[
https://issues.apache.org/jira/browse/PDFBOX-4597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16890124#comment-16890124
]
Michael Klink commented on PDFBOX-4597:
---------------------------------------
{quote}[~tilman] (to add to the mystery, results for non quadrant rotations are
undefined).
{quote}
Well, an attempt is made to treat transformations that dominantly are scaled
rotations (scaling in x and y directions may differ) as the underlying
rotation. This intends to select the multiple of 90° nearest to the actual
direction as long as the result is not mirrored or too badly skewed. Otherwise
there is a default of 0°.
Thus, the results are not _undefined_. For some borderline values, though, the
result might not really be intuitive...
IMO a better implementation would have been to calculate the actual text
direction angle (i.e. apply the matrix to (1,0,1) and calculate the atan),
select the nearest multiple of 90°; then calculate the determinant; if it is
negative, add 180° (mod 360°).
But now changing it might break some applications.
> TextPosition.getX() vs getXDirAdj()
> -----------------------------------
>
> Key: PDFBOX-4597
> URL: https://issues.apache.org/jira/browse/PDFBOX-4597
> Project: PDFBox
> Issue Type: Improvement
> Components: Documentation, Text extraction
> Affects Versions: 2.0.15
> Reporter: Gili
> Priority: Major
> Attachments: rot90.pdf
>
>
> Please explain the difference between "page rotation adjusted coordinates"
> and "text direction adjusted coordinates" as referenced by the TextPosition
> class. You can add an explanation (ideally a visual one) to the class or
> method Javadoc.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
