memorium/routes/api/books/enhance/[name].ts

import { FreshContext, Handlers } from "$fresh/server.ts";
import { fileExtension } from "https://deno.land/x/file_extension@v2.1.0/mod.ts";
import { formatDate, safeFileName, toUrlSafeString } from "@lib/string.ts";
import { createStreamResponse } from "@lib/helpers.ts";
import {
  AccessDeniedError,
  BadRequestError,
  NotFoundError,
} from "@lib/errors.ts";
import { createResource, fetchResource } from "@lib/marka/index.ts";
import { BookResource } from "@lib/marka/schema.ts";
import { webScrape } from "@lib/webScraper.ts";
import * as openai from "@lib/openai.ts";
import { getBookDetails } from "@lib/hardcover.ts";
import { createLogger } from "@lib/log/index.ts";

function ext(str: string) {
  try {
    const u = new URL(str);
    if (u.searchParams.has("fm")) {
      return u.searchParams.get("fm")!;
    }
    return fileExtension(u.pathname);
  } catch (_e) {
    return fileExtension(str);
  }
}

const log = createLogger("api/book/enhance");

async function fetchAndStoreCover(
  imageUrl: string | undefined,
  title: string,
): Promise<string | undefined> {
  if (!imageUrl) return;
  const imagePath = `books/images/${safeFileName(title)}_cover.${
    ext(imageUrl)
  }`;
  try {
    const res = await fetch(imageUrl);
    if (!res.ok) {
      log.error(`Failed to download remote image: ${imageUrl}`, {
        status: res.status,
      });
      return;
    }
    const buffer = await res.arrayBuffer();
    await createResource(imagePath, buffer);
    return `resources/${imagePath}`;
  } catch (err) {
    log.error(`Failed to save image: ${imageUrl}`, err);
    return undefined;
  }
}

async function processEnhanceFromHardcover(
  name: string,
  hardcoverSlug: string,
  streamResponse: ReturnType<typeof createStreamResponse>,
) {
  streamResponse.info("fetching from Hardcover");

  const bookDetails = await getBookDetails(hardcoverSlug);

  if (!bookDetails) {
    throw new NotFoundError("Book not found on Hardcover");
  }

  const book = await fetchResource<BookResource>(`books/${name}`);
  if (!book) {
    throw new NotFoundError();
  }

  const title = bookDetails.title || book.content?.headline || "";
  const authorName = bookDetails.author_names?.[0] || "";

  streamResponse.info("updating cover image");
  if (bookDetails.image && !book.content?.image) {
    const coverPath = await fetchAndStoreCover(bookDetails.image, title);
    if (coverPath) {
      book.content.image = coverPath;
    }
  }

  if (!book.content?.headline || book.content.headline !== bookDetails.title) {
    book.content.headline = bookDetails.title;
  }

  if (!book.content?.subtitle && bookDetails.subtitle) {
    book.content.subtitle = bookDetails.subtitle;
  }

  if (!book.content?.isbn && (bookDetails.isbn13 || bookDetails.isbn)) {
    book.content.isbn = bookDetails.isbn13 || bookDetails.isbn;
  }

  if (bookDetails.rating && !book.content?.reviewRating) {
    book.content.reviewRating = {
      ratingValue: bookDetails.rating,
      bestRating: 5,
      worstRating: 1,
    };
  }

  if (!book.content?.datePublished && bookDetails.release_year) {
    book.content.datePublished = formatDate(`${bookDetails.release_year}-01-01`);
  }

  const newKeywords = [
    ...(bookDetails.genres?.map((g: { name: string }) => g.name) || []),
    ...(bookDetails.tags?.map((t: { name: string }) => t.name) || []),
  ].filter(Boolean);

  if (newKeywords.length > 0) {
    book.content.keywords = [
      ...(book.content.keywords || []),
      ...newKeywords,
    ];
  }

  streamResponse.info("writing to disk");
  await createResource(`books/${name}`, book.content);
  streamResponse.send({ type: "finished", url: name.replace(/$\.md/, "") });
}

async function processEnhanceFromUrl(
  name: string,
  fetchUrl: string,
  streamResponse: ReturnType<typeof createStreamResponse>,
) {
  log.info("enhancing book from url", { url: fetchUrl });
  streamResponse.info("scraping url");
  const result = await webScrape(fetchUrl, streamResponse);

  streamResponse.info("parsing content");

  log.debug("downloaded and parsed", result);

  streamResponse.info("extracting metadata with openai");
  const aiMeta = await openai.extractArticleMetadata(result.markdown);

  const title = result?.title || aiMeta?.headline ||
    name;

  const book = await fetchResource<BookResource>(`books/${name}`);
  if (!book) {
    throw new NotFoundError();
  }

  book.content ??= {
    _type: "Book",
    headline: title,
    url: fetchUrl,
  };

  if (!book.content.bookBody && result.markdown) {
    book.content.bookBody = result.markdown;
  }

  book.content.datePublished ??= formatDate(
    result?.published || aiMeta?.datePublished || undefined,
  );

  if (!book.content.author?.name || book.content.author.name === "") {
    book.content.author = {
      _type: "Person",
      name: (result.schemaOrgData?.author?.name || aiMeta?.author || "")
        .replace("@", "twitter:"),
    };
  }

  if (!book.content.image && result?.image?.length) {
    const coverPath = await fetchAndStoreCover(result.image, title);
    if (coverPath) {
      book.content.image = coverPath;
    }
  }

  log.debug("writing to disk", {
    name: name,
    book: {
      ...book,
      content: {
        ...book.content,
        bookBody: book.content.bookBody?.slice(0, 200),
      },
    },
  });

  streamResponse.info("writing to disk");
  await createResource(`books/${name}`, book.content);
  streamResponse.send({ type: "finished", url: name.replace(/$\.md/, "") });
}

async function processEnhanceBook(
  name: string,
  streamResponse: ReturnType<typeof createStreamResponse>,
) {
  const book = await fetchResource<BookResource>(`books/${name}`);
  if (!book) {
    throw new NotFoundError();
  }

  const hardcoverUrl = book.content?.url;
  if (hardcoverUrl?.includes("hardcover.app")) {
    const match = hardcoverUrl.match(/books\/(.+)/);
    if (match && match[1]) {
      return processEnhanceFromHardcover(name, match[1], streamResponse);
    }
  }

  if (hardcoverUrl) {
    return processEnhanceFromUrl(name, hardcoverUrl, streamResponse);
  }

  throw new BadRequestError("Book has no URL to enhance from.");
}

const POST = (
  _req: Request,
  ctx: FreshContext,
): Response => {
  const session = ctx.state.session;
  if (!session) {
    throw new AccessDeniedError();
  }

  const streamResponse = createStreamResponse();

  processEnhanceBook(ctx.params.name, streamResponse)
    .catch((err) => {
      log.error(err);
      streamResponse.error(err.message);
    })
    .finally(() => {
      streamResponse.cancel();
    });

  return streamResponse.response;
};

export const handler: Handlers = {
  POST,
};