Functions and utils related to VapourSynth
License: MIT License
This module is a collection of functions, utils, types, type-utils, and more aimed at helping at having a common ground between VapourSynth packages, and simplify writing them.
For support you can check out the JET Discord server.
Install vstools with the following command:
pip install vstoolsOr if you want the latest git version, install it with this command:
pip install git+https://github.com/Jaded-Encoding-Thaumaturgy/vs-tools.git![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
When using Timecodes.assume_vfr, it returns the following error:
ValueError: (replace_ranges) RFS: Clip frame rates don't match, enable mismatch if you want variable format.Setting it to mismatch=True seems to still not display it as VFR in vspreview.
replace_ranges also throws a warning:
C:\Program Files\Python311\Lib\site-packages\vstools\utils\ranges.py:168: UserWarning: replace_ranges: use_plugin is deprecated!
warnings.warn('replace_ranges: use_plugin is deprecated!')Talked about this in PMs. There are ways to automatically determine the caller function, so it might be a good idea to implement that so people don't need to worry about it (but they're still able to override it should they want to).
Below is example code from a private module I'm currently working on, which might be reusable here (and can likely be improved upon further, like returning the ClassName.method if the caller originates from a class?):
import inspect
def get_caller_function() -> str | None:
"""Attempt to call the caller function. This is used for exception handling."""
for f in inspect.stack()[1:]:
func = f.function
func_origin = Path(f.filename).name
# Check if private function/method
if func.startswith("_"):
continue
# Check if it's an exception or error class.
if any(x in func for x in ["Exception", "Error"]):
continue
# Check if it originates from an `exceptions.py` or `errors.py` file.
if any(x in func_origin.lower() for x in ["exceptions.py", "error.py"]):
continue
# Finally, return the function name.
return funcAnd a basic example of how I implemented it in my module:
from vstools import CustomError, SupportsString, FuncExceptT
from typing import Any
class BaseCustomError(CustomError):
def __init__(self,
message: SupportsString | None = None,
reason: SupportsString | None = None,
func: FuncExceptT | None = None,
**kwargs: Any
) -> None:
func = func or get_caller_function()
if message:
message = str(message)
return super().__init__(message, func, reason, **kwargs)As mentioned, I'm sure this can be improved upon a lot, but hopefully this serves as a nice jumping-off point.
Just use heuristics
It'd be nice if a directory would be automatically created if it didn't exist.
2023-02-17 18:51:33.464: root: ERROR: [Errno 2] No such file or directory: 'E:\\Anime\\Fansubbing\\Git Repositories\\Encoding Projects\\[Kaleido-subs]\\Work in Progress\\Negima! Series [DVD]_hidden\\S2\\_assets\\NegimaS2DVD_01_scening.txt'
2023-02-17 18:51:33.468: root: ERROR: Traceback (most recent call last):
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vsengine\vpy.py", line 219, in _execute
runpy.run_path(str(script), module.__dict__, module.__name__)
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 291, in run_path
return _run_module_code(code, init_globals, run_name,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 98, in _run_module_code
_run_code(code, mod_globals, init_globals,
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vspreview\core\vsenv.py", line 19, in _monkey_runpy_func
glob_dict = orig_runpy_run_code(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 88, in _run_code
exec(code, run_globals)
File "E:\Anime\Fansubbing\Git Repositories\Encoding Projects\[Kaleido-subs]\Work in Progress\Negima! Series [DVD]_hidden\S2\NegimaS2DVD_01.py", line 323, in <module>
PREFILTER = prefilter(SRC.clip_cut)
^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Anime\Fansubbing\Git Repositories\Encoding Projects\[Kaleido-subs]\Work in Progress\Negima! Series [DVD]_hidden\S2\NegimaS2DVD_01.py", line 46, in prefilter
Keyframes.from_clip(BD.clip_cut).to_file(scening_path)
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\functions\timecodes.py", line 317, in to_file
out_path.touch()
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\pathlib.py", line 1108, in touch
fd = os.open(self, flags, mode)
^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: 'E:\\Anime\\Fansubbing\\Git Repositories\\Encoding Projects\\[Kaleido-subs]\\Work in Progress\\Negima! Series [DVD]_hidden\\S2\\_assets\\NegimaS2DVD_01_scening.txt'QP files require more than just a frame number to function. One of the biggest uses of keyframe functions in VS is to generate qpfiles (either for timing (with Aegisub) or for encoding (passing it to the encoder)). To do this, the ability to append a string after the keyframe number would be neat. See x265 settings/x264 settings.
