polars_io/utils/
mod.rs

1pub mod compression;
2mod other;
3
4pub use other::*;
5#[cfg(any(feature = "async", feature = "cloud"))]
6pub mod byte_source;
7pub mod file;
8pub mod mkdir;
9pub mod slice;
10#[cfg(feature = "async")]
11pub mod stream_buf_reader;
12pub mod sync_on_close;
13
14/// Excludes only the unreserved URI characters in RFC-3986:
15///
16/// <https://datatracker.ietf.org/doc/html/rfc3986#section-2.3>
17///
18/// Characters that are allowed in a URI but do not have a reserved
19/// purpose are called unreserved.  These include uppercase and lowercase
20/// letters, decimal digits, hyphen, period, underscore, and tilde.
21///
22/// unreserved  = ALPHA / DIGIT / "-" / "." / "_" / "~"
23pub const URL_ENCODE_CHARSET: &percent_encoding::AsciiSet = &percent_encoding::NON_ALPHANUMERIC
24    .remove(b'-')
25    .remove(b'.')
26    .remove(b'_')
27    .remove(b'~');
28
29/// Characters to percent-encode for hive values such that they round-trip from bucket storage.
30///
31/// This is much more relaxed than the RFC-3986 URI spec as bucket storage is more permissive of allowed
32/// characters.
33pub const HIVE_VALUE_ENCODE_CHARSET: &percent_encoding::AsciiSet = &percent_encoding::CONTROLS
34    .add(b'/') // Exclude path separator
35    .add(b'=') // Exclude hive `key=value` separator
36    .add(b'%') // Percent itself.
37    // Colon and space are supported by object storage, but are encoded to mimic
38    // the datetime output format from pyarrow:
39    // * i.e. 'date2=2023-01-01 00:00:00.000000' becomes 'date2=2023-01-01%2000%3A00%3A00.000000'
40    .add(b':')
41    .add(b' ');
polars_io/utils/mod.rs

polars_io/utils/
mod.rs
