#-------------------------------------------------------------------------
#
#  Copyright (c) 2007, Enthought, Inc.
#  All rights reserved.
#
#  This software is provided without warranty under the terms of the BSD
#  license included in enthought/LICENSE.txt and may be redistributed only
#  under the conditions described in the aforementioned license.  The license
#  is also available online at http://www.enthought.com/licenses/BSD.txt
#
#  Thanks for using Enthought open source!
#
#  Author: David C. Morrill
#  Date:   06/06/2007
#
#-------------------------------------------------------------------------

""" Class to aid in automatically computing the 'slice' points for a specified
    ImageResource and then drawing it that it can be 'stretched' to fit a larger
    region than the original image.
"""

#-------------------------------------------------------------------------
#  Imports:
#-------------------------------------------------------------------------

from __future__ import absolute_import
import wx

from colorsys \
    import rgb_to_hls

from numpy \
    import reshape, fromstring, uint8

from traits.api \
    import HasPrivateTraits, Instance, Int, List, Color, Enum, Bool

from pyface.image_resource \
    import ImageResource

from .constants \
    import WindowColor

from .constants import is_mac
import traitsui.wx.constants


#-------------------------------------------------------------------------
#  Recursively paint the parent's background if they have an associated image
#  slice.
#-------------------------------------------------------------------------


def paint_parent(dc, window):
    """ Recursively paint the parent's background if they have an associated
        image slice.
    """
    parent = window.GetParent()
    slice = getattr(parent, '_image_slice', None)
    if slice is not None:
        x, y = window.GetPositionTuple()
        dx, dy = parent.GetSizeTuple()
        slice.fill(dc, -x, -y, dx, dy)
    else:
        # Otherwise, just paint the normal window background color:
        dx, dy = window.GetClientSizeTuple()
        if is_mac and hasattr(window, '_border') and window._border:
            dc.SetBackgroundMode(wx.TRANSPARENT)
            dc.SetBrush(wx.Brush(wx.Colour(0, 0, 0, 0)))
        else:
            dc.SetBrush(wx.Brush(parent.GetBackgroundColour()))
        dc.SetPen(wx.TRANSPARENT_PEN)
        dc.DrawRectangle(0, 0, dx, dy)

    return slice

#-------------------------------------------------------------------------
#  'ImageSlice' class:
#-------------------------------------------------------------------------


class ImageSlice(HasPrivateTraits):

    #-- Trait Definitions ----------------------------------------------------

    # The ImageResource to be sliced and drawn:
    image = Instance(ImageResource)

    # The minimum number of adjacent, identical rows/columns needed to identify
    # a repeatable section:
    threshold = Int(10)

    # The maximum number of 'stretchable' rows and columns:
    stretch_rows = Enum(1, 2)
    stretch_columns = Enum(1, 2)

    # Width/height of the image borders:
    top = Int
    bottom = Int
    left = Int
    right = Int

    # Width/height of the extended image borders:
    xtop = Int
    xbottom = Int
    xleft = Int
    xright = Int

    # The color to use for content text:
    content_color = Instance(wx.Colour)

    # The color to use for label text:
    label_color = Instance(wx.Colour)

    # The background color of the image:
    bg_color = Color

    # Should debugging slice lines be drawn?
    debug = Bool(False)

    #-- Private Traits -------------------------------------------------------

    # The current image's opaque bitmap:
    opaque_bitmap = Instance(wx.Bitmap)

    # The current image's transparent bitmap:
    transparent_bitmap = Instance(wx.Bitmap)

    # Size of the current image:
    dx = Int
    dy = Int

    # Size of the current image's slices:
    dxs = List
    dys = List

    # Fixed minimum size of current image:
    fdx = Int
    fdy = Int

    #-- Public Methods -------------------------------------------------------

    def fill(self, dc, x, y, dx, dy, transparent=False):
        """ 'Stretch fill' the specified region of a device context with the
            sliced image.
        """
        # Create the source image dc:
        idc = wx.MemoryDC()
        if transparent:
            idc.SelectObject(self.transparent_bitmap)
        else:
            idc.SelectObject(self.opaque_bitmap)