Behavior relating to scale offsets and limited/full conversion appears to be incorrect.
The current exception thrown when a "clip is too long" isn't always really helpful. It'd be neat if it told the user by how many frames it was too long. This way the user could easily see whether they slightly miscalculated the trim, or if they accidentally passed the wrong clip because the length is obviously wrong.
Additionally, maybe a parameter to allow the user to pass a clip and have it simply trim it by the end if it exceeds the processed clip's length might be nice.
It seems this is traced back to this import line:
vs-tools/vstools/utils/vs_proxy.py
Line 17 in 809df66
R68 seems to no longer be exporting the following items, which is causing this issue:
ccfDisableAutoLoading, ccfDisableLibraryUnloading, ccfEnableGraphInspection, fmFrameState, fmParallel, fmParallelRequests, fmUnordered
Explained in VC, setsu skill issue
(also, typo: "lenghts" -> "lengths")
2023-02-15 21:59:52.443: root: ERROR: (diff) All the lenghts must be equal! ()
2023-02-15 21:59:52.450: root: ERROR: Traceback (most recent call last):
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vsengine\vpy.py", line 219, in _execute
runpy.run_path(str(script), module.__dict__, module.__name__)
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 291, in run_path
return _run_module_code(code, init_globals, run_name,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 98, in _run_module_code
_run_code(code, mod_globals, init_globals,
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vspreview\core\vsenv.py", line 19, in _monkey_runpy_func
glob_dict = orig_runpy_run_code(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\runpy.py", line 88, in _run_code
exec(code, run_globals)
File "E:\Anime\Fansubbing\Git Repositories\Encoding Projects\[Kaleido-subs]\Work in Progress\Joshiraku [BD]_new_hidden\JoshirakuBD_07.py", line 75, in <module>
FILTERED = filterchain(SRC.clip_cut, NCOP.clip_cut, NCED.clip_cut, OP_03.clip_cut)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\Anime\Fansubbing\Git Repositories\Encoding Projects\[Kaleido-subs]\Work in Progress\Joshiraku [BD]_new_hidden\JoshirakuBD_07.py", line 44, in filterchain
lvf.diff(o_u, p_u, thr=80).std.SetFrameProps(Name="diff on U plane")
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\lvsfunc\comparison.py", line 625, in diff
LengthMismatchError.check(diff, a.num_frames, b.num_frames)
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\exceptions\generic.py", line 182, in check
raise cls(func, items, **kwargs)
LengthMismatchError: (diff) All the lenghts must be equal! ()Relevant conversation on Discord
Simply put, when trying to get the Matrix from an HDR source, it attempts to get Matrix(16). Matrix doesn't have a member with the value of 16, so this causes an error. This problem may also persist with Primaries.
2023-11-15 17:18:39.522: root: ERROR: ValueError: 16 is not a valid Matrix
...
File "C:\Users\oliwis\AppData\Local\Programs\Python\Python311\Lib\site-packages\vskernels\kernels\complex.py", line 60, in func
with LinearLight(clip, linear, sigmoid, self, kwargs.pop('format', None)) as ll:
File "C:\Users\oliwis\AppData\Local\Programs\Python\Python311\Lib\site-packages\vskernels\util.py", line 226, in __enter__
self._matrix = Matrix.from_transfer(self._curve)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Related to Jaded-Encoding-Thaumaturgy/vs-dehalo#20
Currently padder only pads with mirror and repeat padding, ideally it should support all modes fillborders provides and additionally act as an std.AddBorders wrapper allowing users to create padding with a specific color.
The check should not be based on res at all, it should be dependant on the colorspace of the input. Top left location is defined by Rec. BT.2020 and as such should only be set to top left if the video is BT.2020, otherwise left should be assumed.
2023-02-15 20:15:50.899: root: ERROR: multiple bases have instance lay-out conflict
2023-02-15 20:15:50.918: root: ERROR: Traceback (most recent call last):
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\lvsfunc\comparison.py", line 651, in diff
raise CustomError[StopIteration]('No differences found!', diff)
~~~~~~~~~~~^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\exceptions\base.py", line 107, in __class_getitem__
class inner_exception(cls, exception): # type: ignore
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\exceptions\base.py", line 34, in __new__
return CustomErrorMeta.setup_exception(type.__new__(cls, *args)) # type: ignore
^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: multiple bases have instance lay-out conflictA handful of method names are inconsistent throughout this package (and other iew packages),
and I cannot really determine any logic behind some of them.
This is the one I thought of right now, but there are more.
If I remember or find more, I will add them in a comment (and reopen if necessary).
If any of the listed ones do have some logic behind them, simply mark those specific instances as wontfix.
From norm_func_name in normalize.py:
if callable(func):
if hasattr(func, '__self__'):
func_name = f'{func.__self__.__name__}.{func_name}'This code path is taken even if func.__name__ or func.__qualname__ exist, and would override those values. I can't really tell if this is intended, but it seems reasonable. However, if called on an instance method, __self__ refers to the instance, which doesn't have a __name__:
>>> from vstools import norm_func_name as norm
>>> class A():
... def foo(self):
... pass
...
>>> norm(A.foo)
'foo'
>>> a = A()
>>> norm(a.foo)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File ".../site-packages/vstools/functions/normalize.py", line 285, in norm_func_name
func_name = f'{func.__self__.__name__}.{func_name}'
^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'A' object has no attribute '__name__'. Did you mean: '__ne__'?The correct behavior would be to access the instance's __class__ property instead:
>>> a.foo.__self__ # instance
<__main__.A object at 0x0000015B8922BA10>
>>> a.foo.__self__.__class__ # class
<class '__main__.A'>
>>> a.foo.__self__.__class__.__name__ # desired name property
'A'Note also that this only happens when referencing the function via an instance of the class (though this is probably much more common than referencing the class's function directly)
I think it would be best to first check if func.__self__ has a __name__ attribute, then if not, check if it has a __class__ attribute, and finally use that one's __name__. (Note that, for a class, A.__class__ is type, so both of these checks would be necessary)
File "C:\Users\Reza\AppData\Roaming\Python\Python311\site-packages\vstools\utils\clips.py", line 68, in finalize_clip
clip = clip.akarin.Expr([
^^^^^^^^^^^^^^^^^^
File "src\\cython\\vapoursynth.pyx", line 2860, in vapoursynth.Function.__call__
vapoursynth.Error: Expr: More expressions given than there are planes
Test the whole function, should give better errors.
When importing lvsfunc or vsaa i get error
AttributeError: 'NoneType' object has no attribute 'isatty'
The full traceback
Traceback (most recent call last):
File "src\cython\vapoursynth.pyx", line 2890, in vapoursynth._vpy_evaluate
File "src\cython\vapoursynth.pyx", line 2891, in vapoursynth._vpy_evaluate
File "F:\test.vpy", line 6, in import lvsfunc
File "C:\Program Files\Python310\lib\site-packages\lvsfunc\__init__.py", line 12, in
from . import (aa, comparison, deblock, dehalo, dehardsub, deinterlace, exceptions, fun, helpers, mask, misc, noise,
File "C:\Program Files\Python310\lib\site-packages\lvsfunc\aa.py", line 7, in
from vskernels import Bicubic, Box, Catrom, Point
File "C:\Program Files\Python310\lib\site-packages\vskernels\__init__.py", line 15, in
from . import exceptions, kernels, util
File "C:\Program Files\Python310\lib\site-packages\vskernels\exceptions.py", line 5, in
from vstools import CustomValueError, FuncExceptT
File "C:\Program Files\Python310\lib\site-packages\vstools\__init__.py", line 1, in
from .enums import * # noqa: F401, F403
File "C:\Program Files\Python310\lib\site-packages\vstools\enums\__init__.py", line 1, in
from .base import * # noqa: F401, F403
File "C:\Program Files\Python310\lib\site-packages\vstools\enums\base.py", line 6, in
from ..exceptions import NotFoundEnumValue
File "C:\Program Files\Python310\lib\site-packages\vstools\exceptions\__init__.py", line 1, in
from .base import * # noqa: F401, F403
File "C:\Program Files\Python310\lib\site-packages\vstools\exceptions\base.py", line 40, in
class CustomError(Exception, metaclass=CustomErrorMeta):
File "C:\Program Files\Python310\lib\site-packages\vstools\exceptions\base.py", line 32, in __new__
if sys.stdout.isatty():
AttributeError: 'NoneType' object has no attribute 'isatty'
I've updated lvsfunc, vsaa and vstools from git passing the --upgrade flag to pip.
