| ... | ... | @@ -3,10 +3,10 @@ const assert = std.debug.assert; |
| 3 | 3 | |
| 4 | 4 | const types = @import("types.zig"); |
| 5 | 5 | const frame = types.frame; |
| 6 | | const Literals = types.compressed_block.Literals; |
| 7 | | const Sequences = types.compressed_block.Sequences; |
| 6 | const LiteralsSection = types.compressed_block.LiteralsSection; |
| 7 | const SequencesSection = types.compressed_block.SequencesSection; |
| 8 | 8 | const Table = types.compressed_block.Table; |
| 9 | | const RingBuffer = @import("RingBuffer.zig"); |
| 9 | pub const RingBuffer = @import("RingBuffer.zig"); |
| 10 | 10 | |
| 11 | 11 | const readInt = std.mem.readIntLittle; |
| 12 | 12 | const readIntSlice = std.mem.readIntSliceLittle; |
| ... | ... | @@ -55,7 +55,7 @@ pub fn decodeFrame(dest: []u8, src: []const u8, verify_checksum: bool) !ReadWrit |
| 55 | 55 | }; |
| 56 | 56 | } |
| 57 | 57 | |
| 58 | | const DecodeState = struct { |
| 58 | pub const DecodeState = struct { |
| 59 | 59 | repeat_offsets: [3]u32, |
| 60 | 60 | |
| 61 | 61 | offset: StateData(8), |
| ... | ... | @@ -70,7 +70,7 @@ const DecodeState = struct { |
| 70 | 70 | |
| 71 | 71 | literal_stream_reader: ReverseBitReader, |
| 72 | 72 | literal_stream_index: usize, |
| 73 | | huffman_tree: ?Literals.HuffmanTree, |
| 73 | huffman_tree: ?LiteralsSection.HuffmanTree, |
| 74 | 74 | |
| 75 | 75 | literal_written_count: usize, |
| 76 | 76 | |
| ... | ... | @@ -84,7 +84,55 @@ const DecodeState = struct { |
| 84 | 84 | }; |
| 85 | 85 | } |
| 86 | 86 | |
| 87 | | fn readInitialState(self: *DecodeState, bit_reader: anytype) !void { |
| 87 | pub fn prepare( |
| 88 | self: *DecodeState, |
| 89 | src: []const u8, |
| 90 | literals: LiteralsSection, |
| 91 | sequences_header: SequencesSection.Header, |
| 92 | ) !usize { |
| 93 | if (literals.huffman_tree) |tree| { |
| 94 | self.huffman_tree = tree; |
| 95 | } else if (literals.header.block_type == .treeless and self.huffman_tree == null) { |
| 96 | return error.TreelessLiteralsFirst; |
| 97 | } |
| 98 | |
| 99 | switch (literals.header.block_type) { |
| 100 | .raw, .rle => {}, |
| 101 | .compressed, .treeless => { |
| 102 | self.literal_stream_index = 0; |
| 103 | switch (literals.streams) { |
| 104 | .one => |slice| try self.initLiteralStream(slice), |
| 105 | .four => |streams| try self.initLiteralStream(streams[0]), |
| 106 | } |
| 107 | }, |
| 108 | } |
| 109 | |
| 110 | if (sequences_header.sequence_count > 0) { |
| 111 | var bytes_read = try self.updateFseTable( |
| 112 | src, |
| 113 | .literal, |
| 114 | sequences_header.literal_lengths, |
| 115 | ); |
| 116 | |
| 117 | bytes_read += try self.updateFseTable( |
| 118 | src[bytes_read..], |
| 119 | .offset, |
| 120 | sequences_header.offsets, |
| 121 | ); |
| 122 | |
| 123 | bytes_read += try self.updateFseTable( |
| 124 | src[bytes_read..], |
| 125 | .match, |
| 126 | sequences_header.match_lengths, |
| 127 | ); |
| 128 | self.fse_tables_undefined = false; |
| 129 | |
| 130 | return bytes_read; |
| 131 | } |
| 132 | return 0; |
| 133 | } |
| 134 | |
| 135 | pub fn readInitialFseState(self: *DecodeState, bit_reader: anytype) !void { |
| 88 | 136 | self.literal.state = try bit_reader.readBitsNoEof(u9, self.literal.accuracy_log); |
| 89 | 137 | self.offset.state = try bit_reader.readBitsNoEof(u8, self.offset.accuracy_log); |
| 90 | 138 | self.match.state = try bit_reader.readBitsNoEof(u9, self.match.accuracy_log); |
| ... | ... | @@ -130,7 +178,7 @@ const DecodeState = struct { |
| 130 | 178 | self: *DecodeState, |
| 131 | 179 | src: []const u8, |
| 132 | 180 | comptime choice: DataType, |
| 133 | | mode: Sequences.Header.Mode, |
| 181 | mode: SequencesSection.Header.Mode, |
| 134 | 182 | ) !usize { |
| 135 | 183 | const field_name = @tagName(choice); |
| 136 | 184 | switch (mode) { |
| ... | ... | @@ -213,7 +261,13 @@ const DecodeState = struct { |
| 213 | 261 | }; |
| 214 | 262 | } |
| 215 | 263 | |
| 216 | | fn executeSequenceSlice(self: *DecodeState, dest: []u8, write_pos: usize, literals: Literals, sequence: Sequence) !void { |
| 264 | fn executeSequenceSlice( |
| 265 | self: *DecodeState, |
| 266 | dest: []u8, |
| 267 | write_pos: usize, |
| 268 | literals: LiteralsSection, |
| 269 | sequence: Sequence, |
| 270 | ) !void { |
| 217 | 271 | try self.decodeLiteralsSlice(dest[write_pos..], literals, sequence.literal_length); |
| 218 | 272 | |
| 219 | 273 | // TODO: should we validate offset against max_window_size? |
| ... | ... | @@ -225,7 +279,12 @@ const DecodeState = struct { |
| 225 | 279 | std.mem.copy(u8, dest[write_pos + sequence.literal_length ..], dest[copy_start..copy_end]); |
| 226 | 280 | } |
| 227 | 281 | |
| 228 | | fn executeSequenceRingBuffer(self: *DecodeState, dest: *RingBuffer, literals: Literals, sequence: Sequence) !void { |
| 282 | fn executeSequenceRingBuffer( |
| 283 | self: *DecodeState, |
| 284 | dest: *RingBuffer, |
| 285 | literals: LiteralsSection, |
| 286 | sequence: Sequence, |
| 287 | ) !void { |
| 229 | 288 | try self.decodeLiteralsRingBuffer(dest, literals, sequence.literal_length); |
| 230 | 289 | // TODO: check that ring buffer window is full enough for match copies |
| 231 | 290 | const copy_slice = dest.sliceAt(dest.write_index + dest.data.len - sequence.offset, sequence.match_length); |
| ... | ... | @@ -234,11 +293,11 @@ const DecodeState = struct { |
| 234 | 293 | for (copy_slice.second) |b| dest.writeAssumeCapacity(b); |
| 235 | 294 | } |
| 236 | 295 | |
| 237 | | fn decodeSequenceSlice( |
| 296 | pub fn decodeSequenceSlice( |
| 238 | 297 | self: *DecodeState, |
| 239 | 298 | dest: []u8, |
| 240 | 299 | write_pos: usize, |
| 241 | | literals: Literals, |
| 300 | literals: LiteralsSection, |
| 242 | 301 | bit_reader: anytype, |
| 243 | 302 | last_sequence: bool, |
| 244 | 303 | ) !usize { |
| ... | ... | @@ -255,10 +314,10 @@ const DecodeState = struct { |
| 255 | 314 | return sequence.match_length + sequence.literal_length; |
| 256 | 315 | } |
| 257 | 316 | |
| 258 | | fn decodeSequenceRingBuffer( |
| 317 | pub fn decodeSequenceRingBuffer( |
| 259 | 318 | self: *DecodeState, |
| 260 | 319 | dest: *RingBuffer, |
| 261 | | literals: Literals, |
| 320 | literals: LiteralsSection, |
| 262 | 321 | bit_reader: anytype, |
| 263 | 322 | last_sequence: bool, |
| 264 | 323 | ) !usize { |
