6960837701

Committed 22 Nov 2023 02:34PM UTC coverage: 73.023% (+0.4%) from 72.607%

Build # 6960837701

Build Type

push

github

Committed by

web-flow

Commit Message

Put the units conversion info in a `ZeroTrie` (#4327)

Run Details

8 of 11 new or added lines in 2 files covered. (72.73%)

1655 existing lines in 63 files now uncovered.

45357 of 62113 relevant lines covered (73.02%)

266109.06 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

64.52

/components/datetime/src/fields/length.rs

// This file is part of ICU4X. For terms of use, please see the file
// called LICENSE at the top level of the ICU4X source tree
// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).

use core::cmp::{Ord, PartialOrd};
use core::fmt;
use displaydoc::Display;
use zerovec::ule::{AsULE, ZeroVecError, ULE};

#[cfg(doc)]
use crate::fields::{Hour, Weekday};

/// An error relating to the length of a field within a date pattern.
#[derive(Display, Debug, PartialEq, Copy, Clone)]
#[non_exhaustive]
pub enum LengthError {
    /// The length of the field string within the pattern is invalid, according to
    /// the field type and its supported field patterns in LDML. See [`FieldLength`].
    #[displaydoc("Invalid length")]
    InvalidLength,
}

#[cfg(feature = "std")]
impl std::error::Error for LengthError {}

/// An enum representing the length of a field within a date or time formatting pattern string,
/// in which the pattern field is represented as a letter occurring 1 or more times in a row, ex:
/// `MMM`, `dd`, `y`.  See the
/// [LDML documentation in UTS 35](https://unicode.org/reports/tr35/tr35-dates.html#Date_Format_Patterns)
/// for more details.
#[derive(Debug, Eq, PartialEq, Clone, Copy, Ord, PartialOrd)]
#[cfg_attr(
    feature = "datagen",
    derive(serde::Serialize, databake::Bake),
    databake(path = icu_datetime::fields),
)]
#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
#[allow(clippy::exhaustive_enums)] // part of data struct
pub enum FieldLength {
    /// Typical style is 1-2 digits. For numeric-only fields.
    One,
    /// Typical style is 2 digits. For numeric-only fields.
    TwoDigit,
    /// Abbreviated (spellout) format.
    Abbreviated,
    /// Wide / Long / Full  (spellout) format.
    Wide,
    /// Narrow / Long / Full  (spellout) format.
    Narrow,
    /// Meaning is field-dependent, for patterns that are 6 characters long. Ex: a [`Weekday`] pattern like
    /// `EEEEEE` means "Short", but `jjjjjj` or `CCCCCC` for [`Hour`] may mean
    /// "Numeric hour (2 digits, zero pad if needed), narrow dayPeriod if used". See the
    /// [LDML documentation in UTS 35](https://unicode.org/reports/tr35/tr35-dates.html#Date_Format_Patterns)
    /// for more details.
    Six,
    /// A fixed size format for numeric-only fields that is at most 127 digits.
    Fixed(u8),
    /// FieldLength::One (numeric), but overridden with a different numbering system
    NumericOverride(FieldNumericOverrides),
}

/// First index used for numeric overrides in compact FieldLength representation
///
/// Currently 17 due to decision in <https://unicode-org.atlassian.net/browse/CLDR-17217>,
/// may become 16 if the `> 16` is updated to a ` >= 16`
const FIRST_NUMERIC_OVERRIDE: u8 = 17;
/// First index used for fixed size formats in compact FieldLength representation
const FIRST_FIXED: u8 = 128;

impl FieldLength {
    #[inline]
    pub(crate) fn idx(&self) -> u8 {
        match self {
            FieldLength::One => 1,
            FieldLength::TwoDigit => 2,
            FieldLength::Abbreviated => 3,
            FieldLength::Wide => 4,
            FieldLength::Narrow => 5,
            FieldLength::Six => 6,
            FieldLength::NumericOverride(o) => FIRST_NUMERIC_OVERRIDE
                .saturating_add(*o as u8)
                .min(FIRST_FIXED - 1),
            FieldLength::Fixed(p) => FIRST_FIXED.saturating_add(*p), /* truncate to at most 127 digits to avoid overflow */
        }
    }

