Fixed Rect results for a sequenced range #7840
Conversation
| function copyRangeRectProperties( rect, source ) { | ||
| const rangeRects = Rect.getDomRangeRects( source ); | ||
| const combinedRect = rangeRects[ 0 ]; | ||
|
|
||
| for ( let i = 1; i < rangeRects.length; i++ ) { | ||
| combinedRect.right = Math.max( combinedRect.right, rangeRects[ i ].right ); | ||
| combinedRect.left = Math.min( combinedRect.left, rangeRects[ i ].left ); | ||
| combinedRect.bottom = Math.max( combinedRect.bottom, rangeRects[ i ].bottom ); | ||
| combinedRect.top = Math.min( combinedRect.top, rangeRects[ i ].top ); | ||
| } | ||
|
|
||
| combinedRect.width = combinedRect.right - combinedRect.left; | ||
| combinedRect.height = combinedRect.bottom - combinedRect.top; | ||
|
|
||
| copyRectProperties( rect, combinedRect ); | ||
| } |
There was a problem hiding this comment.
In the table package we have a helper function that is very similar (the same purpose but different approach). Maybe we could unify those to one helper function?
There was a problem hiding this comment.
Good idea, added #7858 that will also be closed by this PR.
I had a small problem with the naming, first wanted to make it a Rect#expand() member, but the name didn't feel right. Ended up with static Rect.getBoundingRect() which feels just right 👌
Also I simplified mentioned contextualballoon logic a little bit as the way how createBoundingRect internal logic was implemented in the getBalloonCellPositionData function was a little confusing.
There was a problem hiding this comment.
I guess you mean Rect not Range 😉
…turns a `Rect` instance containing all other rectangles.
…on toolbar position.
Suggested merge commit message (convention)
Fix (utils):
Rectutility returns wrong sizes in case of a sequenced range. Closes #7838.Feature (utils): Introduced the
Rect#getBoundingRect()method that returns aRectinstance containing all other rectangles. Closes #7858.Additional information
Initially I was even tempted to simplify it and use
Range.getBoundingClientRect()to do the computation for us, but then I saw a plethora of issues that this method not always give reliable results: