code-loader 1.0.50__tar.gz → 1.0.51__tar.gz

{code_loader-1.0.50 → code_loader-1.0.51}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: code-loader
-Version: 1.0.50
+Version: 1.0.51
 Summary: 
 Home-page: https://github.com/tensorleap/code-loader
 License: MIT
@@ -13,6 +13,7 @@ Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
+Requires-Dist: matplotlib (>=3.3,<3.4)
 Requires-Dist: numpy (>=1.22.3,<2.0.0)
 Requires-Dist: psutil (>=5.9.5,<6.0.0)
 Requires-Dist: pyyaml (>=6.0.2,<7.0.0)

code_loader-1.0.51/code_loader/code_inegration_processes_manager.py

@@ -0,0 +1,83 @@
+# 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)
+

{code_loader-1.0.50 → code_loader-1.0.51}/code_loader/contract/datasetclasses.py

@@ -32,21 +32,39 @@ class PreprocessResponse:
         }
         response = PreprocessResponse(length=len(preprocessed_data), data=preprocessed_data)
     """
-    length: int
-    data: Any
-
-
-SectionCallableInterface = Callable[[int, PreprocessResponse], npt.NDArray[np.float32]]
+    length: Optional[int] = None  # Deprecated. Please use sample_ids instead
+    data: Any = None
+    sample_ids: Optional[Union[List[str], List[int]]] = None
+    state: Optional[DataStateType] = None
+    sample_id_type: Optional[Union[Type[str], Type[int]]] = None
+
+    def __post_init__(self) -> None:
+        if self.length is not None and self.sample_ids is None:
+            self.sample_ids = [i for i in range(self.length)]
+            self.sample_id_type = int
+        elif self.length is None and self.sample_ids is not None:
+            self.length = len(self.sample_ids)
+            if self.sample_id_type is None:
+                self.sample_id_type = str
+        else:
+            raise Exception("length is deprecated.")
+
+    def __len__(self) -> int:
+        assert self.sample_ids is not None
+        return len(self.sample_ids)
+
+
+SectionCallableInterface = Callable[[Union[int, str], PreprocessResponse], npt.NDArray[np.float32]]
 
 MetadataSectionCallableInterface = Union[
-    Callable[[int, PreprocessResponse], int],
-    Callable[[int, PreprocessResponse], Dict[str, int]],
-    Callable[[int, PreprocessResponse], str],
-    Callable[[int, PreprocessResponse], Dict[str, str]],
-    Callable[[int, PreprocessResponse], bool],
-    Callable[[int, PreprocessResponse], Dict[str, bool]],
-    Callable[[int, PreprocessResponse], float],
-    Callable[[int, PreprocessResponse], Dict[str, float]]
+    Callable[[Union[int, str], PreprocessResponse], int],
+    Callable[[Union[int, str], PreprocessResponse], Dict[str, int]],
+    Callable[[Union[int, str], PreprocessResponse], str],
+    Callable[[Union[int, str], PreprocessResponse], Dict[str, str]],
+    Callable[[Union[int, str], PreprocessResponse], bool],
+    Callable[[Union[int, str], PreprocessResponse], Dict[str, bool]],
+    Callable[[Union[int, str], PreprocessResponse], float],
+    Callable[[Union[int, str], PreprocessResponse], Dict[str, float]]
 ]
 
 
@@ -181,5 +199,5 @@ class DatasetSample:
     inputs: Dict[str, npt.NDArray[np.float32]]
     gt: Optional[Dict[str, npt.NDArray[np.float32]]]
     metadata: Dict[str, Union[str, int, bool, float]]
-    index: int
+    index: Union[int, str]
     state: DataStateEnum

{code_loader-1.0.50 → code_loader-1.0.51}/code_loader/inner_leap_binder/leapbinder.py

@@ -389,17 +389,36 @@ class LeapBinder:
         if preprocess is None:
             raise Exception("Please make sure you call the leap_binder.set_preprocess method")
         preprocess_results = preprocess.function()
-        preprocess_result_dict = {
-            DataStateEnum(i): preprocess_result
-            for i, preprocess_result in enumerate(preprocess_results)
-        }
+        preprocess_result_dict = {}
+        for i, preprocess_result in enumerate(preprocess_results):
+            if preprocess_result.state is None:
+                state_enum = DataStateEnum(i)
+                preprocess_result.state = DataStateType(state_enum.name)
+            else:
+                state_enum = DataStateEnum[preprocess_result.state.name]
 
-        unlabeled_preprocess = self.setup_container.unlabeled_data_preprocess
-        if unlabeled_preprocess is not None:
-            preprocess_result_dict[DataStateEnum.unlabeled] = unlabeled_preprocess.function()
+            if state_enum in preprocess_result_dict:
+                raise Exception(f"Duplicate state {state_enum.name} in preprocess results")
+            preprocess_result_dict[state_enum] = preprocess_result
+
+        if DataStateEnum.unlabeled not in preprocess_result_dict:
+            preprocess_unlabeled_result = self.get_preprocess_unlabeled_result()
+            if preprocess_unlabeled_result is not None:
+                preprocess_result_dict[DataStateEnum.unlabeled] = preprocess_unlabeled_result
+
+        if DataStateEnum.training not in preprocess_result_dict:
+            raise Exception("Training data is required")
+        if DataStateEnum.validation not in preprocess_result_dict:
+            raise Exception("Validation data is required")
 
         return preprocess_result_dict
 
+    def get_preprocess_unlabeled_result(self) -> Optional[PreprocessResponse]:
+        unlabeled_preprocess = self.setup_container.unlabeled_data_preprocess
+        if unlabeled_preprocess is not None:
+            return unlabeled_preprocess.function()
+        return None
+
     def _get_all_dataset_base_handlers(self) -> List[Union[DatasetBaseHandler, MetadataHandler]]:
         all_dataset_base_handlers: List[Union[DatasetBaseHandler, MetadataHandler]] = []
         all_dataset_base_handlers.extend(self.setup_container.inputs)
@@ -411,7 +430,8 @@ class LeapBinder:
     def check_handler(
             preprocess_response: PreprocessResponse, test_result: List[DatasetTestResultPayload],
             dataset_base_handler: Union[DatasetBaseHandler, MetadataHandler]) -> List[DatasetTestResultPayload]:
-        raw_result = dataset_base_handler.function(0, preprocess_response)
+        assert preprocess_response.sample_ids is not None
+        raw_result = dataset_base_handler.function(preprocess_response.sample_ids[0], preprocess_response)
         handler_type = 'metadata' if isinstance(dataset_base_handler, MetadataHandler) else None
         if isinstance(dataset_base_handler, MetadataHandler) and isinstance(raw_result, dict):
             metadata_test_result_payloads = [

{code_loader-1.0.50 → code_loader-1.0.51}/code_loader/leaploader.py

@@ -2,10 +2,11 @@
 import importlib.util
 import io
 import sys
+import time
 from contextlib import redirect_stdout
 from functools import lru_cache
 from pathlib import Path
-from typing import Dict, List, Iterable, Union, Any
+from typing import Dict, List, Iterable, Union, Any, Type
 
 import numpy as np
 import numpy.typing as npt
@@ -27,6 +28,8 @@ class LeapLoader:
         self.code_entry_name = code_entry_name
         self.code_path = code_path
 
+        self._preprocess_result_cached = None
+
     @lru_cache()
     def exec_script(self) -> None:
         try:
@@ -103,12 +106,16 @@ class LeapLoader:
             for prediction_type in setup.prediction_types
         }
 
-    def get_sample(self, state: DataStateEnum, idx: int) -> DatasetSample:
+    def get_sample(self, state: DataStateEnum, sample_id: Union[int, str]) -> DatasetSample:
         self.exec_script()
-        sample = DatasetSample(inputs=self._get_inputs(state, idx),
-                               gt=None if state == DataStateEnum.unlabeled else self._get_gt(state, idx),
-                               metadata=self._get_metadata(state, idx),
-                               index=idx,
+        preprocess_result = self._preprocess_result()
+        if state == DataStateEnum.unlabeled and sample_id not in preprocess_result[state].sample_ids:
+            self._preprocess_result(update_unlabeled_preprocess=True)
+
+        sample = DatasetSample(inputs=self._get_inputs(state, sample_id),
+                               gt=None if state == DataStateEnum.unlabeled else self._get_gt(state, sample_id),
+                               metadata=self._get_metadata(state, sample_id),
+                               index=sample_id,
                                state=state)
         return sample
 
@@ -148,6 +155,13 @@ class LeapLoader:
         test_result = DatasetTestResultPayload('preprocess')
         try:
             preprocess_result = self._preprocess_result()
+            if self.get_sample_id_type() is str:
+                max_allowed_item_size = np.dtype('<U256').itemsize
+                for state, preprocess_response in preprocess_result.items():
+                    sample_ids_array = np.array(preprocess_response.sample_ids)
+                    if sample_ids_array.dtype.itemsize > max_allowed_item_size:
+                        raise Exception(f"Sample id are too long. Max allowed length is 256 charecters.")
+
             global_leap_binder.check_preprocess(preprocess_result)
         except Exception as e:
             line_number, file_name, stacktrace = get_root_exception_file_and_line_number()
@@ -279,27 +293,42 @@ class LeapLoader:
         ]
         return ModelSetup(custom_layer_instances)
 
