Ndarray Npy

https://docs.rs/ndarray-npy/0.10.0/
Admin
This crate provides methods to read/write `ndarray`’s `ArrayBase` type from/to `.npy` and `.npz`...
Tokens:42,536
Snippets:400
Trust Score:9.1
Update:3 months ago
Show doc for...
Tokens:
Raw
### NpzWriter Example: Creating and Writing Arrays

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Demonstrates how to create a new .npz file and add multiple arrays of different dimensions to it. This example uses the default uncompressed writing.

```rust
use ndarray::{array, aview0, Array1, Array2};
use ndarray_npy::NpzWriter;
use std::fs::File;

let mut npz = NpzWriter::new(File::create("arrays.npz")?);
let a: Array2<i32> = array![[1, 2, 3], [4, 5, 6]];
let b: Array1<i32> = array![7, 8, 9];
npz.add_array("a", &a)?;
npz.add_array("b", &b)?;
npz.add_array("c", &aview0(&10))?;
npz.finish()?;
# Ok::<_, Box<dyn std::error::Error>>(())
```

--------------------------------

### Example: Viewing a Memory-Mapped .npy File

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/mod.rs.html

Demonstrates opening a readonly memory-mapped file as an ArrayView using the memmap2 crate and the view_npy method. This example requires the target to be little-endian and not running under Miri.

```rust
# // Miri doesn't support mmap, and the file is in little endian format.
# if !cfg!(miri) && cfg!(target_endian = "little") {
use memmap2::Mmap;
use ndarray::ArrayView2;
use ndarray_npy::ViewNpyExt;
use std::fs::File;

let file = File::open("resources/array.npy")?;
let mmap = unsafe { Mmap::map(&file)? };
let view = ArrayView2::<i32>::view_npy(&mmap)?;
# println!("view = {}", view);
# } 
# Ok::<_, Box<dyn std::error::Error>>(())
```

--------------------------------

### Example Usage of write_zeroed_npy

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.write_zeroed_npy.html

An example demonstrating how to use write_zeroed_npy to create a large zeroed .npy file and then memory-map it for modification.

```APIDOC
## Example

In this example, a file containing 64 GiB of zeroed `f64` elements is created. Then, an `ArrayViewMut` is created by memory-mapping the file. Modifications to the data in the `ArrayViewMut` will be applied to the backing file. This works even on systems with less than 64 GiB of physical memory. On filesystems which support sparse files, the disk space that’s actually used depends on how much data is modified.

```rust
use memmap2::MmapMut;
use ndarray::ArrayViewMut3;
use ndarray_npy::{write_zeroed_npy, ViewMutNpyExt};
use std::fs::{File, OpenOptions};

let path = "array.npy";

// Create a (sparse if supported) file containing 64 GiB of zeroed data.
let file = File::create(path)?;
write_zeroed_npy::<f64, _>(&file, (1024, 2048, 4096))?;

// Memory-map the file and create the mutable view.
let file = OpenOptions::new().read(true).write(true).open(path)?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
let mut view_mut = ArrayViewMut3::<f64>::view_mut_npy(&mut mmap)?;

// Modify an element in the view.
view_mut[[500, 1000, 2000]] = 3.14;
```
```

--------------------------------

### Example Usage of ViewNpyExt

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.ViewNpyExt.html

Demonstrates how to use the `view_npy` method to create a readonly memory-mapped ArrayView from an .npy file.

```APIDOC
```rust
use memmap2::Mmap;
use ndarray::ArrayView2;
use ndarray_npy::ViewNpyExt;
use std::fs::File;

let file = File::open("resources/array.npy")?;
let mmap = unsafe { Mmap::map(&file)? };
let view = ArrayView2::<i32>::view_npy(&mmap)?;
```
```

--------------------------------

### Example: Reading Arrays from NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Demonstrates how to open an NPZ file and read arrays by their names. Requires the 'ndarray' and 'ndarray-npy' crates.

```rust
use ndarray::{Array1, Array2};
use ndarray_npy::NpzReader;
use std::fs::File;

let mut npz = NpzReader::new(File::open("arrays.npz")?)?;
let a: Array2<i32> = npz.by_name("a")?;
let b: Array1<i32> = npz.by_name("b")?;
# Ok::<_, Box<dyn std::error::Error>>(())
```

--------------------------------

### Example: Opening a Writable Memory-Mapped File

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.ViewMutNpyExt.html

Demonstrates opening a memory-mapped file for reading and writing using `memmap2` and creating a mutable ArrayView using `view_mut_npy`. Changes to the view will modify the file.

```rust
use memmap2::MmapMut;
use ndarray::ArrayViewMut2;
use ndarray_npy::ViewMutNpyExt;
use std::fs;

let file = fs::OpenOptions::new()
    .read(true)
    .write(true)
    .open("resources/array.npy")?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
let view_mut = ArrayViewMut2::<i32>::view_mut_npy(&mut mmap)?;
```

