args[0];
ocrDir = args[1];
File htrMainDir = new File(htrDir);
File ocrMainDir = new File(ocrDir);
ingestHtrIntoAbbyyForAllFolders(ocrMainDir, htrMainDir);
}
}
private static void ingestHtrIntoAbbyyForAllFolders(File ocrDir, File htrAndImgDir) throws IOException {
if(!ocrDir.exists() || !htrAndImgDir.exists()){
logger.error("One of the start directories does not exist: " + ocrDir + " or " + htrAndImgDir);
usage();
return;
}
if (htrAndImgDir.listFiles().length != ocrDir.listFiles().length){
logger.error("Directories must contain the same number of folders!");
usage();
return;
}
int countAllConverted = 0;
for (File imgFileDir : htrAndImgDir.listFiles()){
File ocrFileDir = new File(ocrDir.getAbsolutePath() + File.separator + imgFileDir.getName() + "/ocr");
logger.debug("ocr folder: " + ocrFileDir.getAbsolutePath());
File[] ocrFiles = ocrFileDir.listFiles();
File htrFileDir = new File(imgFileDir.getAbsolutePath() + "/page");
logger.debug("htr folder: " + htrFileDir.getAbsolutePath());
File[] htrFiles = htrFileDir.listFiles();
logger.debug("img folder: " + imgFileDir.getAbsolutePath());
File[] imgFiles = imgFileDir.listFiles(new FileFilter() {
@Override
public boolean accept(File pathname) {
return pathname.isFile();
}
});
File resultDir = new File(htrAndImgDir.getParentFile().getAbsolutePath() + "/converted" + File.separator + imgFileDir.getName() + File.separator + "ocr");
// if (resultDir.exists()){
// logger.debug(resultDir.getAbsolutePath() + " Result dir exists already - try next one");
// continue;
// }
resultDir.mkdirs();
logger.debug("resultDir folder: " + resultDir.getAbsolutePath());
File sampleDir = new File(htrAndImgDir.getParentFile().getAbsolutePath() + "/samples" + File.separator);
sampleDir.mkdirs();
logger.debug("sampleDir folder: " + sampleDir.getAbsolutePath());
if (htrFiles.length != ocrFiles.length){
logger.error("Directories must contain the same number of files!");
usage();
return;
}
Arrays.sort(htrFiles, (a, b) -> a.getName().compareToIgnoreCase(b.getName()));
Arrays.sort(ocrFiles, (a, b) -> a.getName().compareToIgnoreCase(b.getName()));
Arrays.sort(imgFiles, (a, b) -> a.getName().compareToIgnoreCase(b.getName()));
//
// displayFiles(htrFiles);
// displayFiles(ocrFiles);
// displayFiles(imgFiles);
int nr = 0;
for (File htr : htrFiles){
// String ocrName = findOcrFilename(htr);
// String ocrFn = ocrName != null ? ocrFileDir + File.separator + ocrName + ".xml" : "";
String ocrFn = ocrFileDir + File.separator + FilenameUtils.getBaseName(htr.getName())+ ".xml";
logger.debug("ocrFn " + ocrFn);
File ocrFile = new File(ocrFn);
if (ocrFile.exists()){
// logger.debug("htr file: " + htr.getName());
// logger.debug("ocr file: " + ocrFile.getName());
//
// System.in.read();
File convertedFile = new File(resultDir.getAbsolutePath()+File.separator+htr.getName());
if (convertedFile.exists()){
logger.debug("Already converted (" + countAllConverted++ + ")");
nr++;
continue;
}
//ingestHTRIntoOCR(htr, ocrFiles[nr++]);
File abbyyXml = combineHTRAndOCR(htr, ocrFile, convertedFile);
//take test sample for every 500th page
if (abbyyXml != null && nr % 500 == 400){
String resultPageDir = sampleDir.getAbsolutePath()+"/page/";
new File(resultPageDir).mkdirs();
File pageOutFile = new File(resultPageDir+htr.getName());
BufferedImage img = null;
try
{
img = ImageIO.read(imgFiles[nr].getAbsoluteFile());
FileUtils.copyFileToDirectory(imgFiles[nr], sampleDir);
createPageXml(pageOutFile, true, abbyyXml, true, true, false, imgFiles[nr].getName(), new Dimension(img.getWidth(), img.getHeight()));
}
catch (IOException e)
{
e.printStackTrace();
}
}
nr++;
}
else{
logger.error("No ocr File found :(");
}
}
}
}
private static void startSimpleTestWithFilenames(){
// read from files
