19761 PDF conversion error

[chrisjj]

I have PDFs which are fine in Acrobat X, Win7 Explorer preview and https://www.pdf-online.com/osa/validate.aspx , but of which BC says Conversion Error. It would be nice for this to be fixed.

Sample mailed to [email protected] .

[Aaron]

Please email in sample PDFs to [email protected] with a link back to this forum thread for our reference so we can verify what we're having trouble with.

[chrisjj]

I did - at the time of posting 3 hours ago. Please tell me if you didn't receive.

[Aaron]

Thanks. We did get it. This is an example of no actual text data. It's a picture of text, which we cannot convert. You can verify this by using the Adobe File menu -> Save As Text option, which also generates a blank file. Since a blank file can be actually blank or caused by a conversion error, we default to reporting as a conversion error.

[chrisjj]

> Since a blank file can be actually blank or caused by a conversion error, we default to reporting as a conversion error.

Please pass on my request this be fixed. Just as BC can compare a TXT file containing no text, it should be able to compare a PDF file containing no text.

[Aaron]

Unfortunately, not. The issue is that BC3/4 use 3rd party conversion utilities, and users can also put in their own utilities. Some of these utilities, outside of our control, can error out. When they do, they can create empty files as the output. There is no way to know if the conversion program generated an empty file on purpose or through error.

BC2 would always compare the output, and this lead to false comparisons showing equal when the conversion failed. This was fixed in BC3 and not a change we'll roll back.

[Zoë]

To expand on what Aaron posted a bit, in regards specifically to PDFs, I'd argue that a PDF with just images of text is much more common than completely blank PDFs. It's definitely unfortunate that we can't compare the images, but I think for a less technical user, treating it as a conversion error is more useful and less error prone than silently treating it as empty.

[chrisjj]

The issue is that BC3/4 use 3rd party conversion utilities, and users can also put in their own utilities. Some of these utilities, outside of our control, can error out.

I'm using regular OOTB BC. If that's out of your control, then I think you should get it back in your control and fix this error.

treating it as a conversion error is more useful and less error prone than silently treating it as empty.

No-one is suggesting silently treating it as empty. On the contrary, part of the current problem is that BC is being silent about the fact that it is empty. It is instead falsely reporting a conversion error instead.

treating it as a conversion error is more useful

Telling the truth would be more useful.

[Aaron]

There is currently no method to differentiate between a converted blank file and a converted blank file generated in error by a conversion program. If such support is added to the conversion utility, we'll look into reporting between the two states. As is, we need to assume either blank state is an error to prevent false equivalency.

[chrisjj]

There is currently no method to differentiate between a converted blank file and a converted blank file generated in error by a conversion program.

Uh, what "conversion program"?? I'm using BC.

If I wanted to use a manky converter that doesn't properly report errors, I would. I don't.

[Aaron]

PDF comparisons in BC use different conversion methods, depending on the version of BC you are running. BC4 includes a version of PDFtoText.exe in the main install folder, and that's our default support.

If you would like to use a different program, you can also customize this by creating a new File Format and defining any other command line program for the External Conversion. The higher *.pdf format in the File Formats dialog is the one used automatically.

[chrisjj]

BC4 includes a version of PDFtoText.exe in the main install folder, and that's our default support.

If you would like to use a different program

I wouldn't. What I would like is just BC to work properly, and not false represent an empty PDF as a conversion error.

[Aaron]

Chris,

I'll try to explain this one more time:

BC uses conversion applications for binary files. PDF, MS Word, Excel must be converted to be readable as text in BC. We are using an External Conversion process, where a file format can define any command line utility to be used to perform this conversion, given an input binary file and an output text file.

*We are familiar with various command line utilities that express their errors by generating a blank file.*

Since this is a global construct, we then have to interpret an External Conversion's blank file as an error. In BC2, we didn't anticipate this and ignored it as an error state. This resulted in various conversions failing, giving back blank files for different binary files, and we would erroneously show them as equal in BC2. Adding this as an Error state is the best we can do for now. If we're able to enhance our support for specific, internal conversions, like PDF, we may be able to distinguish between a blank file error and blank file conversion or compare the pictures, but this is not currently supported.

In the current release, providing a PDF that is a Picture to our Text Compare results in a conversion error. The Text Compare is only for comparing text data.

[chrisjj]

Adding this as an Error state is the best we can do for now.

Are you saying the PDFtoText.exe you are using does not return an error code?

[Aaron]

PDFtoText does have Error levels. All External Conversions are currently treated equally. Tracking the errorlevels of specific conversions isn't currently supported.