obsidian-book-tracker/src/data-sources/Goodreads.ts

import { requestUrl } from "obsidian";
import type {
	Author,
	Book as OutputBook,
	Series as OutputSeries,
} from "../types";

interface Ref {
	__ref: string;
}

interface BookContributorEdge {
	__typename: "BookContributorEdge";
	node: Ref;
	role: string;
}

interface BookSeries {
	__typename: "BookSeries";
	userPosition: string;
	series: Ref;
}

interface Genre {
	__typename: "Genre";
	name: string;
	webUrl: string;
}

interface BookGenre {
	__typename: "BookGenre";
	genre: Genre;
}

interface Language {
	__typename: "Language";
	name: string;
}

interface BookDetails {
	__typename: "BookDetails";
	asin: string;
	format: string;
	numPages: number;
	publicationTime: number;
	publisher: string;
	isbn: string;
	isbn13: string;
	language: Language;
}

interface Book {
	__typename: "Book";
	id: string;
	legacyId: number;
	webUrl: string;
	title: string;
	titleComplete: string;
	description: string;
	primaryContributorEdge: BookContributorEdge;
	secondaryContributorEdges: BookContributorEdge[];
	imageUrl: string;
	bookSeries: BookSeries[];
	bookGenres: BookGenre[];
	details: BookDetails;
	work: Ref;
}

interface ContributorWorksConnection {
	__typename: "ContributorWorksConnection";
	totalCount: number;
}

interface ContributorFollowersConnection {
	__typename: "ContributorFollowersConnection";
	totalCount: number;
}

interface Contributor {
	__typename: "Contributor";
	id: string;
	legacyId: number;
	name: string;
	description: string;
	isGrAuthor: boolean;
	works: ContributorWorksConnection;
	profileImageUrl: string;
	webUrl: string;
	followers: ContributorFollowersConnection;
}

interface Series {
	__typename: "Series";
	id: string;
	title: string;
	webUrl: string;
}

interface Query {
	__typename: "Query";
	[key: string]: unknown;
}

interface NextData {
	props: {
		pageProps: {
			apolloState: {
				ROOT_QUERY: Query;
				[key: string]: unknown;
			};
			params: Record<string, string>;
			query: Record<string, string>;
			jwtToken: string;
			dataSource: string;
		};
	};
}

export interface SearchResult {
	legacyId: number;
	title: string;
	authors: string[];
	avgRating: number;
	ratingCount: number;
	publicationYear: number;
	editionCount: number;
	coverImageUrl: string;
}

export class Goodreads {
	async getNextData(legacyId: number): Promise<NextData> {
		const url = "https://www.goodreads.com/book/show/" + legacyId;
		const res = await requestUrl({ url });
		const doc = new DOMParser().parseFromString(res.text, "text/html");
		const nextDataRaw = doc.getElementById("__NEXT_DATA__")?.textContent;
		if (typeof nextDataRaw !== "string") {
			throw new Error("Unable to find next data script in the document.");
		}
		const nextData = JSON.parse(nextDataRaw) as NextData;

		return nextData;
	}

	extractBookFromNextData(nextData: NextData, ref: Ref): OutputBook {
		const apolloState = nextData.props.pageProps.apolloState;
		const bookData = apolloState[ref.__ref] as Book;

		const contributorEdges = [
			bookData.primaryContributorEdge,
			...bookData.secondaryContributorEdges,
		];
		const authors = contributorEdges
			.filter((edge) => edge.role === "Author")
			.map((edge) => apolloState[edge.node.__ref] as Contributor)
			.map<Author>((contributor) => ({
				id: contributor.id,
				legacyId: contributor.legacyId,
				name: contributor.name,
				description: contributor.description,
			}));

		let series: OutputSeries | null = null;
		if (bookData.bookSeries.length > 0) {
			const bookSeries = bookData.bookSeries[0];
			const seriesData = apolloState[bookSeries.series.__ref] as Series;
			series = {
				title: seriesData.title,
				position: parseInt(bookSeries.userPosition, 10),
			};
		}

		let title = bookData.title;
		let subtitle = "";

		if (title.includes(": ")) {
			const parts = title.split(": ");
			subtitle = parts.pop()!;
			title = parts.join(": ");
		}

		return {
			title,
			subtitle,
			description: bookData.description,
			authors,
			series,
			publisher: bookData.details.publisher,
			publishedAt: new Date(bookData.details.publicationTime),
			genres: bookData.bookGenres.map((genre) => genre.genre.name),
			coverImageUrl: bookData.imageUrl,
			pageCount: bookData.details.numPages,
			isbn: bookData.details.isbn,
			isbn13: bookData.details.isbn13,
		};
	}

	async getBookByLegacyId(legacyId: number): Promise<OutputBook> {
		const nextData = await this.getNextData(legacyId);
		const bookRef = nextData.props.pageProps.apolloState.ROOT_QUERY[
			`getBookByLegacyId({"legacyId":"${legacyId}"})`
		] as Ref | undefined;
		if (bookRef === undefined) {
			throw new Error("Could not find reference for book.");
		}

		return this.extractBookFromNextData(nextData, bookRef);
	}

	async search(q: string): Promise<SearchResult[]> {
		const url =
			"https://www.goodreads.com/search?q=" + encodeURIComponent(q);
		const res = await requestUrl({ url });
		const doc = new DOMParser().parseFromString(res.text, "text/html");

		const searchResults: SearchResult[] = [];
		doc.querySelectorAll(
			"table.tableList tr[itemtype='http://schema.org/Book']"
		).forEach((el) => {
			const legacyId = parseInt(
				el.querySelector("div.u-anchorTarget")?.id ?? "",
				10
			);

			const title =
				el.querySelector("a.bookTitle")?.textContent?.trim() || "";

			const authors = Array.from(el.querySelectorAll("a.authorName")).map(
				(a) => a.textContent?.trim() || ""
			);

			const avgRating = parseFloat(
				el
					.querySelector("span.minirating")
					?.textContent?.match(/(\d+\.\d+) avg rating/)?.[1] ?? "0"
			);

			const ratingCount = parseInt(
				el
					.querySelector("span.minirating")
					?.textContent?.match(/(\d[\d,]*) ratings/)?.[1] ?? "0",
				10
			);

			const publicationYear = parseInt(
				el
					.querySelector("span.greyText")
					?.textContent?.match(/published (\d{4})/)?.[1] ?? "0",
				10
			);

			const editionCount = parseInt(
				el
					.querySelector("span.greyText")
					?.textContent?.match(/(\d+) editions/)?.[1] ?? "0",
				10
			);

			const coverImageUrl =
				el.querySelector("img.bookCover")?.getAttribute("src") || "";

			searchResults.push({
				legacyId,
				title,
				authors,
				avgRating,
				ratingCount,
				publicationYear,
				editionCount,
				coverImageUrl,
			});
		});
		return searchResults;
	}
}