def _load_vector(location: int, vector: np.ndarray) -> None:
if len(vector) == 2:
glUniform2f(location, *vector)
elif len(vector) == 3:
glUniform3f(location, *vector)
else:
glUniform4f(location, *vector)
def uniform(self, name, value):
with self:
if type(value) == int or type(value) == long:
glUniform1i(self.uniformLocation(name), value)
elif type(value) == float or type(value) == numpy.float32:
glUniform1f(self.uniformLocation(name), value)
elif len(value) == 4:
value = numpy.array(value, dtype=numpy.float32)
glUniform4f(self.uniformLocation(name), *value)
elif len(value) == 16:
glUniformMatrix4fv(self.uniformLocation(name), 1, GL_FALSE, value)
elif len(value) == 9:
glUniformMatrix3fv(self.uniformLocation(name), 1, GL_FALSE, value)
elif len(value) == 2:
value = numpy.array(value, dtype=numpy.float32)
glUniform2f(self.uniformLocation(name), *value)
else:
print(len(value))
raise Exception('not implemented')
def bind_uniform_vector(self, data, name):
"""Bind uniform vector parameter."""
location = glGetUniformLocation(self.__program, name)
assert location >= 0
glUniform4f(location, *data.flatten())
def bind_uniform_vector(self, data, name):
"""Bind uniform vector parameter."""
location = glGetUniformLocation(self.__program, name)
assert location >= 0
glUniform4f(location, *data.flatten())
def SetUniform4f(self, name, v0, v1, v2, v3):
glUniform4f(self.GetUniformLocation(name), v0, v1, v2, v3)
def main(visualize):
if args.config:
config_path = args.config
else:
config_path = "config.yaml"
with open(config_path) as file:
config = yaml.load(file, Loader=yaml.FullLoader)
teams, boxes, agents = generate_spatial_entities(config)
agents_x_sorted = initial_sort(agents)
if config["sotilaskoti"]["allow"]:
# create queue to the sotilaskoti
q = []
# table of meetings between agents, from the previous simulation step
meets_prev = dict()
if visualize:
# verticies for area borders and inferred width and height of the map.
# I.e. canvas contains map width and height in meters for computations
fences_verts, canvas = generate_map(boxes, config)
# verticies for traingles that represent agents
agents_verts = generate_agents_verticies(config)
if not glfw.init():
return
window = glfw.create_window(config["window"][ "width"],
config["window"]["height"],
config["window"][ "title"],
None, None)
if not window:
glfw.terminate()
return
glfw.make_context_current(window)
# compile shader for on-the-fly configurable trianges
shader = compile_shader()
# create Buffer object in gpu
VBO = glGenBuffers(2)
# bind buffers
glBindBuffer(GL_ARRAY_BUFFER, VBO[0])
glBufferData(GL_ARRAY_BUFFER,
fences_verts.nbytes,
fences_verts,
GL_STATIC_DRAW)
glBindBuffer(GL_ARRAY_BUFFER, VBO[1])
glBufferData(GL_ARRAY_BUFFER,
agents_verts.nbytes,
agents_verts,
GL_STATIC_DRAW)
fences_stride = fences_verts.strides[0]
agents_stride = agents_verts.strides[0]
# get the position from vertex shader
# stride offset
offset = ctypes.c_void_p(0)
init_pos = glGetAttribLocation(shader, 'init_pos')
glVertexAttribPointer(init_pos, 2, GL_FLOAT,
GL_FALSE, agents_stride, offset)
glEnableVertexAttribArray(init_pos)
glUseProgram(shader)
glClearColor(1.0, 1.0, 1.0, 1.0)
"""
Prepare directories to store:
- source configuration files,
- intermediate results in the form of meeting tables
- output statistical reports
"""
if not os.path.exists("output"):
os.makedirs("output")
paths = {
"configs" : os.path.join("output", "configs"),
"meet_tables" : os.path.join("output", "meetings_tables"),
"agents" : os.path.join("output", "agents"),
"out_stats" : os.path.join("output", "stat_results"),
}
for path in paths.values():
if not os.path.exists(path):
os.makedirs(path)
if args.config:
# in this usage scenario all identifiers are set manually
# (unique tags are generated in meta-loop that launches these scripts)
tag = args.name
else:
# in this usage scenario timestamp is autimatically appended to
# distinguish between consequtive manual program launches
timestamp = datetime.now().strftime("%H:%M:%S")
tag = args.name +'_'+ timestamp
# store the config file for the reference
dump_config_path = os.path.join(
paths["configs"], "config_"+ tag +".yaml")
shutil.copy(config_path, dump_config_path)
