[#32] Summarize large arrays with a hash in dump output by default

[SkowronskiAndrew]

Summary

Fixes #32.

When diffing dumps of two builds to find object changes, large arrays of basic types (mesh vertex data, texture bytes, etc.) were a problem either way: dumped in full they drown the output, and --skip-large-arrays printed <Skipped>, hiding real content differences. Following the idea of binary2text's -largebinaryhashonly, the dump command now summarizes arrays of basic types with more than 256 elements as a hash of their raw bytes:

m_IndexBuffer (vector)
  Array<UInt8>[2232]
    ArrayDataHash 3be77a33

This is the new default behavior — a deliberate, modest compatibility break, on the theory that hiding huge arrays is the better default and the hash keeps diffs meaningful. The full content is still available with the new -a / --show-large-arrays option.

Changes

• TextDumperTool: basic-type arrays larger than 256 elements (the same threshold the skip option used) print ArrayDataHash <crc32 hex> instead of their elements, unless ShowLargeArrays is set. The hash is a CRC32 computed via the existing UnityFileReader.ComputeCRC, so no new dependency (binary2text uses Unity's internal 64-bit hash, which isn't available here; for diff detection CRC32 serves the same purpose). The <Skipped> behavior is removed.
• CLI: new -a / --show-large-arrays option. The old -s / --skip-large-arrays is still accepted so existing scripts don't break, but it is ignored (its behavior is essentially the default now) and hidden from --help to keep the help simple.
• Tests: the per-version expected dump snapshots now show hashes; the dump-s snapshots are renamed to dump-a and hold the full-content output (byte-identical to the previous default dump snapshots). New tests cover -a, the ignored -s, and a known-value hash check: the SerializationDemo int array 0..511, whose CRC32 (6feca6e2) was verified against an independent zlib crc32 computation.
• Documentation: command-dump.md gains a "Large Arrays" section and updated option table.

Testing

dotnet test -c Release — full suite green (773 passed, 10 skipped, 0 failed).

Manually verified default, -a, and legacy -s invocations on the LeadingEdge AssetBundles, and that -s no longer appears in dump --help.

[Copilot]

Pull request overview

Updates the dump command’s text output to keep diffs useful while avoiding massive inline array dumps, by default summarizing large basic-type arrays as a CRC32 hash (with an option to print full array contents when needed).

Changes:

• Default dump output now summarizes basic-type arrays over 256 elements as ArrayDataHash <crc32>.
• Adds -a/--show-large-arrays to print full large-array contents; keeps -s/--skip-large-arrays accepted but hidden and ignored for compatibility.
• Updates tests, expected-data generation, and documentation to reflect the new default and options.

Reviewed changes

Copilot reviewed 6 out of 19 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
UnityDataTool/Program.cs	Adds --show-large-arrays, hides legacy --skip-large-arrays, wires new option into dump execution.
UnityDataTool.Tests/UnityDataToolAssetBundleTests.cs	Updates/expands CLI integration tests for new default behavior, -a, and ignored -s.
UnityDataTool.Tests/ExpectedDataGenerator.cs	Renames/regenerates expected dump snapshots to match new defaults and -a.
UnityDataTool.Tests/DumpTests.cs	Adds stdout test coverage for default hash summarization and --show-large-arrays.
TextDumper/TextDumperTool.cs	Implements hash summarization for large basic-type arrays and introduces MaxInlineArraySize.
Documentation/command-dump.md	Documents new default behavior, adds Large Arrays section, updates option table/examples.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.