Hi,
I encountered issues with pdf with thin lines disappearing when rendered by tabula web client. As a result, a vertical line disappears sometimes and two columns get merged. And it is quite random through several pdfs.
I wonder if it could be an aliasing problem when you render the pdf to png, so that increase the definition would solve it. Right now it is hard coded, and I understand increase dpi parameter, would increase the workload for the detection algorithm right after. Maybe should it be an option.
Sadly I cannot publish the faulty pdf ( bank account full of private informations ).
Thank you for your great tool!
|
image = Utils.pageConvertToImage(page.getPDDoc(), pdfPage, 144, ImageType.GRAY); |
and also line 113
Hi,
I encountered issues with pdf with thin lines disappearing when rendered by tabula web client. As a result, a vertical line disappears sometimes and two columns get merged. And it is quite random through several pdfs.
I wonder if it could be an aliasing problem when you render the pdf to png, so that increase the definition would solve it. Right now it is hard coded, and I understand increase dpi parameter, would increase the workload for the detection algorithm right after. Maybe should it be an option.
Sadly I cannot publish the faulty pdf ( bank account full of private informations ).
Thank you for your great tool!
tabula-java/src/main/java/technology/tabula/detectors/NurminenDetectionAlgorithm.java
Line 101 in 8bfa3ad
and also line 113