The ctypes-based simple ImageMagick binding for Python

Exception information:

Traceback (most recent call last):
  File "testss.py", line 1, in <module>
    from wand.image import Image
  File "/Users/hit9/spam/testfirfffff/venv/lib/python2.7/site-packages/wand/image.py", line 20, in <module>
    from .api import MagickPixelPacket, libc, libmagick, library
  File "/Users/hit9/spam/testfirfffff/venv/lib/python2.7/site-packages/wand/api.py", line 156, in <module>
    'Try to install:\n  ' + msg)
ImportError: MagickWand shared library not found.
You probably had not installed ImageMagick library.
Try to install:
  brew install imagemagick

I am using OSX10.8

And I installed imagemagick via brew:

$ brew install imagemagick
Warning: imagemagick-6.8.6-3 already installed

And here is my result run ls in /usr/local/lib :

$ ls -l  /usr/local/lib/ | grep libMagick
lrwxr-xr-x   1 hit9  admin       59 Aug 22 11:04 libMagick++-6.Q16.1.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagick++-6.Q16.1.dylib
lrwxr-xr-x   1 hit9  admin       57 Aug 22 11:04 libMagick++-6.Q16.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagick++-6.Q16.dylib
lrwxr-xr-x   1 hit9  admin       54 Aug 22 11:04 libMagick++-6.Q16.la -> ../Cellar/imagemagick/6.8.6-3/lib/libMagick++-6.Q16.la
lrwxr-xr-x   1 hit9  admin       61 Aug 22 11:04 libMagickCore-6.Q16.1.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickCore-6.Q16.1.dylib
lrwxr-xr-x   1 hit9  admin       59 Aug 22 11:04 libMagickCore-6.Q16.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickCore-6.Q16.dylib
lrwxr-xr-x   1 hit9  admin       56 Aug 22 11:04 libMagickCore-6.Q16.la -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickCore-6.Q16.la
lrwxr-xr-x   1 hit9  admin       61 Aug 22 11:04 libMagickWand-6.Q16.1.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickWand-6.Q16.1.dylib
lrwxr-xr-x   1 hit9  admin       59 Aug 22 11:04 libMagickWand-6.Q16.dylib -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickWand-6.Q16.dylib
lrwxr-xr-x   1 hit9  admin       56 Aug 22 11:04 libMagickWand-6.Q16.la -> ../Cellar/imagemagick/6.8.6-3/lib/libMagickWand-6.Q16.la
lrwxr-xr-x   1 hit9  admin       42 Aug 22 11:42 libMagickWand.dylib -> /usr/local/lib/libMagickWand-6.Q16.1.dylib

I noticed there was a few issues similar to this. But I tried the solutions, and issue still exists. Hope your solution, thanks.

Hello,

Maybe I'm missing a step, but I got the same result on multiple machines:

sudo port install ImageMagick => successful sudo easy_install wand => successful

then in python interactive mode: from wand.image import Image

throws error:

Traceback (most recent call last): File "", line 1, in File "build/bdist.macosx-10.8-intel/egg/wand/image.py", line 19, in File "build/bdist.macosx-10.8-intel/egg/wand/api.py", line 266, in ImportError: MagickWand shared library not found or incompatible

Am I missing a step?

Thanks!

This is a work-in-progress, and should not be merged. I'm opening this pull request for greater visibility, and seek testing help.

This merge request reworked all C-API method mappings into a directory/module structure mimicking ImageMagick's C headers.

The primary benefit was to reduce the complexity of the unorganized wand.api.library, and to incorporate all MagickWand 6/7 methods.

The secondary goal is to include ImageMagick 7 support. For the most part, this has been completed with a few minor TODOs.

Overview of Changes

• C-API methods have been moved to the following structure.

  wand.api.cdefs
                .core              All MagickCore definitions.
                .drawing_wand      <drawing-wand.h> methods.
                .magick_image      <magick-image.h> methods.
                .magick_property   <magick-property.h> methods.
                .magick_wand       <magick-wand.h> methods.
                .pixel_iterator    <pixel-iterrator.h> methods.
                .pixel_wand        <pixel-wand.h> methods.
                .structures        Collection of ctypes.Structure definitions.
                .wandtypes         Ctype portability wrappers like `c_magick_char_p'
  

• PixelInfo replaces MagickPixelPacket
• HDRI support
• Implemented MagickSetImageChannelMask to replace Magick{XXXX}Channel methods.

