专栏：⩑〘2024.02.06〙「第三期」动漫 赛博 科技 AI 壁纸专栏
专栏60张壁纸 云盘60张壁纸
…", "contentLength": 291, "imageUrls": ["https://i0.hdslb.com/bfs/article/c7e39b9fed39c7d186bc8e5e29f2aa7b286431045.jpg", "…"], "discoveredVia": "url", "searchQuery": null, "scrapedAt": "2026-10-05T12:41:13+00:00" } ``` Here is a profile row: ```json { "type": "profile", "authorMid": "244503977", "authorName": "桑葚青梅茶", "authorAvatar": "https://i2.hdslb.com/bfs/face/e22769fd9e1fc4de67ef45d37d99bf615bd6ae20.jpg", "authorSign": "人要有盼头、才会健康", "spaceUrl": "https://space.bilibili.com/244503977", "scrapedAt": "2026-10-05T12:42:21+00:00" } ``` The **Output** tab has four views: **Overview**, **Articles** (text, tags and images), **Engagement** (all the numbers in columns) and **Creators**. You can download the dataset in various formats such as JSON, HTML, CSV, or Excel. #### Data fields | Field | Description | |-------|-------------| | `type` | `article` or `profile` | | `authorMid` / `authorName` | Creator UID and display name | | `articleId` / `opusId` | Bilibili cv article ID and opus ID | | `url` | Link to the article | | `title` / `summary` | Article title and short summary (articles found by creator or keyword) | | `publishedAt` | Publish time, ISO 8601, China time (UTC+8) | | `category` / `tags` | Article category (articles found by creator or keyword) and tags | | `stats` | `view`, `like`, `comment`, `coin`, `favorite`, `forward` (shares) | | `wordCount` | Word count (articles found by creator) | | `contentText` / `contentHtml` | Full text and clean HTML of the article | | `contentLength` | Length of the text in characters. `0` for image-only articles | | `imageUrls` | Image URLs in the article | | `discoveredVia` / `searchQuery` | How the article was found (`creator`, `search` or `url`) and the keyword | | `authorAvatar` / `authorSign` / `spaceUrl` | Profile rows only: avatar, bio and profile URL | | `scrapedAt` | When your run collected the row (UTC) | ### How to use Bilibili Article Scraper #### In Apify Console 1. Open the Actor, fill in the **Input** tab and click **Start**. 2. Watch the run's progress in the log, then open the **Output** tab. 3. Click **Save as a new task** to run the same setup again with one click. On the task, add a **Schedule** to check your creators or keywords every day or week. #### With the Apify API (HTTP) Run the Actor and get its results in one request. Replace `YOUR_API_TOKEN` with your token from **Settings → API & Integrations** in Apify Console. ```bash curl -X POST "https://api.apify.com/v2/acts/hgservices~bilibili-columns-scraper/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \ -H "Content-Type: application/json" \ -d '{"creators": ["244503977"], "maxArticles": 20}' ``` Add `&format=csv` or `&format=xlsx` to the URL to get CSV or Excel instead of JSON. For long runs, start the run with `POST /v2/acts/hgservices~bilibili-columns-scraper/runs`, and then read the items of the run's default dataset when it finishes. #### With the JavaScript client Install the client with `npm install apify-client`, then run: ```javascript import { ApifyClient } from 'apify-client'; const client = new ApifyClient({ token: 'YOUR_API_TOKEN' }); const run = await client.actor('hgservices/bilibili-columns-scraper').call({ searchQueries: ['人工智能'], maxArticles: 50, }); const { items } = await client.dataset(run.defaultDatasetId).listItems(); for (const item of items.filter((i) => i.type === 'article')) { console.log(item.publishedAt, item.title, item.stats.view, item.url); } ``` #### With the Python client Install the client with `pip install apify-client` (version 3.0 or later), then run: ```python from apify_client import ApifyClient client = ApifyClient("YOUR_API_TOKEN") run = client.actor("hgservices/bilibili-columns-scraper").call( run_input={"creators": ["244503977"], "maxArticles": 20} ) if run is None: raise RuntimeError("The run was not found.") for item in client.dataset(run.default_dataset_id).iterate_items(): if item["type"] == "article": print(item["publishedAt"], item["title"], item["url"]) ``` #### With Claude, ChatGPT and other AI assistants (MCP) Connect your AI assistant to the **Apify MCP server**. The assistant can then run this Actor for you and answer from the results. To give the assistant this Actor as a tool, use this server URL: ``` https://mcp.apify.com?tools=hgservices/bilibili-columns-scraper ``` - **Claude (claude.ai or Claude Desktop):** open **Settings → Connectors**, click **Add custom connector** and paste the URL. - **Claude Code:** run `claude mcp add --transport http apify "https://mcp.apify.com?tools=hgservices/bilibili-columns-scraper"`. - **Apify CLI:** run `apify mcp install claude-code --tools hgservices/bilibili-columns-scraper` (also works for `cursor`, `vscode` and `codex`). - **ChatGPT:** in **Settings → Apps & Connectors**, turn on developer mode, create a connector and paste the URL. - **Cursor, VS Code and other MCP clients:** add the URL as a remote (HTTP) MCP server. Sign in with your Apify account when asked. Then ask in plain language, for example: - *"Get the 20 newest Bilibili articles of creator 244503977 and summarize them in English."* - *"Find Bilibili 专栏 articles about 人工智能 and list the 10 with the most likes."* - *"Get the full text of https://www.bilibili.com/read/cv30946284 and translate it into English."* #### With integrations Send the results to **Google Sheets, Slack, email, Zapier, Make, n8n, Airbyte** or any **webhook** from the Actor's **Integrations** tab. For example, run every morning for a list of creators and post new articles to a Slack channel. ### Tips for faster runs - **Start with metadata only.