Fix 8244 -- store timestamps with nanosecond resolution (consistently) (#8337)

# Description

Fix for data ambiguity noted in #8244.

Basic change is to use nanosecond resolution for unix timestamps (stored
in type Int). Previously, a timestamp might have seconds, milliseconds
or nanoseconds, but it turned out there were overlaps in data ranges
between different resolutions, so there wasn't always a unique mapping
back to date/time.

Due to higher precision, the *range* of dates that timestamps can map to
is restricted. Unix timestamps with seconds resolution and 64 bit
storage can cover all dates from the Big Bang to eternity. Timestamps
with seconds resolution and 32 bit storage can only represent dates from
1901-12-13 through 2038-01-19. The nanoseconds resolution and 64 bit
storage used with this fix can represent dates from 1677-09-21T00:12:44
to 2262-04-11T23:47:16, something of a compromise.

# User-Facing Changes
_(List of all changes that impact the user experience here. This helps
us keep track of breaking changes.)_

## `<datetime> | into int`
Converts to nanosecond resolution
```rust
〉date now | into int
1678084730502126846
```
This is the number of non-leap nanoseconds after the unix epoch date:
1970-01-01T00:00:00+00:00.

Conversion fails for dates outside the supported range:
```rust
〉1492-10-12 | into int
Error: nu:🐚:incorrect_value

  × Incorrect value.
   ╭─[entry #51:1:1]
 1 │ 1492-10-12 | into int
   ·              ────┬───
   ·                  ╰── DateTime out of timestamp range 1677-09-21T00:12:43 and 2262-04-11T23:47:16
   ╰────


```

## `<int> | into datetime`
Can no longer fail or produce incorrect results for any 64-bit input:
```rust
〉0 | into datetime 
Thu, 01 Jan 1970 00:00:00 +0000 (53 years ago)
〉"7fffffffffffffff" | into int -r 16 | into datetime
Fri, 11 Apr 2262 23:47:16 +0000 (in 239 years)
〉("7fffffffffffffff" | into int -r 16) * -1 | into datetime
Tue, 21 Sep 1677 00:12:43 +0000 (345 years ago)
```

## `<date> | date to-record` and `<date> | date to-table`
Now both have a `nanosecond` field.  
```rust
〉"7fffffffffffffff" | into int -r 16 | into datetime | date to-record
╭────────────┬───────────╮
│ year       │ 2262      │
│ month      │ 4         │
│ day        │ 11        │
│ hour       │ 23        │
│ minute     │ 47        │
│ second     │ 16        │
│ nanosecond │ 854775807 │
│ timezone   │ +00:00    │
╰────────────┴───────────╯
〉"7fffffffffffffff" | into int -r 16 | into datetime | date to-table
╭───┬──────┬───────┬─────┬──────┬────────┬────────┬────────────┬──────────╮
│ # │ year │ month │ day │ hour │ minute │ second │ nanosecond │ timezone │
├───┼──────┼───────┼─────┼──────┼────────┼────────┼────────────┼──────────┤
│ 0 │ 2262 │     4 │  11 │   23 │     47 │     16 │  854775807 │ +00:00   │
╰───┴──────┴───────┴─────┴──────┴────────┴────────┴────────────┴──────────╯
```

This change was not mandated by the OP problem, but it is nice to be
able to see the nanosecond bits that were present in Nushell `date` type
all along.
# Tests + Formatting

Don't forget to add tests that cover your changes.