Outstanding Items

• [x] Rewrite wand.image.Image.transform as MagickTransformImage has been removed in IM 7.
• [x] Revisite all methods/documentation for MagickCompositeImage. IM 7 is now more "aware" of dynamic channels; so naturally, behavior around Image.composite_channel, Image.composite, Image.transparentize, and Image.watermark should be reviewed.
• [x] Same for Image.alpha_channel property.
• [x] Rewrite regressions test marked XFAIL
• [ ] Documentation. Including all "?.?.?" version placeholders.

Help with...

I personally have run out of free time to finish this effort, and would love some help of the following.

• Performance & Architecture Design. With all MagickWand methods mapped, is wand load time traumatically affected? Is python-caching working in an optimized manner?
• Regression testing & CI. Ensure ImageMagick-6 continues to work as expected, and/or update Travis builds to include ImageMagick-7 + delegates.

The following code eats all your memory in a matter of seconds:

from wand.color import Color
from wand.image import Image

while True:
    ri = Image(width=1000, height=1000, background=Color('red'))
    bi = Image(width=1000, height=1000, background=Color('blue'))
    gi = Image(width=1000, height=1000, background=Color('green'))

    container = Image(width=1000, height=1000)
    container.sequence.append(ri)
    container.sequence.append(bi)
    container.sequence.append(gi)

    container.format = 'gif'
    container.save(filename='/dev/null')

    ri.destroy()
    bi.destroy()
    gi.destroy()
    for frame in container.sequence:
        frame.destroy()
    container.destroy()

    print('saved')

Is it a bug? How can I destroy all the images so the memory is freed? (I'm supposedly destroying them all, but it's not working)

Originally a PU for adding a few methods I needed for whatever project, this drawing_pu turned into a general effort to integrate all ImageMagick's drawing wand methods. This request is open for code-review, feedback, and improved documents.

