11904027177

Committed 19 Nov 2024 12:33AM UTC coverage: 75.477% (+0.3%) from 75.174%

Build # 11904027177

Build Type

push

github

Committed by

web-flow

Commit Message

Move DateTimePattern into pattern module (#5834)

#1317

Also removes `NeoNeverMarker` and fixes #5689

Run Details

258 of 319 new or added lines in 6 files covered. (80.88%)

6967 existing lines in 278 files now uncovered.

54522 of 72237 relevant lines covered (75.48%)

655305.49 hits per line

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

85.71

/components/collections/src/codepointinvliststringlist/mod.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 ).

//! This module provides functionality for querying of sets of Unicode code points and strings.
//!
//! It depends on [`CodePointInversionList`] to efficiently represent Unicode code points, while
//! it also maintains a list of strings in the set.
//!
//! It is an implementation of the existing [ICU4C UnicodeSet API](https://unicode-org.github.io/icu-docs/apidoc/released/icu4c/classicu_1_1UnicodeSet.html).

use crate::codepointinvlist::{
    CodePointInversionList, CodePointInversionListBuilder, CodePointInversionListULE,
};
use alloc::string::{String, ToString};
use alloc::vec::Vec;
use displaydoc::Display;
use yoke::Yokeable;
use zerofrom::ZeroFrom;
use zerovec::{VarZeroSlice, VarZeroVec};

/// A data structure providing a concrete implementation of a set of code points and strings,
/// using an inversion list for the code points.
///
/// This is what ICU4C calls a `UnicodeSet`.
#[zerovec::make_varule(CodePointInversionListAndStringListULE)]
#[zerovec::skip_derive(Ord)]
#[zerovec::derive(Debug)]
#[derive(Debug, Eq, PartialEq, Clone, Yokeable, ZeroFrom)]
// Valid to auto-derive Deserialize because the invariants are weakly held
#[cfg_attr(feature = "serde", derive(serde::Deserialize, serde::Serialize))]
#[cfg_attr(feature = "serde", zerovec::derive(Serialize, Deserialize, Debug))]
pub struct CodePointInversionListAndStringList<'data> {
    #[cfg_attr(feature = "serde", serde(borrow))]
    #[zerovec::varule(CodePointInversionListULE)]
    cp_inv_list: CodePointInversionList<'data>,
    // Invariants (weakly held):
    //   - no input string is length 1 (a length 1 string should be a single code point)
    //   - the string list is sorted
    //   - the elements in the string list are unique
    #[cfg_attr(feature = "serde", serde(borrow))]
    str_list: VarZeroVec<'data, str>,
}

#[cfg(feature = "databake")]
impl databake::Bake for CodePointInversionListAndStringList<'_> {
    fn bake(&self, env: &databake::CrateEnv) -> databake::TokenStream {
        env.insert("icu_collections");
        let cp_inv_list = self.cp_inv_list.bake(env);
        let str_list = self.str_list.bake(env);
        // Safe because our parts are safe.
        databake::quote! {
            icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList::from_parts_unchecked(#cp_inv_list, #str_list)
        }
    }
}

#[cfg(feature = "databake")]
impl databake::BakeSize for CodePointInversionListAndStringList<'_> {
    fn borrows_size(&self) -> usize {
        self.cp_inv_list.borrows_size() + self.str_list.borrows_size()
    }
}

