mayan-edms/mayan/apps/converter/transformations.py

from __future__ import unicode_literals

import hashlib
import logging

from PIL import Image, ImageColor, ImageFilter

from django.utils.text import format_lazy
from django.utils.translation import ugettext_lazy as _
from django.utils.encoding import force_bytes

logger = logging.getLogger(__name__)


class BaseTransformation(object):
    """
    Transformation can modify the appearance of the document's page preview.
    Some transformation available are: Rotate, zoom, resize and crop.
    """
    arguments = ()
    name = 'base_transformation'
    _registry = {}

    @staticmethod
    def combine(transformations):
        result = None

        for transformation in transformations:
            if not result:
                result = hashlib.sha256(transformation.cache_hash())
            else:
                result.update(transformation.cache_hash())

        return result.hexdigest()

    @classmethod
    def get(cls, name):
        return cls._registry[name]

    @classmethod
    def get_label(cls):
        if cls.arguments:
            return format_lazy(
                '{}{}{}', cls.label, ': ', ', '.join(cls.arguments)
            )
        else:
            return cls.label

    @classmethod
    def get_transformation_choices(cls):
        return sorted(
            [
                (name, klass.get_label()) for name, klass in cls._registry.items()
            ]
        )

    @classmethod
    def register(cls, transformation):
        cls._registry[transformation.name] = transformation

    def __init__(self, **kwargs):
        self.kwargs = {}
        for argument_name in self.arguments:
            setattr(self, argument_name, kwargs.get(argument_name))
            self.kwargs[argument_name] = kwargs.get(argument_name)

    def cache_hash(self):
        result = hashlib.sha256(force_bytes(self.name))

        # Sort arguments for guaranteed repeatability
        for key, value in sorted(self.kwargs.items()):
            result.update(force_bytes(key))
            result.update(force_bytes(value))

        return force_bytes(result.hexdigest())

    def execute_on(self, image):
        self.image = image
        self.aspect = 1.0 * image.size[0] / image.size[1]

    def serialize(self):
        kwargs = []
        for key, value in sorted(self.kwargs.items()):
            kwargs.append(
                '{}:{}'.format(key, value)
            )

        return '{};{}'.format(self.name, ','.join(kwargs))


class TransformationCrop(BaseTransformation):
    arguments = ('left', 'top', 'right', 'bottom',)
    label = _('Crop')
    name = 'crop'

    def execute_on(self, *args, **kwargs):
        super(TransformationCrop, self).execute_on(*args, **kwargs)

        try:
            left = int(self.left or '0')
        except ValueError:
            left = 0

        try:
            top = int(self.top or '0')
        except ValueError:
            top = 0

        try:
            right = int(self.right or '0')
        except ValueError:
            right = 0

        try:
            bottom = int(self.bottom or '0')
        except ValueError:
            bottom = 0

        if left < 0:
            left = 0

        if left > self.image.size[0] - 1:
            left = self.image.size[0] - 1

        if top < 0:
            top = 0

        if top > self.image.size[1] - 1:
            top = self.image.size[1] - 1

        if right < 0:
            right = 0

        if right > self.image.size[0] - 1:
            right = self.image.size[0] - 1

        if bottom < 0:
            bottom = 0

        if bottom > self.image.size[1] - 1:
            bottom = self.image.size[1] - 1

        # Invert right value
        # Pillow uses left, top, right, bottom to define a viewport
        # of real coordinates
        # We invert the right and bottom to define a viewport
        # that can crop from the right and bottom borders without
        # having to know the real dimensions of an image
        right = self.image.size[0] - right
        bottom = self.image.size[1] - bottom

        if left > right:
            left = right - 1

        if top > bottom:
            top = bottom - 1

        logger.debug(
            'left: %f, top: %f, right: %f, bottom: %f', left, top, right,
            bottom
        )

        return self.image.crop((left, top, right, bottom))


class TransformationFlip(BaseTransformation):
    arguments = ()
    label = _('Flip')
    name = 'flip'

    def execute_on(self, *args, **kwargs):
        super(TransformationFlip, self).execute_on(*args, **kwargs)

        return self.image.transpose(Image.FLIP_TOP_BOTTOM)


class TransformationGaussianBlur(BaseTransformation):
    arguments = ('radius',)
    label = _('Gaussian blur')
    name = 'gaussianblur'

    def execute_on(self, *args, **kwargs):
        super(TransformationGaussianBlur, self).execute_on(*args, **kwargs)

        return self.image.filter(ImageFilter.GaussianBlur(radius=self.radius))


