teju 0.1.0

Fast floating point to string conversion
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231

//! Rust implementation of [Tejú Jaguá](https://github.com/cassioneri/teju_jagua), a fast algorithm
//! for converting a floating point number to a (decimal) string.
//!
//! The interface mimics that of [Ryu](https://docs.rs/ryu/).
//!
//! ## Usage
//!
//! ```
//! let mut buffer = teju::Buffer::new();
//! let printed = buffer.format_finite(1.234);
//! assert_eq!(printed, "1.234");
//! ```
//!
//! Numbers whose decimal representation is short are written as decimals; numbers with lots of
//! zeroes are written in scientific notation. To force either way, use `format_dec` or
//! `format_exp`, respectively.
//!
//! ```
//! assert_eq!(teju::Buffer::new().format(1e3), "1000.0");
//! assert_eq!(teju::Buffer::new().format_dec(1e3), "1000.0");
//! assert_eq!(teju::Buffer::new().format_exp(1e3), "1e3");
//!
//! assert_eq!(teju::Buffer::new().format(1e30), "1e30");
//! assert_eq!(teju::Buffer::new().format_exp(1e30), "1e30");
//! assert_eq!(teju::Buffer::new().format_dec(1e30), "1000000000000000000000000000000.0");
//! ```
//!
//! ## Performance
//! 
//! ![Microbenchmark chart comparing teju with ryu and std](https://raw.githubusercontent.com/andrepd/teju-jagua-rs/master/microbench.png)

#![cfg_attr(not(test), no_std)]

use core::marker::PhantomData;

mod teju;
pub use teju::float::Float;
use teju::format::{self, Format};

/// Safe API for formatting floating point numbers to text.
///
/// ## Example
///
/// ```
/// let mut buffer = teju::Buffer::new();
/// let printed = buffer.format(1.234);
/// assert_eq!(printed, "1.234");
/// ```
#[derive(Clone, Copy)]
pub struct Buffer<F: Float, Fmt: Format> {
    float: PhantomData<F>,
    bytes: Fmt::Buffer,
}

const POS_INF: &str = "inf";
const NEG_INF: &str = "-inf";
const NAN: &str = "NaN";
const POS_ZERO: &str = "0.0";
const NEG_ZERO: &str = "-0.0";
const POS_ZERO_EXP: &str = "0e0";
const NEG_ZERO_EXP: &str = "-0e0";

impl<F: Float, Fmt: Format> Buffer<F, Fmt> {
    /// This is a cheap operation; you don't need to worry about reusing buffers for efficiency.
    pub fn new() -> Self {
        Buffer { float: PhantomData, bytes: Fmt::new_buffer() }
    }
}

impl<F: Float, Fmt: Format> Default for Buffer<F, Fmt> {
    /// This is a cheap operation; you don't need to worry about reusing buffers for efficiency.
    fn default() -> Self {
        Self::new()
    }
}

impl<F: Float> Buffer<F, format::General> {
    /// Print a floating point `num` into this buffer, and return a reference to its string
    /// representation.
    ///
    /// The number is formatted as a decimal if it fits in a "small" number of characters, or in
    /// scientific notation otherwise.
    ///
    /// This function formats NaN as the string `"NaN"`, positive infinity as `"inf"`, and negative
    /// infinity as `"-inf"`, to match [core::fmt].
    ///
    /// If `num` is known to be finite, you may get better performance by calling the
    /// [Self::format_exp_finite] method instead of format to avoid the checks for special cases.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format(3.14159), "3.14159");
    /// assert_eq!(teju::Buffer::new().format(-1. / 0.), "-inf");
    /// ```
    pub fn format(&mut self, num: F) -> &str {
        match num.classify() {
            teju::float::FloatType::Finite => self.format_finite(num),
            teju::float::FloatType::PosInf => POS_INF,
            teju::float::FloatType::NegInf => NEG_INF,
            teju::float::FloatType::Nan => NAN,
        }
    }

    /// Print a floating point `num` into this buffer, and return a reference to its string
    /// representation, **provided that `num.is_finite()`**.
    ///
    /// The number is formatted as a decimal if it fits in a "small" number of characters, or in
    /// scientific notation otherwise.
    ///
    /// This function **does not** check that `num` is indeed finite, for performance reasons; in
    /// this case it will print an unspecified (but valid) string.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format_finite(3.14159), "3.14159");
    /// ```
    pub fn format_finite(&mut self, num: F) -> &str {
        match num.classify_finite() {
            teju::float::FiniteFloatType::PosZero => return POS_ZERO,
            teju::float::FiniteFloatType::NegZero => return NEG_ZERO,
            teju::float::FiniteFloatType::Nonzero => (),
        }
        let ptr = <format::General as teju::format::Sealed>::buffer_as_ptr(&mut self.bytes);
        let n = unsafe { num.format_general_finite_nonzero(ptr) };
        let slice = unsafe { core::slice::from_raw_parts(ptr, n) };
        debug_assert!(n <= self.bytes.len());
        unsafe { core::str::from_utf8_unchecked(slice) }
    }
}

