Module: Kettle::Drift::DuplicateLineValidator

Defined in:
lib/kettle/drift/duplicate_line_validator.rb

Overview

Scans managed files for repeated adjacent-line chunks that usually signal duplication drift or template corruption.

Constant Summary collapse

DEFAULT_MIN_CHARS = 
6
BINARY_EXTENSIONS = 
Set.new(%w[
  .7z
  .bz2
  .class
  .dll
  .exe
  .gem
  .gif
  .gz
  .ico
  .jar
  .jpeg
  .jpg
  .pdf
  .png
  .so
  .tar
  .tgz
  .ttf
  .woff
  .woff2
  .xz
  .zip
]).freeze
APPRAISALS_DEP_LINE_RE = 
/\A(?:eval_gemfile|gem)\s+["']/
CHANGELOG_METRIC_RE = 
/\A-\s+(?:(?:(?:line|branch)\s+)?coverage:|\d+\.\d+%\s+documented)/i
EXCLUDED_FILENAMES = 
Set.new(["CODE_OF_CONDUCT.md", ".gitlab-ci.yml"]).freeze
KETTLE_JEM_CONFIG_RE = 
/\A(?:strategy|recipe|preference|add_missing|freeze_token|file_type|max_recursion_depth):\s/
RAKEFILE_ENV_ASSIGNMENT_RE = 
/\AENV\[["']/
RESCUE_LOAD_ERROR_RE = 
/\Arescue\s+LoadError/
COVERAGE_MARKER_RE = 
/\A# (?:simplecov:(?:disable|enable)|:nocov:)\z/
CHANGELOG_SUBHEADINGS = 
Set.new([
  "### Added",
  "### Changed",
  "### Deprecated",
  "### Removed",
  "### Fixed",
  "### Security"
]).freeze

Class Method Summary collapse

Class Method Details

.baseline(template_dir: nil, min_chars: DEFAULT_MIN_CHARS) ⇒ Object 



162
163
164
165
166
167
168
169
170
171
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 162

def baseline(template_dir: nil, min_chars: DEFAULT_MIN_CHARS)
  return Set.new unless template_dir && File.directory?(template_dir)

  template_files = Dir.glob(
    File.join(template_dir, "**", "*"),
    File::FNM_DOTMATCH
  ).select { |f| File.file?(f) }

  Set.new(scan(files: template_files, min_chars: min_chars).keys)
end

.binary_content?(path, raw) ⇒ Boolean

Returns:

  • (Boolean) 


248
249
250
251
252
253
254
255
256
257
258
259
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 248

def binary_content?(path, raw)
  return true if BINARY_EXTENSIONS.include?(File.extname(path.to_s).downcase)

  sample = raw.byteslice(0, 4096) || "".b
  return true if sample.include?("\x00")
  return false if sample.empty?

  control_bytes = sample.each_byte.count do |byte|
    (0..8).cover?(byte) || byte == 11 || byte == 12 || (14..31).cover?(byte)
  end
  control_bytes.fdiv(sample.bytesize) > 0.1
end

.compute_fence_lines(content) ⇒ Object 



134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 134

def compute_fence_lines(content)
  in_fence = false
  fence_marker = nil
  fence_lines = Set.new

  content.each_line.with_index(1) do |raw, lineno|
    stripped = raw.strip
    if in_fence
      fence_lines.add(lineno)
      if stripped.match?(/\A#{Regexp.escape(fence_marker)}\s*\z/)
        in_fence = false
        fence_marker = nil
      end
    elsif (match = stripped.match(/\A(`{3,}|~{3,})/))
      fence_marker = match[1]
      in_fence = true
      fence_lines.add(lineno)
    end
  end

  fence_lines
end

.dependabot_schedule_chunk?(path, line1, line2) ⇒ Boolean

Dependabot intentionally has one schedule mapping per ecosystem entry. This exact pair has no independent behavior to de-duplicate.

Returns:

  • (Boolean) 


128
129
130
131
132
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 128

def dependabot_schedule_chunk?(path, line1, line2)
  File.basename(path.to_s) == "dependabot.yml" &&
    File.dirname(path.to_s).end_with?(".github") &&
    line1.delete_prefix("- ") == "schedule:" && line2.start_with?("interval:")
end

.generated_workflow_matrix_chunk?(path, line1, line2) ⇒ Boolean

These generated workflows repeat the same matrix-field pairs once for every Ruby/engine job. Limit the exemption to those exact pairs so a duplicated step or an unrelated workflow regression is still reported.

Returns:

  • (Boolean) 


115
116
117
118
119
120
121
122
123
124
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 115

def generated_workflow_matrix_chunk?(path, line1, line2)
  return false unless %w[dep-heads.yml heads.yml].include?(File.basename(path.to_s))
  return false unless File.dirname(path.to_s).end_with?(".github/workflows")

  matrix_key = line1.delete_prefix("- ")
  return true if matrix_key.start_with?("appraisal: ") && line2.start_with?("exec_cmd: ")
  return true if line1.start_with?("exec_cmd: ") && (line2.start_with?("rubygems: ", "# Run directly from the generated appraisal Gemfile"))

  line1 == "if: ${{ !env.ACT }}" && line2.start_with?("uses: appraisal-rb/setup-ruby-flash@")
end

.ignored_duplicate_chunk?(path, line1, line2) ⇒ Boolean

Returns:

  • (Boolean) 