String htrFilename= "C:/01_Projekte/READ/Projekte/Lehmann/Konvertierung/page/00000101.xml";
String ocrFilename = "C:/01_Projekte/READ/Projekte/Lehmann/Konvertierung/ocr/3697706.xml";
ingestHTRIntoOCR(new File(htrFilename), new File(ocrFilename));
}
private static String findOcrFilename(File htrFile){
try {
FileInputStream fileIsHTR = new FileInputStream(htrFile);
DocumentBuilderFactory builderFactory = DocumentBuilderFactory.newInstance();
DocumentBuilder builder = builderFactory.newDocumentBuilder();
Document htrDocument = builder.parse(fileIsHTR);
XPath xPath = XPathFactory.newInstance().newXPath();
String expressionMd = "//TranskribusMetadata";
NodeList nodeListHTR = (NodeList) xPath.compile(expressionMd).evaluate(htrDocument, XPathConstants.NODESET);
System.out.println("nodeList length " + nodeListHTR.getLength());
for (int i = 0; i < nodeListHTR.getLength(); i++){
NamedNodeMap htrLineString = nodeListHTR.item(i).getAttributes();
Node pageId = nodeListHTR.item(i).getAttributes().getNamedItem("pageId");
if (pageId != null){
return pageId.getTextContent();
}
}
} catch (XPathExpressionException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (ParserConfigurationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (SAXException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (FileNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
return null;
}
private static void ingestHTRIntoOCR(File htrFile, File ocrFile) {
try {
FileInputStream fileIsHTR = new FileInputStream(htrFile);
FileInputStream fileIsOCR = new FileInputStream(ocrFile);
DocumentBuilderFactory builderFactory = DocumentBuilderFactory.newInstance();
DocumentBuilder builder = builderFactory.newDocumentBuilder();
Document htrDocument = builder.parse(fileIsHTR);
Document ocrDocument = builder.parse(fileIsOCR);
XPath xPath = XPathFactory.newInstance().newXPath();
String expressionHTR = "//TextLine/TextEquiv/Unicode";
String expressionOCR = "//line";
//xPath: get all charParams for current node
String expAllCharParamsOfLine = ".//charParams";
NodeList nodeListOCR = (NodeList) xPath.compile(expressionOCR).evaluate(ocrDocument, XPathConstants.NODESET);
NodeList nodeListHTR = (NodeList) xPath.compile(expressionHTR).evaluate(htrDocument, XPathConstants.NODESET);
int countEqualLines = 0;
int countUnequalLines = 0;
int countEqualLineLengths = 0;
int countUnequalLinesLenths = 0;
// String test = "test";
// for (char c : test.toCharArray()){
// System.out.println(c);
//
// }
//System.out.println("result is " + result);
// // and iterate on links
System.out.println("nodeList length " + nodeListHTR.getLength());
for (int i = 0; i < nodeListHTR.getLength(); i++){
String htrLineString = nodeListHTR.item(i).getTextContent();
// NodeList formatting = nodeListOCR.item(i).getChildNodes();
// for (int j = 0; j<formatting.getLength(); j++){
// //NodeList charParams = formatting.item(j).getChildNodes();
//
// logger.debug("node " + formatting.item(j));
Node ocrLine = (Node) nodeListOCR.item(i);
Node formatting = ocrLine.getLastChild();
//logger.debug(ocrLine.getLastChild().getNodeName());
String left = "", right = "", top = "", bottom = "";
NamedNodeMap ocrLineAttributes = ocrLine.getAttributes();
for (int l = 0; l<ocrLineAttributes.getLength(); l++){
Node currAttr = ocrLineAttributes.item(l);
//logger.debug("node map name " + currAttr.getNodeName());
switch (currAttr.getNodeName()){
case "l": left=currAttr.getNodeValue();
case "r": right=currAttr.getNodeValue();
case "t": top=currAttr.getNodeValue();
case "b": bottom=currAttr.getNodeValue();
default: break;
}
}
int lineLeft = Integer.valueOf(left);
int lineRight = Integer.valueOf(right);
int lineTop = Integer.valueOf(top);
int lineBottom = Integer.valueOf(bottom);
