def prove():
# TODO: this is pretty slow, maybe use this to improve performance:
# https://github.com/google/flatbuffers/issues/4668
f = open('computation.zkif', 'wb')
print("*** zkinterface: writing circuit", file=sys.stderr)
builder = flatbuffers.Builder(1024)
vars = write_varlist(builder, pubvals, 1)
Circuit.CircuitStartFieldMaximumVector(builder, 32)
for i in reversed(range(32)):
builder.PrependByte((modulus>>(i*8))&255)
maxi = builder.EndVector(32)
Circuit.CircuitStart(builder)
Circuit.CircuitAddConnections(builder, vars)
Circuit.CircuitAddFreeVariableId(builder, len(pubvals)+len(privvals)+1)
Circuit.CircuitAddR1csGeneration(builder, True)
Circuit.CircuitAddWitnessGeneration(builder, True)
Circuit.CircuitAddFieldMaximum(builder, maxi)
circ = Circuit.CircuitEnd(builder)
Root.RootStart(builder)
Root.RootAddMessageType(builder, Message.Message.Circuit)
Root.RootAddMessage(builder, circ)
root = Root.RootEnd(builder)
builder.FinishSizePrefixed(root)
buf = builder.Output()
f.write(buf)
print("*** zkinterface: writing witness", file=sys.stderr)
# build witness
builder = flatbuffers.Builder(1024)
vars = write_varlist(builder, privvals, len(pubvals)+1)
Witness.WitnessStart(builder)
Witness.WitnessAddAssignedVariables(builder, vars)
wit = Witness.WitnessEnd(builder)
Root.RootStart(builder)
Root.RootAddMessageType(builder, Message.Message.Witness)
Root.RootAddMessage(builder, wit)
root = Root.RootEnd(builder)
builder.FinishSizePrefixed(root)
buf = builder.Output()
f.write(buf)
print("*** zkinterface: writing constraints", file=sys.stderr)
builder = flatbuffers.Builder(1024)
def write_lc(lc):
varls = list(lc.lc.keys())
Variables.VariablesStartVariableIdsVector(builder, len(varls))
for i in reversed(range(len(varls))):
varix = varls[i] if varls[i]>=0 else len(pubvals)-varls[i]
builder.PrependUint64(varix)
vars = builder.EndVector(len(varls))
Variables.VariablesStartValuesVector(builder, 32*len(varls))
for i in reversed(range(len(varls))):
for j in reversed(range(32)):
val=lc.lc[varls[i]]%modulus
builder.PrependByte((val>>(j*8))&255)
vals = builder.EndVector(32*len(varls))
Variables.VariablesStart(builder)
Variables.VariablesAddVariableIds(builder, vars)
Variables.VariablesAddValues(builder, vals)
return Variables.VariablesEnd(builder)
def write_constraint(c):
la = write_lc(c[0])
lb = write_lc(c[1])
lc = write_lc(c[2])
BilinearConstraint.BilinearConstraintStart(builder)
BilinearConstraint.BilinearConstraintAddLinearCombinationA(builder, la)
BilinearConstraint.BilinearConstraintAddLinearCombinationB(builder, lb)
BilinearConstraint.BilinearConstraintAddLinearCombinationC(builder, lc)
return BilinearConstraint.BilinearConstraintEnd(builder)
cs = [write_constraint(c) for c in constraints]
R1CSConstraints.R1CSConstraintsStartConstraintsVector(builder, len(cs))
for i in reversed(range(len(cs))):
builder.PrependUOffsetTRelative(cs[i])
cvec = builder.EndVector(len(cs))
R1CSConstraints.R1CSConstraintsStart(builder)
R1CSConstraints.R1CSConstraintsAddConstraints(builder, cvec)
r1cs = R1CSConstraints.R1CSConstraintsEnd(builder)
Root.RootStart(builder)
Root.RootAddMessageType(builder, Message.Message.R1CSConstraints)
Root.RootAddMessage(builder, r1cs)
root = Root.RootEnd(builder)
builder.FinishSizePrefixed(root)
buf = builder.Output()
f.write(buf)
f.close()
print("*** zkinterface circuit, witness, constraints written to 'computation.zkif', size", len(buf))
def write_flatbuffers_trace(rows,
filename,
directory="",
persistent=True,
with_time=True):
import flatbuffers
from serialization.Timescales import Message, NullableInt64, NullableBool
try:
os.remove(
os.path.join(directory,
'{filename}.bin'.format(filename=filename)))
except OSError:
pass
with open(
os.path.join(directory,
'{filename}.bin'.format(filename=filename)),
'w+b') as f:
previous = dict()
for row in rows:
builder = flatbuffers.Builder(512)
Message.MessageStart(builder)
if with_time:
Message.MessageAddTime(
builder,
NullableInt64.CreateNullableInt64(builder, row.time))
if not persistent:
current = row.data._asdict()
if 'p' in current:
Message.MessageAddPropP(
builder,
NullableBool.CreateNullableBool(builder, current['p']))
if 'q' in current:
Message.MessageAddPropQ(
builder,
NullableBool.CreateNullableBool(builder, current['q']))
if 'r' in current:
Message.MessageAddPropR(
builder,
NullableBool.CreateNullableBool(builder, current['r']))
if 's' in current:
Message.MessageAddPropS(
builder,
NullableBool.CreateNullableBool(builder, current['s']))
else:
current = row.data._asdict()
if 'p' in current and (not 'p' in previous
or current['p'] != previous['p']):
Message.MessageAddPropP(
builder,
NullableBool.CreateNullableBool(builder, current['p']))
if 'q' in current and (not 'q' in previous
or current['q'] != previous['q']):
Message.MessageAddPropQ(
builder,
NullableBool.CreateNullableBool(builder, current['q']))
if 'r' in current and (not 'r' in previous
or current['r'] != previous['r']):
Message.MessageAddPropR(
builder,
NullableBool.CreateNullableBool(builder, current['r']))
if 's' in current and (not 's' in previous
or current['s'] != previous['s']):
Message.MessageAddPropS(
builder,
NullableBool.CreateNullableBool(builder, current['s']))
previous = current
msg = Message.MessageEnd(builder)
builder.FinishSizePrefixed(msg)
buf = builder.Output()
f.write(buf)
def _create_metadata(self):
"""Creates the metadata for a text classifier."""
# Creates model info.
model_meta = _metadata_fb.ModelMetadataT()
model_meta.name = self.model_info.name
model_meta.description = self.model_info.description
model_meta.version = self.model_info.version
model_meta.author = "TensorFlow"
model_meta.license = ("Apache License. Version 2.0 "
"http://www.apache.org/licenses/LICENSE-2.0.")
# Creates input info.
input_meta = _metadata_fb.TensorMetadataT()
input_meta.name = "input_text"
input_meta.description = (
"Embedding vectors representing the input text to be classified. The "
"input need to be converted from raw text to embedding vectors using "
"the attached dictionary file.")
# Create the vocab file.
vocab_file = _metadata_fb.AssociatedFileT()
vocab_file.name = os.path.basename(self.associated_files[1])
vocab_file.description = ("Vocabulary file to convert natural language "
"words to embedding vectors.")
vocab_file.type = _metadata_fb.AssociatedFileType.VOCABULARY
# Create the RegexTokenizer.
tokenizer = _metadata_fb.ProcessUnitT()
tokenizer.optionsType = (
_metadata_fb.ProcessUnitOptions.RegexTokenizerOptions)
tokenizer.options = _metadata_fb.RegexTokenizerOptionsT()
tokenizer.options.delimRegexPattern = self.model_info.delim_regex_pattern
tokenizer.options.vocabFile = [vocab_file]
input_meta.content = _metadata_fb.ContentT()
input_meta.content.contentPropertiesType = (
_metadata_fb.ContentProperties.FeatureProperties)
input_meta.content.contentProperties = _metadata_fb.FeaturePropertiesT()
input_meta.processUnits = [tokenizer]
# Creates output info.
output_meta = _metadata_fb.TensorMetadataT()
output_meta.name = "probability"
output_meta.description = "Probabilities of the labels respectively."
output_meta.content = _metadata_fb.ContentT()
output_meta.content.contentProperties = _metadata_fb.FeaturePropertiesT()
output_meta.content.contentPropertiesType = (
_metadata_fb.ContentProperties.FeatureProperties)
output_stats = _metadata_fb.StatsT()
output_stats.max = [1.0]
output_stats.min = [0.0]
output_meta.stats = output_stats
label_file = _metadata_fb.AssociatedFileT()
label_file.name = os.path.basename(self.associated_files[0])
label_file.description = ("Labels for the categories that the model can "
"classify.")
label_file.type = _metadata_fb.AssociatedFileType.TENSOR_AXIS_LABELS
output_meta.associatedFiles = [label_file]
# Creates subgraph info.
subgraph = _metadata_fb.SubGraphMetadataT()
subgraph.inputTensorMetadata = [input_meta]
subgraph.outputTensorMetadata = [output_meta]
model_meta.subgraphMetadata = [subgraph]
b = flatbuffers.Builder(0)
b.Finish(
model_meta.Pack(b),
_metadata.MetadataPopulator.METADATA_FILE_IDENTIFIER)
self.metadata_buf = b.Output()
def forward(self):
builder = flatbuffers.Builder(64)
# construct MessageBody
ppx_Run.RunStart(builder)
message_body = ppx_Run.RunEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(builder,
ppx_MessageBody.MessageBody().Run)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
while True:
reply = self._requester.receive_reply()
message_body = self._get_message_body(reply)
if isinstance(message_body, ppx_RunResult.RunResult):
result = self._protocol_tensor_to_variable(
message_body.Result())
return result
elif isinstance(message_body, ppx_Sample.Sample):
address = message_body.Address().decode('utf-8')
name = message_body.Name().decode('utf-8')
if name == '':
name = None
control = bool(message_body.Control())
replace = bool(message_body.Replace())
distribution_type = message_body.DistributionType()
if distribution_type == ppx_Distribution.Distribution(
).Uniform:
uniform = ppx_Uniform.Uniform()
uniform.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
low = self._protocol_tensor_to_variable(uniform.Low())
high = self._protocol_tensor_to_variable(uniform.High())
dist = Uniform(low, high)
elif distribution_type == ppx_Distribution.Distribution(
).Normal:
normal = ppx_Normal.Normal()
normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(normal.Mean())
stddev = self._protocol_tensor_to_variable(normal.Stddev())
dist = Normal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Categorical:
categorical = ppx_Categorical.Categorical()
categorical.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
probs = self._protocol_tensor_to_variable(
categorical.Probs())
dist = Categorical(probs)
elif distribution_type == ppx_Distribution.Distribution(
).Poisson:
poisson = ppx_Poisson.Poisson()
poisson.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(poisson.Rate())
dist = Poisson(rate)
else:
raise RuntimeError(
'ppx (Python): Sample from an unexpected distribution requested.'