When pushing a list of range through normalize_ranges, the final frame gets treated as if it's exclusive in a sense.
[(100, 200)]->
[(100, 199)]This causes scenefiltering to be off by one frame.
Hello !
I see requirements.txt says
rich>=12.6.0
And in the docs folder it says
rich==13.0.0
I'm not quire sure how to interpret it, is 13.0.0 for the documentation or something ?
Cheers
It would be nice if Exceptions like for example FormatsRefClipMismatchError would also display the formats of both clips in the reason by default. This makes it much easier to see which clip needs adjusting.
It can't find a mime for .log files, and so runs into an AssertionError.
My xvid keyframe files are in this format:
# XviD 2pass stat file (core version 1.4.-127)
# Please do not modify this file
i 2 920 0 0 2710 2710
p 2 0 0 920 232 232
p 2 0 0 920 232 232
p 2 0 0 920 232 232
p 2 0 0 920 232 232
p 2 0 0 920 232 232
I am not sure whether this is the standard format or not, but it's what my SCXvid.exe spits out.
vstools appears to attempt to parse this sort of file here: https://github.com/Jaded-Encoding-Thaumaturgy/vs-tools/blob/master/vstools/functions/timecodes.py#L458
This fails, but if I change lines[0].lower() to lines[0].lower()[0] and then also modify https://github.com/Jaded-Encoding-Thaumaturgy/vs-tools/blob/master/vstools/functions/timecodes.py#L465 in the same way, it appears to work correctly.
Full keyframe file: attackTwo01_keyframes.txt
With certain files, the wrong filetype appears to be getting parsed.
For example, this m2ts file prases as an AC3 file.
ParsedFile(path=WindowsPath('C:/Users/light/Encoding/temp/maria/00002.m2ts'), ext='.ac3', encoding=None, file_type=<FileType.AUDIO: 'audio'>, mime='application/octet-stream')ffprobe's results:
Input #0, mpegts, from '00002.m2ts':
Duration: 00:25:27.87, start: 5614.566667, bitrate: 42425 kb/s
Program 1
Stream #0:0[0x1011]: Video: h264 (High) (HDMV / 0x564D4448), yuv420p(tv, bt709, top first), 1920x1080 [SAR 1:1 DAR 16:9], 29.97 fps, 29.97 tbr, 90k tbn
Stream #0:1[0x1100]: Audio: pcm_bluray (HDMV / 0x564D4448), 48000 Hz, stereo, s16, 1536 kb/s
Stream #0:2[0x1101]: Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 448 kb/s
The first couple bytes
file_bytes=b'\x0bw\\\xccG@\x00\x10\x00\x00\xb0\x11\x00\x00\xc1\x00\x00\x00'This file gets parsed correctly upon any form of remuxing, even when trying to generate a test clip using the following command:
ffmpeg -to 0:03.00 -i "00002.m2ts" -c copy test.m2tsHaving additional heuristics to determine a file type would be appreciated, but I'm not sure how much can be done here.
One kind of hacky solution I can think of for vssource.source is to check if a video is parsed as anything but VIDEO/IMAGE, and then try simpler heuristics and see if those return a video or image.
This is really an issue with replace_ranges, but I could trace it as far back as invert_ranges I believe.
From what I can gather, it does not properly return an inversion of the ranges, so replace_ranges' a_trims becomes a single clip the length of the original clip as opposed to the two necessary. The b_trims get appended at the end instead of sandwiched between them as it should.
out = replace_ranges(clip, clip.std.BlankClip(keep=True), [(1000, 9000)])
print("og clip", clip.num_frames)
print("out clip", out.num_frames)og clip 34765
out clip 42766
Clips passed to replace_ranges:
VideoNode
Format: YUV420P16
Width: 1920
Height: 1080
Num Frames: 34765
FPS: 24000/1001
VideoNode
Format: YUV420P16
Width: 1920
Height: 1080
Num Frames: 34765
FPS: 24000/1001
a and b_ranges
[(0, 34764)]
[(1000, 9000)]
It would be nice if we could get additional methods to obtain Keyframes and Timecodes objects.