-    @lru_cache()
-    def _preprocess_result(self) -> Dict[DataStateEnum, PreprocessResponse]:
+    def _preprocess_result(self, update_unlabeled_preprocess=False) -> Dict[DataStateEnum, PreprocessResponse]:
         self.exec_script()
-        return global_leap_binder.get_preprocess_result()
+
+        if self._preprocess_result_cached is None:
+            self._preprocess_result_cached = global_leap_binder.get_preprocess_result()
+
+        if update_unlabeled_preprocess:
+            self._preprocess_result_cached[
+                DataStateEnum.unlabeled] = global_leap_binder.get_preprocess_unlabeled_result()
+
+        return self._preprocess_result_cached
+
+    def get_preprocess_sample_ids(self, update_unlabeled_preprocess=False) -> Dict[DataStateEnum, Union[List[int], List[str]]]:
+        preprocess_result = self._preprocess_result(update_unlabeled_preprocess)
+        sample_ids = {}
+        for state, preprocess_response in preprocess_result.items():
+            sample_ids[state] = preprocess_response.sample_ids
+
+        return sample_ids
 
     def _get_dataset_handlers(self, handlers: Iterable[DatasetBaseHandler],
-                              state: DataStateEnum, idx: int) -> Dict[str, npt.NDArray[np.float32]]:
+                              state: DataStateEnum, sample_id: Union[int, str]) -> Dict[str, npt.NDArray[np.float32]]:
         result_agg = {}
         preprocess_result = self._preprocess_result()
         preprocess_state = preprocess_result[state]
         for handler in handlers:
-            handler_result = handler.function(idx, preprocess_state)
+            handler_result = handler.function(sample_id, preprocess_state)
             handler_name = handler.name
             result_agg[handler_name] = handler_result
         return result_agg
 
-    def _get_inputs(self, state: DataStateEnum, idx: int) -> Dict[str, npt.NDArray[np.float32]]:
-        return self._get_dataset_handlers(global_leap_binder.setup_container.inputs, state, idx)
+    def _get_inputs(self, state: DataStateEnum, sample_id: Union[int, str]) -> Dict[str, npt.NDArray[np.float32]]:
+        return self._get_dataset_handlers(global_leap_binder.setup_container.inputs, state, sample_id)
 
-    def _get_gt(self, state: DataStateEnum, idx: int) -> Dict[str, npt.NDArray[np.float32]]:
-        return self._get_dataset_handlers(global_leap_binder.setup_container.ground_truths, state, idx)
+    def _get_gt(self, state: DataStateEnum, sample_id: Union[int, str]) -> Dict[str, npt.NDArray[np.float32]]:
+        return self._get_dataset_handlers(global_leap_binder.setup_container.ground_truths, state, sample_id)
 
     @lru_cache()
     def _metadata_name_to_type(self) -> Dict[str, DatasetMetadataType]:
@@ -334,12 +363,12 @@ class LeapLoader:
 
         return converted_value
 
-    def _get_metadata(self, state: DataStateEnum, idx: int) -> Dict[str, Union[str, int, bool, float]]:
+    def _get_metadata(self, state: DataStateEnum, sample_id: Union[int, str]) -> Dict[str, Union[str, int, bool, float]]:
         result_agg = {}
         preprocess_result = self._preprocess_result()
         preprocess_state = preprocess_result[state]
         for handler in global_leap_binder.setup_container.metadata:
-            handler_result = handler.function(idx, preprocess_state)
+            handler_result = handler.function(sample_id, preprocess_state)
             if isinstance(handler_result, dict):
                 for single_metadata_name, single_metadata_result in handler_result.items():
                     handler_name = f'{handler.name}_{single_metadata_name}'
@@ -349,3 +378,14 @@ class LeapLoader:
                 result_agg[handler_name] = self._convert_metadata_to_correct_type(handler_name, handler_result)
 
         return result_agg
+
+    @lru_cache()
+    def get_sample_id_type(self) -> Type:
+        preprocess_results = list(self._preprocess_result().values())
+        id_type = preprocess_results[0].sample_id_type
+        for preprocess_result in preprocess_results:
+            if preprocess_result.sample_id_type != id_type:
+                raise Exception("Different id types in preprocess results")
+
+        return id_type
+

{code_loader-1.0.50 → code_loader-1.0.51}/code_loader/utils.py

@@ -10,7 +10,7 @@ from code_loader.contract.datasetclasses import SectionCallableInterface, Prepro
 
 
 def to_numpy_return_wrapper(encoder_function: SectionCallableInterface) -> SectionCallableInterface:
-    def numpy_encoder_function(idx: int, samples: PreprocessResponse) -> npt.NDArray[np.float32]:
+    def numpy_encoder_function(idx: Union[int, str], samples: PreprocessResponse) -> npt.NDArray[np.float32]:
         result = encoder_function(idx, samples)
         numpy_result: npt.NDArray[np.float32] = np.array(result)
         return numpy_result

{code_loader-1.0.50 → code_loader-1.0.51}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "code-loader"
-version = "1.0.50"
+version = "1.0.51"
 description = ""
 authors = ["dorhar <doron.harnoy@tensorleap.ai>"]
 license = "MIT"
@@ -15,6 +15,7 @@ include = [
 python = ">=3.8,<3.12"
 numpy = "^1.22.3"
 psutil = "^5.9.5"
+matplotlib = ">=3.3,<3.4"
 requests = "^2.32.3"
 pyyaml = "^6.0.2"