Code snippet for visualizing bounding boxes

visualize_bbox.py

import requests
import os.path
from pdf2image.pdf2image import convert_from_path
from PIL import ImageDraw


pdf_path = os.path.expanduser("~/Downloads/research-highlight.pdf")
upload_form = requests.post("https://v1.api.reducto.ai/upload").json()
requests.put(upload_form["presigned_url"], data=open(pdf_path, "rb"))
response = requests.post(
    "https://v1.api.reducto.ai/parse",
    json={"document_url": upload_form["file_id"]},
    headers={"Authorization": f"Bearer {os.environ['REDUCTO_API_KEY']}"},
)

blocks = [
    block for chunk in response.json()["result"]["chunks"] for block in chunk["blocks"]
]

images = convert_from_path(pdf_path)

for page_num, image in enumerate(images, start=1):
    draw = ImageDraw.Draw(image)
    for block in blocks:
        if block["bbox"]["page"] == page_num:
            bbox = block["bbox"]
            draw.rectangle(
                (
                    bbox["left"] * image.width,
                    bbox["top"] * image.height,
                    (bbox["left"] + bbox["width"]) * image.width,
                    (bbox["top"] + bbox["height"]) * image.height,
                ),
                outline="red",
                width=10,
            )

    image.show()