| ... | ... | @@ -280,7 +339,7 @@ const DecodeState = struct { |
| 280 | 339 | return sequence.match_length + sequence.literal_length; |
| 281 | 340 | } |
| 282 | 341 | |
| 283 | | fn nextLiteralMultiStream(self: *DecodeState, literals: Literals) !void { |
| 342 | fn nextLiteralMultiStream(self: *DecodeState, literals: LiteralsSection) !void { |
| 284 | 343 | self.literal_stream_index += 1; |
| 285 | 344 | try self.initLiteralStream(literals.streams.four[self.literal_stream_index]); |
| 286 | 345 | } |
| ... | ... | @@ -290,7 +349,7 @@ const DecodeState = struct { |
| 290 | 349 | try self.literal_stream_reader.init(bytes); |
| 291 | 350 | } |
| 292 | 351 | |
| 293 | | fn decodeLiteralsSlice(self: *DecodeState, dest: []u8, literals: Literals, len: usize) !void { |
| 352 | pub fn decodeLiteralsSlice(self: *DecodeState, dest: []u8, literals: LiteralsSection, len: usize) !void { |
| 294 | 353 | if (self.literal_written_count + len > literals.header.regenerated_size) return error.MalformedLiteralsLength; |
| 295 | 354 | switch (literals.header.block_type) { |
| 296 | 355 | .raw => { |
| ... | ... | @@ -310,7 +369,7 @@ const DecodeState = struct { |
| 310 | 369 | // const written_bytes_per_stream = (literals.header.regenerated_size + 3) / 4; |
| 311 | 370 | const huffman_tree = self.huffman_tree orelse unreachable; |
| 312 | 371 | const max_bit_count = huffman_tree.max_bit_count; |
| 313 | | const starting_bit_count = Literals.HuffmanTree.weightToBitCount( |
| 372 | const starting_bit_count = LiteralsSection.HuffmanTree.weightToBitCount( |
| 314 | 373 | huffman_tree.nodes[huffman_tree.symbol_count_minus_one].weight, |
| 315 | 374 | max_bit_count, |
| 316 | 375 | ); |
| ... | ... | @@ -345,7 +404,7 @@ const DecodeState = struct { |
| 345 | 404 | }, |
| 346 | 405 | .index => |index| { |
| 347 | 406 | huffman_tree_index = index; |
| 348 | | const bit_count = Literals.HuffmanTree.weightToBitCount( |
| 407 | const bit_count = LiteralsSection.HuffmanTree.weightToBitCount( |
| 349 | 408 | huffman_tree.nodes[index].weight, |
| 350 | 409 | max_bit_count, |
| 351 | 410 | ); |
| ... | ... | @@ -359,7 +418,7 @@ const DecodeState = struct { |
| 359 | 418 | } |
| 360 | 419 | } |
| 361 | 420 | |
| 362 | | fn decodeLiteralsRingBuffer(self: *DecodeState, dest: *RingBuffer, literals: Literals, len: usize) !void { |
| 421 | pub fn decodeLiteralsRingBuffer(self: *DecodeState, dest: *RingBuffer, literals: LiteralsSection, len: usize) !void { |
| 363 | 422 | if (self.literal_written_count + len > literals.header.regenerated_size) return error.MalformedLiteralsLength; |
| 364 | 423 | switch (literals.header.block_type) { |
| 365 | 424 | .raw => { |
| ... | ... | @@ -378,7 +437,7 @@ const DecodeState = struct { |
| 378 | 437 | // const written_bytes_per_stream = (literals.header.regenerated_size + 3) / 4; |
| 379 | 438 | const huffman_tree = self.huffman_tree orelse unreachable; |
| 380 | 439 | const max_bit_count = huffman_tree.max_bit_count; |
| 381 | | const starting_bit_count = Literals.HuffmanTree.weightToBitCount( |
| 440 | const starting_bit_count = LiteralsSection.HuffmanTree.weightToBitCount( |
| 382 | 441 | huffman_tree.nodes[huffman_tree.symbol_count_minus_one].weight, |
