import logging
import weakref
from collections import deque
from contextlib import contextmanager
from ctypes import c_char_p, c_float, c_int, cast
from typing import (Any, Deque, Dict, Iterable, List, Optional, Sequence, Set, Tuple,
Union)
import pyglet
from pyglet import gl
from pyglet.window import Window
from .buffer import Buffer
from .compute_shader import ComputeShader
from .framebuffer import DefaultFrameBuffer, Framebuffer
from .glsl import ShaderSource
from .program import Program
from .query import Query
from .texture import Texture2D
from .types import BufferDescription
from .vertex_array import Geometry
from ..types import BufferProtocol
LOG = logging.getLogger(__name__)
[docs]class Context:
"""
Represents an OpenGL context. This context belongs to a ``pyglet.Window``
normally accessed through ``window.ctx``.
The Context class contains methods for creating resources,
global states and commonly used enums. All enums also exist
in the ``gl`` module. (``ctx.BLEND`` or ``arcade.gl.BLEND``).
"""
#: The active context
active: Optional["Context"] = None
#: The OpenGL api. Usually "gl" or "gles".
gl_api: str = "gl"
# --- Store the most commonly used OpenGL constants
# Texture
#: Texture interpolation: Nearest pixel
NEAREST = 0x2600
#: Texture interpolation: Linear interpolate
LINEAR = 0x2601
#: Texture interpolation: Minification filter for mipmaps
NEAREST_MIPMAP_NEAREST = 0x2700
#: Texture interpolation: Minification filter for mipmaps
LINEAR_MIPMAP_NEAREST = 0x2701
#: Texture interpolation: Minification filter for mipmaps
NEAREST_MIPMAP_LINEAR = 0x2702
#: Texture interpolation: Minification filter for mipmaps
LINEAR_MIPMAP_LINEAR = 0x2703
#: Texture wrap mode: Repeat
REPEAT = gl.GL_REPEAT
# Texture wrap mode: Clamp to border pixel
CLAMP_TO_EDGE = gl.GL_CLAMP_TO_EDGE
# Texture wrap mode: Clamp to border color
CLAMP_TO_BORDER = gl.GL_CLAMP_TO_BORDER
# Texture wrap mode: Repeat mirrored
MIRRORED_REPEAT = gl.GL_MIRRORED_REPEAT
# Flags
#: Context flag: Blending
BLEND = gl.GL_BLEND
#: Context flag: Depth testing
DEPTH_TEST = gl.GL_DEPTH_TEST
#: Context flag: Face culling
CULL_FACE = gl.GL_CULL_FACE
#: Context flag: Enables ``gl_PointSize`` in vertex or geometry shaders.
#:
#: When enabled we can write to ``gl_PointSize`` in the vertex shader to specify the point size
#: for each individual point.
#:
#: If this value is not set in the shader the behavior is undefined. This means the points may
#: or may not appear depending if the drivers enforce some default value for ``gl_PointSize``.
#:
#: When disabled :py:attr:`Context.point_size` is used.
PROGRAM_POINT_SIZE = gl.GL_PROGRAM_POINT_SIZE
# Blend functions
#: Blend function
ZERO = 0x0000
#: Blend function
ONE = 0x0001
#: Blend function
SRC_COLOR = 0x0300
#: Blend function
ONE_MINUS_SRC_COLOR = 0x0301
#: Blend function
SRC_ALPHA = 0x0302
#: Blend function
ONE_MINUS_SRC_ALPHA = 0x0303
#: Blend function
DST_ALPHA = 0x0304
#: Blend function
ONE_MINUS_DST_ALPHA = 0x0305
#: Blend function
DST_COLOR = 0x0306
#: Blend function
ONE_MINUS_DST_COLOR = 0x0307
# Blend equations
#: source + destination
FUNC_ADD = 0x8006
#: Blend equations: source - destination
FUNC_SUBTRACT = 0x800A
#: Blend equations: destination - source
FUNC_REVERSE_SUBTRACT = 0x800B
#: Blend equations: Minimum of source and destination
MIN = 0x8007
#: Blend equations: Maximum of source and destination
MAX = 0x8008
# Blend mode shortcuts
#: Blend mode shortcut for default blend mode: ``SRC_ALPHA, ONE_MINUS_SRC_ALPHA``
BLEND_DEFAULT = 0x0302, 0x0303
#: Blend mode shortcut for additive blending: ``ONE, ONE``
BLEND_ADDITIVE = 0x0001, 0x0001
#: Blend mode shortcut for pre-multiplied alpha: ``SRC_ALPHA, ONE``
BLEND_PREMULTIPLIED_ALPHA = 0x0302, 0x0001
# VertexArray: Primitives
#: Primitive mode
POINTS = gl.GL_POINTS # 0
#: Primitive mode
LINES = gl.GL_LINES # 1
#: Primitive mode
LINE_LOOP = gl.GL_LINE_LOOP # 2
#: Primitive mode
LINE_STRIP = gl.GL_LINE_STRIP # 3
#: Primitive mode
TRIANGLES = gl.GL_TRIANGLES # 4
#: Primitive mode
TRIANGLE_STRIP = gl.GL_TRIANGLE_STRIP # 5
#: Primitive mode
TRIANGLE_FAN = gl.GL_TRIANGLE_FAN # 6
#: Primitive mode
LINES_ADJACENCY = gl.GL_LINES_ADJACENCY # 10
#: Primitive mode
LINE_STRIP_ADJACENCY = gl.GL_LINE_STRIP_ADJACENCY # 11
#: Primitive mode
TRIANGLES_ADJACENCY = gl.GL_TRIANGLES_ADJACENCY # 12
#: Primitive mode
TRIANGLE_STRIP_ADJACENCY = gl.GL_TRIANGLE_STRIP_ADJACENCY # 13
#: Patch mode (tessellation)
PATCHES = gl.GL_PATCHES
# The most common error enums
_errors = {
gl.GL_INVALID_ENUM: "GL_INVALID_ENUM",
gl.GL_INVALID_VALUE: "GL_INVALID_VALUE",
gl.GL_INVALID_OPERATION: "GL_INVALID_OPERATION",
gl.GL_INVALID_FRAMEBUFFER_OPERATION: "GL_INVALID_FRAMEBUFFER_OPERATION",
gl.GL_OUT_OF_MEMORY: "GL_OUT_OF_MEMORY",
gl.GL_STACK_UNDERFLOW: "GL_STACK_UNDERFLOW",