    #[inline]
    pub(crate) fn from_idx(idx: u8) -> Result<Self, LengthError> {
        Ok(match idx {
            1 => Self::One,
            2 => Self::TwoDigit,
            3 => Self::Abbreviated,
            4 => Self::Wide,
            5 => Self::Narrow,
            6 => Self::Six,
            idx => {
                if idx < FIRST_NUMERIC_OVERRIDE {
                    return Err(LengthError::InvalidLength);
                }
                if idx < FIRST_FIXED {
                    Self::NumericOverride((idx - FIRST_NUMERIC_OVERRIDE).try_into()?)
                } else {
                    Self::Fixed(idx - FIRST_FIXED)
                }
            }
        })
    }

    #[inline]
    #[cfg(feature = "datagen")]
    pub(crate) fn to_len(self) -> usize {
        match self {
            FieldLength::One => 1,
            FieldLength::TwoDigit => 2,
            FieldLength::Abbreviated => 3,
            FieldLength::Wide => 4,
            FieldLength::Narrow => 5,
            FieldLength::Six => 6,
            FieldLength::NumericOverride(o) => FIRST_NUMERIC_OVERRIDE as usize + o as usize,
            FieldLength::Fixed(p) => p as usize,
        }
    }

    /// UTS 35 defines several 1 and 2 symbols to be the same as 3 symbols (abbreviated).
    /// For example, 'a' represents an abbreviated day period, the same as 'aaa'.
    ///
    /// This function maps field lengths 1 and 2 to field length 3.
    #[cfg(feature = "experimental")]
    pub(crate) fn numeric_to_abbr(self) -> Self {
        match self {
            FieldLength::One | FieldLength::TwoDigit => FieldLength::Abbreviated,
            other => other,
        }
    }
}

#[repr(transparent)]
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub struct FieldLengthULE(u8);

impl AsULE for FieldLength {
    type ULE = FieldLengthULE;
    fn to_unaligned(self) -> Self::ULE {
        FieldLengthULE(self.idx())
    }
    fn from_unaligned(unaligned: Self::ULE) -> Self {
        #[allow(clippy::unwrap_used)] // OK because the ULE is pre-validated
        Self::from_idx(unaligned.0).unwrap()
    }
}

impl FieldLengthULE {
    #[inline]
    pub(crate) fn validate_byte(byte: u8) -> Result<(), ZeroVecError> {
        FieldLength::from_idx(byte)
            .map(|_| ())
            .map_err(|_| ZeroVecError::parse::<FieldLength>())
    }
}

// Safety checklist for ULE:
//
// 1. Must not include any uninitialized or padding bytes (true since transparent over a ULE).
// 2. Must have an alignment of 1 byte (true since transparent over a ULE).
// 3. ULE::validate_byte_slice() checks that the given byte slice represents a valid slice.
// 4. ULE::validate_byte_slice() checks that the given byte slice has a valid length
//    (true since transparent over a type of size 1).
// 5. All other methods must be left with their default impl.
// 6. Byte equality is semantic equality.
unsafe impl ULE for FieldLengthULE {
    fn validate_byte_slice(bytes: &[u8]) -> Result<(), ZeroVecError> {
        for byte in bytes {
            Self::validate_byte(*byte)?;
        }
        Ok(())
    }
}

