Examples with PDDocument com.tom_roush.pdfbox.pdmodel.PDDocument used on opensource projects

Example 36 with PDDocument

use of com.tom_roush.pdfbox.pdmodel.PDDocument in project PdfBox-Android by TomRoush.

the class PDFMergerUtilityTest method testMergeBogusStructParents2.

/**
 * PDFBOX-4429: merge into destination that has /StructParent(s) entries in the source file but
 * no structure tree.
 *
 * @throws IOException
 */
@Test
public void testMergeBogusStructParents2() throws IOException {
    File inputPdf = TestResourceGenerator.downloadTestResource(TARGETPDFDIR, "PDFBOX-4408.pdf", "https://issues.apache.org/jira/secure/attachment/12952086/form.pdf");
    assumeTrue(inputPdf.exists());
    PDFMergerUtility pdfMergerUtility = new PDFMergerUtility();
    PDDocument src = PDDocument.load(inputPdf);
    PDDocument dst = PDDocument.load(inputPdf);
    src.getDocumentCatalog().setStructureTreeRoot(null);
    src.getPage(0).setStructParents(9999);
    src.getPage(0).getAnnotations().get(0).setStructParent(9998);
    pdfMergerUtility.appendDocument(dst, src);
    checkWithNumberTree(dst);
    checkForPageOrphans(dst);
    src.close();
    dst.close();
}

Example 37 with PDDocument

use of com.tom_roush.pdfbox.pdmodel.PDDocument in project PdfBox-Android by TomRoush.

the class PDFMergerUtilityTest method testStructureTreeMergeIDTree.

/**
 * PDFBOX-4416: Test merging of /IDTree
 * <br>
 * PDFBOX-4009: test merging to empty destination
 *
 * @throws IOException
 */
@Test
public void testStructureTreeMergeIDTree() throws IOException {
    PDFMergerUtility pdfMergerUtility = new PDFMergerUtility();
    PDDocument src = PDDocument.load(testContext.getAssets().open(SRCDIR + "/" + "PDFBOX-4417-001031.pdf"));
    PDDocument dst = PDDocument.load(testContext.getAssets().open(SRCDIR + "/" + "PDFBOX-4417-054080.pdf"));
    PDNameTreeNode<PDStructureElement> srcIDTree = src.getDocumentCatalog().getStructureTreeRoot().getIDTree();
    Map<String, PDStructureElement> srcIDTreeMap = PDFMergerUtility.getIDTreeAsMap(srcIDTree);
    PDNameTreeNode<PDStructureElement> dstIDTree = dst.getDocumentCatalog().getStructureTreeRoot().getIDTree();
    Map<String, PDStructureElement> dstIDTreeMap = PDFMergerUtility.getIDTreeAsMap(dstIDTree);
    int expectedTotal = srcIDTreeMap.size() + dstIDTreeMap.size();
    assertEquals(192, expectedTotal);
    // PDFBOX-4009, test that empty dest doc still merges structure tree
    // (empty dest doc is used in command line app)
    PDDocument emptyDest = new PDDocument();
    pdfMergerUtility.appendDocument(emptyDest, src);
    src.close();
    src = emptyDest;
    assertEquals(4, src.getDocumentCatalog().getStructureTreeRoot().getParentTreeNextKey());
    pdfMergerUtility.appendDocument(dst, src);
    src.close();
    dst.save(new File(TARGETTESTDIR, "PDFBOX-4416-IDTree-merged.pdf"));
    dst.close();
    dst = PDDocument.load(new File(TARGETTESTDIR, "PDFBOX-4416-IDTree-merged.pdf"));
    checkWithNumberTree(dst);
    checkForPageOrphans(dst);
    dstIDTree = dst.getDocumentCatalog().getStructureTreeRoot().getIDTree();
    dstIDTreeMap = PDFMergerUtility.getIDTreeAsMap(dstIDTree);
    assertEquals(expectedTotal, dstIDTreeMap.size());
    dst.close();
    checkStructTreeRootCount(new File(TARGETTESTDIR, "PDFBOX-4416-IDTree-merged.pdf"));
}

Example 38 with PDDocument

use of com.tom_roush.pdfbox.pdmodel.PDDocument in project PdfBox-Android by TomRoush.

the class PDFMergerUtilityTest method testStructureTreeMerge2.