gl.GL_STACK_OVERFLOW: "GL_STACK_OVERFLOW",
}
_valid_apis = ('gl', 'gles')
def __init__(self, window: pyglet.window.Window, gc_mode: str = "context_gc", gl_api: str = "gl"):
self._window_ref = weakref.ref(window)
if gl_api not in self._valid_apis:
raise ValueError(f"Invalid gl_api. Options are: {self._valid_apis}")
self.gl_api = gl_api
self._limits = Limits(self)
self._gl_version = (self._limits.MAJOR_VERSION, self._limits.MINOR_VERSION)
Context.activate(self)
# Texture unit we use when doing operations on textures to avoid
# affecting currently bound textures in the first units
self.default_texture_unit: int = self._limits.MAX_TEXTURE_IMAGE_UNITS - 1
# Detect the default framebuffer
self._screen = DefaultFrameBuffer(self)
# Tracking active program
self.active_program: Optional[Program] = None
# Tracking active framebuffer. On context creation the window is the default render target
self.active_framebuffer: Framebuffer = self._screen
self._stats: ContextStats = ContextStats(warn_threshold=1000)
# Hardcoded states
# This should always be enabled
# gl.glEnable(gl.GL_TEXTURE_CUBE_MAP_SEAMLESS)
# Set primitive restart index to -1 by default
if self.gl_api == "gles":
gl.glEnable(gl.GL_PRIMITIVE_RESTART_FIXED_INDEX)
else:
gl.glEnable(gl.GL_PRIMITIVE_RESTART)
self._primitive_restart_index = -1
self.primitive_restart_index = self._primitive_restart_index
# Detect support for glProgramUniform.
# Assumed to be supported in gles
self._ext_separate_shader_objects_enabled = True
if self.gl_api == "gl":
have_ext = gl.gl_info.have_extension("GL_ARB_separate_shader_objects")
self._ext_separate_shader_objects_enabled = self.gl_version >= (4, 1) or have_ext
# We enable scissor testing by default.
# This is always set to the same value as the viewport
# to avoid background color affecting areas outside the viewport
gl.glEnable(gl.GL_SCISSOR_TEST)
# States
self._blend_func: Union[Tuple[int, int], Tuple[int, int, int, int]] = self.BLEND_DEFAULT
self._point_size = 1.0
self._flags: Set[int] = set()
self._wireframe = False
# Context GC as default. We need to call Context.gc() to free opengl resources
self._gc_mode = "context_gc"
self.gc_mode = gc_mode
#: Collected objects to gc when gc_mode is "context_gc".
#: This can be used during debugging.
self.objects: Deque[Any] = deque()
@property
def info(self) -> "Limits":
"""
Get the Limits object for this context containing information
about hardware/driver limits and other context information.
Example::
>> ctx.info.MAX_TEXTURE_SIZE
(16384, 16384)
>> ctx.info.VENDOR
NVIDIA Corporation
>> ctx.info.RENDERER
NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2
"""
return self._limits
@property
def limits(self) -> "Limits":
"""
Get the Limits object for this context containing information
about hardware/driver limits and other context information.
.. Warning::
This an old alias for :py:attr:`~arcade.gl.Context.info`
and is only around for backwards compatibility.
Example::
>> ctx.limits.MAX_TEXTURE_SIZE
(16384, 16384)
>> ctx.limits.VENDOR
NVIDIA Corporation
>> ctx.limits.RENDERER
NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2
"""
return self._limits
@property
def stats(self) -> "ContextStats":
"""
Get the stats instance containing runtime information
about creation and destruction of OpenGL objects.
Example::
>> ctx.limits.MAX_TEXTURE_SIZE
(16384, 16384)
>> ctx.limits.VENDOR
NVIDIA Corporation
>> ctx.limits.RENDERER
NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2
"""
return self._stats
@property
def window(self) -> Window:
"""
The window this context belongs to.
:type: ``pyglet.Window``
"""
return self._window_ref()
@property
def screen(self) -> Framebuffer:
"""
The framebuffer for the window.
:type: :py:class:`~arcade.Framebuffer`
"""
return self._screen
@property
def fbo(self) -> Framebuffer:
"""
Get the currently active framebuffer.
This property is read-only
:type: :py:class:`arcade.gl.Framebuffer`
"""
return self.active_framebuffer
@property
def gl_version(self) -> Tuple[int, int]:
"""
The OpenGL version as a 2 component tuple.
This is the reported OpenGL version from
drivers and might be a higher version than
you requested.
:type: tuple (major, minor) version
"""
return self._gl_version
[docs] def gc(self) -> int:
"""
Run garbage collection of OpenGL objects for this context.
This is only needed when ``gc_mode`` is ``context_gc``.
:return: The number of resources destroyed
:rtype: int
"""
# Loop the array until all objects are gone.
# Deleting one object might add new ones so we need
# to loop until the deque is empty
num_objects = 0
while len(self.objects):
obj = self.objects.popleft()
obj.delete()
num_objects += 1
return num_objects
@property
def gc_mode(self) -> str:
"""
Set the garbage collection mode for OpenGL resources.