)
result = state.sample(distribution=dist,
control=control,
replace=replace,
name=name,
address=address)
builder = flatbuffers.Builder(64)
result = self._variable_to_protocol_tensor(builder, result)
ppx_SampleResult.SampleResultStart(builder)
ppx_SampleResult.SampleResultAddResult(builder, result)
message_body = ppx_SampleResult.SampleResultEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(
builder,
ppx_MessageBody.MessageBody().SampleResult)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
elif isinstance(message_body, ppx_Observe.Observe):
address = message_body.Address().decode('utf-8')
name = message_body.Name().decode('utf-8')
if name == '':
name = None
value = self._protocol_tensor_to_variable(message_body.Value())
distribution_type = message_body.DistributionType()
if distribution_type == ppx_Distribution.Distribution().NONE:
dist = None
elif distribution_type == ppx_Distribution.Distribution(
).Uniform:
uniform = ppx_Uniform.Uniform()
uniform.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
low = self._protocol_tensor_to_variable(uniform.Low())
high = self._protocol_tensor_to_variable(uniform.High())
dist = Uniform(low, high)
elif distribution_type == ppx_Distribution.Distribution(
).Normal:
normal = ppx_Normal.Normal()
normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(normal.Mean())
stddev = self._protocol_tensor_to_variable(normal.Stddev())
dist = Normal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Categorical:
categorical = ppx_Categorical.Categorical()
categorical.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
probs = self._protocol_tensor_to_variable(
categorical.Probs())
dist = Categorical(probs)
elif distribution_type == ppx_Distribution.Distribution(
).Poisson:
poisson = ppx_Poisson.Poisson()
poisson.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(poisson.Rate())
dist = Poisson(rate)
else:
raise RuntimeError(
'ppx (Python): Sample from an unexpected distribution requested: {}'
.format(distribution_type))
state.observe(distribution=dist,
value=value,
name=name,
address=address)
builder = flatbuffers.Builder(64)
ppx_ObserveResult.ObserveResultStart(builder)
message_body = ppx_ObserveResult.ObserveResultEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(
builder,
ppx_MessageBody.MessageBody().ObserveResult)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
elif isinstance(message_body, ppx_Reset.Reset):
raise RuntimeError(
'ppx (Python): Received a reset request. Protocol out of sync.'
)
else:
raise RuntimeError(
'ppx (Python): Received unexpected message.')
def _serialize_value(self, value):
builder = flatbuffers.Builder(0)
obj = self._build(value, builder)
builder.Finish(obj)
buf = builder.Output()
return bytes(buf)
#params = input("enter params : ")
params = ''
print(my_url + url)
req = requests.get(my_url + url, params=params)
print(req.content)
if selected == '3':
url = 'user/friend'
userid = input('enter userid : ')
targetidx = input('enter targetidx : ')
state = input('enter state : ')
builder = flatbuffers.Builder(128)
proto_headerStart(builder)
proto_headerAddUseridx(builder, int(userid))
header_offset = proto_headerEnd(builder)
proto_responseFriendRequestStart(builder)
proto_responseFriendRequestAddHeader(builder, header_offset)
proto_responseFriendRequestAddTargetidx(builder, int(targetidx))
proto_responseFriendRequestAddState(builder, int(state))
root_offset = proto_responseFriendRequestEnd(builder)
builder.Finish(root_offset)
#res = make_response(builder.Output())
#res.headers['Content-Type'] = 'application/octet-stream'
data = builder.Output()
def convertModelNode(convertContext, data):
"""
Converts a ModelNode. The data map is expected to contain the following elements:
- embeddedResources: optional set of resources to embed with the node. This is a map containing
the elements as expected by SceneResourcesConvert.convertSceneResources().
- modelGeometry: array of model geometry. Each element of the array has the following members:
- type: the name of the geometry type, such as "obj" or "gltf".
- path: the path to the geometry.
- vertexFormat: array of vertex attributes defining the vertex format. Each element of the
array has the following members:
- attrib: the attribute. This can either be an enum value from dsVertexAttrib, removing
the type prefix, or the integer for the attribute.
- format: the attribute format. See the dsGfxFormat enum for values, removing the type
prefix. Only the "standard" formats may be used.
- decoration: the decoration for the format. See the dsGfxFormat enum for values,
removing the type prefix. Only the decorator values may be used.
- indexSize: the size of the index in bytes. This must be either 2 or 4. If not set, no
indices will be produced.
- transforms: optional array of transforms to perform on the vertex values. Each element of
the array has the following members:
- attrib: the attribute, matching one of the attributes in vertexFormat.
- transform: transform to apply on the attribute. Valid values are:
- Identity: leaves the values un-transformed.
- Bounds: normalizes the values based on the original value's bounds
- UNormToSNorm: converts UNorm values to SNorm values.
- SNormToUNorm: converts SNorm values to UNorm values.
- drawInfo: array of definitions for drawing components of the geometry. Each element of the
array has the following members:
- name: the name of the model component. Note that only model components referenced in the
drawInfo array will be included in the final model.
- shader: te name of the shader to draw with.
- material: the name of the material to draw with.
- distanceRange: array of two floats for the minimum and maximum distance to draw at.
Defaults to [0, 3.402823466e38].
- listName The name of the item list to draw the model with.
- models: array of models to draw with manually provided geometry. (i.e. not converted from
the modelGeometry array) Each element of the array has the following members:
- shader: the name of the shader to draw with.
- material: the name of the material to draw with.
- geometry: the name of the geometry to draw.
- distanceRange: array of two floats for the minimum and maximum distance to draw at. Defaults
to [0, 3.402823466e38].
- drawRange: the range of the geometry to draw. This is an object with the following members,
depending on if the geometry is indexed or not:
Indexed geometry:
- indexCount: the number of indices to draw.
- instanceCount: the number of instances to draw. Defaults to 1.
- firstIndex: the first index to draw. Defaults to 0.
- vertexOffset: the offset to apply to each index value. Defaults to 0.
- firstInstance: the first instance to draw. Defaults to 0.
Non-indexed geometry:
- vertexCount: the number of vertices to draw.
- instanceCount: the number of instances to draw. Defaults to 1.
- firstVertex: the first vertex to draw. Defaults to 0.
- firstIstance: the first instance to draw. Defaults to 0.
- primitiveType: the primitive type to draw with. See the dsPrimitiveType enum for values,
removing the type prefix. Defaults to "TriangleList".
- listName The name of the item list to draw the model with.
- extraItemLists: array of extra item list names to add the node to.
- bounds: 2x3 array of float values for the minimum and maximum values for the positions. This
will be automatically calculated from geometry in modelGeometry if unset. Otherwise if unset
the model will have no explicit bounds for culling.
"""
try:
embeddedResources = data.get('embeddedResources', dict())
if not isinstance(embeddedResources, dict):
raise Exception ('ModelNode "embeddedResources" must be an object.')
modelGeometry = data.get('modelGeometry')
if modelGeometry:
models, modelBounds = convertModelNodeGeometry(convertContext, modelGeometry,
embeddedResources)
else:
models = []
modelBounds = None
modelInfoList = data.get('models')
if modelInfoList:
models.extend(convertModelNodeModels(modelInfoList))
extraItemLists = data.get('extraItemLists')
if 'bounds' in data:
modelBounds = data['bounds']
try:
if len(modelBounds) != 2:
raise Exception()
for bound in modelBounds:
if len(bound) != 3:
raise Exception()
for val in bound:
if not isinstance(val, float):
raise Exception()
except:
raise Exception('Invalid model bounds "' + str(modelBounds) + '".')
except (TypeError, ValueError):
raise Exception('ModelNode data must be an object.')
except KeyError as e:
raise Exception('ModelNode data doesn\'t contain element "' + str(e) + '".')
builder = flatbuffers.Builder(0)
if embeddedResources:
embeddedResourcesData = convertSceneResources(convertContext, embeddedResources)
embeddedResourcesOffset = builder.CreateByteVector(embeddedResourcesData)
else:
embeddedResourcesOffset = 0
if extraItemLists:
extraItemListOffsets = []
try:
for item in extraItemLists:
extraItemListOffsets = builder.CreateString(str(item))
except (TypeError, ValueError):
raise Exception('ModelNode "extraItemLists" must be an array of strings.')
ModelNodeStartExtraItemListsVector(builder, len(extraItemListOffsets))
for offset in reversed(extraItemListOffsets):
builder.PrependUOffsetTRelative(offset)
extraItemListsOffset = builder.EndVector(len(extraItemListOffsets))
else:
extraItemListsOffset = 0
modelOffsets = []
for model in models:
shaderOffset = builder.CreateString(model.shader)
materialOffset = builder.CreateString(model.material)
geometryOffset = builder.CreateString(model.geometry)
distanceRangeOffset = CreateVector2f(builder, model.distanceRange[0],
model.distanceRange[1])
drawRange = model.drawRange
if drawRange.rangeType == ModelDrawRange.DrawIndexedRange:
DrawIndexedRangeStart(builder)
DrawIndexedRangeAddIndexCount(builder, drawRange.indexCount)
DrawIndexedRangeAddInstanceCount(builder, drawRange.instanceCount)
DrawIndexedRangeAddFirstIndex(builder, drawRange.firstIndex)
DrawIndexedRangeAddVertexOffset(builder, drawRange.vertexOffset)
DrawIndexedRangeAddFirstInstance(builder, drawRange.firstInstance)
drawRangeOffset = DrawIndexedRangeEnd(builder)
else:
DrawRangeStart(builder)
DrawRangeAddVertexCount(builder, drawRange.vertexCount)
DrawRangeAddInstanceCount(builder, drawRange.instanceCount)
DrawRangeAddFirstVertex(builder, drawRange.firstVertex)
DrawRangeAddFirstInstance(builder, drawRange.firstInstance)
drawRangeOffset = DrawRangeEnd(builder)
listNameOffset = builder.CreateString(model.listName)
ModelInfoStart(builder)
ModelInfoAddShader(builder, shaderOffset)
ModelInfoAddMaterial(builder, materialOffset)
ModelInfoAddGeometry(builder, geometryOffset)
ModelInfoAddDistanceRange(builder, distanceRangeOffset)
ModelInfoAddDrawRangeType(builder, drawRange.rangeType)
ModelInfoAddDrawRange(builder, drawRangeOffset)
ModelInfoAddPrimitiveType(builder, model.primitiveType)