--------------------------------

### Example: Writing an Array to a .npy File

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WriteNpyExt.html

Demonstrates how to use the write_npy method to save an Array2<i32> to a file named 'array.npy'. It's recommended to wrap the file writer in a BufWriter for performance.

```rust
use ndarray::{array, Array2};
use ndarray_npy::WriteNpyExt;
use std::fs::File;
use std::io::BufWriter;

let arr: Array2<i32> = array![[1, 2, 3], [4, 5, 6]];
let writer = BufWriter::new(File::create("array.npy")?);
arr.write_npy(writer)?;
```

--------------------------------

### Writing a Zeroed .npy File and Memory Mapping

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.write_zeroed_npy.html

This example demonstrates creating a large, zeroed .npy file and then memory-mapping it to create a mutable view. Modifications to the view are reflected in the file. This is useful for handling arrays larger than physical memory.

```rust
use memmap2::MmapMut;
use ndarray::ArrayViewMut3;
use ndarray_npy::{write_zeroed_npy, ViewMutNpyExt};
use std::fs::{File, OpenOptions};

let path = "array.npy";

// Create a (sparse if supported) file containing 64 GiB of zeroed data.
let file = File::create(path)?;
write_zeroed_npy::<f64, _>(&file, (1024, 2048, 4096))?;

// Memory-map the file and create the mutable view.
let file = OpenOptions::new().read(true).write(true).open(path)?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
let mut view_mut = ArrayViewMut3::<f64>::view_mut_npy(&mut mmap)?;

// Modify an element in the view.
view_mut[[500, 1000, 2000]] = 3.14;
```

--------------------------------

### View .npy file as ArrayView using memmap2

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.ViewNpyExt.html

Demonstrates opening a readonly memory-mapped .npy file and creating an ArrayView from its buffer. This example uses the memmap2 crate and requires the file to be present at 'resources/array.npy'.

```rust
use memmap2::Mmap;
use ndarray::ArrayView2;
use ndarray_npy::ViewNpyExt;
use std::fs::File;

let file = File::open("resources/array.npy")?;
let mmap = unsafe { Mmap::map(&file)? };
let view = ArrayView2::<i32>::view_npy(&mmap)?;
```

--------------------------------

### Create Mutable Array View from Buffer

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/mod.rs.html

Opens a writable memory-mapped file as an `ArrayViewMut`. Changes to the data in the view will modify the underlying file. This example requires the `memmap2` crate and assumes a little-endian system.

```rust
# // Miri doesn't support mmap, and the file is in little endian format.
# if !cfg!(miri) && cfg!(target_endian = "little") {
use memmap2::MmapMut;
use ndarray::ArrayViewMut2;
use ndarray_npy::ViewMutNpyExt;
use std::fs;

let file = fs::OpenOptions::new()
    .read(true)
    .write(true)
    .open("resources/array.npy")?;
let mut mmap = unsafe { MmapMut::map_mut(&file)? };
let view_mut = ArrayViewMut2::<i32>::view_mut_npy(&mut mmap)?;
# println!("view_mut = {}", view_mut);
# }
# Ok::<_, Box<dyn std::error::Error>>(())
```

--------------------------------

### Reading an .npy file with read_npy

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.read_npy.html

This example demonstrates how to use the read_npy function to load a 2D array of i32 integers from a file named 'resources/array.npy'. Ensure the file exists at the specified path.

```rust
use ndarray::Array2;
use ndarray_npy::read_npy;

let arr: Array2<i32> = read_npy("resources/array.npy")?;
```

--------------------------------

### Writing an Array to an .npy File

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.write_npy.html

Example demonstrating how to use the write_npy function to save a 2D ndarray to 'array.npy'. Ensure you have the necessary imports.

```rust
use ndarray::array;
use ndarray_npy::write_npy;

let arr = array![[1, 2, 3], [4, 5, 6]];
write_npy("array.npy", &arr)?;

```

--------------------------------

### Error::description Implementation for FormatHeaderError (Deprecated)

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Deprecated method to get a string description of the error. Use Display or to_string() instead.

```rust
fn description(&self) -> &str
```

--------------------------------

### Any::type_id Implementation

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Gets the TypeId of the value.

```rust
fn type_id(&self) -> TypeId
```

--------------------------------

### Get Minor Version Number

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/header.rs.html

Returns the minor version number for a given Version enum variant.

```rust
impl Version {
    // ... other methods
    fn minor_version(self) -> u8 {
        match self {
            Version::V1_0 => 0,
            Version::V2_0 => 0,
            Version::V3_0 => 0,
        }
    }
    // ... other methods
}
```

--------------------------------

### Get Major Version Number

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/header.rs.html

Returns the major version number for a given Version enum variant.

```rust
impl Version {
    // ... other methods
    fn major_version(self) -> u8 {
        match self {
            Version::V1_0 => 1,
            Version::V2_0 => 2,
            Version::V3_0 => 3,
        }
    }
    // ... other methods
}
```

