247 lines
5.6 KiB
Python
247 lines
5.6 KiB
Python
import logging
|
|
from typing import Optional
|
|
|
|
import requests
|
|
from bs4 import BeautifulSoup
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
IPDB_URL = "https://www.ipdb.plus/IPDb/puzzle.php?id={id}"
|
|
|
|
|
|
def get_first(key: str, result: dict) -> str:
|
|
obj = ""
|
|
if obj_list := result.get(key):
|
|
obj = obj_list[0]
|
|
return obj
|
|
|
|
|
|
def get_title_from_soup(soup) -> str:
|
|
title = ""
|
|
try:
|
|
title = soup.find("h1").get_text()
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
return title
|
|
|
|
|
|
def get_manufacturer_from_soup(soup) -> str:
|
|
manufacturer = ""
|
|
try:
|
|
manufacturer = (
|
|
soup.find(class_="infobox").div.contents[0].split("|")[0].strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return manufacturer
|
|
|
|
|
|
def get_pieces_count_from_soup(soup) -> int:
|
|
pieces = 0
|
|
try:
|
|
pieces = int(
|
|
soup.find(class_="infobox")
|
|
.div.contents[0]
|
|
.split("|")[1]
|
|
.split(" ")[1]
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return pieces
|
|
|
|
|
|
def get_publish_year_from_soup(soup) -> int:
|
|
year = 1900
|
|
try:
|
|
year = int(
|
|
soup.find(class_="infobox").div.contents[0].split("|")[2].strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return year
|
|
|
|
|
|
def get_material_from_soup(soup) -> str:
|
|
material = ""
|
|
try:
|
|
material = (
|
|
soup.findAll(class_="infobox")[1]
|
|
.findAll("div")[1]
|
|
.contents[2]
|
|
.split("|")[-1]
|
|
.strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return material
|
|
|
|
|
|
def get_cut_style_from_soup(soup) -> str:
|
|
cut_style = ""
|
|
try:
|
|
cut_style = (
|
|
soup.findAll(class_="infobox")[1]
|
|
.findAll("div")[1]
|
|
.contents[2]
|
|
.split("|")[-2]
|
|
.strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return cut_style
|
|
|
|
|
|
def get_barcode_from_soup(soup) -> str:
|
|
barcode = ""
|
|
try:
|
|
barcode = (
|
|
soup.findAll(class_="infobox")[2]
|
|
.findAll("div")[0]
|
|
.contents[0]
|
|
.split("|")[-1]
|
|
.split(":")[-1]
|
|
.strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
if not barcode:
|
|
try:
|
|
barcode = (
|
|
soup.findAll(class_="infobox")[3]
|
|
.findAll("div")[0]
|
|
.contents[0]
|
|
.split("|")[-1]
|
|
.split(":")[-1]
|
|
.strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return barcode
|
|
|
|
|
|
def get_image_url_from_soup(soup) -> str:
|
|
url = ""
|
|
try:
|
|
url = (
|
|
soup.find(class_="image-container").contents[0].contents[0]["src"]
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return url
|
|
|
|
|
|
def get_orientation_from_soup(soup) -> str:
|
|
orientation = ""
|
|
try:
|
|
orientation = (
|
|
soup.findAll(class_="infobox")[1]
|
|
.findAll("div")[1]
|
|
.contents[0]
|
|
.split("|")[0]
|
|
.strip()
|
|
)
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return orientation
|
|
|
|
|
|
def get_dimensions_from_soup(soup) -> str:
|
|
dimensions = ""
|
|
try:
|
|
height = (
|
|
soup.findAll(class_="infobox")[1]
|
|
.findAll("div")[1]
|
|
.contents[1]["data-bs-height"]
|
|
)
|
|
width = (
|
|
soup.findAll(class_="infobox")[1]
|
|
.findAll("div")[1]
|
|
.contents[1]["data-bs-width"]
|
|
)
|
|
|
|
dimensions = "x".join([height, width])
|
|
except AttributeError:
|
|
pass
|
|
except ValueError:
|
|
pass
|
|
except IndexError:
|
|
pass
|
|
return dimensions
|
|
|
|
|
|
def get_rating_from_soup(soup) -> str:
|
|
rating = ""
|
|
try:
|
|
rating = float(
|
|
soup.find(class_="num").get_text().strip("(").strip(")")
|
|
)
|
|
except AttributeError:
|
|
rating = None
|
|
except ValueError:
|
|
rating = None
|
|
return rating
|
|
|
|
|
|
def get_puzzle_from_ipdb_id(ipdb_id: str) -> dict:
|
|
puzzle_url = IPDB_URL.format(id=ipdb_id)
|
|
headers = {"User-Agent": "Vrobbler 0.11.12"}
|
|
response = requests.get(puzzle_url, headers=headers)
|
|
puzzle_dict = {"ipdb_id": ipdb_id}
|
|
|
|
if response.status_code != 200:
|
|
logger.warn(
|
|
"Bad response from untappd.com", extra={"response": response}
|
|
)
|
|
return puzzle_dict
|
|
|
|
soup = BeautifulSoup(response.text, "html.parser")
|
|
puzzle_dict["title"] = get_title_from_soup(soup)
|
|
puzzle_dict["orientation"] = get_orientation_from_soup(soup)
|
|
puzzle_dict["manufacturer"] = get_manufacturer_from_soup(soup)
|
|
puzzle_dict["pieces_count"] = get_pieces_count_from_soup(soup)
|
|
puzzle_dict["ipdb_image_url"] = get_image_url_from_soup(soup)
|
|
puzzle_dict["dimensions_in_inches"] = get_dimensions_from_soup(soup)
|
|
puzzle_dict["publish_year"] = get_publish_year_from_soup(soup)
|
|
puzzle_dict["material"] = get_material_from_soup(soup)
|
|
puzzle_dict["cut_style"] = get_cut_style_from_soup(soup)
|
|
puzzle_dict["barcode"] = get_barcode_from_soup(soup)
|
|
return puzzle_dict
|