Bug in html to docx conversion

[mdolginin]

After update to pandoc 1.19 I can't open result docx files in MS Word (version 15.28) because of error (see the screenshot, some text in russian).

If I use pandoc 1.18 there is no such problem.

[jgm]

Are you using a custom reference.docx? (Or do you perhaps have one in ~/.pandoc?) +++ Maksim [Dec 05 16 03:43 ]:

…

After update to pandoc 1.19 I can't open result docx files in MS Word (version 15.28) because of error (see the screenshot, some text in russian). [1]2016-12-05 15 37 58 If I use pandoc 1.18 there is no such problem. — You are receiving this because you are subscribed to this thread. Reply to this email directly, [2]view it on GitHub, or [3]mute the thread. References 1. https://cloud.githubusercontent.com/assets/9402510/20883699/ef612a9a-bb00-11e6-9d2c-dd026cb6e65f.png 2. #3280 3. https://github.com/notifications/unsubscribe-auth/AAAL5BMwP14JVV2ReA1xdVZ3YLoy-Tupks5rE_j3gaJpZM4LEIe-

[mdolginin]

Yep. I'm using a custom reference file

[jgm]

Did you try pandoc 1.18 with the very same reference.docx file? Your best bet with reference.docx is to start with the default one and modify it minimally, just by changing the styles. If there's any added content, it can cause this sort of corruption. +++ Maksim [Dec 05 16 03:52 ]:

…

Yep. I'm using a custom reference file — You are receiving this because you commented. Reply to this email directly, [1]view it on GitHub, or [2]mute the thread. References 1. #3280 (comment) 2. https://github.com/notifications/unsubscribe-auth/AAAL5HkkoozA933Uj5hkKIwA-gFWcClLks5rE_rjgaJpZM4LEIe-

[mdolginin]

There is same error when I try default reference.docx in pandoc 1.19. When I remove 1.19 and install 1.18 error is gone.

[mdolginin]

But if I try to open result docx in LibreOffice it works fine

[jgm]

Can you send or link to the reference.docx you're using? +++ Maksim [Dec 05 16 22:32 ]:

…

There is same error when I try default reference.docx in pandoc 1.19. When I remove 1.19 and install 1.18 error is gone. — You are receiving this because you commented. Reply to this email directly, [1]view it on GitHub, or [2]mute the thread. References 1. #3280 (comment) 2. https://github.com/notifications/unsubscribe-auth/AAAL5OvNNAVznwVloUvnnqna2ubWAgb1ks5rFQGAgaJpZM4LEIe-

[mdolginin]

It is a default reference that store in pandoc dir by this command pandoc -f html -t doc -o myfile.docx myfile.txt (I don’t know how send it to you) However, I created a custom reference docx file. See it in attach [ref.docx](https://github.com/jgm/pandoc/files/633176/ref.docx) Maksim [email protected]

…

6 дек. 2016 г., в 11:14, John MacFarlane ***@***.***> написал(а): Can you send or link to the reference.docx you're using? +++ Maksim [Dec 05 16 22:32 ]: > There is same error when I try default reference.docx in pandoc 1.19. > When I remove 1.19 and install 1.18 error is gone. > > — > You are receiving this because you commented. > Reply to this email directly, [1]view it on GitHub, or [2]mute the > thread. > >References > > 1. #3280 (comment) > 2. https://github.com/notifications/unsubscribe-auth/AAAL5OvNNAVznwVloUvnnqna2ubWAgb1ks5rFQGAgaJpZM4LEIe- — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3280 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AI94joYO4l3w8k97xyUYRPXokV4Yzsufks5rFQtlgaJpZM4LEIe->.

[jgm]

+++ Maksim [Dec 05 16 23:23 ]:

It is a default reference that store in pandoc dir by this command pandoc -f html -t doc -o myfile.docx myfile.txt (I don’t know how send it to you)

Don't create your default reference.docx that way. It will very likely not work if your document contains anything complex. Instead, I suggest pandoc --print-default-data-file=reference.docx > myfile.docx or maybe echo "" | pandoc -o myfile.docx

[mdolginin]

here is my default reference (after execute pandoc --print-default-data-file=reference.docx > myfile.docx)

myfile.docx

[jgm]

Odd, I had no trouble creating a docx using pandoc 1.19 with your original ref.docx. If you want to repeat what I did, it was

pandoc --reference-docx ref.docx -o my.docx MANUAL.txt

with the pandoc MANUAL.txt. Can you try that and see if it works? To get a copy of the manual you can use pandoc --print-default-data-file MANUAL.txt > MANUAL.txt.

I've got Word for Mac 2011, by the way. Perhaps this is sensitive to Word version?

[jgm]

Also, can you do the conversion you were trying before with the default reference.docx you just generated?

[mdolginin]

with the pandoc MANUAL.txt. Can you try that and see if it works? To get a copy of the manual you can use pandoc --print-default-data-file MANUAL.txt > MANUAL.txt.

It works fine with MANUAL.txt.

But if I try to do this with html file with table (in attachment) I get an error. This HTML doc was generated from asciidoctor file. (In pandoc 1.18 this works fine)

dostup_test.html.zip

P.S. thank you for dealing with this issue. I appreciate it
Pandoc is very usefull tool 👍

[jgm]

OK. It's really not about the reference.docx; you can get the failure with just

pandoc dostup_test.html -o dostup_test.docx

If one does pandoc -t native dostup_test.html it's clear what (one) problem is. Pandoc's HTML reader is producing a bunch of spurious empty table rows.
I conjecture this has something to do with commit 5222572; that's the only thing that affected table parsing in the HTML reader between 1.18 and 1.19.

[jgm]

Oh, I see the problem. I have addEmpties where I should have map addEmpties.

[jgm]

Several mistakes in that code now fixed!
Thanks for reporting this.

[mdolginin]

Thanks man