577850 - A gradient applied to SVG Text element should render as if the objectBoundingBox spans the entire Text content

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Sebastian Markbåge]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/533.4 (KHTML, like Gecko) Chrome/5.0.375.99 Safari/533.4
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:2.0b1) Gecko/20100630 Firefox/4.0b1 (.NET CLR 3.5.30729)

Mozilla doesn't comply with this quote from SVG 1.1 specification (Second Edition): "It is possible to apply a gradient, pattern, clipping path, mask or filter to text. When one of these facilities is applied to text and keyword 'objectBoundingBox' is used (see Object bounding box units) to specify a graphical effect relative to the "object bounding box", then the object bounding box units are computed relative to the entire ‘text’ element in all cases, even when different effects are applied to different ‘tspan’ elements within the same ‘text’ element."

Reproducible: Always

Steps to Reproduce:
Create a text-element with a fill attribute referring an gradient. The referred gradient should have the attribute gradientUnits="objectBoundingBox" (default).

The text-element should have several textPath or tspan children with some offset.
Actual Results:  
The gradient is now treated as if each child (textPath/tspan) have it's own bounding box. It's reapplied for each child.

Expected Results:  
The gradient should be applied as if it's a single bounding box covers all children within entire text-element.

Comment 1

[Robert Longson [:longsonr]]

Can you use the attachment link above to attach a testcase please?

Comment 2

[Sebastian Markbåge]

Attachment: Test case

Here's a test case that shows the gradient repeating for each text segment. Comparing it with WebKit browsers should make the issue apparent.

Comment 3

[Robert Longson [:longsonr]]

Would you mind attaching a screenshot from a WebKit browser?

Comment 4

[Robert Longson [:longsonr]]

The comment 0 quote is from here: http://www.w3.org/TR/SVG/text.html, this text is also present in the 1st edition.

Comment 5

[Sebastian Markbåge]

Attachment: Expected result (from a WebKit browser)

Comment 6

[Robert Longson [:longsonr]]

Attachment: patch with reftest

Comment 7

[Robert Longson [:longsonr]]

Comment on attachment 461779 [details] [diff] [review]
patch with reftest

The patch isn't quite right. While it works for text, tspan and textPath it won't work for glyphs as nsSVGGlyphFrame isn't an nsSVGTextContainerFrame.

Comment 8

[Robert Longson [:longsonr]]

Comment on attachment 461779 [details] [diff] [review]
patch with reftest

comment 7 is wrong, it does work for glyphs as the aFrame->GetContent()->IsNodeOfType(nsINode::eTEXT) line moves us to the nsSVGGlyphFrame parent which is a nsSVGTextContainerFrame.

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 461779 [details] [diff] [review]
patch with reftest

Nice

Comment 10

[Robert Longson [:longsonr]]

Comment on attachment 461779 [details] [diff] [review]
patch with reftest

Simple patch with test. Should be low risk.

Comment 11

[Robert Longson [:longsonr]]

pushed http://hg.mozilla.org/mozilla-central/rev/81c119fb86c7

Comment 12

[Sebastian Markbåge]

In Beta 4, the original test case still fails on the lower textPath samples.

Comment 13

[Robert Longson [:longsonr]]

Please raise a new bug if there are still issues with this functionality.