bilibili/wrapper.py

import bilibili, json, requests
from alive_progress import alive_bar
from cprint import cprint

cprint.warn("PLEASE NOTE:\n\twrapper.py is meant to be used with Warcprox. **NO DATA IS SAVED ANYWHERE.**")

QUEUED = []
QUEUED_IMAGES = []

PROFILE = input("Please enter the profile ID: ")

list(bilibili.userScraper(PROFILE)) # get metadata

cprint.info("Scraping album pagination for URLs...")

with alive_bar() as bar:
    for images, _ in bilibili.albumScraper(PROFILE):
        for image in images["data"]["items"]:
            QUEUED_IMAGES.append((image["dyn_id"], "IMAGE_POST"))
            bar()

cprint.info("Now downloading image metadata...")

with alive_bar(total=len(QUEUED_IMAGES)) as bar:
    for item, typee in QUEUED_IMAGES:
        post = bilibili.postScraper(item)["data"]["card"]
        try:
            QUEUED.append((post["display"]["attach_card"]["cover_url"], "ATTACH_CARD_COVER_URL"))
        except KeyError:
            pass
        try:
            for addoncard in post["display"]["add_on_card_info"]:
                QUEUED.append((addoncard["attach_card"]["cover_url"], "ADDON_CARD_COVER_URL"))
        except KeyError:
            pass
        QUEUED.append((post["desc"]["user_profile"]["info"]["face"], "PFP"))
        card = json.loads(post["card"])
        for picture in card["item"]["pictures"]:
            QUEUED.append((picture["img_src"], "IMAGE"))
        QUEUED.append((card["user"]["head_url"], "PFP"))
        bar()

cprint.info("Finished image collection.")
cprint.info("Downloading articles...")

with alive_bar() as bar:
    for articles, __ in bilibili.articleScraper(PROFILE):
        if not __.startswith("MASTER_LIST_"):
            continue
        for article in articles["data"]["articles"]:
            for url in article["image_urls"]:
                QUEUED.append((url, "ArticleImageUrl"))
                bar()
            for otherurl in article["origin_image_urls"]:
                QUEUED.append((url, "ArticleOriginImgUrl"))
                bar()

cprint.info("Finished article download.")
cprint.info("Downloading queued images and posts...")

with alive_bar(dual_line=True, total=len(QUEUED)) as bar:
    for item, typee in QUEUED:
        bar.text = f"{image} ({typee})"
        requests.get(item)
        bar()