From ba853342bd2baa0a6a8df11b4d34e18b5e584e04 Mon Sep 17 00:00:00 2001
From: Max Richter <jim-x@web.de>
Date: Wed, 2 Aug 2023 15:56:33 +0200
Subject: [PATCH] feat: add ability to scrape youtube video

---
 components/Youtube.tsx              | 24 +-------
 lib/cache/image.ts                  |  4 +-
 lib/env.ts                          |  1 +
 lib/openai.ts                       |  8 ++-
 lib/string.ts                       | 23 ++++++++
 lib/youtube.ts                      | 86 +++++++++++++++++++++++++++++
 routes/api/articles/create/index.ts | 71 +++++++++++++++++++++---
 routes/articles/[name].tsx          |  3 +-
 routes/articles/index.tsx           |  8 ++-
 routes/movies/[name].tsx            |  1 +
 10 files changed, 193 insertions(+), 36 deletions(-)
 create mode 100644 lib/youtube.ts

diff --git a/components/Youtube.tsx b/components/Youtube.tsx
index e70748a..d33d174 100644
--- a/components/Youtube.tsx
+++ b/components/Youtube.tsx
@@ -1,30 +1,10 @@
-export const isYoutubeLink = (link: string) => {
-  try {
-    const url = new URL(link);
-    return ["youtu.be", "youtube.com"].includes(url.hostname);
-  } catch (err) {
-    console.log(err);
-    return false;
-  }
-};
-
-function extractYoutubeId(link: string) {
-  const url = new URL(link);
-  if (url.searchParams.has("v")) {
-    const id = url.searchParams.get("v");
-
-    if (id?.length && id.length > 4) {
-      return id;
-    }
-  }
-
-  return url.pathname.replace(/^\//, "");
-}
+import { extractYoutubeId } from "@lib/string.ts";
 
 export const YoutubePlayer = ({ link }: { link: string }) => {
   const id = extractYoutubeId(link);
   return (
     <iframe
+      class="mb-6"
       width="100%"
       height="400px"
       src={`https://www.youtube-nocookie.com/embed/${id}`}
diff --git a/lib/cache/image.ts b/lib/cache/image.ts
index b6993f6..7a3ff9a 100644
--- a/lib/cache/image.ts
+++ b/lib/cache/image.ts
@@ -69,8 +69,8 @@ export async function setImage(
   const pointerId = await hash(cacheKey);
 
   await cache.set(pointerId, clone);
-  cache.expire(pointerId, 60 * 10);
-  cache.expire(cacheKey, 60 * 10);
+  cache.expire(pointerId, 60 * 60 * 24);
+  cache.expire(cacheKey, 60 * 60 * 24);
 
   await cache.set(
     cacheKey,
diff --git a/lib/env.ts b/lib/env.ts
index 6dd04b2..cdc73c2 100644
--- a/lib/env.ts
+++ b/lib/env.ts
@@ -3,3 +3,4 @@ export const REDIS_HOST = Deno.env.get("REDIS_HOST");
 export const REDIS_PASS = Deno.env.get("REDIS_PASS");
 export const TMDB_API_KEY = Deno.env.get("TMDB_API_KEY");
 export const OPENAI_API_KEY = Deno.env.get("OPENAI_API_KEY");
+export const YOUTUBE_API_KEY = Deno.env.get("YOUTUBE_API_KEY");
diff --git a/lib/openai.ts b/lib/openai.ts
index 36842f9..d903f37 100644
--- a/lib/openai.ts
+++ b/lib/openai.ts
@@ -55,8 +55,12 @@ export async function extractAuthorName(content: string) {
 
   const author = chatCompletion.choices[0].message.content;
 