NodeList nlCharParams = (NodeList) xPath.compile(expAllCharParamsOfLine).evaluate((Node) nodeListOCR.item(i), XPathConstants.NODESET);
// logger.debug("ocr length: " + nlCharParams.getLength());
// logger.debug("htr length: " + htrLineString.length());
boolean differentLength = nlCharParams.getLength() != htrLineString.length();
float newCharWidth = ( (Integer.valueOf(right)-Integer.valueOf(left))/htrLineString.length());
String ocrLineString = "";
for (int k = 0; k<nlCharParams.getLength(); k++){
ocrLineString += nlCharParams.item(k).getTextContent();
}
// logger.debug("htr line text: " + htrLineString);
// logger.debug("ocr line text: " + ocrLineString);
if (htrLineString.length() == ocrLineString.length()){
countEqualLineLengths++;
}
else{
countUnequalLinesLenths++;
}
if (htrLineString.equals(ocrLineString)){
// logger.debug("ocr and htr are the same - keep charparams as they are!!");
countEqualLines++;
continue;
}
else{
countUnequalLines++;
//logger.debug("go on");
}
// if (ocrLineString.length() > htrLineString.length()){
// System.in.read();
// }
for (int k = 0; k<nlCharParams.getLength(); k++){
Node charParamInOCR = nlCharParams.item(k);
Character ocrChar = nlCharParams.item(k).getTextContent().charAt(0);
Character htrChar = k<htrLineString.length()? htrLineString.charAt(k) : null;
// NamedNodeMap nodeMap = charParamInOCR.getAttributes();
// for (int l = 0; l<nodeMap.getLength(); l++){
// logger.debug("node map name " + nodeMap.item(l).getNodeName());
// logger.debug("node map " + nodeMap.item(l).getNodeValue());
// }
/*
* cases to be considered
* (1) ocr line shorter than htr -> add charparams for all additional chars
* (2) htr line shorter than ocr -> delete needless chars
* (3) equal size -> only replace text content for unequal chars
* (4) empty char in htr text but not in ocr
*/
//case 2: pointless ocr charParams need to be deleted
if(htrChar == null){
// logger.debug("remove child: " + charParamInOCR.getTextContent());
//text node of this element needs to be removed as well - otherwise you get empty lines in xml
Node prevElem = charParamInOCR.getPreviousSibling();
if (prevElem != null &&
prevElem .getNodeType() == Node.TEXT_NODE &&
prevElem .getNodeValue().trim().length() == 0) {
charParamInOCR.getParentNode().removeChild(prevElem);
}
//System.in.read();
charParamInOCR.getParentNode().removeChild(charParamInOCR);
continue;
}
//if chars differ: replace ocr with htr (includes case 3, valid for case 1 and 2 too)
if (ocrChar != htrChar){
//if we deal with different length of htr and ocr we newly calculate the coordinates of the bounding box of each character
if (differentLength){
int l = (int) (lineLeft+newCharWidth*k);
int r = (int) (lineLeft+newCharWidth*(k+1));
Node leftCoord = charParamInOCR.getAttributes().getNamedItem("l");
leftCoord.setNodeValue(Integer.toString(l));
Node rightCoord = charParamInOCR.getAttributes().getNamedItem("r");
rightCoord.setNodeValue(Integer.toString(r));
}
//empty space - insert and add wordStart = 1;
if (Character.isSpaceChar(htrChar)){//.equals("\u0020")){
if (charParamInOCR.getNextSibling() != null && charParamInOCR.getNextSibling().getNextSibling() != null){
Node wordStart = charParamInOCR.getNextSibling().getNextSibling().getAttributes().getNamedItem("wordStart");
if (wordStart != null){
wordStart.setNodeValue("1");
}
}
}
//set wordStart=0
else if (Character.isSpaceChar(ocrChar)){
if (charParamInOCR.getNextSibling() != null && charParamInOCR.getNextSibling().getNextSibling() != null){
logger.debug(charParamInOCR.getNextSibling().getNextSibling().getNodeName());
Node wordStart = charParamInOCR.getNextSibling().getNextSibling().getAttributes().getNamedItem("wordStart");
if (wordStart != null){
wordStart.setNodeValue("0");
}
}
}
deleteAllObsoleteAtttributes(charParamInOCR);
