Add function to get extra dims from a dtype & fix unpacking by using it

docs/basic.rst

@@ -14,6 +14,7 @@ that you can use to access to the data.
 .. code:: python
 
     las = pylas.read('somefile.las')
+    print(np.unique(las.classification))
 
     import s3fs
     fs = s3fs.S3FileSystem()
@@ -27,7 +28,6 @@ As the name suggest, this function does not read the whole file, but opens it an
 This is useful if you only need to read the header without loading the whole file in memory.
 
 
-
 Converting
 ==========
 

pylas/point/dims.py

@@ -401,3 +401,16 @@ def is_point_fmt_compatible_with_version(point_format_id, file_version):
         return point_format_id in VERSION_TO_POINT_FMT[str(file_version)]
     except KeyError:
         raise errors.FileVersionNotSupported(file_version)
+
+
+def is_official_dimension(dimension_name, point_fmt):
+    official_names_for_fmt = set(get_dtype_of_format_id(point_fmt, unpacked=True).names)
+    return dimension_name in official_names_for_fmt
+
+
+def get_extra_dimensions_spec(np_dtype, point_format_id):
+    extra_dims_names = [
+        name for name in np_dtype.names
+        if not is_official_dimension(name, point_format_id) and name not in COMPOSED_FIELDS[point_format_id]
+    ]
+    return [(name, np_dtype[name]) for name in extra_dims_names]

pylas/point/packing.py

@@ -2,7 +2,7 @@
 """
 import numpy as np
 
-from .dims import get_dtype_of_format_id, COMPOSED_FIELDS
+from .dims import get_dtype_of_format_id, COMPOSED_FIELDS, get_extra_dimensions_spec
 
 
 def least_significant_bit(val):
@@ -71,9 +71,11 @@ def unpack_sub_fields(data, point_format_id, extra_dims=None):
     Returns:
         A new structured array with the sub-fields de-packed
     """
+    composed_dims = COMPOSED_FIELDS[point_format_id]
+    if extra_dims is None:
+        extra_dims = get_extra_dimensions_spec(data.dtype, point_format_id)
     dtype = get_dtype_of_format_id(
         point_format_id, extra_dims=extra_dims, unpacked=True)
-    composed_dims = COMPOSED_FIELDS[point_format_id]
     point_record = np.zeros_like(data, dtype)
 
     for dim_name in data.dtype.names:

pylastests/test_extrabytes.py

@@ -15,6 +15,16 @@ def las1_4():
     return pylas.read(test1_4_las)
 
 
+def test_extra_dim_spec():
+    extra_dims_specs = [('codification', 'u4')]
+    dtype = pylas.point.dims.get_dtype_of_format_id(0)
+    dtype = pylas.point.dims.dtype_append(dtype, extra_dims_specs)
+
+    found_extra_dims_spec = pylas.point.dims.get_extra_dimensions_spec(dtype, 0)
+
+    assert found_extra_dims_spec == extra_dims_specs
+
+
 def test_extra_names(extrab_las):
     all_dims = set(extrab_las.points_data.array.dtype.names)