feat: get complete lut

Cargo.toml

@@ -24,7 +24,7 @@ async-lock = "3.4.0"
 numpy = "0.26"
 num-traits = "0.2"
 delegate = "0.13"
-omfiles-rs = { git = "https://github.com/open-meteo/rust-omfiles", rev = "91c7a1bfc01c5394332eb1b9b9a035f61554bf07", package = "omfiles", features = ["metadata-tree"] }
+omfiles-rs = { git = "https://github.com/open-meteo/rust-omfiles", rev = "c7e6d4daed18e05bc8bb5d4373185096a493a662", package = "omfiles", features = ["metadata-tree"] }
 om-file-format-sys = { git = "https://github.com/open-meteo/om-file-format", rev = "a484a5a90914ff934fc451dff30b671b9ba8e38e" }
 thiserror = "2.0.12"
 

pyproject.toml

@@ -87,6 +87,9 @@ ignore = [
 [project.entry-points."xarray.backends"]
 om = "omfiles.xarray:OmXarrayEntrypoint"
 
+[project.entry-points."numcodecs.codecs"]
+turbo_pfor = "omfiles._numcodecs:TurboPfor"
+
 [project.entry-points."zarr.codecs"]
 "omfiles.pfor" = "omfiles._zarr3:PforCodec"
 "omfiles.pfor_serializer" = "omfiles._zarr3:PforSerializer"

python/omfiles/_numcodecs.py

@@ -49,6 +49,3 @@ def decode(self, buf, out=None):  # type: ignore
             buf = buf
 
         return self._impl.decode_array(buf, np.dtype(self.dtype), self.chunk_elements)
-
-
-numcodecs.register_codec(TurboPfor)

python/omfiles/omfiles.pyi

@@ -184,6 +184,16 @@ class OmFileReader:
 
         It is safe to call this method multiple times.
         """
+    def get_complete_lut(self) -> builtins.list[builtins.int]:
+        r"""
+        Retrieve the complete lookup table for the variable.
+
+        The lookup table is a monotonically increasing array of u64 values containing
+        n+1 elements, where n is the number of chunks. Each value represents the absolute
+        offset in the file of the end of the chunk, and the first value is the start offset
+        of the first chunk. The size of chunk j can be calculated as lut[j+1] - lut[j]
+        using zero-based indexing.
+        """
     def get_child_by_index(self, index: builtins.int) -> OmFileReader:
         r"""
         Get a child reader at the specified index.

src/codecs.rs

@@ -2,6 +2,7 @@ use numpy::{
     ndarray, IntoPyArray, PyArray1, PyArrayDescr, PyArrayDescrMethods, PyArrayDyn, PyArrayMethods,
     PyUntypedArray, PyUntypedArrayMethods,
 };
+use omfiles_rs::_om_file_format_sys as om_file_format_sys;
 use pyo3::exceptions::{PyTypeError, PyValueError};
 use pyo3::prelude::*;
 use pyo3::types::PyBytes;

src/reader.rs

@@ -356,6 +356,24 @@ impl OmFileReader {
         self.with_reader(|reader| Ok(reader.data_type().is_scalar()))
     }
 
+    /// Retrieve the complete lookup table for the variable.
+    ///
+    /// The lookup table is a monotonically increasing array of u64 values containing
+    /// n+1 elements, where n is the number of chunks. Each value represents the absolute
+    /// offset in the file of the end of the chunk, and the first value is the start offset
+    /// of the first chunk. The size of chunk j can be calculated as lut[j+1] - lut[j]
+    /// using zero-based indexing.
+    fn get_complete_lut(&self) -> PyResult<Vec<u64>> {
+        let lut = self.with_reader(|reader| {
+            reader
+                .expect_array()
+                .map_err(convert_omfilesrs_error)?
+                .get_complete_lut()
+                .map_err(convert_omfilesrs_error)
+        })?;
+        Ok(lut)
+    }
+
     /// Check if the variable is a group (a variable with data type None).
     ///
     /// Returns: