From c7af8cbc9081f349461634db0a1d76bc689d8dd5 Mon Sep 17 00:00:00 2001
From: "Akihiko Odaki (@fn_aki@pawoo.net)" <akihiko.odaki.4i@stu.hosei.ac.jp>
Date: Sun, 4 Jun 2017 21:58:57 +0900
Subject: [PATCH] Remove some arguments of Formatter.instance.format and spec
 (#3541)

* Remove some arguments of Formatter.instance.format

* Improve spec for Formatter
---
 app/lib/formatter.rb       |   7 +-
 spec/lib/formatter_spec.rb | 302 ++++++++++++++++++++++++-------------
 2 files changed, 200 insertions(+), 109 deletions(-)

diff --git a/app/lib/formatter.rb b/app/lib/formatter.rb
index 43d23db96..7b89305ac 100644
--- a/app/lib/formatter.rb
+++ b/app/lib/formatter.rb
@@ -9,7 +9,7 @@ class Formatter
 
   include ActionView::Helpers::TextHelper
 
-  def format(status, attribute = :text, paragraphize = true)
+  def format(status)
     if status.reblog?
       prepend_reblog = status.reblog.account.acct
       status         = status.proper
@@ -17,9 +17,8 @@ class Formatter
       prepend_reblog = false
     end
 
-    raw_content = status.public_send(attribute)
+    raw_content = status.text
 
-    return '' if raw_content.blank?
     return reformat(raw_content) unless status.local?
 
     linkable_accounts = status.mentions.map(&:account)
@@ -28,7 +27,7 @@ class Formatter
     html = raw_content
     html = "RT @#{prepend_reblog} #{html}" if prepend_reblog
     html = encode_and_link_urls(html, linkable_accounts)
-    html = simple_format(html, {}, sanitize: false) if paragraphize
+    html = simple_format(html, {}, sanitize: false)
     html = html.delete("\n")
 
     html.html_safe # rubocop:disable Rails/OutputSafety
diff --git a/spec/lib/formatter_spec.rb b/spec/lib/formatter_spec.rb
index ec61eaa43..cc32f7fd6 100644
--- a/spec/lib/formatter_spec.rb
+++ b/spec/lib/formatter_spec.rb
@@ -1,193 +1,285 @@
 require 'rails_helper'
 
 RSpec.describe Formatter do
-  let(:account)       { Fabricate(:account, username: 'alice') }
-  let(:local_text)    { 'Hello world http://google.com' }
-  let(:local_status)  { Fabricate(:status, text: local_text, account: account) }
-  let(:remote_status) { Fabricate(:status, text: '<script>alert("Hello")</script> Beep boop', uri: 'beepboop', account: account) }
-
-  let(:local_text_with_mention) { "@#{account.username} @#{account.username}@example.com #{local_text}?x=@#{account.username} #hashtag" }
-
-  let(:local_status_with_mention) do
-    Fabricate(
-      :status,
-      text: local_text_with_mention,
-      account: account,
-      mentions: [Fabricate(:mention, account: account)]
-    )
-  end
+  let(:local_account)  { Fabricate(:account, domain: nil, username: 'alice') }
+  let(:remote_account) { Fabricate(:account, domain: 'remote', username: 'bob', url: 'https://remote/') }
 
-  describe '#format' do
-    subject { Formatter.instance.format(local_status) }
+  shared_examples 'encode and link URLs' do
+    context 'matches a stand-alone medium URL' do
+      let(:text) { 'https://hackernoon.com/the-power-to-build-communities-a-response-to-mark-zuckerberg-3f2cac9148a4' }
 
-    context 'with standalone status' do
-      it 'returns a string' do
-        expect(subject).to be_a String
+      it 'has valid URL' do
+        is_expected.to include 'href="https://hackernoon.com/the-power-to-build-communities-a-response-to-mark-zuckerberg-3f2cac9148a4"'
       end
+    end
 
-      it 'contains plain text' do
-        expect(subject).to match('Hello world')
-      end
+    context 'matches a stand-alone google URL' do
+      let(:text) { 'http://google.com' }
 
-      it 'contains a link' do
-        expect(subject).to match('<a href="http://google.com/" rel="nofollow noopener" target="_blank"><span class="invisible">http://</span><span class="">google.com/</span><span class="invisible"></span></a>')
+      it 'has valid URL' do
+        is_expected.to include 'href="http://google.com/"'
       end
+    end
+
+    context 'matches a stand-alone IDN URL' do
+      let(:text) { 'https://nic.みんな/' }
 