Make sure you've run and fixed any issues with these commands:

- `cargo fmt --all -- --check` to check standard code formatting (`cargo
fmt --all` applies these changes)
- `cargo clippy --workspace -- -D warnings -D clippy::unwrap_used -A
clippy::needless_collect` to check that you're using the standard code
style
- `cargo test --workspace` to check that all tests pass

# After Submitting

If your PR had any user-facing changes, update [the
documentation](https://github.com/nushell/nushell.github.io) after the
PR is merged, if necessary. This will help us keep the docs up to date.
This commit is contained in:
Bob Hyman 2023-03-07 18:02:15 -05:00 committed by GitHub
parent f34034ae58
commit 2ad0fcb377
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 198 additions and 109 deletions

View File

@ -1,6 +1,6 @@
use crate::input_handler::{operate, CmdArgument};
use crate::{generate_strftime_list, parse_date_from_string};
use chrono::{DateTime, FixedOffset, Local, LocalResult, TimeZone, Utc};
use chrono::{DateTime, FixedOffset, Local, TimeZone, Utc};
use nu_engine::CallExt;
use nu_protocol::ast::Call;
use nu_protocol::ast::CellPath;
@ -83,12 +83,12 @@ impl Command for SubCommand {
.named(
"format",
SyntaxShape::String,
"Specify an expected format for parsing strings to datetimes. Use --list to see all possible options",
"Specify expected format of string input to parse to datetime. Use --list to see options",
Some('f'),
)
.switch(
"list",
"Show all possible variables for use with the --format flag",
"Show all possible variables for use in --format flag",
Some('l'),
)
.rest(
@ -125,10 +125,12 @@ impl Command for SubCommand {
span: zone.span,
}),
};
let format_options = call
.get_flag::<String>(engine_state, stack, "format")?
.as_ref()
.map(|fmt| DatetimeFormat(fmt.to_string()));
let args = Arguments {
format_options,
zone_options,
@ -139,7 +141,7 @@ impl Command for SubCommand {
}
fn usage(&self) -> &str {
"Convert text into a datetime."
"Convert text or timestamp into a datetime."
}
fn search_terms(&self) -> Vec<&str> {
@ -147,53 +149,44 @@ impl Command for SubCommand {
}
fn examples(&self) -> Vec<Example> {
let example_result_1 = |secs: i64, nsecs: u32| match Utc.timestamp_opt(secs, nsecs) {
LocalResult::Single(dt) => Some(Value::Date {
val: dt.into(),
let example_result_1 = |nanos: i64| {
Some(Value::Date {
val: Utc.timestamp_nanos(nanos).into(),
span: Span::test_data(),
}),
_ => panic!("datetime: help example is invalid"),
};
let example_result_2 = |millis: i64| match Utc.timestamp_millis_opt(millis) {
LocalResult::Single(dt) => Some(Value::Date {
val: dt.into(),
span: Span::test_data(),
}),
_ => panic!("datetime: help example is invalid"),
})
};
vec![
Example {
description: "Convert to datetime",
description: "Convert any standard timestamp string to datetime",
example: "'27.02.2021 1:55 pm +0000' | into datetime",
result: example_result_1(1614434100,0)
#[allow(clippy::inconsistent_digit_grouping)]
result: example_result_1(1614434100_000000000),
},
Example {