ModelInfoAddListName(builder, listNameOffset)
modelOffsets.append(ModelInfoEnd(builder))
ModelNodeStartModelsVector(builder, len(modelOffsets))
for offset in reversed(modelOffsets):
builder.PrependUOffsetTRelative(offset)
modelsOffset = builder.EndVector(len(modelOffsets))
if modelBounds:
center = []
halfExtents = []
for i in range(0, 3):
center.append((modelBounds[0][i] + modelBounds[1][i])/2)
halfExtents.append((modelBounds[1][i] - modelBounds[0][i])/2)
boundsOffset = CreateOrientedBox3f(builder, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0,
center[0], center[1], center[2], halfExtents[0], halfExtents[1], halfExtents[2])
else:
boundsOffset = 0
ModelNodeStart(builder)
ModelNodeAddEmbeddedResources(builder, embeddedResourcesOffset)
ModelNodeAddExtraItemLists(builder, extraItemListsOffset)
ModelNodeAddModels(builder, modelsOffset)
ModelNodeAddBounds(builder, boundsOffset)
builder.Finish(ModelNodeEnd(builder))
return builder.Output()
def make_postprocess_model(self,
max_detections=10,
detections_per_class=100,
max_classes_per_detection=1,
use_regular_nms=True,
nms_score_threshold=0.3,
nms_iou_threshold=0.6,
num_classes=90,
x_scale=10.0,
y_scale=10.0,
w_scale=5.0,
h_scale=5.0):
"""Returns the bytes of a tflite model containing a single TFLite_Detection_PostProcess op"""
builder = flatbuffers.Builder(1024)
# op_code
custom_code = builder.CreateString("TFLite_Detection_PostProcess")
OperatorCode.OperatorCodeStart(builder)
OperatorCode.OperatorCodeAddDeprecatedBuiltinCode(
builder, BuiltinOperator.CUSTOM)
OperatorCode.OperatorCodeAddCustomCode(builder, custom_code)
OperatorCode.OperatorCodeAddBuiltinCode(builder,
BuiltinOperator.CUSTOM)
op_code = OperatorCode.OperatorCodeEnd(builder)
# op_codes
Model.ModelStartOperatorCodesVector(builder, 1)
builder.PrependUOffsetTRelative(op_code)
op_codes = builder.EndVector(1)
# Make tensors
# [names, shape, type tensors]
ts = []
inputs_info = [('box_encodings', [-1, -1, 4]),
('class_predictions', [-1, -1, -1]),
('anchors', [-1, 4])]
outputs_info = [('detection_boxes', [-1, -1, 4]),
('detection_classes', [-1, -1]),
('detection_scores', [-1, -1]),
('num_detections', [-1])]
for name_info, shape_info in inputs_info + outputs_info:
name = builder.CreateString(name_info)
shape = builder.CreateNumpyVector(
np.maximum(np.array(shape_info, np.int32), 1))
shape_signature = builder.CreateNumpyVector(
np.array(shape_info, np.int32))
Tensor.TensorStart(builder)
Tensor.TensorAddShape(builder, shape)
Tensor.TensorAddType(builder, TensorType.FLOAT32)
Tensor.TensorAddName(builder, name)
Tensor.TensorAddShapeSignature(builder, shape_signature)
ts.append(Tensor.TensorEnd(builder))
SubGraph.SubGraphStartTensorsVector(builder, len(ts))
for tensor in reversed(ts):
builder.PrependUOffsetTRelative(tensor)
tensors = builder.EndVector(len(ts))
# inputs
SubGraph.SubGraphStartInputsVector(builder, 3)
for inp in reversed([0, 1, 2]):
builder.PrependInt32(inp)
inputs = builder.EndVector(3)
# outputs
SubGraph.SubGraphStartOutputsVector(builder, 4)
for out in reversed([3, 4, 5, 6]):
builder.PrependInt32(out)
outputs = builder.EndVector(4)
flexbuffer = \
b'y_scale\x00nms_score_threshold\x00max_detections\x00x_scale\x00w_scale\x00nms_iou_threshold' \
b'\x00use_regular_nms\x00h_scale\x00max_classes_per_detection\x00num_classes\x00detections_per_class' \
b'\x00\x0b\x16E>\x88j\x9e([v\x7f\xab\x0b\x00\x00\x00\x01\x00\x00\x00\x0b\x00\x00\x00*attr4**attr7*' \
b'*attr10**attr9**attr1**attr2**attr3**attr11*\x00\x00\x00*attr8**attr5**attr6*\x06\x0e\x06\x06\x0e' \
b'\x0e\x06j\x0e\x0e\x0e7&\x01'
flexbuffer = flexbuffer.replace(b'*attr1*',
struct.pack('<f', nms_iou_threshold))
flexbuffer = flexbuffer.replace(b'*attr2*',
struct.pack('<f', nms_score_threshold))
flexbuffer = flexbuffer.replace(b'*attr3*',
struct.pack('<i', num_classes))
flexbuffer = flexbuffer.replace(
b'*attr4*', struct.pack('<i', detections_per_class))
flexbuffer = flexbuffer.replace(b'*attr5*', struct.pack('<f', x_scale))
flexbuffer = flexbuffer.replace(b'*attr6*', struct.pack('<f', y_scale))
flexbuffer = flexbuffer.replace(b'*attr7*', struct.pack('<f', h_scale))
flexbuffer = flexbuffer.replace(b'*attr8*', struct.pack('<f', w_scale))
flexbuffer = flexbuffer.replace(b'*attr9*',
struct.pack('<i', max_detections))
flexbuffer = flexbuffer.replace(
b'*attr10*', struct.pack('<i', max_classes_per_detection))
flexbuffer = flexbuffer.replace(b'*attr11*',
struct.pack('<b', use_regular_nms))
custom_options = builder.CreateNumpyVector(
np.array(bytearray(flexbuffer)))
# operator
Operator.OperatorStart(builder)
Operator.OperatorAddOpcodeIndex(builder, 0)
Operator.OperatorAddInputs(builder, inputs)
Operator.OperatorAddOutputs(builder, outputs)
Operator.OperatorAddCustomOptions(builder, custom_options)
Operator.OperatorAddCustomOptionsFormat(
builder, CustomOptionsFormat.FLEXBUFFERS)
operator = Operator.OperatorEnd(builder)
# operators
SubGraph.SubGraphStartOperatorsVector(builder, 1)
builder.PrependUOffsetTRelative(operator)
operators = builder.EndVector(1)
# subgraph
SubGraph.SubGraphStart(builder)
SubGraph.SubGraphAddTensors(builder, tensors)
SubGraph.SubGraphAddInputs(builder, inputs)
SubGraph.SubGraphAddOutputs(builder, outputs)
SubGraph.SubGraphAddOperators(builder, operators)
subgraph = SubGraph.SubGraphEnd(builder)
# subgraphs
Model.ModelStartSubgraphsVector(builder, 1)
builder.PrependUOffsetTRelative(subgraph)
subgraphs = builder.EndVector(1)
description = builder.CreateString("Model for tflite testing")
Buffer.BufferStartDataVector(builder, 0)
data = builder.EndVector(0)
Buffer.BufferStart(builder)
Buffer.BufferAddData(builder, data)
buffer = Buffer.BufferEnd(builder)
Model.ModelStartBuffersVector(builder, 1)
builder.PrependUOffsetTRelative(buffer)
buffers = builder.EndVector(1)
# model
Model.ModelStart(builder)
Model.ModelAddVersion(builder, 3)
Model.ModelAddOperatorCodes(builder, op_codes)
Model.ModelAddSubgraphs(builder, subgraphs)
Model.ModelAddDescription(builder, description)
Model.ModelAddBuffers(builder, buffers)
model = Model.ModelEnd(builder)
builder.Finish(model, b"TFL3")
return builder.Output()
def create_from_metadata(
cls,
model_buffer: bytearray,
model_metadata: Optional[_metadata_fb.ModelMetadataT] = None,
input_metadata: Optional[List[_metadata_fb.TensorMetadataT]] = None,
output_metadata: Optional[List[_metadata_fb.TensorMetadataT]] = None,
associated_files: Optional[List[str]] = None,
input_process_units: Optional[List[_metadata_fb.ProcessUnitT]] = None,
output_process_units: Optional[List[_metadata_fb.ProcessUnitT]] = None):
"""Creates MetadataWriter based on the metadata Flatbuffers Python Objects.
Args:
model_buffer: valid buffer of the model file.
model_metadata: general model metadata [1]. The subgraph_metadata will be
refreshed with input_metadata and output_metadata.
input_metadata: a list of metadata of the input tensors [2].
output_metadata: a list of metadata of the output tensors [3].
associated_files: path to the associated files to be populated.
input_process_units: a lits of metadata of the input process units [4].
output_process_units: a lits of metadata of the output process units [5].
[1]:
https://github.com/tensorflow/tflite-support/blob/b80289c4cd1224d0e1836c7654e82f070f9eefaa/tensorflow_lite_support/metadata/metadata_schema.fbs#L640-L681
[2]:
https://github.com/tensorflow/tflite-support/blob/b80289c4cd1224d0e1836c7654e82f070f9eefaa/tensorflow_lite_support/metadata/metadata_schema.fbs#L590
[3]:
https://github.com/tensorflow/tflite-support/blob/b80289c4cd1224d0e1836c7654e82f070f9eefaa/tensorflow_lite_support/metadata/metadata_schema.fbs#L599
[4]:
https://github.com/tensorflow/tflite-support/blob/b5cc57c74f7990d8bc055795dfe8d50267064a57/tensorflow_lite_support/metadata/metadata_schema.fbs#L646
[5]:
https://github.com/tensorflow/tflite-support/blob/b5cc57c74f7990d8bc055795dfe8d50267064a57/tensorflow_lite_support/metadata/metadata_schema.fbs#L650
Returns:
A MetadataWriter Object.
"""
# Create empty tensor metadata when input_metadata/output_metadata are None
# to bypass MetadataPopulator verification.
if not input_metadata:
model = _schema_fb.Model.GetRootAsModel(model_buffer, 0)
num_input_tensors = model.Subgraphs(0).InputsLength()
input_metadata = [
_metadata_fb.TensorMetadataT() for i in range(num_input_tensors)
]
if not output_metadata:
model = _schema_fb.Model.GetRootAsModel(model_buffer, 0)
num_output_tensors = model.Subgraphs(0).OutputsLength()
output_metadata = [
_metadata_fb.TensorMetadataT() for i in range(num_output_tensors)
]
_fill_default_tensor_names(
input_metadata, writer_utils.get_input_tensor_names(model_buffer))
_fill_default_tensor_names(
output_metadata, writer_utils.get_output_tensor_names(model_buffer))
subgraph_metadata = _metadata_fb.SubGraphMetadataT()
subgraph_metadata.inputTensorMetadata = input_metadata
subgraph_metadata.outputTensorMetadata = output_metadata
subgraph_metadata.inputProcessUnits = input_process_units
subgraph_metadata.outputProcessUnits = output_process_units
if model_metadata is None:
model_metadata = _metadata_fb.ModelMetadataT()
model_metadata.subgraphMetadata = [subgraph_metadata]
b = flatbuffers.Builder(0)
b.Finish(
model_metadata.Pack(b),
_metadata.MetadataPopulator.METADATA_FILE_IDENTIFIER)
return cls(model_buffer, b.Output(), associated_files)
def serialize_tensor(ndarray):
size = ndarray.nbytes + 100
builder = flatbuffers.Builder(size)
tensor = make_tensor(builder, ndarray)
return output(builder, tensor)
def reply_proposal(self, proposal=None):
# allocate buffer
builder = flatbuffers.Builder(64)
if proposal is None:
infcomp.protocol.ProposalReply.ProposalReplyStart(builder)
infcomp.protocol.ProposalReply.ProposalReplyAddSuccess(
builder, False)
message_body = infcomp.protocol.ProposalReply.ProposalReplyEnd(
builder)
else:
if isinstance(proposal, UniformDiscrete):
# construct probabilities
proposal_probabilities = Tensor_to_NDArray(
builder, proposal.proposal_probabilities)
# construct UniformDiscrete
infcomp.protocol.UniformDiscrete.UniformDiscreteStart(builder)
infcomp.protocol.UniformDiscrete.UniformDiscreteAddPriorMin(
builder, proposal.prior_min)
infcomp.protocol.UniformDiscrete.UniformDiscreteAddPriorSize(
builder, proposal.prior_size)
infcomp.protocol.UniformDiscrete.UniformDiscreteAddProposalProbabilities(
builder, proposal_probabilities)
distribution = infcomp.protocol.UniformDiscrete.UniformDiscreteEnd(
builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).UniformDiscrete
elif isinstance(proposal, Normal):
# construct Normal
infcomp.protocol.Normal.NormalStart(builder)
infcomp.protocol.Normal.NormalAddPriorMean(
builder, proposal.prior_mean)
infcomp.protocol.Normal.NormalAddPriorStd(
builder, proposal.prior_std)