class TransformationLineArt(BaseTransformation):
    label = _('Line art')
    name = 'lineart'

    def execute_on(self, *args, **kwargs):
        super(TransformationLineArt, self).execute_on(*args, **kwargs)

        return self.image.convert('L').point(lambda x: 0 if x < 128 else 255, '1')


class TransformationMirror(BaseTransformation):
    arguments = ()
    label = _('Mirror')
    name = 'mirror'

    def execute_on(self, *args, **kwargs):
        super(TransformationMirror, self).execute_on(*args, **kwargs)

        return self.image.transpose(Image.FLIP_LEFT_RIGHT)


class TransformationResize(BaseTransformation):
    arguments = ('width', 'height')
    label = _('Resize')
    name = 'resize'

    def execute_on(self, *args, **kwargs):
        super(TransformationResize, self).execute_on(*args, **kwargs)

        width = int(self.width)
        height = int(self.height or 1.0 * width / self.aspect)

        factor = 1
        while self.image.size[0] / factor > 2 * width and self.image.size[1] * 2 / factor > 2 * height:
            factor *= 2

        if factor > 1:
            self.image.thumbnail(
                (self.image.size[0] / factor, self.image.size[1] / factor),
                Image.NEAREST
            )

        # Resize the image with best quality algorithm ANTI-ALIAS
        self.image.thumbnail((width, height), Image.ANTIALIAS)

        return self.image


class TransformationRotate(BaseTransformation):
    arguments = ('degrees', 'fillcolor')
    label = _('Rotate')
    name = 'rotate'

    def execute_on(self, *args, **kwargs):
        super(TransformationRotate, self).execute_on(*args, **kwargs)

        self.degrees %= 360

        if self.degrees == 0:
            return self.image

        fillcolor_value = getattr(self, 'fillcolor', None)
        if fillcolor_value:
            fillcolor = ImageColor.getrgb(fillcolor_value)
        else:
            fillcolor = None

        return self.image.rotate(
            angle=360 - self.degrees, resample=Image.BICUBIC, expand=True,
            fillcolor=fillcolor
        )


class TransformationRotate90(TransformationRotate):
    arguments = ()
    degrees = 90
    label = _('Rotate 90 degrees')
    name = 'rotate90'

    def __init__(self, **kwargs):
        super(TransformationRotate90, self).__init__()
        self.kwargs['degrees'] = 90


class TransformationRotate180(TransformationRotate):
    arguments = ()
    degrees = 180
    label = _('Rotate 180 degrees')
    name = 'rotate180'

    def __init__(self, **kwargs):
        super(TransformationRotate180, self).__init__()
        self.kwargs['degrees'] = 180


class TransformationRotate270(TransformationRotate):
    arguments = ()
    degrees = 270
    label = _('Rotate 270 degrees')
    name = 'rotate270'

    def __init__(self, **kwargs):
        super(TransformationRotate270, self).__init__()
        self.kwargs['degrees'] = 270


class TransformationUnsharpMask(BaseTransformation):
    arguments = ('radius', 'percent', 'threshold')
    label = _('Unsharp masking')
    name = 'unsharpmask'

    def execute_on(self, *args, **kwargs):
        super(TransformationUnsharpMask, self).execute_on(*args, **kwargs)

        return self.image.filter(
            ImageFilter.UnsharpMask(
                radius=self.radius, percent=self.percent,
                threshold=self.threshold
            )
        )


class TransformationZoom(BaseTransformation):
    arguments = ('percent',)
    label = _('Zoom')
    name = 'zoom'

    def execute_on(self, *args, **kwargs):
        super(TransformationZoom, self).execute_on(*args, **kwargs)

        if self.percent == 100:
            return self.image

        decimal_value = float(self.percent) / 100
        return self.image.resize(
            (
                int(self.image.size[0] * decimal_value),
                int(self.image.size[1] * decimal_value)
            ), Image.ANTIALIAS
        )


BaseTransformation.register(TransformationCrop)
BaseTransformation.register(TransformationFlip)
BaseTransformation.register(TransformationGaussianBlur)
BaseTransformation.register(TransformationLineArt)
BaseTransformation.register(TransformationMirror)
BaseTransformation.register(TransformationResize)
BaseTransformation.register(TransformationRotate)
BaseTransformation.register(TransformationRotate90)
BaseTransformation.register(TransformationRotate180)
BaseTransformation.register(TransformationRotate270)
BaseTransformation.register(TransformationUnsharpMask)
BaseTransformation.register(TransformationZoom)