nu-parser + nu-protocol: switch to metric for KB, MB, GB, add KiB, MiB, GiB units (#3035)

fixes inconsistency with formatting/rendering which uses standard Rust byte_unit
https://en.wikipedia.org/wiki/Byte#Multiple-byte_units

crates/nu-command/src/commands/into_int.rs

@@ -31,11 +31,18 @@ impl WholeStreamCommand for IntoInt {
     }
 
     fn examples(&self) -> Vec<Example> {
-        vec![Example {
+        vec![
+            Example {
                 description: "Convert filesize to integer",
-            example: "into-int 1kb | each { = $it / 1024 }",
+                example: "into-int 1kb | each { = $it / 1000 }",
                 result: Some(vec![UntaggedValue::int(1).into()]),
-        }]
+            },
+            Example {
+                description: "Convert filesize to integer",
+                example: "into-int 1kib | each { = $it / 1024 }",
+                result: Some(vec![UntaggedValue::int(1).into()]),
+            },
+        ]
     }
 }
 

crates/nu-command/tests/commands/into_int.rs

@@ -5,7 +5,19 @@ fn into_int_filesize() {
     let actual = nu!(
         cwd: ".", pipeline(
         r#"
-        into-int 1kb | each {= $it / 1024 }
+        into-int 1kb | each {= $it / 1000 }
+        "#
+    ));
+
+    assert!(actual.out.contains('1'));
+}
+
+#[test]
+fn into_int_filesize2() {
+    let actual = nu!(
+        cwd: ".", pipeline(
+        r#"
+        into-int 1kib | each {= $it / 1024 }
         "#
     ));
 

crates/nu-command/tests/commands/where_.rs

@@ -7,7 +7,7 @@ use nu_test_support::pipeline;
 fn filters_by_unit_size_comparison() {
     let actual = nu!(
         cwd: "tests/fixtures/formats",
-        "ls | where size > 1kb | sort-by size | get name | first 1 | str trim"
+        "ls | where size > 1kib | sort-by size | get name | first 1 | str trim"
     );
 
     assert_eq!(actual.out, "cargo_sample.toml");

crates/nu-parser/src/parse.rs

@@ -332,6 +332,9 @@ fn parse_unit(lite_arg: &Spanned<String>) -> (SpannedExpression, Option<ParseErr
         (Unit::Gigabyte, vec!["gb", "GB", "Gb", "gB"]),
         (Unit::Terabyte, vec!["tb", "TB", "Tb", "tB"]),
         (Unit::Petabyte, vec!["pb", "PB", "Pb", "pB"]),
+        (Unit::Kibibyte, vec!["KiB", "kib", "kiB", "Kib"]),
+        (Unit::Mebibyte, vec!["MiB", "mib", "miB", "Mib"]),
+        (Unit::Gibibyte, vec!["GiB", "gib", "giB", "Gib"]),
         (Unit::Nanosecond, vec!["ns"]),
         (Unit::Microsecond, vec!["us"]),
         (Unit::Millisecond, vec!["ms"]),
@@ -2215,6 +2218,41 @@ fn unit_parse_byte_units() -> Result<(), ParseError> {
             value: 27,
             unit: Unit::Petabyte,
         },
+        TestCase {
+            string: String::from("10kib"),
+            value: 10,
+            unit: Unit::Kibibyte,
+        },
+        TestCase {
+            string: String::from("123KiB"),
+            value: 123,
+            unit: Unit::Kibibyte,
+        },
+        TestCase {
+            string: String::from("24kiB"),
+            value: 24,
+            unit: Unit::Kibibyte,
+        },
+        TestCase {
+            string: String::from("10mib"),
+            value: 10,
+            unit: Unit::Mebibyte,
+        },
+        TestCase {
+            string: String::from("123MiB"),
+            value: 123,
+            unit: Unit::Mebibyte,
+        },
+        TestCase {
+            string: String::from("10gib"),
+            value: 10,
+            unit: Unit::Gibibyte,
+        },
+        TestCase {
+            string: String::from("123GiB"),
+            value: 123,
+            unit: Unit::Gibibyte,
+        },
     ];
 
     for case in cases.iter() {

crates/nu-protocol/src/hir.rs

@@ -346,7 +346,7 @@ impl HasSpan for ExternalCommand {
 
 #[derive(Debug, Ord, PartialOrd, Eq, PartialEq, Clone, Hash, Copy, Deserialize, Serialize)]
 pub enum Unit {
-    // Filesize units
+    // Filesize units: metric
     Byte,
     Kilobyte,
     Megabyte,
@@ -354,6 +354,11 @@ pub enum Unit {
     Terabyte,
     Petabyte,
 