| 383 | 442 | max_bit_count, |
| 384 | 443 | ); |
| ... | ... | @@ -413,7 +472,7 @@ const DecodeState = struct { |
| 413 | 472 | }, |
| 414 | 473 | .index => |index| { |
| 415 | 474 | huffman_tree_index = index; |
| 416 | | const bit_count = Literals.HuffmanTree.weightToBitCount( |
| 475 | const bit_count = LiteralsSection.HuffmanTree.weightToBitCount( |
| 417 | 476 | huffman_tree.nodes[index].weight, |
| 418 | 477 | max_bit_count, |
| 419 | 478 | ); |
| ... | ... | @@ -647,54 +706,6 @@ fn decodeRleBlockRingBuffer(dest: *RingBuffer, src: []const u8, block_size: u21, |
| 647 | 706 | return block_size; |
| 648 | 707 | } |
| 649 | 708 | |
| 650 | | fn prepareDecodeState( |
| 651 | | decode_state: *DecodeState, |
| 652 | | src: []const u8, |
| 653 | | literals: Literals, |
| 654 | | sequences_header: Sequences.Header, |
| 655 | | ) !usize { |
| 656 | | if (literals.huffman_tree) |tree| { |
| 657 | | decode_state.huffman_tree = tree; |
| 658 | | } else if (literals.header.block_type == .treeless and decode_state.huffman_tree == null) { |
| 659 | | return error.TreelessLiteralsFirst; |
| 660 | | } |
| 661 | | |
| 662 | | switch (literals.header.block_type) { |
| 663 | | .raw, .rle => {}, |
| 664 | | .compressed, .treeless => { |
| 665 | | decode_state.literal_stream_index = 0; |
| 666 | | switch (literals.streams) { |
| 667 | | .one => |slice| try decode_state.initLiteralStream(slice), |
| 668 | | .four => |streams| try decode_state.initLiteralStream(streams[0]), |
| 669 | | } |
| 670 | | }, |
| 671 | | } |
| 672 | | |
| 673 | | if (sequences_header.sequence_count > 0) { |
| 674 | | var bytes_read = try decode_state.updateFseTable( |
| 675 | | src, |
| 676 | | .literal, |
| 677 | | sequences_header.literal_lengths, |
| 678 | | ); |
| 679 | | |
| 680 | | bytes_read += try decode_state.updateFseTable( |
| 681 | | src[bytes_read..], |
| 682 | | .offset, |
| 683 | | sequences_header.offsets, |
| 684 | | ); |
| 685 | | |
| 686 | | bytes_read += try decode_state.updateFseTable( |
| 687 | | src[bytes_read..], |
| 688 | | .match, |
| 689 | | sequences_header.match_lengths, |
| 690 | | ); |
| 691 | | decode_state.fse_tables_undefined = false; |
| 692 | | |
| 693 | | return bytes_read; |
| 694 | | } |
| 695 | | return 0; |
| 696 | | } |
| 697 | | |
| 698 | 709 | pub fn decodeBlock( |
| 699 | 710 | dest: []u8, |
| 700 | 711 | src: []const u8, |
| ... | ... | @@ -715,7 +726,7 @@ pub fn decodeBlock( |
| 715 | 726 | const literals = try decodeLiteralsSection(src, &bytes_read); |
| 716 | 727 | const sequences_header = try decodeSequencesHeader(src[bytes_read..], &bytes_read); |
| 717 | 728 | |
| 718 | | bytes_read += try prepareDecodeState(decode_state, src[bytes_read..], literals, sequences_header); |
| 729 | bytes_read += try decode_state.prepare(src[bytes_read..], literals, sequences_header); |
| 719 | 730 | |
| 720 | 731 | var bytes_written: usize = 0; |
| 721 | 732 | if (sequences_header.sequence_count > 0) { |
| ... | ... | @@ -723,7 +734,7 @@ pub fn decodeBlock( |
| 723 | 734 | var bit_stream: ReverseBitReader = undefined; |
| 724 | 735 | try bit_stream.init(bit_stream_bytes); |
| 725 | 736 | |
