qoi consistent style pass#356
Conversation
| # allocate a buffer 3/4 the size of the pathological encoding | ||
|
|
||
| result.add(qoiSignature) | ||
| when cpuEndian == bigEndian: |
There was a problem hiding this comment.
no tests, untested. we do not support big endian anywhere else and have not / cannot? ensure this actually is all that is needed. rm.
| inc run | ||
| if run == 62 or off == qoi.data.high: | ||
| result.addUint8(opRun or pred(run)) | ||
| reset run |
There was a problem hiding this comment.
had to look this up, not worth it
| if px == pxPrev: | ||
| inc run | ||
| if run == 62 or off == qoi.data.high: | ||
| result.addUint8(opRun or pred(run)) |
There was a problem hiding this comment.
had to look this up too, not worth it
|
|
||
| result = Qoi( | ||
| data: newSeq[ColorRGBA](int width * height), | ||
| width: int width, |
There was a problem hiding this comment.
we by convention append our type conversions to the end
| copyMem(result.data[0].addr, qoi.data[0].addr, qoi.data.len * 4) | ||
| result.data.toPremultipliedAlpha() | ||
|
|
||
| func toQoi*(img: Image; channels: range[3..4]): Qoi = |
There was a problem hiding this comment.
be consistent with png's raw api, do not take in channels without tests (this is not run in any tests)
| var | ||
| width, height: uint32 | ||
| channels, colorspace: uint8 | ||
| block: |
There was a problem hiding this comment.
see endian comment below
| inc p | ||
| of opRun: | ||
| run = b0 and 0x3f | ||
| else: assert false |
There was a problem hiding this comment.
this seems bad, throw for real?
| px.r = data.readUint8(p+0) | ||
| px.g = data.readUint8(p+1) | ||
| px.b = data.readUint8(p+2) | ||
| inc(p, 3) |
There was a problem hiding this comment.
out of convension we only use inc and dec for +1 and -1, otherwise just use +=
| ## Decodes data in the QOI file format to an `Image`. | ||
| decompressQoi(data).toImage() | ||
|
|
||
| proc decodeQoi*(data: seq[uint8]): Image {.inline, raises: [PixieError].} = |
There was a problem hiding this comment.
gave up on seq[uint8] a long time ago. nim's native representation for bytes is string, see standard lib readFile
| data[pos] = value | ||
| value = rand(255).uint8 | ||
| data[pos] = value.char | ||
| echo &"{i} {pos} {value}" |
There was a problem hiding this comment.
important to echo, so if fuzzing crashes we can reproduce the case immediately
| @@ -1,5 +1,4 @@ | |||
| import std/endians, chroma, flatty/binny | |||
There was a problem hiding this comment.
morepretty does not understand [] imports so we do not use them (support could be added tho, not opposed to them if it is automatic)
2 participants
Uh oh!
There was an error while loading. Please reload this page.