+    // Filesize units: ISO/IEC 80000
+    Kibibyte,
+    Mebibyte,
+    Gibibyte,
+
     // Duration units
     Nanosecond,
     Microsecond,
@@ -540,6 +545,9 @@ impl Unit {
             Unit::Gigabyte => "GB",
             Unit::Terabyte => "TB",
             Unit::Petabyte => "PB",
+            Unit::Kibibyte => "KiB",
+            Unit::Mebibyte => "MiB",
+            Unit::Gibibyte => "GiB",
             Unit::Nanosecond => "ns",
             Unit::Microsecond => "us",
             Unit::Millisecond => "ms",
@@ -558,13 +566,18 @@ impl Unit {
 
         match self {
             Unit::Byte => filesize(convert_number_to_u64(&size)),
-            Unit::Kilobyte => filesize(convert_number_to_u64(&size) * 1024),
-            Unit::Megabyte => filesize(convert_number_to_u64(&size) * 1024 * 1024),
-            Unit::Gigabyte => filesize(convert_number_to_u64(&size) * 1024 * 1024 * 1024),
-            Unit::Terabyte => filesize(convert_number_to_u64(&size) * 1024 * 1024 * 1024 * 1024),
+            Unit::Kilobyte => filesize(convert_number_to_u64(&size) * 1000),
+            Unit::Megabyte => filesize(convert_number_to_u64(&size) * 1000 * 1000),
+            Unit::Gigabyte => filesize(convert_number_to_u64(&size) * 1000 * 1000 * 1000),
+            Unit::Terabyte => filesize(convert_number_to_u64(&size) * 1000 * 1000 * 1000 * 1000),
             Unit::Petabyte => {
-                filesize(convert_number_to_u64(&size) * 1024 * 1024 * 1024 * 1024 * 1024)
+                filesize(convert_number_to_u64(&size) * 1000 * 1000 * 1000 * 1000 * 1000)
             }
+
+            Unit::Kibibyte => filesize(convert_number_to_u64(&size) * 1024),
+            Unit::Mebibyte => filesize(convert_number_to_u64(&size) * 1024 * 1024),
+            Unit::Gibibyte => filesize(convert_number_to_u64(&size) * 1024 * 1024 * 1024),
+
             Unit::Nanosecond => duration(size.to_bigint().expect("Conversion should never fail.")),
             Unit::Microsecond => {
                 duration(size.to_bigint().expect("Conversion should never fail.") * 1000)

crates/nu-protocol/src/value/value_structure.rs

@@ -3,7 +3,10 @@ use nu_errors::ShellError;
 use std::path::{Component, Path, PathBuf};
 
 fn is_value_tagged_dir(value: &Value) -> bool {
-    matches!(&value.value, UntaggedValue::Row(_) | UntaggedValue::Table(_))
+    matches!(
+        &value.value,
+        UntaggedValue::Row(_) | UntaggedValue::Table(_)
+    )
 }
 
 #[derive(Debug, Eq, Ord, PartialEq, PartialOrd)]

tests/shell/pipeline/commands/internal.rs

@@ -759,7 +759,9 @@ fn filesize_math() {
         "#
     );
 
-    assert_eq!(actual.out, "1.0 MB");
+    assert_eq!(actual.out, "1000.0 KB");
+    // why 1000.0 KB instead of 1.0 MB?
+    // looks like `byte.get_appropriate_unit(false)` behaves this way
 }
 
 #[test]
@@ -783,7 +785,7 @@ fn filesize_math3() {
         "#
     );
 
-    assert_eq!(actual.out, "10.2 KB");
+    assert_eq!(actual.out, "10.0 KB");
 }
 #[test]
 fn filesize_math4() {
@@ -794,7 +796,43 @@ fn filesize_math4() {
         "#
     );
 
-    assert_eq!(actual.out, "512.0 KB");
+    assert_eq!(actual.out, "500.0 KB");
+}
+
+#[test]
+fn filesize_math5() {
+    let actual = nu!(
+        cwd: ".",
+        r#"
+        = 1001 * 1kb
+        "#
+    );
+
+    assert_eq!(actual.out, "1.0 MB");
+}
+
+#[test]
+fn filesize_math6() {
+    let actual = nu!(
+        cwd: ".",
+        r#"
+        = 1001 * 1mb
+        "#
+    );
+
+    assert_eq!(actual.out, "1.0 GB");
+}
+
+#[test]
+fn filesize_math7() {
+    let actual = nu!(
+        cwd: ".",
+        r#"
+        = 1001 * 1gb
+        "#
+    );
+
+    assert_eq!(actual.out, "1.0 TB");
 }
 
 #[test]