impl<'data> CodePointInversionListAndStringList<'data> {
    /// Returns a new [`CodePointInversionListAndStringList`] from both a [`CodePointInversionList`] for the
    /// code points and a [`VarZeroVec`]`<`[`str`]`>` of strings.
    pub fn try_from(
        cp_inv_list: CodePointInversionList<'data>,
        str_list: VarZeroVec<'data, str>,
    ) -> Result<Self, InvalidStringList> {
        // Verify invariants:
        // Do so by using the equivalent of str_list.iter().windows(2) to get
        // overlapping windows of size 2. The above putative code is not possible
        // because `.windows()` exists on a slice, but VarZeroVec cannot return a slice
        // because the non-fixed size elements necessitate at least some type
        // of allocation.
        {
            let mut it = str_list.iter();
            if let Some(mut x) = it.next() {
                if x.len() == 1 {
                    return Err(InvalidStringList::InvalidStringLength(x.to_string()));
                }
                for y in it {
                    if x.len() == 1 {
                        return Err(InvalidStringList::InvalidStringLength(x.to_string()));
                    } else if x == y {
                        return Err(InvalidStringList::StringListNotUnique(x.to_string()));
                    } else if x > y {
                        return Err(InvalidStringList::StringListNotSorted(
                            x.to_string(),
                            y.to_string(),
                        ));
                    }

                    // Next window begins. Update `x` here, `y` will be updated in next loop iteration.
                    x = y;
                }
            }
        }

        Ok(CodePointInversionListAndStringList {
            cp_inv_list,
            str_list,
        })
    }

    #[doc(hidden)] // databake internal
    pub const fn from_parts_unchecked(
        cp_inv_list: CodePointInversionList<'data>,
        str_list: VarZeroVec<'data, str>,
    ) -> Self {
        CodePointInversionListAndStringList {
            cp_inv_list,
            str_list,
        }
    }

    /// Returns the number of elements in this set (its cardinality).
    /// Note than the elements of a set may include both individual
    /// codepoints and strings.
    pub fn size(&self) -> usize {
        self.cp_inv_list.size() + self.str_list.len()
    }

    /// Return true if this set contains multi-code point strings or the empty string.
    pub fn has_strings(&self) -> bool {
        !self.str_list.is_empty()
    }

    ///
    /// # Examples
    /// ```
    /// use icu::collections::codepointinvlist::CodePointInversionList;
    /// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
    /// use zerovec::VarZeroVec;
    ///
    /// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
    /// let cp_list =
    ///    CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
    /// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
    /// let str_list = VarZeroVec::<str>::from(str_slice);
    ///
    /// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
    ///
    /// assert!(cpilsl.contains_str("bmp_max"));
    /// assert!(cpilsl.contains_str(""));
    /// assert!(cpilsl.contains_str("A"));
    /// assert!(cpilsl.contains_str("ቔ"));  // U+1254 ETHIOPIC SYLLABLE QHEE
    /// assert!(!cpilsl.contains_str("bazinga!"));
    /// ```
    pub fn contains_str(&self, s: &str) -> bool {
        let mut chars = s.chars();
        if let Some(first_char) = chars.next() {
            if chars.next().is_none() {
                return self.contains(first_char);
            }
        }
        self.str_list.binary_search(s).is_ok()
    }

    ///
    /// # Examples
    /// ```
    /// use icu::collections::codepointinvlist::CodePointInversionList;
    /// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
    /// use zerovec::VarZeroVec;
    ///
    /// let cp_slice = &[0, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];
    /// let cp_list =
    ///     CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
    /// let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];
    /// let str_list = VarZeroVec::<str>::from(str_slice);
    ///
    /// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
    ///
    /// assert!(cpilsl.contains32(0));
    /// assert!(cpilsl.contains32(0x0042));
    /// assert!(!cpilsl.contains32(0x0080));
    /// ```
    pub fn contains32(&self, cp: u32) -> bool {
        self.cp_inv_list.contains32(cp)
    }

    ///
    /// # Examples
    /// ```
    /// use icu::collections::codepointinvlist::CodePointInversionList;
    /// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
    /// use zerovec::VarZeroVec;
    ///
    /// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
    /// let cp_list =
    ///    CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
    /// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
    /// let str_list = VarZeroVec::<str>::from(str_slice);
    ///
    /// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
    ///
    /// assert!(cpilsl.contains('A'));
    /// assert!(cpilsl.contains('ቔ'));  // U+1254 ETHIOPIC SYLLABLE QHEE
    /// assert!(!cpilsl.contains('\u{1_0000}'));
    /// assert!(!cpilsl.contains('🨫'));  // U+1FA2B NEUTRAL CHESS TURNED QUEEN
    pub fn contains(&self, ch: char) -> bool {
        self.contains32(ch as u32)
    }