        # Set up the drawing parameters:
        sdx, sdy = self.dx, self.dx
        dxs, dys = self.dxs, self.dys
        tdx, tdy = dx - self.fdx, dy - self.fdy

        # Calculate vertical slice sizes to use for source and destination:
        n = len(dxs)
        if n == 1:
            pdxs = [(0, 0), (1, max(1, tdx / 2)), (sdx - 2, sdx - 2),
                    (1, max(1, tdx - (tdx / 2))), (0, 0)]
        elif n == 3:
            pdxs = [(dxs[0], dxs[0]), (dxs[1], max(0, tdx)), (0, 0),
                    (0, 0), (dxs[2], dxs[2])]
        else:
            pdxs = [(dxs[0], dxs[0]), (dxs[1], max(0, tdx / 2)),
                    (dxs[2], dxs[2]), (dxs[3], max(0, tdx - (tdx / 2))),
                    (dxs[4], dxs[4])]

        # Calculate horizontal slice sizes to use for source and destination:
        n = len(dys)
        if n == 1:
            pdys = [(0, 0), (1, max(1, tdy / 2)), (sdy - 2, sdy - 2),
                    (1, max(1, tdy - (tdy / 2))), (0, 0)]
        elif n == 3:
            pdys = [(dys[0], dys[0]), (dys[1], max(0, tdy)), (0, 0),
                    (0, 0), (dys[2], dys[2])]
        else:
            pdys = [(dys[0], dys[0]), (dys[1], max(0, tdy / 2)),
                    (dys[2], dys[2]), (dys[3], max(0, tdy - (tdy / 2))),
                    (dys[4], dys[4])]

        # Iterate over each cell, performing a stretch fill from the source
        # image to the destination window:
        last_x, last_y = x + dx, y + dy
        y0, iy0 = y, 0
        for idy, wdy in pdys:
            if y0 >= last_y:
                break

            if wdy != 0:
                x0, ix0 = x, 0
                for idx, wdx in pdxs:
                    if x0 >= last_x:
                        break

                    if wdx != 0:
                        self._fill(idc, ix0, iy0, idx, idy,
                                   dc, x0, y0, wdx, wdy)
                        x0 += wdx
                    ix0 += idx
                y0 += wdy
            iy0 += idy

        if self.debug:
            dc.SetPen(wx.Pen(wx.RED))
            dc.DrawLine(x, y + self.top, last_x, y + self.top)
            dc.DrawLine(x, last_y - self.bottom - 1,
                        last_x, last_y - self.bottom - 1)
            dc.DrawLine(x + self.left, y, x + self.left, last_y)
            dc.DrawLine(last_x - self.right - 1, y,
                        last_x - self.right - 1, last_y)

    #-- Event Handlers -------------------------------------------------------

    def _image_changed(self, image):
        """ Handles the 'image' trait being changed.
        """
        # Save the original bitmap as the transparent version:
        self.transparent_bitmap = bitmap = \
            image.create_image().ConvertToBitmap()

        # Save the bitmap size information:
        self.dx = dx = bitmap.GetWidth()
        self.dy = dy = bitmap.GetHeight()

        # Create the opaque version of the bitmap:
        self.opaque_bitmap = wx.EmptyBitmap(dx, dy)
        mdc2 = wx.MemoryDC()
        mdc2.SelectObject(self.opaque_bitmap)
        mdc2.SetBrush(wx.Brush(WindowColor))
        mdc2.SetPen(wx.TRANSPARENT_PEN)
        mdc2.DrawRectangle(0, 0, dx, dy)
        mdc = wx.MemoryDC()
        mdc.SelectObject(bitmap)
        mdc2.Blit(0, 0, dx, dy, mdc, 0, 0, useMask=True)
        mdc.SelectObject(wx.NullBitmap)
        mdc2.SelectObject(wx.NullBitmap)

