dell_standalone.py

"""Parsers for Dell LCA PDF.

See an example here https://i.dell.com/sites/csdocuments/CorpComm_Docs/en/carbon-footprint-wyse-3030.pdf
"""

import logging
import re
import datetime
from typing import BinaryIO, Iterator
import os

from tools.parsers.lib import data
from tools.parsers.lib.image import crop, find_text_in_image, image_to_text
from tools.parsers.lib import loader
from tools.parsers.lib import pdf
from tools.parsers.lib import text
import argparse
import requests
import io
from tools.parsers import dell_laptop


argparser = argparse.ArgumentParser()
result = data.DeviceCarbonFootprintData()
argparser.add_argument("-s", "--source", required=True, help="URL to .pdf to be converted")
args = vars(argparser.parse_args())
pdf_path = args["source"]
url = ""
rm_tempfile = False
if re.search('http(s)*\:\/\/*.', pdf_path):
    open('./tempfile.pdf', 'wb').write(requests.get(pdf_path).content)
    url=pdf_path
    pdf_path = "./tempfile.pdf"
    rm_tempfile = True

with open(pdf_path, 'rb') as fh:
     for result in dell_laptop.parse(io.BytesIO(fh.read()), url):
        result.data['sources_hash']=data.md5_file(pdf_path)
        result.data['sources']=url
        print(result.as_csv_row())
if rm_tempfile:
    os.remove('./tempfile.pdf')
quit()