format='%(asctime)s - %(name)s - %(levelname)s: %(message)s',
level=logging.INFO
)
logger = logging.getLogger(__name__)
LIMIT = 500
DELAY = 1.0
RETRIES = 3
PROVIDER = prov.NYPL_DEFAULT_PROVIDER
BASE_ENDPOINT = "http://api.repo.nypl.org/api/v1/items/search"
METADATA_ENDPOINT = "http://api.repo.nypl.org/api/v1/items/item_details/"
NYPL_API = os.getenv("NYPL_API_KEY")
TOKEN = f"Token token={NYPL_API}"
delay_request = DelayedRequester(delay=DELAY)
image_store = ImageStore(provider=PROVIDER)
DEFAULT_QUERY_PARAM = {
"q": "CC_0",
"field": "use_rtxt_s",
"page": 1,
"per_page": LIMIT
}
HEADERS = {
"Authorization": TOKEN
}
IMAGE_URL_DIMENSIONS = [
"g", "v", "q", "w", "r"
]
import time
import logging
import re
import json
from urllib.parse import urlparse, parse_qs
DELAY = 1.0 #time delay (in seconds)
PROVIDER = 'rawpixel'
FILE = 'rawpixel_{}.tsv'.format(int(time.time()))
logging.basicConfig(
format='%(asctime)s: [%(levelname)s - RawPixel API] =======> %(message)s',
level=logging.INFO)
delayed_requester = DelayedRequester(DELAY)
image_store = ImageStore(provider=PROVIDER, output_file=FILE)
def sanitizeString(_data):
if _data is None:
return ''
else:
_data = str(_data)
_data = _data.strip()
_data = _data.replace('"', "'")
_data = re.sub(r'\n|\r', ' ', _data)
#_data = re.escape(_data)
backspaces = re.compile('\b+')
_data = backspaces.sub('', _data)