twinside / svg-tree Goto Github PK

See https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/clip-path#Usage_notes and https://developer.mozilla.org/en-US/docs/Web/CSS/basic-shape

This is a low priority issue for me.

Hi,

We recently discovered that the value None in the Tree datatype is serialized as "< />" and this element is not validated by some XML libraries (for instance libxml2). Such SVG files are rejected by libraries like librsvg2 or Firefox. For instance, this svg file.

This issue was found using QuickFuzz (we have a new engine to generate arbitrary instances in using svg-tree, thanks to @martinceresa ).

Looking at the SVG spec, and the documents produced by Inkscape, I see things like:

    <ellipse
       style="fill:none;fill-opacity:1;stroke:#ff0000;stroke-width:31.24999809;stroke-linecap:round;stroke-linejoin:miter;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1"
       id="whatever"
       cx="-1514.915"
       cy="-19145.889"
       rx="185.75783"
       ry="181.22716"
       inkscape:label="#unk"
       transform="matrix(0.93969262,0.34202015,-0.34202015,0.93969262,0,0)"
       clip-path="none">
      <desc
         id="desc11832">test</desc>
    </ellipse>

I think that desc should be parsed and added to DrawAttributes? I couldn't immediately figure out how when looking at the code though.

See Twinside/Juicy.Pixels#136

It would be useful for reanimate[1] if Graphics.Svg.NamedColors was exported. Would you accept such as PR?

[1] https://github.com/Lemmih/reanimate

Noticed with Stackage, trying to build locally to reproduce

After generating some random svg values (Document) some of them are failing if i try to show them with this error:

Exception thrown while printing test case: '<<loop>>'

Unfortunately, i don't know which constructors are used, because i can't print them, but the Arbitrary instances to reproduce this issue are available here:

https://github.com/CIFASIS/QuickFuzz/blob/master/src/Svg.hs

I would like to have an element inside a text element (See https://developer.mozilla.org/en-US/docs/Web/SVG/Element/text).

Is there a way to do this using the library?

All elements can have an ID but not all elements may have drawing attributes.

Would you please consider changing the documentSize function to return (Double, Double) instead of (Int, Int)? The rounding causes an annoying precision loss in my gcode converter.

I was just wondering about this as I am interfacing my library (which uses Double) to yours. I could change my library to use Float which would avoid using float2Double everywhere so I was wondering why you chose Float over Double? I think the SVG spec says floats should be double precision.

A lot of the draw attributes (such as strokeWidth and strokeColor) use the 'Last' monoid. As I see it, the only purpose of this is to tell when attributes have been omitted. Are there any use-cases that require this or can we simply fill in missing attributes with their default values?

I have the following simple code:

module SvgTree.IO where
import Graphics.Svg
import Data.Maybe

fn="data/input/00.jpg.svg"
svgTest = loadSvgFile fn
save= do
  l<-svgTest
  saveXmlFile "testSvgTree.svg" $ fromJust l

if I have in data/input/00.jpg.svg the following:

<?xml version="1.0" encoding="utf-8"?>
<!-- Generator: Adobe Illustrator 18.0.0, SVG Export Plug-In . SVG Version: 6.00 Build 0)  -->
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
<svg version="1.1" id="Layer_1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" x="0px" y="0px"
     viewBox="0 1023.8 612 382.5" enable-background="new 0 1023.8 612 382.5" xml:space="preserve">
<image overflow="visible" enable-background="new    " width="1680" height="1050" xlink:href="00.jpg"  transform="matrix(0.3643 0 0 0.3643 0 1023.75)">
</image>
<rect x="289.4" y="1221.2" fill="#FFFFFF" stroke="#FFFFFF" stroke-miterlimit="10" width="7" height="9.1"/>
<rect x="319" y="1141.9" fill="#FFFFFF" width="10.2" height="12.2"/>
<rect x="327.3" y="1315.9" fill="#FFFFFF" width="9.1" height="15"/>
<rect x="277.1" y="1089.2" fill="#FFFFFF" width="9.1" height="12.6"/>
</svg>

then in testSvgTree.svg I get this :

<?xml version='1.0' ?>
<svg viewBox="0 1023 612 382" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" version="1.1">
  <image transform="matrix(0.3643, 0.0, 0.0, 0.3643, 0.0, 1023.75)" class="" width="1680.0" height="1050.0" xlink:href="#00.jpg" />
  <rect stroke="#FFFFFF" fill="#FFFFFF" stroke-miterlimit="10.0" class="" width="7.0" height="9.1" x="289.4" y="1221.2" />
  <rect fill="#FFFFFF" class="" width="10.2" height="12.2" x="319.0" y="1141.9" />
  <rect fill="#FFFFFF" class="" width="9.1" height="15.0" x="327.3" y="1315.9" />
  <rect fill="#FFFFFF" class="" width="9.1" height="12.6" x="277.1" y="1089.2" />
</svg>

which has the problem that xlink:href="#00.jpg" has an extra # in front of the file name.

