From 5b2a12ca6dd9573a0dd4770f07e304616266c648 Mon Sep 17 00:00:00 2001
From: Nagy Tibor <xnagytibor@protonmail.com>
Date: Wed, 27 Jul 2022 21:23:08 +0200
Subject: [PATCH] strtof, strtod: Fix parsing floats with exponents
---
src/macros.rs | 64 +++++++++--
tests/expected/stdlib/strtod.stdout | 162 ++++++++++++++++++++++++++++
tests/stdlib/strtod.c | 52 ++++++++-
3 files changed, 270 insertions(+), 8 deletions(-)
diff --git a/src/macros.rs b/src/macros.rs
index 6f9db24f..88383d65 100644
--- a/src/macros.rs
+++ b/src/macros.rs
@@ -233,6 +233,8 @@ macro_rules! strto_float_impl {
let mut s = $s;
let endptr = $endptr;
+ // TODO: Handle named floats: NaN, Inf...
+
while ctype::isspace(*s as c_int) != 0 {
s = s.offset(1);
}
@@ -240,16 +242,16 @@ macro_rules! strto_float_impl {
let mut result: $type = 0.0;
let mut radix = 10;
- let negative = match *s as u8 {
+ let result_sign = match *s as u8 {
b'-' => {
s = s.offset(1);
- true
+ -1.0
}
b'+' => {
s = s.offset(1);
- false
+ 1.0
}
- _ => false,
+ _ => 1.0,
};
if *s as u8 == b'0' && *s.offset(1) as u8 == b'x' {
@@ -274,6 +276,54 @@ macro_rules! strto_float_impl {
}
}
+ let s_before_exponent = s;
+
+ let exponent = match (*s as u8, radix) {
+ (b'e' | b'E', 10) | (b'p' | b'P', 16) => {
+ s = s.offset(1);
+
+ let is_exponent_positive = match *s as u8 {
+ b'-' => {
+ s = s.offset(1);
+ false
+ }
+ b'+' => {
+ s = s.offset(1);
+ true
+ }
+ _ => true,
+ };
+
+ // Exponent digits are always in base 10.
+ if (*s as u8 as char).is_digit(10) {
+ let mut exponent_value = 0;
+
+ while let Some(digit) = (*s as u8 as char).to_digit(10) {
+ exponent_value *= 10;
+ exponent_value += digit;
+ s = s.offset(1);
+ }
+
+ let exponent_base = match radix {
+ 10 => 10u128,
+ 16 => 2u128,
+ _ => unreachable!(),
+ };
+
+ if is_exponent_positive {
+ Some(exponent_base.pow(exponent_value) as $type)
+ } else {
+ Some(1.0 / (exponent_base.pow(exponent_value) as $type))
+ }
+ } else {
+ // Exponent had no valid digits after 'e'/'p' and '+'/'-', rollback
+ s = s_before_exponent;
+ None
+ }
+ }
+ _ => None,
+ };
+
if !endptr.is_null() {
// This is stupid, but apparently strto* functions want
// const input but mut output, yet the man page says
@@ -282,10 +332,10 @@ macro_rules! strto_float_impl {
*endptr = s as *mut _;
}
- if negative {
- -result
+ if let Some(exponent) = exponent {
+ result_sign * result * exponent
} else {
- result
+ result_sign * result
}
}};
}
diff --git a/tests/expected/stdlib/strtod.stdout b/tests/expected/stdlib/strtod.stdout
index 68d889ec..5ed8e2ec 100644
--- a/tests/expected/stdlib/strtod.stdout
+++ b/tests/expected/stdlib/strtod.stdout
@@ -7,3 +7,165 @@ d: -5.300000 Endptr: ""
d: 16.071045 Endptr: ""
