[
https://issues.apache.org/jira/browse/PDFBOX-5522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17612144#comment-17612144
]
Stefan Ziegler edited comment on PDFBOX-5522 at 10/2/22 6:40 PM:
-----------------------------------------------------------------
We process a lot of PDF files created by external programs. There are several
reasons for using an own COSWriter, e.g. for the following:
1) PDF files are not always completely correct and every now and then we have
to intervene to correct them. We can not change the external application.
2) Another point is a recompression of COSStreams. We often also process PDFs
that are not or not optimally compressed. We can then reprocess all streams,
determine the best compression and rewrite them.
3) A third point are extractions, e.g. we can extract all images, including
some hidden one that we can't access through the pages objects.
For all these things a custom COSWriter makes sense here. We also tried to use
COSDOcument#getObjectsByType( COSName type ) to retrieve a list of all objects
of a specific type, but this is not possible because when you load a PDF,
objects are not in the objectPool at this time and the returned list is empty.
If all objects would be in objectPool I think this would also be bad from an
memory point of view.
A very simple COSWriter which add compression to streams would be this:
{code:java}
public class CompressingCosWriter extends COSWriter {
public CompressingCosWriter(File file) throws FileNotFoundException {
super(new BufferedOutputStream(new FileOutputStream(file)));
}
public CompressingCosWriter(OutputStream outputStream) {
super(outputStream);
}
@Override
public Object visitFromStream(COSStream obj) throws IOException {
PDStream stream = new PDStream(obj);
List<COSName> filters = stream.getFilters();
if ((filters == null || filters.size() == 0) && stream.getLength() >
20) {
obj.removeItem(COSName.DECODE_PARMS);
OutputStream out = null;
try {
byte[] bytes = IOUtils.toByteArray(stream.createInputStream());
out = stream.createOutputStream(COSName.FLATE_DECODE);
out.write(bytes);
} catch (IOException e) {
throw new RuntimeException(e);
} finally {
IOUtils.closeQuietly(out);
}
}
return super.visitFromStream(obj);
}
}
{code}
was (Author: JIRAUSER289113):
We process a lot of PDF files created by external programs. There are several
reasons for using an own COSWriter, e.g. for the following:
1) PDF files are not always completely correct and every now and then we have
to intervene to correct them. We can not change the external application.
2) Another point is a recompression of COSStreams. We often also process PDFs
that are not or not optimally compressed. We can then reprocess all streams,
determine the best compression and rewrite them.
3) A third point are extractions, e.g. we can extract all images, including
some hidden one that we can't access through the pages objects.
For all these things a custom COSWriter makes sense here. We also tried to use
COSDOcument#getObjectsByType( COSName type ) to retrieve a list of all objects
of a specific type, but this is not possible because when you load a PDF,
objects are not in the objectPool at this time and the returned list is empty.
If all objects would be in objectPool I think this would also be bad from an
memory point of view.
A very simple COSWriter which add compression to streams would be this:
{code:java}
public class CompressingCosWriter extends COSWriter {
public CompressingCosWriter(File file) throws FileNotFoundException {
super(new BufferedOutputStream(new FileOutputStream(file)));
}
public CompressingCosWriter(OutputStream outputStream) {
super(outputStream);
}
@Override
public Object visitFromStream(COSStream obj) throws IOException {
PDStream stream = new PDStream(obj);
List<COSName> filters = stream.getFilters();
if ((filters == null || filters.size() == 0) && stream.getLength() >
20) {
obj.removeItem(COSName.DECODE_PARMS);
OutputStream out = null;
try {
byte[] bytes = IOUtils.toByteArray(stream.createInputStream());
out = stream.createOutputStream(COSName.FLATE_DECODE);
out.write(bytes);
} catch (IOException e) {
throw new RuntimeException(e);
} finally {
IOUtils.closeQuietly(out);
}
}
return super.visitFromStream(obj);
}
}
{code}
> Add public void save(COSWriter writer) to PDDocument
> ----------------------------------------------------
>
> Key: PDFBOX-5522
> URL: https://issues.apache.org/jira/browse/PDFBOX-5522
> Project: PDFBox
> Issue Type: New Feature
> Components: PDModel
> Affects Versions: 2.0.28, 3.0.0 PDFBox
> Reporter: Stefan Ziegler
> Priority: Minor
> Attachments: PDDocument.java.patch
>
>
> Please add the following method to PDDocument:
>
> {code:java}
> public void save(COSWriter writer){code}
>
> Why?
> This gives us the possibility to use a custom COSWriter when saving the PDF
> file. Inside the custom COSWriter we can add some checks and convert some
> data structures if required.
>
> Patch is attached.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
