Truncate all non-blob markdown to 1MB by default

and prepend a user message if the limit is over a
certain threshold

Changelog: security

app/helpers/markup_helper.rb

@@ -118,6 +118,7 @@ module MarkupHelper
 
   def markup(file_name, text, context = {})
     context[:project] ||= @project
+    context[:text_source] ||= :blob
     html = context.delete(:rendered) || markup_unsafe(file_name, text, context)
     prepare_for_rendering(html, context)
   end

lib/banzai/filter/truncate_source_filter.rb

@@ -3,12 +3,29 @@
 module Banzai
   module Filter
     class TruncateSourceFilter < HTML::Pipeline::TextFilter
+      CHARACTER_COUNT_LIMIT = 1.megabyte
+      USER_MSG_LIMIT = 10_000
+
       def call
-        return text unless context.key?(:limit)
+        # don't truncate if it's a :blob and no limit is set
+        return text if context[:text_source] == :blob && !context.key?(:limit)
+
+        limit = context[:limit] || CHARACTER_COUNT_LIMIT
+
+        # no sense in allowing `truncate_bytes` to duplicate a large
+        # string unless it's too big
+        return text if text.bytesize <= limit
 
         # Use three dots instead of the ellipsis Unicode character because
         # some clients show the raw Unicode value in the merge commit.
-        text.truncate_bytes(context[:limit], omission: '...')
+        trunc = text.truncate_bytes(limit, omission: '...')
+
+        # allows us to indicate to the user that what they see is a truncated copy
+        if limit > USER_MSG_LIMIT
+          trunc.prepend("_The text is longer than #{limit} characters and has been visually truncated._\n\n")
+        end
+
+        trunc
       end
     end
   end

spec/helpers/markup_helper_spec.rb

@@ -418,6 +418,13 @@ FooBar
   describe '#markup' do
     let(:content) { 'Noël' }
 
+    it 'sets the :text_source to :blob in the context' do
+      context = {}
+      helper.markup('foo.md', content, context)
+
+      expect(context).to include(text_source: :blob)
+    end
+
     it 'preserves encoding' do
       expect(content.encoding.name).to eq('UTF-8')
       expect(helper.markup('foo.rst', content).encoding.name).to eq('UTF-8')

spec/lib/banzai/filter/truncate_source_filter_spec.rb

@@ -8,24 +8,68 @@ RSpec.describe Banzai::Filter::TruncateSourceFilter do
   let(:short_text) { 'foo' * 10 }
   let(:long_text) { ([short_text] * 10).join(' ') }
 
-  it 'does nothing when limit is unspecified' do
-    output = filter(long_text)
-
-    expect(output).to eq(long_text)
+  before do
+    stub_const("#{described_class}::CHARACTER_COUNT_LIMIT", 50)
+    stub_const("#{described_class}::USER_MSG_LIMIT", 20)
   end
 
-  it 'does nothing to a short-enough text' do
-    output = filter(short_text, limit: short_text.bytesize)
+  context 'when markdown belongs to a blob' do
+    it 'does nothing when limit is unspecified' do
+      output = filter(long_text, text_source: :blob)
+
+      expect(output).to eq(long_text)
+    end
+
+    it 'truncates normally when limit specified' do
+      truncated = 'foofoof...'
+
+      output = filter(long_text, text_source: :blob, limit: 10)
 
-    expect(output).to eq(short_text)
+      expect(output).to eq(truncated)
+    end
   end
 
-  it 'truncates UTF-8 text by bytes, on a character boundary' do
-    utf8_text = '日本語の文字が大きい'
-    truncated = '日...'
+  context 'when markdown belongs to a field (non-blob)' do
+    it 'does nothing when limit is greater' do
+      output = filter(long_text, limit: 1.megabyte)
+
+      expect(output).to eq(long_text)
+    end
+
+    it 'truncates to the default when limit is unspecified' do
+      stub_const("#{described_class}::USER_MSG_LIMIT", 200)
+      truncated = 'foofoofoofoofoofoofoofoofoofoo foofoofoofoofoof...'
+
+      output = filter(long_text)
+
+      expect(output).to eq(truncated)
+    end
+
+    it 'prepends the user message' do
+      truncated = <<~TEXT
+        _The text is longer than 50 characters and has been visually truncated._
+
+        foofoofoofoofoofoofoofoofoofoo foofoofoofoofoof...
+      TEXT
+
+      output = filter(long_text)
+
+      expect(output).to eq(truncated.strip)
+    end
+
+    it 'does nothing to a short-enough text' do
+      output = filter(short_text, limit: short_text.bytesize)
+
+      expect(output).to eq(short_text)
+    end
+
+    it 'truncates UTF-8 text by bytes, on a character boundary' do
+      utf8_text = '日本語の文字が大きい'
+      truncated = '日...'
 
-    expect(filter(utf8_text, limit: truncated.bytesize)).to eq(truncated)
-    expect(filter(utf8_text, limit: utf8_text.bytesize)).to eq(utf8_text)
-    expect(filter(utf8_text, limit: utf8_text.mb_chars.size)).not_to eq(utf8_text)
+      expect(filter(utf8_text, limit: truncated.bytesize)).to eq(truncated)
+      expect(filter(utf8_text, limit: utf8_text.bytesize)).to eq(utf8_text)
+      expect(filter(utf8_text, limit: utf8_text.mb_chars.size)).not_to eq(utf8_text)
+    end
   end
 end