--------------------------------

### Get Number of Arrays in NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Returns the total count of arrays stored within the NPZ file.

```rust
pub fn len(&self) -> usize {
    self.zip.len()
}
```

--------------------------------

### Error::cause Implementation for FormatHeaderError (Deprecated)

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Deprecated method to get the underlying cause of the error. Use Error::source() instead.

```rust
fn cause(&self) -> Option<&dyn Error>
```

--------------------------------

### ndarray-npy Crate Overview

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/lib.rs.html

This snippet shows the main crate-level documentation, outlining its purpose, features, and links to external resources like the repository and NumPy documentation.

```rust
#![cfg_attr(docsrs, feature(doc_cfg))]

//! This crate provides methods to read/write [`ndarray`]'s
//! [`ArrayBase`](ndarray::ArrayBase) type from/to [`.npy`] and [`.npz`] files.
//!
//! See the [repository] for information about the default features and how to
//! use this crate with Cargo.
//!
//! [repository]: https://github.com/jturner314/ndarray-npy
//! [`ndarray`]: https://github.com/bluss/ndarray
//! [`.npy`]: https://docs.scipy.org/doc/numpy/reference/generated/numpy.lib.format.html
//! [`.npz`]: https://docs.scipy.org/doc/numpy/reference/generated/numpy.savez.html
//!
```

--------------------------------

### Writing Multiple Arrays to an NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzWriter.html

Demonstrates how to create a new .npz file and add multiple arrays (2D, 1D, and scalar) to it using NpzWriter. Ensure the 'npz' feature is enabled.

```rust
use ndarray::{array, aview0, Array1, Array2};
use ndarray_npy::NpzWriter;
use std::fs::File;

let mut npz = NpzWriter::new(File::create("arrays.npz")?);
let a: Array2<i32> = array![[1, 2, 3], [4, 5, 6]];
let b: Array1<i32> = array![7, 8, 9];
npz.add_array("a", &a)?;
npz.add_array("b", &b)?;
npz.add_array("c", &aview0(&10))?;
npz.finish()?;
```

--------------------------------

### Reading Arrays from NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzReader.html

Demonstrates how to open an NPZ file and read arrays by their names. Ensure the 'npz' feature is enabled.

```rust
use ndarray::{Array1, Array2};
use ndarray_npy::NpzReader;
use std::fs::File;

let mut npz = NpzReader::new(File::open("arrays.npz")?)?;
let a: Array2<i32> = npz.by_name("a")?;
let b: Array1<i32> = npz.by_name("b")?;
```

--------------------------------

### Get Header Length Number of Bytes

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/header.rs.html

Determines the number of bytes used to represent the header length based on the .npy file version.

```rust
impl Version {
    // ... other methods
    fn header_len_num_bytes(self) -> usize {
        match self {
            Version::V1_0 => 2,
            Version::V2_0 | Version::V3_0 => 4,
        }
    }
    // ... other methods
}
```

--------------------------------

### WritableElement Trait Definition

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

The WritableElement trait provides methods for writing individual elements and slices to a writer, along with a method to get a type descriptor for the header.

```APIDOC
## Trait WritableElement

An array element type that can be written to an `.npy` or `.npz` file.

### Required Methods

#### fn type_descriptor() -> PyValue
Returns a descriptor of the type that can be used in the header.

#### fn write<W: Write>(&self, writer: W) -> Result<(), WriteDataError>
Writes a single instance of `Self` to the writer.

#### fn write_slice<W: Write>(
    slice: &[Self],
    writer: W,
) -> Result<(), WriteDataError>
Writes a slice of `Self` to the writer.
```

--------------------------------

### From<T> for T Implementation

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Returns the argument unchanged.

```rust
fn from(t: T) -> T
```

--------------------------------

### NpzWriter::new_with_options

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzWriter.html

Creates a new .npz file writer with specified options for all subsequent arrays.

```APIDOC
## NpzWriter::new_with_options

### Description
Creates a new `.npz` file with the specified options to be used for each array. This allows you to use a custom compression method, such as zstd, or set other options. Make sure to enable the relevant features of the `zip` crate.

### Method
`new_with_options`

### Signature
`pub fn new_with_options(writer: W, options: SimpleFileOptions) -> NpzWriter<W>`

### Parameters
- **writer**: The underlying writer implementing `Write` and `Seek`.
- **options**: `SimpleFileOptions` to configure the compression and other settings for arrays.
```

--------------------------------

### Implement ViewElement for Multi-Byte Types

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/mod.rs.html

Macro to implement `ViewElement` for multi-byte types, handling native and non-native byte orderings based on type description.

```rust
macro_rules! impl_view_and_view_mut_always_valid_cast_multi_byte {
    ($elem:ty, $native_desc:expr, $non_native_desc:expr) => {
        impl $crate::ViewElement for $elem {
            fn bytes_as_slice<'a>(
                bytes: &'a [u8],
                type_desc: &::py_literal::Value,
                len: usize,
            ) -> Result<&'a [Self], $crate::ViewDataError> {
                match *type_desc {
                    ::py_literal::Value::String(ref s) if s == $native_desc => unsafe {