/// Various numeric overrides for datetime patterns
/// as found in CLDR
#[derive(Debug, Eq, PartialEq, Clone, Copy, Ord, PartialOrd)]
#[cfg_attr(
    feature = "datagen",
    derive(serde::Serialize, databake::Bake),
    databake(path = icu_datetime::fields),
)]
#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
#[non_exhaustive]
pub enum FieldNumericOverrides {
    /// `hanidec`
    Hanidec = 0,
    /// `hanidays`
    Hanidays = 1,
    /// `hebr`
    Hebr = 2,
    /// `romanlow`
    Romanlow = 3,
    /// `jpnyear`
    Jpnyear = 4,
}

impl TryFrom<u8> for FieldNumericOverrides {
    type Error = LengthError;
    fn try_from(other: u8) -> Result<Self, LengthError> {
        Ok(match other {
            0 => Self::Hanidec,
            1 => Self::Hanidays,
            2 => Self::Hebr,
            3 => Self::Romanlow,
            4 => Self::Jpnyear,
            _ => return Err(LengthError::InvalidLength),
        })
    }
}

impl FieldNumericOverrides {
    /// Conver this to the corresponding string code
    pub fn as_str(self) -> &'static str {
        match self {
            Self::Hanidec => "hanidec",
            Self::Hanidays => "hanidays",
            Self::Hebr => "hebr",
            Self::Romanlow => "romanlow",
            Self::Jpnyear => "jpnyear",
        }
    }
}

impl fmt::Display for FieldNumericOverrides {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        self.as_str().fmt(f)
    }
}