-      it 'contains a mention' do
-        result = Formatter.instance.format(local_status_with_mention)
-        expect(result).to match "<a href=\"#{TagManager.instance.url_for(account)}\" class=\"u-url mention\">@<span>#{account.username}</span></a></span>"
-        expect(result).to match %r{href=\"http://google.com/\?x=@#{account.username}}
-        expect(result).not_to match "href=\"https://example.com/@#{account.username}"
+      it 'has valid URL' do
+        is_expected.to include 'href="https://nic.xn--q9jyb4c/"'
       end
 
-      it 'contains a hashtag' do
-        result = Formatter.instance.format(local_status_with_mention)
-        expect(result).to match('/tags/hashtag" class="mention hashtag" rel="tag">#<span>hashtag</span></a>')
+      it 'has display URL' do
+        is_expected.to include '<span class="">nic.みんな/</span>'
       end
     end
 
-    context 'with cashtag' do
-      let(:local_text) { 'Hello world $AAPL' }
+    context 'matches a URL without trailing period' do
+      let(:text) { 'http://www.mcmansionhell.com/post/156408871451/50-states-of-mcmansion-hell-scottsdale-arizona. ' }
 
-      it 'skip cashtag' do
-        expect(subject).to match '<p>Hello world $AAPL</p>'
+      it 'has valid URL' do
+        is_expected.to include 'href="http://www.mcmansionhell.com/post/156408871451/50-states-of-mcmansion-hell-scottsdale-arizona"'
       end
     end
 
-    context 'with reblog' do
-      let(:local_status) { Fabricate(:status, account: account, reblog: Fabricate(:status, text: 'Hello world', account: account)) }
+    context 'matches a URL without closing paranthesis' do
+      let(:text) { '(http://google.com/)' }
 
-      it 'contains credit to original author' do
-        expect(subject).to include("RT <span class=\"h-card\"><a href=\"#{TagManager.instance.url_for(account)}\" class=\"u-url mention\">@<span>#{account.username}</span></a></span> Hello world")
+      it 'has valid URL' do
+        is_expected.to include 'href="http://google.com/"'
       end
     end
 
-    context 'matches a stand-alone medium URL' do
-      let(:local_text) { 'https://hackernoon.com/the-power-to-build-communities-a-response-to-mark-zuckerberg-3f2cac9148a4' }
+    context 'matches a URL without exclamation point' do
+      let(:text) { 'http://www.google.com!' }
 
-      it 'has valid url' do
-        expect(subject).to include('href="https://hackernoon.com/the-power-to-build-communities-a-response-to-mark-zuckerberg-3f2cac9148a4"')
+      it 'has valid URL' do
+        is_expected.to include 'href="http://www.google.com/"'
       end
     end
 
-    context 'matches a stand-alone google URL' do
-      let(:local_text) { 'http://google.com' }
+    context 'matches a URL without single quote' do
+      let(:text) { "http://www.google.com'" }
 
-      it 'has valid url' do
-        expect(subject).to include('href="http://google.com/"')
+      it 'has valid URL' do
+        is_expected.to include 'href="http://www.google.com/"'
       end
     end
 
-    context 'matches a stand-alone IDN URL' do
-      let(:local_text) { 'https://nic.みんな/' }
+    context 'matches a URL without angle brackets' do
+      let(:text) { 'http://www.google.com>' }
 
-      it 'has valid url' do
-        expect(subject).to include('href="https://nic.xn--q9jyb4c/"')
+      it 'has valid URL' do
+        is_expected.to include 'href="http://www.google.com/"'
       end
+    end
 
-      it 'has display url' do
-        expect(subject).to include('<span class="">nic.みんな/</span>')
+    context 'matches a URL with a query string' do
+      let(:text) { 'https://www.ruby-toolbox.com/search?utf8=%E2%9C%93&q=autolink' }
+
+      it 'has valid URL' do
+        is_expected.to include 'href="https://www.ruby-toolbox.com/search?utf8=%E2%9C%93&amp;q=autolink"'
       end
     end
 
-    context 'matches a URL without trailing period' do
-      let(:local_text) { 'http://www.mcmansionhell.com/post/156408871451/50-states-of-mcmansion-hell-scottsdale-arizona. ' }
+    context 'matches a URL with parenthesis in it' do
+      let(:text) { 'https://en.wikipedia.org/wiki/Diaspora_(software)' }
 