    /// Access the underlying [`CodePointInversionList`].
    pub fn code_points(&self) -> &CodePointInversionList<'data> {
        &self.cp_inv_list
    }

    /// Access the contained strings.
    pub fn strings(&self) -> &VarZeroSlice<str> {
        &self.str_list
    }
}

impl<'a> FromIterator<&'a str> for CodePointInversionListAndStringList<'_> {
    fn from_iter<I>(it: I) -> Self
    where
        I: IntoIterator<Item = &'a str>,
    {
        let mut builder = CodePointInversionListBuilder::new();
        let mut strings = Vec::<&str>::new();
        for s in it {
            let mut chars = s.chars();
            if let Some(first_char) = chars.next() {
                if chars.next().is_none() {
                    builder.add_char(first_char);
                    continue;
                }
            }
            strings.push(s);
        }

        // Ensure that the string list is sorted. If not, the binary search that
        // is used for `.contains(&str)` will return garbage output.
        strings.sort_unstable();
        strings.dedup();

        let cp_inv_list = builder.build();
        let str_list = VarZeroVec::<str>::from(&strings);

        CodePointInversionListAndStringList {
            cp_inv_list,
            str_list,
        }
    }
}

/// Custom Errors for [`CodePointInversionListAndStringList`].
#[derive(Display, Debug)]
pub enum InvalidStringList {
    /// A string in the string list had an invalid length
    #[displaydoc("Invalid string length for string: {0}")]
    InvalidStringLength(String),
    /// A string in the string list appears more than once
    #[displaydoc("String list has duplicate: {0}")]
    StringListNotUnique(String),
    /// Two strings in the string list compare to each other opposite of sorted order
    #[displaydoc("Strings in string list not in sorted order: ({0}, {1})")]
    StringListNotSorted(String, String),
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_size_has_strings() {
        let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];
        let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
        let str_slice = &["ascii_max", "bmp_max", "unicode_max", "zero"];
        let str_list = VarZeroVec::<str>::from(str_slice);

        let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();

        assert!(cpilsl.has_strings());
        assert_eq!(8, cpilsl.size());
    }

    #[test]
    fn test_empty_string_allowed() {
        let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];
        let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
        let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];
        let str_list = VarZeroVec::<str>::from(str_slice);

        let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();

        assert!(cpilsl.has_strings());
        assert_eq!(9, cpilsl.size());
    }

    #[test]
    fn test_invalid_string() {
        let cp_slice = &[0, 1];
        let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
        let str_slice = &["a"];
        let str_list = VarZeroVec::<str>::from(str_slice);

        let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);

        assert!(matches!(
            cpilsl,
            Err(InvalidStringList::InvalidStringLength(_))
        ));
    }

    #[test]
    fn test_invalid_string_list_has_duplicate() {
        let cp_slice = &[0, 1];
        let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
        let str_slice = &["abc", "abc"];
        let str_list = VarZeroVec::<str>::from(str_slice);

        let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);

        assert!(matches!(
            cpilsl,
            Err(InvalidStringList::StringListNotUnique(_))
        ));
    }

    #[test]
    fn test_invalid_string_list_not_sorted() {
        let cp_slice = &[0, 1];
        let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
        let str_slice = &["xyz", "abc"];
        let str_list = VarZeroVec::<str>::from(str_slice);

        let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);

        assert!(matches!(
            cpilsl,
            Err(InvalidStringList::StringListNotSorted(_, _))
        ));
    }