```

--------------------------------

### Get Names of Arrays in NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Retrieves a list of all array names stored within the NPZ file. The '.npy' suffix is automatically stripped from each name, mirroring NumPy's behavior.

```APIDOC
## names

### Description
Returns the names of all of the arrays in the file. The '.npy' suffix (if present) will be stripped from each name, matching NumPy's behavior.

### Signature
`pub fn names(&mut self) -> Result<Vec<String>, ReadNpzError>`

### Returns
- `Result<Vec<String>, ReadNpzError>`: A `Result` containing a vector of strings (the array names) on success, or a `ReadNpzError` on failure.
```

--------------------------------

### Create New NPZ Reader

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Initializes a new NpzReader from a given Read + Seek source. This prepares the reader to access arrays within the NPZ archive.

```rust
pub fn new(reader: R) -> Result<NpzReader<R>, ReadNpzError> {
    Ok(NpzReader {
        zip: ZipArchive::new(reader)?,
    })
}
```

--------------------------------

### Create a new NPY file

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.create_new_npy.html

Demonstrates how to use `create_new_npy` to write an array to a new file and shows that it returns an error if the file already exists. Ensure you have the `ndarray` and `ndarray-npy` crates in your Cargo.toml.

```rust
use ndarray::array;
use ndarray_npy::create_new_npy;

let arr = array![[1, 2, 3], [4, 5, 6]];
create_new_npy("new_array.npy", &arr)?;
assert!(create_new_npy("new_array.npy", &arr).is_err());
```

--------------------------------

### Get all array names from NPZ file

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Retrieves a list of all array names stored within an NPZ file. The '.npy' suffix is automatically stripped from each name to match NumPy's behavior.

```rust
pub fn names(&mut self) -> Result<Vec<String>, ReadNpzError> {
        Ok((0..self.zip.len())
            .map(|i| {
                let file = self.zip.by_index(i)?;
                let name = file.name();
                let stripped = name.strip_suffix(".npy").unwrap_or(name);
                Ok(stripped.to_owned())
            })
            .collect::<Result<_, ZipError>>()?)
    }
```

--------------------------------

### Reading and Writing .npz Files

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/lib.rs.html

Provides structures for reading from and writing to .npz (NumPy compressed archive) files.

```APIDOC
## Reading and Writing .npz Files

### Description
Provides structures for reading from and writing to .npz (NumPy compressed archive) files.

### Structures
- `NpzReader`: For reading data from .npz files.
- `NpzWriter`: For writing data to .npz files.

### Errors
- `ReadNpzError`: Errors encountered during .npz file reading.
- `WriteNpzError`: Errors encountered during .npz file writing.
```

--------------------------------

### NpzWriter::new

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzWriter.html

Creates a new, uncompressed .npz file writer. This is equivalent to numpy.savez.

```APIDOC
## NpzWriter::new

### Description
Creates a new `.npz` file without compression. See `numpy.savez`.

### Method
`new`

### Signature
`pub fn new(writer: W) -> NpzWriter<W>`

### Parameters
- **writer**: The underlying writer implementing `Write` and `Seek`.
```

--------------------------------

### Read NPY file into Array2

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/mod.rs.html

Use the `read_npy` method from the `ReadNpyExt` trait to load data from a file into an `Array2` instance. Ensure the file path is correct and the file exists. This example requires `std::fs::File` and `ndarray_npy::ReadNpyExt`.

```rust
use ndarray::Array2;
use ndarray_npy::ReadNpyExt;
use std::fs::File;
# use ndarray_npy::ReadNpyError;

let reader = File::open("resources/array.npy")?;
let arr = Array2::<i32>::read_npy(reader)?;
# println!("arr = {}", arr);
# Ok::<_, ReadNpyError>(())
```

--------------------------------

### Header Implementations

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/struct.Header.html

Provides methods for interacting with the Header struct, including deserializing from a reader, serializing to bytes, and writing to a writer.

```APIDOC
## impl Header

### `pub fn from_reader<R: Read>(reader: &mut R) -> Result<Self, ReadHeaderError>`

Deserializes a header from the provided reader.

### `pub fn to_bytes(&self) -> Result<Vec<u8>, FormatHeaderError>`

Returns the serialized representation of the header.

### `pub fn write<W: Write>(&self, writer: W) -> Result<(), WriteHeaderError>`

Writes the serialized representation of the header to the provided writer.
```

--------------------------------

### Writing .npz Files

Source: https://docs.rs/ndarray-npy

Offers functionality to write data to .npz (compressed archive) files.

```APIDOC
## Writing .npz Files

This section details writing data to `.npz` files.

### Writer
- `NpzWriter`: A struct for writing `.npz` files.

### Errors
- `WriteNpzError`: Errors that can occur when writing a `.npz` file.
```

