From b31d4edd720ccbe73c41524443cff2aad99e9e63 Mon Sep 17 00:00:00 2001
From: "Dewa M. Widyakumara" <widyakumara@users.noreply.github.com>
Date: Sun, 21 Apr 2019 16:52:05 +0700
Subject: [PATCH] add: instagram/tag (#1964)

add support to fetch instagram's hashtags. eg:
* [#designertoys](https://www.instagram.com/explore/tags/designertoys/)
* [#urbantoys](https://www.instagram.com/explore/tags/urbantoys/)
---
 docs/en/README.md           |   2 +
 lib/router.js               |   1 +
 lib/routes/instagram/tag.js | 109 ++++++++++++++++++++++++++++++++++++
 3 files changed, 112 insertions(+)
 create mode 100644 lib/routes/instagram/tag.js

diff --git a/docs/en/README.md b/docs/en/README.md
index 0767e2c12f..25f20367c3 100644
--- a/docs/en/README.md
+++ b/docs/en/README.md
@@ -225,6 +225,8 @@ If no matching results were found, the server returns only a HTTP status code `2
 
 <RouteEn name="User" path="/instagram/user/:id" example="/instagram/user/diygod" :paramsDesc="['Instagram id']" />
 
+<RouteEn name="Hashtag" author="widyakumara" path="/instagram/tag/:tag" example="/instagram/tag/urbantoys" :paramsDesc="['Instagram hashtag']" />
+
 ### Youtube
 
 <RouteEn name="User" path="/youtube/user/:username/:embed?" example="/youtube/user/JFlaMusic" :paramsDesc="['YouTuber id', 'Default to embed the video, set to any value to disable embedding']" />
diff --git a/lib/router.js b/lib/router.js
index 3bf2652937..4f7aee8c71 100755
--- a/lib/router.js
+++ b/lib/router.js
@@ -258,6 +258,7 @@ if (config.twitter && config.twitter.consumer_key && config.twitter.consumer_sec
 
 // Instagram
 router.get('/instagram/user/:id', require('./routes/instagram/user'));
+router.get('/instagram/tag/:tag', require('./routes/instagram/tag'));
 
 // Youtube
 if (config.youtube && config.youtube.key) {
diff --git a/lib/routes/instagram/tag.js b/lib/routes/instagram/tag.js
new file mode 100644
index 0000000000..6e7bd29954
--- /dev/null
+++ b/lib/routes/instagram/tag.js
@@ -0,0 +1,109 @@
+const axios = require('../../utils/axios');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const tag = ctx.params.tag;
+
+    const response = await axios({
+        method: 'get',
+        url: `https://www.instagram.com/explore/tags/${tag}/`,
+        headers: {
+            Referer: `https://www.instagram.com/explore/tags/${tag}/`,
+        },
+    });
+
+    const data = JSON.parse(response.data.match(/<script type="text\/javascript">window._sharedData = (.*);<\/script>/)[1]) || {};
+    const list = data.entry_data.TagPage[0].graphql.hashtag.edge_hashtag_to_media.edges;
+
+    const $ = cheerio.load(response.data);
+
+    // retrieve media objects
+    const media = await Promise.all(
+        list.map(async (item) => {
+            item = item.node;
+
+            const url = `https://www.instagram.com/p/${item.shortcode}`;
+            const cache = await ctx.cache.get(url);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+
+            if (item.__typename === 'GraphImage') {
+                const single = {
+                    image: item.display_url,
+                };
+
+                ctx.cache.set(url, JSON.stringify([single]), 24 * 60 * 60);
+                return Promise.resolve([single]);
+            } else if (item.__typename === 'GraphSidecar') {
+                const response = await axios({
+                    method: 'get',
+                    url,
+                });
+                const data = JSON.parse(response.data.match(/<script type="text\/javascript">window._sharedData = (.*);<\/script>/)[1]) || {};
+                const single = data.entry_data.PostPage[0].graphql.shortcode_media.edge_sidecar_to_children.edges.map((item) => ({
+                    image: item.node.display_url,
+                    video: item.node.video_url,
+                }));
+
+                ctx.cache.set(url, JSON.stringify(single), 24 * 60 * 60);
+                return Promise.resolve(single);
+            } else if (item.__typename === 'GraphVideo') {
+                const response = await axios({
+                    method: 'get',
+                    url,
+                });
+                const data = JSON.parse(response.data.match(/<script type="text\/javascript">window._sharedData = (.*);<\/script>/)[1]) || {};
+                const single = {
+                    image: data.entry_data.PostPage[0].graphql.shortcode_media.display_url,
+                    video: data.entry_data.PostPage[0].graphql.shortcode_media.video_url,
+                };
+
+                ctx.cache.set(url, JSON.stringify([single]), 24 * 60 * 60);
+                return Promise.resolve([single]);
+            }
+        })
+    );
+
+    ctx.state.data = {
+        title: `#${tag} hashtag on Instagram`,
+        link: `https://www.instagram.com/explore/tags/${tag}/`,
+        description: $('meta[name="description"]').attr('content'),
+        item: list.map((item, index) => {
+            item = item.node;
+            let image = 0;
+            let video = 0;
+
+            let content = '';
+
+            for (let i = 0; i < media[index].length; i++) {
+                if (media[index][i].image) {
+                    content += `<img referrerpolicy="no-referrer" src="${media[index][i].image}"><br>`;
+                    image++;
+                }
+
+                if (media[index][i].video) {
+                    content += `<video width="100%" controls="controls"> <source src="${media[index][i].video}" type="video/mp4"> Your RSS reader does not support video playback. </video>`;
+                    video++;
+                }
+            }
+
+            let title = (item.edge_media_to_caption.edges && item.edge_media_to_caption.edges[0] && item.edge_media_to_caption.edges[0].node.text) || '无题/Untitled';
+
+            if (image > 1) {
+                title = `[组图/Carousel]${title}`;
+            }
+
+            if (video > 0) {
+                title = `[视频/Video]${title}`;
+            }
+
+            return {
+                title: `${title}`,
+                description: `${title}<br>${content}`,
+                pubDate: new Date(item.taken_at_timestamp * 1000).toUTCString(),
+                link: `https://www.instagram.com/p/${item.shortcode}/`,
+            };
+        }),
+    };
+};