Pdf-lib: TypeError: Cannot read property 'clone' of undefined
Hi,
When trying to copy a specific PDF, i have the error : TypeError: Cannot read property 'clone' of undefined
Here is a not working example :
https://runkit.com/gautiert/5d81e279596dfc0015e16a7c
Probably related to #186
GautierT
All 11 comments
@GautierT yes I think it's the same issue
mlecoq
on 20 Sep 2019
GautierT
on 20 Sep 2019
I will have a look tomorrow
mlecoq
on 22 Sep 2019
It's funny, pdf is still readable but barcodes are missing
mlecoq
on 23 Sep 2019
I have tried several use cases : merging this pdf with another one, adding content on it.... Barcodes are always missing. The generated pdf is slightly different from the original one, for example version is updated : PDF-1.4 to PDF-1.7
mlecoq
on 23 Sep 2019
I think there is a second issue in the library
mlecoq
on 23 Sep 2019
Barcodes were added with Hummusjs
GautierT
on 23 Sep 2019
This does appear to be a separate issue. I verified that the barcodes are still missing after merging #199.
Hopding
on 25 Sep 2019
I was able to fix the problem in #200. It will go out in the next pdf-lib release.
Incase, you're interested, the problem was due to a missing newline. The PDF states that:
The last line of the file shall contain only the end-of-file marker, %%EOF.
This seems pretty clear cut. But it becomes ambiguous when you start dealing with incremental update sections (the file you shared here has one of these):
Each trailer shall be terminated by its own end-of-file (%%EOF) marker.
But it doesn't really state whether the previous %%EOF marker should be followed by a newline before the update section's content begins. All examples in the spec do have a newline. And every other updated PDF I've come across does as well.
However, the PDF you shared here does not follow this convention:
285 | 17201^M
286 | %%EOF28 0 obj^M
287 | <<^M
Notice line 286. That's what caused the issue. pdf-lib's parser was implicitly expecting a newline after the %%EOF. And when it didn't find one, that caused it to enter error recovery mode and skip object 28.
Gotta love the PDF spec's ambiguity and lack of clarity 馃檮. Anyhow, it's fixed now 馃憤
Hopding
on 25 Sep 2019
Thanks a lot @Hopding !!
Bravo for this awesome lib ! 馃憦
GautierT
on 25 Sep 2019
Version 1.1.1 is now published. It contains the fix for this issue. The full release notes are available here.
You can install this new version with npm:
npm install [email protected]
It's also available on unpkg:
Hopding
on 26 Sep 2019
Related issues
dhollenbeck
路
4Comments
zimt28
路
5Comments
zimt28
路
5Comments
lepidotteri
路
3Comments
DanielJackson-Oslo
路
3Comments
Most helpful comment
I was able to fix the problem in #200. It will go out in the next pdf-lib release.
Incase, you're interested, the problem was due to a missing newline. The PDF states that:
This seems pretty clear cut. But it becomes ambiguous when you start dealing with incremental update sections (the file you shared here has one of these):
But it doesn't really state whether the previous %%EOF marker should be followed by a newline before the update section's content begins. All examples in the spec do have a newline. And every other updated PDF I've come across does as well.
However, the PDF you shared here does not follow this convention:
Notice line
286. That's what caused the issue.pdf-lib's parser was implicitly expecting a newline after the %%EOF. And when it didn't find one, that caused it to enter error recovery mode and skip object 28.Gotta love the PDF spec's ambiguity and lack of clarity 馃檮. Anyhow, it's fixed now 馃憤