infcomp.protocol.Normal.NormalAddProposalMean(
builder, proposal.proposal_mean)
infcomp.protocol.Normal.NormalAddProposalStd(
builder, proposal.proposal_std)
distribution = infcomp.protocol.Normal.NormalEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Normal
elif isinstance(proposal, Flip):
# construct Flip
infcomp.protocol.Flip.FlipStart(builder)
infcomp.protocol.Flip.FlipAddProposalProbability(
builder, proposal.proposal_probability)
distribution = infcomp.protocol.Flip.FlipEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Flip
elif isinstance(proposal, Discrete):
# construct probabilities
proposal_probabilities = Tensor_to_NDArray(
builder, proposal.proposal_probabilities)
# construct Discrete
infcomp.protocol.Discrete.DiscreteStart(builder)
infcomp.protocol.Discrete.DiscreteAddPriorSize(
builder, proposal.prior_size)
infcomp.protocol.Discrete.DiscreteAddProposalProbabilities(
builder, proposal_probabilities)
distribution = infcomp.protocol.Discrete.DiscreteEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Discrete
elif isinstance(proposal, Categorical):
# construct probabilities
proposal_probabilities = Tensor_to_NDArray(
builder, proposal.proposal_probabilities)
# construct Categorical
infcomp.protocol.Categorical.CategoricalStart(builder)
infcomp.protocol.Categorical.CategoricalAddPriorSize(
builder, proposal.prior_size)
infcomp.protocol.Categorical.CategoricalAddProposalProbabilities(
builder, proposal_probabilities)
distribution = infcomp.protocol.Categorical.CategoricalEnd(
builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Categorical
elif isinstance(proposal, UniformContinuous):
# construct UniformContinuous
infcomp.protocol.UniformContinuous.UniformContinuousStart(
builder)
infcomp.protocol.UniformContinuous.UniformContinuousAddPriorMin(
builder, proposal.prior_min)
infcomp.protocol.UniformContinuous.UniformContinuousAddPriorMax(
builder, proposal.prior_max)
infcomp.protocol.UniformContinuous.UniformContinuousAddProposalMode(
builder, proposal.proposal_mode)
infcomp.protocol.UniformContinuous.UniformContinuousAddProposalCertainty(
builder, proposal.proposal_certainty)
distribution = infcomp.protocol.UniformContinuous.UniformContinuousEnd(
builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).UniformContinuous
elif isinstance(proposal, UniformContinuousAlt):
# construct proposal parameters
# print('means, stds, coeffs')
# print(proposal.proposal_means)
# print(proposal.proposal_stds)
# print(proposal.proposal_coeffs)
proposal_means = Tensor_to_NDArray(builder,
proposal.proposal_means)
proposal_stds = Tensor_to_NDArray(builder,
proposal.proposal_stds)
proposal_coeffs = Tensor_to_NDArray(builder,
proposal.proposal_coeffs)
# construct UniformContinuousAlt
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltStart(
builder)
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltAddPriorMin(
builder, proposal.prior_min)
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltAddPriorMax(
builder, proposal.prior_max)
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltAddProposalMeans(
builder, proposal_means)
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltAddProposalStds(
builder, proposal_stds)
infcomp.protocol.UniformContinuousAlt.UniformContinuousAltAddProposalCoeffs(
builder, proposal_coeffs)
distribution = infcomp.protocol.UniformContinuousAlt.UniformContinuousAltEnd(
builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).UniformContinuousAlt
elif isinstance(proposal, Laplace):
# construct Laplace
infcomp.protocol.Laplace.LaplaceStart(builder)
infcomp.protocol.Laplace.LaplaceAddPriorLocation(
builder, proposal.prior_location)
infcomp.protocol.Laplace.LaplaceAddPriorScale(
builder, proposal.prior_scale)
infcomp.protocol.Laplace.LaplaceAddProposalLocation(
builder, proposal.proposal_location)
infcomp.protocol.Laplace.LaplaceAddProposalScale(
builder, proposal.proposal_scale)
distribution = infcomp.protocol.Laplace.LaplaceEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Laplace
elif isinstance(proposal, Gamma):
# construct Gamma
infcomp.protocol.Gamma.GammaStart(builder)
infcomp.protocol.Gamma.GammaAddProposalLocation(
builder, proposal.proposal_location)
infcomp.protocol.Gamma.GammaAddProposalScale(
builder, proposal.proposal_scale)
distribution = infcomp.protocol.Gamma.GammaEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Gamma
elif isinstance(proposal, Beta):
# construct Beta
infcomp.protocol.Beta.BetaStart(builder)
infcomp.protocol.Beta.BetaAddProposalMode(
builder, proposal.proposal_mode)
infcomp.protocol.Beta.BetaAddProposalCertainty(
builder, proposal.proposal_certainty)
distribution = infcomp.protocol.Beta.BetaEnd(builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).Beta
elif isinstance(proposal, MultivariateNormal):
# construct prior_mean, prior_cov, proposal_mean, proposal_vars
prior_mean = Tensor_to_NDArray(builder, proposal.prior_mean)
prior_cov = Tensor_to_NDArray(builder, proposal.prior_cov)
proposal_mean = Tensor_to_NDArray(builder,
proposal.proposal_mean)
proposal_vars = Tensor_to_NDArray(builder,
proposal.proposal_vars)
# construct MultivariateNormal
infcomp.protocol.MultivariateNormal.MultivariateNormalStart(
builder)
infcomp.protocol.MultivariateNormal.MultivariateNormalAddPriorMean(
builder, prior_mean)
infcomp.protocol.MultivariateNormal.MultivariateNormalAddPriorCov(
builder, prior_cov)
infcomp.protocol.MultivariateNormal.MultivariateNormalAddProposalMean(
builder, proposal_mean)
infcomp.protocol.MultivariateNormal.MultivariateNormalAddProposalVars(
builder, proposal_vars)
distribution = infcomp.protocol.MultivariateNormal.MultivariateNormalEnd(
builder)
distribution_type = infcomp.protocol.Distribution.Distribution(
).MultivariateNormal
else:
util.logger.log_error(
'reply_proposal: Unsupported proposal distribution: {0}'.
format(proposal))
# construct message body (ProposalReply)
infcomp.protocol.ProposalReply.ProposalReplyStart(builder)
infcomp.protocol.ProposalReply.ProposalReplyAddSuccess(
builder, True)
infcomp.protocol.ProposalReply.ProposalReplyAddDistributionType(
builder, distribution_type)
infcomp.protocol.ProposalReply.ProposalReplyAddDistribution(
builder, distribution)
message_body = infcomp.protocol.ProposalReply.ProposalReplyEnd(
builder)
# construct message
infcomp.protocol.Message.MessageStart(builder)
infcomp.protocol.Message.MessageAddBodyType(
builder,
infcomp.protocol.MessageBody.MessageBody().ProposalReply)
infcomp.protocol.Message.MessageAddBody(builder, message_body)
message = infcomp.protocol.Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._replier.send_reply(message)
def serialize_metadata_response(metadata):
builder = flatbuffers.Builder(4096)
r = make_metadata_response(builder, metadata)
return output(builder, r)
def serialize_metadata_request():
builder = flatbuffers.Builder(1024)
r = make_metadata_request(builder)
return output(builder, r)
import flatbuffers as fb
from python.proto import Order
# encode
b = fb.Builder(0)
Order.OrderStart(b)
Order.OrderAddId(b, 100)
o = Order.OrderEnd(b)
b.Finish(o)
out = bytes(b.Output())
# s = ""
# for i in out:
# s += "0x{:02x}, ".format(i)
# print(s)
# decode self
order = Order.Order.GetRootAsOrder(out, 0)
print(order.Id())
# decode C-encoded Order
c_enc = b'\x08\x00\x00\x00\x76\x30\x30\x31\xf8\xff\xff\xff\x64\x00\x00\x00\x06\x00\x08\x00\x04\x00'
order = Order.Order.GetRootAsOrder(out, 0)
print(order.Id())
def convertVectorShaders(convertContext, data):
"""
Converts vector shaders used in a scene. The data map is expected to contain the following
elements:
- modules: array of versioned shader modules. The appropriate model based on the graphics API
version being used will be chosen at runtime. Each element of the array has the following
members:
- version: the version of the shader as a standard config. (e.g. glsl-4.1, spirv-1.0)
- module: path to the shader module or base64 encoded data prefixed with "base64:". The
module is expected to have been compiled with Modular Shader Language (MSL).
- output: the path to the location to copy the shader module to. This can be omitted to
embed the shader module directly.
- outputRelativeDir: the directory relative to output path. This will be removed from the
path before adding the reference.
- resourceType: the resource type. See the dsFileResourceType for values, removing the type
prefix. Defaults to "Embedded".
- extraElements: list of extra meterial elements to add for the material description. Each
element of the array has the following members:
- name: the name of the element.
- type: the type of the element. See dsMaterialType enum for values, removing the type prefix.
- count: the number of array elements. If 0 or omitted, this is not an array.
- binding: the binding type for the element. See the dsMaterialBinding enum for values,
removing the type prefix. This is only used for texture, image, buffer, and shader variable
group types.
- shaderVariableGroupDesc: the name of the shader variable group description when the type
is a shader variable group.
- materialDesc: the name of the material description to register. This can be referenced by
other objects, such as creating materials used for drawing vector images.
- fillColor: the name of the shader for filling with a solid color. Defaults to
"dsVectorFillColor".
- fillLinearGradient: the name of the shader for filling with a linear gradient. Defaults to
"dsVectorFillLinearGradient".
- fillRadialGradient: the name of the shader for filling with a radial gradient. Defaults to
"dsVectorFillRadialGradient".
- line: the name of the shader for a line with a color or gradient. Defaults to
"dsVectorLine".
- image: the name of the shader for a texture applied as an image. Defaults to "dsVectorImage".
- textColor: name of the shader for standard single-color text. Defaults to "dsVectorTextColor".
- textColorOutline: name of the shader for standard single-color text with a single-colored
outline. Defaults to "dsVectorTextColorOutline".
- textGradient: name of the shader for text using a gradient. Defaults to
"dsVectorTextGradient".
- textGradientOutline: name of the shader for text with an outline using a gradient. Defaults to
"dsVectorTextGradientOutline".
"""
def createOptionalString(builder, string):
if string:
return builder.CreateString(string)
else:
return 0
builder = flatbuffers.Builder(0)
try:
modules = data['modules']
versionedModules = []
try:
for versionedModuleData in modules:
version = str(versionedModuleData['version'])
moduleStr = str(versionedModuleData['module'])
try:
modulePath, moduleContents = readDataOrPath(moduleStr)
except TypeError:
raise Exception(
'VectorShaders shader module "module" uses incorrect base64 encoding.'