-  if (author !== "not found") return author;
-  return "";
+  if (
+    author?.toLowerCase().includes("not") &&
+    author?.toLowerCase().includes("found")
+  ) return "";
+
+  return author;
 }
 
 export async function createTags(content: string) {
diff --git a/lib/string.ts b/lib/string.ts
index 7849f57..561c116 100644
--- a/lib/string.ts
+++ b/lib/string.ts
@@ -30,3 +30,26 @@ export function extractHashTags(inputString: string) {
 
   return hashtags;
 }
+
+export const isYoutubeLink = (link: string) => {
+  try {
+    const url = new URL(link);
+    return ["youtu.be", "youtube.com","www.youtube.com" ].includes(url.hostname);
+  } catch (err) {
+    console.log(err);
+    return false;
+  }
+};
+
+export function extractYoutubeId(link: string) {
+  const url = new URL(link);
+  if (url.searchParams.has("v")) {
+    const id = url.searchParams.get("v");
+
+    if (id?.length && id.length > 4) {
+      return id;
+    }
+  }
+
+  return url.pathname.replace(/^\//, "");
+}
diff --git a/lib/youtube.ts b/lib/youtube.ts
new file mode 100644
index 0000000..c276a8e
--- /dev/null
+++ b/lib/youtube.ts
@@ -0,0 +1,86 @@
+import { YOUTUBE_API_KEY } from "@lib/env.ts";
+
+const BASE_URL = "https://youtube.googleapis.com/youtube/v3/";
+
+export interface APIResponse {
+  kind: string;
+  etag: string;
+  items: Item[];
+  pageInfo: PageInfo;
+}
+
+export interface Item {
+  kind: string;
+  etag: string;
+  id: string;
+  snippet: Snippet;
+  contentDetails: ContentDetails;
+  statistics: Statistics;
+}
+
+export interface Snippet {
+  publishedAt: string;
+  channelId: string;
+  title: string;
+  description: string;
+  thumbnails: Thumbnails;
+  channelTitle: string;
+  tags: string[];
+  categoryId: string;
+  liveBroadcastContent: string;
+  localized: Localized;
+}
+
+export interface Thumbnails {
+  default: Resolution;
+  medium: Resolution;
+  high: Resolution;
+  standard: Resolution;
+  maxres: Resolution;
+}
+
+export interface Resolution {
+  url: string;
+  width: number;
+  height: number;
+}
+
+export interface Localized {
+  title: string;
+  description: string;
+}
+
+export interface ContentDetails {
+  duration: string;
+  dimension: string;
+  definition: string;
+  caption: string;
+  licensedContent: boolean;
+  contentRating: ContentRating;
+  projection: string;
+}
+
+export interface ContentRating {}
+
+export interface Statistics {
+  viewCount: string;
+  likeCount: string;
+  favoriteCount: string;
+  commentCount: string;
+}
+
+export interface PageInfo {
+  totalResults: number;
+  resultsPerPage: number;
+}
+
+export async function getYoutubeVideoDetails(
+  id: string,
+): Promise<Item> {
+  const response = await fetch(
+    `${BASE_URL}videos?part=snippet%2CcontentDetails%2Cstatistics&id=${id}&key=${YOUTUBE_API_KEY}`,
+  );
+  const json = await response.json();
+
+  return json?.items[0];
+}
diff --git a/routes/api/articles/create/index.ts b/routes/api/articles/create/index.ts
index 7e049d8..89fe23c 100644
--- a/routes/api/articles/create/index.ts
+++ b/routes/api/articles/create/index.ts
@@ -7,7 +7,9 @@ import * as openai from "@lib/openai.ts";
 
 import tds from "https://cdn.skypack.dev/turndown@7.1.1";
 //import { gfm } from "https://cdn.skypack.dev/@guyplusplus/turndown-plugin-gfm@1.0.7";
-import { createArticle } from "@lib/resource/articles.ts";
+import { Article, createArticle } from "@lib/resource/articles.ts";
+import { getYoutubeVideoDetails } from "@lib/youtube.ts";
+import { extractYoutubeId, formatDate, isYoutubeLink } from "@lib/string.ts";
 
 const parser = new DOMParser();
 
@@ -124,6 +126,47 @@ async function processCreateArticle(
   streamResponse.enqueue("id: " + newArticle.id);
 }
 
