package com.tom_roush.pdfbox.pdmodel.common;
import java.io.IOException;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.NoSuchElementException;
import java.util.TreeMap;
import com.tom_roush.pdfbox.cos.COSArray;
import com.tom_roush.pdfbox.cos.COSBase;
import com.tom_roush.pdfbox.cos.COSDictionary;
import com.tom_roush.pdfbox.cos.COSInteger;
import com.tom_roush.pdfbox.cos.COSName;
import com.tom_roush.pdfbox.pdmodel.PDDocument;
/**
* Represents the page label dictionary of a document.
*
* @author Igor Podolskiy
*/
public class PDPageLabels implements COSObjectable
{
private Map<Integer, PDPageLabelRange> labels;
private PDDocument doc;
/**
* Creates an empty page label dictionary for the given document.
*
* <p>
* Note that the page label dictionary won't be automatically added to the
* document; you will still need to do it manually (see
* {@link com.tom_roush.pdfbox.pdmodel.PDDocumentCatalog#setPageLabels(PDPageLabels)}.
* </p>
*
* @param document
* The document the page label dictionary is created for.
* @see com.tom_roush.pdfbox.pdmodel.PDDocumentCatalog#setPageLabels(PDPageLabels)
*/
public PDPageLabels(PDDocument document)
{
labels = new TreeMap<Integer, PDPageLabelRange>();
this.doc = document;
PDPageLabelRange defaultRange = new PDPageLabelRange();
defaultRange.setStyle(PDPageLabelRange.STYLE_DECIMAL);
labels.put(0, defaultRange);
}
/**
* Creates an page label dictionary for a document using the information in
* the given COS dictionary.
*
* <p>
* Note that the page label dictionary won't be automatically added to the
* document; you will still need to do it manually (see
* {@link com.tom_roush.pdfbox.pdmodel.PDDocumentCatalog#setPageLabels(PDPageLabels)}.
* </p>
*
* @param document
* The document the page label dictionary is created for.
* @param dict
* an existing page label dictionary
* @see com.tom_roush.pdfbox.pdmodel.PDDocumentCatalog#setPageLabels(PDPageLabels)
* @throws IOException
* If something goes wrong during the number tree conversion.
*/
public PDPageLabels(PDDocument document, COSDictionary dict) throws IOException
{
this(document);
if (dict == null)
{
return;
}
PDNumberTreeNode root = new PDNumberTreeNode(dict, COSDictionary.class);
findLabels(root);
}
private void findLabels(PDNumberTreeNode node) throws IOException
{
if (node.getKids() != null)
{
List<PDNumberTreeNode> kids = node.getKids();
for (PDNumberTreeNode kid : kids)
{
findLabels(kid);
}
}
else if (node.getNumbers() != null)
{
Map<Integer, COSObjectable> numbers = node.getNumbers();
for (Entry<Integer, COSObjectable> i : numbers.entrySet())
{
if(i.getKey() >= 0)
{
labels.put(i.getKey(), new PDPageLabelRange((COSDictionary)i.getValue()));
}
}
}
}
/**
* Returns the number of page label ranges.
*
* <p>
* This will be always >= 1, as the required default entry for the page
* range starting at the first page is added automatically by this
* implementation (see PDF32000-1:2008, p. 375).
* </p>
*
* @return the number of page label ranges.
*/
public int getPageRangeCount()
{
return labels.size();
}
/**
* Returns the page label range starting at the given page, or {@code null}
* if no such range is defined.
*
* @param startPage
* the 0-based page index representing the start page of the page
* range the item is defined for.
* @return the page label range or {@code null} if no label range is defined
* for the given start page.
*/
public PDPageLabelRange getPageLabelRange(int startPage)
{
return labels.get(startPage);
}
/**
* Sets the page label range beginning at the specified start page.
*
* @param startPage
* the 0-based index of the page representing the start of the
* page label range.
* @param item
* the page label item to set.
* @throws IllegalArgumentException if the startPage parameter is < 0.
*/
public void setLabelItem(int startPage, PDPageLabelRange item)
{
if (startPage < 0)
{
throw new IllegalArgumentException("startPage parameter of setLabelItem may not be < 0");
}
labels.put(startPage, item);
}
/**
* {@inheritDoc}
*/
@Override
public COSBase getCOSObject()
{
COSDictionary dict = new COSDictionary();
COSArray arr = new COSArray();
for (Entry<Integer, PDPageLabelRange> i : labels.entrySet())
{
arr.add(COSInteger.get(i.getKey()));
arr.add(i.getValue());
}
dict.setItem(COSName.NUMS, arr);
return dict;
}
/**
* Returns a mapping with computed page labels as keys and corresponding
* 0-based page indices as values. The returned map will contain at most as
* much entries as the document has pages.
*
* <p>
* <strong>NOTE:</strong> If the document contains duplicate page labels,
* the returned map will contain <em>less</em> entries than the document has
* pages. The page index returned in this case is the <em>highest</em> index
* among all pages sharing the same label.
* </p>
*
* @return a mapping from labels to 0-based page indices.
*/
public Map<String, Integer> getPageIndicesByLabels()
{
final Map<String, Integer> labelMap =
new HashMap<String, Integer>(doc.getNumberOfPages());
computeLabels(new LabelHandler()
{
@Override
public void newLabel(int pageIndex, String label)
{
labelMap.put(label, pageIndex);
}
});
return labelMap;
}
/**
* Returns a mapping with 0-based page indices as keys and corresponding
* page labels as values as an array. The array will have exactly as much
* entries as the document has pages.