Supported modes are::
# Default:
# Defer garbage collection until ctx.gc() is called
# This can be useful to enforce the main thread to
# run garbage collection of opengl resources
ctx.gc_mode = "context_gc"
# Auto collect is similar to python garbage collection.
# This is a risky mode. Know what you are doing before using this.
ctx.gc_mode = "auto"
"""
return self._gc_mode
@gc_mode.setter
def gc_mode(self, value: str):
modes = ["auto", "context_gc"]
if value not in modes:
raise ValueError("Unsupported gc_mode. Supported modes are:", modes)
self._gc_mode = value
@property
def error(self) -> Union[str, None]:
"""Check OpenGL error
Returns a string representation of the occurring error
or ``None`` of no errors has occurred.
Example::
err = ctx.error
if err:
raise RuntimeError("OpenGL error: {err}")
:type: str
"""
err = gl.glGetError()
if err == gl.GL_NO_ERROR:
return None
return self._errors.get(err, "GL_UNKNOWN_ERROR")
[docs] @classmethod
def activate(cls, ctx: "Context"):
"""
Mark a context as the currently active one.
.. Warning:: Never call this unless you know exactly what you are doing.
"""
cls.active = ctx
[docs] def enable(self, *flags):
"""
Enables one or more context flags::
# Single flag
ctx.enable(ctx.BLEND)
# Multiple flags
ctx.enable(ctx.DEPTH_TEST, ctx.CULL_FACE)
"""
self._flags.update(flags)
for flag in flags:
gl.glEnable(flag)
[docs] def enable_only(self, *args):
"""
Enable only some flags. This will disable all other flags.
This is a simple way to ensure that context flag states
are not lingering from other sections of your code base::
# Ensure all flags are disabled (enable no flags)
ctx.enable_only()
# Make sure only blending is enabled
ctx.enable_only(ctx.BLEND)
# Make sure only depth test and culling is enabled
ctx.enable_only(ctx.DEPTH_TEST, ctx.CULL_FACE)
"""
self._flags = set(args)
if self.BLEND in self._flags:
gl.glEnable(self.BLEND)
else:
gl.glDisable(self.BLEND)
if self.DEPTH_TEST in self._flags:
gl.glEnable(self.DEPTH_TEST)
else:
gl.glDisable(self.DEPTH_TEST)
if self.CULL_FACE in self._flags:
gl.glEnable(self.CULL_FACE)
else:
gl.glDisable(self.CULL_FACE)
if self.gl_api == "gl":
if self.PROGRAM_POINT_SIZE in self._flags:
gl.glEnable(self.PROGRAM_POINT_SIZE)
else:
gl.glDisable(self.PROGRAM_POINT_SIZE)
[docs] @contextmanager
def enabled(self, *flags):
"""
Temporarily change enabled flags.
Flags that was enabled initially will stay enabled.
Only new enabled flags will be reversed when exiting
the context.
Example::
with ctx.enabled(ctx.BLEND, ctx.CULL_FACE):
# Render something
"""
flags = set(flags)
new_flags = flags - self._flags
self.enable(*flags)
try:
yield
finally:
self.disable(*new_flags)
[docs] @contextmanager
def enabled_only(self, *flags):
"""
Temporarily change enabled flags.
Only the supplied flags with be enabled in
in the context. When exiting the context
the old flags will be restored.
Example::
with ctx.enabled_only(ctx.BLEND, ctx.CULL_FACE):
# Render something
"""
old_flags = self._flags
self.enable_only(*flags)
try:
yield
finally:
self.enable_only(*old_flags)
[docs] def disable(self, *args):
"""
Disable one or more context flags::
# Single flag
ctx.disable(ctx.BLEND)
# Multiple flags
ctx.disable(ctx.DEPTH_TEST, ctx.CULL_FACE)
"""
self._flags -= set(args)
for flag in args:
gl.glDisable(flag)
[docs] def is_enabled(self, flag) -> bool:
"""
Check if a context flag is enabled
:type: bool
"""
return flag in self._flags
@property
def viewport(self) -> Tuple[int, int, int, int]:
"""
Get or set the viewport for the currently active framebuffer.
The viewport simply describes what pixels of the screen
OpenGL should render to. Normally it would be the size of
the window's framebuffer::
# 4:3 screen
ctx.viewport = 0, 0, 800, 600
# 1080p
ctx.viewport = 0, 0, 1920, 1080
# Using the current framebuffer size
ctx.viewport = 0, 0, *ctx.screen.size
:type: tuple (x, y, width, height)
"""
return self.active_framebuffer.viewport
@viewport.setter
def viewport(self, value: Tuple[int, int, int, int]):
self.active_framebuffer.viewport = value
@property
def scissor(self) -> Optional[Tuple[int, int, int, int]]:
"""
Get or set the scissor box for the active framebuffer.
This is a shortcut for :py:meth:`~arcade.gl.Framebuffer.scissor`.
By default the scissor box is disabled and has no effect
and will have an initial value of ``None``. The scissor
box is enabled when setting a value and disabled when
set to ``None``.
Example::
# Set and enable scissor box only drawing
# in a 100 x 100 pixel lower left area
ctx.scissor = 0, 0, 100, 100
# Disable scissoring
ctx.scissor = None
:type: tuple (x, y, width, height)
"""
return self.fbo.scissor
@scissor.setter
def scissor(self, value):
self.fbo.scissor = value
@property
def blend_func(self) -> Union[Tuple[int, int], Tuple[int, int, int, int]]:
"""
Get or set the blend function.
This is tuple specifying how the color and
alpha blending factors are computed for the source
and destination pixel.
When using a two component tuple you specify the
blend function for the source and the destination.
