start extraction of binary file - not finished

paint/util/binary_to_h5.py

@@ -0,0 +1,156 @@
+import struct
+from pathlib import Path
+from typing import Tuple, Union
+
+import h5py
+import torch
+
+
+class BinaryToH5Converter:
+    """
+    Implement a converter that converts binary data to HDF5 format.
+
+    Attributes
+    ----------
+    input_file_path : str
+        The file path to the binary data file that will be converted.
+    output_file_path : str
+        The file path to save the converted h5 file.
+    surface_header_name : str
+        The name for the surface header in the binary file.
+    facet_header_name : str
+        The name for the facet header in the binary file.
+    points_on_facet_struct_name : str
+        The name of the point on facet structure in the binary file.
+
+    Methods
+    -------
+    nwu_to_enu()
+        Cast from an NWU to an ENU coordinate system.
+    convert_to_h5_and_extract_properties()
+        Convert binary data to h5 and extract heliostat properties not to be saved in the deflectometry file.
+    """
+
+    def __init__(
+        self,
+        input_file_path: Union[str, Path],
+        output_file_path: Union[str, Path],
+        surface_header_name: str,
+        facet_header_name: str,
+        points_on_facet_struct_name: str,
+    ) -> None:
+        """
+        Initialize the converter.
+
+        Parameters
+        ----------
+        input_file_path : str
+            The file path to the binary data file that will be converted.
+        output_file_path : str
+            The file path to save the converted h5 file.
+        surface_header_name : str
+            The name for the surface header in the binary file.
+        facet_header_name : str
+            The name for the facet header in the binary file.
+        points_on_facet_struct_name : str
+            The name of the point on facet structure in the binary file.
+        """
+        self.input_file_path = Path(input_file_path)
+        self.output_file_path = Path(output_file_path)
+        self.surface_header_name = surface_header_name
+        self.facet_header_name = facet_header_name
+        self.points_on_facet_struct_name = points_on_facet_struct_name
+
+    @staticmethod
+    def nwu_to_enu(nwu_tensor: torch.Tensor) -> torch.Tensor:
+        """
+        Cast the coordinate system from NWU to ENU.
+
+        Parameters
+        ----------
+        nwu_tensor : torch.Tensor
+            The tensor in the NWU coordinate system.
+
+        Returns
+        -------
+        torch.Tensor
+            The converted tensor in the ENU coordinate system.
+        """
+        return torch.tensor(
+            [-nwu_tensor[1], nwu_tensor[0], nwu_tensor[2]], dtype=torch.float
+        )
+
+    def convert_to_h5_and_extract_properties(
+        self,
+    ) -> Tuple[int, torch.Tensor, torch.Tensor, torch.Tensor]:
+        """Convert binary data to h5 and extract heliostat properties not to be saved in the deflectometry file."""
+        # Create structures for reading binary file correctly.
+        surface_header_struct = struct.Struct(self.surface_header_name)
+        facet_header_struct = struct.Struct(self.facet_header_name)
+        points_on_facet_struct = struct.Struct(self.points_on_facet_struct_name)
+
+        with open(f"{self.input_file_path}.binp", "rb") as file:
+            surface_header_data = surface_header_struct.unpack_from(
+                file.read(surface_header_struct.size)
+            )
+            # Load width and height.
+            width, height = surface_header_data[3:5]
+
+            # Calculate the number of facets.
+            n_xy = surface_header_data[5:7]
+            number_of_facets = int(n_xy[0] * n_xy[1])
+
+            # Create empty tensors for storing data.
+            facet_translation_vectors = torch.empty(number_of_facets, 3)
+            canting_e = torch.empty(number_of_facets, 3)
+            canting_n = torch.empty(number_of_facets, 3)
+            surface_points_with_facets = torch.empty(0)
+            surface_normals_with_facets = torch.empty(0)
+            for f in range(number_of_facets):
+                facet_header_data = facet_header_struct.unpack_from(
+                    file.read(facet_header_struct.size)
+                )
+
+                facet_translation_vectors[f] = torch.tensor(
+                    facet_header_data[1:4], dtype=torch.float
+                )
+                canting_n[f] = self.nwu_to_enu(
+                    torch.tensor(
+                        facet_header_data[4:7],
+                        dtype=torch.float,
+                    )
+                )
+                canting_e[f] = self.nwu_to_enu(
+                    torch.tensor(
+                        facet_header_data[7:10],
+                        dtype=torch.float,
+                    )
+                )
+                number_of_points = facet_header_data[10]
+                if f == 0:
+                    surface_points_with_facets = torch.empty(
+                        number_of_facets, number_of_points, 3
+                    )
+                    surface_normals_with_facets = torch.empty(
+                        number_of_facets, number_of_points, 3
+                    )
+
+                points_data = points_on_facet_struct.iter_unpack(
+                    file.read(points_on_facet_struct.size * number_of_points)
+                )
+                for i, point_data in enumerate(points_data):
+                    surface_points_with_facets[f, i, :] = torch.tensor(
+                        point_data[:3], dtype=torch.float
+                    )
+                    surface_normals_with_facets[f, i, :] = torch.tensor(
+                        point_data[3:6], dtype=torch.float
+                    )
+
+        # to maintain consistency we cast the west direction to east direction
+        canting_e[:, 0] = -canting_e[:, 0]
+
+        with h5py.File(self.output_file_path, "w") as file:
+            # TODO Save h5py
+            pass
+
+        return number_of_facets, facet_translation_vectors, canting_e, canting_n