Overhaul the thousands module.

It currently only inserts separators into `usize`s, because that's all
that has been needed so far. `-Zmacro-stats` will need `isize` and `f64`
handling, so this commit adds that.

Author: nnethercote

compiler/rustc_data_structures/src/thousands/mod.rs

@@ -1,16 +1,37 @@
-//! This is an extremely bare-bones alternative to the `thousands` crate on
-//! crates.io, for printing large numbers in a readable fashion.
+//! This is a bare-bones alternative to the `thousands` crate on crates.io, for
+//! printing large numbers in a readable fashion.
 
 #[cfg(test)]
 mod tests;
 
-// Converts the number to a string, with underscores as the thousands separator.
-pub fn format_with_underscores(n: usize) -> String {
-    let mut s = n.to_string();
-    let mut i = s.len();
-    while i > 3 {
+fn format_with_underscores(mut s: String) -> String {
+    // Ignore a leading '-'.
+    let start = if s.starts_with('-') { 1 } else { 0 };
+
+    // Stop after the first non-digit, e.g. '.' or 'e' for floats.
+    let non_digit = s[start..].find(|c: char| !c.is_digit(10));
+    let end = if let Some(non_digit) = non_digit { start + non_digit } else { s.len() };
+
+    // Insert underscores within `start..end`.
+    let mut i = end;
+    while i > start + 3 {
         i -= 3;
         s.insert(i, '_');
     }
     s
 }
+
+/// Print a `usize` with underscore separators.
+pub fn usize_with_underscores(n: usize) -> String {
+    format_with_underscores(format!("{n}"))
+}
+
+/// Print an `isize` with underscore separators.
+pub fn isize_with_underscores(n: isize) -> String {
+    format_with_underscores(format!("{n}"))
+}
+
+/// Print an `f64` with precision 1 (one decimal place) and underscore separators.
+pub fn f64p1_with_underscores(n: f64) -> String {
+    format_with_underscores(format!("{n:.1}"))
+}

compiler/rustc_data_structures/src/thousands/tests.rs

@@ -2,13 +2,51 @@ use super::*;
 
 #[test]
 fn test_format_with_underscores() {
-    assert_eq!("0", format_with_underscores(0));
-    assert_eq!("1", format_with_underscores(1));
-    assert_eq!("99", format_with_underscores(99));
-    assert_eq!("345", format_with_underscores(345));
-    assert_eq!("1_000", format_with_underscores(1_000));
-    assert_eq!("12_001", format_with_underscores(12_001));
-    assert_eq!("999_999", format_with_underscores(999_999));
-    assert_eq!("1_000_000", format_with_underscores(1_000_000));
-    assert_eq!("12_345_678", format_with_underscores(12_345_678));
+    assert_eq!("", format_with_underscores("".to_string()));
+    assert_eq!("0", format_with_underscores("0".to_string()));
+    assert_eq!("12_345.67e14", format_with_underscores("12345.67e14".to_string()));
+    assert_eq!("-1_234.5678e10", format_with_underscores("-1234.5678e10".to_string()));
+    assert_eq!("------", format_with_underscores("------".to_string()));
+    assert_eq!("abcdefgh", format_with_underscores("abcdefgh".to_string()));
+    assert_eq!("-1b", format_with_underscores("-1b".to_string()));
+    assert_eq!("-3_456xyz", format_with_underscores("-3456xyz".to_string()));
+}
+
+#[test]
+fn test_usize_with_underscores() {
+    assert_eq!("0", usize_with_underscores(0));
+    assert_eq!("1", usize_with_underscores(1));
+    assert_eq!("99", usize_with_underscores(99));
+    assert_eq!("345", usize_with_underscores(345));
+    assert_eq!("1_000", usize_with_underscores(1_000));
+    assert_eq!("12_001", usize_with_underscores(12_001));
+    assert_eq!("999_999", usize_with_underscores(999_999));
+    assert_eq!("1_000_000", usize_with_underscores(1_000_000));
+    assert_eq!("12_345_678", usize_with_underscores(12_345_678));
+}
+
+#[test]
+fn test_isize_with_underscores() {
+    assert_eq!("0", isize_with_underscores(0));
+    assert_eq!("-1", isize_with_underscores(-1));
+    assert_eq!("99", isize_with_underscores(99));
+    assert_eq!("345", isize_with_underscores(345));
+    assert_eq!("-1_000", isize_with_underscores(-1_000));
+    assert_eq!("12_001", isize_with_underscores(12_001));
+    assert_eq!("-999_999", isize_with_underscores(-999_999));
+    assert_eq!("1_000_000", isize_with_underscores(1_000_000));
+    assert_eq!("-12_345_678", isize_with_underscores(-12_345_678));
+}
+
+#[test]
+fn test_f64p1_with_underscores() {
+    assert_eq!("0.0", f64p1_with_underscores(0f64));
+    assert_eq!("0.0", f64p1_with_underscores(0.00000001));
+    assert_eq!("-0.0", f64p1_with_underscores(-0.00000001));
+    assert_eq!("1.0", f64p1_with_underscores(0.9999999));
+    assert_eq!("-1.0", f64p1_with_underscores(-0.9999999));
+    assert_eq!("345.5", f64p1_with_underscores(345.4999999));
+    assert_eq!("-100_000.0", f64p1_with_underscores(-100_000f64));
+    assert_eq!("123_456_789.1", f64p1_with_underscores(123456789.123456789));
+    assert_eq!("-123_456_789.1", f64p1_with_underscores(-123456789.123456789));
 }

