refactor: DRY Page class, fix bugs, enforce autoload convention

.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2026-05-13 08:31:40 UTC using RuboCop version 1.86.1.
+# on 2026-05-13 13:11:16 UTC using RuboCop version 1.86.1.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
@@ -19,14 +19,6 @@ Layout/LineLength:
     - 'lib/archaeo/url_rewriter.rb'
     - 'spec/archaeo/asset_extractor_spec.rb'
 
-# Offense count: 2
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyle, IndentationWidth.
-# SupportedStyles: aligned, indented, indented_relative_to_receiver
-Layout/MultilineMethodCallIndentation:
-  Exclude:
-    - 'lib/archaeo/page.rb'
-
 # Offense count: 3
 Lint/UselessConstantScoping:
   Exclude:
@@ -46,7 +38,7 @@ Metrics/AbcSize:
     - 'lib/archaeo/cli.rb'
     - 'lib/archaeo/configuration.rb'
     - 'lib/archaeo/content_tracker.rb'
-    - 'lib/archaeo/coverage_report.rb'
+    - 'lib/archaeo/coverage_analyzer.rb'
     - 'lib/archaeo/local_rewriter.rb'
     - 'lib/archaeo/parallel_cdx.rb'
     - 'lib/archaeo/pattern_filter.rb'
@@ -114,14 +106,6 @@ Performance/StringBytesize:
   Exclude:
     - 'lib/archaeo/encoding_detector.rb'
 
-# Offense count: 1
-# This cop supports unsafe autocorrection (--autocorrect-all).
-# Configuration parameters: SkipBlocks, EnforcedStyle, OnlyStaticConstants.
-# SupportedStyles: described_class, explicit
-RSpec/DescribedClass:
-  Exclude:
-    - 'spec/archaeo/page_microposts_spec.rb'
-
 # Offense count: 131
 # Configuration parameters: CountAsOne.
 RSpec/ExampleLength:
@@ -178,12 +162,3 @@ RSpec/RepeatedExampleGroupDescription:
 RSpec/StubbedMock:
   Exclude:
     - 'spec/archaeo/parallel_cdx_spec.rb'
-
-# Offense count: 3
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyleForMultiline.
-# SupportedStylesForMultiline: comma, consistent_comma, diff_comma, no_comma
-Style/TrailingCommaInArguments:
-  Exclude:
-    - 'lib/archaeo/cli.rb'
-    - 'lib/archaeo/page.rb'

lib/archaeo.rb

@@ -71,4 +71,7 @@ def initialize(message, status_code:, url:, page:)
   autoload :SearchResult, "archaeo/archive_search"
   autoload :LocalRewriter, "archaeo/local_rewriter"
   autoload :LocalRewriteSummary, "archaeo/local_rewriter"
+  autoload :CoverageAnalyzer, "archaeo/coverage_analyzer"
+  autoload :HealthReport, "archaeo/health_report"
+  autoload :HealthDetail, "archaeo/health_report"
 end

lib/archaeo/archive_health_check.rb

@@ -1,20 +1,6 @@
 # frozen_string_literal: true
 
 module Archaeo
-  # Verifies that archived snapshots are still accessible.
-  #
-  # Checks each snapshot by performing HEAD requests to the
-  # archive URL and reporting accessibility status.
-  HealthReport = Struct.new(
-    :total, :accessible, :missing, :errors, :details,
-    keyword_init: true
-  )
-
-  HealthDetail = Struct.new(
-    :snapshot, :status, :error,
-    keyword_init: true
-  )
-
   class ArchiveHealthCheck
     def initialize(client: HttpClient.new, cdx_api: nil)
       @client = client

lib/archaeo/bulk_downloader.rb

@@ -245,9 +245,6 @@ def fetch_and_save(snapshot)
       page = fetch_page(snapshot)
       validate_page_status(page, snapshot)
       write_page_file(page, snapshot)
-    rescue StandardError
-      FileUtils.rm_f(tmp_path) if defined?(tmp_path)
-      raise
     end
 
     def fetch_page(snapshot)
@@ -272,6 +269,9 @@ def write_page_file(page, snapshot)
       File.binwrite(tmp_path, page.content)
       File.rename(tmp_path, filename)
       page.content
+    rescue StandardError
+      FileUtils.rm_f(tmp_path) if defined?(tmp_path)
+      raise
     end
 