d: 1.136719 Endptr: ""
d: 3.128906 Endptr: ""
+d: 100000.000000 Endptr: ""
+d: 100000.000000 Endptr: ""
+d: 0.000010 Endptr: ""
+d: 100000.000000 Endptr: " "
+d: 100000.000000 Endptr: " "
+d: 0.000010 Endptr: " "
+d: 10000000000.000000 Endptr: ""
+d: 1.000000 Endptr: "eXXXX"
+d: 1.000000 Endptr: "e"
+d: 1.000000 Endptr: "e "
+d: 10000000000.000000 Endptr: ""
+d: 1.000000 Endptr: "e+XXXX"
+d: 1.000000 Endptr: "e+"
+d: 1.000000 Endptr: "e+ "
+d: 0.000000 Endptr: ""
+d: 1.000000 Endptr: "e-XXXX"
+d: 1.000000 Endptr: "e-"
+d: 1.000000 Endptr: "e- "
+d: -100000.000000 Endptr: ""
+d: -100000.000000 Endptr: ""
+d: -0.000010 Endptr: ""
+d: -100000.000000 Endptr: " "
+d: -100000.000000 Endptr: " "
+d: -0.000010 Endptr: " "
+d: -10000000000.000000 Endptr: ""
+d: -1.000000 Endptr: "eXXXX"
+d: -1.000000 Endptr: "e"
+d: -1.000000 Endptr: "e "
+d: -10000000000.000000 Endptr: ""
+d: -1.000000 Endptr: "e+XXXX"
+d: -1.000000 Endptr: "e+"
+d: -1.000000 Endptr: "e+ "
+d: -0.000000 Endptr: ""
+d: -1.000000 Endptr: "e-XXXX"
+d: -1.000000 Endptr: "e-"
+d: -1.000000 Endptr: "e- "
+d: 1234000.000000 Endptr: ""
+d: 1234000.000000 Endptr: ""
+d: 0.000123 Endptr: ""
+d: 1234000.000000 Endptr: " "
+d: 1234000.000000 Endptr: " "
+d: 0.000123 Endptr: " "
+d: 123400000000.000000 Endptr: ""
+d: 12.340000 Endptr: "eXXXX"
+d: 12.340000 Endptr: "e"
+d: 12.340000 Endptr: "e "
+d: 123400000000.000000 Endptr: ""
+d: 12.340000 Endptr: "e+XXXX"
+d: 12.340000 Endptr: "e+"
+d: 12.340000 Endptr: "e+ "
+d: 0.000000 Endptr: ""
+d: 12.340000 Endptr: "e-XXXX"
+d: 12.340000 Endptr: "e-"
+d: 12.340000 Endptr: "e- "
+d: -1234000.000000 Endptr: ""
+d: -1234000.000000 Endptr: ""
+d: -0.000123 Endptr: ""
+d: -1234000.000000 Endptr: " "
+d: -1234000.000000 Endptr: " "
+d: -0.000123 Endptr: " "
+d: -123400000000.000000 Endptr: ""
+d: -12.340000 Endptr: "eXXXX"
+d: -12.340000 Endptr: "e"
+d: -12.340000 Endptr: "e "
+d: -123400000000.000000 Endptr: ""
+d: -12.340000 Endptr: "e+XXXX"
+d: -12.340000 Endptr: "e+"
+d: -12.340000 Endptr: "e+ "
+d: -0.000000 Endptr: ""
+d: -12.340000 Endptr: "e-XXXX"
+d: -12.340000 Endptr: "e-"
+d: -12.340000 Endptr: "e- "
+d: 192.000000 Endptr: ""
+d: -192.000000 Endptr: ""
+d: 0.005859 Endptr: ""
+d: -0.005859 Endptr: ""
+d: 10.000000 Endptr: ""
+d: 0.156250 Endptr: ""
+d: -10.000000 Endptr: ""
+d: -0.156250 Endptr: ""
+d: 16.062500 Endptr: ""
+d: 16.062500 Endptr: ""
+d: -16.062500 Endptr: ""
+d: -16.062500 Endptr: ""
+d: 0.500000 Endptr: ""
+d: 5.000000 Endptr: ""
+d: 50.000000 Endptr: ""
+d: 500.000000 Endptr: ""
+d: 5000.000000 Endptr: ""
+d: 50000.000000 Endptr: ""
+d: 500000.000000 Endptr: ""
+d: 5000000.000000 Endptr: ""
+d: 50000000.000000 Endptr: ""