description: "Convert to datetime",
example: "'2021-02-27T13:55:40+00:00' | into datetime",
result: example_result_1(1614434140, 0)
description: "Convert any standard timestamp string to datetime",
example: "'2021-02-27T13:55:40.2246+00:00' | into datetime",
#[allow(clippy::inconsistent_digit_grouping)]
result: example_result_1(1614434140_224600000),
},
Example {
description: "Convert to datetime using a custom format",
description:
"Convert non-standard timestamp string to datetime using a custom format",
example: "'20210227_135540+0000' | into datetime -f '%Y%m%d_%H%M%S%z'",
result: example_result_1(1614434140, 0)
},
Example {
description: "Convert timestamp (no larger than 8e+12) to a UTC datetime",
example: "1614434140 | into datetime",
result: example_result_1(1614434140, 0)
#[allow(clippy::inconsistent_digit_grouping)]
result: example_result_1(1614434140_000000000),
},
Example {
description:
"Convert timestamp (no larger than 8e+12) to datetime using a specified timezone offset (between -12 and 12)",
example: "1614434140 | into datetime -o +9",
result: None,
"Convert nanosecond-precision unix timestamp to a datetime with offset from UTC",
example: "1614434140123456789 | into datetime --offset -5",
#[allow(clippy::inconsistent_digit_grouping)]
result: example_result_1(1614434140_123456789),
},
Example {
description:
"Convert a millisecond-precise timestamp",
example: "1656165681720 | into datetime",
result: example_result_2(1656165681720)
description: "Convert standard (seconds) unix timestamp to a UTC datetime",
example: "1614434140 * 1_000_000_000 | into datetime",
#[allow(clippy::inconsistent_digit_grouping)]
result: example_result_1(1614434140_000000000),
},
]
}
@ -205,6 +198,9 @@ struct DatetimeFormat(String);
fn action(input: &Value, args: &Arguments, head: Span) -> Value {
let timezone = &args.zone_options;
let dateformat = &args.format_options;
const HOUR: i32 = 60 * 60;
// Check to see if input looks like a Unix timestamp (i.e. can it be parsed to an int?)
let timestamp = match input {
Value::Int { val, .. } => Ok(*val),
@ -224,66 +220,36 @@ fn action(input: &Value, args: &Arguments, head: Span) -> Value {
};
if let Ok(ts) = timestamp {
const TIMESTAMP_BOUND: i64 = 8.2e+12 as i64;
const HOUR: i32 = 3600;
if ts.abs() > TIMESTAMP_BOUND {
return Value::Error {
error: ShellError::UnsupportedInput(
"timestamp is out of range; it should between -8e+12 and 8e+12".to_string(),
format!("timestamp is {ts:?}"),
head,
// Again, can safely unwrap this from here on
input.expect_span(),
),
};
}
macro_rules! match_datetime {
($expr:expr) => {
match $expr {
LocalResult::Single(dt) => Value::Date {
dt => Value::Date {
val: dt.into(),
span: head,
},
_ => {
return Value::Error {
error: ShellError::UnsupportedInput(
"The given local datetime representation is invalid.".into(),
format!("timestamp is {:?}", ts),
head,
head,
),
};
}
}
};
}
return match timezone {
// note all these `.timestamp_nanos()` could overflow if we didn't check range in `<date> | into int`.
