[poppler] Annot borders

[Leonard Rosenthol]

Sounds like good changes to me!

Leonard

On 11/26/13 3:11 PM, "Carlos Garcia Campos" <carlosgc at gnome.org> wrote:

>
>I've noticed some small issues in the way we handle the border of
>annots compared to what the spec says and what acroread does. Attached
>are 4 patches:
>
>[PATCH 1/4] annots: Remove unused AnnotBorderType from AnnotBorder
>
>This is simply a cleanup, we have a base class AnnotBorder that allows
>to create instances of it with the unknown type. It doesn't make much
>sense and we are not using it at all, so I've removed the
>AnnotBorderType, and made the constructor protected.
>
>[PATCH 2/4] annots: Add helper function Annot::setLineStyleForBorder
>
>The code to set the line width and dash was duplicated in several draw()
>methods, so I've moved it to a helper function. This patch also changes
>the way we are using it. We were always ignoring the border when the
>border with was 0, but it doesn't seem to be always correct. The PDF
>spec says that in both Border and BS entries when the width is 0, no
>border should be drawn at all. But some annotations like lines,
>geometry, polygons, etc. don't use the border entry to actually draw a
>border, but to set the line with and dash for the stroke operations. For
>example, we were not drawing lines for annots with no border or with a
>border width = 0, but acroread does, because in this case the border
>entry is used to set the line with and dash pattern, and the PDF spec
>also says that a line width of 0 should be drawn as the thinnest line that
>can be rendered at device resolution: 1 device pixel wide.
>So, for FreeText annotations where we actually draw a border, we only
>call Annot::setLineStyleForBorder if we have a border and the width is
>greater than 0, and in all other cases we always call it when we have a
>border. Of course, this only affects annotations not having an AP entry.
>
>[PATCH 3/4] annots: Use a default border for annots that can have a BS
>entry
>
>According to the PDF spec if neither the Border nor the BS entry is
>present, the border shall be drawn as a solid line with a width of 1
>point. But again, acroread seems to only apply this rule for annotations
>that can have a BS entry. This patch moves the parsing of the BS entry
>From the base Annot class to the specific annot classes that can have a
>BS entry, and it creates a default border object when neither Border nor
>Bs is present. It also removes the border passed to Gfx::drawAnnot() in
>AnnotFileAttachment::draw and AnnotSound::draw because acroread ignores
>the Border entry also for annots that can't have a BS entry.
>This ensures that we always draw line, geometry, polygon, etc, even if
>there's no border specified.
>
>[PATCH 4/4] annots: Make Annot::setBorder receive an AnnotBorder object
>
>Currently we can only set AnnotBorderArray objects to annots. This might
>have no effect if the annots already has a BS entry, for example,
>because the BS takes precedence over Border. This patches changes
>Annot::setBorder to receive an AnnotBorder object, so that you can
>either pass an AnnotBorderArray or an  AnnotBorderBS. Frontends should
>always use BS when updating an annotation that can have BS entries. The
>patch also completes the implementation of writeToObject method for
>array borders and adds an impementation for BS borders too.
>
>
>I've passed my tests with no regressions, but I can't test the qt
>frontend (that uses setBorder(), for example), so it would be great if
>someone could test these patches to make sure they don't introduce any
>regression in qt (it should build in any case).
>
>Leonard, please, feel free to correct me if I'm wrong in any of my
>interpretations of the PDF spec.
>
>Regards,