McEs, A Hacker Life: Justified text with Pango

Wednesday, May 02, 2007

Justified text with Pango

...is not unsupported anymore.

The first paragraph has positive letter spacing, and then justified. As you see the third line didn't hvae any space chars and so is justified by applying more letter spacing. Now the Latin part has unfairly long words and short lines, but Arabic works much better than I expected.

It's a very simple greedy algorithm. Mathias Hasselmann start writing patches, and hours later and three ~100 line patches, I've got it all working and nicely interacting with letter-spacing too.

Six year old bugzilla entry here.

Labels: pango, screenshot, typography

¶ 9:34 PM

Comments:

Now there is just the hyphenation left ;-)

# posted by

Anonymous : May 02, 2007 10:50 PM

Yeah, thinking about that.

# posted by

behdad : May 02, 2007 11:31 PM

Simple (greedy) algorithms are often the best, but I remember being told that you can apply Bresenham's algorithm to justify text evenly, such as breaking 8px into three gaps as 3+2+3 rather than the unbalanced 3+3+2 that a naive greedy algorithm would deliver. In this case, we would be calculating widths of inter-word spaces, rather than counting contiguous pixels for each row of a rendered line segment (for a line that is wider than it is tall), but the idea is the same.

I don't know if it's worth the effort of coding it up, but I thought you might be interested by the analogy.

# posted by

Anonymous : May 03, 2007 3:51 AM

Thanks for the comment. My implementation gives 3+2+3. It works by:

round(1 * 8./3) - 0 = 3
round(2 * 8./3) - 3 = 2
round(3 * 8./3) - 5 = 3

# posted by

behdad : May 03, 2007 4:30 AM

Rock! You can't imagine how much pain the lack of justification caused Tomas when porting abiword 2.5 to Pango! Very cool (it's too late to get it in AbiWord 2.6 though).

Now if only you can add a proper UCS4 API to pango instead of the current UTF8 only one, that'd be awesome :D (we have to do constant conversions when rendering, as AbiWord is UCS4 only internally for speed). This is one of the reasons why our pango renderer is slower (no hard numbers, sorry) than our own renderer. - uwog

# posted by

Anonymous : May 03, 2007 5:39 AM

Thanks a lot for your amazing work. Fonts in gnome are improving a lot.

# posted by

Nacho : May 03, 2007 6:29 AM

Excellent, many thanks Behdad. I assume proper justification for Arabic (not using word spacing) is in the works ?

# posted by

Anonymous : May 03, 2007 11:48 AM

Totally in the works.

UTF-32 API too, I'll give it a shot.

# posted by

behdad : May 03, 2007 3:47 PM

What about kashida-justified text as imagined by Microsoft?

# posted by

Anonymous : May 03, 2007 4:04 PM

Soon.

# posted by

behdad : May 03, 2007 4:08 PM

Congratulations for making gnome better with your work!

# posted by

Elías Torres : May 03, 2007 7:12 PM

Great. Thanks for your work. It works well for Thai, too, except that I wish character spacing were applied in every line, not just the ones without space. As spaces are rare in Thai texts, when there is one, it's extraordinarily wide. Distributing spaces in characters as well would help reduce it. (screenshot)

# posted by

Thep : May 14, 2007 5:52 AM

What about the TeX line-breaking algorith? What about hyphenation?

# posted by

Anonymous : May 20, 2007 4:36 PM

TeX-style h&j at some point... who knows..

# posted by

behdad : May 22, 2007 12:54 AM

I grabbed the latest pango release today
(1.17.3) but I can't get it work. when
I run pango-view --justify
test-justify.txt I get a hughe window and
the span is not removed. No line breaks
at all. I thought the justification code
is in this release. Whats wrong?

Regards,
yaleo

# posted by

Anonymous : June 21, 2007 8:20 AM

Ah, sorry, I fount out how it works.

Regards,
yaleo

# posted by

Anonymous : June 21, 2007 8:25 AM

You need to set a width too. Try --width=500

# posted by

behdad : June 21, 2007 9:44 AM

About Me

Twitter Updates