code-loader 1.0.65__tar.gz → 1.0.65.dev1__tar.gz

{code_loader-1.0.65 → code_loader-1.0.65.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: code-loader
-Version: 1.0.65
+Version: 1.0.65.dev1
 Summary: 
 Home-page: https://github.com/tensorleap/code-loader
 License: MIT

code_loader-1.0.65.dev1/code_loader/inner_leap_binder/inner_classes.py

@@ -0,0 +1,34 @@
+
+from typing import List, Dict, Union, Callable
+from enum import Enum, IntEnum
+from dataclasses import dataclass, field
+from typing import Union
+
+class FunctionType(IntEnum):
+    gt = 0
+    visualizer = 1
+    input = 2
+    metric = 3
+    loss = 4
+    model_input = 5
+    model_output = 6
+
+@dataclass(frozen=True)
+class TLFunction:
+    FuncType: FunctionType
+    FuncName: str
+
+MappingList = List[Dict[TLFunction, Dict[str, Union[Callable, int]]]]
+
+def leap_input(idx):
+    def dummy():
+        return None
+    dummy.tl_func = TLFunction(FunctionType.model_input, idx)
+    return dummy
+
+def leap_output(idx):
+    def dummy():
+        return None
+    dummy.tl_func = TLFunction(FunctionType.model_output, idx)
+    return dummy
+

{code_loader-1.0.65 → code_loader-1.0.65.dev1}/code_loader/inner_leap_binder/leapbinder.py

@@ -1,5 +1,5 @@
 import inspect
-from typing import Callable, List, Optional, Dict, Any, Type, Union
+from typing import Callable, List, Optional, Dict, Any, Type, Union, Tuple
 
 import numpy as np
 import numpy.typing as npt
@@ -20,6 +20,7 @@ from code_loader.visualizers.default_visualizers import DefaultVisualizer, \
     default_graph_visualizer, \
     default_image_visualizer, default_horizontal_bar_visualizer, default_word_visualizer, \
     default_image_mask_visualizer, default_text_mask_visualizer, default_raw_data_visualizer
+from code_loader.inner_leap_binder.inner_classes import FunctionType, TLFunction, MappingList, leap_output, leap_input
 
 
 class LeapBinder:
@@ -35,6 +36,7 @@ class LeapBinder:
     def __init__(self) -> None:
         self.setup_container = DatasetIntegrationSetup()
         self.cache_container: Dict[str, Any] = {"word_to_index": {}}
+        self._mappings: MappingList = list()
         self._visualizer_names: List[str] = list()
         self._encoder_names: List[str] = list()
         self._extend_with_default_visualizers()
@@ -65,7 +67,8 @@ class LeapBinder:
     def set_visualizer(self, function: VisualizerCallableInterface,
                        name: str,
                        visualizer_type: LeapDataType,
-                       heatmap_visualizer: Optional[Callable[..., npt.NDArray[np.float32]]] = None) -> None:
+                       heatmap_visualizer: Optional[Callable[..., npt.NDArray[np.float32]]] = None,
+                       connects_to = None) -> None: #TODO add types
         """
         Set a visualizer for a specific data type.
 
@@ -132,10 +135,11 @@ class LeapBinder:
                 raise Exception(
                     f'The return type of function {function.__name__} is invalid. current return type: {return_type}, '
                     f'should be {expected_return_type}')
-
         self.setup_container.visualizers.append(
             VisualizerHandler(VisualizerHandlerData(name, visualizer_type, arg_names), function, heatmap_visualizer))
         self._visualizer_names.append(name)
+        if connects_to is not None:
+            self._mappings.append({TLFunction(FunctionType.visualizer, name): connects_to})
 
     def set_preprocess(self, function: Callable[[], List[PreprocessResponse]]) -> None:
         """
@@ -240,7 +244,8 @@ class LeapBinder:
                           CustomMultipleReturnCallableInterfaceMultiArgs,
                           ConfusionMatrixCallableInterfaceMultiArgs],
                           name: str,
-                          direction: Optional[MetricDirection] = MetricDirection.Downward) -> None:
+                          direction: Optional[MetricDirection] = MetricDirection.Downward,
+                          connects_to = None) -> None: #TODO TOM add types
         """
         Add a custom metric to the setup.
 