)
dataType, dataOffset = convertFileOrData(
builder, modulePath, moduleContents,
versionedModuleData.get('output'),
versionedModuleData.get('outputRelativeDir'),
versionedModuleData.get('resourceType'))
versionedModules.append((version, dataType, dataOffset))
except KeyError as e:
raise Exception(
'Versioned shader module data doesn\'t contain element ' +
str(e) + '.')
except (TypeError, ValueError):
raise Exception(
'Versioned shader module list must be an array of objects.')
extraElementsData = data.get('extraElements', [])
extraElements = []
try:
for elementData in extraElementsData:
element = Object()
element.name = str(elementData['name'])
typeStr = str(elementData['type'])
try:
element.type = getattr(MaterialType, typeStr)
except AttributeError:
raise Exception('Invalid material type "' + typeStr + '".')
countValue = elementData.get('count', 0)
try:
element.count = int(countValue)
if element.count < 0:
raise Exception(
) # Common error handling in except block.
except:
raise Exception('Invalid vector shader element count "' +
str(countValue) + '".')
bindingStr = str(elementData['binding'])
try:
element.binding = getattr(MaterialBinding, bindingStr)
except AttributeError:
raise Exception('Invalid material binding "' + bindingStr +
'".')
element.shaderVariableGroupDesc = str(
elementData.get('shaderVariableGroupDesc', ''))
extraElements.append(element)
except KeyError as e:
raise Exception(
'Versioned shader module data doesn\'t contain element ' +
str(e) + '.')
except (TypeError, ValueError):
raise Exception(
'Versioned shader module list must be an array of objects.')
materialDescName = str(data['materialDesc'])
fillColor = str(data.get('fillColor', ''))
fillLinearGradient = str(data.get('fillLinearGradient', ''))
fillRadialGradient = str(data.get('fillRadialGradient', ''))
line = str(data.get('line', ''))
image = str(data.get('image', ''))
textColor = str(data.get('textColor', ''))
textColorOutline = str(data.get('textColorOutline', ''))
textGradient = str(data.get('textGradient', ''))
textGradientOutline = str(data.get('textGradientOutline', ''))
except KeyError as e:
raise Exception('VectorShaders doesn\'t contain element ' + str(e) +
'.')
except (AttributeError, TypeError, ValueError):
raise Exception('VectorShaders must be an object.')
modulesOffsets = []
for version, dataType, dataOffset in versionedModules:
versionOffset = builder.CreateString(version)
VersionedShaderModule.Start(builder)
VersionedShaderModule.AddVersion(builder, versionOffset)
VersionedShaderModule.AddDataType(builder, dataType)
VersionedShaderModule.AddData(builder, dataOffset)
modulesOffsets.append(VersionedShaderModule.End(builder))
VectorShaders.StartModulesVector(builder, len(modulesOffsets))
for offset in reversed(modulesOffsets):
builder.PrependUOffsetTRelative(offset)
modulesOffset = builder.EndVector()
extraElementsOffsets = []
for element in extraElements:
elementNameOffset = builder.CreateString(element.name)
shaderVariableGroupDescOffset = createOptionalString(
builder, element.shaderVariableGroupDesc)
MaterialElement.Start(builder)
MaterialElement.AddName(builder, elementNameOffset)
MaterialElement.AddType(builder, element.type)
MaterialElement.AddCount(builder, element.count)
MaterialElement.AddBinding(builder, element.binding)
MaterialElement.AddShaderVariableGroupDesc(
builder, shaderVariableGroupDescOffset)
extraElementsOffsets.append(MaterialElement.End(builder))
VectorShaders.StartExtraElementsVector(builder, len(extraElementsOffsets))
for offset in reversed(extraElementsOffsets):
builder.PrependUOffsetTRelative(offset)
extraElementsOffset = builder.EndVector()
materialDescNameOffset = builder.CreateString(materialDescName)
fillColorOffset = createOptionalString(builder, fillColor)
fillLinearGradientOffset = createOptionalString(builder,
fillLinearGradient)
fillRadialGradientOffset = createOptionalString(builder,
fillRadialGradient)
lineOffset = createOptionalString(builder, line)
imageOffset = createOptionalString(builder, image)
textColorOffset = createOptionalString(builder, textColor)
textColorOutlineOffset = createOptionalString(builder, textColorOutline)
textGradientOffset = createOptionalString(builder, textGradient)
textGradientOutlineOffset = createOptionalString(builder,
textGradientOutline)
VectorShaders.Start(builder)
VectorShaders.AddModules(builder, modulesOffset)
VectorShaders.AddExtraElements(builder, extraElementsOffset)
VectorShaders.AddMaterialDesc(builder, materialDescNameOffset)
VectorShaders.AddFillColor(builder, fillColorOffset)
VectorShaders.AddFillLinearGradient(builder, fillLinearGradientOffset)
VectorShaders.AddFillRadialGradient(builder, fillRadialGradientOffset)
VectorShaders.AddLine(builder, lineOffset)
VectorShaders.AddImage(builder, imageOffset)
VectorShaders.AddTextColor(builder, textColorOffset)
VectorShaders.AddTextColorOutline(builder, textColorOutlineOffset)
VectorShaders.AddTextGradient(builder, textGradientOffset)
VectorShaders.AddTextGradientOutline(builder, textGradientOutlineOffset)
builder.Finish(VectorShaders.End(builder))
return builder.Output()
def convertVectorImage(convertContext, data):
"""
Converts a VectorImage. The data map is expected to contain the following elements:
- image: path to the vector image or base64 encoded data prefixed with "base64:".
- output: the path to the output the vector image. This can be omitted if the vector image is
embedded.
- outputRelativeDir: the directory relative to output path. This will be removed from the path
before adding the reference.
- resourceType: the resource type. See the dsFileResourceType for values, removing the type
prefix. Defaults to "Embedded".
- targetSize: the target size of the vector image for the tessellation quality as an array of
two floats. Defaults to the original image size.
- sharedMaterials: the name of the vector material set for shared material data.
- vectorShaders: the name of the vector material set for shared material data.
- vectorResources: list of strings for the names of the vector resources to get textures and
fonts from.
- srgb: bool for whether or not the embedded materials should be treated as sRGB and converted
to linear when drawing. Defaults to false.
"""
builder = flatbuffers.Builder(0)
try:
imageStr = str(data['image'])
try:
imagePath, imageContents = readDataOrPath(imageStr)
except TypeError:
raise Exception('VectorImage "data" uses incorrect base64 encoding.')
imageType, imageOffset = convertFileOrData(builder, imagePath, imageContents,
data.get('output'), data.get('outputRelativeDir'), data.get('resourceType'))
size = data.get('targetSize')
if size:
try:
if len(size) != 2:
raise Exception() # Common error handling in except block.
size[0] = float(size[0])
size[1] = float(size[1])
except:
raise Exception('Invalid vector image target size "' + str(size) + '".')
sharedMaterials = str(data.get('sharedMaterials', ''))
shaders = str(data['vectorShaders'])
resources = data.get('resources', [])
if not isinstance(resources, list):
raise Exception('Invalid vector image resources "' + str(resources) + '".')
srgb = bool(data.get('srgb'))
except KeyError as e:
raise Exception('VectorImage doesn\'t contain element ' + str(e) + '.')
except (AttributeError, TypeError, ValueError):
raise Exception('VectorImage must be an object.')
if sharedMaterials:
sharedMaterialsOffset = builder.CreateString(sharedMaterials)
else:
sharedMaterialsOffset = 0
shadersOffset = builder.CreateString(shaders)
resourceOffsets = []
for resource in resources:
resourceOffsets.append(builder.CreateString(resource))
VectorImage.StartResourcesVector(builder, len(resourceOffsets))
for offset in reversed(resourceOffsets):
builder.PrependUOffsetTRelative(offset)
resourcesOffset = builder.EndVector()
VectorImage.Start(builder)
VectorImage.AddImageType(builder, imageType)
VectorImage.AddImage(builder, imageOffset)
if size:
sizeOffset = CreateVector2f(builder, size[0], size[1])
else:
sizeOffset = 0
VectorImage.AddTargetSize(builder, sizeOffset)
VectorImage.AddSharedMaterials(builder, sharedMaterialsOffset)
VectorImage.AddVectorShaders(builder, shadersOffset)
VectorImage.AddResources(builder, resourcesOffset)
VectorImage.AddSrgb(builder, srgb)
builder.Finish(VectorImage.End(builder))
return builder.Output()
score.content.range.min = 2
num_of_detection = _metadata_fb.TensorMetadataT()
num_of_detection.name = "number of detections"
num_of_detection.description = "The number of the detected boxes. "
num_of_detection.content = _metadata_fb.ContentT()
num_of_detection.content.contentPropertiesType = (_metadata_fb.ContentProperties.FeatureProperties)
num_of_detection.content.contentProperties = (_metadata_fb.FeaturePropertiesT())
subgraph = _metadata_fb.SubGraphMetadataT()
subgraph.outputTensorGroups = [output_tensorgroups]
subgraph.inputTensorMetadata = [input_meta]
subgraph.outputTensorMetadata = [location, category, score, num_of_detection]
model_meta.subgraphMetadata = [subgraph]
b = flatbuffers.Builder(0)
b.Finish(
model_meta.Pack(b),
_metadata.MetadataPopulator.METADATA_FILE_IDENTIFIER)
metadata_buf = b.Output()
populator = _metadata.MetadataPopulator.with_model_file(exported_model_path)
populator.load_metadata_buffer(metadata_buf)
populator.load_associated_files([labelmap_file])
populator.populate()
displayer = _metadata.MetadataDisplayer.with_model_file(exported_model_path)
export_json_file = os.path.join('C:/Users/anywh/Desktop/cetification/android/object_detection/android/app/src/main/assets', "colab_v2.json")
json_file = displayer.get_metadata_json()
def to_fb_stream(self, f: BinaryIO):
builder = flatbuffers.Builder(1024)
fbclips = self.to_fb(builder)
builder.Finish(fbclips)
buf = builder.Output()
f.write(buf)
cmd = sys.argv[1]
filename = sys.argv[2]
do_output = "-o" in sys.argv or "--output" in sys.argv
use_flatcc = "-c" in sys.argv or "--flatcc" in sys.argv
if use_flatcc:
import irondb.flatcc as irondb_flatbuf
else:
import irondb.flatbuf as irondb_flatbuf
if cmd.startswith("read"):
eprint("Using Flatbuffer module: " + irondb_flatbuf.__name__)
if cmd == "create_find_data":
num_entries = int(sys.argv[3])
leaf_cutoff = num_entries - (num_entries / 10)
builder = flatbuffers.Builder(0)
leaf_arr = []
for x in range(0, num_entries):
e = None
if x > leaf_cutoff:
uuid = builder.CreateString(
"11111111-1111-1111-1111-111111111111")
metric = builder.CreateString("dummy metric " + str(x))
check = builder.CreateString("dummy check " + str(x))
category = builder.CreateString("graphite")
egress = builder.CreateString("avg")
LeafData.LeafDataStart(builder)
LeafData.LeafDataAddUuid(builder, uuid)
LeafData.LeafDataAddMetricName(builder, metric)
LeafData.LeafDataAddCheckName(builder, check)
LeafData.LeafDataAddCategory(builder, category)
def read_join_response(command):
union_join = registrar.Registrar.Join.Join()
union_join.Init(command.Message().Bytes, command.Message().Pos)
room = union_join.Room()
print('CLIENT: Joined room with name ' + str(room.Name()))
clients_length = union_join.ClientsLength()
print('CLIENT: Joined room that has ' + str(clients_length) + ' other clients')
for i in range(clients_length):
client = union_join.Clients(i)
print('CLIENT: Client ' + str(i) + ' in room has ip ' + str(client.Ip()))
############ LIST REQUEST ############
builder = flatbuffers.Builder(1024)
offset = build_command(
builder,
registrar.Registrar.Message.Message().List,
0)
print('CLIENT: Sending a List request');
response = send_request(builder, req_socket, offset)
command = registrar.Registrar.Command.Command.GetRootAsCommand(response, 0)
first_room_guid = None
if command.MessageType() == registrar.Registrar.Message.Message().List:
print('CLIENT: Received List response from server')
first_room_guid = read_list_response(command)
else:
def main():
builder = flatbuffers.Builder(0)
# Create some weapons for our Monster ('Sword' and 'Axe').
weapon_one = builder.CreateString('Sword')
weapon_two = builder.CreateString('Axe')
MyGame.Sample.Weapon.WeaponStart(builder)
MyGame.Sample.Weapon.WeaponAddName(builder, weapon_one)
MyGame.Sample.Weapon.WeaponAddDamage(builder, 3)
sword = MyGame.Sample.Weapon.WeaponEnd(builder)
MyGame.Sample.Weapon.WeaponStart(builder)
MyGame.Sample.Weapon.WeaponAddName(builder, weapon_two)
MyGame.Sample.Weapon.WeaponAddDamage(builder, 5)
axe = MyGame.Sample.Weapon.WeaponEnd(builder)
# Serialize the FlatBuffer data.
name = builder.CreateString('Orc')
MyGame.Sample.Monster.MonsterStartInventoryVector(builder, 10)
# Note: Since we prepend the bytes, this loop iterates in reverse order.
for i in reversed(range(0, 10)):
builder.PrependByte(i)
inv = builder.EndVector(10)
MyGame.Sample.Monster.MonsterStartWeaponsVector(builder, 2)
# Note: Since we prepend the data, prepend the weapons in reverse order.
builder.PrependUOffsetTRelative(axe)
builder.PrependUOffsetTRelative(sword)
weapons = builder.EndVector(2)
pos = MyGame.Sample.Vec3.CreateVec3(builder, 1.0, 2.0, 3.0)
MyGame.Sample.Monster.MonsterStart(builder)
MyGame.Sample.Monster.MonsterAddPos(builder, pos)
MyGame.Sample.Monster.MonsterAddHp(builder, 300)
MyGame.Sample.Monster.MonsterAddName(builder, name)
MyGame.Sample.Monster.MonsterAddInventory(builder, inv)
MyGame.Sample.Monster.MonsterAddColor(builder,
MyGame.Sample.Color.Color().Red)
MyGame.Sample.Monster.MonsterAddWeapons(builder, weapons)
MyGame.Sample.Monster.MonsterAddEquippedType(
builder,
MyGame.Sample.Equipment.Equipment().Weapon)
MyGame.Sample.Monster.MonsterAddEquipped(builder, axe)
orc = MyGame.Sample.Monster.MonsterEnd(builder)
builder.Finish(orc)