-      it 'has valid url' do
-        expect(subject).to include('href="http://www.mcmansionhell.com/post/156408871451/50-states-of-mcmansion-hell-scottsdale-arizona"')
+      it 'has valid URL' do
+        is_expected.to include 'href="https://en.wikipedia.org/wiki/Diaspora_(software)"'
       end
     end
 
-    xit 'matches a URL without closing paranthesis' do
-      expect(subject.match('(http://google.com/)')[0]).to eq 'http://google.com'
+    context 'contains HTML (script tag)' do
+      let(:text) { '<script>alert("Hello")</script>' }
+
+      it 'has escaped HTML' do
+        is_expected.to include '<p>&lt;script&gt;alert(&quot;Hello&quot;)&lt;/script&gt;</p>'
+      end
     end
 
-    context 'matches a URL without exclamation point' do
-      let(:local_text) { 'http://www.google.com!' }
+    context 'contains HTML (XSS attack)' do
+      let(:text) { %q{<img src="javascript:alert('XSS');">} }
 
-      it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com/"')
+      it 'has escaped HTML' do
+        is_expected.to include '<p>&lt;img src=&quot;javascript:alert(&apos;XSS&apos;);&quot;&gt;</p>'
       end
     end
 
-    context 'matches a URL without single quote' do
-      let(:local_text) { "http://www.google.com'" }
+    context 'contains invalid URL' do
+      let(:text) { 'http://www\.google\.com' }
 
-      it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com/"')
+      it 'has raw URL' do
+        is_expected.to eq '<p>http://www\.google\.com</p>'
       end
     end
 
-    context 'matches a URL without angle brackets' do
-      let(:local_text) { 'http://www.google.com>' }
+    context 'contains a hashtag' do
+      let(:text)  { '#hashtag' }
 
-      it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com/"')
+      it 'has a link' do
+        is_expected.to include '/tags/hashtag" class="mention hashtag" rel="tag">#<span>hashtag</span></a>'
       end
     end
+  end
 
-    context 'matches a URL with a query string' do
-      let(:local_text) { 'https://www.ruby-toolbox.com/search?utf8=%E2%9C%93&q=autolink' }
+  describe '#format' do
+    subject { Formatter.instance.format(status) }
 
-      it 'has valid url' do
-        expect(subject).to include('href="https://www.ruby-toolbox.com/search?utf8=%E2%9C%93&amp;q=autolink"')
+    context 'with local status' do
+      context 'with reblog' do
+        let(:reblog) { Fabricate(:status, account: local_account, text: 'Hello world', uri: nil) }
+        let(:status) { Fabricate(:status, reblog: reblog) }
+
+        it 'returns original status with credit to its author' do
+          is_expected.to include 'RT <span class="h-card"><a href="https://cb6e6126.ngrok.io/@alice" class="u-url mention">@<span>alice</span></a></span> Hello world'
+        end
+      end
+
+      context 'contains plain text' do
+        let(:status)  { Fabricate(:status, text: 'text', uri: nil) }
+
+        it 'paragraphizes' do
+          is_expected.to eq '<p>text</p>'
+        end
+      end
+
+      context 'contains line feeds' do
+        let(:status)  { Fabricate(:status, text: "line\nfeed", uri: nil) }
+
+        it 'removes line feeds' do
+          is_expected.not_to include "\n"
+        end
+      end
+
+      context 'contains linkable mentions' do
+        let(:status) { Fabricate(:status, mentions: [ Fabricate(:mention, account: local_account) ], text: '@alice') }
+
+        it 'links' do
+          is_expected.to include '<a href="https://cb6e6126.ngrok.io/@alice" class="u-url mention">@<span>alice</span></a></span>'
+        end
+      end
+
+      context 'contains unlinkable mentions' do
+        let(:status) { Fabricate(:status, text: '@alice', uri: nil) }
+
+        it 'does not link' do
+          is_expected.to include '@alice'
+        end
+      end
+
+      context do
+        subject do
+          status = Fabricate(:status, text: text, uri: nil)
+          Formatter.instance.format(status)
+        end
+
+        include_examples 'encode and link URLs'
       end
     end
 
-    context 'matches a URL with parenthesis in it' do
-      let(:local_text) { 'https://en.wikipedia.org/wiki/Diaspora_(software)' }
+    context 'with remote status' do
+      let(:status) { Fabricate(:status, text: 'Beep boop', uri: 'beepboop') }
 