+d: 500000000.000000 Endptr: ""
+d: 5000000000.000000 Endptr: ""
+d: 50000000000.000000 Endptr: ""
+d: 500000000000.000000 Endptr: ""
+d: 5000000000000.000000 Endptr: ""
+d: 50000000000000.000000 Endptr: ""
+d: 500000000000000.000000 Endptr: ""
+d: 5000000000000000.000000 Endptr: ""
+d: 50000000000000000.000000 Endptr: ""
+d: 500000000000000000.000000 Endptr: ""
+d: 5000000000000000000.000000 Endptr: ""
+d: 50000000000000000000.000000 Endptr: ""
+d: 500000000000000000000.000000 Endptr: ""
+d: 5000000000000000000000.000000 Endptr: ""
+d: 49999999999999995805696.000000 Endptr: ""
+d: 499999999999999991611392.000000 Endptr: ""
+d: 5000000000000000452984832.000000 Endptr: ""
+d: 50000000000000002382364672.000000 Endptr: ""
+d: 500000000000000006643777536.000000 Endptr: ""
+d: 4999999999999999791559868416.000000 Endptr: ""
+d: 49999999999999995716575428608.000000 Endptr: ""
+d: 500000000000000009942312419328.000000 Endptr: ""
+d: 4999999999999999817948147482624.000000 Endptr: ""
+d: 50000000000000002683081102196736.000000 Endptr: ""
+d: 499999999999999972787615493521408.000000 Endptr: ""
+d: 4999999999999999727876154935214080.000000 Endptr: ""
+d: 49999999999999998431683053958987776.000000 Endptr: ""
+d: 500000000000000021210318687008980992.000000 Endptr: ""
+d: 4999999999999999769381329101060571136.000000 Endptr: ""
+d: 49999999999999998874404911728017014784.000000 Endptr: ""
+d: -0.500000 Endptr: ""
+d: -5.000000 Endptr: ""
+d: -50.000000 Endptr: ""
+d: -500.000000 Endptr: ""
+d: -5000.000000 Endptr: ""
+d: -50000.000000 Endptr: ""
+d: -500000.000000 Endptr: ""
+d: -5000000.000000 Endptr: ""
+d: -50000000.000000 Endptr: ""
+d: -500000000.000000 Endptr: ""
+d: -5000000000.000000 Endptr: ""
+d: -50000000000.000000 Endptr: ""
+d: -500000000000.000000 Endptr: ""
+d: -5000000000000.000000 Endptr: ""
+d: -50000000000000.000000 Endptr: ""
+d: -500000000000000.000000 Endptr: ""
+d: -5000000000000000.000000 Endptr: ""
+d: -50000000000000000.000000 Endptr: ""
+d: -500000000000000000.000000 Endptr: ""
+d: -5000000000000000000.000000 Endptr: ""
+d: -50000000000000000000.000000 Endptr: ""
+d: -500000000000000000000.000000 Endptr: ""
+d: -5000000000000000000000.000000 Endptr: ""
+d: -49999999999999995805696.000000 Endptr: ""
+d: -499999999999999991611392.000000 Endptr: ""
+d: -5000000000000000452984832.000000 Endptr: ""
+d: -50000000000000002382364672.000000 Endptr: ""
+d: -500000000000000006643777536.000000 Endptr: ""
+d: -4999999999999999791559868416.000000 Endptr: ""
+d: -49999999999999995716575428608.000000 Endptr: ""
+d: -500000000000000009942312419328.000000 Endptr: ""
+d: -4999999999999999817948147482624.000000 Endptr: ""