Any idea how to fix this ? Which part of the library causes the issue ?

I wonder if it were possible to create some kind of QuickCheck test with the property that whatever was parsed in should produce the same output ? Maybe that would have catched this bug ?

Hi,

Even in the last revision, it is still possible to serialize a svg and obtain an invalid svg because of a duplicated id. For instance:

Document {_viewBox = Just (10.93378015676586,-2.5002198377193974,1.1076888622404122,1.3056209094151552), _width = Nothing, _height = Nothing, _elements = [], _definitions = fromList [("d",ElementClipPath (ClipPath {_clipPathDrawAttributes = DrawAttributes {_strokeWidth = Last {getLast = Nothing}, _strokeColor = Last {getLast = Just (TextureRef "d")}, _strokeOpacity = Nothing, _strokeLineCap = Last {getLast = Just CapButt}, _strokeLineJoin = Last {getLast = Just JoinMiter}, _strokeMiterLimit = Last {getLast = Just 0.608029056458719}, _fillColor = Last {getLast = Just (TextureRef "a")}, _fillOpacity = Nothing, _groupOpacity = Just (-1.1104697), _transform = Just [], _fillRule = Last {getLast = Nothing}, _maskRef = Last {getLast = Nothing}, _clipPathRef = Last {getLast = Nothing}, _clipRule = Last {getLast = Nothing}, _attrClass = ["b"], _attrId = Just "e", _strokeOffset = Last {getLast = Nothing}, _strokeDashArray = Last {getLast = Nothing}, _fontSize = Last {getLast = Nothing}, _fontFamily = Last {getLast = Just ["e"]}, _fontStyle = Last {getLast = Just FontStyleItalic}, _textAnchor = Last {getLast = Just TextAnchorStart}, _markerStart = Last {getLast = Nothing}, _markerMid = Last {getLast = Nothing}, _markerEnd = Last {getLast = Just (Ref "d")}}, _clipPathUnits = CoordBoundingBox, _clipPathContent = [CircleTree (Circle {_circleDrawAttributes = DrawAttributes {_strokeWidth = Last {getLast = Nothing}, _strokeColor = Last {getLast = Nothing}, _strokeOpacity = Nothing, _strokeLineCap = Last {getLast = Nothing}, _strokeLineJoin = Last {getLast = Nothing}, _strokeMiterLimit = Last {getLast = Nothing}, _fillColor = Last {getLast = Just (TextureRef "c")}, _fillOpacity = Just 0.0, _groupOpacity = Just 0.0, _transform = Just [], _fillRule = Last {getLast = Nothing}, _maskRef = Last {getLast = Nothing}, _clipPathRef = Last {getLast = Just RefNone}, _clipRule = Last {getLast = Just FillNonZero}, _attrClass = [], _attrId = Just "d", _strokeOffset = Last {getLast = Nothing}, _strokeDashArray = Last {getLast = Nothing}, _fontSize = Last {getLast = Just (Num 0.0)}, _fontFamily = Last {getLast = Nothing}, _fontStyle = Last {getLast = Just FontStyleNormal}, _textAnchor = Last {getLast = Just TextAnchorEnd}, _markerStart = Last {getLast = Just (Ref "e")}, _markerMid = Last {getLast = Just RefNone}, _markerEnd = Last {getLast = Just (Ref "a")}}, _circleCenter = (Num 0.0,Point 0.0), _circleRadius = Mm 0.0})]}))], _description = "d", _styleRules = [], _documentLocation = "b"}

I'm not sure if this is a bug in svg-tree.

I was going to offer PR for svg-tree, but before doing so I reviewed existing PRs and forks, and noticed the https://github.com/reanimate/reanimate-svg fork (preferably I'd file this issue there but it has GitHub issues disabled).

It would be nice to know what are the differences between the two. Preferably reanimate-svg would explain it (as it initiated the fork) but lacking that, an explanation in svg-tree's readme would clear things up!