# We now have a FlatBuffer that we could store on disk or send over a network.
# ...Saving to file or sending over a network code goes here...
# Instead, we are going to access this buffer right away (as if we just
# received it).
buf = builder.Output()
# Note: We use `0` for the offset here, since we got the data using the
# `builder.Output()` method. This simulates the data you would store/receive
# in your FlatBuffer. If you wanted to read from the `builder.Bytes` directly,
# you would need to pass in the offset of `builder.Head()`, as the builder
# actually constructs the buffer backwards.
monster = MyGame.Sample.Monster.Monster.GetRootAsMonster(buf, 0)
# Note: We did not set the `Mana` field explicitly, so we get a default value.
assert monster.Mana() == 150
assert monster.Hp() == 300
assert monster.Name() == 'Orc'
assert monster.Color() == MyGame.Sample.Color.Color().Red
assert monster.Pos().X() == 1.0
assert monster.Pos().Y() == 2.0
assert monster.Pos().Z() == 3.0
# Get and test the `inventory` FlatBuffer `vector`.
for i in xrange(monster.InventoryLength()):
assert monster.Inventory(i) == i
# Get and test the `weapons` FlatBuffer `vector` of `table`s.
expected_weapon_names = ['Sword', 'Axe']
expected_weapon_damages = [3, 5]
for i in xrange(monster.WeaponsLength()):
assert monster.Weapons(i).Name() == expected_weapon_names[i]
assert monster.Weapons(i).Damage() == expected_weapon_damages[i]
# Get and test the `equipped` FlatBuffer `union`.
assert monster.EquippedType() == MyGame.Sample.Equipment.Equipment().Weapon
# An example of how you can appropriately convert the table depending on the
# FlatBuffer `union` type. You could add `elif` and `else` clauses to handle
# the other FlatBuffer `union` types for this field.
if monster.EquippedType() == MyGame.Sample.Equipment.Equipment().Weapon:
# `monster.Equipped()` returns a `flatbuffers.Table`, which can be used
# to initialize a `MyGame.Sample.Weapon.Weapon()`, in this case.
union_weapon = MyGame.Sample.Weapon.Weapon()
union_weapon.Init(monster.Equipped().Bytes, monster.Equipped().Pos)
assert union_weapon.Name() == "Axe"
assert union_weapon.Damage() == 5
print 'The FlatBuffer was successfully created and verified!'
def pocketfft(c, a, *, fft_type: FftType, fft_lengths: List[int]):
"""PocketFFT kernel for CPU."""
shape = c.get_shape(a)
n = len(shape.dimensions())
dtype = shape.element_type()
builder = flatbuffers.Builder(128)
fft_lengths = list(fft_lengths)
assert len(fft_lengths) >= 1
assert len(fft_lengths) <= n, (fft_lengths, n)
forward = fft_type in (FftType.FFT, FftType.RFFT)
if fft_type == FftType.RFFT:
pocketfft_type = pd.PocketFftType.R2C
assert dtype in (np.float32, np.float64), dtype
out_dtype = np.dtype(np.complex64 if dtype ==
np.float32 else np.complex128)
pocketfft_dtype = (pd.PocketFftDtype.COMPLEX64 if dtype == np.float32
else pd.PocketFftDtype.COMPLEX128)
assert list(shape.dimensions())[-len(fft_lengths):] == fft_lengths, (
shape, fft_lengths)
out_shape = list(shape.dimensions())
out_shape[-1] = out_shape[-1] // 2 + 1
elif fft_type == FftType.IRFFT:
pocketfft_type = pd.PocketFftType.C2R
assert np.issubdtype(dtype, np.complexfloating), dtype
out_dtype = np.dtype(np.float32 if dtype ==
np.complex64 else np.float64)
pocketfft_dtype = (pd.PocketFftDtype.COMPLEX64 if dtype == np.complex64
else pd.PocketFftDtype.COMPLEX128)
assert list(
shape.dimensions())[-len(fft_lengths):-1] == fft_lengths[:-1]
out_shape = list(shape.dimensions())
out_shape[-1] = fft_lengths[-1]
assert (out_shape[-1] // 2 + 1) == shape.dimensions()[-1]
else:
pocketfft_type = pd.PocketFftType.C2C
assert np.issubdtype(dtype, np.complexfloating), dtype
out_dtype = dtype
pocketfft_dtype = (pd.PocketFftDtype.COMPLEX64 if dtype == np.complex64
else pd.PocketFftDtype.COMPLEX128)
assert list(shape.dimensions())[-len(fft_lengths):] == fft_lengths, (
shape, fft_lengths)
out_shape = shape.dimensions()
# PocketFft does not allow size 0 dimensions.
if 0 in shape.dimensions() or 0 in out_shape:
return xla_client.ops.Broadcast(
xla_client.ops.Constant(c, np.array(0, dtype=out_dtype)),
out_shape)
# Builds a PocketFftDescriptor flatbuffer. This descriptor is passed to the
# C++ kernel to describe the FFT to perform.
pd.PocketFftDescriptorStartShapeVector(builder, n)
for d in reversed(
shape.dimensions() if fft_type != FftType.IRFFT else out_shape):
builder.PrependUint64(d)
if flatbuffers_version_2:
pocketfft_shape = builder.EndVector()
else:
pocketfft_shape = builder.EndVector(n)
pd.PocketFftDescriptorStartStridesInVector(builder, n)
stride = dtype.itemsize
for d in reversed(shape.dimensions()):
builder.PrependUint64(stride)
stride *= d
if flatbuffers_version_2:
strides_in = builder.EndVector()
else:
strides_in = builder.EndVector(n)
pd.PocketFftDescriptorStartStridesOutVector(builder, n)
stride = out_dtype.itemsize
for d in reversed(out_shape):
builder.PrependUint64(stride)
stride *= d
if flatbuffers_version_2:
strides_out = builder.EndVector()
else:
strides_out = builder.EndVector(n)
pd.PocketFftDescriptorStartAxesVector(builder, len(fft_lengths))
for d in range(len(fft_lengths)):
builder.PrependUint32(n - d - 1)
if flatbuffers_version_2:
axes = builder.EndVector()
else:
axes = builder.EndVector(len(fft_lengths))
scale = 1. if forward else (1. / np.prod(fft_lengths))
pd.PocketFftDescriptorStart(builder)
pd.PocketFftDescriptorAddDtype(builder, pocketfft_dtype)
pd.PocketFftDescriptorAddFftType(builder, pocketfft_type)
pd.PocketFftDescriptorAddShape(builder, pocketfft_shape)
pd.PocketFftDescriptorAddStridesIn(builder, strides_in)
pd.PocketFftDescriptorAddStridesOut(builder, strides_out)
pd.PocketFftDescriptorAddAxes(builder, axes)
pd.PocketFftDescriptorAddForward(builder, forward)
pd.PocketFftDescriptorAddScale(builder, scale)
descriptor = pd.PocketFftDescriptorEnd(builder)
builder.Finish(descriptor)
descriptor_bytes = builder.Output()
return xla_client.ops.CustomCallWithLayout(
c,
b"pocketfft",
operands=(
xla_client.ops.Constant(
c, np.frombuffer(descriptor_bytes, dtype=np.uint8)),
a,
),
shape_with_layout=xla_client.Shape.array_shape(
out_dtype, out_shape, tuple(range(n - 1, -1, -1))),
operand_shapes_with_layout=(
xla_client.Shape.array_shape(np.dtype(np.uint8),
(len(descriptor_bytes), ), (0, )),
xla_client.Shape.array_shape(dtype, shape.dimensions(),
tuple(range(n - 1, -1, -1))),
))
async def game_loop(self):
last_tick_game_time = None # What the tick time of the last observed tick was
last_call_real_time = datetime.now() # When we last called the Agent
packet = GameTickPacket()
# Run until main process tells to stop
while not self.quit_event.is_set():
before = datetime.now()
self.game_interface.update_live_data_packet(packet)
# Run the Agent only if the gameInfo has updated.
tick_game_time = packet.game_info.seconds_elapsed
worth_communicating = tick_game_time != last_tick_game_time or \
datetime.now() - last_call_real_time >= MAX_AGENT_CALL_PERIOD
ball = packet.game_ball
if ball is not None and worth_communicating and max(
self.running_indices) < packet.num_cars:
last_tick_game_time = tick_game_time
last_call_real_time = datetime.now()
tiny_player_offsets = []
builder = flatbuffers.Builder(0)
for i in range(packet.num_cars):
tiny_player_offsets.append(
self.copy_player(packet.game_cars[i], builder))
TinyPacket.TinyPacketStartPlayersVector(
builder, packet.num_cars)
for i in reversed(range(0, len(tiny_player_offsets))):
rlbot_index = self.get_rlbot_index(i)
builder.PrependUOffsetTRelative(
tiny_player_offsets[rlbot_index])
players_offset = builder.EndVector(len(tiny_player_offsets))
ballOffset = self.copy_ball(ball, builder)
TinyPacket.TinyPacketStart(builder)
TinyPacket.TinyPacketAddPlayers(builder, players_offset)
TinyPacket.TinyPacketAddBall(builder, ballOffset)
packet_offset = TinyPacket.TinyPacketEnd(builder)
builder.Finish(packet_offset)
buffer = bytes(builder.Output())
filtered_sockets = {s for s in self.current_sockets if s.open}
for socket in filtered_sockets:
await socket.send(buffer)
self.current_sockets = filtered_sockets
after = datetime.now()
duration = (after - before).total_seconds()
sleep_secs = 1 / 60 - duration
if sleep_secs > 0:
await asyncio.sleep(sleep_secs)
def _create_metadata(self):
"""Creates the metadata for an image classifier."""
# Creates model info.
model_meta = _metadata_fb.ModelMetadataT()
model_meta.name = self.model_info.name
model_meta.description = ("Identify the most prominent object in the "
"image from a set of %d categories." %
self.model_info.num_classes)
model_meta.version = self.model_info.version
model_meta.author = "TensorFlow"
model_meta.license = ("Apache License. Version 2.0 "
"http://www.apache.org/licenses/LICENSE-2.0.")