//Element additionalCharParam = ocrDocument.createElement("charParams");
//test if this is sufficient
charParamInOCR.setTextContent(htrChar.toString());
//charParamInOCR.getParentNode().replaceChild(charParamInOCR, charParamInOCR);
}
//nlCharParams.item(k).getParentNode().replaceChild(newChild, oldChild)
// logger.debug("node name: " + nlCharParams.item(k).getNodeName());
// if (nlCharParams.item(k).getNodeName().equals("charParams"))
// logger.debug("charParams: " + nlCharParams.item(k).getTextContent());
// if (k<htrLineString.length())
// logger.debug("vs. HTR char " + htrLineString.charAt(k));
}
//case 1:
int remaining = htrLineString.length()-ocrLineString.length();
if (remaining > 0){
//logger.debug("append node to line " + ocrLine.getTextContent());
for (int j = ocrLineString.length(); j<htrLineString.length(); j++){
/*
* calculate the left and right coordinates from lineWidth/numberOfHTRChars
* top and bottom are taken from line
*/
int l = (int) (lineLeft+newCharWidth*j);
int r = (int) (lineLeft+newCharWidth*(j+1));
Character htrChar = htrLineString.charAt(j);
Element additionalCharParam = ocrDocument.createElement("charParams");
additionalCharParam.setAttribute("l", Integer.toString(l));
additionalCharParam.setAttribute("t", top);
additionalCharParam.setAttribute("r", Integer.toString(r));
additionalCharParam.setAttribute("b", bottom);
additionalCharParam.appendChild(ocrDocument.createTextNode(htrChar.toString()));
Node node = formatting.appendChild(additionalCharParam);
// logger.debug("node added " + node.getTextContent());
//System.in.read();
}
}
//}
// for (char c : currLineString.toCharArray()){
//
// }
//logger.debug("ocr line length: " + nodeListOCR.item(i).getTextContent());
//logger.debug("child nodes of line: " + nodeListOCR.item(i).getFirstChild().getTextContent());
//logger.debug("ocr line length: " + nodeListOCR.item(i).getTextContent().length());
}
TransformerFactory tff = TransformerFactory.newInstance();
Transformer transformer = tff.newTransformer();
DOMSource xmlSource = new DOMSource(ocrDocument);
StreamResult outputTarget = new StreamResult(new File(ocrFile.getParentFile().getAbsolutePath() + File.separatorChar + htrFile.getName()));
transformer.transform(xmlSource, outputTarget);
logger.debug("Statistics of file " + htrFile.getAbsolutePath());
logger.debug("countEqualLines: "+countEqualLines);
logger.debug("countUnequalLines: " +countUnequalLines);
logger.debug("countEqualLineLengths: " +countEqualLineLengths);
logger.debug("countUnequalLinesLenths: " + countUnequalLinesLenths);
// System.out.println("nodeListOCR length " + nodeListOCR.getLength());
// for (int i = 0; i < nodeListOCR.getLength(); i++){
// logger.debug("node value: " + nodeListOCR.item(i).getTextContent().length());
// }
// Document doc1 = expandedData1.getOwnerDocument();
// // insert the nodes
// Node expandedData2 = (Node) xpath.evaluate("//expandedData", ocr, NODE);
// expandedData1.getParentNode()
// .replaceChild(doc1.adoptNode(expandedData2), expandedData1);
// // print results
// TransformerFactory.newInstance()
// .newTransformer()
// .transform(new DOMSource(doc1), new StreamResult(System.out));
} catch (XPathExpressionException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (ParserConfigurationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (SAXException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (FileNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (TransformerConfigurationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (TransformerException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
/*
* experiment: combine OCR and HTR: the longest string wins
* seems to bring best results
* TODO: calculate error rates and compare!
*/
private static File combineHTRAndOCR(File htrFile, File ocrFile, File convertedFile) {
try {
FileInputStream fileIsHTR = new FileInputStream(htrFile);
FileInputStream fileIsOCR = new FileInputStream(ocrFile);
DocumentBuilderFactory builderFactory = DocumentBuilderFactory.newInstance();
DocumentBuilder builder = builderFactory.newDocumentBuilder();
Document htrDocument = builder.parse(fileIsHTR);
Document ocrDocument = builder.parse(fileIsOCR);
XPath xPath = XPathFactory.newInstance().newXPath();
String expressionHTR = "//TextLine/TextEquiv/Unicode";
String expressionOCR = "//block[@blockType='Text']//line";
String expressionOCRTables = "//block[@blockType='Table']//line";
//xPath: get all charParams for current node
String expAllCharParamsOfLine = ".//charParams";
NodeList nodeListOCR = (NodeList) xPath.compile(expressionOCR).evaluate(ocrDocument, XPathConstants.NODESET);
NodeList nodeListOCRTables = (NodeList) xPath.compile(expressionOCRTables).evaluate(ocrDocument, XPathConstants.NODESET);
NodeList nodeListHTR = (NodeList) xPath.compile(expressionHTR).evaluate(htrDocument, XPathConstants.NODESET);
int countEqualLines = 0;
int countUnequalLines = 0;
int countEqualLineLengths = 0;
int countUnequalLinesLenths = 0;
// String test = "test";
// for (char c : test.toCharArray()){
// System.out.println(c);
//
// }
int tableNr = 0;
//System.out.println("result is " + result);
// // and iterate on links
//System.out.println("nodeList length " + nodeListHTR.getLength());
for (int i = 0; i < nodeListHTR.getLength(); i++){
String htrLineString = nodeListHTR.item(i).getTextContent();
// NodeList formatting = nodeListOCR.item(i).getChildNodes();
// for (int j = 0; j<formatting.getLength(); j++){
// //NodeList charParams = formatting.item(j).getChildNodes();
//
// logger.debug("node " + formatting.item(j));
Node ocrLine = (Node) nodeListOCR.item(i);
Node formatting = null;
if (ocrLine != null){
//logger.debug("ocr line not null");
formatting = ocrLine.getLastChild();
}
else{
logger.debug("ocr line is null - try table lines");
ocrLine = (Node) nodeListOCRTables.item(tableNr++);
}
if (ocrLine == null){
logger.debug("ocr line is null");
logger.debug("htr file name: " + htrFile.getName());
logger.debug("ocr file name: " + ocrFile.getName());
}
//logger.debug(ocrLine.getLastChild().getNodeName());
String left = "", right = "", top = "", bottom = "";
NamedNodeMap ocrLineAttributes = ocrLine.getAttributes();
for (int l = 0; l<ocrLineAttributes.getLength(); l++){
Node currAttr = ocrLineAttributes.item(l);
//logger.debug("node map name " + currAttr.getNodeName());
switch (currAttr.getNodeName()){
case "l": left=currAttr.getNodeValue();break;
case "r": right=currAttr.getNodeValue();break;
case "t": top=currAttr.getNodeValue();break;
case "b": bottom=currAttr.getNodeValue();break;
default: break;
}
}
int lineLeft = Integer.valueOf(left);
int lineRight = Integer.valueOf(right);
int lineTop = Integer.valueOf(top);
int lineBottom = Integer.valueOf(bottom);
NodeList nlCharParams = (NodeList) xPath.compile(expAllCharParamsOfLine).evaluate(ocrLine, XPathConstants.NODESET);
// logger.debug("ocr length: " + nlCharParams.getLength());
// logger.debug("htr length: " + htrLineString.length());
String ocrLineString = "";
String penaltyString = "";
boolean ocrIsWorse = false;
for (int k = 0; k<nlCharParams.getLength(); k++){
ocrLineString += nlCharParams.item(k).getTextContent();
// Node penalty = nlCharParams.item(k).getAttributes().getNamedItem("wordPenalty");
// Node charConf = nlCharParams.item(k).getAttributes().getNamedItem("charConfidence");
//
// if (penalty != null){
// penaltyString = penalty.getTextContent();
// }
//
// if (penalty != null && Integer.valueOf(penalty.getTextContent())>20){
// logger.debug("word Penalty: " + penalty.getTextContent());
//
// ocrIsWorse = true;
// }
// if (charConf != null && charConf.getTextContent().equals("-1")){
// logger.debug("char Confidence: " + charConf.getTextContent());
// ocrIsWorse = true;
// }
}
// logger.debug("htr line text: " + htrLineString);
// logger.debug("ocr line text: " + ocrLineString);
if (htrLineString.length() == ocrLineString.length()){
countEqualLineLengths++;
}
else{
countUnequalLinesLenths++;
}
if (htrLineString.equals(ocrLineString)){
// logger.debug("ocr and htr are the same - keep charparams as they are!!");
countEqualLines++;
continue;
}
countUnequalLines++;
/*
* this HTR correction with better OCR does not bring better results for an already good HTR
*/