--------------------------------

### WritableElement for i64

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 64-bit signed integers.

```rust
impl WritableElement for i64
```

--------------------------------

### Writing .npz Files

Source: https://docs.rs/ndarray-npy/0.10.0/index.html

Offers functionality to write data to .npz files.

```APIDOC
## Writing .npz Files

### Description
Offers functionality to write data to .npz files.

### Structs
- `NpzWriter`: Writer for `.npz` files.

### Errors
- `WriteNpzError`: An error writing a `.npz` file.
```

--------------------------------

### WritableElement for f64

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 64-bit floating-point numbers.

```rust
impl WritableElement for f64
```

--------------------------------

### Error::provide Implementation for FormatHeaderError (Nightly)

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Nightly-only experimental API for providing type-based access to error context.

```rust
fn provide<'a>(&'a self, request: &mut Request<'a>)
```

--------------------------------

### WritableElement for i32

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 32-bit signed integers.

```rust
impl WritableElement for i32
```

--------------------------------

### npy::create_new_npy

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/all.html

Creates a new NPY file with specified dimensions and data type, initializing it with zeros.

```APIDOC
## npy::create_new_npy

### Description
Creates a new NPY file with specified dimensions and data type, initializing it with zeros. This function is useful for pre-allocating space for NPY data.

### Function Signature
```rust
pub fn create_new_npy<P: AsRef<Path>, T: WritableElement>(path: P, shape: &[usize], dtype: DType) -> Result<(), WriteNpyError>
```

### Parameters
#### Path Parameter
- **path** (*P*): A type that can be converted into a reference to a Path, representing the file path for the new NPY file.

#### Generic Type Parameters
- **T**: The type of elements to be written to the NPY file. Must implement the `WritableElement` trait.

#### Arguments
- **shape** (*&[usize]*): A slice representing the dimensions of the NPY array.
- **dtype** (*DType*): The data type of the elements in the NPY file.

### Returns
- `Result<(), WriteNpyError>`: Ok(()) if the file was created successfully, or a `WriteNpyError` if an error occurred during file creation or writing.
```

--------------------------------

### WritableElement for u64

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 64-bit unsigned integers.

```rust
impl WritableElement for u64
```

--------------------------------

### WritableElement for i8

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 8-bit signed integers.

```rust
impl WritableElement for i8
```

--------------------------------

### write() Method

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Writes a single instance of the element to the provided writer.

```rust
fn write<W: Write>(&self, writer: W) -> Result<(), WriteDataError>;
```

--------------------------------

### From<FormatHeaderError> for WriteHeaderError

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.WriteHeaderError.html

Converts a FormatHeaderError into a WriteHeaderError.

```rust
impl From<FormatHeaderError> for WriteHeaderError {
    fn from(err: FormatHeaderError) -> WriteHeaderError {
        // ... implementation details ...
    }
}
```

--------------------------------

### ViewNpyExt::view_npy

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.ViewNpyExt.html

Creates an ArrayView from a buffer containing an .npy file. This is useful for memory-mapped files where direct viewing is preferred over full loading.

```APIDOC
## fn view_npy(buf: &'a [u8]) -> Result<Self, ViewNpyError>

### Description
Creates an `ArrayView` from a buffer containing an `.npy` file.

### Parameters
#### Path Parameters
- `buf` (&'a [u8]): A byte slice representing the content of an .npy file.

### Returns
- `Result<Self, ViewNpyError>`: An `ArrayView` of the appropriate type and dimension, or a `ViewNpyError` if the buffer cannot be interpreted as an .npy file.
```

--------------------------------

### TryInto<U> for T Trait Implementation

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.ReadHeaderError.html

Demonstrates the generic `try_into` conversion trait implementation.

```rust
fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>
Performs the conversion.
```

--------------------------------

### WritableElement for f32

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 32-bit floating-point numbers.

```rust
impl WritableElement for f32
```

--------------------------------

### NpzReader Methods

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzReader.html

This section details the public methods available on the NpzReader struct for interacting with .npz files.

```APIDOC
## NpzReader

Reader for `.npz` files.

### `new(reader: R) -> Result<NpzReader<R>, ReadNpzError>`

Creates a new `.npz` file reader.

### `is_empty(&self) -> bool`

Returns `true` iff the `.npz` file doesn’t contain any arrays.

### `len(&self) -> usize`

Returns the number of arrays in the `.npz` file.

### `names(&mut self) -> Result<Vec<String>, ReadNpzError>`

Returns the names of all of the arrays in the file. Note that a single ".npy" suffix (if present) will be stripped from each name; this matches NumPy’s behavior.

### `by_name<S, D>(&mut self, name: &str) -> Result<ArrayBase<S, D>, ReadNpzError>`

Reads an array by name. Note that this first checks for `name` in the `.npz` file, and if that is not present, checks for `format!("{name}.npy")`. This matches NumPy’s behavior.

### `by_index<S, D>(&mut self, index: usize) -> Result<ArrayBase<S, D>, ReadNpzError>`

Reads an array by index in the `.npz` file.
```

--------------------------------

### WritableElement for i16

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 16-bit signed integers.

```rust
impl WritableElement for i16
```

--------------------------------

### Implement i16 primitive type

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/num.rs.html

Applies the `impl_primitive_multi_byte` macro to i16. This configures its reading, writing, and viewing capabilities using little-endian ('<i2') and big-endian ('>i2') descriptors.

```Rust
impl_primitive_multi_byte!(i16, "<i2", ">i2", 0, read_i16_into);
```

--------------------------------

### Writing .npz Files

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/index.html

Provides functionality to write data to .npz files, enabling the creation of archives containing multiple .npy files.

```APIDOC
## NpzWriter

### Description
A writer for `.npz` files, allowing the creation of archives containing multiple `.npy` arrays.

### Methods
- `NpzWriter::new()`: Creates a new `NpzWriter` instance.
- Methods for adding `.npy` files to the archive (details not specified).

### Errors
- `WriteNpzError`: If there is an error writing the `.npz` file.
```

--------------------------------

### Implementing ViewElement for bool

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/bool.rs.html

Provides a safe way to view raw bytes as a slice of bools. It validates that each byte is 0 or 1 before casting.

```rust
impl ViewElement for bool {
    fn bytes_as_slice<'a>(
        bytes: &'a [u8],
        type_desc: &PyValue,
        len: usize,
    ) -> Result<&'a [Self], ViewDataError> {
        match *type_desc {
            PyValue::String(ref s) if s == "|b1" => {
                check_valid_for_bool(bytes)?;
                unsafe { bytes_as_slice(bytes, len) }
            }
            ref other => Err(ViewDataError::WrongDescriptor(other.clone())),
        }
    }
}
```

--------------------------------

### Implement ViewElement and ViewMutElement for one-byte primitive types

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/num.rs.html

Implements `ViewElement` and `ViewMutElement` traits for one-byte primitive types. This allows viewing the raw bytes of the data as slices of the primitive type, based on matching type descriptors.

```Rust
macro_rules! impl_view_and_view_mut_primitive_one_byte {
    ($elem:ty, [$($desc:expr),*]) => {
        impl ViewElement for $elem {
            fn bytes_as_slice<'a>(
                bytes: &'a [u8],
                type_desc: &PyValue,
                len: usize,
            ) -> Result<&'a [Self], ViewDataError> {
                match *type_desc {
                    PyValue::String(ref s) if $(s == $desc)||* => unsafe {
                        bytes_as_slice(bytes, len)
                    }
                    ref other => Err(ViewDataError::WrongDescriptor(other.clone())),
                }
            }
        }

        impl ViewMutElement for $elem {
            fn bytes_as_mut_slice<'a>(
                bytes: &'a mut [u8],
                type_desc: &PyValue,
                len: usize,
            ) -> Result<&'a mut [Self], ViewDataError> {
                match *type_desc {
                    PyValue::String(ref s) if $(s == $desc)||* => unsafe {
                        bytes_as_mut_slice(bytes, len)
                    }
                    ref other => Err(ViewDataError::WrongDescriptor(other.clone())),
                }
            }
        }
    };
}
```

--------------------------------

### NpzWriter::new_compressed

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/struct.NpzWriter.html

Creates a new .npz file writer with Deflated compression. This is equivalent to numpy.savez_compressed.

```APIDOC
## NpzWriter::new_compressed

### Description
Creates a new `.npz` file with `Deflated` compression. See `numpy.savez_compressed`.
For other compression algorithms, use `NpzWriter::new_with_options` or `NpzWriter::add_array_with_options`.

### Method
`new_compressed`

### Signature
`pub fn new_compressed(writer: W) -> NpzWriter<W>`

### Parameters
- **writer**: The underlying writer implementing `Write` and `Seek`.
```

--------------------------------

### ViewMutElement Implementations

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.ViewMutElement.html

Shows implementations of the ViewMutElement trait for various primitive and complex number types.

```APIDOC
### impl ViewMutElement for Complex<f32>

Available on **crate feature`num-complex-0_4`** only.

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for Complex<f64>

Available on **crate feature`num-complex-0_4`** only.

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for bool

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &PyValue,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for f32

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for f64

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for i8

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &PyValue,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for i16

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for i32

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for i64

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for u8

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &PyValue,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for u16

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for u32

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

```APIDOC
### impl ViewMutElement for u64

#### fn bytes_as_mut_slice<'a>(
    bytes: &'a mut [u8],
    type_desc: &Value,
    len: usize,
) -> Result<&'a mut [Self], ViewDataError>
```

--------------------------------

### Implement primitive types with one-byte descriptors

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/num.rs.html

A convenience macro to implement `WritableElement`, `ReadableElement`, `ViewElement`, and `ViewMutElement` for primitive types that use one-byte descriptors. It takes the element type, write descriptor, read descriptors, zero value, and read function.

