Am 09.07.2015 um 18:35 schrieb Allison, Timothy B.:
Are the tiff files generated with 1.8.9 valid and really tiff files?
I.e. non empty, and start with "II" or "MM"?
The file extracted with 1.8.9 app is non-empty, starts with II and when double
clicked in windows opens up to show an image of a lightbulb.
The file extracted with 2.0.0 app is empty.
Could it be that jai_imageio is packaged somewhere else in your classpath?
Not sure how jai-imageio would sneak onto my classpath with this:
java -jar pdfbox-app-1.8.9.jar ExtractImages testPDF_childAttachments.pdf
but not with this:
java -jar pdfbox-app-2.0.0-20150707.080851-1479.jar ExtractImages
testPDF_childAttachments.pdf
Sorry, I didn't read your text properly. And I have a good answer too
this time.
In 1.8.9 tiff files were written with write2OutputStream, which uses
TiffWrapper, which creates a TIFF file itself. This was deleted in
https://issues.apache.org/jira/browse/PDFBOX-2653
https://svn.apache.org/viewvc?view=revision&revision=r1660716
(I proposed the deletion)
Tilman
Could it be that jai_imageio is packaged somewhere else in your classpath?
-----Original Message-----
From: Tilman Hausherr [mailto:[email protected]]
Sent: Thursday, July 09, 2015 12:24 PM
To: [email protected]
Subject: Re: migrating Tika to 2.0.0 -- tiff files in app's ExtractImages
Am 09.07.2015 um 18:11 schrieb Allison, Timothy B.:
That error comes when jai_imageio.jar is missing. Standard java can't
write to tiff. And yes, the prebuilt pdfbox-app can't write to tiff,
because we can't bundle jai_imageio.jar due to license issues.
Ok, but to confirm, it isn't the case that 1.8.9 app has jai_imageio.jar
bundled but 2.0.0 doesn't?
Prebuilt-1.8.9 app _does_ write to tiff, but prebuilt-2.0.0 app doesn't.
What's the difference?
Are the tiff files generated with 1.8.9 valid and really tiff files?
I.e. non empty, and start with "II" or "MM"?
Could it be that jai_imageio is packaged somewhere else in your classpath?
Tilman
Am 09.07.2015 um 16:42 schrieb Allison, Timothy B.:
Hi Tilman,
Thank you for your quick response. I'm not sure this is an issue of
dependencies (although thank you for that reminder!).
If I download a prebuilt pdfbox-app-1.8.9 and call ExtractImages, I get
two image files, on jpg and one tiff...both are actual image files. Nothing is
logged to stdout.
If I download a prebuilt nightly build of pdfbox-app-2.0.0 and call
ExtractImages, I get one actual image file for the jpeg but then an empty (zero
byte) tiff file.
The log message from 2.0.0 is:
java -jar
pdfbox-app-2.0.0-20150709.140349-1486.jar ExtractImages
testPDF_childAttachment
s.pdf
Writing image: testPDF_childAttachments-1
Writing image: testPDF_childAttachments-2
Jul 09, 2015 10:37:13 AM org.apache.pdfbox.tools.imageio.ImageIOUtil writeImage
SEVERE: No ImageWriter found for 'tiff' format
Jul 09, 2015 10:37:13 AM org.apache.pdfbox.tools.imageio.ImageIOUtil writeImage
SEVERE: Supported formats: JPG jpg bmp BMP gif GIF WBMP png PNG wbmp jpeg JPEG
That error comes when jai_imageio.jar is missing. Standard java can't
write to tiff. And yes, the prebuilt pdfbox-app can't write to tiff,
because we can't bundle jai_imageio.jar due to license issues.
Tilman
Tilman
Jul 09, 2015 10:37:13 AM org.apache.pdfbox.pdmodel.font.PDSimpleFont toUnicode
WARNING: No Unicode mapping for f_i (31) in font SCZFMD+HelveticaNeueLTStd-Roman
-----Original Message-----
From: Tilman Hausherr [mailto:[email protected]]
Sent: Wednesday, July 08, 2015 2:59 AM
To: [email protected]
Subject: Re: migrating Tika to 2.0.0
Am 08.07.2015 um 04:19 schrieb Allison, Timothy B.:
There are two embedded/inline images (not regular attachments) that are
processed by pdfbox app's ExtractImages.
In 1.8.9, there's a tiff (lightbulb) and a jpeg (flag/fireworks). With trunk,
there is a log warning saying that tiff isn't supported and then an empty tiff
file and a jpeg.
You need to attach jai_imageio.jar to your build.
And also the levigo jbig2 plugin. Like in the 1.8 version.
https://pdfbox.apache.org/1.8/dependencies.html
If it still doesn't work, could you please post the log message?
Tilman
-----Original Message-----
From: Tilman Hausherr [mailto:[email protected]]
Sent: Tuesday, July 07, 2015 3:48 PM
To: [email protected]
Subject: Re: migrating Tika to 2.0.0
Am 07.07.2015 um 21:39 schrieb Allison, Timothy B.:
Thank you, Andreas. I opened PDFBox-2856.
How about tiffs not being handled by ExtractImages...is this expected?
I also noticed that the tiff file is no longer extracted (2.0.0 logger
says tiff not handled, but a tiff is extracted with 1.8.9). Is this expected?
What tiff? When displaying it with Adobe Reader, I see a word file and a
joboptions file.
Tilman
Thank you, again.
Best,
Tim
-----Original Message-----
From: Andreas Lehmkuehler [mailto:[email protected]]
Sent: Tuesday, July 07, 2015 3:08 PM
To: [email protected]
Subject: Re: migrating Tika to 2.0.0
Hi,
Am 07.07.2015 um 18:59 schrieb Allison, Timothy B.:
All,
As part of TIKA-1285, I updated Jeremy Anderson's original patch for
our wrapper for PDFBox 2.0.0 on Tika. I'm having some problems running the
unit tests because at least one of our files [0] is causing hefty resource
utilization, which sends my laptop into paging. The parse does eventually
stop, and content is extracted.
What version of PDFBox are you using, I guess the lastest SNAPSHOT?
I also tried this file outside of Tika and used the straight PDFBox-app
( both ExtractImages and ExtractText), and performance is also far, far slower
when compared with 1.8.9.
I ran some quick tests and I can confirm that 2.0.0 is 4-5 times slower than
1.8.9 when extracting the text from the given pdf.
Many apologies if this issue has already been identified.
AFAIK, it was unknown until now. Please create a JIRA ticket and thanks for
reporting.
I also noticed that the tiff file is no longer extracted (2.0.0 logger
says tiff not handled, but a tiff is extracted with 1.8.9). Is this expected?
Thank you!
Best,
Tim
[0]
https://issues.apache.org/jira/secure/attachment/12743988/testPDF_childAttachments.pdf
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
BR
Andreas
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