     EXTENSION_MAP = {

lib/archaeo/coverage_analyzer.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+module Archaeo
+  # Builds a CoverageReport from CDX snapshot data.
+  class CoverageAnalyzer
+    def initialize(cdx_api: nil)
+      @cdx_api = cdx_api
+    end
+
+    def analyze(url, from: nil, to: nil)
+      cdx = @cdx_api || CdxApi.new
+      snapshots = cdx.snapshots(url, from: from, to: to).to_a
+
+      unique_urls = snapshots.map(&:original_url).uniq
+      status_dist = compute_status_distribution(snapshots)
+      gaps = compute_temporal_gaps(snapshots)
+
+      CoverageReport.new(
+        url: url,
+        total_urls: unique_urls.size,
+        archived_urls: snapshots.count(&:success?),
+        status_distribution: status_dist,
+        temporal_gaps: gaps,
+      )
+    end
+
+    private
+
+    def compute_status_distribution(snapshots)
+      snapshots.each_with_object(Hash.new(0)) do |snap, counts|
+        counts[snap.status_code] += 1
+      end
+    end
+
+    def compute_temporal_gaps(snapshots)
+      return [] if snapshots.size < 2
+
+      sorted = snapshots.sort_by(&:timestamp)
+      gaps = []
+      sorted.each_cons(2) do |a, b|
+        diff_days = (b.timestamp.to_time - a.timestamp.to_time) / 86400
+        next unless diff_days > 30
+
+        gaps << { from: a.timestamp.to_s, to: b.timestamp.to_s,
+                  gap_days: diff_days.round }
+      end
+      gaps
+    end
+  end
+end

lib/archaeo/coverage_report.rb

@@ -51,51 +51,4 @@ def as_json(*)
       to_h
     end
   end
-
-  # Builds a CoverageReport from CDX snapshot data.
-  class CoverageAnalyzer
-    def initialize(cdx_api: nil)
-      @cdx_api = cdx_api
-    end
-
-    def analyze(url, from: nil, to: nil)
-      cdx = @cdx_api || CdxApi.new
-      snapshots = cdx.snapshots(url, from: from, to: to).to_a
-
-      unique_urls = snapshots.map(&:original_url).uniq
-      status_dist = compute_status_distribution(snapshots)
-      gaps = compute_temporal_gaps(snapshots)
-
-      CoverageReport.new(
-        url: url,
-        total_urls: unique_urls.size,
-        archived_urls: snapshots.count(&:success?),
-        status_distribution: status_dist,
-        temporal_gaps: gaps,
-      )
-    end
-
-    private
-
-    def compute_status_distribution(snapshots)
-      snapshots.each_with_object(Hash.new(0)) do |snap, counts|
-        counts[snap.status_code] += 1
-      end
-    end
-
-    def compute_temporal_gaps(snapshots)
-      return [] if snapshots.size < 2
-
-      sorted = snapshots.sort_by(&:timestamp)
-      gaps = []
-      sorted.each_cons(2) do |a, b|
-        diff_days = (b.timestamp.to_time - a.timestamp.to_time) / 86400
-        next unless diff_days > 30
-
-        gaps << { from: a.timestamp.to_s, to: b.timestamp.to_s,
-                  gap_days: diff_days.round }
-      end
-      gaps
-    end
-  end
 end

lib/archaeo/download_state.rb

@@ -73,18 +73,15 @@ def file_exists?(timestamp, base_dir: @output_dir)
     def stale_entries(base_dir: @output_dir)
       @mutex.synchronize do
         entries.reject do |e|
-          find_file(base_dir,
-                    e["ts"]) && File.exist?(find_file(base_dir, e["ts"]))
+          path = find_file(base_dir, e["ts"])
+          path && File.exist?(path)
         end
       end
     end
 
     def cleanup_stale(base_dir: @output_dir)
       @mutex.synchronize do
-        stale = entries.reject do |e|
-          path = find_file(base_dir, e["ts"])
-          path && File.exist?(path)
-        end
+        stale = stale_entries(base_dir: base_dir)
         @entries = entries - stale
         @entries_key = nil
         save

lib/archaeo/health_report.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Archaeo
+  HealthReport = Struct.new(
+    :total, :accessible, :missing, :errors, :details,
+    keyword_init: true
+  )
+
+  HealthDetail = Struct.new(
+    :snapshot, :status, :error,
+    keyword_init: true
+  )
+end