+d: -50000000000000002683081102196736.000000 Endptr: ""
+d: -499999999999999972787615493521408.000000 Endptr: ""
+d: -4999999999999999727876154935214080.000000 Endptr: ""
+d: -49999999999999998431683053958987776.000000 Endptr: ""
+d: -500000000000000021210318687008980992.000000 Endptr: ""
+d: -4999999999999999769381329101060571136.000000 Endptr: ""
+d: -49999999999999998874404911728017014784.000000 Endptr: ""
diff --git a/tests/stdlib/strtod.c b/tests/stdlib/strtod.c
index d6826f67..fa1013ed 100644
--- a/tests/stdlib/strtod.c
+++ b/tests/stdlib/strtod.c
@@ -10,7 +10,57 @@ int main(void) {
char* inputs[] = {
"a 1 hello", " 1 hello", "1 hello 2",
"10.123", "010.123", "-5.3",
- "0x10.123", "0x1.23", "0x3.21"
+ "0x10.123", "0x1.23", "0x3.21",
+
+ "1e5", "1e+5", "1e-5",
+ "1e5 ", "1e+5 ", "1e-5 ",
+
+ "1e10", "1eXXXX", "1e", "1e ",
+ "1e+10", "1e+XXXX", "1e+", "1e+ ",
+ "1e-10", "1e-XXXX", "1e-", "1e- ",
+
+ "-1e5", "-1e+5", "-1e-5",
+ "-1e5 ", "-1e+5 ", "-1e-5 ",
+
+ "-1e10", "-1eXXXX", "-1e", "-1e ",
+ "-1e+10", "-1e+XXXX", "-1e+", "-1e+ ",
+ "-1e-10", "-1e-XXXX", "-1e-", "-1e- ",
+
+ "12.34e5", "12.34e+5", "12.34e-5",
+ "12.34e5 ", "12.34e+5 ", "12.34e-5 ",
+
+ "12.34e10", "12.34eXXXX", "12.34e", "12.34e ",
+ "12.34e+10", "12.34e+XXXX", "12.34e+", "12.34e+ ",
+ "12.34e-10", "12.34e-XXXX", "12.34e-", "12.34e- ",
+
+ "-12.34e5", "-12.34e+5", "-12.34e-5",
+ "-12.34e5 ", "-12.34e+5 ", "-12.34e-5 ",
+
+ "-12.34e10", "-12.34eXXXX", "-12.34e", "-12.34e ",
+ "-12.34e+10", "-12.34e+XXXX", "-12.34e+", "-12.34e+ ",
+ "-12.34e-10", "-12.34e-XXXX", "-12.34e-", "-12.34e- ",
+
+ "0x0.3p10", "-0x0.3p10", "0x0.3p-5", "-0x0.3p-5",
+ "0x1.4p3", "0x1.4p-3", "-0x1.4p3", "-0x1.4p-3",
+ "0x10.1p0", "0x10.1p-0", "-0x10.1p0", "-0x10.1p-0",
+
+ "0.5e0", "0.5e1", "0.5e2", "0.5e3", "0.5e4",
+ "0.5e5", "0.5e6", "0.5e7", "0.5e8", "0.5e9",
+ "0.5e10", "0.5e11", "0.5e12", "0.5e13", "0.5e14",
+ "0.5e15", "0.5e16", "0.5e17", "0.5e18", "0.5e19",
+ "0.5e20", "0.5e21", "0.5e22", "0.5e23", "0.5e24",
+ "0.5e25", "0.5e26", "0.5e27", "0.5e28", "0.5e29",
+ "0.5e30", "0.5e31", "0.5e32", "0.5e33", "0.5e34",
+ "0.5e35", "0.5e36", "0.5e37", "0.5e38",
+
+ "-0.5e0", "-0.5e1", "-0.5e2", "-0.5e3", "-0.5e4",
+ "-0.5e5", "-0.5e6", "-0.5e7", "-0.5e8", "-0.5e9",
+ "-0.5e10", "-0.5e11", "-0.5e12", "-0.5e13", "-0.5e14",
+ "-0.5e15", "-0.5e16", "-0.5e17", "-0.5e18", "-0.5e19",
+ "-0.5e20", "-0.5e21", "-0.5e22", "-0.5e23", "-0.5e24",
+ "-0.5e25", "-0.5e26", "-0.5e27", "-0.5e28", "-0.5e29",
+ "-0.5e30", "-0.5e31", "-0.5e32", "-0.5e33", "-0.5e34",
+ "-0.5e35", "-0.5e36", "-0.5e37", "-0.5e38",
};
for (int i = 0; i < sizeof(inputs) / sizeof(char*); i += 1) {
d = strtod(inputs[i], &endptr);
--
GitLab