    #[test]
    fn test_from_iter_invariants() {
        let in_strs_1 = ["a", "abc", "xyz", "abc"];
        let in_strs_2 = ["xyz", "abc", "a", "abc"];

        let cpilsl_1 = CodePointInversionListAndStringList::from_iter(in_strs_1);
        let cpilsl_2 = CodePointInversionListAndStringList::from_iter(in_strs_2);

        assert_eq!(cpilsl_1, cpilsl_2);

        assert!(cpilsl_1.has_strings());
        assert!(cpilsl_1.contains_str("abc"));
        assert!(cpilsl_1.contains_str("xyz"));
        assert!(!cpilsl_1.contains_str("def"));

        assert_eq!(1, cpilsl_1.cp_inv_list.size());
        assert!(cpilsl_1.contains('a'));
        assert!(!cpilsl_1.contains('0'));
        assert!(!cpilsl_1.contains('q'));

        assert_eq!(3, cpilsl_1.size());
    }
}

1	// This file is part of ICU4X. For terms of use, please see the file
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	//! This module provides functionality for querying of sets of Unicode code points and strings.
6	//!
7	//! It depends on [`CodePointInversionList`] to efficiently represent Unicode code points, while
8	//! it also maintains a list of strings in the set.
9	//!
10	//! It is an implementation of the existing [ICU4C UnicodeSet API](https://unicode-org.github.io/icu-docs/apidoc/released/icu4c/classicu_1_1UnicodeSet.html).
11
12	use crate::codepointinvlist::{
13	CodePointInversionList, CodePointInversionListBuilder, CodePointInversionListULE,
14	};
15	use alloc::string::{String, ToString};
16	use alloc::vec::Vec;
17	use displaydoc::Display;
18	use yoke::Yokeable;
19	use zerofrom::ZeroFrom;
20	use zerovec::{VarZeroSlice, VarZeroVec};
21
22	/// A data structure providing a concrete implementation of a set of code points and strings,
23	/// using an inversion list for the code points.
24	///
25	/// This is what ICU4C calls a `UnicodeSet`.
26	#[zerovec::make_varule(CodePointInversionListAndStringListULE)]	963,185✔
27	#[zerovec::skip_derive(Ord)]
28	#[zerovec::derive(Debug)]
29	#[derive(Debug, Eq, PartialEq, Clone, Yokeable, ZeroFrom)]	409✔
30	// Valid to auto-derive Deserialize because the invariants are weakly held
31	#[cfg_attr(feature = "serde", derive(serde::Deserialize, serde::Serialize))]	273✔
32	#[cfg_attr(feature = "serde", zerovec::derive(Serialize, Deserialize, Debug))]
33	pub struct CodePointInversionListAndStringList<'data> {
34	#[cfg_attr(feature = "serde", serde(borrow))]
35	#[zerovec::varule(CodePointInversionListULE)]
36	cp_inv_list: CodePointInversionList<'data>,	316✔
37	// Invariants (weakly held):
38	// - no input string is length 1 (a length 1 string should be a single code point)
39	// - the string list is sorted
40	// - the elements in the string list are unique
41	#[cfg_attr(feature = "serde", serde(borrow))]
42	str_list: VarZeroVec<'data, str>,	316✔
43	}
44
45	#[cfg(feature = "databake")]
46	impl databake::Bake for CodePointInversionListAndStringList<'_> {
47	fn bake(&self, env: &databake::CrateEnv) -> databake::TokenStream {	×
48	env.insert("icu_collections");	×
49	let cp_inv_list = self.cp_inv_list.bake(env);	×
UNCOV 50	let str_list = self.str_list.bake(env);	×
51	// Safe because our parts are safe.
UNCOV 52	databake::quote! {	×
53	icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList::from_parts_unchecked(#cp_inv_list, #str_list)
54	}
UNCOV 55	}	×
56	}
57
58	#[cfg(feature = "databake")]
59	impl databake::BakeSize for CodePointInversionListAndStringList<'_> {
UNCOV 60	fn borrows_size(&self) -> usize {	×