| 726 | | try decode_state.readInitialState(&bit_stream); |
| 737 | try decode_state.readInitialFseState(&bit_stream); |
| 727 | 738 | |
| 728 | 739 | var i: usize = 0; |
| 729 | 740 | while (i < sequences_header.sequence_count) : (i += 1) { |
| ... | ... | @@ -780,7 +791,7 @@ pub fn decodeBlockRingBuffer( |
| 780 | 791 | const literals = try decodeLiteralsSection(src, &bytes_read); |
| 781 | 792 | const sequences_header = try decodeSequencesHeader(src[bytes_read..], &bytes_read); |
| 782 | 793 | |
| 783 | | bytes_read += try prepareDecodeState(decode_state, src[bytes_read..], literals, sequences_header); |
| 794 | bytes_read += try decode_state.prepare(src[bytes_read..], literals, sequences_header); |
| 784 | 795 | |
| 785 | 796 | var bytes_written: usize = 0; |
| 786 | 797 | if (sequences_header.sequence_count > 0) { |
| ... | ... | @@ -788,7 +799,7 @@ pub fn decodeBlockRingBuffer( |
| 788 | 799 | var bit_stream: ReverseBitReader = undefined; |
| 789 | 800 | try bit_stream.init(bit_stream_bytes); |
| 790 | 801 | |
| 791 | | try decode_state.readInitialState(&bit_stream); |
| 802 | try decode_state.readInitialFseState(&bit_stream); |
| 792 | 803 | |
| 793 | 804 | var i: usize = 0; |
| 794 | 805 | while (i < sequences_header.sequence_count) : (i += 1) { |
| ... | ... | @@ -928,7 +939,7 @@ pub fn decodeBlockHeader(src: *const [3]u8) frame.ZStandard.Block.Header { |
| 928 | 939 | }; |
| 929 | 940 | } |
| 930 | 941 | |
| 931 | | pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals { |
| 942 | pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !LiteralsSection { |
| 932 | 943 | // TODO: we probably want to enable safety for release-fast and release-small (or insert custom checks) |
| 933 | 944 | var bytes_read: usize = 0; |
| 934 | 945 | const header = decodeLiteralsHeader(src, &bytes_read); |
| ... | ... | @@ -936,7 +947,7 @@ pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals |
| 936 | 947 | .raw => { |
| 937 | 948 | const stream = src[bytes_read .. bytes_read + header.regenerated_size]; |
| 938 | 949 | consumed_count.* += header.regenerated_size + bytes_read; |
| 939 | | return Literals{ |
| 950 | return LiteralsSection{ |
| 940 | 951 | .header = header, |
| 941 | 952 | .huffman_tree = null, |
| 942 | 953 | .streams = .{ .one = stream }, |
| ... | ... | @@ -945,7 +956,7 @@ pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals |
| 945 | 956 | .rle => { |
| 946 | 957 | const stream = src[bytes_read .. bytes_read + 1]; |
| 947 | 958 | consumed_count.* += 1 + bytes_read; |
| 948 | | return Literals{ |
| 959 | return LiteralsSection{ |
| 949 | 960 | .header = header, |
| 950 | 961 | .huffman_tree = null, |
| 951 | 962 | .streams = .{ .one = stream }, |
| ... | ... | @@ -966,7 +977,7 @@ pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals |
| 966 | 977 | const stream = src[bytes_read .. bytes_read + total_streams_size]; |
| 967 | 978 | bytes_read += total_streams_size; |
| 968 | 979 | consumed_count.* += bytes_read; |
| 969 | | return Literals{ |
| 980 | return LiteralsSection{ |
| 970 | 981 | .header = header, |
| 971 | 982 | .huffman_tree = huffman_tree, |
| 972 | 983 | .streams = .{ .one = stream }, |
