Implement #771

CHANGELOG.md

@@ -2,6 +2,10 @@ ChangeLog (Release Notes)
 =========================
 主に notable な changes を書いていきます
 
+unreleased
+----------
+* New: ユーザーページによく使うドメインを表示 (#771)
+
 2566 (2017/09/07)
 -----------------
 * New: 投稿することの多いキーワードをユーザーページに表示する (#768)

locales/en.yml

@@ -498,6 +498,7 @@ mobile:
       images: "Images"
       activity: "Activity"
       keywords: "Keywords"
+      domains: "Domains"
       followers-you-know: "Followers you know"
       last-used-at: "Latest used at"
 
@@ -512,6 +513,9 @@ mobile:
     mk-user-overview-keywords:
       no-keywords: "No keywords"
 
+    mk-user-overview-domains:
+      no-domains: "No domains"
+
     mk-user-overview-followers-you-know:
       loading: "Loading"
       no-users: "No users"

locales/ja.yml

@@ -498,6 +498,7 @@ mobile:
       images: "画像"
       activity: "アクティビティ"
       keywords: "キーワード"
+      domains: "頻出ドメイン"
       followers-you-know: "知り合いのフォロワー"
       last-used-at: "最終ログイン"
 
@@ -512,6 +513,9 @@ mobile:
     mk-user-overview-keywords:
       no-keywords: "キーワードはありません(十分な数の投稿をしていない可能性があります)"
 
+    mk-user-overview-domains:
+      no-domains: "よく表れるドメインは検出されませんでした"
+
     mk-user-overview-followers-you-know:
       loading: "読み込み中"
       no-users: "知り合いのユーザーはいません"

src/tools/analysis/extract-user-domains.ts

@@ -0,0 +1,120 @@
+import * as URL from 'url';
+
+import Post from '../../api/models/post';
+import User from '../../api/models/user';
+import parse from '../../api/common/text';
+
+process.on('unhandledRejection', console.dir);
+
+function tokenize(text: string) {
+	if (text == null) return [];
+
+	// パース
+	const ast = parse(text);
+
+	const domains = ast
+		// URLを抽出
+		.filter(t => t.type == 'url' || t.type == 'link')
+		.map(t => URL.parse(t.url).hostname);
+
+	return domains;
+}
+
+// Fetch all users
+User.find({}, {
+	fields: {
+		_id: true
+	}
+}).then(users => {
+	let i = -1;
+
+	const x = cb => {
+		if (++i == users.length) return cb();
+		extractDomainsOne(users[i]._id).then(() => x(cb), err => {
+			console.error(err);
+			setTimeout(() => {
+				i--;
+				x(cb);
+			}, 1000);
+		});
+	};
+
+	x(() => {
+		console.log('complete');
+	});
+});
+
+function extractDomainsOne(id) {
+	return new Promise(async (resolve, reject) => {
+		process.stdout.write(`extracting domains of ${id} ...`);
+
+		// Fetch recent posts
+		const recentPosts = await Post.find({
+			user_id: id,
+			text: {
+				$exists: true
+			}
+		}, {
+			sort: {
+				_id: -1
+			},
+			limit: 10000,
+			fields: {
+				_id: false,
+				text: true
+			}
+		});
+
+		// 投稿が少なかったら中断
+		if (recentPosts.length < 100) {
+			process.stdout.write(' >>> -\n');
+			return resolve();
+		}
+
+		const domains = {};
+
+		// Extract domains from recent posts
+		recentPosts.forEach(post => {
+			const domainsOfPost = tokenize(post.text);
+
+			domainsOfPost.forEach(domain => {
+				if (domains[domain]) {
+					domains[domain]++;
+				} else {
+					domains[domain] = 1;
+				}
+			});
+		});
+
+		// Calc peak
+		let peak = 0;
+		Object.keys(domains).forEach(domain => {
+			if (domains[domain] > peak) peak = domains[domain];
+		});
+
+		// Sort domains by frequency
+		const domainsSorted = Object.keys(domains).sort((a, b) => domains[b] - domains[a]);
+
+		// Lookup top 10 domains
+		const topDomains = domainsSorted.slice(0, 10);
+
+		process.stdout.write(' >>> ' + topDomains.join(', ') + '\n');
+
+		// Make domains object (includes weights)
+		const domainsObj = topDomains.map(domain => ({
+			domain: domain,
+			weight: domains[domain] / peak
+		}));
+
+		// Save
+		User.update({ _id: id }, {
+			$set: {
+				domains: domainsObj
+			}
+		}).then(() => {
+			resolve();
+		}, err => {
+			reject(err);
+		});
+	});
+}

src/web/app/mobile/tags/user.tag

@@ -240,6 +240,12 @@
 			<mk-user-overview-keywords user={ user }/>
 		</div>
 	</section>
+	<section class="domains">
+		<h2><i class="fa fa-globe"></i>%i18n:mobile.tags.mk-user-overview.domains%</h2>
+		<div>
+			<mk-user-overview-domains user={ user }/>
+		</div>
+	</section>
 	<section class="followers-you-know" if={ SIGNIN && I.id !== user.id }>
 		<h2><i class="fa fa-users"></i>%i18n:mobile.tags.mk-user-overview.followers-you-know%</h2>
 		<div>
@@ -579,6 +585,40 @@
 	</script>
 </mk-user-overview-keywords>
 
+<mk-user-overview-domains>
+	<div if={ user.domains != null && user.domains.length > 1 }>
+		<virtual each={ domain in user.domains }>
+			<a style="opacity: { 0.5 + (domain.weight / 2) }">{ domain.domain }</a>
+		</virtual>
+	</div>
+	<p class="empty" if={ user.domains == null || user.domains.length == 0 }>%i18n:mobile.tags.mk-user-overview-domains.no-domains%</p>
+	<style>
+		:scope
+			display block
+
+			> div
+				padding 4px
+
+				> a
+					display inline-block
+					margin 4px
+					color #555
+
+			> .empty
+				margin 0
+				padding 16px
+				text-align center
+				color #aaa
+
+				> i
+					margin-right 4px
+
+	</style>
+	<script>
+		this.user = this.opts.user;
+	</script>
+</mk-user-overview-domains>
+
 <mk-user-overview-followers-you-know>
 	<p class="initializing" if={ initializing }><i class="fa fa-spinner fa-pulse fa-fw"></i>%i18n:mobile.tags.mk-user-overview-followers-you-know.loading%<mk-ellipsis/></p>
 	<div if={ !initializing && users.length > 0 }>