@@ -260,6 +265,8 @@ class LeapBinder:
         """
         arg_names = inspect.getfullargspec(function)[0]
         self.setup_container.metrics.append(MetricHandler(MetricHandlerData(name, arg_names, direction), function))
+        if connects_to is not None:
+            self._mappings.append({TLFunction(FunctionType.metric, name): connects_to})
 
     def add_prediction(self, name: str, labels: List[str], channel_dim: int = -1) -> None:
         """
@@ -476,6 +483,9 @@ class LeapBinder:
                     continue
                 self.check_handler(preprocess_response, test_result, dataset_base_handler)
 
+    def set_model_inputs(self, connections: Tuple[Callable[..., Any], ...]):
+        self._mappings.append({TLFunction(FunctionType.model_input, "model"): {str(i): (connections[i], ) for i in range(len(connections))}})
+
     def check(self) -> None:
         preprocess_result = self.get_preprocess_result()
         self.check_preprocess(preprocess_result)

{code_loader-1.0.65 → code_loader-1.0.65.dev1}/code_loader/inner_leap_binder/leapbinder_decorators.py

@@ -13,9 +13,10 @@ from code_loader.contract.enums import MetricDirection, LeapDataType
 from code_loader import leap_binder
 from code_loader.contract.visualizer_classes import LeapImage, LeapImageMask, LeapTextMask, LeapText, LeapGraph, \
     LeapHorizontalBar, LeapImageWithBBox, LeapImageWithHeatmap
+from code_loader.inner_leap_binder.inner_classes import TLFunction, FunctionType
 
-
-def tensorleap_custom_metric(name: str, direction: Optional[MetricDirection] = MetricDirection.Downward):
+def tensorleap_custom_metric(name: str, direction: Optional[MetricDirection] = MetricDirection.Downward,
+                             connects_to = None): #TODO TOM add type
     def decorating_function(
             user_function: Union[CustomCallableInterfaceMultiArgs,
             CustomMultipleReturnCallableInterfaceMultiArgs,
@@ -26,7 +27,7 @@ def tensorleap_custom_metric(name: str, direction: Optional[MetricDirection] = M
                 raise Exception(f'Metric with name {name} already exists. '
                                 f'Please choose another')
 
-        leap_binder.add_custom_metric(user_function, name, direction)
+        leap_binder.add_custom_metric(user_function, name, direction, connects_to=connects_to)
 
         def _validate_input_args(*args, **kwargs) -> None:
             for i, arg in enumerate(args):
@@ -91,14 +92,15 @@ def tensorleap_custom_metric(name: str, direction: Optional[MetricDirection] = M
 
 
 def tensorleap_custom_visualizer(name: str, visualizer_type: LeapDataType,
-                                 heatmap_function: Optional[Callable[..., npt.NDArray[np.float32]]] = None):
+                                 heatmap_function: Optional[Callable[..., npt.NDArray[np.float32]]] = None,
+                                 connects_to = None): #TODO TOM ADD TYPES
     def decorating_function(user_function: VisualizerCallableInterface):
         for viz_handler in leap_binder.setup_container.visualizers:
             if viz_handler.visualizer_handler_data.name == name:
                 raise Exception(f'Visualizer with name {name} already exists. '
                                 f'Please choose another')
 
-        leap_binder.set_visualizer(user_function, name, visualizer_type, heatmap_function)
+        leap_binder.set_visualizer(user_function, name, visualizer_type, heatmap_function, connects_to)
 
         def _validate_input_args(*args, **kwargs):
             for i, arg in enumerate(args):
@@ -137,6 +139,7 @@ def tensorleap_custom_visualizer(name: str, visualizer_type: LeapDataType,
             result = user_function(*args, **kwargs)
             _validate_result(result)
             return result
+        inner.tl_func = TLFunction(FunctionType.visualizer, name)
 
         return inner
 
@@ -285,6 +288,7 @@ def tensorleap_input_encoder(name: str, channel_dim=-1):
             result = user_function(sample_id, preprocess_response)
             _validate_result(result)
             return result
+        inner.tl_func = TLFunction(FunctionType.input, name)
 
         return inner
 
@@ -325,7 +329,7 @@ def tensorleap_gt_encoder(name: str):
             result = user_function(sample_id, preprocess_response)
             _validate_result(result)
             return result
-
+        inner.tl_func = TLFunction(FunctionType.gt, name)
         return inner
 
     return decorating_function
@@ -379,6 +383,7 @@ def tensorleap_custom_loss(name: str):
             _validate_result(result)
             return result
 
