pjfanning opened a new pull request, #1535: URL: https://github.com/apache/poi/pull/1535
`DrawSimpleShape.drawShadow` drew shadows of flipped shapes on the wrong side. It added 180 degrees for a vertical flip, which dates from 2011 with no bug reference, and ignored horizontal flips, parent group flips and `rotWithShape`. ## PowerPoint's behaviour I rendered a test deck of flipped and rotated shapes in PowerPoint for the web and measured the shadow offsets from a screenshot. All shapes used `dir` = 45 degrees (down-right on the page). | Case | no flip | flipH | flipV | flipH + flipV | |---|---|---|---|---| | `rotWithShape=0`, any rotation | 45° | 45° | 45° | 45° | | `rotWithShape=1`, no rotation | 45° | 135° | -45° | 225° | | `rotWithShape=1`, rotated 30° | 75° | 165° | -15° | 255° | - With `rotWithShape=0`, the shadow keeps its direction on the page. - With `rotWithShape=1`, the offset goes through the shape's rotation and flips. The schema default is `true`, so an unset value behaves the same way. - A flip on a parent group gave the same results as a flip on the shape. ## Changes - Add `Shadow.isRotateWithShape()`, a default method returning `false`. `XSLFShadow` overrides it from `rotWithShape`. - Add `DrawSimpleShape.getShadowOffset`. It keeps the offset as-is when the shadow rotates with the shape, and otherwise reverts the rotation and flips of the shape and its parent groups. This also changes rotated XSLF shapes that have no flip: with `rotWithShape=1` or unset, their shadow now rotates with the shape. Before, it always kept its page direction. ## Not verified against PowerPoint - **HSLF:** the `.ppt` test deck wouldn't open in PowerPoint for the web, so that file gave no results. HSLF has no `rotWithShape`, so it uses the page direction it already used for rotation, now also for flips. - **Rotated parent groups:** the deck only covered flipped groups. Reverting parent rotation follows from the same rule but wasn't observed. ## Relationship to #1534 This PR doesn't depend on #1534 and can merge first, but flipped shapes won't render correctly until #1534 is in, because trunk's `DrawShape` flips every shape. With #1534's main code applied temporarily, rendering the deck through POI gave the same shadow directions as PowerPoint on all three XSLF slides. ## Tests `TestXSLFShadow` asserts the table above, including flipped groups, and `TestHSLFShadow` asserts the page direction. Both passed locally; I ran only these two classes. I found no related issue on GitHub. ASF Bugzilla's search needs a login, so I couldn't search it directly. 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