//if (ocrLineString.replaceAll("\\s","").length() > htrLineString.replaceAll("\\s","").length()){
// if (ocrLineString.length() > htrLineString.length()
// && (ocrLineString.length() - htrLineString.length()>2)){
// logger.debug("OCR String longer then HTR string - OCR could be better");
// logger.debug("is ocr better? " + !ocrIsWorse);
// //System.in.read();
// //continue;
// }
//
// if (ocrLineString.length() > htrLineString.length()
// && (ocrLineString.length() - htrLineString
// .length()>2) && !ocrIsWorse){
// logger.debug("OCR String longer then HTR string - OCR wins");
// logger.debug("penaltyString " + penaltyString);
// //System.in.read();
// continue;
// }
// if(wordPenalty){
// System.in.read();
// }
boolean differentLength = nlCharParams.getLength() != htrLineString.length();
int nrOfChars = (htrLineString.length()>0 ? htrLineString.length() : ocrLineString.length());
double newCharWidth = Math.ceil(( (Integer.valueOf(right)-Integer.valueOf(left))/nrOfChars));
//logger.debug("new CharWidth is " + newCharWidth);
// if (ocrLineString.length() > htrLineString.length()){
// System.in.read();
// }
for (int k = 0; k<nlCharParams.getLength(); k++){
Node charParamInOCR = nlCharParams.item(k);
Character ocrChar = nlCharParams.item(k).getTextContent().charAt(0);
Character htrChar = k<htrLineString.length()? htrLineString.charAt(k) : null;
// NamedNodeMap nodeMap = charParamInOCR.getAttributes();
// for (int l = 0; l<nodeMap.getLength(); l++){
// logger.debug("node map name " + nodeMap.item(l).getNodeName());
// logger.debug("node map " + nodeMap.item(l).getNodeValue());
// }
/*
* cases to be considered
* (1) ocr line shorter than htr -> add charparams for all additional chars
* (2) htr line shorter than ocr -> delete needless chars
* (3) equal size -> only replace text content for unequal chars
* (4) empty char in htr text but not in ocr
*/
//case 2: pointless ocr charParams need to be deleted
if(htrChar == null){
// logger.debug("remove child: " + charParamInOCR.getTextContent());
//text node of this element needs to be removed as well - otherwise you get empty lines in xml
Node prevElem = charParamInOCR.getPreviousSibling();
if (prevElem != null &&
prevElem .getNodeType() == Node.TEXT_NODE &&
prevElem .getNodeValue().trim().length() == 0) {
charParamInOCR.getParentNode().removeChild(prevElem);
}
//System.in.read();
charParamInOCR.getParentNode().removeChild(charParamInOCR);
continue;
}
//if we deal with different length of htr and ocr we newly calculate the coordinates of the bounding box of each character
if (differentLength){
int l = (int) (lineLeft+newCharWidth*k);
int r = (int) (lineLeft+newCharWidth*(k+1));
Node leftCoord = charParamInOCR.getAttributes().getNamedItem("l");
leftCoord.setNodeValue(Integer.toString(l));
Node rightCoord = charParamInOCR.getAttributes().getNamedItem("r");
rightCoord.setNodeValue(Integer.toString(r));
}
//if chars differ: replace ocr with htr (includes case 3, valid for case 1 and 2 too)
if (ocrChar != htrChar){
//empty space - insert and add wordStart = 1;
if (Character.isSpaceChar(htrChar)){//.equals("\u0020")){
if (charParamInOCR.getNextSibling() != null && charParamInOCR.getNextSibling().getNextSibling() != null){
Node wordStart = charParamInOCR.getNextSibling().getNextSibling().getAttributes().getNamedItem("wordStart");
if (wordStart != null){
wordStart.setNodeValue("1");
}
else{
Element currNode = (Element) charParamInOCR.getNextSibling().getNextSibling();
currNode.setAttribute("wordStart", "1");
}
}
}
//set wordStart=0
else if (Character.isSpaceChar(ocrChar)){
if (charParamInOCR.getNextSibling() != null && charParamInOCR.getNextSibling().getNextSibling() != null){
//logger.debug(charParamInOCR.getNextSibling().getNextSibling().getNodeName());
Node wordStart = charParamInOCR.getNextSibling().getNextSibling().getAttributes().getNamedItem("wordStart");
if (wordStart != null){
wordStart.setNodeValue("0");
}
else{
Element currNode = (Element) charParamInOCR.getNextSibling().getNextSibling();
currNode.setAttribute("wordStart", "0");
}
}
}
deleteAllObsoleteAtttributes(charParamInOCR);
//Element additionalCharParam = ocrDocument.createElement("charParams");
//test if this is sufficient
charParamInOCR.setTextContent(htrChar.toString());
//charParamInOCR.getParentNode().replaceChild(charParamInOCR, charParamInOCR);
}
//nlCharParams.item(k).getParentNode().replaceChild(newChild, oldChild)
// logger.debug("node name: " + nlCharParams.item(k).getNodeName());