/**
 * PDFBOX-3999: check that no streams are kept from the source document by the destination
 * document, despite orphan annotations remaining in the structure tree.
 *
 * @throws IOException
 */
@Test
public void testStructureTreeMerge2() throws IOException {
    File inputPdf = TestResourceGenerator.downloadTestResource(TARGETPDFDIR, "PDFBOX-3999-GeneralForbearance.pdf", "https://issues.apache.org/jira/secure/attachment/12896905/GeneralForbearance.pdf");
    assumeTrue(inputPdf.exists());
    PDFMergerUtility pdfMergerUtility = new PDFMergerUtility();
    PDDocument doc = PDDocument.load(inputPdf);
    doc.getDocumentCatalog().getAcroForm().flatten();
    doc.save(new File(TARGETTESTDIR, "PDFBOX-3999-GeneralForbearance-flattened.pdf"));
    ElementCounter elementCounter = new ElementCounter();
    elementCounter.walk(doc.getDocumentCatalog().getStructureTreeRoot().getK());
    int singleCnt = elementCounter.cnt;
    int singleSetSize = elementCounter.set.size();
    assertEquals(134, singleCnt);
    assertEquals(134, singleSetSize);
    doc.close();
    PDDocument src = PDDocument.load(new File(TARGETTESTDIR, "PDFBOX-3999-GeneralForbearance-flattened.pdf"));
    PDDocument dst = PDDocument.load(new File(TARGETTESTDIR, "PDFBOX-3999-GeneralForbearance-flattened.pdf"));
    pdfMergerUtility.appendDocument(dst, src);
    // before solving PDFBOX-3999, the close() below brought
    // IOException: COSStream has been closed and cannot be read.
    src.close();
    dst.save(new File(TARGETTESTDIR, "PDFBOX-3999-GeneralForbearance-flattened-merged.pdf"));
    dst.close();
    doc = PDDocument.load(new File(TARGETTESTDIR, "PDFBOX-3999-GeneralForbearance-flattened-merged.pdf"));
    checkForPageOrphans(doc);
    // Assume that the merged tree has double element count
    elementCounter = new ElementCounter();
    elementCounter.walk(doc.getDocumentCatalog().getStructureTreeRoot().getK());
    assertEquals(singleCnt * 2, elementCounter.cnt);
    assertEquals(singleSetSize * 2, elementCounter.set.size());
    doc.close();
}

Example 39 with PDDocument

use of com.tom_roush.pdfbox.pdmodel.PDDocument in project PdfBox-Android by TomRoush.

the class PDFMergerUtilityTest method testMergeBogusStructParents1.

/**
 * PDFBOX-4429: merge into destination that has /StructParent(s) entries in the destination file
 * but no structure tree.
 *
 * @throws IOException
 */
@Test
public void testMergeBogusStructParents1() throws IOException {
    File inputPdf = TestResourceGenerator.downloadTestResource(TARGETPDFDIR, "PDFBOX-4408.pdf", "https://issues.apache.org/jira/secure/attachment/12952086/form.pdf");
    assumeTrue(inputPdf.exists());
    PDFMergerUtility pdfMergerUtility = new PDFMergerUtility();
    PDDocument src = PDDocument.load(inputPdf);
    PDDocument dst = PDDocument.load(inputPdf);
    dst.getDocumentCatalog().setStructureTreeRoot(null);
    dst.getPage(0).setStructParents(9999);
    dst.getPage(0).getAnnotations().get(0).setStructParent(9998);
    pdfMergerUtility.appendDocument(dst, src);
    checkWithNumberTree(dst);
    checkForPageOrphans(dst);
    src.close();
    dst.close();
}

Example 40 with PDDocument

use of com.tom_roush.pdfbox.pdmodel.PDDocument in project PdfBox-Android by TomRoush.

the class PDFMergerUtilityTest method testMissingParentTreeNextKey.

/**
 * PDFBOX-4009: Test that ParentTreeNextKey is recalculated correctly.
 */
@Test
public void testMissingParentTreeNextKey() throws IOException {
    PDFMergerUtility pdfMergerUtility = new PDFMergerUtility();
    File inputPdf = TestResourceGenerator.downloadTestResource(TARGETPDFDIR, "PDFBOX-4418-000314.pdf", "https://issues.apache.org/jira/secure/attachment/12953423/000314.pdf");
    assumeTrue(inputPdf.exists());
    PDDocument src = PDDocument.load(inputPdf);
    PDDocument dst = PDDocument.load(inputPdf);
    // existing numbers are 321..327; ParentTreeNextKey is 408.
    // After deletion, it is recalculated in the merge 328.
    // That value is added to all numbers of the destination,
    // so the new numbers should be 321+328..327+328, i.e. 649..655,
    // and this ParentTreeNextKey is 656 at the end.
    dst.getDocumentCatalog().getStructureTreeRoot().getCOSObject().removeItem(COSName.PARENT_TREE_NEXT_KEY);
    pdfMergerUtility.appendDocument(dst, src);
    src.close();
    dst.save(new File(TARGETTESTDIR, "PDFBOX-4418-000314-merged.pdf"));
    dst.close();
    dst = PDDocument.load(new File(TARGETTESTDIR, "PDFBOX-4418-000314-merged.pdf"));
    assertEquals(656, dst.getDocumentCatalog().getStructureTreeRoot().getParentTreeNextKey());
    dst.close();
}