| 1 | //! Parser for transfer-encoding: chunked. |
| 2 | |
| 3 | const ChunkParser = @This(); |
| 4 | const std = @import("std"); |
| 5 | |
| 6 | state: State, |
| 7 | chunk_len: u64, |
| 8 | |
| 9 | pub const init: ChunkParser = .{ |
| 10 | .state = .head_size, |
| 11 | .chunk_len = 0, |
| 12 | }; |
| 13 | |
| 14 | pub const State = enum { |
| 15 | head_size, |
| 16 | head_ext, |
| 17 | head_r, |
| 18 | data, |
| 19 | data_suffix, |
| 20 | data_suffix_r, |
| 21 | invalid, |
| 22 | }; |
| 23 | |
| 24 | /// Returns the number of bytes consumed by the chunk size. This is always |
| 25 | /// less than or equal to `bytes.len`. |
| 26 | /// |
| 27 | /// After this function returns, `chunk_len` will contain the parsed chunk size |
| 28 | /// in bytes when `state` is `data`. Alternately, `state` may become `invalid`, |
| 29 | /// indicating a syntax error in the input stream. |
| 30 | /// |
| 31 | /// If the amount returned is less than `bytes.len`, the parser is in the |
| 32 | /// `chunk_data` state and the first byte of the chunk is at `bytes[result]`. |
| 33 | /// |
| 34 | /// Asserts `state` is neither `data` nor `invalid`. |
| 35 | pub fn feed(p: *ChunkParser, bytes: []const u8) usize { |
| 36 | for (bytes, 0..) |c, i| switch (p.state) { |
| 37 | .data_suffix => switch (c) { |
| 38 | '\r' => p.state = .data_suffix_r, |
| 39 | '\n' => p.state = .head_size, |
| 40 | else => { |
| 41 | p.state = .invalid; |
| 42 | return i; |
| 43 | }, |
| 44 | }, |
| 45 | .data_suffix_r => switch (c) { |
| 46 | '\n' => p.state = .head_size, |
| 47 | else => { |
| 48 | p.state = .invalid; |
| 49 | return i; |
| 50 | }, |
| 51 | }, |
| 52 | .head_size => { |
| 53 | const digit = switch (c) { |
| 54 | '0'...'9' => |b| b - '0', |
| 55 | 'A'...'Z' => |b| b - 'A' + 10, |
| 56 | 'a'...'z' => |b| b - 'a' + 10, |
| 57 | '\r' => { |
| 58 | p.state = .head_r; |
| 59 | continue; |
| 60 | }, |
| 61 | '\n' => { |
| 62 | p.state = .data; |
| 63 | return i + 1; |
| 64 | }, |
| 65 | else => { |
| 66 | p.state = .head_ext; |
| 67 | continue; |
| 68 | }, |
| 69 | }; |
| 70 | |
| 71 | const new_len = p.chunk_len *% 16 +% digit; |
| 72 | if (new_len <= p.chunk_len and p.chunk_len != 0) { |
| 73 | p.state = .invalid; |
| 74 | return i; |
| 75 | } |
| 76 | |
| 77 | p.chunk_len = new_len; |
| 78 | }, |
| 79 | .head_ext => switch (c) { |
| 80 | '\r' => p.state = .head_r, |
| 81 | '\n' => { |
| 82 | p.state = .data; |
| 83 | return i + 1; |
| 84 | }, |
| 85 | else => continue, |
| 86 | }, |
| 87 | .head_r => switch (c) { |
| 88 | '\n' => { |
| 89 | p.state = .data; |
| 90 | return i + 1; |
| 91 | }, |
| 92 | else => { |
| 93 | p.state = .invalid; |
| 94 | return i; |
| 95 | }, |
| 96 | }, |
| 97 | .data => unreachable, |
| 98 | .invalid => unreachable, |
| 99 | }; |
| 100 | return bytes.len; |
| 101 | } |
| 102 | |
| 103 | test feed { |
| 104 | const testing = std.testing; |
| 105 | |
| 106 | const data = "Ff\r\nf0f000 ; ext\n0\r\nffffffffffffffffffffffffffffffffffffffff\r\n"; |
| 107 | |
| 108 | var p = init; |
| 109 | const first = p.feed(data[0..]); |
| 110 | try testing.expectEqual(@as(u32, 4), first); |
| 111 | try testing.expectEqual(@as(u64, 0xff), p.chunk_len); |
| 112 | try testing.expectEqual(.data, p.state); |
| 113 | |
| 114 | p = init; |
| 115 | const second = p.feed(data[first..]); |
| 116 | try testing.expectEqual(@as(u32, 13), second); |
| 117 | try testing.expectEqual(@as(u64, 0xf0f000), p.chunk_len); |
| 118 | try testing.expectEqual(.data, p.state); |
| 119 | |
| 120 | p = init; |
| 121 | const third = p.feed(data[first + second ..]); |
| 122 | try testing.expectEqual(@as(u32, 3), third); |
| 123 | try testing.expectEqual(@as(u64, 0), p.chunk_len); |
| 124 | try testing.expectEqual(.data, p.state); |
| 125 | |
| 126 | p = init; |
| 127 | const fourth = p.feed(data[first + second + third ..]); |
| 128 | try testing.expectEqual(@as(u32, 16), fourth); |
| 129 | try testing.expectEqual(@as(u64, 0xffffffffffffffff), p.chunk_len); |
| 130 | try testing.expectEqual(.invalid, p.state); |
| 131 | } |