I want to extract drawings as images

[rafayaar]

I am building a solution where when provided with PDF it can parse it and make it editable

I have extract text and images along with there bounding box's

What I want to achieve is, to actually extract drawings / shapes / graphics which is left on pdf and save them as png images
Of course, for this as well I would want bbox.

One of the approaches I tried is to use .get_drawings() function and map it on pixmap and save as Pixel file to image. But cant do it

Help required

[JorjMcKie]

If your vector graphics are made of drawing paths that are close enough to each other, the Page method cluster_drawings should deliver what you want.

Its parameters allow you some flexibility:

• You can preprocess the get_drawings() output to build some desired subset, which you can hand into the method. Otherwise, this parameter is also useful for avoiding multiple vector graphic extractions.
• You can control the "sensitivity" for regarding drawings to be "neighbours" by manipulating the tolerance values. The default is 3 points: two paths in get_drawings() are assumed to belong to the same cluster if their minimum distance is below 3.

[JorjMcKie]

I forgot to mention the obvious:
cluster_drawings() delivers a list of rectangles. You can take each of these bboxes and make an image from it, that can be saved:

bboxes = page.cluster_drawings()
for i, bbox in enumerate(bboxes):
    pix = page.get_pixmap(clip=bbox)
    filename = f"{doc.name}-{page.number}-{i}.png"
    pix.save(filename)

So for the 5th vector graphics on page 3, a PNG image with the name "input.pdf-3-5.png" will be saved.

[JorjMcKie]

... ignoring 0-based numbers above 🤷‍♂️