92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 92

def ignored_duplicate_chunk?(path, line1, line2)
  basename = File.basename(path.to_s)

  return true if generated_workflow_matrix_chunk?(path, line1, line2)
  return true if dependabot_schedule_chunk?(path, line1, line2)

  if basename == "Appraisals"
    return true if APPRAISALS_DEP_LINE_RE.match?(line1) && APPRAISALS_DEP_LINE_RE.match?(line2)
    return true if line1.start_with?("#") && APPRAISALS_DEP_LINE_RE.match?(line2)
  end

  return true if basename == "CHANGELOG.md" && CHANGELOG_METRIC_RE.match?(line1) && CHANGELOG_METRIC_RE.match?(line2)
  return true if basename == "Rakefile" && RAKEFILE_ENV_ASSIGNMENT_RE.match?(line1) && RAKEFILE_ENV_ASSIGNMENT_RE.match?(line2)
  return true if RESCUE_LOAD_ERROR_RE.match?(line1) && COVERAGE_MARKER_RE.match?(line2)
  return true if File.extname(path.to_s) == ".md" && line1.start_with?("|") && line2.start_with?("|")
  return true if basename == ".kettle-jem.yml" && KETTLE_JEM_CONFIG_RE.match?(line1) && KETTLE_JEM_CONFIG_RE.match?(line2)

  false
end

.read_text_content(path) ⇒ Object 



236
237
238
239
240
241
242
243
244
245
246
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 236

def read_text_content(path)
  raw = File.binread(path)
  return if binary_content?(path, raw)

  content = raw.dup.force_encoding(Encoding::UTF_8)
  return content if content.valid_encoding?

  content.scrub
rescue
  nil
end

.report_summary(results, project_root: nil) ⇒ Object 



212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 212

def report_summary(results, project_root: nil)
  return "No duplicate lines detected.\n" if results.empty?

  lines = ["### Duplicate Line Report\n", "| Chunk (line1 ↵ line2) | File | Start Lines |", "|---|---|---|"]

  results.each do |content, entries|
    display = content.gsub("\n", " ↵ ")
    display = "#{display[0, 77]}..." if display.length > 80
    display = display.gsub("|", "\\|")

    entries.each do |entry|
      file = Kettle::Drift.display_path(entry[:file])
      if project_root
        display_root = Kettle::Drift.display_path(project_root)
        file = file.sub(%r{^#{Regexp.escape(display_root)}/?}, "")
      end
      lines << "| `#{display}` | #{file} | #{entry[:lines].join(", ")} |"
    end
  end

  lines << ""
  lines.join("\n")
end

.scan(files:, min_chars: DEFAULT_MIN_CHARS) ⇒ Object 



54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 54

def scan(files:, min_chars: DEFAULT_MIN_CHARS)
  duplicates = {}

  files.each do |path|
    next unless File.file?(path)
    next if EXCLUDED_FILENAMES.include?(File.basename(path.to_s))

    content = read_text_content(path)
    next unless content

    fence_lines = (File.extname(path.to_s) == ".md") ? compute_fence_lines(content) : Set.new
    indexed = content.each_line.map.with_index(1) { |raw, n| [n, raw.strip] }

    chunk_map = Hash.new { |h, k| h[k] = [] }
    indexed.each_cons(2) do |(lineno1, line1), (lineno2, line2)|
      next if line1.gsub(/\s/, "").length <= min_chars
      next if line2.gsub(/\s/, "").length <= min_chars
      next if CHANGELOG_SUBHEADINGS.include?(line1)
      next if fence_lines.include?(lineno1) && fence_lines.include?(lineno2)
      next if ignored_duplicate_chunk?(path, line1, line2)

      chunk_map["#{line1}\n#{line2}"] << lineno1
    end

    chunk_map.each do |chunk_content, start_lines|
      next if start_lines.size < 2

      duplicates[chunk_content] ||= []
      duplicates[chunk_content] << {
        file: path,
        lines: start_lines
      }
    end
  end

  duplicates
end

.scan_template_results(template_results:, min_chars: DEFAULT_MIN_CHARS) ⇒ Object 



157
158
159
160
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 157

def scan_template_results(template_results:, min_chars: DEFAULT_MIN_CHARS)
  written_files = template_results.select { |_path, rec| %i[create replace].include?(rec[:action]) }.keys
  scan(files: written_files, min_chars: min_chars)
end

.subtract_baseline(results, baseline_set:) ⇒ Object 



173
174
175
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 173

def subtract_baseline(results, baseline_set:)
  results.except(*baseline_set)
end

.template_managed_files(project_root:, template_dir: nil) ⇒ Object 



177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 177

def template_managed_files(project_root:, template_dir: nil)
  template_dir ||= File.join(project_root, "template")
  return [] unless File.directory?(template_dir)

  managed = []
  Dir.glob(File.join(template_dir, "**", "*"), File::FNM_DOTMATCH).each do |src|
    next unless File.file?(src)

    rel = src.sub(%r{^#{Regexp.escape(template_dir)}/?}, "")
    rel = rel.sub(/\.example\z/, "")
    next if rel.include?(".no-osc")

    dest = File.join(project_root, rel)
    managed << dest if File.file?(dest)
  end

  managed.uniq
end

.to_json(results) ⇒ Object 



200
201
202
203
204
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 200

def to_json(results)
  JSON.pretty_generate(results.transform_values do |entries|
    entries.map { |entry| {file: Kettle::Drift.display_path(entry[:file]), lines: entry[:lines]} }
  end)
end

.warning_count(results) ⇒ Object 



196
197
198
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 196

def warning_count(results)
  results.values.flatten.size
end

.write_json(results, json_path) ⇒ Object 



206
207
208
209
210
 
# File 'lib/kettle/drift/duplicate_line_validator.rb', line 206

def write_json(results, json_path)
  FileUtils.mkdir_p(File.dirname(json_path))
  File.write(json_path, to_json(results))
  json_path
end