+async function processCreateYoutubeVideo(
+  { fetchUrl, streamResponse }: {
+    fetchUrl: string;
+    streamResponse: ReturnType<typeof createStreamResponse>;
+  },
+) {
+  console.log("[api/article] create youtube article from url", {
+    url: fetchUrl,
+  });
+
+  streamResponse.enqueue("getting video infos from youtube api");
+
+  const id = extractYoutubeId(fetchUrl);
+
+  const video = await getYoutubeVideoDetails(id);
+
+  streamResponse.enqueue("shortening title with openai");
+  const newId = await openai.shortenTitle(video.snippet.title);
+
+  const newArticle: Article = {
+    name: video.snippet.title,
+    id: newId || video.snippet.title,
+    content: video.snippet.description,
+    tags: video.snippet.tags.slice(0, 5),
+    meta: {
+      status: "not-finished",
+      link: fetchUrl,
+      author: video.snippet.channelTitle,
+      date: new Date(video.snippet.publishedAt),
+    },
+  };
+
+  streamResponse.enqueue("creating article");
+
+  await createArticle(newArticle);
+
+  streamResponse.enqueue("finished");
+
+  streamResponse.enqueue("id: " + newArticle.id);
+}
+
 export const handler: Handlers = {
   GET(req) {
     const url = new URL(req.url);
@@ -135,13 +178,25 @@ export const handler: Handlers = {
 
     const streamResponse = createStreamResponse();
 
-    processCreateArticle({ fetchUrl, streamResponse }).then((article) => {
-      console.log({ article });
-    }).catch((err) => {
-      console.log(err);
-    }).finally(() => {
-      streamResponse.cancel();
-    });
+    if (isYoutubeLink(fetchUrl)) {
+      processCreateYoutubeVideo({ fetchUrl, streamResponse }).then(
+        (article) => {
+          console.log({ article });
+        },
+      ).catch((err) => {
+        console.log(err);
+      }).finally(() => {
+        streamResponse.cancel();
+      });
+    } else {
+      processCreateArticle({ fetchUrl, streamResponse }).then((article) => {
+        console.log({ article });
+      }).catch((err) => {
+        console.log(err);
+      }).finally(() => {
+        streamResponse.cancel();
+      });
+    }
 
     return streamResponse.response;
   },
diff --git a/routes/articles/[name].tsx b/routes/articles/[name].tsx
index f471aa5..77e7154 100644
--- a/routes/articles/[name].tsx
+++ b/routes/articles/[name].tsx
@@ -3,8 +3,9 @@ import { MainLayout } from "@components/layouts/main.tsx";
 import { Article, getArticle } from "@lib/resource/articles.ts";
 import { RecipeHero } from "@components/RecipeHero.tsx";
 import { KMenu } from "@islands/KMenu.tsx";
-import { isYoutubeLink, YoutubePlayer } from "@components/Youtube.tsx";
+import { YoutubePlayer } from "@components/Youtube.tsx";
 import { HashTags } from "@components/HashTags.tsx";
+import { isYoutubeLink } from "@lib/string.ts";
 
 export const handler: Handlers<Article | null> = {
   async GET(_, ctx) {
diff --git a/routes/articles/index.tsx b/routes/articles/index.tsx
index a42d968..2ff2211 100644
--- a/routes/articles/index.tsx
+++ b/routes/articles/index.tsx
@@ -31,7 +31,13 @@ export default function Greet(props: PageProps<Article[] | null>) {
       <KMenu type="main" context={false} />
       <Grid>
         {props.data?.map((doc) => {
-          return <Card link={`/articles/${doc.id}`} title={doc.name} />;
+          return (
+            <Card
+              image={"/placeholder.svg"}
+              link={`/articles/${doc.id}`}
+              title={doc.name}
+            />
+          );
         })}
       </Grid>
     </MainLayout>
diff --git a/routes/movies/[name].tsx b/routes/movies/[name].tsx
index d4c94cf..bb4eeec 100644
--- a/routes/movies/[name].tsx
+++ b/routes/movies/[name].tsx
@@ -22,6 +22,7 @@ export default function Greet(props: PageProps<Movie>) {
       <RecipeHero
         data={movie}
         subline={[author, date.toString()]}
+        editLink={`https://notes.max-richter.dev/Media/movies/${movie.id}`}
         backlink="/movies"
       />
       <KMenu type="main" context={movie} />