# store agents for the further move speed / infection spread correlating
agents_souls_path = os.path.join(
paths["agents"], "spatial_agents_"+ tag +".bin")
with open(agents_souls_path, 'wb') as file:
pickle.dump(agents, file)
# create the file with agent meetings
# originally a .bin file, is later compressed to the .bin.tar.bz2 format
meets_table_path = os.path.join(
paths["meet_tables"], "meet_table_"+ tag +".bin")
with open(meets_table_path, 'wb') as file:
# run until the end of the set simulation period
T = config["simulationDuration"] * 24*60*60
dt = config[ "minSimulationStep"]
eval_times = np.arange(0, T, dt)
for eval_time in tqdm(eval_times):
"""
Transition agents between service and leave
"""
entities = (teams, boxes, agents)
# some agents prefer to stay on the base during holidays
stay_chance = config.get('dontGoOffDuty', 0.0)
rotate_teams(entities, stay_chance, eval_time, dt)
"""
Transition agents to "Sotilaskoti" cafeteria and back
"""
if config["sotilaskoti"]["allow"]:
queue_sotilaskoti(entities, q, eval_time, dt, config)
"""
Update agent positions (along one time step)
"""
increment_agent_positions(agents)
"""
Refresh the sorting of agents after the positions update
"""
x_sort(agents_x_sorted)
"""
Register new meetings between agents and export them to file
"""
meets_curr = detect_meetings(agents_x_sorted, eval_time,
config, visualize)
# each key is a meeting link between two agents
# in the form {agent1_idx, agent2_idx}
links_curr = set( meets_curr.keys() )
links_prev = set( meets_prev.keys() )
meets_new = dict()
for link in links_curr:
if link not in links_prev:
meets_new[link] = meets_curr[link]
if meets_new:
timeline = {"timestamp" : eval_time,
"meetings" : meets_new}
pickle.dump(timeline, file)
meets_prev = meets_curr
"""
Plot canvas if not specified otherwise (--no-visual option)
"""
if visualize:
if glfw.window_should_close(window):
break
time_zero = time.time()
glClear(GL_COLOR_BUFFER_BIT)
"""
Indicate current day in window title
"""
day_n = eval_time // (24*60*60) + 1
dayly_title = config["window"]["title"] +", day: "+ str(day_n)
glfw.set_window_title(window, dayly_title)
"""
Draw borders (i.e. boxes, i.e. fences) - 1 px black outlines
"""
glBindBuffer(GL_ARRAY_BUFFER, VBO[0])
glVertexAttribPointer(init_pos, 2, GL_FLOAT,
GL_FALSE, fences_stride, offset)
glEnableVertexAttribArray(init_pos)
transformLoc = glGetUniformLocation(shader, "dyn_pos")
glUniform2f(transformLoc, 0.0, 0.0)
transformLoc = glGetUniformLocation(shader, "dyn_color")
glUniform4f(transformLoc, 0.0, 0.0, 0.0, 0.0)
glDrawArrays(GL_TRIANGLES, 0, len(fences_verts))
"""
Draw agents (i.e. conscripts and civilians)
"""
glBindBuffer(GL_ARRAY_BUFFER, VBO[1])
glVertexAttribPointer(init_pos, 2, GL_FLOAT,
GL_FALSE, agents_stride, offset)
glEnableVertexAttribArray(init_pos)
for i, agent in enumerate(agents):
poly_prop = np.zeros(1, [( "pos" , np.float32, 2),
("color", np.float32, 4)])
# absolute to relative coordinates, meters -> fractions
x = (agent.x/canvas[ "width"]*2 - 1)*0.99
y = (agent.y/canvas["height"]*2 - 1)*0.99
poly_prop["pos"] = (x, y)
transformLoc = glGetUniformLocation(shader, "dyn_pos")
glUniform2f(transformLoc, *poly_prop["pos"].T)
"""
Agent triangle marker filling
"""
poly_prop["color"] = agent.color
transformLoc = glGetUniformLocation(shader, "dyn_color")
glUniform4f(transformLoc, *poly_prop["color"].T)
if agent.conscripted:
glDrawArrays(GL_TRIANGLES, 3, 6)
else:
glDrawArrays(GL_TRIANGLES, 0, 3)
"""
Marker outline
"""
transformLoc = glGetUniformLocation(shader, "dyn_color")
glUniform4f(transformLoc, 0.0, 0.0, 0.0, 1.0) # black
if agent.conscripted:
glDrawArrays(GL_LINE_LOOP, 3, 6)
else:
glDrawArrays(GL_LINE_LOOP, 0, 3)
glfw.swap_buffers(window)
# FPS limited to 60
while(time.time() - time_zero < 1/60):
time.sleep(0.001)
glfw.poll_events()
if visualize:
glfw.terminate()
"""
Compress output file to save space
"""
compressed_path = os.path.join(
paths["meet_tables"], "meet_table_"+ tag +".bin.tar.bz2")
with tarfile.open(compressed_path, "w:bz2") as tar:
tar.add(meets_table_path)
# in case compressing went successful, remove the source file
if os.path.exists(compressed_path):
os.remove(meets_table_path)