// if (nlCharParams.item(k).getNodeName().equals("charParams"))
// logger.debug("charParams: " + nlCharParams.item(k).getTextContent());
// if (k<htrLineString.length())
// logger.debug("vs. HTR char " + htrLineString.charAt(k));
}
//case 1:
int remaining = htrLineString.length()-ocrLineString.length();
if (remaining > 0){
//logger.debug("append node to line " + ocrLine.getTextContent());
for (int j = ocrLineString.length(); j<htrLineString.length(); j++){
/*
* calculate the left and right coordinates from lineWidth/numberOfHTRChars
* top and bottom are taken from line
*/
int l = (int) (lineLeft+newCharWidth*j);
int r = (int) (lineLeft+newCharWidth*(j+1));
Character htrChar = htrLineString.charAt(j);
Element additionalCharParam = ocrDocument.createElement("charParams");
additionalCharParam.setAttribute("l", Integer.toString(l));
additionalCharParam.setAttribute("t", top);
additionalCharParam.setAttribute("r", Integer.toString(r));
additionalCharParam.setAttribute("b", bottom);
additionalCharParam.appendChild(ocrDocument.createTextNode(htrChar.toString()));
if (formatting != null){
formatting.appendChild(additionalCharParam);
//logger.debug("node added " + node.getTextContent());
//System.in.read();
}
else{
ocrLine.appendChild(additionalCharParam);
}
}
}
//}
// for (char c : currLineString.toCharArray()){
//
// }
//logger.debug("ocr line length: " + nodeListOCR.item(i).getTextContent());
//logger.debug("child nodes of line: " + nodeListOCR.item(i).getFirstChild().getTextContent());
//logger.debug("ocr line length: " + nodeListOCR.item(i).getTextContent().length());
}
TransformerFactory tff = TransformerFactory.newInstance();
Transformer transformer = tff.newTransformer();
DOMSource xmlSource = new DOMSource(ocrDocument);
//File convertedAbbyyXml = new File(htrFile.getParentFile().getParentFile().getAbsolutePath() + File.separatorChar +"ocr" + File.separatorChar + htrFile.getName());
StreamResult outputTarget = new StreamResult(convertedFile);
transformer.transform(xmlSource, outputTarget);
logger.debug("Statistics of file " + ocrFile.getAbsolutePath() );
logger.debug("countEqualLines: "+countEqualLines);
logger.debug("countUnequalLines: " +countUnequalLines);
logger.debug("countEqualLineLengths: " +countEqualLineLengths);
logger.debug("countUnequalLinesLenths: " + countUnequalLinesLenths);
//System.in.read();
return convertedFile;
// System.out.println("nodeListOCR length " + nodeListOCR.getLength());
// for (int i = 0; i < nodeListOCR.getLength(); i++){
// logger.debug("node value: " + nodeListOCR.item(i).getTextContent().length());
// }
// Document doc1 = expandedData1.getOwnerDocument();
// // insert the nodes
// Node expandedData2 = (Node) xpath.evaluate("//expandedData", ocr, NODE);
// expandedData1.getParentNode()
// .replaceChild(doc1.adoptNode(expandedData2), expandedData1);
// // print results
// TransformerFactory.newInstance()
// .newTransformer()
// .transform(new DOMSource(doc1), new StreamResult(System.out));
} catch (XPathExpressionException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (ParserConfigurationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (SAXException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (FileNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (TransformerConfigurationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (TransformerException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
return null;
}
private static void deleteAllObsoleteAtttributes(Node charParamInOCR) throws IOException
{
//logger.debug("deleteAllObsoleteAtttributes " + charParamInOCR.getTextContent());
NamedNodeMap charParamAttributes = charParamInOCR.getAttributes();
ArrayList<String> attrName =new ArrayList<String>();
/*
* copy the attribute names - charParamAttributes shrinks with removing the attributes
* and with the attribute names we can find and remove the attribute
*/
for (int l = 0; l<charParamAttributes.getLength(); l++){
attrName.add(charParamAttributes.item(l).getNodeName());
}
for (int l = 0; l<attrName.size(); l++){
// logger.debug("l " + l);
// logger.debug("curr Attr: " + attrName.get(l));
if (!attrName.get(l).matches("l|r|b|t|wordStart")){
//logger.debug("delete: " + charParamAttributes.getNamedItem(attrName.get(l)));
charParamAttributes.removeNamedItem(attrName.get(l));
}
}
//System.in.read();
}
private static void compareVersions() {
StringUtils.getLevenshteinDistance("fly", "ant");
}
/**
* Method will create a PAGE XML from the given source files at pageOutFile.