Drawing API Integration Status

• [x] ClearDrawingWand
• [x] CloneDrawingWand
• [x] DestroyDrawingWand
• [x] DrawAffine
• [x] DrawAnnotation
• [x] DrawArc
• [x] DrawBezier
• [x] DrawCircle
• [x] DrawClearException
• [x] DrawComposite
• [x] DrawColor
• [x] DrawComment
• [x] DrawEllipse
• [x] DrawGetBorderColor
• [x] DrawGetClipPath
• [x] DrawGetClipRule
• [x] DrawGetClipUnits
• [x] DrawGetException
• [ ] DrawGetExceptionType (Never needed as python manages exceptions)
• [x] DrawGetFillColor
• [x] DrawGetFillOpacity
• [x] DrawGetFillRule
• [x] DrawGetFont
• [x] DrawGetFontFamily
• [x] DrawGetFontResolution
• [x] DrawGetFontSize
• [x] DrawGetFontStretch
• [x] DrawGetFontStyle
• [x] DrawGetFontWeight
• [x] DrawGetGravity
• [x] DrawGetOpacity
• [x] DrawGetStrokeAntialias
• [x] DrawGetStrokeColor
• [x] DrawGetStrokeDashArray
• [x] DrawGetStrokeDashOffset
• [x] DrawGetStrokeLineCap
• [x] DrawGetStrokeLineJoin
• [x] DrawGetStrokeMiterLimit
• [x] DrawGetStrokeOpacity
• [x] DrawGetStrokeWidth
• [x] DrawGetTextAlignment
• [x] DrawGetTextAntialias
• [x] DrawGetTextDecoration
• [x] DrawGetTextDirection
• [x] DrawGetTextEncoding
• [x] DrawGetTextKerning
• [x] DrawGetTextInterlineSpacing
• [x] DrawGetTextInterwordSpacing
• [x] DrawGetVectorGraphics
• [x] DrawGetTextUnderColor
• [x] DrawLine
• [x] DrawMatte
• [x] DrawPathClose
• [x] DrawPathCurveToAbsolute
• [x] DrawPathCurveToRelative
• [x] DrawPathCurveToQuadraticBezierAbsolute
• [x] DrawPathCurveToQuadraticBezierRelative
• [x] DrawPathCurveToQuadraticBezierSmoothAbsolute
• [x] DrawPathCurveToQuadraticBezierSmoothRelative
• [x] DrawPathCurveToSmoothAbsolute
• [x] DrawPathCurveToSmoothRelative
• [x] DrawPathEllipticArcAbsolute
• [x] DrawPathEllipticArcRelative
• [x] DrawPathFinish
• [x] DrawPathLineToAbsolute
• [x] DrawPathLineToRelative
• [x] DrawPathLineToHorizontalAbsolute
• [x] DrawPathLineToHorizontalRelative
• [x] DrawPathLineToVerticalAbsolute
• [x] DrawPathLineToVerticalRelative
• [x] DrawPathMoveToAbsolute
• [x] DrawPathMoveToRelative
• [x] DrawPathStart
• [x] DrawPoint
• [x] DrawPolygon
• [x] DrawPolyline
• [x] DrawPopClipPath
• [x] DrawPopDefs
• [x] DrawPopPattern
• [x] DrawPushClipPath
• [x] DrawPushDefs
• [x] DrawPushPattern
• [x] DrawRectangle
• [x] DrawResetVectorGraphics
• [x] DrawRotate
• [x] DrawRoundRectangle (Integrate with rectangle method)
• [x] DrawScale
• [x] DrawSetBorderColor
• [x] DrawSetClipPath
• [x] DrawSetClipRule
• [x] DrawSetClipUnits
• [x] DrawSetFillColor
• [x] DrawSetFillOpacity
• [x] DrawSetFontResolution
• [x] DrawSetOpacity
• [x] DrawSetFillPatternURL
• [x] DrawSetFillRule
• [x] DrawSetFont
• [x] DrawSetFontFamily
• [x] DrawSetFontSize
• [x] DrawSetFontStretch
• [x] DrawSetFontStyle
• [x] DrawSetFontWeight
• [x] DrawSetGravity
• [x] DrawSetStrokeColor
• [x] DrawSetStrokePatternURL
• [x] DrawSetStrokeAntialias
• [x] DrawSetStrokeDashArray
• [x] DrawSetStrokeDashOffset
• [x] DrawSetStrokeLineCap
• [x] DrawSetStrokeLineJoin
• [x] DrawSetStrokeMiterLimit
• [x] DrawSetStrokeOpacity
• [x] DrawSetStrokeWidth
• [x] DrawSetTextAlignment
• [x] DrawSetTextAntialias
• [x] DrawSetTextDecoration
• [x] DrawSetTextDirection
• [x] DrawSetTextEncoding
• [x] DrawSetTextKerning
• [x] DrawSetTextInterlineSpacing
• [x] DrawSetTextInterwordSpacing
• [x] DrawSetTextUnderColor
• [x] DrawSetVectorGraphics
• [x] DrawSkewX
• [x] DrawSkewY
• [x] DrawTranslate
• [x] DrawSetViewbox
• [x] IsDrawingWand
• [x] NewDrawingWand
• [ ] ~~~PeekDrawingWand~~~ (requires building DrawInfo struct, and this is not helpful until other drawing-info methods are exposed to the C-API)
• [x] PopDrawingWand
• [x] PushDrawingWand

Original PU Overview

• Created PointInfo struct reference in API type casting.
• Assigned / mapped arguments for wand library methods.
  • DrawPolygon
  • DrawPolyline
  • DrawBezier
• Completed initial unit test & documentation.