When using a four component tuple you specify the
blend function for the source color, source alpha
destination color and destination alpha. (separate blend
functions for color and alpha)
Supported blend functions are::
ZERO
ONE
SRC_COLOR
ONE_MINUS_SRC_COLOR
DST_COLOR
ONE_MINUS_DST_COLOR
SRC_ALPHA
ONE_MINUS_SRC_ALPHA
DST_ALPHA
ONE_MINUS_DST_ALPHA
# Shortcuts
DEFAULT_BLENDING # (SRC_ALPHA, ONE_MINUS_SRC_ALPHA)
ADDITIVE_BLENDING # (ONE, ONE)
PREMULTIPLIED_ALPHA # (SRC_ALPHA, ONE)
These enums can be accessed in the ``arcade.gl``
module or simply as attributes of the context object.
The raw enums from ``pyglet.gl`` can also be used.
Example::
# Using constants from the context object
ctx.blend_func = ctx.ONE, ctx.ONE
# from the gl module
from arcade import gl
ctx.blend_func = gl.ONE, gl.ONE
:type: tuple (src, dst)
"""
return self._blend_func
@blend_func.setter
def blend_func(self, value: Union[Tuple[int, int], Tuple[int, int, int, int]]):
self._blend_func = value
if len(value) == 2:
gl.glBlendFunc(*value)
elif len(value) == 4:
gl.glBlendFuncSeparate(*value)
else:
ValueError("blend_func takes a tuple of 2 or 4 values")
# def blend_equation(self)
# def front_face(self)
# def cull_face(self)
@property
def wireframe(self) -> bool:
"""
Get or set the wireframe mode.
When enabled all primitives will be rendered as lines.
:type: bool
"""
return self._wireframe
@wireframe.setter
def wireframe(self, value: bool):
self._wireframe = value
if value:
gl.glPolygonMode(gl.GL_FRONT_AND_BACK, gl.GL_LINE)
else:
gl.glPolygonMode(gl.GL_FRONT_AND_BACK, gl.GL_FILL)
@property
def patch_vertices(self) -> int:
"""
Get or set number of vertices that will be used to make up a single patch primitive.
Patch primitives are consumed by the tessellation control shader (if present)
and subsequently used for tessellation.
:type: int
"""
value = c_int()
gl.glGetIntegerv(gl.GL_PATCH_VERTICES, value)
return value.value
@patch_vertices.setter
def patch_vertices(self, value: int):
if not isinstance(value, int):
raise TypeError("patch_vertices must be an integer")
gl.glPatchParameteri(gl.GL_PATCH_VERTICES, value)
@property
def point_size(self) -> float:
"""
Set or get the point size. Default is `1.0`.
Point size changes the pixel size of rendered points. The min and max values
are limited by :py:attr:`~arcade.gl.Context.info.POINT_SIZE_RANGE`.
This value usually at least ``(1, 100)``, but this depends on the drivers/vendors.
If variable point size is needed you can enable :py:attr:`~arcade.gl.Context.PROGRAM_POINT_SIZE`
and write to ``gl_PointSize`` in the vertex or geometry shader.
.. Note::
Using a geometry shader to create triangle strips from points is often a safer
way to render large points since you don't have have any size restrictions.
"""
return self._point_size
@point_size.setter
def point_size(self, value: float):
if self.gl_api == "gl":
gl.glPointSize(self._point_size)
self._point_size = value
@property
def primitive_restart_index(self) -> int:
"""
Get or set the primitive restart index. Default is ``-1``.
The primitive restart index can be used in index buffers
to restart a primitive. This is for example useful when you
use triangle strips or line strips and want to start on
a new strip in the same buffer / draw call.
"""
return self._primitive_restart_index
@primitive_restart_index.setter
def primitive_restart_index(self, value: int):
self._primitive_restart_index = value
if self.gl_api == "gl":
gl.glPrimitiveRestartIndex(value)
[docs] def finish(self) -> None:
"""
Wait until all OpenGL rendering commands are completed.
This function will actually stall until all work is done
and may have severe performance implications.
"""
gl.glFinish()
[docs] def flush(self) -> None:
"""
A suggestion to the driver to execute all the queued
drawing calls even if the queue is not full yet.
This is not a blocking call and only a suggestion.
This can potentially be used for speedups when
we don't have anything else to render.
"""
gl.glFlush()
# Various utility methods
[docs] def copy_framebuffer(self, src: Framebuffer, dst: Framebuffer):
"""
Copies/blits a framebuffer to another one.
This operation has many restrictions to ensure it works across
different platforms and drivers:
* The source and destination framebuffer must be the same size
* The formats of the attachments must be the same
* Only the source framebuffer can be multisampled
* Framebuffers cannot have integer attachments
:param Framebuffer src: The framebuffer to copy from
:param Framebuffer dst: The framebuffer we copy to
"""
gl.glBindFramebuffer(gl.GL_READ_FRAMEBUFFER, src._glo)
gl.glBindFramebuffer(gl.GL_DRAW_FRAMEBUFFER, dst._glo)
gl.glBlitFramebuffer(
0, 0, src.width, src.height, # Make source and dest size the same
0, 0, src.width, src.height,
gl.GL_COLOR_BUFFER_BIT | gl.GL_DEPTH_BUFFER_BIT,
gl.GL_NEAREST,
)
self.active_framebuffer.use(force=True)
# --- Resource methods ---
[docs] def buffer(
self, *, data: Optional[BufferProtocol] = None, reserve: int = 0, usage: str = "static"
) -> Buffer:
"""
Create an OpenGL Buffer object. The buffer will contain all zero-bytes if no data is supplied.
Examples::
# Create 1024 byte buffer
ctx.buffer(reserve=1024)
# Create a buffer with 1000 float values using python's array.array
from array import array
ctx.buffer(data=array('f', [i for in in range(1000)])
# Create a buffer with 1000 random 32 bit floats using numpy
self.ctx.buffer(data=np.random.random(1000).astype("f4"))
The ``data`` parameter can be anything that implements the
`Buffer Protocol <https://docs.python.org/3/c-api/buffer.html>`_.