| ... | ... | @@ -988,7 +999,7 @@ pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals |
| 988 | 999 | |
| 989 | 1000 | consumed_count.* += total_streams_size + bytes_read; |
| 990 | 1001 | |
| 991 | | return Literals{ |
| 1002 | return LiteralsSection{ |
| 992 | 1003 | .header = header, |
| 993 | 1004 | .huffman_tree = huffman_tree, |
| 994 | 1005 | .streams = .{ .four = .{ |
| ... | ... | @@ -1002,7 +1013,7 @@ pub fn decodeLiteralsSection(src: []const u8, consumed_count: *usize) !Literals |
| 1002 | 1013 | } |
| 1003 | 1014 | } |
| 1004 | 1015 | |
| 1005 | | fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !Literals.HuffmanTree { |
| 1016 | fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !LiteralsSection.HuffmanTree { |
| 1006 | 1017 | var bytes_read: usize = 0; |
| 1007 | 1018 | bytes_read += 1; |
| 1008 | 1019 | const header = src[0]; |
| ... | ... | @@ -1094,7 +1105,7 @@ fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !Literals.HuffmanT |
| 1094 | 1105 | weights[symbol_count - 1] = @intCast(u4, std.math.log2_int(u16, next_power_of_two - weight_power_sum) + 1); |
| 1095 | 1106 | log.debug("weights[{d}] = {d}", .{ symbol_count - 1, weights[symbol_count - 1] }); |
| 1096 | 1107 | |
| 1097 | | var weight_sorted_prefixed_symbols: [256]Literals.HuffmanTree.PrefixedSymbol = undefined; |
| 1108 | var weight_sorted_prefixed_symbols: [256]LiteralsSection.HuffmanTree.PrefixedSymbol = undefined; |
| 1098 | 1109 | for (weight_sorted_prefixed_symbols[0..symbol_count]) |_, i| { |
| 1099 | 1110 | weight_sorted_prefixed_symbols[i] = .{ |
| 1100 | 1111 | .symbol = @intCast(u8, i), |
| ... | ... | @@ -1104,7 +1115,7 @@ fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !Literals.HuffmanT |
| 1104 | 1115 | } |
| 1105 | 1116 | |
| 1106 | 1117 | std.sort.sort( |
| 1107 | | Literals.HuffmanTree.PrefixedSymbol, |
| 1118 | LiteralsSection.HuffmanTree.PrefixedSymbol, |
| 1108 | 1119 | weight_sorted_prefixed_symbols[0..symbol_count], |
| 1109 | 1120 | weights, |
| 1110 | 1121 | lessThanByWeight, |
| ... | ... | @@ -1137,7 +1148,7 @@ fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !Literals.HuffmanT |
| 1137 | 1148 | } |
| 1138 | 1149 | } |
| 1139 | 1150 | consumed_count.* += bytes_read; |
| 1140 | | const tree = Literals.HuffmanTree{ |
| 1151 | const tree = LiteralsSection.HuffmanTree{ |
| 1141 | 1152 | .max_bit_count = max_number_of_bits, |
| 1142 | 1153 | .symbol_count_minus_one = @intCast(u8, prefixed_symbol_count - 1), |
| 1143 | 1154 | .nodes = weight_sorted_prefixed_symbols, |
| ... | ... | @@ -1148,8 +1159,8 @@ fn decodeHuffmanTree(src: []const u8, consumed_count: *usize) !Literals.HuffmanT |
| 1148 | 1159 | |
| 1149 | 1160 | fn lessThanByWeight( |
| 1150 | 1161 | weights: [256]u4, |
| 1151 | | lhs: Literals.HuffmanTree.PrefixedSymbol, |
| 1152 | | rhs: Literals.HuffmanTree.PrefixedSymbol, |
| 1162 | lhs: LiteralsSection.HuffmanTree.PrefixedSymbol, |
| 1163 | rhs: LiteralsSection.HuffmanTree.PrefixedSymbol, |
| 1153 | 1164 | ) bool { |
| 1154 | 1165 | // NOTE: this function relies on the use of a stable sorting algorithm, |
