#sanitizer #rfc822 #rfc2822

rfc822_sanitizer

rfc822/2822 sanitizer/normalizer

9 releases

0.3.6 Jan 11, 2022
0.3.4 Aug 17, 2020
0.3.3 Nov 5, 2017
0.3.2 Oct 7, 2017
0.1.2 Sep 28, 2017

#77 in Date and time

Download history 156/week @ 2023-11-20 77/week @ 2023-11-27 86/week @ 2023-12-04 94/week @ 2023-12-11 75/week @ 2023-12-18 51/week @ 2023-12-25 95/week @ 2024-01-01 92/week @ 2024-01-08 154/week @ 2024-01-15 103/week @ 2024-01-22 102/week @ 2024-01-29 195/week @ 2024-02-05 132/week @ 2024-02-12 86/week @ 2024-02-19 303/week @ 2024-02-26 183/week @ 2024-03-04

732 downloads per month
Used in 6 crates (3 directly)

GPL-3.0+

64KB
2K SLoC

RFC822/2822_sanitizer

pipeline status Build Status Build status Crates.io codecov Dependency Status

Documentation

A helper function that does what you wanted not what you said!

The world is full of broken code and everyone seem to reinvent daytimes.

If it encounters an invalid rfc822/2822 daytime it tries to fix it first before returning Errors.

This crates soul purpose is to try to fix the mess of the generators that do stuff like this:

Thu, 05 Aug 2016 06:00:00 -0400

Mon, 31 July 2017 16:00:00 PDT

Wed, 20 Sep 2017 10:00:00 -0000

30 Aug 2017 1:30:00 PDT

All of the above look fine at first glance. But all are invalid and would fail to be parsed with chrono::Datetime::parse_from_rfc2822

Playground link.

Err(ParseError(Impossible))
Err(ParseError(Invalid))
Err(ParseError(NotEnough))
Err(ParseError(Invalid))
  • The first one Thu, 05 Aug 2016 06:00:00 -0400 is actually a Friday.

  • The second Mon, 31 July 2017 16:00:00 PDT uses full lenght month Name.

  • The third Wed, 20 Sep 2017 10:00:00 -0000 has -0000 as the timezone, which is sort of undefined behaviour. For more see #102.

  • The forth 30 Aug 2017 1:30:00 PDT uses single digit notation for Hour.

The dates above have been encountered while trying to parse rss feeds from the wild west eer the Internet.

The RSS spec specifies the use of RFC822 for the date format, which is forward compatible with RFC2822. This crate proves that people/generators still get wrong a format witch was published in 1982.

Now if we were to use the sanitizer, we would actually get a correct datetime.

Usage:

Add the following to your Cargo.toml.

[dependencies]
rfc822_sanitizer = "0.3"

from examples/simple.rs

extern crate rfc822_sanitizer;
use rfc822_sanitizer::parse_from_rfc2822_with_fallback;

fn main() {
    let dates = vec![
        "Thu, 05 Aug 2016 06:00:00 -0400",
        "Mon, 31 July 2017 16:00:00 PDT",
        "Wed, 20 Sep 2017 10:00:00 -0000",
        "30 Aug 2017 1:30:00 PDT",
    ];

    for foo in dates.iter() {
        let fallback = parse_from_rfc2822_with_fallback(&foo);
        println!("{:?}", fallback);
    }
}

Output:

Ok(2016-08-05T06:00:00-04:00)
Ok(2017-07-31T16:00:00-07:00)
Ok(2017-09-20T10:00:00+00:00)
Ok(2017-08-30T01:30:00-07:00)

Though keep in mind that it would consume more resources.

$ cargo +nightly bench -q
running 4 tests
test bench_correct_dates_normal_parse        ... bench:      20,741 ns/iter (+/- 1,000)
test bench_correct_dates_with_fallback       ... bench:      21,133 ns/iter (+/- 282)
test bench_parse_invalid_dates_with_fallback ... bench:     119,116 ns/iter (+/- 3,706)
test mixed_sample_case                       ... bench:     431,814 ns/iter (+/- 8,526)

Dependencies

~3–4.5MB
~72K SLoC