|
| 1 | +# spec/streaming_128kb_cliff_spec.rb |
| 2 | +require "spec_helper" |
| 3 | +require "stringio" |
| 4 | +require "zstd-ruby" |
| 5 | + |
| 6 | +RSpec.describe "Zstd streaming 128KB cliff" do |
| 7 | + MAGIC = [0x28, 0xB5, 0x2F, 0xFD].freeze |
| 8 | + |
| 9 | + def hex(str) |
| 10 | + str.unpack1("H*") |
| 11 | + end |
| 12 | + |
| 13 | + def has_magic?(bin) |
| 14 | + bytes = bin.bytes |
| 15 | + bytes[0, 4] == MAGIC |
| 16 | + end |
| 17 | + |
| 18 | + shared_examples "round-trip streaming compress" do |size| |
| 19 | + it "round-trips #{size} bytes and starts with zstd magic" do |
| 20 | + # Produce data |
| 21 | + src = "a" * size |
| 22 | + |
| 23 | + sc = Zstd::StreamingCompress.new |
| 24 | + sc << src |
| 25 | + compressed = sc.finish |
| 26 | + |
| 27 | + expect(has_magic?(compressed)).to be(true), "missing magic: #{hex(compressed)[0, 16]}" |
| 28 | + expect(Zstd.decompress(compressed)).to eq(src) |
| 29 | + end |
| 30 | + end |
| 31 | + |
| 32 | + context "exactly around 128 KiB" do |
| 33 | + include_examples "round-trip streaming compress", 131_071 |
| 34 | + include_examples "round-trip streaming compress", 131_072 # the cliff |
| 35 | + include_examples "round-trip streaming compress", 131_073 |
| 36 | + end |
| 37 | + |
| 38 | + context "multiple writes crossing the boundary" do |
| 39 | + it "64KiB + 64KiB (exact boundary) works" do |
| 40 | + part = "x" * 65_536 |
| 41 | + sc = Zstd::StreamingCompress.new |
| 42 | + sc << part |
| 43 | + sc << part |
| 44 | + compressed = sc.finish |
| 45 | + expect(has_magic?(compressed)).to be(true) |
| 46 | + expect(Zstd.decompress(compressed)).to eq(part * 2) |
| 47 | + end |
| 48 | + |
| 49 | + it "64KiB + 64KiB + 1 works" do |
| 50 | + a = "a" * 65_536 |
| 51 | + b = "b" * 65_536 |
| 52 | + c = "c" |
| 53 | + sc = Zstd::StreamingCompress.new |
| 54 | + sc << a << b << c |
| 55 | + compressed = sc.finish |
| 56 | + expect(has_magic?(compressed)).to be(true) |
| 57 | + expect(Zstd.decompress(compressed)).to eq(a + b + c) |
| 58 | + end |
| 59 | + end |
| 60 | + |
| 61 | + context "flush/end draining" do |
| 62 | + it "returns all produced bytes across flush and finish" do |
| 63 | + sc = Zstd::StreamingCompress.new |
| 64 | + sc << ("a" * 70_000) |
| 65 | + out = sc.flush |
| 66 | + expect(out).not_to be_empty |
| 67 | + sc << ("b" * 70_000) |
| 68 | + out << sc.finish |
| 69 | + expect(has_magic?(out)).to be(true) |
| 70 | + expect(Zstd.decompress(out)).to eq(("a" * 70_000) + ("b" * 70_000)) |
| 71 | + end |
| 72 | + end |
| 73 | + |
| 74 | + context "GC.compact interaction" do |
| 75 | + it "survives compaction around the boundary" do |
| 76 | + sc = Zstd::StreamingCompress.new |
| 77 | + sc << ("a" * 80_000) |
| 78 | + GC.compact if GC.respond_to?(:compact) |
| 79 | + sc << ("b" * 60_000) # total now > 128KiB |
| 80 | + compressed = sc.finish |
| 81 | + expect(has_magic?(compressed)).to be(true) |
| 82 | + expect(Zstd.decompress(compressed)).to eq(("a" * 80_000) + ("b" * 60_000)) |
| 83 | + end |
| 84 | + end |
| 85 | + |
| 86 | + context "larger payload sanity" do |
| 87 | + it "round-trips ~1 MiB" do |
| 88 | + src = "z" * 1_048_576 |
| 89 | + sc = Zstd::StreamingCompress.new(level: 3) |
| 90 | + sc << src |
| 91 | + compressed = sc.finish |
| 92 | + expect(has_magic?(compressed)).to be(true) |
| 93 | + expect(Zstd.decompress(compressed)).to eq(src) |
| 94 | + end |
| 95 | + end |
| 96 | + |
| 97 | + describe Zstd::StreamWriter do |
| 98 | + it "produces a valid frame and round-trips at exactly 128KiB" do |
| 99 | + io = StringIO.new |
| 100 | + sw = Zstd::StreamWriter.new(io) |
| 101 | + sw.write("a" * 131_072) |
| 102 | + sw.finish |
| 103 | + |
| 104 | + io.rewind |
| 105 | + bin = io.read |
| 106 | + expect(has_magic?(bin)).to be(true), "missing magic: #{hex(bin)[0, 16]}" |
| 107 | + |
| 108 | + io.rewind |
| 109 | + sr = Zstd::StreamReader.new(io) |
| 110 | + out = sr.read(2_000_000) |
| 111 | + expect(out.size).to eq(131_072) |
| 112 | + expect(out).to eq("a" * 131_072) |
| 113 | + io.close |
| 114 | + end |
| 115 | + |
| 116 | + it "handles boundary-crossing writes with flush in between" do |
| 117 | + io = StringIO.new |
| 118 | + sw = Zstd::StreamWriter.new(io) |
| 119 | + sw.write("a" * 70_000) |
| 120 | + sw.write("b" * 70_000) # crosses 128KiB internally |
| 121 | + sw.finish |
| 122 | + |
| 123 | + io.rewind |
| 124 | + bin = io.read |
| 125 | + expect(has_magic?(bin)).to be(true) |
| 126 | + io.rewind |
| 127 | + sr = Zstd::StreamReader.new(io) |
| 128 | + out = sr.read(1_000_000) |
| 129 | + expect(out).to eq("a" * 70_000 + "b" * 70_000) |
| 130 | + io.close |
| 131 | + end |
| 132 | + |
| 133 | + it "survives GC.compact mid-stream" do |
| 134 | + io = StringIO.new |
| 135 | + sw = Zstd::StreamWriter.new(io) |
| 136 | + sw.write("x" * 90_000) |
| 137 | + GC.compact if GC.respond_to?(:compact) |
| 138 | + sw.write("y" * 50_000) |
| 139 | + sw.finish |
| 140 | + |
| 141 | + io.rewind |
| 142 | + bin = io.read |
| 143 | + expect(has_magic?(bin)).to be(true) |
| 144 | + io.rewind |
| 145 | + sr = Zstd::StreamReader.new(io) |
| 146 | + out = sr.read(200_000) |
| 147 | + expect(out).to eq("x" * 90_000 + "y" * 50_000) |
| 148 | + end |
| 149 | + end |
| 150 | +end |
0 commit comments