.from_listThe first suggestion would be for Keyframes: .from_list.
This would simply call __init__ and pass the iterator.
The purpose of this is more so just to have a convenience function for people looking for it who aren't aware you can just pass a list directly to the class.
.from_wobblyThe next one is a convenience method for wobbly usage and to speed up parsing, for both Keyframes and Timecodes.
This would go together with eventually(?) wobbly/wibbly updates in vsdeinterlace.
This should simply read the .wob file and construct the relevant object (from which the user can then write it to a file).
Currently DPIR and Waifu2x both have their own implementations of mod_padding (with DPIR's giving incorrect results). Instead of duplicating code just implement the functionality into padder.
foo.py
from vstools import vs
import vapoursynth
resizer = vs.core.resize.Bicubic
def get_resizer():
print('VS core:', vapoursynth.core)
print('vstools core:', vs.core)
return resizerUntitled.vpy
from foo import *
clip = vs.core.std.BlankClip()
clip = get_resizer()(clip)
clip.set_output()vsedit Untitled.vpy with F6 script checking (where VS core is acquired from the newly created script, and pointers are printed to stdout):
VS core: <vapoursynth._CoreProxy object at 0x7f0cf0188bb0>
vstools core: <vstools.utils.vs_proxy.VSCoreProxy object at 0x7f0ceaa61120>
Current core: 0x55f2b65b1f60
Current script: 0x55f2b65cd130
VS core: <vapoursynth._CoreProxy object at 0x7f0cf0188bb0>
vstools core: <vstools.utils.vs_proxy.VSCoreProxy object at 0x7f0ceaa61120>
Current core: 0x55f2b5be6ca0
Current script: 0x7f0cec007060
This error pops up when using normalize_ranges (such as in replace_ranges) with specific input. The error originates from normalize_list_to_ranges.
Input:
[45, 46, 47, 48, 49, 50, 51, 210, 211, 212, 213, 214, 215, 216, 820, 824, 825, 829, 830, 834, 835, 839, 840, 1584, 1585, 1586, 1587, 1596, 1597, 1598, 1599, 1600, 1609, 1610, 1611, 1612, 1613, 1623, 1624, 1625, 1626, 1636, 1637, 1638, 1639, 1651, 1652, 1653, 1654, 1677, 1678, 1679, 1680, 1681, 1682, 1683, 1713, 1714, 1715, 1717, 1723, 2354, 2412]normalize_ranges output that gets passed on to normalize_list_to_ranges:
[(45, 45), (46, 46), (47, 47), (48, 48), (49, 49), (50, 50), (51, 51), (210, 210), (211, 211), (212, 212), (213, 213), (214, 214), (215, 215), (216, 216), (820, 820), (824, 824), (825, 825), (829, 829), (830, 830), (834, 834), (835, 835), (839, 839), (840, 840), (1584, 1584), (1585, 1585), (1586, 1586), (1587, 1587), (1596, 1596), (1597, 1597), (1598, 1598), (1599, 1599), (1600, 1600), (1609, 1609), (1610, 1610), (1611, 1611), (1612, 1612), (1613, 1613), (1623, 1623), (1624, 1624), (1625, 1625), (1626, 1626), (1636, 1636), (1637, 1637), (1638, 1638), (1639, 1639), (1651, 1651), (1652, 1652), (1653, 1653), (1654, 1654), (1677, 1677), (1678, 1678), (1679, 1679), (1680, 1680), (1681, 1681), (1682, 1682), (1683, 1683), (1713, 1713), (1714, 1714), (1715, 1715), (1717, 1717), (1723, 1723), (2354, 2354), (2412, 2412)]Error in normalize_list_to_ranges:
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\utils\ranges.py", line 82, in replace_ranges
nranges = normalize_ranges(clip_b, ranges)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\functions\normalize.py", line 251, in normalize_ranges
return normalize_list_to_ranges([
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\functions\normalize.py", line 192, in normalize_list_to_ranges
[i[-1 if j == last_idx else -2] for j, i in enumerate(flist4)]
File "C:\Users\light\AppData\Local\Programs\Python\Python311\Lib\site-packages\vstools\functions\normalize.py", line 192, in <listcomp>
[i[-1 if j == last_idx else -2] for j, i in enumerate(flist4)]
~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
IndexError: list index out of range
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.