// default to UTC
None => {
// be able to convert chrono::Utc::now()
match ts.to_string().len() {
x if x > 13 => Value::Date {
val: Utc.timestamp_nanos(ts).into(),
span: head,
},
x if x > 10 => match_datetime!(Utc.timestamp_millis_opt(ts)),
_ => match_datetime!(Utc.timestamp_opt(ts, 0)),
}
}
None => Value::Date {
val: Utc.timestamp_nanos(ts).into(),
span: head,
},
Some(Spanned { item, span }) => match item {
Zone::Utc => match_datetime!(Utc.timestamp_opt(ts, 0)),
Zone::Local => match_datetime!(Local.timestamp_opt(ts, 0)),
Zone::Utc => match_datetime!(Utc.timestamp_nanos(ts)),
Zone::Local => match_datetime!(Local.timestamp_nanos(ts)),
Zone::East(i) => match FixedOffset::east_opt((*i as i32) * HOUR) {
Some(eastoffset) => match_datetime!(eastoffset.timestamp_opt(ts, 0)),
Some(eastoffset) => match_datetime!(eastoffset.timestamp_nanos(ts)),
None => Value::Error {
error: ShellError::DatetimeParseError(input.debug_value(), *span),
},
},
Zone::West(i) => match FixedOffset::west_opt((*i as i32) * HOUR) {
Some(westoffset) => match_datetime!(westoffset.timestamp_opt(ts, 0)),
Some(westoffset) => match_datetime!(westoffset.timestamp_nanos(ts)),
None => Value::Error {
error: ShellError::DatetimeParseError(input.debug_value(), *span),
},
@ -297,7 +263,7 @@ fn action(input: &Value, args: &Arguments, head: Span) -> Value {
},
},
};
}
};
// If input is not a timestamp, try parsing it as a string
match input {
@ -386,7 +352,7 @@ mod tests {
#[test]
fn takes_timestamp_offset() {
let date_str = Value::test_string("1614434140");
let date_str = Value::test_string("1614434140000000000");
let timezone_option = Some(Spanned {
item: Zone::East(8),
span: Span::test_data(),
@ -408,7 +374,7 @@ mod tests {
#[test]
fn takes_timestamp_offset_as_int() {
let date_int = Value::test_int(1614434140);
let date_int = Value::test_int(1614434140_000000000);
let timezone_option = Some(Spanned {
item: Zone::East(8),
span: Span::test_data(),
@ -430,7 +396,7 @@ mod tests {
#[test]
fn takes_timestamp() {
let date_str = Value::test_string("1614434140");
let date_str = Value::test_string("1614434140000000000");
let timezone_option = Some(Spanned {
item: Zone::Local,
span: Span::test_data(),
@ -451,7 +417,7 @@ mod tests {
#[test]
fn takes_timestamp_without_timezone() {
let date_str = Value::test_string("1614434140");
let date_str = Value::test_string("1614434140000000000");
let args = Arguments {
zone_options: None,
format_options: None,
@ -467,23 +433,6 @@ mod tests {
assert_eq!(actual, expected)
}
#[test]
fn takes_invalid_timestamp() {
let date_str = Value::test_string("10440970000000");
let timezone_option = Some(Spanned {
item: Zone::Utc,
span: Span::test_data(),
});
let args = Arguments {
zone_options: timezone_option,
format_options: None,
cell_paths: None,
};
let actual = action(&date_str, &args, Span::test_data());
assert_eq!(actual.get_type(), Error);
}
#[test]
fn communicates_parsing_error_given_an_invalid_datetimelike_string() {
let date_str = Value::test_string("16.11.1984 8:00 am Oops0000");

