/** Serializes the given (non-text) element. */
protected String serializeNonTextElement(PdfElement element) {
if (element == null) {
return null;
}
List<String> fields = new ArrayList<>();
// Add feature name.
PdfFeature feature = element.getFeature();
fields.add(feature != null ? feature.getFieldName() : "");
// Add bounding box.
Rectangle rect = element.getRectangle();
fields.add(rect != null ? String.valueOf(rect.getMinX()) : "");
fields.add(rect != null ? String.valueOf(rect.getMinY()) : "");
fields.add(rect != null ? String.valueOf(rect.getMaxX()) : "");
fields.add(rect != null ? String.valueOf(rect.getMaxY()) : "");
// Add color.
PdfColor color = element.getColor();
fields.add(color != null ? color.getId() : "");
// Add role.
PdfRole role = element.getRole();
fields.add(role != null ? role.name : "");
return CollectionUtils.join(fields, "\t");
}
/** Serializes the given elements with respect to the given roles. */
protected List<String> serializeElementsWithRespectToRoles(List<? extends PdfElement> elements) {
if (elements == null) {
return null;
}
List<String> lines = new ArrayList<>();
for (PdfElement element : elements) {
// Serialize the element only if its role is included in the given roles.
if (getRolesToSerialize().contains(element.getRole())) {
String serialized = serializeElement(element);
if (serialized != null && !serialized.isEmpty()) {
lines.add(serialized);
}
}
}
return lines;
}
/** Serializes the given page. */
protected List<String> serializePage(PdfPage page) {
if (page == null) {
return null;
}
List<String> lines = new ArrayList<>();
// Obtain the features to serialize.
List<PdfFeature> features = getFeaturesToSerialize();
if (features != null) {
for (PdfFeature f : features) {
/**
* This feature was added for David. He needs the paragraphs with the associated lines.
* Remove it if it not needed anymore.
*/
if (f == PdfFeature.paragraphs_with_lines) {
List<PdfTextParagraph> paragraphs = page.getParagraphs();
for (PdfElement para : paragraphs) {
lines.add(serializeElement(para));
for (PdfTextLine line : para.getTextLines()) {
lines.add(serializeElement(line));
}
}
continue;
}
// Obtain the elements by feature.
List<? extends PdfElement> elements = page.getElementsByFeature(f);
List<String> serialized = serializeElementsWithRespectToRoles(elements);
if (serialized != null && !serialized.isEmpty()) {
lines.addAll(serialized);
}
}
}
return lines;
}
/** Serializes the given element. */
protected String serializeElement(PdfElement element) {
if (element == null) {
return null;
}
// TODO: Get rid of ignore method. Is still needed for dehyphenation.
if (element.ignore()) {
return null;
}
if (element instanceof PdfTextElement) {
return serializeTextElement((PdfTextElement) element);
} else {
return serializeNonTextElement(element);
}
}
