Class: Banzai::Filter::BaseSanitizationFilter

Inherits:
HTML::Pipeline::SanitizationFilter
  • Object
show all
Extended by:
Gitlab::Utils::SanitizeNodeLink
Includes:
Concerns::TimeoutFilterHandler, Gitlab::Utils::StrongMemoize
Defined in:
lib/banzai/filter/base_sanitization_filter.rb

Overview

Sanitize HTML produced by markup languages (Markdown, AsciiDoc…). Specific rules are implemented in dedicated filters:

  • Banzai::Filter::SanitizationFilter (Markdown)

  • Banzai::Filter::AsciiDocSanitizationFilter (AsciiDoc/Asciidoctor)

  • Banzai::Filter::BroadcastMessageSanitizationFilter (Markdown with styled links and line breaks)

  • Banzai::Filter::MinimumMarkdownSanitizationFilter (only inline elements)

Extends HTML::Pipeline::SanitizationFilter with common rules.

Constant Summary collapse

UNSAFE_PROTOCOLS =
%w[data javascript vbscript].freeze

Constants included from Gitlab::Utils::SanitizeNodeLink

Gitlab::Utils::SanitizeNodeLink::ATTRS_TO_SANITIZE

Constants included from Concerns::TimeoutFilterHandler

Concerns::TimeoutFilterHandler::COMPLEX_MARKDOWN_MESSAGE, Concerns::TimeoutFilterHandler::RENDER_TIMEOUT, Concerns::TimeoutFilterHandler::SANITIZATION_RENDER_TIMEOUT

Class Method Summary collapse

Instance Method Summary collapse

Methods included from Gitlab::Utils::SanitizeNodeLink

permit_url?, remove_unsafe_links, safe_protocol?, sanitize_unsafe_links

Class Method Details

.remove_relObject



88
89
90
91
92
93
94
95
96
97
98
# File 'lib/banzai/filter/base_sanitization_filter.rb', line 88

def remove_rel
  ->(env) do
    if env[:node_name] == 'a'
      # we allow rel="license" to support the Rel-license microformat
      # http://microformats.org/wiki/rel-license
      unless env[:node].attribute('rel')&.value == 'license'
        env[:node].remove_attribute('rel')
      end
    end
  end
end

Instance Method Details

#allowlistObject



25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
# File 'lib/banzai/filter/base_sanitization_filter.rb', line 25

def allowlist
  strong_memoize(:allowlist) do
    allowlist = super.deep_dup

    # Allow span elements
    allowlist[:elements].push('span')

    # Allow data-math-style attribute in order to support LaTeX formatting
    allowlist[:attributes]['span'] = %w[data-math-style]
    allowlist[:attributes]['code'] = %w[data-math-style]
    allowlist[:attributes]['pre'] = %w[data-canonical-lang data-lang-params
      data-math-style data-mermaid-style data-kroki-style]

    # Allow data-placeholder from gitlab-glfm-markdown
    allowlist[:attributes]['span'].push('data-placeholder')

    # Allow html5 details/summary elements
    allowlist[:elements].push('details')
    allowlist[:elements].push('summary')

    # Allow abbr elements with title attribute
    allowlist[:elements].push('abbr')
    allowlist[:attributes]['abbr'] = %w[title]

    # Disallow `name` attribute globally, allow on `a`
    allowlist[:attributes][:all].delete('name')
    allowlist[:attributes]['a'].push('name')

    allowlist[:attributes]['a'].push('data-wikilink')
    allowlist[:attributes]['a'].push('data-placeholder')

    allowlist[:attributes]['img'].push('data-diagram')
    allowlist[:attributes]['img'].push('data-diagram-src')
    allowlist[:attributes]['img'].push('data-placeholder')

    # Allow any protocol in `a` elements
    # and then remove links with unsafe protocols in SanitizeLinkFilter
    allowlist[:protocols].delete('a')

    # Remove `rel` attribute from `a` elements
    allowlist[:transformers].push(self.class.remove_rel)

    customize_allowlist(allowlist)
  end
end

#callObject



21
22
23
# File 'lib/banzai/filter/base_sanitization_filter.rb', line 21

def call
  Sanitize.clean_node!(doc, allowlist)
end

#customize_allowlist(allowlist) ⇒ Object

Raises:

  • (NotImplementedError)


71
72
73
# File 'lib/banzai/filter/base_sanitization_filter.rb', line 71

def customize_allowlist(allowlist)
  raise NotImplementedError
end