```Rust
macro_rules! impl_primitive_one_byte {
    ($elem:ty, $write_desc:expr, [$($read_desc:expr),*], $zero:expr, $read_into:ident) => {
        impl_writable_element_always_valid_cast!($elem, $write_desc, $write_desc);
        impl_readable_primitive_one_byte!($elem, [$($read_desc),*], $zero, $read_into);
        impl_view_and_view_mut_primitive_one_byte!($elem, [$($read_desc),*]);
    };
}
```

--------------------------------

### Display Format for Header

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/struct.Header.html

Implements the Display trait for the Header struct, enabling user-friendly string representation.

```rust
fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>
```

--------------------------------

### WritableElement Implementations

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

This section details the implementations of the WritableElement trait for various primitive types and complex numbers, indicating compatibility with specific feature flags.

```APIDOC
### impl WritableElement for Complex<f32>
Available on **crate feature`num-complex-0_4`** only.

### impl WritableElement for Complex<f64>
Available on **crate feature`num-complex-0_4`** only.

### impl WritableElement for bool

### impl WritableElement for f32

### impl WritableElement for f64

### impl WritableElement for i8

### impl WritableElement for i16

### impl WritableElement for i32

### impl WritableElement for i64

### impl WritableElement for u8

### impl WritableElement for u16

### impl WritableElement for u32

### impl WritableElement for u64
```

--------------------------------

### Reading .npz Files

Source: https://docs.rs/ndarray-npy

Provides functionality to read data from .npz (compressed archive) files.

```APIDOC
## Reading .npz Files

This section covers reading data from `.npz` files.

### Reader
- `NpzReader`: A struct for reading `.npz` files.

### Errors
- `ReadNpzError`: Errors that can occur when reading a `.npz` file.
```

--------------------------------

### WritableElement for u32

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 32-bit unsigned integers.

```rust
impl WritableElement for u32
```

--------------------------------

### TryInto<U> for T Implementation

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Performs a fallible conversion into another type.

```rust
fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>
```

--------------------------------

### HeaderLengthInfo

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/header.rs.html

Information about the calculated header length.

```APIDOC
struct HeaderLengthInfo {
    /// Total header length (including magic string, version number, header
    /// length value, array format description, padding, and final newline).
    total_len: usize,
    /// Formatted `HEADER_LEN` value. (This is the number of bytes in the array
    /// format description, padding, and final newline.)
    formatted_header_len: Vec<u8>,
}

```

--------------------------------

### Implement i8 and u8 primitive types

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npy/elements/num.rs.html

Applies the `impl_primitive_one_byte` macro to i8 and u8. This sets up their reading, writing, and viewing capabilities with their respective NumPy descriptors and zero values.

```Rust
impl_primitive_one_byte!(i8, "|i1", ["|i1", "i1", "b"], 0, read_i8_into);
impl_primitive_one_byte!(u8, "|u1", ["|u1", "u1", "B"], 0, read_exact);
```

--------------------------------

### create_new_npy

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/fn.create_new_npy.html

Writes an array to a new .npy file at the specified path. Returns an error if the file already exists. This function is a shortcut for creating a new file and then writing the array using `WriteNpyExt`.

```APIDOC
## Function create_new_npy

### Description
Writes an array to a new `.npy` file at the specified path; error if the file exists. This is a convenience function for `BufWriter::new(File::create_new(path)?)` followed by `WriteNpyExt::write_npy`.

### Signature
```rust
pub fn create_new_npy<P, T>(path: P, array: &T) -> Result<(), WriteNpyError>
where
    P: AsRef<Path>,
    T: WriteNpyExt + ?Sized,
```

### Example
```rust
use ndarray::array;
use ndarray_npy::create_new_npy;

let arr = array![[1, 2, 3], [4, 5, 6]];
create_new_npy("new_array.npy", &arr)?;
assert!(create_new_npy("new_array.npy", &arr).is_err());
```
```

--------------------------------

### Reading .npz Files

Source: https://docs.rs/ndarray-npy/0.10.0/index.html

Provides functionality to read data from .npz files.

```APIDOC
## Reading .npz Files

### Description
Provides functionality to read data from .npz files.

### Structs
- `NpzReader`: Reader for `.npz` files.

### Errors
- `ReadNpzError`: An error reading a `.npz` file.
```

--------------------------------

### From<WriteNpyError> for WriteNpzError Conversion (npz feature)

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/enum.WriteNpyError.html

Shows the conversion from WriteNpyError to WriteNpzError, available when the 'npz' crate feature is enabled. This allows for error handling across both .npy and .npz file writing operations.

```rust
impl From<WriteNpyError> for WriteNpzError {
    fn from(err: WriteNpyError) -> WriteNpzError {
        WriteNpzError::Io(err.into())
    }
}
```

--------------------------------

### TryFrom<U> for T Implementation

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/enum.FormatHeaderError.html

Performs a fallible conversion from one type to another.

