Add adaptive max_depth and max_items handling in diff_objects

lib/difftastic/differ.rb

@@ -2,8 +2,15 @@
 
 class Difftastic::Differ
 	DEFAULT_TAB_WIDTH = 2
-
-	def initialize(background: nil, color: nil, syntax_highlight: nil, context: nil, width: nil, tab_width: nil, parse_error_limit: nil, underline_highlights: true, left_label: nil, right_label: nil, display: "side-by-side-show-both")
+	DEFAULT_MAX_DEPTH = 5
+	DEFAULT_MAX_ITEMS = 10
+	DEFAULT_MAX_DEPTH_CAP = 20
+	DEFAULT_MAX_ITEMS_CAP = 40
+	MAX_DEPTH_INCREMENT = 5
+	MAX_ITEMS_INCREMENT = 10
+	DIFF_UNAVAILABLE_MESSAGE = "[Diff unavailable: exceeded depth/size display limits]"
+
+	def initialize(background: nil, color: nil, syntax_highlight: nil, context: nil, width: nil, tab_width: nil, parse_error_limit: nil, underline_highlights: true, left_label: nil, right_label: nil, display: "side-by-side-show-both", max_depth: nil, max_items: nil, max_depth_cap: nil, max_items_cap: nil, max_depth_increment: nil, max_items_increment: nil)
 		@show_paths = false
 		@background = background => :dark | :light | nil
 		@color = color => :always | :never | :auto | nil
@@ -16,15 +23,44 @@ def initialize(background: nil, color: nil, syntax_highlight: nil, context: nil,
 		@left_label = left_label => String | nil
 		@right_label = right_label => String | nil
 		@display = display
+		@max_depth = max_depth => Integer | nil
+		@max_items = max_items => Integer | nil
+		@max_depth_cap = max_depth_cap => Integer | nil
+		@max_items_cap = max_items_cap => Integer | nil
+		@max_depth_increment = max_depth_increment => Integer | nil
+		@max_items_increment = max_items_increment => Integer | nil
 	end
 
 	def diff_objects(old, new)
 		tab_width = @tab_width || DEFAULT_TAB_WIDTH
+		max_depth = @max_depth || DEFAULT_MAX_DEPTH
+		max_items = @max_items || DEFAULT_MAX_ITEMS
+		max_depth_cap = @max_depth_cap || DEFAULT_MAX_DEPTH_CAP
+		max_items_cap = @max_items_cap || DEFAULT_MAX_ITEMS_CAP
+		max_depth_increment = @max_depth_increment || MAX_DEPTH_INCREMENT
+		max_items_increment = @max_items_increment || MAX_ITEMS_INCREMENT
+
+		loop do
+			old_str = Difftastic.pretty(old, tab_width:, max_depth:, max_items:)
+			new_str = Difftastic.pretty(new, tab_width:, max_depth:, max_items:)
+
+			# If prettified strings don't differ, the strings probably were truncated to max_depth and/or max_items.
+			# PrettyPlease then correctly did not return non-matching strings.
+			# In that case we increase max_depth & max_items in the loop up to DEFAULT_MAX_ITEMS_CAP or DEFAULT_MAX_DEPTH_CAP.
+			if old_str != new_str
+				return diff_strings(old_str, new_str, file_extension: "rb")
+			end
 
-		old = Difftastic.pretty(old, tab_width:)
-		new = Difftastic.pretty(new, tab_width:)
+			# If we've hit both caps, stop trying
+			if max_depth >= max_depth_cap && max_items >= max_items_cap
+				return DIFF_UNAVAILABLE_MESSAGE
+			end
 
-		diff_strings(old, new, file_extension: "rb")
+			# Increase limits and retry, while never increasing to more than max_depth_cap/max_items_cap.
+			# Both values are matched to perform 3 retries, but neither value would increase beyond its cap anyway.
+			max_depth = [max_depth + max_depth_increment, max_depth_cap].min
+			max_items = [max_items + max_items_increment, max_items_cap].min
+		end
 	end
 
 	def diff_ada(old, new)