Examples

Drawing.polygon

with Image(filename="crosshatch.png") as img:
    with Drawing() as draw:
        draw.fill_color = Color("#fff")
        draw.stroke_color = Color("#000")
        draw.stroke_width = 2
        points = [(5,5),(45,40),(45,10),(5,45)]
        draw.polygon(points)
        draw.draw(img)
        img.save(filename="polygon.png")

Drawing.polyline

with Image(filename="crosshatch.png") as img:
    with Drawing() as draw:
        draw.fill_color = Color("transparent")
        draw.stroke_color = Color("#000")
        draw.stroke_width = 2
        points = [(5,5),(45,40),(45,10),(5,45)]
        draw.polyline(points)
        draw.draw(img)
        img.save(filename="polyline.png")

Drawing.bezier

with Image(filename="crosshatch.png") as img:
    with Drawing() as draw:
        draw.fill_color = Color("transparent")
        draw.stroke_color = Color("#000")
        draw.stroke_width = 2
        points = [(5,5), (0,50), (50,0), (45,45)]
        draw.bezier(points)
        draw.draw(img)
        img.save(filename="bezier.png")

The "colorMap" and "colorMapSize" attributes of the Image class are not available for reading / writing (or I missed them in the docs) so you can't get / set the palette of an image.

This is related to #468 so if you want to close this and re-open that, that's fine by me.

I'm using wand 0.6.2, and when I use the PDF still attached to #468 , and run that example code I posted, (I tried both setting alpha_channel to off and False) I get something far weirder (see attached). The original fix that was posted to stackoverflow also no longer works (but I think that's expected because you now handle it internally).

Some sample code:

This works:

with Image(filename='5-page-pdf.pdf') as img:
    img.type = "grayscale"
    img.format = "tiff"
    img.compression = "lzw"
    img.alpha_channel = 'off'
    img.save(filename="test.tiff")

This does not:

with Image(filename='5-page-pdf.pdf') as img:
    img.type = "grayscale"
    img.format = "tiff"
    img.compression = "lzw"
    img.alpha_channel = False
    img.save(filename="test.tiff")

I've noticed that some TIFF viewers enhance images more than others, the Preview application on macOS is pretty barebones and doesn't do anything to make the TIFF nicer, so the difference between the working and non working examples is that the non working example only has the first and last page with the alpha channel correctly set to off. (When Windows loads the TIFF using Windows Photo Viewer, it looks correct, but I believe its not showing the raw tiff). Not a major deal, I can just use "off", I just expected False to work based on the docs.

This also does not work, but the result is far weirder and likely a bug (TIFF output also attached):

blob = Image(filename='5-page-pdf.pdf').make_blob()

with Image(blob=blob) as img:
    img.type = "grayscale"
    img.format = "tiff"
    img.compression = "lzw"
    img.alpha_channel = 'off'
    img.save(filename="test.tiff")

As you can see, the alpha channel is removed, but each page is scaled down and repeated 3 times horizontally on its respective page. Very bizarre

The tiff showing the difference between "off" and False. (The version using off looks correct so I didnt bother posting it).

The tiff showing the result of loading an image as a blob and then trying to remove the alpha channel

I am using wand to convert a CMYK PDF to an sRGB png.

Here's a sample CMYK PDF: https://dl.dropboxusercontent.com/u/15672/dndel/in.pdf (sorry, github doesn't allow me to add a pdf directly to an issue).

When I use imagemagick on the command line:

convert in.pdf out1.png

I get:

which is bad. Instead, I have to change the colorspace, so if I do

convert -colorspace sRGB in.pdf out2.png

I get:

which is good.

As far as I can tell, there's no way to set colorspace in Wand.

When I run

from wand.image import Image

with Image(filename='in.pdf') as img:
    with img.convert('png') as converted:
        converted.save(filename='out3.png')

I get the same as the first convert:

It would be nice if Wand supported colorspace-- many PDFs and other print documents use CMYK.

I think these requirements conflict with the setup.py ones.

https://github.com/pypa/pip/issues/1197 <- I added the test deps to extras_require so pip install -e .[test] works as expected.

It would be nice if you could provide a wand image to numpy.asarray() and have it automatically create a properly sized and shaped numpy matrix.

Right now if you try to pass an image in, things bog down and 100% CPU is used. I suspect this is due to the iteration interface on wand.image.Image creating a Color object for every pixel.