impl<F: Float> Buffer<F, format::Scientific> {
    /// Print a floating point `num` into this buffer in scientific notation, and return a
    /// reference to its string representation.
    ///
    /// The number is always formatted in the form `[mantissa]e[exponent]`, where `mantissa` is a
    /// number between 1 (inclusive) and 10 (exclusive), even if `exponent` is `0`.
    /// 
    /// This function formats NaN as the string `"NaN"`, positive infinity as `"inf"`, and negative
    /// infinity as `"-inf"`, to match [core::fmt].
    ///
    /// If `num` is known to be finite, you may get better performance by calling the
    /// [Self::format_exp_finite] method instead of format to avoid the checks for special cases.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format(137.035999177), "137.035999177");
    /// assert_eq!(teju::Buffer::new().format_exp(137.035999177), "1.37035999177e2");
    /// ```
    pub fn format_exp(&mut self, num: F) -> &str {
        match num.classify() {
            teju::float::FloatType::Finite => self.format_exp_finite(num),
            teju::float::FloatType::PosInf => POS_INF,
            teju::float::FloatType::NegInf => NEG_INF,
            teju::float::FloatType::Nan => NAN,
        }
    }

    /// Print a floating point `num` into this buffer in scientific notation, and return a
    /// reference to its string representation, provied that `num.is_finite()`.
    /// 
    /// The number is always formatted in the form `[mantissa]e[exponent]`, where `mantissa` is a
    /// number between 1 (inclusive) and 10 (exclusive), even if `exponent` is `0`.
    /// 
    /// This function **does not** check that `num` is indeed finite, for performance reasons; in
    /// this case it will print an unspecified (but valid) string.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format_finite(137.035999177), "137.035999177");
    /// assert_eq!(teju::Buffer::new().format_exp_finite(137.035999177), "1.37035999177e2");
    /// ```
    pub fn format_exp_finite(&mut self, num: F) -> &str {
        match num.classify_finite() {
            teju::float::FiniteFloatType::PosZero => return POS_ZERO_EXP,
            teju::float::FiniteFloatType::NegZero => return NEG_ZERO_EXP,
            teju::float::FiniteFloatType::Nonzero => (),
        }
        let ptr = <format::Scientific as teju::format::Sealed>::buffer_as_ptr(&mut self.bytes);
        let n = unsafe { num.format_exp_finite_nonzero(ptr) };
        let slice = unsafe { core::slice::from_raw_parts(ptr, n) };
        debug_assert!(n <= self.bytes.len());
        unsafe { core::str::from_utf8_unchecked(slice) }
    }
}

impl<F: Float> Buffer<F, format::Decimal> {
    /// Print a floating point `num` into this buffer in decimal notation, and return a reference
    /// to its string representation.
    /// 
    /// The number is always formatted as `[integral part].[fractional part]`.
    /// 
    /// This function formats NaN as the string `"NaN"`, positive infinity as `"inf"`, and negative
    /// infinity as `"-inf"`, to match [core::fmt].
    ///
    /// If `num` is known to be finite, you may get better performance by calling the
    /// [Self::format_dec_finite] method instead of format to avoid the checks for special cases.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format(1.602176634e-19), "1.602176634e-19");
    /// assert_eq!(teju::Buffer::new().format_dec(1.602176634e-19), "0.0000000000000000001602176634");
    /// ```
    pub fn format_dec(&mut self, num: F) -> &str {
        match num.classify() {
            teju::float::FloatType::Finite => self.format_dec_finite(num),
            teju::float::FloatType::PosInf => POS_INF,
            teju::float::FloatType::NegInf => NEG_INF,
            teju::float::FloatType::Nan => NAN,
        }
    }

    /// Print a floating point `num` into this buffer in decimal notation, and return a reference
    /// to its string representation, provied that `num.is_finite()`.
    /// 
    /// The number is always formatted as `[integral part].[fractional part]`.
    /// 
    /// This function **does not** check that `num` is indeed finite, for performance reasons; in
    /// this case it will print an unspecified (but valid) string.
    ///
    /// ```
    /// assert_eq!(teju::Buffer::new().format(1.602176634e-19), "1.602176634e-19");
    /// assert_eq!(teju::Buffer::new().format_dec(1.602176634e-19), "0.0000000000000000001602176634");
    /// ```
    pub fn format_dec_finite(&mut self, num: F) -> &str {
        match num.classify_finite() {
            teju::float::FiniteFloatType::PosZero => return POS_ZERO,
            teju::float::FiniteFloatType::NegZero => return NEG_ZERO,
            teju::float::FiniteFloatType::Nonzero => (),
        }
        let ptr = <format::Decimal as teju::format::Sealed>::buffer_as_ptr(&mut self.bytes);
        let n = unsafe { num.format_dec_finite_nonzero(ptr) };
        let slice = unsafe { core::slice::from_raw_parts(ptr, n) };
        debug_assert!(n <= self.bytes.len());
        unsafe { core::str::from_utf8_unchecked(slice) }
    }
}