        # Finally, analyze the image to find out its characteristics:
        self._analyze_bitmap()

    #-- Private Methods ------------------------------------------------------

    def _analyze_bitmap(self):
        """ Analyzes the bitmap.
        """
        # Get the image data:
        threshold = self.threshold
        bitmap = self.opaque_bitmap
        dx, dy = self.dx, self.dy
        image = bitmap.ConvertToImage()

        # Convert the bitmap data to a numpy array for analysis:
        data = reshape(fromstring(image.GetData(), uint8), (dy, dx, 3))

        # Find the horizontal slices:
        matches = []
        y, last = 0, dy - 1
        max_diff = 0.10 * dx
        while y < last:
            y_data = data[y]
            for y2 in range(y + 1, dy):
                if abs(y_data - data[y2]).sum() > max_diff:
                    break

            n = y2 - y
            if n >= threshold:
                matches.append((y, n))

            y = y2

        n = len(matches)
        if n == 0:
            if dy > 50:
                matches = [(0, dy)]
            else:
                matches = [(dy / 2, 1)]
        elif n > self.stretch_rows:
            matches.sort(key=lambda x: x[1], reverse=True)
            matches = matches[: self.stretch_rows]

        # Calculate and save the horizontal slice sizes:
        self.fdy, self.dys = self._calculate_dxy(dy, matches)

        # Find the vertical slices:
        matches = []
        x, last = 0, dx - 1
        max_diff = 0.10 * dy
        while x < last:
            x_data = data[:, x]
            for x2 in range(x + 1, dx):
                if abs(x_data - data[:, x2]).sum() > max_diff:
                    break

            n = x2 - x
            if n >= threshold:
                matches.append((x, n))

            x = x2

        n = len(matches)
        if n == 0:
            if dx > 50:
                matches = [(0, dx)]
            else:
                matches = [(dx / 2, 1)]
        elif n > self.stretch_columns:
            matches.sort(key=lambda x: x[1], reverse=True)
            matches = matches[: self.stretch_columns]

        # Calculate and save the vertical slice sizes:
        self.fdx, self.dxs = self._calculate_dxy(dx, matches)

        # Save the border size information:
        self.top = min(dy / 2, self.dys[0])
        self.bottom = min(dy / 2, self.dys[-1])
        self.left = min(dx / 2, self.dxs[0])
        self.right = min(dx / 2, self.dxs[-1])

        # Find the optimal size for the borders (i.e. xleft, xright, ... ):
        self._find_best_borders(data)

        # Save the background color:
        x, y = (dx / 2), (dy / 2)
        r, g, b = data[y, x]
        self.bg_color = (0x10000 * r) + (0x100 * g) + b

        # Find the best contrasting text color (black or white):
        self.content_color = self._find_best_color(data, x, y)

        # Find the best contrasting label color:
        if self.xtop >= self.xbottom:
            self.label_color = self._find_best_color(data, x, self.xtop / 2)
        else:
            self.label_color = self._find_best_color(
                data, x, dy - (self.xbottom / 2) - 1)

    def _fill(self, idc, ix, iy, idx, idy, dc, x, y, dx, dy):
        """ Performs a stretch fill of a region of an image into a region of a
            window device context.
        """
        last_x, last_y = x + dx, y + dy
        while y < last_y:
            ddy = min(idy, last_y - y)
            x0 = x
            while x0 < last_x:
                ddx = min(idx, last_x - x0)
                dc.Blit(x0, y, ddx, ddy, idc, ix, iy, useMask=True)
                x0 += ddx
            y += ddy

    def _calculate_dxy(self, d, matches):
        """ Calculate the size of all image slices for a specified set of
            matches.
        """
        if len(matches) == 1:
            d1, d2 = matches[0]