1	// This file is part of ICU4X. For terms of use, please see the file	2✔
2	// called LICENSE at the top level of the ICU4X source tree
3	// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
4
5	use core::cmp::{Ord, PartialOrd};
6	use core::fmt;
7	use displaydoc::Display;
8	use zerovec::ule::{AsULE, ZeroVecError, ULE};
9
10	#[cfg(doc)]
11	use crate::fields::{Hour, Weekday};
12
13	/// An error relating to the length of a field within a date pattern.
14	#[derive(Display, Debug, PartialEq, Copy, Clone)]	×
15	#[non_exhaustive]
16	pub enum LengthError {
17	/// The length of the field string within the pattern is invalid, according to
18	/// the field type and its supported field patterns in LDML. See [`FieldLength`].
19	#[displaydoc("Invalid length")]
20	InvalidLength,
21	}
22
23	#[cfg(feature = "std")]
24	impl std::error::Error for LengthError {}
25
26	/// An enum representing the length of a field within a date or time formatting pattern string,
27	/// in which the pattern field is represented as a letter occurring 1 or more times in a row, ex:
28	/// `MMM`, `dd`, `y`. See the
29	/// [LDML documentation in UTS 35](https://unicode.org/reports/tr35/tr35-dates.html#Date_Format_Patterns)
30	/// for more details.
31	#[derive(Debug, Eq, PartialEq, Clone, Copy, Ord, PartialOrd)]	41,002✔
32	#[cfg_attr(
33	feature = "datagen",
34	derive(serde::Serialize, databake::Bake),	10✔
35	databake(path = icu_datetime::fields),
36	)]
37	#[cfg_attr(feature = "serde", derive(serde::Deserialize))]	20✔
38	#[allow(clippy::exhaustive_enums)] // part of data struct
39	pub enum FieldLength {	5✔
40	/// Typical style is 1-2 digits. For numeric-only fields.
41	One,
42	/// Typical style is 2 digits. For numeric-only fields.
43	TwoDigit,
44	/// Abbreviated (spellout) format.
45	Abbreviated,
46	/// Wide / Long / Full (spellout) format.
47	Wide,
48	/// Narrow / Long / Full (spellout) format.
49	Narrow,
50	/// Meaning is field-dependent, for patterns that are 6 characters long. Ex: a [`Weekday`] pattern like
51	/// `EEEEEE` means "Short", but `jjjjjj` or `CCCCCC` for [`Hour`] may mean
52	/// "Numeric hour (2 digits, zero pad if needed), narrow dayPeriod if used". See the
53	/// [LDML documentation in UTS 35](https://unicode.org/reports/tr35/tr35-dates.html#Date_Format_Patterns)
54	/// for more details.
55	Six,
56	/// A fixed size format for numeric-only fields that is at most 127 digits.
57	Fixed(u8),	53✔
58	/// FieldLength::One (numeric), but overridden with a different numbering system
59	NumericOverride(FieldNumericOverrides),	×
60	}
61
62	/// First index used for numeric overrides in compact FieldLength representation
63	///
64	/// Currently 17 due to decision in <https://unicode-org.atlassian.net/browse/CLDR-17217>,
65	/// may become 16 if the `> 16` is updated to a ` >= 16`
66	const FIRST_NUMERIC_OVERRIDE: u8 = 17;
67	/// First index used for fixed size formats in compact FieldLength representation
68	const FIRST_FIXED: u8 = 128;
69
70	impl FieldLength {
71	#[inline]
72	pub(crate) fn idx(&self) -> u8 {	3,375✔
73	match self {	3,375✔
74	FieldLength::One => 1,	1,136✔
75	FieldLength::TwoDigit => 2,	983✔
76	FieldLength::Abbreviated => 3,	208✔
77	FieldLength::Wide => 4,	874✔
78	FieldLength::Narrow => 5,	109✔
79	FieldLength::Six => 6,	57✔
UNCOV 80	FieldLength::NumericOverride(o) => FIRST_NUMERIC_OVERRIDE	×
UNCOV 81	.saturating_add(*o as u8)	×
82	.min(FIRST_FIXED - 1),
83	FieldLength::Fixed(p) => FIRST_FIXED.saturating_add(p), / truncate to at most 127 digits to avoid overflow */	8✔
84	}
85	}	3,375✔
86
87	#[inline]
88	pub(crate) fn from_idx(idx: u8) -> Result<Self, LengthError> {	16,831✔
89	Ok(match idx {	33,660✔
90	1 => Self::One,	7,736✔
91	2 => Self::TwoDigit,	4,775✔
92	3 => Self::Abbreviated,	1,092✔
93	4 => Self::Wide,	2,617✔
94	5 => Self::Narrow,	416✔
95	6 => Self::Six,	143✔
96	idx => {	52✔
97	if idx < FIRST_NUMERIC_OVERRIDE {	52✔
98	return Err(LengthError::InvalidLength);	2✔
99	}
100	if idx < FIRST_FIXED {	50✔
UNCOV 101	Self::NumericOverride((idx - FIRST_NUMERIC_OVERRIDE).try_into()?)	×
102	} else {
103	Self::Fixed(idx - FIRST_FIXED)	50✔
104	}
105	}
106	})
107	}	16,831✔
108
109	#[inline]
