9457158389

Committed 10 Jun 2024 11:45PM UTC coverage: 75.174% (+0.05%) from 75.121%

Build # 9457158389

Build Type

push

github

Committed by

web-flow

Commit Message

Add constructing TinyAsciiStr from utf16 (#5025)

Introduces TinyAsciiStr constructors from utf16 and converges on the
consensus from #4931.

---------

Co-authored-by: Robert Bastian <4706271+robertbastian@users.noreply.github.com>

Run Details

65 of 82 new or added lines in 14 files covered. (79.27%)

3441 existing lines in 141 files now uncovered.

52850 of 70304 relevant lines covered (75.17%)

563298.06 hits per line

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

87.67

/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 `UnicodeSet`
/// (which represents a set of code points and strings) using an inversion list for the code points and a simple
/// list-like structure to store and iterate over the strings.
#[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)
        }
    }
}

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_clone_from_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("bmp_max"));
    /// assert!(cpilsl.contains(""));
    /// assert!(cpilsl.contains("A"));
    /// assert!(cpilsl.contains("ቔ"));  // U+1254 ETHIOPIC SYLLABLE QHEE
    /// assert!(!cpilsl.contains("bazinga!"));
    /// ```
    pub fn contains(&self, s: &str) -> bool {
        let mut chars = s.chars();
        if let Some(first_char) = chars.next() {
            if chars.next().is_none() {
                return self.contains_char(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_clone_from_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_clone_from_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_char('A'));
    /// assert!(cpilsl.contains_char('ቔ'));  // U+1254 ETHIOPIC SYLLABLE QHEE
    /// assert!(!cpilsl.contains_char('\u{1_0000}'));
    /// assert!(!cpilsl.contains_char('🨫'));  // U+1FA2B NEUTRAL CHESS TURNED QUEEN
    pub fn contains_char(&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 garbase otuput.
        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_clone_from_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_clone_from_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_clone_from_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_clone_from_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_clone_from_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("abc"));
        assert!(cpilsl_1.contains("xyz"));
        assert!(!cpilsl_1.contains("def"));