This includes ``bytes``, ``bytearray``, ``array.array``, and
more. You may need to use typing workarounds for non-builtin
types. See :ref:`prog-guide-gl-buffer-protocol-typing` for more
information.
The ``usage`` parameter enables the GL implementation to make more intelligent
decisions that may impact buffer object performance. It does not add any restrictions.
If in doubt, skip this parameter and revisit when optimizing. The result
are likely to be different between vendors/drivers or may not have any effect.
The available values mean the following::
stream
The data contents will be modified once and used at most a few times.
static
The data contents will be modified once and used many times.
dynamic
The data contents will be modified repeatedly and used many times.
:param BufferProtocol data: The buffer data. This can be a ``bytes`` instance or any
any other object supporting the buffer protocol.
:param int reserve: The number of bytes to reserve
:param str usage: Buffer usage. 'static', 'dynamic' or 'stream'
:rtype: :py:class:`~arcade.gl.Buffer`
"""
return Buffer(self, data, reserve=reserve, usage=usage)
[docs] def framebuffer(
self,
*,
color_attachments: Optional[Union[Texture2D, List[Texture2D]]] = None,
depth_attachment: Optional[Texture2D] = None
) -> Framebuffer:
"""Create a Framebuffer.
:param List[arcade.gl.Texture] color_attachments: List of textures we want to render into
:param arcade.gl.Texture depth_attachment: Depth texture
:rtype: :py:class:`~arcade.gl.Framebuffer`
"""
return Framebuffer(
self, color_attachments=color_attachments, depth_attachment=depth_attachment
)
[docs] def texture(
self,
size: Tuple[int, int],
*,
components: int = 4,
dtype: str = "f1",
data: Optional[BufferProtocol] = None,
wrap_x: Optional[gl.GLenum] = None,
wrap_y: Optional[gl.GLenum] = None,
filter: Optional[Tuple[gl.GLenum, gl.GLenum]] = None,
samples: int = 0,
immutable: bool = False,
) -> Texture2D:
"""Create a 2D Texture.
Wrap modes: ``GL_REPEAT``, ``GL_MIRRORED_REPEAT``, ``GL_CLAMP_TO_EDGE``, ``GL_CLAMP_TO_BORDER``
Minifying filters: ``GL_NEAREST``, ``GL_LINEAR``, ``GL_NEAREST_MIPMAP_NEAREST``, ``GL_LINEAR_MIPMAP_NEAREST``
``GL_NEAREST_MIPMAP_LINEAR``, ``GL_LINEAR_MIPMAP_LINEAR``
Magnifying filters: ``GL_NEAREST``, ``GL_LINEAR``
:param Tuple[int, int] size: The size of the texture
:param int components: Number of components (1: R, 2: RG, 3: RGB, 4: RGBA)
:param str dtype: The data type of each component: f1, f2, f4 / i1, i2, i4 / u1, u2, u4
:param BufferProtocol data: The texture data (optional). Can be ``bytes``
or any object supporting the buffer protocol.
:param GLenum wrap_x: How the texture wraps in x direction
:param GLenum wrap_y: How the texture wraps in y direction
:param Tuple[GLenum,GLenum] filter: Minification and magnification filter
:param int samples: Creates a multisampled texture for values > 0
:param bool immutable: Make the storage (not the contents) immutable. This can sometimes be
required when using textures with compute shaders.
"""
return Texture2D(
self,
size,
components=components,
data=data,
dtype=dtype,
wrap_x=wrap_x,
wrap_y=wrap_y,
filter=filter,
samples=samples,
immutable=immutable,
)
[docs] def depth_texture(self, size: Tuple[int, int], *, data: Optional[BufferProtocol] = None) -> Texture2D:
"""
Create a 2D depth texture. Can be used as a depth attachment
in a :py:class:`~arcade.gl.Framebuffer`.
:param Tuple[int, int] size: The size of the texture
:param BufferProtocol data: The texture data (optional). Can be
``bytes`` or any object supporting
the buffer protocol.
"""
return Texture2D(self, size, data=data, depth=True)
[docs] def geometry(
self,
content: Optional[Sequence[BufferDescription]] = None,
index_buffer: Optional[Buffer] = None,
mode: Optional[int] = None,
index_element_size: int = 4,
):
"""
Create a Geometry instance. This is Arcade's version of a vertex array adding
a lot of convenience for the user. Geometry objects are fairly light. They are
mainly responsible for automatically map buffer inputs to your shader(s)
and provide various methods for rendering or processing this geometry,
The same geometry can be rendered with different
programs as long as your shader is using one or more of the input attribute.
This means geometry with positions and colors can be rendered with a program
only using the positions. We will automatically map what is necessary and
cache these mappings internally for performace.
In short, the geometry object is a light object that describes what buffers
contains and automatically negotiate with shaders/programs. This is a very
complex field in OpenGL so the Geometry object provides substantial time
savings and greatly reduces the complexity of your code.
Geometry also provide rendering methods supporting the following:
* Rendering geometry with and without index buffer
* Rendering your geometry using instancing. Per instance buffers can be provided
or the current instance can be looked up using ``gl_InstanceID`` in shaders.
* Running transform feedback shaders that writes to buffers instead the screen.
This can write to one or multiple buffer.
* Render your geometry with indirect rendering. This means packing
multiple meshes into the same buffer(s) and batch drawing them.
