Apache OpenOffice (AOO) Bugzilla – Issue 28231
Import of word document formats text improperly
Last modified: 2013-08-07 14:41:36 UTC
Imported a word document using OOo 680m34 lays out the text in an ugly way. Inter-letter spacing is bad. This works in OOo 1.1.1 both run using fedora core 2 test 2. Issue 28013 already addresses a regression in importing bulleted items from the same doc file.
Created attachment 14694 [details] Original word document that is being imported
Created attachment 14695 [details] relevant screen shot from 1.1.1
Created attachment 14696 [details] screen shot showing improper text spacing
Hi, I'm under WinXP and I can't reproduce your issue. I hope that a Linux guy can reproduce it. You've noticed that doesn't appear the field with the phone number in the 1.1.1?
I rechecked the problem with m36 on linux and the problem is still there. Maybe the differnce between linux and windows is a font rendering problem, or maybe it is a kernel 2.6 issue. And yes, it has already been noted that 2.0 imports more of the document than 1.1.1
Subcomponent changed.
MRU->US: This looks like a font replacement problem of the "Arial" or "Arial Black" font in src680m36. On my Sun JDS Linux (where "Arial Black" and "Arial Narrow" are available), the document looks good.
On RH and Fedora there is no "Arial Black" font available and thus OOo falls back to Helvetica which is an unscalable Bitmap font. Additionally the character spacing is set to condensed which leads to the reported spacing. Turn off condensed (Format/Character: tab page "Position") and the spacing problem is gone. The underlaying problem is the used Bitmap font in favor of an Type1 font which is (also) aliased to Helvetica in fonts.dir file. Refer to issue 23601 on this.
Closing Resolved/WFM.
Answer is a little incomplete. Why is the formatting more readable in 1.1.1 in the same environment? It also shouldn't be able to find Arial Black. Also, the recommended workaround is not very useful when it is someone else's document. Is issue 23601 going to eventually result in picking a ttf instead of a bitmap font when doing font substitution? Seems like a much more reasonable choice.