To support this, Wand would need to have a __array_interface__ property which returned some information about the data. You can find some information on the interface at http://scipy-lectures.github.com/advanced/advanced_numpy/index.html#array-interface-protocol

Basically, I think something like this would work, at least for raw RGB:

with Image.open(filename="/tmp/some.jpg") as img:
    array = numpy.asarray(img)

And the __array_interface__ would be something like:

@property
def __array_interface__(self):
    return dict(data=self.make_blob("RGB"),
                shape=(self.width, self.height, 3),
                typestr="|u1")

You might want to use the image's internal format so you could support things like RGBA; I'm not sure what the right thing to do here is. (And in that case, shape would be 4 for the third argument.)

A new Sequence object is created as image.sequence property in Image constructor.

Here are some use cases of api:

>>> # Load 4-imaged apple's icon
>>> import requests
>>> r = requests.get('http://apple.com/favicon.ico')
>>> from wand.image import Image
>>> img = Image(blob=r.content, format='ico')
>>> # Get sequenses length
>>> print len(img.sequence)
4
>>> # Print images sizes at sequence
>>> for i in img.sequence:
...     print (i, img.size)
...     
... 
(0L, (32L, 32L))
(1L, (16L, 16L))
(2L, (32L, 32L))
(3L, (16L, 16L))
>>> # Set third image as current (every changes would be applied to it)
>>> img.sequence.index = 2
>>> # Delete all images exept the first one
>>> for i in range(3, 0, -1):
...     img.sequence.index = i
...     del img.sequence.index
...     
... 
>>> print len(img.sequence)
1
>>> # Append a copy of our image before current image
>>> img2 = img.clone()
>>> img.sequence.append(img.clone(), True)
>>> print(len(img.sequence))
2

Hi, In addition to https://github.com/emcconville/wand/pull/607, py.local is deprecated and pathlib or pathlib2 should be used instead of this: https://github.com/emcconville/wand/blob/3eeb75734ac6b6f4d703ba3562c517a20b19be3c/tests/conftest.py#L11

From the pytest doc:

The tmpdir and tmpdir_factory fixtures are similar to tmp_path and tmp_path_factory, but use/return legacy py.path.local_ objects rather than standard :class:pathlib.Path objects.

From the py doc:

Note: The ‘py’ library is in “maintenance mode” and so is not recommended for new projects. Please check out pathlib or pathlib2 for path operations.

I'm running Ubuntu 21.10, Python 3.9.7 and Wand 0.6.7. There is an issue if my python code first does some pyplot stuff and then drawing an rounded rectangle using wand.

The following code does not throw any error message, however the resulting file output.png is empty (NO green rectangle is drawn):

from wand.image import Image
from wand.drawing import Drawing
from wand.color import Color

# do some pyplot stuff
import matplotlib.pyplot as plt
plt.plot([1, 2, 3, 4])

# Do some wand stuff
img = Image(width=850, height=600) # This dimension does NOT work
with Drawing() as drw:
	drw.fill_color = Color('green')
	drw.rectangle(left=0,
				top=0,
				width=img.width,
				height=img.height,
				radius=img.width*0.05)
	drw(img)
img.save(filename='output.png')

Now the strange part. The code works fine if I remove seemingly unrelated pyplot stuff at the beginning:

from wand.image import Image
from wand.drawing import Drawing
from wand.color import Color

# Do some wand stuff
img = Image(width=850, height=600) # This dimension does NOT work
with Drawing() as drw:
	drw.fill_color = Color('green')
	drw.rectangle(left=0,
				top=0,
				width=img.width,
				height=img.height,
				radius=img.width*0.05)
	drw(img)
img.save(filename='output.png')

Now the super crazy thing: I don't need to remove the pyplot stuff, because I can also make the code running by simply changing the dimensions:

from wand.image import Image
from wand.drawing import Drawing
from wand.color import Color

# do some pyplot stuff
import matplotlib.pyplot as plt
plt.plot([1, 2, 3, 4])

# Do some wand stuff
#img = Image(width=850, height=600) # This dimension does not work
img = Image(width=800, height=600) # This dimension does work
with Drawing() as drw:
	drw.fill_color = Color('green')
	drw.rectangle(left=0,
				top=0,
				width=img.width,
				height=img.height,
				radius=img.width*0.05)
	drw(img)
img.save(filename='output.png')

Thanks for your help!