UNCOV 61	self.cp_inv_list.borrows_size() + self.str_list.borrows_size()	×
UNCOV 62	}	×
63	}
64
65	impl<'data> CodePointInversionListAndStringList<'data> {
66	/// Returns a new [`CodePointInversionListAndStringList`] from both a [`CodePointInversionList`] for the
67	/// code points and a [`VarZeroVec`]`<`[`str`]`>` of strings.
68	pub fn try_from(	646✔
69	cp_inv_list: CodePointInversionList<'data>,
70	str_list: VarZeroVec<'data, str>,
71	) -> Result<Self, InvalidStringList> {
72	// Verify invariants:
73	// Do so by using the equivalent of str_list.iter().windows(2) to get
74	// overlapping windows of size 2. The above putative code is not possible
75	// because `.windows()` exists on a slice, but VarZeroVec cannot return a slice
76	// because the non-fixed size elements necessitate at least some type
77	// of allocation.
78	{
79	let mut it = str_list.iter();	646✔
80	if let Some(mut x) = it.next() {	644✔
81	if x.len() == 1 {	68✔
82	return Err(InvalidStringList::InvalidStringLength(x.to_string()));	1✔
83	}
84	for y in it {	502✔
85	if x.len() == 1 {	436✔
UNCOV 86	return Err(InvalidStringList::InvalidStringLength(x.to_string()));	×
87	} else if x == y {	436✔
88	return Err(InvalidStringList::StringListNotUnique(x.to_string()));	1✔
89	} else if x > y {	435✔
90	return Err(InvalidStringList::StringListNotSorted(	1✔
91	x.to_string(),	1✔
92	y.to_string(),	1✔
UNCOV 93	));	×
94	}
95
96	// Next window begins. Update `x` here, `y` will be updated in next loop iteration.
97	x = y;	434✔
98	}
99	}
100	}
101
102	Ok(CodePointInversionListAndStringList {	641✔
103	cp_inv_list,	641✔
104	str_list,	641✔
105	})
106	}	644✔
107
108	#[doc(hidden)] // databake internal
109	pub const fn from_parts_unchecked(	×
110	cp_inv_list: CodePointInversionList<'data>,
111	str_list: VarZeroVec<'data, str>,
112	) -> Self {
UNCOV 113	CodePointInversionListAndStringList {	×
114	cp_inv_list,
115	str_list,
116	}
UNCOV 117	}	×
118
119	/// Returns the number of elements in this set (its cardinality).
120	/// Note than the elements of a set may include both individual
121	/// codepoints and strings.
122	pub fn size(&self) -> usize {	147✔
123	self.cp_inv_list.size() + self.str_list.len()	147✔
124	}	147✔
125
126	/// Return true if this set contains multi-code point strings or the empty string.
127	pub fn has_strings(&self) -> bool {	80✔
128	!self.str_list.is_empty()	80✔
129	}	80✔
130
131	///
132	/// # Examples
133	/// ```
134	/// use icu::collections::codepointinvlist::CodePointInversionList;
135	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
136	/// use zerovec::VarZeroVec;
137	///
138	/// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
139	/// let cp_list =
140	/// CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
141	/// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
142	/// let str_list = VarZeroVec::<str>::from(str_slice);
143	///
144	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
145	///
146	/// assert!(cpilsl.contains_str("bmp_max"));
147	/// assert!(cpilsl.contains_str(""));
148	/// assert!(cpilsl.contains_str("A"));
149	/// assert!(cpilsl.contains_str("ቔ")); // U+1254 ETHIOPIC SYLLABLE QHEE
150	/// assert!(!cpilsl.contains_str("bazinga!"));
151	/// ```
152	pub fn contains_str(&self, s: &str) -> bool {	15,903✔
153	let mut chars = s.chars();	15,903✔
154	if let Some(first_char) = chars.next() {	15,903✔
155	if chars.next().is_none() {	7,975✔
156	return self.contains(first_char);	7,941✔
157	}
158	}
159	self.str_list.binary_search(s).is_ok()	7,962✔