+        inner.tl_func = TLFunction(FunctionType.loss, name)
         return inner
 
     return decorating_function

code_loader-1.0.65.dev1/code_loader/inner_leap_binder/mapping.py

@@ -0,0 +1,219 @@
+from typing import Optional, Callable, Any, Dict, List, Set, OrderedDict
+from copy import deepcopy
+import numpy as np
+from collections import OrderedDict as od
+from code_loader.inner_leap_binder.inner_classes import FunctionType, TLFunction, MappingList
+from code_loader.contract.datasetclasses import DatasetIntegrationSetup, VisualizerHandler, MetricHandler
+from code_loader.contract.datasetclasses import DatasetBaseHandler, GroundTruthHandler, InputHandler
+
+
+class LeapMapping:
+
+    def __init__(self, setup_container: DatasetIntegrationSetup, mapping: MappingList):
+        self.container : DatasetIntegrationSetup = setup_container
+        self.mapping: MappingList = mapping
+        self.max_id = 20000
+        self.current_id = self.max_id
+        self.mapping_json = []
+        self.handle_dict = {
+            FunctionType.gt : self.container.ground_truths,
+            FunctionType.metric : self.container.metrics,
+            FunctionType.visualizer: self.container.visualizers,
+            FunctionType.input: self.container.inputs,
+        }
+        self.handle_to_op_type = {
+            GroundTruthHandler: "GroundTruth",
+            InputHandler: "Input",
+            VisualizerHandler: "Visualizer"
+
+        }
+        self.function_to_json : Dict[TLFunction, List[Dict[str, Any]]] = {}
+
+    def create_mapping_list(self, connections_list): #TODO add types
+        pass
+
+    def create_connect_nodes(self):
+        for element in self.mapping:
+            for sink in element.keys():
+                for node_name, connected_func_list in element[sink].items():
+                    tl_connected_functions = [ connected_func.tl_func for connected_func in connected_func_list
+                                               if not isinstance(connected_func, int)]
+                    for i, source in enumerate(tl_connected_functions):
+                        print(sink, source, node_name)
+                        sink_node = self.create_node(sink, name=node_name)
+                        self.add_connections(sink_node, source, input_idx=i)
+        print(1)
+        import yaml
+        with open('temp.yaml', 'w') as yf:
+            yaml.safe_dump(({'decoder': self.mapping_json}), yf, default_flow_style=False, sort_keys=False)
+
+    def create_partial_mapping(self): #TODO add types
+        referenced = self._get_referenced_sinks()
+        for func in referenced:
+            kwargs = {}
+            if func.FuncType == FunctionType.model_output:
+                kwargs = {'name': func.FuncName}
+            self.create_node(func, **kwargs)
+        self.create_connect_nodes()
+
+    def _get_referenced_sinks(self) -> Set[TLFunction]:
+        referenced : Set[TLFunction] = set()
+        references_types = {FunctionType.input, FunctionType.gt, FunctionType.model_output}
+        for element in self.mapping:
+            for func in element.keys():
+                for node_name, connected_func_list in element[func].items():
+                    tl_connected_functions = [ connected_func.tl_func for connected_func in connected_func_list
+                                               if not isinstance(connected_func, int)]
+                    for connected_func in tl_connected_functions:
+                        if connected_func.FuncType in references_types:
+                            referenced.add(connected_func)
+        return referenced
+
+    def create_gt_input(self, handle: DatasetBaseHandler, **kwargs) -> Dict[str, Any]:
+        op_type = self.handle_to_op_type[type(handle)]
+        json_node = {'operation': op_type,
+                     'data': {'type': op_type, 'output_name': handle.name},
+                     'id': f'{str(self.current_id)}',
+                     'inputs': {},
+                     'outputs': {handle.name: []}}
+        return json_node
+
+    def create_visualize(self, visualize_handle: VisualizerHandler ,**kwargs) -> Dict[str, Any]:
+        op_type = "Visualizer"
+        json_node = {'operation': op_type,
+                     'data': {'type': op_type,
+                              'name': visualize_handle.name,
+                              'visualizer_name': visualize_handle.name,
+                              'visualizer_type': visualize_handle.type.name,
+                              'arg_names': deepcopy(visualize_handle.arg_names),
+                              'user_unique_name': kwargs['name']},
+                     'id': f'{str(self.current_id)}',
+                     'inputs': {arg: [] for arg in visualize_handle.arg_names},
+                     'outputs': {} }
+        return json_node
+
+    def create_metric(self, metric_handle: MetricHandler, **kwargs) -> Dict[str, Any]:
+        op_type = "Metric"
+        json_node = {'operation': op_type,
+                     'data': {'type': op_type,
+                              'name': metric_handle.name,
+                              'metric_name': metric_handle.name,
+                              'arg_names': deepcopy(metric_handle.arg_names),
+                              'user_unique_name': kwargs['name']},