Examples::
# Single buffer geometry with a vec2 vertex position attribute
ctx.geometry([BufferDescription(buffer, '2f', ["in_vert"])], mode=ctx.TRIANGLES)
# Single interlaved buffer with two attributes. A vec2 position and vec2 velocity
ctx.geometry([
BufferDescription(buffer, '2f 2f', ["in_vert", "in_velocity"])
],
mode=ctx.POINTS,
)
# Geometry with index buffer
ctx.geometry(
[BufferDescription(buffer, '2f', ["in_vert"])],
index_buffer=ibo,
mode=ctx.TRIANGLES,
)
# Separate buffers
ctx.geometry([
BufferDescription(buffer_pos, '2f', ["in_vert"])
BufferDescription(buffer_vel, '2f', ["in_velocity"])
],
mode=ctx.POINTS,
)
# Providing per-instance data for instancing
ctx.geometry([
BufferDescription(buffer_pos, '2f', ["in_vert"])
BufferDescription(buffer_instance_pos, '2f', ["in_offset"], instanced=True)
],
mode=ctx.POINTS,
)
:param list content: List of :py:class:`~arcade.gl.BufferDescription` (optional)
:param Buffer index_buffer: Index/element buffer (optional)
:param int mode: The default draw mode (optional)
:param int mode: The default draw mode (optional)
:param int index_element_size: Byte size of a single index/element in the index buffer.
In other words, the index buffer can be 8, 16 or 32 bit integers.
Can be 1, 2 or 4 (8, 16 or 32 bit unsigned integer)
"""
return Geometry(
self,
content,
index_buffer=index_buffer,
mode=mode,
index_element_size=index_element_size,
)
[docs] def program(
self,
*,
vertex_shader: str,
fragment_shader: Optional[str] = None,
geometry_shader: Optional[str] = None,
tess_control_shader: Optional[str] = None,
tess_evaluation_shader: Optional[str] = None,
common: Optional[List[str]] = None,
defines: Optional[Dict[str, str]] = None,
varyings: Optional[Sequence[str]] = None,
varyings_capture_mode: str = "interleaved",
) -> Program:
"""Create a :py:class:`~arcade.gl.Program` given the vertex, fragment and geometry shader.
:param str vertex_shader: vertex shader source
:param str fragment_shader: fragment shader source (optional)
:param str geometry_shader: geometry shader source (optional)
:param str tess_control_shader: tessellation control shader source (optional)
:param str tess_evaluation_shader: tessellation evaluation shader source (optional)
:param list common: Common shader sources injected into all shaders
:param dict defines: Substitute #defines values in the source (optional)
:param Optional[Sequence[str]] varyings: The name of the out attributes in a transform shader.
This is normally not necessary since we auto detect them,
but some more complex out structures we can't detect.
:param str varyings_capture_mode: The capture mode for transforms.
``"interleaved"`` means all out attribute will be written to a single buffer.
``"separate"`` means each out attribute will be written separate buffers.
Based on these settings the `transform()` method will accept a single
buffer or a list of buffer.
:rtype: :py:class:`~arcade.gl.Program`
"""
source_vs = ShaderSource(self, vertex_shader, common, gl.GL_VERTEX_SHADER)
source_fs = (
ShaderSource(self, fragment_shader, common, gl.GL_FRAGMENT_SHADER)
if fragment_shader
else None
)
source_geo = (
ShaderSource(self, geometry_shader, common, gl.GL_GEOMETRY_SHADER)
if geometry_shader
else None
)
source_tc = (
ShaderSource(self, tess_control_shader, common, gl.GL_TESS_CONTROL_SHADER)
if tess_control_shader
else None
)
source_te = (
ShaderSource(self, tess_evaluation_shader, common, gl.GL_TESS_EVALUATION_SHADER)
if tess_evaluation_shader
else None
)
# If we don't have a fragment shader we are doing transform feedback.
# When a geometry shader is present the out attributes will be located there
out_attributes = list(varyings) if varyings is not None else [] # type: List[str]
if not source_fs and not out_attributes:
if source_geo:
out_attributes = source_geo.out_attributes
else:
out_attributes = source_vs.out_attributes
return Program(
self,
vertex_shader=source_vs.get_source(defines=defines),
fragment_shader=source_fs.get_source(defines=defines)
if source_fs
else None,
geometry_shader=source_geo.get_source(defines=defines)
if source_geo
else None,
tess_control_shader=source_tc.get_source(defines=defines)
if source_tc
else None,
tess_evaluation_shader=source_te.get_source(defines=defines)
if source_te
else None,
varyings=out_attributes,
varyings_capture_mode=varyings_capture_mode,
)
[docs] def query(self, *, samples=True, time=True, primitives=True) -> Query:
"""
Create a query object for measuring rendering calls in opengl.
:param bool samples: Collect written samples
:param bool time: Measure rendering duration
:param bool primitives: Collect the number of primitives emitted
:rtype: :py:class:`~arcade.gl.Query`
"""
return Query(self, samples=samples, time=time, primitives=primitives)
[docs] def compute_shader(self, *, source: str, common: Iterable[str] = ()) -> ComputeShader:
"""
Create a compute shader.
:param str source: The glsl source
:param Iterable[str] common: Common / library source injected into compute shader
"""
src = ShaderSource(self, source, common, gl.GL_COMPUTE_SHADER)
return ComputeShader(self, src.get_source())
[docs]class ContextStats:
"""
Runtime allocation statistics of OpenGL objects.
"""
def __init__(self, warn_threshold=100):
self.warn_threshold = warn_threshold
#: Textures (created, freed)
self.texture = (0, 0)
#: Framebuffers (created, freed)
self.framebuffer = (0, 0)
#: Buffers (created, freed)
self.buffer = (0, 0)
#: Programs (created, freed)
self.program = (0, 0)
#: Vertex Arrays (created, freed)
self.vertex_array = (0, 0)
#: Geometry (created, freed)
self.geometry = (0, 0)
#: Compute Shaders (created, freed)
self.compute_shader = (0, 0)
#: Queries (created, freed)
self.query = (0, 0)
[docs] def incr(self, key: str) -> None:
"""
Increments a counter.
:param str key: The attribute name / counter to increment.
