Extracting images

[tulas75]

Hi guys.

I tried to extract images from pdfs, but I don't understand how to manipulate the output format from images.

I used the following code

import openparse
temp_file = 'myfile.pdf'
parser = openparse.DocumentParser(
table_args={
"parsing_algorithm": "pymupdf",
"table_output_format": "markdown"
}
)
import json

chunks = parsed_basic_doc.model_dump_json()
chunks = json.loads(chunks)
print (chunks)

Printing the chunks I can extract all the information(text, tokens, bbox, etc etc) from nodes but for image I don't understand the format. this is the cutted output

{'embedding': None, 'node_id': 'df849f05-4c01-45e4-8ae1-407f4cf6c78e', 'variant': ['image'], 'tokens': 512, 'images': [{'text': '', 'bbox': {'page': 0, 'page_height': 842.0, 'page_width': 596.0, 'x0': 439.5, 'y0': 595.25, 'x1': 612.75002, 'y1': 833.0}, 'image': 'QEBAPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ozs7Ozs7AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
...
...

The pdf has been generated from Google Documents.

How could extract the image from the node?

Thank you
.andrea

[tulas75]

It seems that problem is related to PNG format. Documents with jpeg images are parsed correctly.
If I use the tool(pdf2txt.py) of the same library(pdfminer)used in openparse seems to work correctly, extracting all the images(jpegs and pngs)

pdf2txt.py myfile.pdf --output-dir .

Any tips?
.a

[tulas75]

just noticed that "pdf2txt.py myfile.pdf --output-dir ." transforms all the png images in jpg.

[tulas75]

I have just installed new release 0.6.1, but the problem with png is even present. Could I please have a feedback on it, if I am wrong or the problem that I encountered is a bug?

thank you
.t