| 1155 | 1166 | // otherwise a special case of if (weights[lhs] == weights[rhs]) return lhs < rhs; |
| ... | ... | @@ -1157,11 +1168,11 @@ fn lessThanByWeight( |
| 1157 | 1168 | return weights[lhs.symbol] < weights[rhs.symbol]; |
| 1158 | 1169 | } |
| 1159 | 1170 | |
| 1160 | | pub fn decodeLiteralsHeader(src: []const u8, consumed_count: *usize) Literals.Header { |
| 1171 | pub fn decodeLiteralsHeader(src: []const u8, consumed_count: *usize) LiteralsSection.Header { |
| 1161 | 1172 | // TODO: we probably want to enable safety for release-fast and release-small (or insert custom checks) |
| 1162 | 1173 | const start = consumed_count.*; |
| 1163 | 1174 | const byte0 = src[0]; |
| 1164 | | const block_type = @intToEnum(Literals.BlockType, byte0 & 0b11); |
| 1175 | const block_type = @intToEnum(LiteralsSection.BlockType, byte0 & 0b11); |
| 1165 | 1176 | const size_format = @intCast(u2, (byte0 & 0b1100) >> 2); |
| 1166 | 1177 | var regenerated_size: u20 = undefined; |
| 1167 | 1178 | var compressed_size: ?u18 = null; |
| ... | ... | @@ -1220,7 +1231,7 @@ pub fn decodeLiteralsHeader(src: []const u8, consumed_count: *usize) Literals.He |
| 1220 | 1231 | compressed_size, |
| 1221 | 1232 | }, |
| 1222 | 1233 | ); |
| 1223 | | return Literals.Header{ |
| 1234 | return LiteralsSection.Header{ |
| 1224 | 1235 | .block_type = block_type, |
| 1225 | 1236 | .size_format = size_format, |
| 1226 | 1237 | .regenerated_size = regenerated_size, |
| ... | ... | @@ -1228,7 +1239,7 @@ pub fn decodeLiteralsHeader(src: []const u8, consumed_count: *usize) Literals.He |
| 1228 | 1239 | }; |
| 1229 | 1240 | } |
| 1230 | 1241 | |
| 1231 | | fn decodeSequencesHeader(src: []const u8, consumed_count: *usize) !Sequences.Header { |
| 1242 | pub fn decodeSequencesHeader(src: []const u8, consumed_count: *usize) !SequencesSection.Header { |
| 1232 | 1243 | var sequence_count: u24 = undefined; |
| 1233 | 1244 | |
| 1234 | 1245 | var bytes_read: usize = 0; |
| ... | ... | @@ -1237,7 +1248,7 @@ fn decodeSequencesHeader(src: []const u8, consumed_count: *usize) !Sequences.Hea |
| 1237 | 1248 | bytes_read += 1; |
| 1238 | 1249 | log.debug("decoded sequences header '{}': sequence count = 0", .{std.fmt.fmtSliceHexUpper(src[0..bytes_read])}); |
| 1239 | 1250 | consumed_count.* += bytes_read; |
| 1240 | | return Sequences.Header{ |
| 1251 | return SequencesSection.Header{ |
| 1241 | 1252 | .sequence_count = 0, |
| 1242 | 1253 | .offsets = undefined, |
| 1243 | 1254 | .match_lengths = undefined, |
| ... | ... | @@ -1258,9 +1269,9 @@ fn decodeSequencesHeader(src: []const u8, consumed_count: *usize) !Sequences.Hea |
| 1258 | 1269 | bytes_read += 1; |
| 1259 | 1270 | |
| 1260 | 1271 | consumed_count.* += bytes_read; |
| 1261 | | const matches_mode = @intToEnum(Sequences.Header.Mode, (compression_modes & 0b00001100) >> 2); |
| 1262 | | const offsets_mode = @intToEnum(Sequences.Header.Mode, (compression_modes & 0b00110000) >> 4); |
| 1263 | | const literal_mode = @intToEnum(Sequences.Header.Mode, (compression_modes & 0b11000000) >> 6); |
| 1272 | const matches_mode = @intToEnum(SequencesSection.Header.Mode, (compression_modes & 0b00001100) >> 2); |