"""
created, freed = getattr(self, key)
setattr(self, key, (created + 1, freed))
if created % self.warn_threshold == 0 and created > 0:
LOG.debug(
"%s allocations passed threshold (%s) [created = %s] [freed = %s] [active = %s]",
key,
self.warn_threshold,
created,
freed,
created - freed,
)
[docs] def decr(self, key):
"""
Decrement a counter.
:param str key: The attribute name / counter to decrement.
"""
created, freed = getattr(self, key)
setattr(self, key, (created, freed + 1))
[docs]class Limits:
"""OpenGL Limitations"""
def __init__(self, ctx):
self._ctx = ctx
#: Minor version number of the OpenGL API supported by the current context
self.MINOR_VERSION = self.get(gl.GL_MINOR_VERSION)
#: Major version number of the OpenGL API supported by the current context.
self.MAJOR_VERSION = self.get(gl.GL_MAJOR_VERSION)
#: The vendor string. For example "NVIDIA Corporation"
self.VENDOR = self.get_str(gl.GL_VENDOR)
#: The renderer things. For example "NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2"
self.RENDERER = self.get_str(gl.GL_RENDERER)
#: Value indicating the number of sample buffers associated with the framebuffer
self.SAMPLE_BUFFERS = self.get(gl.GL_SAMPLE_BUFFERS)
#: An estimate of the number of bits of subpixel resolution
#: that are used to position rasterized geometry in window coordinates
self.SUBPIXEL_BITS = self.get(gl.GL_SUBPIXEL_BITS)
#: Minimum required alignment for uniform buffer sizes and offset
self.UNIFORM_BUFFER_OFFSET_ALIGNMENT = self.get(
gl.GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT
)
#: Value indicates the maximum number of layers allowed in an array texture, and must be at least 256
self.MAX_ARRAY_TEXTURE_LAYERS = self.get(gl.GL_MAX_ARRAY_TEXTURE_LAYERS)
#: A rough estimate of the largest 3D texture that the GL can handle. The value must be at least 64
self.MAX_3D_TEXTURE_SIZE = self.get(gl.GL_MAX_3D_TEXTURE_SIZE)
#: Maximum number of color attachments in a framebuffer
self.MAX_COLOR_ATTACHMENTS = self.get(gl.GL_MAX_COLOR_ATTACHMENTS)
#: Maximum number of samples in a color multisample texture
self.MAX_COLOR_TEXTURE_SAMPLES = self.get(gl.GL_MAX_COLOR_TEXTURE_SAMPLES)
#: the number of words for fragment shader uniform variables in all uniform blocks
self.MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS
)
#: Number of words for geometry shader uniform variables in all uniform blocks
self.MAX_COMBINED_GEOMETRY_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_COMBINED_GEOMETRY_UNIFORM_COMPONENTS
)
#: Maximum supported texture image units that can be used to access texture maps from the vertex shader
self.MAX_COMBINED_TEXTURE_IMAGE_UNITS = self.get(
gl.GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS
)
#: Maximum number of uniform blocks per program
self.MAX_COMBINED_UNIFORM_BLOCKS = self.get(gl.GL_MAX_COMBINED_UNIFORM_BLOCKS)
#: Number of words for vertex shader uniform variables in all uniform blocks
self.MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS
)
#: A rough estimate of the largest cube-map texture that the GL can handle
self.MAX_CUBE_MAP_TEXTURE_SIZE = self.get(gl.GL_MAX_CUBE_MAP_TEXTURE_SIZE)
#: Maximum number of samples in a multisample depth or depth-stencil texture
self.MAX_DEPTH_TEXTURE_SAMPLES = self.get(gl.GL_MAX_DEPTH_TEXTURE_SAMPLES)
#: Maximum number of simultaneous outputs that may be written in a fragment shader
self.MAX_DRAW_BUFFERS = self.get(gl.GL_MAX_DRAW_BUFFERS)
#: Recommended maximum number of vertex array indices
self.MAX_ELEMENTS_INDICES = self.get(gl.GL_MAX_ELEMENTS_INDICES)
#: Recommended maximum number of vertex array vertices
self.MAX_ELEMENTS_VERTICES = self.get(gl.GL_MAX_ELEMENTS_VERTICES)
#: Maximum number of components of the inputs read by the fragment shader
self.MAX_FRAGMENT_INPUT_COMPONENTS = self.get(
gl.GL_MAX_FRAGMENT_INPUT_COMPONENTS
)
#: Maximum number of individual floating-point, integer, or boolean values that can be
#: held in uniform variable storage for a fragment shader
self.MAX_FRAGMENT_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_FRAGMENT_UNIFORM_COMPONENTS
)
#: maximum number of individual 4-vectors of floating-point, integer,
#: or boolean values that can be held in uniform variable storage for a fragment shader
self.MAX_FRAGMENT_UNIFORM_VECTORS = self.get(gl.GL_MAX_FRAGMENT_UNIFORM_VECTORS)
#: Maximum number of uniform blocks per fragment shader.