            return (d - d2, [d1, d2, d - d1 - d2])

        d1, d2 = matches[0]
        d3, d4 = matches[1]

        return (d - d2 - d4, [d1, d2, d3 - d1 - d2, d4, d - d3 - d4])

    def _find_best_borders(self, data):
        """ Find the best set of image slice border sizes (e.g. for images with
            rounded corners, there should exist a better set of borders than
            the ones computed by the image slice algorithm.
        """
        # Make sure the image size is worth bothering about:
        dx, dy = self.dx, self.dy
        if (dx < 5) or (dy < 5):
            return

        # Calculate the starting point:
        left = right = dx / 2
        top = bottom = dy / 2

        # Calculate the end points:
        last_y = dy - 1
        last_x = dx - 1

        # Mark which edges as 'scanning':
        t = b = l = r = True

        # Keep looping while at last one edge is still 'scanning':
        while l or r or t or b:

            # Calculate the current core area size:
            height = bottom - top + 1
            width = right - left + 1

            # Try to extend all edges that are still 'scanning':
            nl = (l and (left > 0) and
                  self._is_equal(data, left - 1, top, left, top, 1, height))

            nr = (r and (right < last_x) and
                  self._is_equal(data, right + 1, top, right, top, 1, height))

            nt = (t and (top > 0) and
                  self._is_equal(data, left, top - 1, left, top, width, 1))

            nb = (b and (bottom < last_y) and
                  self._is_equal(data, left, bottom + 1, left, bottom,
                                 width, 1))

            # Now check the corners of the edges:
            tl = ((not nl) or (not nt) or
                  self._is_equal(data, left - 1, top - 1, left, top, 1, 1))

            tr = ((not nr) or (not nt) or
                  self._is_equal(data, right + 1, top - 1, right, top, 1, 1))

            bl = ((not nl) or (not nb) or
                  self._is_equal(data, left - 1, bottom + 1, left, bottom,
                                 1, 1))

            br = ((not nr) or (not nb) or
                  self._is_equal(data, right + 1, bottom + 1, right, bottom,
                                 1, 1))

            # Calculate the new edge 'scanning' values:
            l = nl and tl and bl
            r = nr and tr and br
            t = nt and tl and tr
            b = nb and bl and br

            # Adjust the coordinate of an edge if it is still 'scanning':
            left -= l
            right += r
            top -= t
            bottom += b

        # Now compute the best set of image border sizes using the current set
        # and the ones we just calculated:
        self.xleft = min(self.left, left)
        self.xright = min(self.right, dx - right - 1)
        self.xtop = min(self.top, top)
        self.xbottom = min(self.bottom, dy - bottom - 1)

    def _find_best_color(self, data, x, y):
        """ Find the best contrasting text color for a specified pixel
            coordinate.
        """
        r, g, b = data[y, x]
        h, l, s = rgb_to_hls(r / 255.0, g / 255.0, b / 255.0)
        text_color = wx.BLACK
        if l < 0.50:
            text_color = wx.WHITE

        return text_color

    def _is_equal(self, data, x0, y0, x1, y1, dx, dy):
        """ Determines if two identically sized regions of an image array are
            'the same' (i.e. within some slight color variance of each other).
        """
        return (abs(data[y0: y0 + dy, x0: x0 + dx] -
                    data[y1: y1 + dy, x1: x1 + dx]).sum() < 0.10 * dx * dy)


#-------------------------------------------------------------------------
#  Returns a (possibly cached) ImageSlice:
#-------------------------------------------------------------------------

image_slice_cache = {}


def image_slice_for(image):
    """ Returns a (possibly cached) ImageSlice.
    """
    global image_slice_cache

    result = image_slice_cache.get(image)
    if result is None:
        image_slice_cache[image] = result = ImageSlice(image=image)

    return result