# Creates input info.
input_meta = _metadata_fb.TensorMetadataT()
input_meta.name = "image"
input_meta.description = (
"Input image to be classified. The expected image is {0} x {1}, with "
"three channels (red, blue, and green) per pixel. Each value in the "
"tensor is a single byte between {2} and {3}.".format(
self.model_info.image_width, self.model_info.image_height,
self.model_info.image_min, self.model_info.image_max))
input_meta.content = _metadata_fb.ContentT()
input_meta.content.contentProperties = _metadata_fb.ImagePropertiesT()
input_meta.content.contentProperties.colorSpace = (
_metadata_fb.ColorSpaceType.RGB)
input_meta.content.contentPropertiesType = (
_metadata_fb.ContentProperties.ImageProperties)
input_normalization = _metadata_fb.ProcessUnitT()
input_normalization.optionsType = (
_metadata_fb.ProcessUnitOptions.NormalizationOptions)
input_normalization.options = _metadata_fb.NormalizationOptionsT()
input_normalization.options.mean = self.model_info.mean
input_normalization.options.std = self.model_info.std
input_meta.processUnits = [input_normalization]
input_stats = _metadata_fb.StatsT()
input_stats.max = [self.model_info.image_max]
input_stats.min = [self.model_info.image_min]
input_meta.stats = input_stats
# Creates output info.
output_meta = _metadata_fb.TensorMetadataT()
output_meta.name = "probability"
output_meta.description = "Probabilities of the %d labels respectively." % self.model_info.num_classes
output_meta.content = _metadata_fb.ContentT()
output_meta.content.content_properties = _metadata_fb.FeaturePropertiesT(
)
output_meta.content.contentPropertiesType = (
_metadata_fb.ContentProperties.FeatureProperties)
output_stats = _metadata_fb.StatsT()
output_stats.max = [1.0]
output_stats.min = [0.0]
output_meta.stats = output_stats
label_file = _metadata_fb.AssociatedFileT()
label_file.name = os.path.basename(self.label_file_path)
label_file.description = "Labels for objects that the model can recognize."
label_file.type = _metadata_fb.AssociatedFileType.TENSOR_AXIS_LABELS
output_meta.associatedFiles = [label_file]
# Creates subgraph info.
subgraph = _metadata_fb.SubGraphMetadataT()
subgraph.inputTensorMetadata = [input_meta]
subgraph.outputTensorMetadata = [output_meta]
model_meta.subgraphMetadata = [subgraph]
b = flatbuffers.Builder(0)
b.Finish(model_meta.Pack(b),
_metadata.MetadataPopulator.METADATA_FILE_IDENTIFIER)
self.metadata_buf = b.Output()
def write_calibration_table(calibration_cache):
'''
Helper function to write calibration table to files.
'''
import json
import flatbuffers
import onnxruntime.quantization.CalTableFlatBuffers.TrtTable as TrtTable
import onnxruntime.quantization.CalTableFlatBuffers.KeyValue as KeyValue
logging.info("calibration cache: {}".format(calibration_cache))
with open("calibration.json", 'w') as file:
file.write(json.dumps(
calibration_cache)) # use `json.loads` to do the reverse
# Serialize data using FlatBuffers
builder = flatbuffers.Builder(1024)
key_value_list = []
for key in sorted(calibration_cache.keys()):
values = calibration_cache[key]
value = str(max(abs(values[0]), abs(values[1])))
flat_key = builder.CreateString(key)
flat_value = builder.CreateString(value)
KeyValue.KeyValueStart(builder)
KeyValue.KeyValueAddKey(builder, flat_key)
KeyValue.KeyValueAddValue(builder, flat_value)
key_value = KeyValue.KeyValueEnd(builder)
key_value_list.append(key_value)
TrtTable.TrtTableStartDictVector(builder, len(key_value_list))
for key_value in key_value_list:
builder.PrependUOffsetTRelative(key_value)
main_dict = builder.EndVector(len(key_value_list))
TrtTable.TrtTableStart(builder)
TrtTable.TrtTableAddDict(builder, main_dict)
cal_table = TrtTable.TrtTableEnd(builder)
builder.Finish(cal_table)
buf = builder.Output()
with open("calibration.flatbuffers", 'wb') as file:
file.write(buf)
# Deserialize data (for validation)
if False:
cal_table = TrtTable.TrtTable.GetRootAsTrtTable(buf, 0)
dict_len = cal_table.DictLength()
for i in range(dict_len):
key_value = cal_table.Dict(i)
logging.info(key_value.Key())
logging.info(key_value.Value())
# write plain text
with open("calibration.cache", 'w') as file:
for key in sorted(calibration_cache.keys()):
value = calibration_cache[key]
s = key + ' ' + str(max(abs(value[0]), abs(value[1])))
file.write(s)
file.write('\n')
def convertText(convertContext, data):
"""
Converts text for use in a scene. The data map is expected to contain the following elements:
- resources: the name of the vector resources to get the font from.
- font: the name of the font to use if not provided by the text XML element. Defaults to
'serif'.
- text: the text. This can either be a path to a .xml file or embedded XML string. The XML
contents should be a single <text> SVG element with any number of <tspan> embedded elements.
(see https://www.w3.org/TR/SVG2/text.html#TextElement for details) Only solid colors are
allowed for stroke and fill. When a position is provided, only a relative offset for the
vertical position is supported.
"""
builder = flatbuffers.Builder(0)
try:
vectorResources = str(data['vectorResources'])
defaultFont = str(data.get('font', 'serif'))
textData = str(data['text']).strip()
except KeyError as e:
raise Exception('SceneText doesn\'t contain element ' + str(e) + '.')
except (AttributeError, TypeError, ValueError):
raise Exception('SceneText must be an object.')
if textData.startswith('<'):
if sys.version_info < (3, 0):
textData = unicode(textData)
textXml = minidom.parse(io.StringIO(textData))
else:
textXml = minidom.parse(textData)
materials = Materials('text')
font, text, ranges = readText(textXml.firstChild, defaultFont, [0.0, 0.0],
0.0, materials)
if not text:
raise Exception('Invalid SVG text data.')
colorMaterials = dict()
for i in range(0, len(materials.colors)):
colorMaterials[materials.getColorName(i)] = materials.colors[i]
vectorResourcesOffset = builder.CreateString(vectorResources)
fontOffset = builder.CreateString(font.font)
VectorResourceRef.Start(builder)
VectorResourceRef.AddResources(builder, vectorResourcesOffset)
VectorResourceRef.AddName(builder, fontOffset)
vectorResourceRefOffset = VectorResourceRef.End(builder)
textOffset = builder.CreateString(text)
rangeOffsets = []
for textRange in ranges:
style = textRange.style
SceneTextStyle.Start(builder)
SceneTextStyle.AddStart(builder, textRange.start)
SceneTextStyle.AddCount(builder, textRange.count)
SceneTextStyle.AddSize(builder, style.font.size)
SceneTextStyle.AddEmbolden(builder, style.font.embolden)
SceneTextStyle.AddSlant(builder, style.font.slant)
SceneTextStyle.AddOutlineWidth(
builder, style.stroke.width if style.stroke else 0.0)
SceneTextStyle.AddFuziness(builder, 1.0)
SceneTextStyle.AddVerticalOffset(
builder, textRange.position[1]
if textRange.positionType == TextPosition.Offset else 0.0)
if style.fill:
colorOffset = createColor(colorMaterials, style.fill, builder)
else:
colorOffset = 0
SceneTextStyle.AddColor(builder, colorOffset)
if style.stroke:
colorOffset = createColor(colorMaterials, style.stroke, builder)
else:
colorOffset = 0
SceneTextStyle.AddOutlineColor(builder, colorOffset)
rangeOffsets.append(SceneTextStyle.End(builder))
SceneText.StartStylesVector(builder, len(rangeOffsets))
for offset in reversed(rangeOffsets):
builder.PrependUOffsetTRelative(offset)
stylesOffset = builder.EndVector()
SceneText.Start(builder)
SceneText.AddFont(builder, vectorResourceRefOffset)
SceneText.AddText(builder, textOffset)
SceneText.AddStyles(builder, stylesOffset)
builder.Finish(SceneText.End(builder))
return builder.Output()
def convertShadowManager(convertContext, data):
"""
Converts a shadow manager for a scene. The data map is expected to contain the following
elements:
- lightSet: the name of the light set to query the light from. If set, this will be the default
for elements in the shadows array.
- shadows: array of objects for the shadows the shadow manager will manage. Each element is
expected to have the following members:
- name: name of the shadows.
- lightSet: name of the light set to query the light from.
- lightType: type of the light to shadow. See dsSceneLightType enum for values, removing the
type prefix.
- light: name of the light to shadow. May be unset to disable initially until set at runtime.
- transformGroupDesc: name of the shader variable group description for the transform group.
- transformGroupName: name of the transform group to set as view global data. This may be
omitted if not used as global data on a view.
- maxCascades: the maximum number of cascades for cascaded directional light shadows. Defaults
to 4.
- maxFirstSplitDistance: maximum distance for the first split for cascaded shadows. Defaults
to 100.
- cascadeExpFactor: exponential factor for cascaded shadows in the range [0, 1], where 0 uses
linear distances between the splits and 1 is fully exponential. Defaults to 0.5.
- minDepthRanges: minimim distance between the near and far planes for each cascade. Spot and
point light shadows only use the first value. Can either be an array to set the cascade
values or a float to set all 4 possible cascade values.
- fadeStartDistance: the distance to start fading out shadows. Defaults to 1000000, which is a
large distance less likely to break GPUs that use limited precision floats.
- maxDistance: the maximum distance to display shadows. Defaults to 1000000, which is a large
distance less likely to break GPUs that use limited precision floats.
"""
largeDistance = 100000000.0
def readFloat(value, name, minVal=None, maxVal=None):
try:
floatVal = float(value)
if (minVal is not None and floatVal < minVal) or \
(maxVal is not None and floatVal > maxVal):
raise Exception() # Common error handling in except block.
return floatVal
except:
raise Exception('Invalid ' + name + ' value "' + str(value) + '".')
try:
shadowsData = data['shadows']
defaultLightSet = data.get('lightSet', '')
shadows = []
try:
for shadowData in shadowsData:
try:
shadow = Object()
shadow.name = str(shadowData['name'])
shadow.lightSet = str(
shadowData.get('lightSet', defaultLightSet))
if not shadow.lightSet:
raise KeyError('lightSet')
lightTypeStr = str(shadowData['lightType'])
try:
shadow.lightType = getattr(LightType, lightTypeStr)
except AttributeError:
raise Exception('Invalid light type "' + lightTypeStr +
'".')
shadow.light = str(shadowData.get('light', ''))
shadow.transformGroupDesc = str(
shadowData['transformGroupDesc'])
shadow.transformGroupName = str(
shadowData.get('transformGroupName', ''))
maxCascadesVal = shadowData.get('maxCascades', 4)
try:
shadow.maxCascades = int(maxCascadesVal)
if shadow.maxCascades < 1 or shadow.maxCascades > 4:
raise Exception(
) # Common error handling in except block.
except:
raise Exception('Invalid max cascade count "' +
str(maxCascadesVal) + '".')
minDepthRangesVal = shadowData.get('minDepthRanges')
shadow.minDepthRanges = []
if isinstance(minDepthRangesVal, list):
for val in minDepthRangesVal:
shadow.minDepthRanges.append(
readFloat(val, 'minDepthRanges'))
elif minDepthRangesVal:
floatVal = readFloat(minDepthRangesVal,
'minDepthRanges')
shadow.minDepthRanges = [
floatVal, floatVal, floatVal, floatVal
]
shadow.maxFirstSplitDistance = readFloat(
shadowData.get('maxFirstSplitDistance', 100.0),
'max first split distance', 0.1)
shadow.cascadeExpFactor = readFloat(
shadowData.get('cascadeExpFactor', 0.5),
'cascade exp factor', 0.0, 1.0)
shadow.fadeStartDistance = readFloat(
shadowData.get('fadeStartDistance', largeDistance),
'fade start distance', 0.0)
shadow.maxDistance = readFloat(
shadowData.get('maxDistance', largeDistance),
'max distance', 0.1)
shadows.append(shadow)
except KeyError as e:
raise Exception(
'ShadowManager shadows doesn\'t contain element ' +
str(e) + '.')