View File

@ -1,3 +1,5 @@
use chrono::{FixedOffset, TimeZone};
use crate::input_handler::{operate, CmdArgument};
use nu_engine::CallExt;
use nu_protocol::{
@ -32,7 +34,7 @@ impl Command for SubCommand {
(Type::String, Type::Int),
(Type::Number, Type::Int),
(Type::Bool, Type::Int),
// Unix timestamp in seconds
// Unix timestamp in nanoseconds
(Type::Date, Type::Int),
// TODO: Users should do this by dividing a Filesize by a Filesize explicitly
(Type::Filesize, Type::Int),
@ -124,9 +126,9 @@ impl Command for SubCommand {
}),
},
Example {
description: "Convert date to integer (Unix timestamp)",
example: "2022-02-02 | into int",
result: Some(Value::test_int(1643760000)),
description: "Convert date to integer (Unix nanosecond timestamp)",
example: "1983-04-13T12:09:14.123456789-05:00 | into int",
result: Some(Value::test_int(419101754123456789)),
},
Example {
description: "Convert to integer from binary",
@ -217,10 +219,30 @@ fn action(input: &Value, args: &Arguments, span: Span) -> Value {
Value::Int { val: 0, span }
}
}
Value::Date { val, .. } => Value::Int {
val: val.timestamp(),
span,
},
Value::Date { val, .. } => {
if val
< &FixedOffset::east_opt(0)
.expect("constant")
.with_ymd_and_hms(1677, 9, 21, 0, 12, 44)
.unwrap()
|| val
> &FixedOffset::east_opt(0)
.expect("constant")
.with_ymd_and_hms(2262, 4, 11, 23, 47, 16)
.unwrap()
{
Value::Error {
error: ShellError::IncorrectValue(
"DateTime out of range for timestamp: 1677-09-21T00:12:43Z to 2262-04-11T23:47:16".to_string(),
span),
}
} else {
Value::Int {
val: val.timestamp_nanos(),
span,
}
}
}
Value::Duration { val, .. } => Value::Int { val: *val, span },
Value::Binary { val, span } => {
use byteorder::{BigEndian, ByteOrder, LittleEndian};
@ -381,6 +403,9 @@ fn int_from_string(a_string: &str, span: Span) -> Result<i64, ShellError> {
#[cfg(test)]
mod test {
use chrono::{DateTime, FixedOffset};
use rstest::rstest;
use super::Value;
use super::*;
use nu_protocol::Type::Error;
@ -455,4 +480,57 @@ mod test {
assert_eq!(actual.get_type(), Error)
}
#[rstest]
#[case("2262-04-11T23:47:16+00:00", 0x7fffffff_ffffffff)]
#[case("1970-01-01T00:00:00+00:00", 0)]
#[case("1677-09-21T00:12:44+00:00", -0x7fffffff_ffffffff)]
fn datetime_to_int_values_that_work(
#[case] dt_in: DateTime<FixedOffset>,
#[case] int_expected: i64,
) {
let s = Value::test_date(dt_in);
let actual = action(
&s,
&Arguments {
radix: 10,
cell_paths: None,
little_endian: false,
},
Span::test_data(),
);
// ignore fractional seconds -- I don't want to hard code test values that might vary due to leap nanoseconds.
let exp_truncated = (int_expected / 1_000_000_000) * 1_000_000_000;
assert_eq!(actual, Value::test_int(exp_truncated));
}
#[rstest]
#[case("2262-04-11T23:47:17+00:00", "DateTime out of range for timestamp")]
#[case("1677-09-21T00:12:43+00:00", "DateTime out of range for timestamp")]
fn datetime_to_int_values_that_fail(
#[case] dt_in: DateTime<FixedOffset>,
#[case] err_expected: &str,
) {
let s = Value::test_date(dt_in);
let actual = action(
&s,
&Arguments {
radix: 10,
cell_paths: None,
little_endian: false,
},
Span::test_data(),
);
if let Value::Error {
error: ShellError::IncorrectValue(e, ..),
} = actual
{
assert!(
e.contains(err_expected),
"{e:?} doesn't contain {err_expected}"
);
} else {
panic!("Unexpected actual value {actual:?}")
}
}
}

View File

@ -59,6 +59,7 @@ impl Command for SubCommand {
"hour".into(),
"minute".into(),
"second".into(),
"nanosecond".into(),
"timezone".into(),
];
let vals = vec![
@ -68,6 +69,10 @@ impl Command for SubCommand {
Value::Int { val: 22, span },
Value::Int { val: 10, span },
Value::Int { val: 57, span },
Value::Int {
val: 123_000_000,
span,
},
Value::String {
val: "+02:00".to_string(),
span,
@ -89,7 +94,7 @@ impl Command for SubCommand {
},
Example {
description: "Convert a date string into a record.",
example: "'2020-04-12 22:10:57 +0200' | date to-record",
example: "'2020-04-12T22:10:57.123+02:00' | date to-record",
result: example_result_1(),
},
// TODO: This should work but does not; see https://github.com/nushell/nushell/issues/7032
@ -110,6 +115,7 @@ fn parse_date_into_table(date: Result<DateTime<FixedOffset>, Value>, head: Span)
"hour".into(),
"minute".into(),
"second".into(),
"nanosecond".into(),
"timezone".into(),
];
match date {
@ -121,6 +127,7 @@ fn parse_date_into_table(date: Result<DateTime<FixedOffset>, Value>, head: Span)
Value::int(x.hour() as i64, head),
Value::int(x.minute() as i64, head),
Value::int(x.second() as i64, head),
Value::int(x.nanosecond() as i64, head),
Value::string(x.offset().to_string(), head),
];
Value::Record {

