ImageForm 类型的对象不是 JSON 可序列化的

Question

我想将图像处理从视图发布到 celery，但它向我展示了一个与 JSON 相关的问题“ImageForm 类型的对象不是 JSON 可序列化的”

这是来自models.py的代码

class ImageModel(models.Model):
    title = models.CharField(max_length=20, null=True)
    sec_title = models.CharField(max_length=20, null=True)
    img = models.ImageField(upload_to='upload_p/', blank=False)
    slug = models.SlugField(max_length=250, null=True)

    def delete(self, *args, **kwargs):
        self.img.delete()
        super().delete(*args, **kwargs)

表格.py

class ImageForm(forms.ModelForm):
    class Meta:
        model = ImageModel
        fields = ('title', 'img', 'sec_title')

视图.py

def upload_image(request):
    if request.method == 'POST':
        form_w = ImageForm(request.POST, request.FILES)
        if form_w.is_valid():
            water_mark.delay(form_w)
    else:
        form = ImageForm()
    return render(request, 'Luki/upload_img.html', {
        'form': form,
    })

任务.py

@shared_task
def water_mark(form_w):
    instance = form_w.save(commit=False)
    cd = form_w.cleaned_data['img']

    if instance.img:
        im = Image.open(instance.img)

        width, height = im.size
        draw = ImageDraw.Draw(im)
        text = "TEST WATERMARK"

        font = ImageFont.truetype('arial.ttf', 36)
        textwidth, textheight = draw.textsize(text, font)

        # calculate the x,y coordinates of the text
        margin = 10
        x = width - textwidth - margin
        y = height - textheight - margin

        draw.text((x, y), text, font=font)


        thumb_io = BytesIO()
        print('BYTES IO: ', thumb_io)
        im.save(thumb_io, im.format, quality=100)
        instance.img.save(str(cd), ContentFile(thumb_io.getvalue()), save=False)

    instance.save()

    return redirect('Luki:gallery')

当然，所有的库都被导入，并且没有 celera 的视图中的代码被执行，它看起来像这样。 这样可以使您在表单中添加的照片获得水印、保存并带您进入图库。

视图.py

def upload_image(request):
    if request.method == 'POST':
        form_w = ImageForm(request.POST, request.FILES)
        if form_w.is_valid():
            instance = form_w.save(commit=False)
            cd = form_w.cleaned_data['img']

            if instance.img:
                im = Image.open(instance.img)

                width, height = im.size
                draw = ImageDraw.Draw(im)
                text = "TEST WATERMARK"

                font = ImageFont.truetype('arial.ttf', 36)
                textwidth, textheight = draw.textsize(text, font)

                # calculate the x,y coordinates of the text
                margin = 10
                x = width - textwidth - margin
                y = height - textheight - margin

                # draw watermark in the bottom right corner
                draw.text((x, y), text, font=font)

                thumb_io = BytesIO()
                im.save(thumb_io, im.format, quality=100)
                instance.img.save(str(cd), ContentFile(thumb_io.getvalue()), save=False)

            instance.save()


            return redirect('Luki:gallery')
    else:
        form_w = ImageForm()
    return render(request, 'Luki/upload_img.html', {
        'form_w': form_w,
    })

Answer 1

根据此博客，您不能在 celery 中传递对象。

由于 Celery 是一个分布式系统，你无法知道任务会在哪个进程，甚至在哪台机器上运行。 所以你不应该将 Django 模型对象作为参数传递给任务，从数据库中重新获取对象几乎总是更好，因为可能涉及竞争条件。

因此，在您的情况下，将图像保存在模型中，并将模型的 pk 传递给 celery 任务，然后再次从模型中获取这些详细信息。

Answer 2

这是带有tasks.py的代码的样子，但我仍然有错误。

def upload_image(request):
    if request.method == 'POST':
        form_w = ImageForm(request.POST, request.FILES)
        # files = request.FILES.getlist('img')
        if form_w.is_valid():
            wm = form_w.save()
            water_mark.delay(wm.pk)


            return redirect('Luki:gallery')
    else:
        form_w = ImageForm()
    return render(request, 'Luki/upload_img.html', {
        'form_w': form_w,
    })

和tasks.py

def water_mark(wm):
    instance = ImageModel.objects.get(id=wm)

    # im = Image.open(instance.img)
    # print('im before saving: ', im)

    # Writes the image correctly
    # im.save('Luki/media/upload_p/zzzta.jpg')


    if instance.img:
        # upload_p/img-8945.jpg
        print('img before opening: ', instance.img)
        # returns upload_p/img-8945.jpg

        im = Image.open(instance.img)
        print('img after: ', im)
        # returns <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=1920x1080 at 0x5AD75F0>


        width, height = im.size
        draw = ImageDraw.Draw(im)
        text = "TEST WATERMARK"

        font = ImageFont.truetype('arial.ttf', 36)
        textwidth, textheight = draw.textsize(text, font)

        # calculate the x,y coordinates of the text
        margin = 10
        x = width - textwidth - margin
        y = height - textheight - margin

        # draw watermark in the bottom right corner
        draw.text((x, y), text, font=font)

        thumb_io = BytesIO()
        print('BYTES IO: ', thumb_io)
        im.save(thumb_io, im.format, quality=100)

        # FieldFile.save(name, content, save=True)
        instance.img.save(str('zzjca.jpg'), ContentFile(thumb_io.getvalue()), save=False)

    instance.save()

它返回一个错误File "f:\\python\\portfoliolk\\venv\\lib\\site-packages\\eventlet\\greenio\\base.py", line 100, in set_nonblocking raise NotImplementedError("set_nonblocking() on a file object " NotImplementedError: set_nonblocking() on a file object with no setblocking() method (Windows pipes don't support non-blocking I/O)

不知道是不是windows的问题...