Re[2]: [Gnash-commit] gnash/server shape.cpp

[Udo Giacomozzi]

Hello strk,

Monday, March 5, 2007, 1:17:20 PM, you wrote:
s> So, shape_character_def should find closed paths itself... being
s> closed path sequences with *all*  paths having either a left a fill 
specified.
s> Sounds as we'd need more testcases for this.

Yes, in theory it should work similarly to the "path" version, but
walking all edges in all paths.

I think it should be relatively simple:

- being X and Y the coordinates of the desired point (transformed by
  the reverse matrix, i.e. normalized to the character coordinates!):

- for each path:

  - for each edge in the path:

    - find X1 := the edge X coordinate crossing Y

    - go to next edge if no crossing

    - when matchX < X1 < X:

      - set matchX := X1

      - choose m_fill0 if edge goes upward or m_fill1 if edge goes
        downward

      - temporary function result is true when choosen fill style is
        not zero

  - next edge

- next path


Note the "matchX" way avoids sorting of all X coordinates since we
must walk all edges anyway.


There are some tricky parts:

- calculation of curve crossings (a bit of math, should be already
  implemented in the current point_test()

- a curve may cross the imaginary ray at two points:

                  ...
               .       .
       ------.-----------.-------
           .               .


  It that case both points must be tested and m_fill0 be used for one
  crossing and m_fill1 for the other.



Basically the algorithm says: The point is inside the shape when the
rightmost edge (in that line), which is still at the left of the test
point, has a fill style at it's right side. For a line that goes from
bottom to top, that's m_fill1. For a line that goes from top to
bottom, that's m_fill0.


Udo