memorium/routes/api/articles/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 } 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 { ArticleResource } from "@lib/marka/schema.ts";
import { webScrape } from "@lib/webScraper.ts";
import * as openai from "@lib/openai.ts";
import * as unsplash from "@lib/unsplash.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/article/enhance");

async function getUnsplashCoverImage(
  content: string,
  streamResponse: ReturnType<typeof createStreamResponse>,
): Promise<string | undefined> {
  try {
    streamResponse.info("creating unsplash search term");
    const searchTerm = await openai.createUnsplashSearchTerm(content);
    if (!searchTerm) return;
    streamResponse.info(`searching for ${searchTerm}`);
    const unsplashUrl = await unsplash.getImageBySearchTerm(searchTerm);
    return unsplashUrl;
  } catch (e) {
    log.error("Failed to get unsplash cover image", e);
    return undefined;
  }
}

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

async function processEnhanceArticle(
  name: string,
  streamResponse: ReturnType<typeof createStreamResponse>,
) {
  const article = await fetchResource<ArticleResource>(
    `articles/${name}`,
  );
  if (!article) {
    throw new NotFoundError();
  }

  const fetchUrl = article.content?.url;
  if (!fetchUrl) {
    throw new BadRequestError("Article has no URL to enhance from.");
  }

  log.info("enhancing article 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 ||
    article.content?.headline || "";

  article.content ??= {
    _type: "Article",
    headline: title,
    url: fetchUrl,
  };

  article.content.articleBody = result.markdown;
  article.content.datePublished ??= formatDate(
    result?.published || aiMeta?.datePublished || undefined,
  );

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

  if (!article.content.image) {
    let coverImagePath: string | undefined = undefined;
    if (result?.image?.length) {
      log.debug("using local image for cover image", { image: result.image });
      coverImagePath = await fetchAndStoreCover(
        result.image,
        title,
        streamResponse,
      );
    } else {
      const urlPath = await getUnsplashCoverImage(
        result.content,
        streamResponse,
      );
      coverImagePath = await fetchAndStoreCover(urlPath, title, streamResponse);
      log.debug("using unsplash for cover image", { image: coverImagePath });
    }
    if (coverImagePath) {
      article.content.image = coverImagePath;
    }
  }

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

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

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

  const streamResponse = createStreamResponse();

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

  return streamResponse.response;
};

export const handler: Handlers = {
  POST,
};