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

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

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	//! 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 `UnicodeSet`
23	/// (which represents a set of code points and strings) using an inversion list for the code points and a simple
24	/// list-like structure to store and iterate over the strings.
25	#[zerovec::make_varule(CodePointInversionListAndStringListULE)]	961,770✔
26	#[zerovec::skip_derive(Ord)]
27	#[zerovec::derive(Debug)]
28	#[derive(Debug, Eq, PartialEq, Clone, Yokeable, ZeroFrom)]	407✔
29	// Valid to auto-derive Deserialize because the invariants are weakly held
30	#[cfg_attr(feature = "serde", derive(serde::Deserialize, serde::Serialize))]	273✔
31	#[cfg_attr(feature = "serde", zerovec::derive(Serialize, Deserialize, Debug))]
32	pub struct CodePointInversionListAndStringList<'data> {
33	#[cfg_attr(feature = "serde", serde(borrow))]
34	#[zerovec::varule(CodePointInversionListULE)]
35	cp_inv_list: CodePointInversionList<'data>,	313✔
36	// Invariants (weakly held):
37	// - no input string is length 1 (a length 1 string should be a single code point)
38	// - the string list is sorted
39	// - the elements in the string list are unique
40	#[cfg_attr(feature = "serde", serde(borrow))]
41	str_list: VarZeroVec<'data, str>,	313✔
42	}
43
44	#[cfg(feature = "databake")]
45	impl databake::Bake for CodePointInversionListAndStringList<'_> {
UNCOV 46	fn bake(&self, env: &databake::CrateEnv) -> databake::TokenStream {	×
47	env.insert("icu_collections");	×
48	let cp_inv_list = self.cp_inv_list.bake(env);	×
49	let str_list = self.str_list.bake(env);	×
50	// Safe because our parts are safe.
UNCOV 51	databake::quote! {	×
52	icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList::from_parts_unchecked(#cp_inv_list, #str_list)
53	}
UNCOV 54	}	×
55	}
56
57	impl<'data> CodePointInversionListAndStringList<'data> {
58	/// Returns a new [`CodePointInversionListAndStringList`] from both a [`CodePointInversionList`] for the
59	/// code points and a [`VarZeroVec`]`<`[`str`]`>` of strings.
60	pub fn try_from(	765✔
61	cp_inv_list: CodePointInversionList<'data>,
62	str_list: VarZeroVec<'data, str>,
63	) -> Result<Self, InvalidStringList> {
64	// Verify invariants:
65	// Do so by using the equivalent of str_list.iter().windows(2) to get
66	// overlapping windows of size 2. The above putative code is not possible
67	// because `.windows()` exists on a slice, but VarZeroVec cannot return a slice
68	// because the non-fixed size elements necessitate at least some type
69	// of allocation.
70	{	765✔
71	let mut it = str_list.iter();	765✔
72	if let Some(mut x) = it.next() {	639✔
73	if x.len() == 1 {	66✔
74	return Err(InvalidStringList::InvalidStringLength(x.to_string()));	1✔
75	}
76	for y in it {	501✔
77	if x.len() == 1 {	436✔
UNCOV 78	return Err(InvalidStringList::InvalidStringLength(x.to_string()));	×
79	} else if x == y {	436✔
80	return Err(InvalidStringList::StringListNotUnique(x.to_string()));	1✔
81	} else if x > y {	435✔
82	return Err(InvalidStringList::StringListNotSorted(	1✔
83	x.to_string(),	1✔
84	y.to_string(),	1✔
85	));	×
86	}
87
88	// Next window begins. Update `x` here, `y` will be updated in next loop iteration.
89	x = y;	434✔
90	}
91	}
92	}	635✔
93
94	Ok(CodePointInversionListAndStringList {	636✔
95	cp_inv_list,	636✔
96	str_list,	636✔
97	})
98	}	761✔
99
100	#[doc(hidden)] // databake internal
UNCOV 101	pub const fn from_parts_unchecked(	×
102	cp_inv_list: CodePointInversionList<'data>,
103	str_list: VarZeroVec<'data, str>,
104	) -> Self {
UNCOV 105	CodePointInversionListAndStringList {	×
106	cp_inv_list,
107	str_list,
108	}
UNCOV 109	}	×
110
111	/// Returns the number of elements in this set (its cardinality).
112	/// Note than the elements of a set may include both individual
113	/// codepoints and strings.
114	pub fn size(&self) -> usize {	147✔
115	self.cp_inv_list.size() + self.str_list.len()	147✔
116	}	147✔
117
118	/// Return true if this set contains multi-code point strings or the empty string.
119	pub fn has_strings(&self) -> bool {	78✔
