NBSP in Word, part 2

This is a follow-on to my previous post on the subject.  These are my notes as I’m working, so they don’t read very well as text 🙂 .

Another test, on a different file exhibiting the same problem:

In the UI:

  • Removed all the new-style equations (OMath objects)
  • Changed all OpenType fancy features to defaults (in the Character Properties)
  • Changed the proofing language for all text to English

In word/settings.xml:

  • Removed FELayout
  • Removed all w:compat/w:compatSetting except for [@w:name=”compatibilityMode”]
  • Left m:mathPr for now, but I may remove that another time. Edit: I removed that below.

Did `find . -name \*xml | xargs grep -il asia`.  That gave me word/{document,numbering,settings,styles}.xml.

Used XML Notepad 2007 to move tags onto their own lines — just open the file, then save it.

settings.xml: it was w:themeFontLang.  Leaving that in for now.

styles.xml: It’s a bunch of <w:lang/> and <w:rFonts/> (`grep -i asia word/styles.xml |sort|uniq|less` is my friend!).  Leaving them in for now.

numbering.xml: Again, just <w:rFonts/>.  Leaving them for now.

document.xml: There were two lonely <w:rFonts/> tags!  Those I left in at first.

Zipping it

I tried using 7z, both from the context menu and with `7z a -tzip -mm=Deflate -mx=0 ../draft8e.docx *` (7z command-line options reference).  However, Word 2013 didn’t like either.  I used Send To | Compressed Folder, and that was OK.  A problem for another day.

More changes

Result?  Still no luck.  Time to strip the rFonts!

  1. Remove the rFonts referencing asia (/asia/g, in fact) from document.xml.  Result? No luck.
  2.  Remove that pesky m:mathPr from settings.xml.  You know, I wonder…  Result?  No luck.
  3. Remove w:themeFontLang from settings.  In styles, leave w:val, but remove w:eastAsia and w:bidi.  Result: success!!!!

So if you do all of the above, you’ll probably be OK 🙂 .

 

 

Music of the Day

“Moonmadness” by Camel.

Wow!  I had heard of Camel before, but this is the first full album I’ve listened to.  It will be in my collection as soon as I can conveniently arrange it 🙂 .  Solid prog rock in a fantastic style.  Well worth a listen, and I suspect it will stand up to repeated play.

More details at Prog Archives.

Word 2013 non-breaking space width: an end to the insanity!

Pre-Word 2013, a persistent annoyance was that non-breaking spaces (Chr(160)) were fixed width, even in justified text.  Word 2013 changed that.  Now, in justified text, regular spacing and non-breaking spaces are the same width.

Sometimes.

If you’re lucky.

User slasza on Microsoft Answers posted this test case from Word 2013, in which everything should have been right, but the nonbreaking spaces still didn’t show up:

NBSP failure – image by slasza

I ran some test files and took a look at the Word XML, and found the culprit at last!  It is the <w:useFELayout/> tag in the <w:compat> group in word/settings.xml.  If that tag is present in a Word 2013 document, nonbreaking spaces are fixed-width.  If it is absent, nonbreaking spaces are variable-width.

More detailed instructions on how to fix this are coming soon.  But rejoice that at least now you know what the problem is!

Edit 1

I just fixed one legacy file by going on an odyssey through the UI and the XML.  I:

  1. In the UI, turned off all the fancy OpenType features
  2. Set the proofing language to US-English for all text
  3. In word/settings.xml, removed FELayout, everything under w:compat except for w:compatSetting, and I think a few other things under /w:settings.
  4. In word/*.xml, removed just about every mention of the word “asia” (case-insensitive)

And my NBSPs are finally variable-width!

vim: cleaning up mixed indents

Thanks to the wiki for this.  If you run across a file that mixes tabs and spaces (ewww!), set the tab settings the way you want (e.g., ts=4 sts=4 sw=4 et ai) and run

:retab

to convert all the tabs to the right number of spaces.  This takes tab stops into account, which :%s/^I/    /g won’t.

This replaces what I used to do, which was /^I s<Tab><Esc> followed by a whole lot of n.n.n.n.n.n.n.n. … .  One command is much better 🙂 .  I’m not even going to try to count the keystroke savings on my current project (pym, a preprocessor written in Python).

Vim: joining lines having continuation markers

This handy file-renaming script at Perl Monks has source code with red “+” marks marking wrapped lines. I pasted it into Vim and wanted to wrap those lines back to the way they should be. After a bit of fiddling, I got:

:g/^+/execute "norm 0x" | .-1,.j!

The `g/^+/’ finds the lines beginning with a “+”. Then the `execute “norm 0x”‘ deletes the “+” (“0” moves to beginning of line; “x” deletes), “|” marks the next command, and `.-1,.j!’ joins (“j”) the current line (“.”) with the previous line (“.-1”) without whitespace (“!”) added between the lines.

I leave it to you whether I have too much time on my hands.