docuseal/bin/rebrand-check

#!/usr/bin/env ruby
# frozen_string_literal: true
#
# bin/rebrand-check — fails (exit 1) if any unintended "docuseal" /
# "Docuseal" / "DocuSeal" references survive in the tree.
#
# Intended survivors are documented in REBRANDING.md:
#   - AGPL §7(b) attribution (NOTICE, README, LICENSE_ADDITIONAL_TERMS,
#     _powered_by, _email_attribution, completed.vue's credit, the
#     based_on i18n keys, Wabosign::UPSTREAM_NAME / UPSTREAM_URL)
#   - SDK custom-element names <docuseal-form>, <docuseal-builder>
#   - npm packages @docuseal/{react,vue,angular,embed}
#   - upstream binary URLs github.com/docusealco/{fields-detection,
#     pdfium-binaries,turbo}
#   - DocuSeal LLC copyright header in calculator.js
#
# Any other occurrence is treated as an accidental survivor and printed.
#
# Wired into CI via .github/workflows/ci.yml.

require 'find'
require 'set'

ROOT = File.expand_path('..', __dir__)

DENY_PREFIXES = [
  '.git/', '.github/', 'node_modules/', 'vendor/bundle/', 'vendor/cache/',
  'tmp/', 'log/', 'pg_data/', 'public/packs/', 'public/packs-test/',
  'wabosign/', '.claude/', 'coverage/'
].freeze

BINARY_EXT = Set.new(%w[
  .png .jpg .jpeg .gif .ico .svg .pdf .zip .tgz .gz .bz2 .xz
  .woff .woff2 .ttf .eot .otf
  .pkcs12 .p12 .pem .crt .cer .key
  .db .sqlite .sqlite3 .so .dylib .dll .exe .onnx
]).freeze

# Whole files we ignore — every "docuseal" hit in them is intentional.
ALLOW_FILES = Set.new([
  'NOTICE',
  'LICENSE',
  'LICENSE_ADDITIONAL_TERMS',
  'REBRANDING.md',
  'CHANGELOG.md',
  'README.md',
  'GOOGLE_SSO.md',
  'SMS.md',
  'app/javascript/submission_form/calculator.js',
  'app/javascript/submission_form/completed.vue',
  'app/views/shared/_powered_by.html.erb',
  'app/views/shared/_email_attribution.html.erb',
  'bin/rebrand-sync',
  'bin/rebrand-check',
  'bin/sync-upstream',
  # Fork-invariant guard + its manifest deliberately name "docuseal" tokens
  # (e.g. lib/docuseal.rb, docuseal-form) as markers to assert on.
  'bin/fork-check',
  'config/fork_invariants.yml',
  'config/brand_assets.sha256',
  'lib/wabosign.rb',
  'lib/docuseal.rb',
  # Migration that finds rows by the legacy docuseal_aatl name.
  'db/migrate/20260515183000_rename_docuseal_aatl_cert.rb'
]).freeze

ALLOW_FILE_PREFIXES = ['docs/embedding/', 'docs/api/'].freeze

# Per-line allowed patterns. If every "docuseal" hit on a line is
# inside one of these, the line is tolerated.
ALLOW_PATTERNS = [
  # SDK custom element tags — bare, attributed, HTML-encoded, quoted
  %r{<docuseal-(form|builder)\b},
  %r{</docuseal-(form|builder)>},
  %r{&lt;/?docuseal-(form|builder)},
  %r{['"]docuseal-(form|builder)['"]},
  # SDK class names exported by the npm packages
  %r{\bDocuseal(Form|Builder)(Component)?\b},
  # npm package paths
  %r{@docuseal/(react|vue|angular|embed)},
  # Upstream binary release URLs
  %r{docusealco/fields-detection},
  %r{docusealco/pdfium-binaries},
  %r{docusealco/turbo},
  # Programmatic custom-element registration
  %r{customElements\.define\(['"]docuseal-},
  # Constants holding the AGPL §7(b) attribution
  %r{Wabosign::UPSTREAM_(NAME|URL)},
  # AGPL attribution idioms — preserve the literal "DocuSeal" credit
  %r{\b(?:fork(?:ed)? of|forked from|based on|derived from)\s+DocuSeal\b},
  # JS calculator copyright header
  %r{DocuSeal,?\s+LLC},
  # i18n key carrying the AGPL credit value
  %r{based_on:},
  # i18n keys (underscore-separated) — these are identifiers, not user text
  # Matches both YAML key lines and t() calls
  %r{[a-zA-Z_0-9]*docuseal[a-zA-Z_0-9]*:\s},
  %r{t\(['"][a-zA-Z_0-9_]*docuseal[a-zA-Z_0-9_]*['"]\)},
  # i18n keys with leading spaces (nested YAML)
  %r{\s+[a-zA-Z_0-9_]*docuseal[a-zA-Z_0-9_]*:},
  # t() calls referencing docuseal i18n keys
  %r{t\(['"][a-z_]*docuseal[a-z_]*['"]\)},
  # Twitter handles and org references in meta tags
  %r{twitter:(?:creator|site).*content=".*docusealco"},
  # ENV variable names (uppercase)
  %r{DOCUSEAL_API_KEY},
  # github.com/docusealco URLs (upstream binary/ repo references)
  %r{github\.com/docusealco}
].freeze

def deny_listed?(rel)
  return true if ALLOW_FILES.include?(rel)
  return true if ALLOW_FILE_PREFIXES.any? { |p| rel.start_with?(p) }
  DENY_PREFIXES.any? { |p| rel.start_with?(p) }
end

def likely_binary?(path)
  return true if BINARY_EXT.include?(File.extname(path).downcase)
  head = File.binread(path, 1024)
  head.include?("\x00")
rescue StandardError
  false
end

HIT_RE = /docuseal/i

def line_is_tolerated?(line)
  scrubbed = line.dup
  ALLOW_PATTERNS.each { |re| scrubbed = scrubbed.gsub(re, '') }
  !scrubbed.match?(HIT_RE)
end

violations = []

Find.find(ROOT) do |path|
  rel = path.sub(%r{\A#{Regexp.escape(ROOT)}/}, '')
  if File.directory?(path)
    Find.prune if rel != '' && DENY_PREFIXES.any? { |p| "#{rel}/".start_with?(p) }
    next
  end
  next if deny_listed?(rel)
  next if likely_binary?(path)

  begin
    File.foreach(path, encoding: 'UTF-8').with_index(1) do |line, lineno|
      next unless line.match?(HIT_RE)
      next if line_is_tolerated?(line)

      violations << "#{rel}:#{lineno}: #{line.chomp}"
    end
  rescue Encoding::InvalidByteSequenceError, ArgumentError
    next
  end
end

if violations.empty?
  puts 'rebrand-check: ok'
  exit 0
else
  warn "rebrand-check: #{violations.size} unintended DocuSeal reference(s):"
  violations.each { |v| warn "  #{v}" }
  exit 1
end