Ref: https://github.com/emcconville/wand/issues/564

I think the following codes are messing with outline.

##########   main func for image rendering   ##############

def export_caption(caption, filename):
    '''Creates a .png file <filename> suitable for bdsup2sub that displays
    the given caption.'''
    img = Image(width=XSIZE, height=YSIZE)
    # by default wand will use the text's baseline as y-coord which is bad,
    # because we'd never know where we end, so set the gravity to north_west
    # Beware: it seems like if we set text_alignment='center' so we could
    # use wand's multiline text feature the gravity setting will be lost
    # irrevocably, so we need to deal with multiline captions ourselves
    img.gravity = 'north_west'
    box = Drawing()
    # Color() handles 'none' by itself, case-independently
    box.fill_color = Color(OPTS.BOXCOLOR)

    draw = Drawing()
    draw.fill_color = Color(OPTS.FILLCOLOR)
    draw.font = FONT
    draw.font_size = FONTSIZE
    draw.text_encoding = ENCODING
    if OPTS.ANTIALIASING:
        draw.text_antialias = True
    else:
        draw.text_antialias = False
    if OUTLINEWIDTH:
        draw.stroke_color = "BLACK"
        draw.stroke_width = OUTLINEWIDTH
        if OPTS.ANTIALIASING:
            draw.stroke_antialias = True
        else:
            draw.stroke_antialias = False

    lines = [line.strip() for line in caption.splitlines() if line.strip()]

    # padding for the text's bounding box
    padx, pady = BOXPADDING

    # calculate the max. allowed text width, and don't forget the outline but
    # ignore the box; keep a few extra px. (1% of XSIZE) in reserve,
    # just to be on the safe side in case we have rounding errors
    max_textwidth = int(round(XSIZE * MAXLINESIZE / 100)) - \
                    2 * OUTLINEWIDTH - int(round(0.01 * XSIZE))

    # now loop through the list of strings, calculate each string's width and
    # if necessary split it in two;
    # repeat this until no line needs to be wrapped
    lines, wmax, line_widths = fix_caption(lines, max_textwidth, img, draw)

    # collect info for which lines we need y box padding;
    # this seems useful to avoid a too big offset between text lines,
    # which may look odd, esp. when the box is transparent
    i = 0
    padinfo = []
    for lw in line_widths:
        a, b, c = 0, 0, 0 # top padding, bottom padding, Y incr. of next line
        if i == 0:
            # the first line needs top padding
            a = pady
        elif line_widths[i-1] < lw:
            # previous line is smaller, so we need top padding
            a = pady
        if i == len(line_widths) - 1:
            # the last line needs bottom padding
            b = pady
        elif line_widths[i+1] < lw:
            # next line is smaller, so we need bottom padding
            b = pady
        elif line_widths[i+1] > lw:
            # next line is bigger, shift the Y-increment by pady
            c = pady
        i += 1
        padinfo.append((a, b, c))

    # width of the box == max. textwidth + 2* (boxpad + outline width)
    wmax = wmax + 2*padx + 2*OUTLINEWIDTH

    # now all lines should be safe to fit on the screen,
    # so start drawing
    Y = 0
    i = 0
    bottom = 0
    for line in lines:
        s = line.strip()
        m = draw.get_font_metrics(img, line)
        w, h = int(m.text_width), int(m.text_height)

        # make sure the smaller line appears centered relative to the
        # bigger line
        if wmax > XSIZE:
            # may happen if an insane x box padding was given;
            # make sure the text will be centered anyway, the excess box
            # will be chopped off later
            X = int((XSIZE - w) / 2)
        else:
            X = int((wmax - w) / 2)

        a, b, c = padinfo[i]
        # draw the bounding box
        if OPTS.BOXCOLOR.lower() != 'none':
            box.rectangle(X - OUTLINEWIDTH - padx,
                          Y - OUTLINEWIDTH - a,
                          X + w + OUTLINEWIDTH + padx,
                          Y + h + OUTLINEWIDTH + b)
            box(img)