-      it 'has valid url' do
-        expect(subject).to include('href="https://en.wikipedia.org/wiki/Diaspora_(software)"')
+      it 'reformats' do
+        is_expected.to eq 'Beep boop'
       end
     end
+  end
+
+  describe '#reformat' do
+    subject { Formatter.instance.reformat(text) }
 
-    context 'contains html (script tag)' do
-      let(:local_text) { '<script>alert("Hello")</script>' }
+    context 'contains plain text' do
+      let(:text) { 'Beep boop' }
 
-      it 'has valid url' do
-        expect(subject).to match '<p>&lt;script&gt;alert(&quot;Hello&quot;)&lt;/script&gt;</p>'
+      it 'contains plain text' do
+        is_expected.to include 'Beep boop'
       end
     end
 
-    context 'contains html (xss attack)' do
-      let(:local_text) { %q{<img src="javascript:alert('XSS');">} }
+    context 'contains scripts' do
+      let(:text) { '<script>alert("Hello")</script>' }
 
-      it 'has valid url' do
-        expect(subject).to match '<p>&lt;img src=&quot;javascript:alert(&apos;XSS&apos;);&quot;&gt;</p>'
+      it 'strips scripts' do
+        is_expected.to_not include '<script>alert("Hello")</script>'
       end
     end
+  end
 
-    context 'contains invalid URL' do
-      let(:local_text) { 'http://www\.google\.com' }
+  describe '#plaintext' do
+    subject { Formatter.instance.plaintext(status) }
+
+    context 'with local status' do
+      let(:status)  { Fabricate(:status, text: '<p>a text by a nerd who uses an HTML tag in text</p>', uri: nil) }
 
-      it 'has valid url' do
-        expect(subject).to eq '<p>http://www\.google\.com</p>'
+      it 'returns raw text' do
+        is_expected.to eq '<p>a text by a nerd who uses an HTML tag in text</p>'
       end
     end
 
-    context 'concatenates hashtag and URL' do
-      let(:local_text) { '#hashtaghttps://www.google.com' }
+    context 'with remote status' do
+      let(:status)  { Fabricate(:status, text: '<script>alert("Hello")</script>', uri: 'beep boop') }
 
-      it 'has valid hashtag' do
-        expect(subject).to match('/tags/hashtag" class="mention hashtag" rel="tag">#<span>hashtag</span></a>')
+      it 'returns tag-stripped text' do
+        is_expected.to eq ''
       end
     end
   end
 
-  describe '#reformat' do
-    subject { Formatter.instance.format(remote_status) }
+  describe '#simplified_format' do
+    subject { Formatter.instance.simplified_format(account) }
+
+    context 'with local status' do
+      let(:account) { Fabricate(:account, domain: nil, note: text) }
+
+      context 'contains linkable mentions for local accounts' do
+        let(:text) { '@alice' }
+
+        before { local_account }
+
+        it 'links' do
+          is_expected.to eq '<p><span class="h-card"><a href="https://cb6e6126.ngrok.io/@alice" class="u-url mention">@<span>alice</span></a></span></p>'
+        end
+      end
+
+      context 'contains linkable mentions for remote accounts' do
+        let(:text) { '@bob@remote' }
+
+        before { remote_account }
+
+        it 'links' do
+          is_expected.to eq '<p><span class="h-card"><a href="https://remote/" class="u-url mention">@<span>bob</span></a></span></p>'
+        end
+      end
+
+      context 'contains unlinkable mentions' do
+        let(:text) { '@alice' }
 
-    it 'returns a string' do
-      expect(subject).to be_a String
+        it 'returns raw mention texts' do
+          is_expected.to eq '<p>@alice</p>'
+        end
+      end
+
+      include_examples 'encode and link URLs'
     end
 
-    it 'contains plain text' do
-      expect(subject).to match('Beep boop')
+    context 'with remote status' do
+      let(:text) { '<script>alert("Hello")</script>' }
+      let(:account) { Fabricate(:account, domain: 'remote', note: text) }
+
+      it 'reformats' do
+        is_expected.to_not include '<script>alert("Hello")</script>'
+      end
     end
+  end
+
+  describe '#sanitize' do
+    let(:html) { '<script>alert("Hello")</script>' }
+
+    subject { Formatter.instance.sanitize(html, Sanitize::Config::MASTODON_STRICT) }
 
-    it 'does not contain scripts' do
-      expect(subject).to_not match('<script>alert("Hello")</script>')
+    it 'sanitizes' do
+      is_expected.to eq 'alert("Hello")'
     end
   end
 end