160	}	15,903✔
161
162	///
163	/// # Examples
164	/// ```
165	/// use icu::collections::codepointinvlist::CodePointInversionList;
166	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
167	/// use zerovec::VarZeroVec;
168	///
169	/// let cp_slice = &[0, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];
170	/// let cp_list =
171	/// CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
172	/// let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];
173	/// let str_list = VarZeroVec::<str>::from(str_slice);
174	///
175	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
176	///
177	/// assert!(cpilsl.contains32(0));
178	/// assert!(cpilsl.contains32(0x0042));
179	/// assert!(!cpilsl.contains32(0x0080));
180	/// ```
181	pub fn contains32(&self, cp: u32) -> bool {	240,634✔
182	self.cp_inv_list.contains32(cp)	240,634✔
183	}	240,634✔
184
185	///
186	/// # Examples
187	/// ```
188	/// use icu::collections::codepointinvlist::CodePointInversionList;
189	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
190	/// use zerovec::VarZeroVec;
191	///
192	/// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
193	/// let cp_list =
194	/// CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();
195	/// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
196	/// let str_list = VarZeroVec::<str>::from(str_slice);
197	///
198	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
199	///
200	/// assert!(cpilsl.contains('A'));
201	/// assert!(cpilsl.contains('ቔ')); // U+1254 ETHIOPIC SYLLABLE QHEE
202	/// assert!(!cpilsl.contains('\u{1_0000}'));
203	/// assert!(!cpilsl.contains('🨫')); // U+1FA2B NEUTRAL CHESS TURNED QUEEN
204	pub fn contains(&self, ch: char) -> bool {	240,630✔
205	self.contains32(ch as u32)	240,630✔
206	}	240,630✔
207
208	/// Access the underlying [`CodePointInversionList`].
209	pub fn code_points(&self) -> &CodePointInversionList<'data> {	568✔
210	&self.cp_inv_list	568✔
211	}	568✔
212
213	/// Access the contained strings.
214	pub fn strings(&self) -> &VarZeroSlice<str> {	232,870✔
215	&self.str_list	232,870✔
216	}	232,870✔
217	}
218
219	impl<'a> FromIterator<&'a str> for CodePointInversionListAndStringList<'_> {
220	fn from_iter<I>(it: I) -> Self	94✔
221	where
222	I: IntoIterator<Item = &'a str>,
223	{
224	let mut builder = CodePointInversionListBuilder::new();	94✔
225	let mut strings = Vec::<&str>::new();	94✔
226	for s in it {	4,921✔
227	let mut chars = s.chars();	4,827✔
228	if let Some(first_char) = chars.next() {	4,827✔
229	if chars.next().is_none() {	4,827✔
230	builder.add_char(first_char);	4,794✔
231	continue;
232	}
233	}
234	strings.push(s);	33✔
235	}	94✔
236
237	// Ensure that the string list is sorted. If not, the binary search that
238	// is used for `.contains(&str)` will return garbage output.
239	strings.sort_unstable();	94✔
240	strings.dedup();	94✔
241
242	let cp_inv_list = builder.build();	94✔
243	let str_list = VarZeroVec::<str>::from(&strings);	94✔
244
245	CodePointInversionListAndStringList {	94✔
246	cp_inv_list,	94✔
247	str_list,
248	}
249	}	94✔
250	}
251
252	/// Custom Errors for [`CodePointInversionListAndStringList`].
UNCOV 253	#[derive(Display, Debug)]	×
254	pub enum InvalidStringList {
255	/// A string in the string list had an invalid length
UNCOV 256	#[displaydoc("Invalid string length for string: {0}")]	×
UNCOV 257	InvalidStringLength(String),	×
258	/// A string in the string list appears more than once
UNCOV 259	#[displaydoc("String list has duplicate: {0}")]	×