except (AttributeError, TypeError, ValueError):
raise Exception(
'ShadowManager shadows must be an array of objects.')
except KeyError as e:
raise Exception('ShadowManager doesn\'t contain element ' + str(e) +
'.')
except (AttributeError, TypeError, ValueError):
raise Exception('ShadowManager must be an object.')
builder = flatbuffers.Builder(0)
shadowOffsets = []
for shadow in shadows:
nameOffset = builder.CreateString(shadow.name)
lightSetOffset = builder.CreateString(shadow.lightSet)
if shadow.light:
lightOffset = builder.CreateString(shadow.light)
else:
lightOffset = 0
transformGroupDescOffset = builder.CreateString(
shadow.transformGroupDesc)
if shadow.transformGroupName:
transformGroupNameOffset = builder.CreateString(
shadow.transformGroupName)
else:
transformGroupNameOffset = 0
if shadow.minDepthRanges:
SceneLightShadows.StartMinDepthRangesVector(
builder, len(shadow.minDepthRanges))
for val in reversed(shadow.minDepthRanges):
builder.PrependFloat32(val)
minDepthRangesOffset = builder.EndVector()
else:
minDepthRangesOffset = 0
SceneLightShadows.Start(builder)
SceneLightShadows.AddName(builder, nameOffset)
SceneLightShadows.AddLightSet(builder, lightSetOffset)
SceneLightShadows.AddLightType(builder, shadow.lightType)
SceneLightShadows.AddLight(builder, lightOffset)
SceneLightShadows.AddTransformGroupDesc(builder,
transformGroupDescOffset)
SceneLightShadows.AddTransformGroupName(builder,
transformGroupNameOffset)
SceneLightShadows.AddMaxCascades(builder, shadow.maxCascades)
SceneLightShadows.AddMaxFirstSplitDistance(
builder, shadow.maxFirstSplitDistance)
SceneLightShadows.AddCascadeExpFactor(builder, shadow.cascadeExpFactor)
SceneLightShadows.AddMinDepthRanges(builder, minDepthRangesOffset)
SceneLightShadows.AddFadeStartDistance(builder,
shadow.fadeStartDistance)
SceneLightShadows.AddMaxDistance(builder, shadow.maxDistance)
shadowOffsets.append(SceneLightShadows.End(builder))
SceneShadowManager.StartShadowsVector(builder, len(shadowOffsets))
for offset in reversed(shadowOffsets):
builder.PrependUOffsetTRelative(offset)
shadowsOffset = builder.EndVector()
SceneShadowManager.Start(builder)
SceneShadowManager.AddShadows(builder, shadowsOffset)
builder.Finish(SceneShadowManager.End(builder))
return builder.Output()
def build_command(builder, message_type, message):
registrar.Registrar.Command.CommandStart(builder)
registrar.Registrar.Command.CommandAddMessageType(builder, message_type)
registrar.Registrar.Command.CommandAddMessage(builder, message)
return registrar.Registrar.Command.CommandEnd(builder)
while True:
# Wait for next request from client
request = rep_socket.recv()
Command = registrar.Registrar.Command.Command.GetRootAsCommand(request, 0)
# Switch on the message type to send a response.
req_builder = flatbuffers.Builder(1024)
pub_builder = flatbuffers.Builder(1024)
message_offset = 0
message_type = Command.MessageType()
publish_offset = None
channel = None
if message_type == registrar.Registrar.Message.Message().List:
print('SERVER: Received a List command')
rooms = Room.select()
message_offset = list_api.response(req_builder, rooms)
elif message_type == registrar.Registrar.Message.Message().Create:
print('SERVER: Received a Create command')
union_create = registrar.Registrar.Create.Create()
union_create.Init(Command.Message().Bytes, Command.Message().Pos)
client = union_create.Client()
room = None
def builder():
_builder = flatbuffers.Builder(0)
return _builder
def forward(self):
builder = flatbuffers.Builder(64)
# construct MessageBody
ppx_Run.RunStart(builder)
message_body = ppx_Run.RunEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(builder,
ppx_MessageBody.MessageBody().Run)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
# Bradley: This is very inefficient. Should replace
# with a hash map. But I don't think this will be used
# once everything is connected.
while True:
reply = self._requester.receive_reply()
message_body = self._get_message_body(reply)
if isinstance(message_body, ppx_RunResult.RunResult):
result = self._protocol_tensor_to_variable(
message_body.Result())
return result
elif isinstance(message_body, ppx_Sample.Sample):
address = message_body.Address().decode('utf-8')
name = message_body.Name().decode('utf-8')
if name == '':
name = None
control = bool(message_body.Control())
replace = bool(message_body.Replace())
distribution_type = message_body.DistributionType()
if distribution_type == ppx_Distribution.Distribution(
).Uniform:
uniform = ppx_Uniform.Uniform()
uniform.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
low = self._protocol_tensor_to_variable(uniform.Low())
high = self._protocol_tensor_to_variable(uniform.High())
dist = Uniform(low, high)
elif distribution_type == ppx_Distribution.Distribution(
).Normal:
normal = ppx_Normal.Normal()
normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(normal.Mean())
stddev = self._protocol_tensor_to_variable(normal.Stddev())
dist = Normal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Categorical:
categorical = ppx_Categorical.Categorical()
categorical.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
probs = self._protocol_tensor_to_variable(
categorical.Probs())
dist = Categorical(probs)
elif distribution_type == ppx_Distribution.Distribution(
).Poisson:
poisson = ppx_Poisson.Poisson()
poisson.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(poisson.Rate())
dist = Poisson(rate)
elif distribution_type == ppx_Distribution.Distribution(
).Gamma:
gamma = ppx_Gamma.Gamma()
gamma.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
concentration = self._protocol_tensor_to_variable(
gamma.Concentration())
rate = self._protocol_tensor_to_variable(gamma.Rate())
dist = Gamma(concentration, rate)
elif distribution_type == ppx_Distribution.Distribution(
).LogNormal:
log_normal = ppx_LogNormal.LogNormal()
log_normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(log_normal.Mean())
stddev = self._protocol_tensor_to_variable(
log_normal.Stddev())
dist = LogNormal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Exponential:
exponential = ppx_Exponential.Exponential()
exponential.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(
exponential.Rate())
dist = Exponential(rate)
elif distribution_type == ppx_Distribution.Distribution(
).Weibull:
weibull = ppx_Weibull.Weibull()
weibull.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
scale = self._protocol_tensor_to_variable(weibull.Scale())
concentration = self._protocol_tensor_to_variable(
weibull.Concentration())
dist = Weibull(scale, concentration)
else:
raise RuntimeError(
'ppx (Python): Sample from an unexpected distribution requested.'
)
result = state.sample(distribution=dist,
control=control,
replace=replace,
name=name,
address=address)
# print(" Debug statement in pyprob.Remote.forward(). \n Printing the sampling statement \n \
# Result : {0} \n \
# Distribution : {1}".format(result,dist))
builder = flatbuffers.Builder(64)
result = self._variable_to_protocol_tensor(builder, result)
ppx_SampleResult.SampleResultStart(builder)
ppx_SampleResult.SampleResultAddResult(builder, result)
message_body = ppx_SampleResult.SampleResultEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(
builder,
ppx_MessageBody.MessageBody().SampleResult)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
elif isinstance(message_body, ppx_Observe.Observe):
address = message_body.Address().decode('utf-8')
name = message_body.Name().decode('utf-8')
if name == '':
name = None
value = self._protocol_tensor_to_variable(message_body.Value())
distribution_type = message_body.DistributionType()
if distribution_type == ppx_Distribution.Distribution().NONE:
dist = None
elif distribution_type == ppx_Distribution.Distribution(
).Uniform:
uniform = ppx_Uniform.Uniform()
uniform.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
low = self._protocol_tensor_to_variable(uniform.Low())
high = self._protocol_tensor_to_variable(uniform.High())
dist = Uniform(low, high)
elif distribution_type == ppx_Distribution.Distribution(
).Normal:
normal = ppx_Normal.Normal()
normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(normal.Mean())
stddev = self._protocol_tensor_to_variable(normal.Stddev())
dist = Normal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Categorical:
categorical = ppx_Categorical.Categorical()
categorical.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
probs = self._protocol_tensor_to_variable(
categorical.Probs())
dist = Categorical(probs)
elif distribution_type == ppx_Distribution.Distribution(
).Poisson:
poisson = ppx_Poisson.Poisson()
poisson.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(poisson.Rate())
dist = Poisson(rate)
elif distribution_type == ppx_Distribution.Distribution(
).Gamma:
gamma = ppx_Gamma.Gamma()
gamma.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
concentration = self._protocol_tensor_to_variable(
gamma.Concentration())
rate = self._protocol_tensor_to_variable(gamma.Rate())
dist = Gamma(concentration, rate)
elif distribution_type == ppx_Distribution.Distribution(
).LogNormal:
log_normal = ppx_LogNormal.LogNormal()
log_normal.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
mean = self._protocol_tensor_to_variable(log_normal.Mean())
stddev = self._protocol_tensor_to_variable(
log_normal.Stddev())
dist = LogNormal(mean, stddev)
elif distribution_type == ppx_Distribution.Distribution(
).Exponential:
exponential = ppx_Exponential.Exponential()
exponential.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
rate = self._protocol_tensor_to_variable(
exponential.Rate())
dist = Exponential(rate)
elif distribution_type == ppx_Distribution.Distribution(
).Weibull:
weibull = ppx_Weibull.Weibull()
weibull.Init(message_body.Distribution().Bytes,
message_body.Distribution().Pos)
scale = self._protocol_tensor_to_variable(weibull.Scale())
concentration = self._protocol_tensor_to_variable(
weibull.Concetration())
dist = Weibull(scale, concentration)
else:
raise RuntimeError(
'ppx (Python): Sample from an unexpected distribution requested: {}'
.format(distribution_type))
state.observe(distribution=dist,
value=value,
name=name,
address=address)
builder = flatbuffers.Builder(64)
ppx_ObserveResult.ObserveResultStart(builder)
message_body = ppx_ObserveResult.ObserveResultEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(
builder,
ppx_MessageBody.MessageBody().ObserveResult)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
elif isinstance(message_body, ppx_Tag.Tag):
address = message_body.Address().decode('utf-8')
name = message_body.Name().decode('utf-8')
if name == '':
name = None
value = self._protocol_tensor_to_variable(message_body.Value())
state.tag(value=value, name=name, address=address)
builder = flatbuffers.Builder(64)
ppx_TagResult.TagResultStart(builder)
message_body = ppx_TagResult.TagResultEnd(builder)
# construct Message
ppx_Message.MessageStart(builder)
ppx_Message.MessageAddBodyType(
builder,
ppx_MessageBody.MessageBody().TagResult)
ppx_Message.MessageAddBody(builder, message_body)
message = ppx_Message.MessageEnd(builder)
builder.Finish(message)
message = builder.Output()
self._requester.send_request(message)
elif isinstance(message_body, ppx_Reset.Reset):
raise RuntimeError(
'ppx (Python): Received a reset request. Protocol out of sync.'
)
else:
raise RuntimeError(
'ppx (Python): Received unexpected message.')