View File

@ -59,6 +59,7 @@ impl Command for SubCommand {
"hour".into(),
"minute".into(),
"second".into(),
"nanosecond".into(),
"timezone".into(),
];
let vals = vec![
@ -68,6 +69,7 @@ impl Command for SubCommand {
Value::Int { val: 22, span },
Value::Int { val: 10, span },
Value::Int { val: 57, span },
Value::Int { val: 789, span },
Value::String {
val: "+02:00".to_string(),
span,
@ -92,7 +94,9 @@ impl Command for SubCommand {
},
Example {
description: "Convert a given date into a table.",
example: "'2020-04-12 22:10:57 +0200' | date to-table",
//todo: resolve https://github.com/bspeice/dtparse/issues/40, which truncates nanosec bits
// for now, change the example to use date literal rather than string conversion, as workaround
example: "2020-04-12T22:10:57.000000789+02:00 | date to-table",
result: example_result_1(),
},
// TODO: This should work but does not; see https://github.com/nushell/nushell/issues/7032
@ -113,6 +117,7 @@ fn parse_date_into_table(date: Result<DateTime<FixedOffset>, Value>, head: Span)
"hour".into(),
"minute".into(),
"second".into(),
"nanosecond".into(),
"timezone".into(),
];
match date {
@ -124,6 +129,7 @@ fn parse_date_into_table(date: Result<DateTime<FixedOffset>, Value>, head: Span)
Value::int(x.hour() as i64, head),
Value::int(x.minute() as i64, head),
Value::int(x.second() as i64, head),
Value::int(x.nanosecond() as i64, head),
Value::string(x.offset().to_string(), head),
];
Value::List {

View File

@ -1,3 +1,6 @@
use chrono::{DateTime, FixedOffset, NaiveDate, TimeZone};
use rstest::rstest;
use nu_test_support::{nu, pipeline};
#[test]
@ -47,3 +50,40 @@ fn into_int_binary() {
assert!(actual.out.contains("16843009"));
}
#[test]
#[ignore]
fn into_int_datetime1() {
let dt = DateTime::parse_from_rfc3339("1983-04-13T12:09:14.123456789+00:00");
eprintln!("dt debug {:?}", dt);
assert_eq!(
dt,
Ok(FixedOffset::east_opt(0)
.unwrap()
.from_local_datetime(
&NaiveDate::from_ymd_opt(1983, 4, 13)
.unwrap()
.and_hms_nano_opt(12, 9, 14, 123456789)
.unwrap()
)
.unwrap())
);
let dt_nano = dt.expect("foo").timestamp_nanos();
assert_eq!(dt_nano % 1_000_000_000, 123456789);
}
#[rstest]
#[case("1983-04-13T12:09:14.123456789-05:00", "419101754123456789")] // full precision
#[case("1983-04-13T12:09:14.456789-05:00", "419101754456789000")] // microsec
#[case("1983-04-13T12:09:14-05:00", "419101754000000000")] // sec
#[case("2052-04-13T12:09:14.123456789-05:00", "2596640954123456789")] // future date > 2038 epoch
#[case("1902-04-13T12:09:14.123456789-05:00", "-2137042245876543211")] // past date < 1970
fn into_int_datetime(#[case] time_in: &str, #[case] int_out: &str) {
let actual = nu!(
cwd: ".", pipeline(
&format!(r#""{time_in}" | into datetime --format "%+" | into int"#)
));
assert_eq!(int_out, actual.out);
}

View File

@ -1632,6 +1632,15 @@ impl Value {
span: Span::test_data(),
}
}
/// Note: Only use this for test data, *not* live data, as it will point into unknown source
/// when used in errors.
pub fn test_date(val: DateTime<FixedOffset>) -> Value {
Value::Date {
val,
span: Span::test_data(),
}
}
}
impl Default for Value {