UNCOV 260	StringListNotUnique(String),	×
261	/// Two strings in the string list compare to each other opposite of sorted order
UNCOV 262	#[displaydoc("Strings in string list not in sorted order: ({0}, {1})")]	×
UNCOV 263	StringListNotSorted(String, String),	×
264	}
265
266	#[cfg(test)]
267	mod tests {
268	use super::*;
269
270	#[test]
271	fn test_size_has_strings() {	2✔
272	let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];	1✔
273	let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();	1✔
274	let str_slice = &["ascii_max", "bmp_max", "unicode_max", "zero"];	1✔
275	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
276
277	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();	1✔
278
279	assert!(cpilsl.has_strings());	1✔
280	assert_eq!(8, cpilsl.size());	1✔
281	}	2✔
282
283	#[test]
284	fn test_empty_string_allowed() {	2✔
285	let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];	1✔
286	let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();	1✔
287	let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];	1✔
288	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
289
290	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();	1✔
291
292	assert!(cpilsl.has_strings());	1✔
293	assert_eq!(9, cpilsl.size());	1✔
294	}	2✔
295
296	#[test]
297	fn test_invalid_string() {	2✔
298	let cp_slice = &[0, 1];	1✔
299	let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();	1✔
300	let str_slice = &["a"];	1✔
301	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
302
303	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
304
305	assert!(matches!(	1✔
306	cpilsl,	1✔
307	Err(InvalidStringList::InvalidStringLength(_))
308	));
309	}	2✔
310
311	#[test]
312	fn test_invalid_string_list_has_duplicate() {	2✔
313	let cp_slice = &[0, 1];	1✔
314	let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();	1✔
315	let str_slice = &["abc", "abc"];	1✔
316	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
317
318	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
319
320	assert!(matches!(	1✔
321	cpilsl,	1✔
322	Err(InvalidStringList::StringListNotUnique(_))
323	));
324	}	2✔
325
326	#[test]
327	fn test_invalid_string_list_not_sorted() {	2✔
328	let cp_slice = &[0, 1];	1✔
329	let cp_list = CodePointInversionList::try_from_u32_inversion_list_slice(cp_slice).unwrap();	1✔
330	let str_slice = &["xyz", "abc"];	1✔
331	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
332
333	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
334
335	assert!(matches!(	1✔
336	cpilsl,	1✔
337	Err(InvalidStringList::StringListNotSorted(_, _))
338	));
339	}	2✔
340
341	#[test]
342	fn test_from_iter_invariants() {	2✔
343	let in_strs_1 = ["a", "abc", "xyz", "abc"];	1✔
344	let in_strs_2 = ["xyz", "abc", "a", "abc"];	1✔
345
346	let cpilsl_1 = CodePointInversionListAndStringList::from_iter(in_strs_1);	1✔
347	let cpilsl_2 = CodePointInversionListAndStringList::from_iter(in_strs_2);	1✔
348
349	assert_eq!(cpilsl_1, cpilsl_2);	1✔
350
351	assert!(cpilsl_1.has_strings());	1✔
352	assert!(cpilsl_1.contains_str("abc"));	1✔
353	assert!(cpilsl_1.contains_str("xyz"));	1✔
354	assert!(!cpilsl_1.contains_str("def"));	1✔
355
356	assert_eq!(1, cpilsl_1.cp_inv_list.size());	1✔
357	assert!(cpilsl_1.contains('a'));	1✔
358	assert!(!cpilsl_1.contains('0'));	1✔
359	assert!(!cpilsl_1.contains('q'));	1✔
360
361	assert_eq!(3, cpilsl_1.size());	1✔
362	}	2✔
363	}

zbraniecki / icu4x / 11904027177

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