/**
* Takes the raw page lines represented as one continuous line and sorts the text by the y access
* of the word bounds. The words are then sliced into separate lines base on y changes. And
* finally each newly sorted line is sorted once more by each words x coordinate.
*/
public void sortAndFormatText() {
ArrayList<LineText> visiblePageLines = new ArrayList<LineText>(pageLines);
// create new array for storing the sorted lines
ArrayList<LineText> sortedPageLines = sortLinesVertically(visiblePageLines);
// try and insert the option words on existing lines
if (sortedPageLines.size() == 0) {
sortedPageLines = getVisiblePageLines(true);
} else {
insertOptionalLines(sortedPageLines);
}
// sort again
sortedPageLines = sortLinesVertically(sortedPageLines);
// do a rough check for duplicate strings that are sometimes generated
// by Chrystal Reports. Enable with
// -Dorg.icepdf.core.views.page.text.trim.duplicates=true
if (checkForDuplicates) {
for (final LineText lineText : sortedPageLines) {
final List<WordText> words = lineText.getWords();
if (words.size() > 0) {
final List<WordText> trimmedWords = new ArrayList<WordText>();
final Set<String> refs = new HashSet<String>();
for (final WordText wordText : words) {
// use regular rectangle so get a little rounding.
final String key = wordText.getText() + wordText.getBounds().getBounds();
if (refs.add(key)) {
trimmedWords.add(wordText);
}
}
lineText.setWords(trimmedWords);
}
}
}
// sort each line by x coordinate.
if (sortedPageLines.size() > 0) {
for (LineText lineText : sortedPageLines) {
Collections.sort(lineText.getWords(), new WordPositionComparator());
}
}
// recalculate the line bounds.
if (sortedPageLines.size() > 0) {
for (LineText lineText : sortedPageLines) {
lineText.getBounds();
}
}
// sort the lines
if (sortedPageLines.size() > 0 && !preserveColumns) {
Collections.sort(sortedPageLines, new LinePositionComparator());
}
// assign back the sorted lines.
this.sortedPageLines = sortedPageLines;
}
public String toString() {
StringBuilder extractedText = new StringBuilder();
for (LineText lineText : pageLines) {
for (WordText wordText : lineText.getWords()) {
extractedText.append(wordText.getText());
}
extractedText.append('\n');
}
return extractedText.toString();
}
/**
* Utility method to normalize text created in a Xform content stream and is only called from the
* contentParser when parsing 'Do' token.
*
* @param transform do matrix transform
*/
public void applyXObjectTransform(AffineTransform transform) {
for (LineText lineText : pageLines) {
lineText.clearBounds();
for (WordText wordText : lineText.getWords()) {
wordText.clearBounds();
for (GlyphText glyph : wordText.getGlyphs()) {
glyph.normalizeToUserSpace(transform, null);
}
}
}
}
/**
* Sorts the given pageLines vertically (y coordinate) in page space. .
*
* @param pageLines page lines to sort, not directly sorted, new array is created for sorted data.
* @return new array of sorted pages lines
*/
private ArrayList<LineText> sortLinesVertically(ArrayList<LineText> pageLines) {
ArrayList<LineText> sortedPageLines = new ArrayList<LineText>(64);
// move over all
for (LineText pageLine : pageLines) {
// all page words will be on one line
java.util.List<WordText> words = pageLine.getWords();
if (words != null && words.size() > 0) {
// break the words into lines on every change of y
double lastY = Math.round(words.get(0).getTextExtractionBounds().y);
int start = 0, end = 0;
double currentY, diff;
for (WordText word : words) {
currentY = Math.round(word.getTextExtractionBounds().getY());
// little bit of tolerance for detecting a line, basically anything that is
// > then half the current word height / 2 will be marked as a break.
// this works well enough sub and super script and inconsistencies
// on table base text.
diff = Math.abs(currentY - lastY);
if (diff != 0 && diff > word.getTextExtractionBounds().getHeight() / 2) {
LineText lineText = new LineText();
lineText.addAll(words.subList(start, end));
sortedPageLines.add(lineText);
start = end;
}
end++;
lastY = currentY;
}
if (start < end) {
LineText lineText = new LineText();
lineText.addAll(words.subList(start, end));
sortedPageLines.add(lineText);
}
}
}
return sortedPageLines;
}