self.MAX_FRAGMENT_UNIFORM_BLOCKS = self.get(gl.GL_MAX_FRAGMENT_UNIFORM_BLOCKS)
#: Maximum number of components of inputs read by a geometry shader
self.MAX_GEOMETRY_INPUT_COMPONENTS = self.get(
gl.GL_MAX_GEOMETRY_INPUT_COMPONENTS
)
#: Maximum number of components of outputs written by a geometry shader
self.MAX_GEOMETRY_OUTPUT_COMPONENTS = self.get(
gl.GL_MAX_GEOMETRY_OUTPUT_COMPONENTS
)
#: Maximum supported texture image units that can be used to access texture maps from the geometry shader
self.MAX_GEOMETRY_TEXTURE_IMAGE_UNITS = self.get(
gl.GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS
)
#: Maximum number of uniform blocks per geometry shader
self.MAX_GEOMETRY_UNIFORM_BLOCKS = self.get(gl.GL_MAX_GEOMETRY_UNIFORM_BLOCKS)
#: Maximum number of individual floating-point, integer, or boolean values that can
#: be held in uniform variable storage for a geometry shader
self.MAX_GEOMETRY_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_GEOMETRY_UNIFORM_COMPONENTS
)
#: Maximum number of samples supported in integer format multisample buffers
self.MAX_INTEGER_SAMPLES = self.get(gl.GL_MAX_INTEGER_SAMPLES)
#: Maximum samples for a framebuffer
self.MAX_SAMPLES = self.get(gl.GL_MAX_SAMPLES)
#: Maximum supported size for renderbuffers
self.MAX_RENDERBUFFER_SIZE = self.get(gl.GL_MAX_RENDERBUFFER_SIZE)
#: Maximum number of sample mask words
self.MAX_SAMPLE_MASK_WORDS = self.get(gl.GL_MAX_SAMPLE_MASK_WORDS)
#: Maximum number of uniform buffer binding points on the context
self.MAX_UNIFORM_BUFFER_BINDINGS = self.get(gl.GL_MAX_UNIFORM_BUFFER_BINDINGS)
#: Maximum number of uniform buffer binding points on the context
self.MAX_UNIFORM_BUFFER_BINDINGS = self.get(gl.GL_MAX_UNIFORM_BUFFER_BINDINGS)
#: The value gives a rough estimate of the largest texture that the GL can handle
self.MAX_TEXTURE_SIZE = self.get(gl.GL_MAX_TEXTURE_SIZE)
#: Maximum number of uniform buffer binding points on the context
self.MAX_UNIFORM_BUFFER_BINDINGS = self.get(gl.GL_MAX_UNIFORM_BUFFER_BINDINGS)
#: Maximum size in basic machine units of a uniform block
self.MAX_UNIFORM_BLOCK_SIZE = self.get(gl.GL_MAX_UNIFORM_BLOCK_SIZE)
#: The number 4-vectors for varying variables
self.MAX_VARYING_VECTORS = self.get(gl.GL_MAX_VARYING_VECTORS)
#: Maximum number of 4-component generic vertex attributes accessible to a vertex shader.
self.MAX_VERTEX_ATTRIBS = self.get(gl.GL_MAX_VERTEX_ATTRIBS)
#: Maximum supported texture image units that can be used to access texture maps from the vertex shader.
self.MAX_VERTEX_TEXTURE_IMAGE_UNITS = self.get(
gl.GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS
)
#: Maximum number of individual floating-point, integer, or boolean values that
#: can be held in uniform variable storage for a vertex shader
self.MAX_VERTEX_UNIFORM_COMPONENTS = self.get(
gl.GL_MAX_VERTEX_UNIFORM_COMPONENTS
)
#: Maximum number of 4-vectors that may be held in uniform variable storage for the vertex shader
self.MAX_VERTEX_UNIFORM_VECTORS = self.get(gl.GL_MAX_VERTEX_UNIFORM_VECTORS)
#: Maximum number of components of output written by a vertex shader
self.MAX_VERTEX_OUTPUT_COMPONENTS = self.get(gl.GL_MAX_VERTEX_OUTPUT_COMPONENTS)
#: Maximum number of uniform blocks per vertex shader.
self.MAX_VERTEX_UNIFORM_BLOCKS = self.get(gl.GL_MAX_VERTEX_UNIFORM_BLOCKS)
# self.MAX_VERTEX_ATTRIB_RELATIVE_OFFSET = self.get(gl.GL_MAX_VERTEX_ATTRIB_RELATIVE_OFFSET)
# self.MAX_VERTEX_ATTRIB_BINDINGS = self.get(gl.GL_MAX_VERTEX_ATTRIB_BINDINGS)
self.MAX_TEXTURE_IMAGE_UNITS = self.get(gl.GL_MAX_TEXTURE_IMAGE_UNITS)
#: The highest supported anisotropy value. Usually 8.0 or 16.0.
self.MAX_TEXTURE_MAX_ANISOTROPY = self.get_float(gl.GL_MAX_TEXTURE_MAX_ANISOTROPY, 1.0)
#: The maximum support window or framebuffer viewport.
#: This is usually the same as the maximum texture size
self.MAX_VIEWPORT_DIMS = self.get_int_tuple(gl.GL_MAX_VIEWPORT_DIMS, 2)
#: How many buffers we can have as output when doing a transform(feedback).
#: This is usually 4
self.MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS = self.get(gl.GL_MAX_TRANSFORM_FEEDBACK_SEPARATE_ATTRIBS)
#: The minimum and maximum point size
self.POINT_SIZE_RANGE = self.get_int_tuple(gl.GL_POINT_SIZE_RANGE, 2)
err = self._ctx.error
if err:
from warnings import warn
warn("Error happened while querying of limits. Moving on ..")
[docs] def get_int_tuple(self, enum: gl.GLenum, length: int):
"""Get an enum as an int tuple"""
try:
values = (c_int * length)()
gl.glGetIntegerv(enum, values)
return tuple(values)
except gl.lib.GLException:
return tuple([0] * length)
[docs] def get(self, enum: gl.GLenum, default=0) -> int:
"""Get an integer limit"""
try:
value = c_int()
gl.glGetIntegerv(enum, value)
return value.value
except gl.lib.GLException:
return default
[docs] def get_float(self, enum: gl.GLenum, default=0.0) -> float:
"""Get a float limit"""
try:
value = c_float()
gl.glGetFloatv(enum, value)
return value.value
except gl.lib.GLException:
return default
[docs] def get_str(self, enum: gl.GLenum) -> str:
"""Get a string limit"""
try:
return cast(gl.glGetString(enum), c_char_p).value.decode() # type: ignore
except gl.lib.GLException:
return "Unknown"