Simplify format_integer_with_underscore_sep

Only ever needs to handle decimal reprs

Author: yotamofek

src/librustdoc/clean/utils.rs

@@ -1,7 +1,7 @@
 use std::assert_matches::debug_assert_matches;
 use std::fmt::{self, Display, Write as _};
-use std::mem;
 use std::sync::LazyLock as Lazy;
+use std::{iter, mem};
 
 use rustc_ast::tokenstream::TokenTree;
 use rustc_hir::def::{DefKind, Res};
@@ -391,30 +391,13 @@ pub(crate) fn print_evaluated_const(
     })
 }
 
-fn format_integer_with_underscore_sep(num: &str) -> String {
-    let num_chars: Vec<_> = num.chars().collect();
-    let mut num_start_index = if num_chars.first() == Some(&'-') { 1 } else { 0 };
-    let chunk_size = match &num.as_bytes()[num_start_index..] {
-        [b'0', b'b' | b'x', ..] => {
-            num_start_index += 2;
-            4
-        }
-        [b'0', b'o', ..] => {
-            num_start_index += 2;
-            let remaining_chars = num_chars.len() - num_start_index;
-            if remaining_chars <= 6 {
-                // don't add underscores to Unix permissions like 0755 or 100755
-                return num.to_string();
-            }
-            3
-        }
-        _ => 3,
-    };
-
-    num_chars[..num_start_index]
-        .iter()
-        .chain(num_chars[num_start_index..].rchunks(chunk_size).rev().intersperse(&['_']).flatten())
-        .collect()
+fn format_integer_with_underscore_sep(num: u128, is_negative: bool) -> String {
+    let num = num.to_string();
+    iter::chain(
+        is_negative.then_some('-'),
+        num.as_bytes().rchunks(3).rev().intersperse(b"_").flatten().copied().map(char::from),
+    )
+    .collect()
 }
 
 fn print_const_with_custom_print_scalar<'tcx>(
@@ -428,7 +411,10 @@ fn print_const_with_custom_print_scalar<'tcx>(
     match (ct, ct.ty().kind()) {
         (mir::Const::Val(mir::ConstValue::Scalar(int), _), ty::Uint(ui)) => {
             let mut output = if with_underscores {
-                format_integer_with_underscore_sep(&int.to_string())
+                format_integer_with_underscore_sep(
+                    int.assert_scalar_int().to_bits_unchecked(),
+                    false,
+                )
             } else {
                 int.to_string()
             };
@@ -445,7 +431,10 @@ fn print_const_with_custom_print_scalar<'tcx>(
                 .size;
             let sign_extended_data = int.assert_scalar_int().to_int(size);
             let mut output = if with_underscores {
-                format_integer_with_underscore_sep(&sign_extended_data.to_string())
+                format_integer_with_underscore_sep(
+                    sign_extended_data.unsigned_abs(),
+                    sign_extended_data.is_negative(),
+                )
             } else {
                 sign_extended_data.to_string()
             };

src/librustdoc/clean/utils/tests.rs

@@ -2,40 +2,10 @@ use super::*;
 
 #[test]
 fn int_format_decimal() {
-    assert_eq!(format_integer_with_underscore_sep("12345678"), "12_345_678");
-    assert_eq!(format_integer_with_underscore_sep("123"), "123");
-    assert_eq!(format_integer_with_underscore_sep("123459"), "123_459");
-    assert_eq!(format_integer_with_underscore_sep("-12345678"), "-12_345_678");
-    assert_eq!(format_integer_with_underscore_sep("-123"), "-123");
-    assert_eq!(format_integer_with_underscore_sep("-123459"), "-123_459");
-}
-
-#[test]
-fn int_format_hex() {
-    assert_eq!(format_integer_with_underscore_sep("0xab3"), "0xab3");
-    assert_eq!(format_integer_with_underscore_sep("0xa2345b"), "0xa2_345b");
-    assert_eq!(format_integer_with_underscore_sep("0xa2e6345b"), "0xa2e6_345b");
-    assert_eq!(format_integer_with_underscore_sep("-0xab3"), "-0xab3");
-    assert_eq!(format_integer_with_underscore_sep("-0xa2345b"), "-0xa2_345b");
-    assert_eq!(format_integer_with_underscore_sep("-0xa2e6345b"), "-0xa2e6_345b");
-}
-
-#[test]
-fn int_format_binary() {
-    assert_eq!(format_integer_with_underscore_sep("0o12345671"), "0o12_345_671");
-    assert_eq!(format_integer_with_underscore_sep("0o123"), "0o123");
-    assert_eq!(format_integer_with_underscore_sep("0o123451"), "0o123451");
-    assert_eq!(format_integer_with_underscore_sep("-0o12345671"), "-0o12_345_671");
-    assert_eq!(format_integer_with_underscore_sep("-0o123"), "-0o123");
-    assert_eq!(format_integer_with_underscore_sep("-0o123451"), "-0o123451");
-}
-
-#[test]
-fn int_format_octal() {
-    assert_eq!(format_integer_with_underscore_sep("0b101"), "0b101");
-    assert_eq!(format_integer_with_underscore_sep("0b101101011"), "0b1_0110_1011");
-    assert_eq!(format_integer_with_underscore_sep("0b01101011"), "0b0110_1011");
-    assert_eq!(format_integer_with_underscore_sep("-0b101"), "-0b101");
-    assert_eq!(format_integer_with_underscore_sep("-0b101101011"), "-0b1_0110_1011");
-    assert_eq!(format_integer_with_underscore_sep("-0b01101011"), "-0b0110_1011");
+    assert_eq!(format_integer_with_underscore_sep(12345678, false), "12_345_678");
+    assert_eq!(format_integer_with_underscore_sep(123, false), "123");
+    assert_eq!(format_integer_with_underscore_sep(123459, false), "123_459");
+    assert_eq!(format_integer_with_underscore_sep(12345678, true), "-12_345_678");
+    assert_eq!(format_integer_with_underscore_sep(123, true), "-123");
+    assert_eq!(format_integer_with_underscore_sep(123459, true), "-123_459");
 }

src/librustdoc/lib.rs

@@ -10,6 +10,7 @@
 #![feature(format_args_nl)]
 #![feature(if_let_guard)]
 #![feature(impl_trait_in_assoc_type)]
+#![feature(iter_chain)]
 #![feature(iter_intersperse)]
 #![feature(never_type)]
 #![feature(round_char_boundary)]