Parsing pdfs

from llama_index.langchain_helpers.text_splitter import SentenceSplitter
from gpt_index.node_parser.simple import SimpleNodeParser

node_parser = SimpleNodeParser(text_splitter=SentenceSplitter())
service_context = Service_context.from_defaults(node_parser=node_parser)

import json
import fitz

def get_block_type(block):
    if block[6] == 0:
        return "text"
    elif block[6] == 1:
        return "heading"
    elif block[6] == 2:
        return "image"
    elif block[6] == 3:
        return "list"
    else:
        return "unknown"

def get_page_blocks(page):
    blocks = []
    for block in page.get_text("blocks"):
        block_type = get_block_type(block)
        if block_type == "text" or block_type == "heading":
            text = block[4]
        elif block_type == "list":
            text = [span["text"] for line in block["lines"] for span in line["spans"]]
        else:
            text = ""
        blocks.append({
            "type": block_type,
            "text": text,
            "coordinates": block[:4]
        })
    return blocks

def get_pdf_content(pdf_path):
    doc = fitz.open(pdf_path)
    content = {
        "document_metadata": {
            "title": doc.metadata["title"],
            "author": doc.metadata["author"],
            "creation_date": doc.metadata["creationDate"],
            "modification_date": doc.metadata["modDate"]
        },
        "pages": []
    }
    for page in doc:
        content["pages"].append({
            "page_number": page.number,
            "page_metadata": {
                "width": page.rect.width,
                "height": page.rect.height,
                
            },
            "blocks": get_page_blocks(page)
        })
    doc.close()
    return content

if __name__ == '__main__':
    pdf_content = get_pdf_content('test.pdf')
    json_content = json.dumps(pdf_content, indent=4)
    #write json content to output.json file
    with open('output.json', 'w') as f:
        f.write(json_content)