110	#[cfg(feature = "datagen")]
111	pub(crate) fn to_len(self) -> usize {	219✔
112	match self {	219✔
113	FieldLength::One => 1,	118✔
114	FieldLength::TwoDigit => 2,	46✔
115	FieldLength::Abbreviated => 3,	16✔
116	FieldLength::Wide => 4,	17✔
117	FieldLength::Narrow => 5,	13✔
118	FieldLength::Six => 6,	2✔
UNCOV 119	FieldLength::NumericOverride(o) => FIRST_NUMERIC_OVERRIDE as usize + o as usize,	×
120	FieldLength::Fixed(p) => p as usize,	7✔
121	}
122	}	219✔
123
124	/// UTS 35 defines several 1 and 2 symbols to be the same as 3 symbols (abbreviated).
125	/// For example, 'a' represents an abbreviated day period, the same as 'aaa'.
126	///
127	/// This function maps field lengths 1 and 2 to field length 3.
128	#[cfg(feature = "experimental")]
129	pub(crate) fn numeric_to_abbr(self) -> Self {	150✔
130	match self {	150✔
131	FieldLength::One \| FieldLength::TwoDigit => FieldLength::Abbreviated,	46✔
132	other => other,	104✔
133	}
134	}	150✔
135	}
136
137	#[repr(transparent)]
UNCOV 138	#[derive(Debug, Copy, Clone, PartialEq, Eq)]	×
UNCOV 139	pub struct FieldLengthULE(u8);	×
140
141	impl AsULE for FieldLength {
142	type ULE = FieldLengthULE;
143	fn to_unaligned(self) -> Self::ULE {	6✔
144	FieldLengthULE(self.idx())	6✔
145	}	6✔
146	fn from_unaligned(unaligned: Self::ULE) -> Self {	4✔
147	#[allow(clippy::unwrap_used)] // OK because the ULE is pre-validated
148	Self::from_idx(unaligned.0).unwrap()	4✔
149	}	4✔
150	}
151
152	impl FieldLengthULE {
153	#[inline]
154	pub(crate) fn validate_byte(byte: u8) -> Result<(), ZeroVecError> {	120✔
155	FieldLength::from_idx(byte)	120✔
156	.map(\|_\| ())	120✔
UNCOV 157	.map_err(\|_\| ZeroVecError::parse::<FieldLength>())	×
158	}	120✔
159	}
160
161	// Safety checklist for ULE:
162	//
163	// 1. Must not include any uninitialized or padding bytes (true since transparent over a ULE).
164	// 2. Must have an alignment of 1 byte (true since transparent over a ULE).
165	// 3. ULE::validate_byte_slice() checks that the given byte slice represents a valid slice.
166	// 4. ULE::validate_byte_slice() checks that the given byte slice has a valid length
167	// (true since transparent over a type of size 1).
168	// 5. All other methods must be left with their default impl.
169	// 6. Byte equality is semantic equality.
170	unsafe impl ULE for FieldLengthULE {
171	fn validate_byte_slice(bytes: &[u8]) -> Result<(), ZeroVecError> {	2✔
172	for byte in bytes {	4✔
173	Self::validate_byte(*byte)?;	2✔
174	}
175	Ok(())	2✔
176	}	2✔
177	}
178
179	/// Various numeric overrides for datetime patterns
180	/// as found in CLDR
UNCOV 181	#[derive(Debug, Eq, PartialEq, Clone, Copy, Ord, PartialOrd)]	×
182	#[cfg_attr(
183	feature = "datagen",
184	derive(serde::Serialize, databake::Bake),	×
185	databake(path = icu_datetime::fields),
186	)]
187	#[cfg_attr(feature = "serde", derive(serde::Deserialize))]	×
188	#[non_exhaustive]
189	pub enum FieldNumericOverrides {	×
190	/// `hanidec`
191	Hanidec = 0,
192	/// `hanidays`
193	Hanidays = 1,
194	/// `hebr`
195	Hebr = 2,
196	/// `romanlow`
197	Romanlow = 3,
198	/// `jpnyear`
199	Jpnyear = 4,
200	}
201
202	impl TryFrom<u8> for FieldNumericOverrides {
203	type Error = LengthError;
204	fn try_from(other: u8) -> Result<Self, LengthError> {	×
UNCOV 205	Ok(match other {	×
UNCOV 206	0 => Self::Hanidec,	×
UNCOV 207	1 => Self::Hanidays,	×
208	2 => Self::Hebr,	×
209	3 => Self::Romanlow,	×
210	4 => Self::Jpnyear,	×
UNCOV 211	_ => return Err(LengthError::InvalidLength),	×
212	})
UNCOV 213	}	×
214	}
215
216	impl FieldNumericOverrides {
217	/// Conver this to the corresponding string code
UNCOV 218	pub fn as_str(self) -> &'static str {	×
UNCOV 219	match self {	×
UNCOV 220	Self::Hanidec => "hanidec",	×
UNCOV 221	Self::Hanidays => "hanidays",	×
UNCOV 222	Self::Hebr => "hebr",	×
UNCOV 223	Self::Romanlow => "romanlow",	×
UNCOV 224	Self::Jpnyear => "jpnyear",	×
225	}
UNCOV 226	}	×
227	}
228
229	impl fmt::Display for FieldNumericOverrides {
UNCOV 230	fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {	×
UNCOV 231	self.as_str().fmt(f)	×
UNCOV 232	}	×
233	}

zbraniecki / icu4x / 6960837701

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous