Books Scraper (Google Books + Open Library)

Books Scraper: book metadata from Open Library or Google Books

Search by keywords, a title, a subject or an ISBN and get one row per book: title, subtitle, authors, publisher, year, ISBN, page count, subjects, rating, language and a cover image. Both sources come out in the same shape, so you can switch between them without rewriting anything downstream.

The two sources are not equivalent. Open Library answers with no key and no setup. Google Books carries the description and the retail price, and to use it reliably you need a free Google key of your own, which takes about two minutes to create.

🔍 What Books Scraper does

It runs one search against the catalogue you pick and pages through the results until it has as many unique books as you asked for.

Open Library is the default and the one to start with. No key, no account, nothing to set up. Rows come back with the subjects list, the rating and vote count, the cover and the work page URL. Open Library holds no blurb and no price, so description and price are null on every row from it.

Google Books adds the publisher's description and, where the book is sold through Google, a price object with the amount, currency and a buy link. Without a key those calls run on an anonymous allowance shared by everyone using it, which is usually spent, so the run switches to Open Library and writes one uncharged notice row telling you it did. Put your own key in googleApiKey and the call runs on your own allowance instead.

Books arrive deduplicated: by ISBN where there is one, by title plus first author where there is not. A record with no title is dropped before it is counted or charged.

📥 What you give it

{
  "source": "openlibrary",
  "query": "clean code",
  "maxItems": 50
}

Never paste a key into a published Apify task. Task inputs are readable by anyone who opens the page. Put it in the run input, or supply it per call.

📤 What you get back

A real row from a recent run:

{
  "ok": true,
  "source": "openlibrary",
  "sourceId": "/works/OL17618370W",
  "title": "Clean Code",
  "subtitle": "A Handbook of Agile Software Craftsmanship",
  "authors": ["Robert C. Martin"],
  "publisher": "Prentice Hall",
  "publishedDate": "2008",
  "year": 2008,
  "isbn": "9780136083221",
  "pageCount": 444,
  "categories": ["Agile software development", "Reliability", "Computer software", "Computer software, development", "Coding theory"],
  "averageRating": 4.44,
  "ratingsCount": 43,
  "language": "eng",
  "description": null,
  "coverImage": "https://covers.openlibrary.org/b/id/8065615-L.jpg",
  "url": "https://openlibrary.org/works/OL17618370W",
  "price": null
}

🧾 Reading the output

Real book rows carry ok: true. Anything with ok: false is a note or a problem, and none of those are charged.

The RATE_LIMITED notice arrives as the first row in the dataset, before the books. A run that opens with what looks like an error is usually a run that worked.

If Google Books returns nothing at all with a key set, suspect the key before the query. A key Google will not accept reads back as a search with no matches, not as a key error.

▶️ How to run it

1. Open Books Scraper and click Try for free. 2. Leave Source on Open Library for your first run. 3. Type into Search query. A subject like machine learning works as well as a title. 4. Set Max books, then click Start. 5. Download the dataset as JSON, CSV, Excel or XML.

For Google Books, create a free API key in the Google Cloud console, switch on the Books API for that project, and paste the key into Google Books API key.

💰 How much does it cost?

$1.00 per 1,000 books. Flat on every Apify plan, no volume tiers, and the same price whichever catalogue answered.

You are billed per unique book row. Duplicates dropped along the way, records with no title, the fallback notice row and every diagnostic row are not charged.

💡 What people use it for

• Filling in a reading list or a library catalogue that only has titles, using ISBN as the join key.
• Pulling everything under a subject, like every book Open Library files under machine learning, to

see what exists before buying.

• Building a price and description list for a shortlist of titles through Google Books.
• Checking page counts and publication years across a set of editions before choosing one.

🚧 What it does not do

• No reviews and no review text. Ratings and vote counts only.
• No editions list. One row per work, not one per printing, so the ISBN you get is a

representative one rather than the specific edition on your shelf.

• No full text, no excerpts, no tables of contents.
• Open Library rows have no description and no price. That is the catalogue, not a setting you

can change.

• Google Books keyless is not dependable. Expect the fallback unless you supply a key.
• Covers are links, not files. The image stays on the catalogue's servers.
• Search quality is the catalogue's. A loose query returns loosely related books, and those rows

are charged like any other, so keep maxItems low while you find the right wording.

🧭 Which catalogue scraper do you need?

❓ Questions people ask

Do I need a key? Not for Open Library. For Google Books, yes in practice. It is free and takes a couple of minutes in the Google Cloud console.

Can I search by ISBN? Yes. Put isbn:9780132350884 in the query. That works on Google Books.

Why does my row say openlibrary when I picked Google? The run fell back, and the notice row at the top of the dataset says so. Add your own key to stay on Google.

How do I avoid paying for the same book twice across runs? Dedupe on isbn, falling back to title plus the first entry in authors, which is what the run does internally.

Why did I get fewer books than I asked for? The catalogue ran out of matches, or duplicates were removed. You pay for what arrived.

Is this legal? Open Library and Google Books both publish this data through open APIs meant to be read, and book metadata is not personal data. Apify's write-up on scraping and the law is a good starting point, and we are not lawyers.

🆘 If something breaks

Open the Issues tab on the actor page. Send the query, the source you picked and the run id. The errorCode on the diagnostic row usually names the problem by itself.