*
* @return an array mapping from 0-based page indices to labels.
*/
public String[] getLabelsByPageIndices()
{
final String[] map = new String[doc.getNumberOfPages()];
computeLabels(new LabelHandler()
{
@Override
public void newLabel(int pageIndex, String label)
{
if(pageIndex < doc.getNumberOfPages())
{
map[pageIndex] = label;
}
}
});
return map;
}
/**
* Internal interface for the control flow support.
*
* @author Igor Podolskiy
*/
private interface LabelHandler
{
void newLabel(int pageIndex, String label);
}
private void computeLabels(LabelHandler handler)
{
Iterator<Entry<Integer, PDPageLabelRange>> iterator =
labels.entrySet().iterator();
if (!iterator.hasNext())
{
return;
}
int pageIndex = 0;
Entry<Integer, PDPageLabelRange> lastEntry = iterator.next();
while (iterator.hasNext())
{
Entry<Integer, PDPageLabelRange> entry = iterator.next();
int numPages = entry.getKey() - lastEntry.getKey();
LabelGenerator gen = new LabelGenerator(lastEntry.getValue(),
numPages);
while (gen.hasNext())
{
handler.newLabel(pageIndex, gen.next());
pageIndex++;
}
lastEntry = entry;
}
LabelGenerator gen = new LabelGenerator(lastEntry.getValue(),
doc.getNumberOfPages() - lastEntry.getKey());
while (gen.hasNext())
{
handler.newLabel(pageIndex, gen.next());
pageIndex++;
}
}
/**
* Generates the labels in a page range.
*
* @author Igor Podolskiy
*
*/
private static class LabelGenerator implements Iterator<String>
{
private final PDPageLabelRange labelInfo;
private final int numPages;
private int currentPage;
LabelGenerator(PDPageLabelRange label, int pages)
{
this.labelInfo = label;
this.numPages = pages;
this.currentPage = 0;
}
@Override
public boolean hasNext()
{
return currentPage < numPages;
}
@Override
public String next()
{
if (!hasNext())
{
throw new NoSuchElementException();
}
StringBuilder buf = new StringBuilder();
if (labelInfo.getPrefix() != null)
{
String label = labelInfo.getPrefix();
// there may be some labels with some null bytes at the end
// which will lead to an incomplete output, see PDFBOX-1047
while (label.lastIndexOf(0) != -1)
{
label = label.substring(0, label.length()-1);
}
buf.append(label);
}
if (labelInfo.getStyle() != null)
{
buf.append(getNumber(labelInfo.getStart() + currentPage,
labelInfo.getStyle()));
}
currentPage++;
return buf.toString();
}
private String getNumber(int pageIndex, String style)
{
if (PDPageLabelRange.STYLE_DECIMAL.equals(style))
{
return Integer.toString(pageIndex);
}
else if (PDPageLabelRange.STYLE_LETTERS_LOWER.equals(style))
{
return makeLetterLabel(pageIndex);
}
else if (PDPageLabelRange.STYLE_LETTERS_UPPER.equals(style))
{
return makeLetterLabel(pageIndex).toUpperCase();
}
else if (PDPageLabelRange.STYLE_ROMAN_LOWER.equals(style))
{
return makeRomanLabel(pageIndex);
}
else if (PDPageLabelRange.STYLE_ROMAN_UPPER.equals(style))
{
return makeRomanLabel(pageIndex).toUpperCase();
}
else
{
// Fall back to decimals.
return Integer.toString(pageIndex);
}
}
/**
* Lookup table used by the {@link #makeRomanLabel(int)} method.
*/
private static final String[][] ROMANS = new String[][] {
{ "", "i", "ii", "iii", "iv", "v", "vi", "vii", "viii", "ix" },
{ "", "x", "xx", "xxx", "xl", "l", "lx", "lxx", "lxxx", "xc" },
{ "", "c", "cc", "ccc", "cd", "d", "dc", "dcc", "dccc", "cm" }
};
private static String makeRomanLabel(int pageIndex)
{
StringBuilder buf = new StringBuilder();
int power = 0;
while (power < 3 && pageIndex > 0)
{
buf.insert(0, ROMANS[power][pageIndex % 10]);
pageIndex = pageIndex / 10;
power++;
}
// Prepend as many m as there are thousands (which is
// incorrect by the roman numeral rules for numbers > 3999,
// but is unbounded and Adobe Acrobat does it this way).
// This code is somewhat inefficient for really big numbers,
// but those don't occur too often (and the numbers in those cases
// would be incomprehensible even if we and Adobe
// used strict Roman rules).
for (int i = 0; i < pageIndex; i++)
{
buf.insert(0, 'm');
}
return buf.toString();
}
/**
* A..Z, AA..ZZ, AAA..ZZZ ... labeling as described in PDF32000-1:2008,
* Table 159, Page 375.
*/
private static String makeLetterLabel(int num)
{
StringBuilder buf = new StringBuilder();
int numLetters = num / 26 + Integer.signum(num % 26);
int letter = num % 26 + 26 * (1 - Integer.signum(num % 26)) + 64;
for (int i = 0; i < numLetters; i++)
{
buf.appendCodePoint(letter);
}
return buf.toString();
}
@Override
public void remove()
{
// This is a generator, no removing allowed.
throw new UnsupportedOperationException();
}
}
}