        # draw the text and shift the Y position to the correct
        # coordinate for the following line (if any)
        draw.text(X, Y, s)
        draw(img)

        # store bottom Y coord
        bottom = Y + h + OUTLINEWIDTH + b
        Y = bottom + c + OUTLINEWIDTH + 2 # 1 px. extra so the boxes won't overlap
        i += 1

    x0, y0, x1, y1 = 0, 0, wmax + 1, bottom + 1
    # cut the text box from the surrounding void
    if y1 > YSIZE:
        # may happen if the font is too big or an insane y boxpadding is set
        printerror('\nError: generated subtitle image exceeds screen height')
        printerror('The offending subtitle text was:')
        for line in caption.splitlines():
            printerror('    ' + line)
        printerror('This subtitle will not look as expected.')
        y1 = YSIZE

    if x1 > XSIZE:
        # may happen if the user set insane boxpadding values, making
        # it impossible to wrap the text to fit into one line
        printerror('\nError: generated subtitle image exceeds screen width')
        printerror('The offending subtitle text was:')
        for line in caption.splitlines():
            printerror('    ' + line)
        printerror('This subtitle may not look as expected.')
        x1 = XSIZE

    img.crop(x0, y0, x1, y1)

    # write file and return the size for our xml content
    img.format = 'png'
    img.save(filename=filename)
    box.destroy()
    draw.destroy()
    img.destroy()
    return wmax+1, bottom+1

Whole file: srt2vobsub.zip

I am ending up with an error when I use the below code when its run in docker.

try:
     with skew(blob=img_png) as img_png_1:           
         img_png_1.deskew(0.5 * img_png_1.quantum_range)
         pil_image = Image.open(io.BytesIO(img_png_1.make_blob("png")))            
         img_png = image_to_byte_array(pil_image)            
     log.info("Corrected the Skewness for %s", file_name)
 except Exception as e:
     log.exception(f'Exception:{e} occured while Corrected the Skewness for {file_name}')
     pass

Error Message: wand.exceptions.WandRuntimeError: MagickReadImage returns false, but did not raise ImageMagick exception. This can occur when a delegate is missing, or returns EXIT_SUCCESS without generating a raster.

When I run this in local windows, I don't see any error, but failed in docker(linux). It keeps working for few images and fails for few. Please advice.

In Ubuntu 18.04 LTS, the libmagickwand-6.q16-3 package installs the wand libraries at:

• /usr/lib/x86_64-linux-gnu/libMagickCore-6.Q16.so.3
• /usr/lib/x86_64-linux-gnu/libMagickWand-6.Q16.so.3

import wand finds these, but it takes ~700ms to do because it runs ctypes.find_library (apparently several times?)

Looking at the source code, it seemed like we could avoid this by setting MAGICK_HOME and WAND_MAGICK_LIBRARY_SUFFIX in our containers.

However, this doesn't actually help because wand hardcodes the .so extension, and .so.3 doesn't match.

Perhaps wand should handle a couple of environment variables containing the full paths to the libraries?

The following code behaves as expected:

from wand.image import Image # MUST BE RUN BEFORE `from transformers import ... ` or will produce silent failure
from transformers import PreTrainedTokenizer # MUST BE RUN AFTER `from wand.image import... ` or will produce silent failure

with Image(filename='/path/to/filename.pdf') as wand_img:
    for page in wand_img.sequence:
        page_img = Image(page)
        print('foo')

However, if the libraries are imported in the opposite order, Wand will silently terminate on line 6:

from transformers import PreTrainedTokenizer # MUST BE RUN AFTER `from wand.image import... ` or will produce silent failure
from wand.image import Image # MUST BE RUN BEFORE `from transformers import ... ` or will produce silent failure

with Image(filename='/path/to/filename.pdf') as wand_img:
    for page in wand_img.sequence:
        page_img = Image(page) # crashes
        print('foo')

transformers version 4.6.1 Wand version 0.6.6