120	!self.str_list.is_empty()	78✔
121	}	78✔
122
123	///
124	/// # Examples
125	/// ```
126	/// use icu::collections::codepointinvlist::CodePointInversionList;
127	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
128	/// use zerovec::VarZeroVec;
129	///
130	/// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
131	/// let cp_list =
132	/// CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();
133	/// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
134	/// let str_list = VarZeroVec::<str>::from(str_slice);
135	///
136	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
137	///
138	/// assert!(cpilsl.contains("bmp_max"));
139	/// assert!(cpilsl.contains(""));
140	/// assert!(cpilsl.contains("A"));
141	/// assert!(cpilsl.contains("ቔ")); // U+1254 ETHIOPIC SYLLABLE QHEE
142	/// assert!(!cpilsl.contains("bazinga!"));
143	/// ```
144	pub fn contains(&self, s: &str) -> bool {	15,897✔
145	let mut chars = s.chars();	15,897✔
146	if let Some(first_char) = chars.next() {	15,897✔
147	if chars.next().is_none() {	7,972✔
148	return self.contains_char(first_char);	7,938✔
149	}
150	}
151	self.str_list.binary_search(s).is_ok()	7,959✔
152	}	15,897✔
153
154	///
155	/// # Examples
156	/// ```
157	/// use icu::collections::codepointinvlist::CodePointInversionList;
158	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
159	/// use zerovec::VarZeroVec;
160	///
161	/// let cp_slice = &[0, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];
162	/// let cp_list =
163	/// CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();
164	/// let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];
165	/// let str_list = VarZeroVec::<str>::from(str_slice);
166	///
167	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
168	///
169	/// assert!(cpilsl.contains32(0));
170	/// assert!(cpilsl.contains32(0x0042));
171	/// assert!(!cpilsl.contains32(0x0080));
172	/// ```
173	pub fn contains32(&self, cp: u32) -> bool {	240,279✔
174	self.cp_inv_list.contains32(cp)	240,279✔
175	}	240,279✔
176
177	///
178	/// # Examples
179	/// ```
180	/// use icu::collections::codepointinvlist::CodePointInversionList;
181	/// use icu::collections::codepointinvliststringlist::CodePointInversionListAndStringList;
182	/// use zerovec::VarZeroVec;
183	///
184	/// let cp_slice = &[0, 0x1_0000, 0x10_FFFF, 0x11_0000];
185	/// let cp_list =
186	/// CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();
187	/// let str_slice = &["", "bmp_max", "unicode_max", "zero"];
188	/// let str_list = VarZeroVec::<str>::from(str_slice);
189	///
190	/// let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();
191	///
192	/// assert!(cpilsl.contains_char('A'));
193	/// assert!(cpilsl.contains_char('ቔ')); // U+1254 ETHIOPIC SYLLABLE QHEE
194	/// assert!(!cpilsl.contains_char('\u{1_0000}'));
195	/// assert!(!cpilsl.contains_char('🨫')); // U+1FA2B NEUTRAL CHESS TURNED QUEEN
196	pub fn contains_char(&self, ch: char) -> bool {	240,274✔
197	self.contains32(ch as u32)	240,274✔
198	}	240,274✔
199
200	/// Access the underlying [`CodePointInversionList`].
201	pub fn code_points(&self) -> &CodePointInversionList<'data> {	566✔
202	&self.cp_inv_list
203	}	566✔
204
205	/// Access the contained strings.
206	pub fn strings(&self) -> &VarZeroSlice<str> {	232,516✔
207	&self.str_list	232,516✔
208	}	232,516✔
209	}
210
211	impl<'a> FromIterator<&'a str> for CodePointInversionListAndStringList<'_> {
212	fn from_iter<I>(it: I) -> Self	94✔
213	where
214	I: IntoIterator<Item = &'a str>,
215	{
216	let mut builder = CodePointInversionListBuilder::new();	94✔
217	let mut strings = Vec::<&str>::new();	94✔
218	for s in it {	4,921✔
219	let mut chars = s.chars();	4,827✔
220	if let Some(first_char) = chars.next() {	4,827✔
221	if chars.next().is_none() {	4,827✔
222	builder.add_char(first_char);	4,794✔
223	continue;
224	}
225	}
226	strings.push(s);	33✔
227	}
228
229	// Ensure that the string list is sorted. If not, the binary search that
230	// is used for `.contains(&str)` will return garbase otuput.
231	strings.sort_unstable();	94✔
232	strings.dedup();	94✔
233
234	let cp_inv_list = builder.build();	94✔
235	let str_list = VarZeroVec::<str>::from(&strings);	94✔
