Add support for CSS property: word-break

[qwfy]

Currently, word-break is not supported, it's needed if you want to do fine-grained text layout, especially for CJK text.

[liZe]

Hello!

Strange, word-break is documented as supported, but it’s definitely not. It should be possible to add the feature with the possibilities given by Pango.

[aschmitz]

It seems as though this is a particularly acute problem in tables, as it can't be worked around with overflow-wrap in tables the way it can be in body text:

<style>
html {
  overflow-wrap: break-word;
}
</style>

<table><tr><td>LONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONG</td></tr></table>

<p>LONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONGLONG</p>

I suspect this is probably about to cause some issues for my uses. If you have a few minutes, can you point me towards where in the code I might try to poke to get either overflow-wrap to work in tables or word-break to work generally?

(edit: ignore the space before the last G in the screenshot - that was a mistake on my part)

[Bolik]

Hi, have you any plans about this feature?

[liZe]

It seems as though this is a particularly acute problem in tables

That’s right, it would be really useful.

Hi, have you any plans about this feature?

We have plans to rewrite the text management in the future, mostly to handle right-to-left and bidirectional text. But there’s nothing about that in the milestones yet.

If you have a few minutes, can you point me towards where in the code I might try to poke to get either overflow-wrap to work in tables or word-break to work generally?

As far as I can remember, overflow-wrap should work but we had some problems with Pango’s wrap mode that wasn’t working as expected. You can look at this code, that’s supposed to break lines at characters boundaries when needed:

WeasyPrint/weasyprint/text/line_break.py

Lines 539 to 556 in 22efdc8

	if not minimum and overflow_wrap == 'break-word' and space < 0:
	# Is it really OK to remove hyphenation for word-break ?
	hyphenated = False
	# TODO: Modify code to preserve W3C condition:
	# "Shaping characters are still shaped as if the word were not broken"
	# The way new lines are processed in this function (one by one with no
	# memory of the last) prevents shaping characters (arabic, for
	# instance) from keeping their shape when wrapped on the next line with
	# pango layout. Maybe insert Unicode shaping characters in text?
	layout.set_text(text)
	pango.pango_layout_set_width(
	layout.layout, units_from_double(max_width))
	pango.pango_layout_set_wrap(
	layout.layout, PANGO_WRAP_MODE['WRAP_CHAR'])
	first_line, index = layout.get_first_line()
	resume_index = index or first_line.length
	if resume_index >= len(text.encode('utf-8')):
	resume_index = None

[oshmoun]

Is this still an issue? I tested overflow-wrap: break-word on a couple documents I have where some long URLs would overflow in a column layout, and it worked quite well.

[liZe]

Is this still an issue? I tested overflow-wrap: break-word on a couple documents I have where some long URLs would overflow in a column layout, and it worked quite well.

You’re right, it’s supported now. And if it works "quite well" in columns, we can assume that it works well enough everywhere else 😄.

Thanks a lot!