TCIP download script

get_tcip.py

# Usage:
# For a particular TCIP book
#     (such as the one whose viewer URL begins http://access.bl.uk/item/viewer/ark:/81055/vdc_100086389865.0x000001)
# run
# python3 get_tcip.py vdc_100086389865


import io
import requests
import socket
import os
import tqdm
from time import sleep
from queue import Queue
from PIL import Image
from sys import argv, exit
from threading import Thread


def download_tile():
	pagenum = 0
	image_width = 0
	image_height = 0
	i = 0
	j = 0
	image_url = ''
	while True:
		(pagenum,image_width,image_height,i,j,image_url) = q.get()
		while True:
			try:
				image_request = requests.get(image_url)
				im = Image.open(io.BytesIO(image_request.content))
				break
			except Exception as e:
				print(pagenum,image_width,image_height,i,j,image_url,e)
				sleep(3)
				continue
		new_im.paste(im,(i,j))
		q.task_done()

concurrent_threads = 8
q = Queue(3*concurrent_threads)
for i in range(concurrent_threads):
	t = Thread(target=download_tile)
	t.daemon = True
	t.start()

ark_id = argv[1]

r = requests.get(f"https://api.bl.uk/metadata/iiif/ark:/81055/{ark_id}.0x000001/manifest.json")
rjson = r.json()
if(rjson["sequences"] == []):
	print("No images available for", rjson["metadata"][0]["value"], rjson["metadata"][1]["value"])
	exit(0)
else:
	try:
		os.mkdir(ark_id)
	except FileExistsError:
		pass
os.chdir(ark_id)
pagenums = len(rjson["sequences"][0]["canvases"])
print("Starting", rjson["metadata"][0]["value"], rjson["metadata"][1]["value"])

tq = tqdm.tqdm(range(1,pagenums+1))
for pagenum in tq:
	toplevelimg = rjson["sequences"][0]["canvases"][pagenum-1]["images"][0]["resource"]["service"]["@id"]
	out_image = f'image_{pagenum}_.jpg'
	if(os.path.isfile(out_image)):
		continue
	cur_rjson = rjson["sequences"][0]["canvases"][pagenum-1]
	image_width = cur_rjson['width']
	image_height = cur_rjson['height']
	tile = max(cur_rjson['images'][0]['resource']['service']['tiles'], key=lambda x: x['width'])
	tile_width = tile['width']

	new_im = Image.new('RGB',(image_width,image_height))

	try:
		for i in range(0,image_width,tile_width):
			for j in range(0,image_height,tile_width):
				cur_width = min(tile_width,image_width-i)
				cur_height = min(tile_width,image_height-j)
				tile_url = f"{toplevelimg}/{i},{j},{tile_width},{tile_width}/{cur_width},{cur_height}/0/default.jpg"
				q.put((pagenum,image_width,image_height,i,j,tile_url))
		q.join()
	except KeyboardInterrupt:
		exit(1)

	new_im.save(out_image)
	tq.set_postfix(out_image=out_image)
	sleep(3) # Adjust this at will.
os.chdir("..")