Extract Images / use processFulltextAssetDocument service

[mariAFLJ]

Hi GROBID team,

I want to extract the images from pdfs together with the conversion of the PDF (to TEI-XML) using GROBID. Using the batch command everything works fine.

The grobid client is also working as it should, but I miss the function of extracting the images. Is it possible to add the "processFulltextAssetDocument" service to the python grobid-client to extract images as well? In the documentation is written: "There is a web service doing the same, returning everything in a big zip file, processFulltextAssetDocument, still usable but deprecated." When I use it as service

service="processFulltextAssetDocument"

calling

"client.process(service, input, output)"

it fails with ERROR 406.

Thanks in advance!

[thejiangcj]

I never use it to extract image, but you can extract image by coordinate and pymupdf:

for example, you can reference my code:

import fitz
def get_image_from_corrd(pdf_path, coord, to_path, fig_type):
    doc = fitz.open(pdf_path)

    page_map = {}
    for cor in coord:
        cor2 = cor.split(",")
        page = int(cor2[0])
        x1 = float(cor2[1])
        y1 = float(cor2[2])
        x2 = float(cor2[1])+float(cor2[3])
        y2 = float(cor2[2])+float(cor2[4])

        if page_map.get(page) is None:
            page_map[page] = {
                "x1": x1,
                "y1": y1,
                "x2": x2,
                "y2": y2
            }
        else:
            page_map[page] = {
                "x1": min(page_map[page]["x1"], x1),
                "y1": min(page_map[page]["y1"], y1),
                "x2": max(page_map[page]["x2"], x2),
                "y2": max(page_map[page]["y2"], y2)
            }

    to_paths = []
    for page_number, data in page_map.items():

        # Get the page (page_number is 1-based, fitz uses 0-based indexing)
        page_number, x, y, x2, y2 = int(page_number), float(data["x1"]), float(data["y1"]), float(data["x2"]), float(data["y2"])
        page = doc.load_page(int(page_number) - 1)

        # Define the rectangle for the image extraction
        rect = fitz.Rect(x, y, x2, y2)

        # Clip the page using the rectangle and get the image
        pix = page.get_pixmap(clip=rect)

        # Save the image to the specified path
        if fig_type == "figure":
            pix.save(os.path.join(to_path, f"{fig_type}_{self.fig_num}.png"))
            to_paths.append(os.path.join(to_path, f"{fig_type}_{self.fig_num}.png"))
            self.fig_num += 1
        elif fig_type == "table":
            pix.save(os.path.join(to_path, f"{fig_type}_{self.table_num}.png"))
            to_paths.append(os.path.join(to_path, f"{fig_type}_{self.table_num}.png"))
            self.table_num += 1

    # Close the document
    doc.close()
    return to_paths

get_image_from_corrd(pdf_path, coords.split(";"), img_path, types)