[evolution-patches] Re: gtkhtml, mail content should be read line by line

[Radek Doulík <rodo ximian com>]

Hi Eric,

your patch doesn't work very well. It assumes that there's only one text
object on each line. That's quite not true, try to read a message with
an inline image. The text before image is not displayed.

I think you may better work directly with the text object rather than
trying to move cursor. The text object of the gtkhtml's document are
always structured like this: HTMLText, HTMLTextSlave,
HTMLTextSlave, ....

The HTMLText object contains the whole text. The slaves are basically
consecutive intervals of text, where there's no more than 1
HTMLTextSlave of the same HTMLText on each line. Calculating line begin
and end offsets thus means finding the right TextSlave and using
posStart and posStart + posLen. The right slave you get with
html_text_get_slave_at_offset.

Hope this helps.

Cheers
Radek

On Mon, 2004-03-29 at 15:48 +0800, Eric Zhao wrote:
> Hi, 
>    Would you please help review this patch? thanks!
> 
>    This is a patch for bug #55250, which is a accessibility related bug.
> 
> The bug's reason is that gail_text_util doesn't know how lines are
> broken in gtkhtml, so gail_text_util_get_text function will treat the
> whole paragraph as a single line. In this patch, we implement the
> function when boundary_type is ATK_TEXT_BOUNDARY_LINE_START or
> ATK_TEXT_BOUNDARY_LINE_END, and leave the rest to gail_text_util.
> 
> Regards,
> Eric