compiler/rustc_metadata/src/rmeta/encoder.rs

@@ -10,7 +10,7 @@ use rustc_data_structures::fx::{FxIndexMap, FxIndexSet};
 use rustc_data_structures::memmap::{Mmap, MmapMut};
 use rustc_data_structures::sync::{join, par_for_each_in};
 use rustc_data_structures::temp_dir::MaybeTempDir;
-use rustc_data_structures::thousands::format_with_underscores;
+use rustc_data_structures::thousands::usize_with_underscores;
 use rustc_feature::Features;
 use rustc_hir as hir;
 use rustc_hir::def_id::{CRATE_DEF_ID, CRATE_DEF_INDEX, LOCAL_CRATE, LocalDefId, LocalDefIdSet};
@@ -789,7 +789,7 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
                     "{} {:<23}{:>10} ({:4.1}%)",
                     prefix,
                     label,
-                    format_with_underscores(size),
+                    usize_with_underscores(size),
                     perc(size)
                 );
             }
@@ -798,7 +798,7 @@ impl<'a, 'tcx> EncodeContext<'a, 'tcx> {
                 "{} {:<23}{:>10} (of which {:.1}% are zero bytes)",
                 prefix,
                 "Total",
-                format_with_underscores(total_bytes),
+                usize_with_underscores(total_bytes),
                 perc(zero_bytes)
             );
             eprintln!("{prefix}");

compiler/rustc_passes/src/input_stats.rs

@@ -5,7 +5,7 @@
 use rustc_ast::visit::BoundKind;
 use rustc_ast::{self as ast, NodeId, visit as ast_visit};
 use rustc_data_structures::fx::{FxHashMap, FxHashSet};
-use rustc_data_structures::thousands::format_with_underscores;
+use rustc_data_structures::thousands::usize_with_underscores;
 use rustc_hir::{self as hir, AmbigArg, HirId, intravisit as hir_visit};
 use rustc_middle::ty::TyCtxt;
 use rustc_span::Span;
@@ -140,10 +140,10 @@ impl<'k> StatCollector<'k> {
                 "{} {:<18}{:>10} ({:4.1}%){:>14}{:>14}",
                 prefix,
                 label,
-                format_with_underscores(size),
+                usize_with_underscores(size),
                 percent(size, total_size),
-                format_with_underscores(node.stats.count),
-                format_with_underscores(node.stats.size)
+                usize_with_underscores(node.stats.count),
+                usize_with_underscores(node.stats.size)
             );
             if !node.subnodes.is_empty() {
                 // We will soon sort, so the initial order does not matter.
@@ -159,9 +159,9 @@ impl<'k> StatCollector<'k> {
                         "{} - {:<18}{:>10} ({:4.1}%){:>14}",
                         prefix,
                         label,
-                        format_with_underscores(size),
+                        usize_with_underscores(size),
                         percent(size, total_size),
-                        format_with_underscores(subnode.count),
+                        usize_with_underscores(subnode.count),
                     );
                 }
             }
@@ -171,8 +171,8 @@ impl<'k> StatCollector<'k> {
             "{} {:<18}{:>10}        {:>14}",
             prefix,
             "Total",
-            format_with_underscores(total_size),
-            format_with_underscores(total_count),
+            usize_with_underscores(total_size),
+            usize_with_underscores(total_count),
         );
         eprintln!("{prefix}");
     }