Module `Datasets.Dataset_utils`Source

Utilities for downloading and managing datasets.

Sourceval get_cache_dir : string -> string

Return the platform-specific cache directory path for the given dataset.

The default location is "~/.cache/ocannl/datasets/dataset_name/".

Parameters

dataset_name: the name of the dataset.

Returns

the cache directory path, including trailing slash.

Sourceval download_file : string -> string -> unit

Download a file from a URL to a destination path.

Creates parent directories as needed, downloads the file from url, and saves it to dest_path.

Parameters

url: the source URL of the file.
dest_path: local path to save the downloaded file.

Raises

Failure on download or write error.

Sourceval ensure_file : string -> string -> unit

Ensure a file exists at the given path, downloading if necessary.

Checks if dest_path exists. If not, downloads the file from url.

Parameters

url: the source URL of the file.
dest_path: local path to ensure the file exists.

Raises

Failure on download or write error.

Source

val ensure_extracted_archive : 
  url:string ->
  archive_path:string ->
  extract_dir:string ->
  check_file:string ->
  unit

Ensure an archive is downloaded, extracted, and a file exists.

Checks if check_file (relative to extract_dir) exists. If not, downloads the archive from url to archive_path, extracts it into extract_dir, and verifies check_file is present. Currently supports only .tar.gz archives.

Parameters

url: the source URL of the archive.
archive_path: local path for the downloaded archive.
extract_dir: directory to extract the archive into.
check_file: relative path under extract_dir to verify extraction.

Raises

Failure on download, extraction, or missing check_file.

Sourceval ensure_decompressed_gz : gz_path:string -> target_path:string -> bool

Ensure a gzip-compressed file is decompressed to a target path.

If target_path exists, does nothing and returns true. Otherwise, if gz_path exists, decompresses it to target_path.

Parameters

gz_path: the path to the .gz file to decompress.
target_path: the destination path for the decompressed file.

Returns

true if target_path exists after the operation.
false if gz_path does not exist.

Raises

Failure on gzip decompression error.

Sourceval parse_float_cell : context:(unit -> string) -> string -> float

Parse a CSV cell as a float.

Attempts to convert value to a float. On failure, raises Failure with a descriptive message including context ().

Parameters

context: a function returning context information for error messages.
value: the string to parse as a float.

Returns

the parsed float.

Raises

Failure if value cannot be parsed as a float.

Sourceval parse_int_cell : context:(unit -> string) -> string -> int

Parse a CSV cell as an integer.

Attempts to convert value to an int. On failure, raises Failure with a descriptive message including context ().

Parameters

context: a function returning context information for error messages.
value: the string to parse as an int.

Returns

the parsed integer.

Raises

Failure if value cannot be parsed as an int.

Sourceval mkdir_p : string -> unit

Recursively create a directory and its parents.

Creates the directory at path, along with any missing parent directories. If path already exists as a directory, does nothing.

Parameters

path: the directory path to create.

Raises

Unix.Unix_error if creation fails for other reasons.

package neural_nets_lib

Module Datasets.Dataset_utilsSource

Module `Datasets.Dataset_utils`Source