```rust
fn try_from(value: U) -> Result<T, <T as TryFrom<U>>::Error>
```

--------------------------------

### Serialize Header to Bytes

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/struct.Header.html

Returns the serialized byte representation of the header. Useful for writing .npy files.

```rust
pub fn to_bytes(&self) -> Result<Vec<u8>, FormatHeaderError>
```

--------------------------------

### WritableElement for u8

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 8-bit unsigned integers.

```rust
impl WritableElement for u8
```

--------------------------------

### Reading and Writing .npy Files

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/lib.rs.html

This section details the extension traits and convenience functions available for reading and writing .npy files, including options for zeroed data and readonly/mutable viewing.

```rust
#!/usr/bin/env 
// This crate provides methods to read/write [`ndarray`]'s
//! [`ArrayBase`](ndarray::ArrayBase) type from/to [`.npy`] and [`.npz`] files.
//!
//! See the [repository] for information about the default features and how to
//! use this crate with Cargo.
//!
//! [repository]: https://github.com/jturner314/ndarray-npy
//! [`ndarray`]: https://github.com/bluss/ndarray
//! [`.npy`]: https://docs.scipy.org/doc/numpy/reference/generated/numpy.lib.format.html
//! [`.npz`]: https://docs.scipy.org/doc/numpy/reference/generated/numpy.savez.html
//!
```

--------------------------------

### WritableElement for u16

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/trait.WritableElement.html

Implementation for writing 16-bit unsigned integers.

```rust
impl WritableElement for u16
```

--------------------------------

### Generic CloneToUninit Trait Implementation (Nightly)

Source: https://docs.rs/ndarray-npy/0.10.0/ndarray_npy/npy/header/struct.Header.html

Implements the nightly-only CloneToUninit trait for cloning data into uninitialized memory.

```rust
unsafe fn clone_to_uninit(&self, dest: *mut u8)
```

--------------------------------

### Finish Writing NPZ File

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Finalizes the NPZ file by calling finish on the zip writer and flush on the underlying writer. Essential for proper error handling.

```rust
pub fn finish(self) -> Result<W, WriteNpzError> {
    let mut writer = self.zip.finish()?;
    writer.flush().map_err(ZipError::from)?;
    Ok(writer)
}
```

--------------------------------

### finish

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Finalizes the .npz file by writing zip structures and flushing the underlying writer. This method is crucial for error handling during the finalization process.

```APIDOC
## finish

### Description
Completes the writing of the `.npz` file by calling `.finish()` on the internal zip writer and `.flush()` on the underlying writer. This ensures all data is written and zip structures are finalized. It's recommended to call this method explicitly to handle potential errors during finalization, as dropping the writer might silently ignore them.

### Method Signature
`pub fn finish(self) -> Result<W, WriteNpzError>`

### Type Parameters
* `W`: The type of the underlying writer, which must implement `Write` and `Seek`.

### Returns
* `Result<W, WriteNpzError>`: The underlying writer if the process completed successfully, or an error if any part of the finalization failed.
```

--------------------------------

### add_array_with_options

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Adds an array to the .npz file with specified options, overriding any default options. The .npy extension is automatically appended to the name.

```APIDOC
## add_array_with_options<N, T, U>

### Description
Adds an array to the .npz file with custom file options. These options override any defaults set during the `NpzWriter` initialization. A `.npy` extension is appended to the provided name.

### Method Signature
`pub fn add_array_with_options<N, T, U>(&mut self, name: N, array: &T, options: FileOptions<'_, U>) -> Result<(), WriteNpzError>`

### Type Parameters
* `N`: Type that can be converted into a `String` (used for the array name).
* `T`: Type that implements `WriteNpyExt` (the array to be written).
* `U`: Type that implements `FileOptionExtension` (specifies extension options for the file).

### Parameters
* **name**: `N` - The name of the array to be added.
* **array**: `&T` - A reference to the array to be written.
* **options**: `FileOptions<'_, U>` - The file options to use for writing this specific array.

### Returns
* `Result<(), WriteNpzError>`: Ok if the array was added successfully, or an error if writing failed.
```

--------------------------------

### add_array

Source: https://docs.rs/ndarray-npy/0.10.0/src/ndarray_npy/npz.rs.html

Adds an array to the .npz file with default options. The .npy extension is automatically appended to the name.

```APIDOC
## add_array<N, T>

### Description
Adds an array to the .npz file using the default file options.

### Method Signature
`pub fn add_array<N, T>(&mut self, name: N, array: &T) -> Result<(), WriteNpzError>`

### Type Parameters
* `N`: Type that can be converted into a `String` (used for the array name).
* `T`: Type that implements `WriteNpyExt` (the array to be written).

### Parameters
* **name**: `N` - The name of the array to be added.
* **array**: `&T` - A reference to the array to be written.

### Returns
* `Result<(), WriteNpzError>`: Ok if the array was added successfully, or an error if writing failed.
```
Tokens:
Raw