| 1273 | const offsets_mode = @intToEnum(SequencesSection.Header.Mode, (compression_modes & 0b00110000) >> 4); |
| 1274 | const literal_mode = @intToEnum(SequencesSection.Header.Mode, (compression_modes & 0b11000000) >> 6); |
| 1264 | 1275 | log.debug("decoded sequences header '{}': (sc={d},o={s},m={s},l={s})", .{ |
| 1265 | 1276 | std.fmt.fmtSliceHexUpper(src[0..bytes_read]), |
| 1266 | 1277 | sequence_count, |
| ... | ... | @@ -1270,7 +1281,7 @@ fn decodeSequencesHeader(src: []const u8, consumed_count: *usize) !Sequences.Hea |
| 1270 | 1281 | }); |
| 1271 | 1282 | if (compression_modes & 0b11 != 0) return error.ReservedBitSet; |
| 1272 | 1283 | |
| 1273 | | return Sequences.Header{ |
| 1284 | return SequencesSection.Header{ |
| 1274 | 1285 | .sequence_count = sequence_count, |
| 1275 | 1286 | .offsets = offsets_mode, |
| 1276 | 1287 | .match_lengths = matches_mode, |
| ... | ... | @@ -1428,25 +1439,25 @@ const ReversedByteReader = struct { |
| 1428 | 1439 | } |
| 1429 | 1440 | }; |
| 1430 | 1441 | |
| 1431 | | const ReverseBitReader = struct { |
| 1442 | pub const ReverseBitReader = struct { |
| 1432 | 1443 | byte_reader: ReversedByteReader, |
| 1433 | 1444 | bit_reader: std.io.BitReader(.Big, ReversedByteReader.Reader), |
| 1434 | 1445 | |
| 1435 | | fn init(self: *ReverseBitReader, bytes: []const u8) !void { |
| 1446 | pub fn init(self: *ReverseBitReader, bytes: []const u8) !void { |
| 1436 | 1447 | self.byte_reader = ReversedByteReader.init(bytes); |
| 1437 | 1448 | self.bit_reader = std.io.bitReader(.Big, self.byte_reader.reader()); |
| 1438 | 1449 | while (0 == self.readBitsNoEof(u1, 1) catch return error.BitStreamHasNoStartBit) {} |
| 1439 | 1450 | } |
| 1440 | 1451 | |
| 1441 | | fn readBitsNoEof(self: *@This(), comptime U: type, num_bits: usize) !U { |
| 1452 | pub fn readBitsNoEof(self: *@This(), comptime U: type, num_bits: usize) !U { |
| 1442 | 1453 | return self.bit_reader.readBitsNoEof(U, num_bits); |
| 1443 | 1454 | } |
| 1444 | 1455 | |
| 1445 | | fn readBits(self: *@This(), comptime U: type, num_bits: usize, out_bits: *usize) !U { |
| 1456 | pub fn readBits(self: *@This(), comptime U: type, num_bits: usize, out_bits: *usize) !U { |
| 1446 | 1457 | return try self.bit_reader.readBits(U, num_bits, out_bits); |
| 1447 | 1458 | } |
| 1448 | 1459 | |
| 1449 | | fn alignToByte(self: *@This()) void { |
| 1460 | pub fn alignToByte(self: *@This()) void { |
| 1450 | 1461 | self.bit_reader.alignToByte(); |
| 1451 | 1462 | } |
| 1452 | 1463 | }; |
| ... | ... | @@ -1514,7 +1525,7 @@ fn dumpFseTable(prefix: []const u8, table: []const Table.Fse) void { |
| 1514 | 1525 | } |
| 1515 | 1526 | } |
| 1516 | 1527 | |
| 1517 | | fn dumpHuffmanTree(tree: Literals.HuffmanTree) void { |
| 1528 | fn dumpHuffmanTree(tree: LiteralsSection.HuffmanTree) void { |
| 1518 | 1529 | log.debug("Huffman tree: max bit count = {}, symbol count = {}", .{ tree.max_bit_count, tree.symbol_count_minus_one + 1 }); |
| 1519 | 1530 | for (tree.nodes[0 .. tree.symbol_count_minus_one + 1]) |node| { |
| 1520 | 1531 | log.debug("symbol = {[symbol]d}, prefix = {[prefix]d}, weight = {[weight]d}", node); |
Dominic <4678790+dweiller@users.noreply.github.com> 2023-01-23 16:26:03+11:00