+                     'id': f'{str(self.current_id)}',
+                     'inputs': {arg: [] for arg in metric_handle.arg_names},
+                     'outputs': {} }
+        return  json_node
+
+
+    def create_temp_model_node(self, node_type: TLFunction, **kwargs) -> Dict[str, Any]:
+        default_name = self.get_default_name(node_type.FuncType, kwargs['name'])
+        op_type = "PlaceHolder"
+        json_node = {'operation': op_type,
+                     'data': {'type': op_type,
+                              'name': default_name,
+                              'arg_names': [default_name],
+                              'user_unique_name': default_name},
+                     'id': f'{default_name}',
+                     'inputs': {},
+                     'outputs': {} }
+        if node_type.FuncType == FunctionType.model_input:
+            json_node['inputs'] = {default_name: []}
+        else:
+            json_node['outputs'] = {default_name: []}
+        return json_node
+
+    def _find_handle(self, node: TLFunction) -> DatasetBaseHandler:
+        handle_list = self.handle_dict[node.FuncType]
+        chosen_handle: Optional[DatasetBaseHandler] = None
+        for handle in handle_list:
+            if handle.name == node.FuncName:
+                chosen_handle = handle
+        assert chosen_handle is not None
+        return chosen_handle
+
+    def _find_json_node(self, node: TLFunction, **kwargs) -> Dict[str, Any]:
+        candidates = self.function_to_json[node]
+        if len(candidates) == 1:
+            return candidates[0]
+        elif len(candidates) > 1:
+            raise Exception("More than 1 candidate - solve")
+        else:
+            raise Exception(f"No json node creted for node {node}")
+
+    def _get_node_id(self, node: TLFunction, **kwargs):
+        node = self._find_json_node(node, **kwargs)
+        return node['id']
+
+    def create_node(self, node_type: TLFunction, **kwargs) -> Dict[str, Any]:
+        if node_type.FuncType in [FunctionType.model_output, FunctionType.model_input]:
+            json_node = self.create_temp_model_node(node_type, **kwargs)
+        else:
+            node_creator_map: Dict[FunctionType, Callable[..., Any]] = { FunctionType.gt : self.create_gt_input,
+                                                                         FunctionType.input: self.create_gt_input,
+                                                                         FunctionType.visualizer: self.create_visualize,
+                                                                         FunctionType.metric: self.create_metric}
+            handle = self._find_handle(node_type)
+            json_node = node_creator_map[node_type.FuncType](handle, **kwargs)
+            self.current_id += 1
+        self.function_to_json[node_type] =  self.function_to_json.get(node_type, []) + [json_node]
+        self.mapping_json.append(json_node)
+        return  json_node
+
+    def get_default_name(self, f_type: FunctionType, name: str):
+        return f"{str(f_type)}_{name}"
+
+    def _get_node_output_name(self, node: TLFunction):
+        if node.FuncType not in {FunctionType.model_output, FunctionType.model_input}:
+            from_handle = self._find_handle(node)
+            output_name = from_handle.name
+        else:
+            output_name = self.get_default_name(node.FuncType, node.FuncName)
+        return output_name
+
+
+    def add_connections(self, to_json: Dict[str, Any], from_node: TLFunction, input_idx: int):
+        try:
+            input_name = to_json['data']['arg_names'][input_idx]
+        except Exception as e:
+            raise Exception(f"arg names mismatch when adding input for {to_json['operation']}")
+        output_key = self._get_node_output_name(from_node)
+        from_json = self._find_json_node(from_node)
+        operation = from_json['operation']
+        from_id = from_json['id']
+        # Add Input
+        to_json['inputs'][input_name].append({'outputKey': output_key,
+                                              'operation': operation,
+                                              'id': from_id
+                                              })
+        # Add Output
+        from_json['outputs'][output_key].append({'inputKey': input_name,
+                                              'operation': to_json['operation'],
+                                              'id': to_json['id']
+                                              })
+
+    def add_output(self, input_node, output_node):
+        pass
+
+def create_mapping(model_parser_map, connections_list):
+
+    referenced_gt = []
+    referenced_inputs = []
+    for element in connections_list:
+        for func in element.keys():
+            for node_name, connected_func_list in element[func].items():
+                tl_connected_functions = [ connected_func.tl_func if not isinstance(connected_func, int) else connected_func
+                                      for connected_func in connected_func_list ]
+
+
+
+
+
+    # Check Max Node ID
+    # Get Preds Node ID
+
+    # TODO Create Inputs (if referenced)
+    # TODO Create Gts (if referenced)
+    # TODO Create Visualizers
+    # TODO Create Metrics
+    # TODO Create losses