test/diff_objects_adaptive_test.rb

@@ -0,0 +1,249 @@
+# frozen_string_literal: true
+
+require_relative "test_helper"
+
+class DiffObjectsAdaptiveTest < Minitest::Spec
+	# Use smaller values than the gem's defaults to keep tests fast.
+	# The adaptive logic works the same regardless of the actual values.
+	TEST_MAX_DEPTH = 3
+	TEST_MAX_ITEMS = 5
+	TEST_MAX_DEPTH_CAP = 10
+	TEST_MAX_ITEMS_CAP = 12
+	TEST_INCREMENT = 1
+	DIFF_UNAVAILABLE_MESSAGE = Difftastic::Differ::DIFF_UNAVAILABLE_MESSAGE
+
+	def differ(**options)
+		Difftastic::Differ.new(
+			color: :never,
+			max_depth: TEST_MAX_DEPTH,
+			max_items: TEST_MAX_ITEMS,
+			max_depth_cap: TEST_MAX_DEPTH_CAP,
+			max_items_cap: TEST_MAX_ITEMS_CAP,
+			max_depth_increment: TEST_INCREMENT,
+			max_items_increment: TEST_INCREMENT,
+			**options # overrides defaults when provided
+		)
+	end
+
+	# Example:
+	#   nested_hash(4, "x")
+	#   # => { l1: { l2: { l3: { l4: "x" } } } }
+	def nested_hash(depth, nested_value)
+		(1..depth)
+			.reverse_each
+			.reduce(nested_value) do |inner, i|
+				{ "l#{i}": inner }
+			end
+	end
+
+	# Example:
+	#   array_with_diff_at(4, "x")
+	#   # => [1, 2, 3, "x"]
+	def array_with_diff_at(position, value)
+		(1...position).to_a + [value]
+	end
+
+	describe "adaptive max_depth" do
+		it "shows diff at exactly TEST_MAX_DEPTH" do
+			old = nested_hash(TEST_MAX_DEPTH, "old")
+			new = nested_hash(TEST_MAX_DEPTH, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "shows diff at TEST_MAX_DEPTH + 1 (requires adaptation)" do
+			depth = TEST_MAX_DEPTH + 1
+			old = nested_hash(depth, "old")
+			new = nested_hash(depth, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "shows diff at 2 * TEST_MAX_DEPTH (requires multiple adaptations)" do
+			depth = 2 * TEST_MAX_DEPTH
+			old = nested_hash(depth, "old")
+			new = nested_hash(depth, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+	end
+
+	describe "adaptive max_items" do
+		it "shows diff at exactly TEST_MAX_ITEMS" do
+			old = array_with_diff_at(TEST_MAX_ITEMS, "old")
+			new = array_with_diff_at(TEST_MAX_ITEMS, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "shows diff at TEST_MAX_ITEMS + 1 (requires adaptation)" do
+			position = TEST_MAX_ITEMS + 1
+			old = array_with_diff_at(position, "old")
+			new = array_with_diff_at(position, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "shows diff at 2 * TEST_MAX_ITEMS (requires multiple adaptations)" do
+			position = 2 * TEST_MAX_ITEMS
+			old = array_with_diff_at(position, "old")
+			new = array_with_diff_at(position, "new")
+
+			output = differ.diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+	end
+
+	describe "configurable starting values" do
+		it "respects custom max_depth" do
+			depth = TEST_MAX_DEPTH_CAP - 1
+			old = nested_hash(depth, "old")
+			new = nested_hash(depth, "new")
+
+			output = differ(max_depth: 1).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "respects custom max_items" do
+			position = TEST_MAX_ITEMS_CAP - 1
+			old = array_with_diff_at(position, "old")
+			new = array_with_diff_at(position, "new")
+
+			output = differ(max_items: 1).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+	end
+
+	describe "configurable caps" do
+		it "respects custom max_depth_cap" do
+			depth = TEST_MAX_DEPTH - 1
+			old = nested_hash(depth, "old")
+			new = nested_hash(depth, "new")
+
+			output = differ(max_depth_cap: TEST_MAX_DEPTH).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "respects custom max_items_cap" do
+			position = TEST_MAX_ITEMS - 1
+			old = array_with_diff_at(position, "old")
+			new = array_with_diff_at(position, "new")
+
+			output = differ(max_items_cap: TEST_MAX_ITEMS).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "returns unavailable when depth exceeds custom max_depth_cap" do
+			cap = 2
+			old = nested_hash(cap, "old")
+			new = nested_hash(cap, "new")
+
+			# Both caps must be hit (&&), and starting max must be smaller than cap to trigger truncation
+			output = differ(max_depth: 1, max_depth_cap: cap, max_items_cap: 1).diff_objects(old, new)
+
+			assert_includes output, DIFF_UNAVAILABLE_MESSAGE
+			refute_includes output, "old"
+			refute_includes output, "new"
+		end
+
+		it "returns unavailable when position exceeds custom max_items_cap" do
+			cap = 2
+			old = array_with_diff_at(cap, "old")
+			new = array_with_diff_at(cap, "new")
+
+			# Both caps must be hit (&&), and starting max must be smaller than cap to trigger truncation
+			output = differ(max_items: 1, max_depth_cap: 1, max_items_cap: cap).diff_objects(old, new)
+
+			assert_includes output, DIFF_UNAVAILABLE_MESSAGE
+			refute_includes output, "old"
+			refute_includes output, "new"
+		end
+
+		it "uses starting max_items when higher than cap without adaptation" do
+			cap = 2
+			old = array_with_diff_at(cap, "old")
+			new = array_with_diff_at(cap, "new")
+
+			# With max_items: 5 (default) > max_items_cap: 2, no truncation occurs, diff found immediately
+			output = differ(max_items_cap: cap).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+
+		it "uses starting max_depth when higher than cap without adaptation" do
+			cap = 2
+			old = nested_hash(cap, "old")
+			new = nested_hash(cap, "new")
+
+			# With max_depth: 3 (default) > max_depth_cap: 2, no truncation occurs, diff found immediately
+			output = differ(max_depth_cap: cap).diff_objects(old, new)
+
+			refute_includes output, DIFF_UNAVAILABLE_MESSAGE
+			assert_includes output, "old"
+			assert_includes output, "new"
+		end
+	end
+
+	describe "loop termination at caps" do
+		it "returns unavailable message when depth exceeds max_depth_cap" do
+			depth = TEST_MAX_DEPTH_CAP + 1
+			old = nested_hash(depth, "old")
+			new = nested_hash(depth, "new")
+
+			output = differ.diff_objects(old, new)
+
+			assert_includes output, DIFF_UNAVAILABLE_MESSAGE
+			refute_includes output, "old"
+			refute_includes output, "new"
+		end
+
+		it "returns unavailable message when position exceeds max_items_cap" do
+			position = TEST_MAX_ITEMS_CAP + 1
+			old = array_with_diff_at(position, "old")
+			new = array_with_diff_at(position, "new")
+
+			output = differ.diff_objects(old, new)
+
+			assert_includes output, DIFF_UNAVAILABLE_MESSAGE
+			refute_includes output, "old"
+			refute_includes output, "new"
+		end
+	end
+end