merfrei

"""
Scrap the full list of active vessels form MAERSK
"""

import asyncio
import re
import csv
from typing import IO
import requests
from w3lib.url import add_or_replace_parameter

merfrei

def fix_json_unescaped_quotes(json_str):
    """For example:
    "name":"Under Armour Women's On 3" Shorts"
    =>
    "name":"Under Armour Women's On 3\" Shorts"
    """
    new_json_str = ''
    is_open = False  # When : is found in a declaration
    is_started = False  # When " is found after :
    for j_ix, j_chr in enumerate(json_str):

merfrei

# Custom blocking detection and retries with puppeteer API example

import logging

from functools import wraps

logger = logging.getLogger(__name__)


BLOCKED_CODES = [403, 500, 502, 503, 504]

merfrei

#!/usr/bin/env python

import urllib2
import json

def check_envio(num_envio):
    url = 'http://www.enviosoca.com/Tracking/GetLastTrackingData/default.aspx?trackingNumber=%s' % (num_envio,)
    resp = urllib2.urlopen(url)
    json_data = resp.read()
    return json.loads(json_data)

merfrei

# -*- coding: utf-8 -*-

from scrapy.spider import BaseSpider
from scrapy.selector import HtmlXPathSelector
from scrapy.item import Item, Field
from scrapy.contrib.loader import XPathItemLoader
from scrapy.contrib.loader.processor import MapCompose, TakeFirst
from scrapy.contrib.exporter import CsvItemExporter

from scrapy.conf import settings