{code_loader-1.0.65 → code_loader-1.0.65.dev1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "code-loader"
-version = "1.0.65"
+version = "1.0.65.dev1"
 description = ""
 authors = ["dorhar <doron.harnoy@tensorleap.ai>"]
 license = "MIT"

code_loader-1.0.65/code_loader/code_inegration_processes_manager.py

@@ -1,83 +0,0 @@
-# mypy: ignore-errors
-import traceback
-from dataclasses import dataclass
-
-from typing import List, Tuple, Optional
-
-from multiprocessing import Process, Queue
-
-from code_loader.leap_loader_parallelized_base import LeapLoaderParallelizedBase
-from code_loader.leaploader import LeapLoader
-from code_loader.contract.enums import DataStateEnum
-from code_loader.metric_calculator_parallelized import MetricCalculatorParallelized
-from code_loader.samples_generator_parallelized import SamplesGeneratorParallelized
-
-
-@dataclass
-class SampleSerializableError:
-    state: DataStateEnum
-    index: int
-    leap_script_trace: str
-    exception_as_str: str
-
-
-class CodeIntegrationProcessesManager:
-    def __init__(self, code_path: str, code_entry_name: str, n_workers: Optional[int] = 2,
-                 max_samples_in_queue: int = 128) -> None:
-        self.metric_calculator_parallelized = MetricCalculatorParallelized(code_path, code_entry_name)
-        self.samples_generator_parallelized = SamplesGeneratorParallelized(code_path, code_entry_name)
-
-    def _create_and_start_process(self) -> Process:
-        process = self.multiprocessing_context.Process(
-            target=CodeIntegrationProcessesManager._process_func,
-            args=(self.code_path, self.code_entry_name, self._inputs_waiting_to_be_process,
-                  self._ready_processed_results))
-        process.daemon = True
-        process.start()
-        return process
-
-    def _run_and_warm_first_process(self):
-        process = self._create_and_start_process()
-        self.processes = [process]
-
-        # needed in order to make sure the preprocess func runs once in nonparallel
-        self._start_process_inputs([(DataStateEnum.training, 0)])
-        self._get_next_ready_processed_result()
-
-    def _operation_decider(self):
-        if self.metric_calculator_parallelized._ready_processed_results.empty() and not \
-            self.metric_calculator_parallelized._inputs_waiting_to_be_process.empty():
-            return 'metric'
-
-        if self.samples_generator_parallelized._ready_processed_results.empty() and not \
-            self.samples_generator_parallelized._inputs_waiting_to_be_process.empty():
-            return 'dataset'
-
-
-
-
-    @staticmethod
-    def _process_func(code_path: str, code_entry_name: str,
-                      samples_to_process: Queue, ready_samples: Queue,
-                      metrics_to_process: Queue, ready_metrics: Queue) -> None:
-        import os
-        os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
-
-        leap_loader = LeapLoader(code_path, code_entry_name)
-        while True:
-
-            # decide on sample or metric to process
-            state, idx = samples_to_process.get(block=True)
-            leap_loader._preprocess_result()
-            try:
-                sample = leap_loader.get_sample(state, idx)
-            except Exception as e:
-                leap_script_trace = traceback.format_exc().split('File "<string>"')[-1]
-                ready_samples.put(SampleSerializableError(state, idx, leap_script_trace, str(e)))
-                continue
-
-            ready_samples.put(sample)
-
-    def generate_samples(self, sample_identities: List[Tuple[DataStateEnum, int]]):
-        return self.start_process_inputs(sample_identities)
-