236
237	CodePointInversionListAndStringList {	94✔
238	cp_inv_list,	94✔
239	str_list,	94✔
240	}
241	}	94✔
242	}
243
244	/// Custom Errors for [`CodePointInversionListAndStringList`].
UNCOV 245	#[derive(Display, Debug)]	×
246	pub enum InvalidStringList {
247	/// A string in the string list had an invalid length
UNCOV 248	#[displaydoc("Invalid string length for string: {0}")]	×
UNCOV 249	InvalidStringLength(String),	×
250	/// A string in the string list appears more than once
UNCOV 251	#[displaydoc("String list has duplicate: {0}")]	×
UNCOV 252	StringListNotUnique(String),	×
253	/// Two strings in the string list compare to each other opposite of sorted order
UNCOV 254	#[displaydoc("Strings in string list not in sorted order: ({0}, {1})")]	×
UNCOV 255	StringListNotSorted(String, String),	×
256	}
257
258	#[cfg(test)]
259	mod tests {
260	use super::*;
261
262	#[test]
263	fn test_size_has_strings() {	2✔
264	let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];	2✔
265	let cp_list =
266	CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();	1✔
267	let str_slice = &["ascii_max", "bmp_max", "unicode_max", "zero"];	1✔
268	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
269
270	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();	1✔
271
272	assert!(cpilsl.has_strings());	1✔
273	assert_eq!(8, cpilsl.size());	1✔
274	}	2✔
275
276	#[test]
277	fn test_empty_string_allowed() {	2✔
278	let cp_slice = &[0, 1, 0x7F, 0x80, 0xFFFF, 0x1_0000, 0x10_FFFF, 0x11_0000];	2✔
279	let cp_list =
280	CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();	1✔
281	let str_slice = &["", "ascii_max", "bmp_max", "unicode_max", "zero"];	1✔
282	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
283
284	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list).unwrap();	1✔
285
286	assert!(cpilsl.has_strings());	1✔
287	assert_eq!(9, cpilsl.size());	1✔
288	}	2✔
289
290	#[test]
291	fn test_invalid_string() {	2✔
292	let cp_slice = &[0, 1];	2✔
293	let cp_list =
294	CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();	1✔
295	let str_slice = &["a"];	1✔
296	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
297
298	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
299
300	assert!(matches!(	1✔
301	cpilsl,	1✔
302	Err(InvalidStringList::InvalidStringLength(_))
303	));
304	}	2✔
305
306	#[test]
307	fn test_invalid_string_list_has_duplicate() {	2✔
308	let cp_slice = &[0, 1];	2✔
309	let cp_list =
310	CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();	1✔
311	let str_slice = &["abc", "abc"];	1✔
312	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
313
314	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
315
316	assert!(matches!(	1✔
317	cpilsl,	1✔
318	Err(InvalidStringList::StringListNotUnique(_))
319	));
320	}	2✔
321
322	#[test]
323	fn test_invalid_string_list_not_sorted() {	2✔
324	let cp_slice = &[0, 1];	2✔
325	let cp_list =
326	CodePointInversionList::try_clone_from_inversion_list_slice(cp_slice).unwrap();	1✔
327	let str_slice = &["xyz", "abc"];	1✔
328	let str_list = VarZeroVec::<str>::from(str_slice);	1✔
329
330	let cpilsl = CodePointInversionListAndStringList::try_from(cp_list, str_list);	1✔
331
332	assert!(matches!(	1✔
333	cpilsl,	1✔
334	Err(InvalidStringList::StringListNotSorted(_, _))
335	));
336	}	2✔
337
338	#[test]
339	fn test_from_iter_invariants() {	2✔
340	let in_strs_1 = ["a", "abc", "xyz", "abc"];	1✔
341	let in_strs_2 = ["xyz", "abc", "a", "abc"];	1✔
342
343	let cpilsl_1 = CodePointInversionListAndStringList::from_iter(in_strs_1);	1✔
344	let cpilsl_2 = CodePointInversionListAndStringList::from_iter(in_strs_2);	1✔
345
346	assert_eq!(cpilsl_1, cpilsl_2);	1✔
347
348	assert!(cpilsl_1.has_strings());	1✔
349	assert!(cpilsl_1.contains("abc"));	1✔
350	assert!(cpilsl_1.contains("xyz"));	1✔
351	assert!(!cpilsl_1.contains("def"));	1✔
352
353	assert_eq!(1, cpilsl_1.cp_inv_list.size());	1✔
354	assert!(cpilsl_1.contains_char('a'));	1✔
355	assert!(!cpilsl_1.contains_char('0'));	1✔
356	assert!(!cpilsl_1.contains_char('q'));	1✔
357
358	assert_eq!(3, cpilsl_1.size());	1✔
359	}	2✔
360	}

zbraniecki / icu4x / 9457158389

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