* if no supported source file exists (abbyy/alto/txt), then a skeleton will be created if possible.
* <br/><br/>
* This method must NEVER return null. Many mechanisms in Transkribus
* depend on this method reliably creating a file.
*
* @param pageOutFile
* @param doOverwrite
* @param abbyyXml
* @param altoXml
* @param txtFile
* @param preserveOcrFontFamily
* @param preserveOcrTxtStyles
* @param replaceBadChars
* @param imgFile
* @param dim
* @return
* @throws FileNotFoundException
* @throws IOException
*/
protected static File createPageXml(File pageOutFile, boolean doOverwrite, File abbyyXml, boolean preserveOcrFontFamily, boolean preserveOcrTxtStyles,
boolean replaceBadChars, final String imgFileName, Dimension dim)
throws FileNotFoundException, IOException {
if(pageOutFile == null) {
throw new IllegalArgumentException("PAGE XML output File is null.");
}
if(pageOutFile.exists() && !doOverwrite) {
throw new IOException("PAGE XML already exists at: " + pageOutFile.getAbsolutePath());
}
if(StringUtils.isEmpty(imgFileName)) {
throw new IllegalArgumentException("Image filename must not be empty");
}
PcGtsType pc = null;
if(abbyyXml != null){
//try find Abbyy XML
pc = createPageFromAbbyy(imgFileName, abbyyXml, preserveOcrTxtStyles, preserveOcrFontFamily, replaceBadChars);
}
//from here we need the dimension of the image
if(dim == null) {
//set (0,0) here in order to make the following work
dim = new Dimension();
}
//if still null, there is no suitable file for this page yet => create one
if (pc == null) {
logger.warn("No Transcript XML found for img: " + FilenameUtils.getBaseName(imgFileName));
logger.info("Creating empty PageXml.");
pc = PageXmlUtils.createEmptyPcGtsType(imgFileName, dim);
}
//create the file
try{
JaxbUtils.marshalToFile(pc, pageOutFile);
} catch (JAXBException je) {
throw new IOException("Could not create PageXml on disk!", je);
}
return pageOutFile;
}
private static PcGtsType createPageFromAbbyy(final String imgFileName, File abbyyXml, boolean preserveOcrTxtStyles,
boolean preserveOcrFontFamily, boolean replaceBadChars) throws IOException {
try{
XmlFormat xmlFormat = XmlUtils.getXmlFormat(abbyyXml);
if(xmlFormat.equals(XmlFormat.ABBYY_10)){
logger.info(abbyyXml.getAbsolutePath() + ": Transforming Finereader10/11 XML to PAGE XML.");
PcGtsType pc = PageXmlUtils.createPcGtsTypeFromAbbyy(
abbyyXml, imgFileName,
preserveOcrTxtStyles, preserveOcrFontFamily,
replaceBadChars
);
return pc;
}
throw new IOException("Not a valid Finereader10/11 XML file.");
} catch(IOException | TransformerException ioe){
logger.error(ioe.getMessage(), ioe);
throw new IOException("Could not migrate file: " + abbyyXml.getAbsolutePath(), ioe);
} catch (ParserConfigurationException | SAXException xmle) {
logger.error(xmle.getMessage(), xmle);
throw new IOException("Could not transform XML file!", xmle);
} catch (JAXBException je) {
/* TODO This exception is only thrown when the pageXML is unmarshalled
* for inserting the image filename which is not included in the abbyy xml! */
logger.error(je.getMessage(), je);
throw new IOException("Transformation output is not a valid page XML!", je);
}
}
private static void usage() {
System.out.println("Use: java -jar jarFileName htrDirectoryName ocrDirectoryName\n"
+ "folders contain htr and ocr results and must contain the same number of files!");
return;
}
public static void displayFiles(File[] files) {
for (File file : files) {
System.out.printf("File: %-20s Last Modified:" + "\n", file.getName());
}
}
}