** Turn off *Include full article content*. The run is about 10 times faster and shows what a creator or keyword returns. Then run again with content for the articles you need. Metadata-only rows have no tags, coins or shares. - **Use the limit.** *Max articles per creator / keyword* stops each creator and keyword early. Results are newest first for creators. - **Monitor creators on a schedule.** Run a saved task daily or weekly and remove duplicates by `articleId` in your own system. In one run, an article that matches more than one creator, keyword or URL appears only once. - **Give large runs more time.** With full content, the Actor collects about 30 articles per minute. A creator with 2,000 articles needs more than 1 hour. Before you start, increase **Timeout** in the run options. If a run stops at the timeout, the articles that it collected stay in the dataset. - **Keep the default memory.** More memory does not make the run faster, because the speed depends on Bilibili. It only makes the run cost more. - **Blocked requests.** Bilibili sometimes blocks requests. The Actor then tries again automatically. If a run reports that Bilibili blocked it, run it again later. - **Interrupted runs continue.** If the platform restarts a run, the run continues where it stopped. Articles that are already in the dataset are not scraped again. ### FAQ, disclaimers and support **Do I need a Bilibili account?** No. The Actor reads only public articles. You don't need a login, cookie or API key. **Why do some fields have `null`?** Bilibili doesn't show every field in every place. The `summary` and `category` fields are filled for articles found by creator or keyword, and `wordCount` for articles found by creator. Image-only articles have an empty `contentText` and a `contentLength` of `0`. If you turn off *Include full article content*, the rows have no `tags`, and `coin` and `forward` are `null`. Articles found by keyword also have `favorite` set to `null`. Turn the content on to get these fields. **Why do some emoji show as `?`?** Bilibili's article list and search replace emoji with `?`. With *Include full article content* on, the Actor gets the correct title from the article itself. The `summary` field comes only from the list and search, so its emoji can stay `?`. **Why did my keyword return unrelated articles?** The results come from Bilibili's own article search, which also matches parts of your keyword. Use more specific keywords for more precise results. **Can it scrape Bilibili videos, comments or danmaku?** No. This Actor scrapes articles (专栏, plus opus posts given by URL) only. Use a Bilibili video or comments scraper for other data. **Can it scrape private, deleted or paid articles?** No. If you give the link of such an article, the Actor skips it. If a creator's list or a search still shows a deleted article, you get a row with its title and numbers, but with no content. The log tells you which articles had no content. With *Include full article content* on, you don't pay for a row without content. This also applies when Bilibili blocks the content of an article. **Is it legal to scrape Bilibili?** The Actor collects only data that is publicly available. You are responsible for using the data in line with Bilibili's terms of service, copyright law and data protection laws such as the GDPR and China's PIPL. Don't republish authors' articles without permission. **I found a problem or need a feature.** Open an issue on the **Issues** tab, and we will look into it. Custom solutions for Bilibili and other Chinese websites are available on request. # Actor input Schema ## `creators` (type: `array`): Bilibili creator UIDs (mid) or profile URLs, one per line. Accepts `244503977`, `https://space.bilibili.com/244503977` or `https://m.bilibili.com/space/244503977`. The Actor gets the creator's articles, newest first. ## `articleUrls` (type: `array`): Bilibili article links, one per line. Accepts `https://www.bilibili.com/read/cv30946284`, opus links (`https://www.bilibili.com/opus/…`), mobile links, b23.tv share links, or bare cv IDs (`30946284`). Articles given by URL always include the full content. ## `searchQueries` (type: `array`): Keywords to search Bilibili articles for, one per line, for example `人工智能` or `genshin`. Chinese, English, Japanese and Korean keywords work. ## `maxArticles` (type: `integer`): Cap the number of articles scraped per creator and per search keyword (newest first for creators). Defaults to 50 to keep a first run small and fast. Set to 0 for no limit. Bilibili shows at most 1,000 search results per keyword. ## `includeArticleContent` (type: `boolean`): Include each article's full text, HTML and image URLs. Turn off for a metadata-only run, which is much faster. A metadata-only run has the title, summary, author, date, category, views, likes and comments, but no tags, coins or shares. Favorites are included only for articles found by creator. ## Actor input object example ```json { "creators": [ "244503977" ], "maxArticles": 50, "includeArticleContent": true } ``` # Actor output Schema ## `overview` (type: `string`): Creator profiles and articles with their key fields. ## `articles` (type: `string`): Article text, tags and images. ## `engagement` (type: `string`): Views, likes, comments, coins, favorites and shares. ## `profiles` (type: `string`): One row per creator. # API You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup. ## JavaScript example ```javascript import { ApifyClient } from 'apify-client'; // Initialize the ApifyClient with your Apify API token // Replace the '