feat(function): Use the "IEC 80000-13" to replace the decimal based unit

kB/KB/MB always leads to a lot of misunderstandings, we just use the KiB

common/functions/tests/it/scalars/others/humanize.rs

@@ -22,27 +22,27 @@ use crate::scalars::scalar_function_test::ScalarFunctionTest;
 fn test_humanize_size_function() -> Result<()> {
     let tests = vec![
         ScalarFunctionTest {
-            name: "humanize_size(1000)",
-            columns: vec![Series::from_data(vec![1000_u32])],
-            expect: Series::from_data(vec!["1 KB"]),
+            name: "humanize_size(1024)",
+            columns: vec![Series::from_data(vec![1024_u32])],
+            expect: Series::from_data(vec!["1.00 KiB"]),
             error: "",
         },
         ScalarFunctionTest {
-            name: "humanize_size(-1000)",
-            columns: vec![Series::from_data(vec![-1000_i32])],
-            expect: Series::from_data(vec!["-1 KB"]),
+            name: "humanize_size(-1024)",
+            columns: vec![Series::from_data(vec![-1024_i32])],
+            expect: Series::from_data(vec!["-1.00 KiB"]),
             error: "",
         },
         ScalarFunctionTest {
             name: "humanize_size('abc')",
             columns: vec![Series::from_data(vec!["abc"])],
-            expect: Series::from_data(vec!["-1 KB"]),
+            expect: Series::from_data(vec!["-1 KiB"]),
             error: "Expected a numeric type, but got String",
         },
         ScalarFunctionTest {
             name: "humanize_size(true)",
             columns: vec![Series::from_data(vec![true])],
-            expect: Series::from_data(vec!["-1 KB"]),
+            expect: Series::from_data(vec!["-1 KiB"]),
             error: "Expected a numeric type, but got Boolean",
         },
     ];
@@ -54,8 +54,8 @@ fn test_humanize_size_function() -> Result<()> {
 fn test_humanize_size_nullable() -> Result<()> {
     let tests = vec![ScalarFunctionTest {
         name: "humanize_size(null)",
-        columns: vec![Series::from_data(vec![Some(1_000_000_i32), None])],
-        expect: Series::from_data(vec![Some("1 MB"), None]),
+        columns: vec![Series::from_data(vec![Some(1_048_576_i32), None])],
+        expect: Series::from_data(vec![Some("1.00 MiB"), None]),
         error: "",
     }];
 

common/io/src/utils.rs

@@ -22,19 +22,16 @@ use common_exception::Result;
 pub fn convert_byte_size(num: f64) -> String {
     let negative = if num.is_sign_positive() { "" } else { "-" };
     let num = num.abs();
-    let units = ["B", "KB", "MB", "GB", "TB", "PB", "EB", "ZB", "YB"];
+    let units = ["B", "KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "ZiB", "YiB"];
     if num < 1_f64 {
-        return format!("{}{} {}", negative, num, "B");
+        return format!("{}{:.02} {}", negative, num, "B");
     }
-    let delimiter = 1000_f64;
+    let delimiter = 1024_f64;
     let exponent = cmp::min(
         (num.ln() / delimiter.ln()).floor() as i32,
         (units.len() - 1) as i32,
     );
-    let pretty_bytes = format!("{:.2}", num / delimiter.powi(exponent))
-        .parse::<f64>()
-        .unwrap()
-        * 1_f64;
+    let pretty_bytes = format!("{:.02}", num / delimiter.powi(exponent));
     let unit = units[exponent as usize];
     format!("{}{} {}", negative, pretty_bytes, unit)
 }

common/io/tests/it/utils.rs

@@ -16,9 +16,16 @@ use common_io::prelude::*;
 
 #[test]
 fn convert_test() {
-    assert_eq!(convert_byte_size(1_f64), "1 B");
-    assert_eq!(convert_byte_size(1022_f64), "1.02 KB");
-    assert_eq!(convert_byte_size(1022_f64 * 10000000f64), "10.22 GB");
+    assert_eq!(convert_byte_size(0_f64), "0.00 B");
+    assert_eq!(convert_byte_size(0.1_f64), "0.10 B");
+    assert_eq!(convert_byte_size(1_f64), "1.00 B");
+    assert_eq!(convert_byte_size(1023_f64), "1023.00 B");
+    assert_eq!(convert_byte_size(1024_f64), "1.00 KiB");
+    assert_eq!(convert_byte_size(1229_f64), "1.20 KiB");
+    assert_eq!(
+        convert_byte_size(1024_f64 * 1024_f64 * 1024_f64),
+        "1.00 GiB"
+    );
 
     assert_eq!(convert_number_size(1_f64), "1");
     assert_eq!(convert_number_size(1022_f64), "1.02 thousand");

docs/doc/30-reference/20-functions/120-other-functions/humanize-size.md

@@ -2,7 +2,7 @@
 title: HUMANIZE_SIZE
 ---
 
-Returns the readable size with a suffix(KB, MB, etc).
+Returns the readable size with a suffix(KiB, MiB, etc).
 
 ## Syntax
 
@@ -24,10 +24,10 @@ String.
 ## Examples
 
 ```sql
-SELECT HUMANIZE_SIZE(1000 * 1000)
+SELECT HUMANIZE_SIZE(1024 * 1024)
 +-------------------------+
-| HUMANIZE_SIZE((1000 * 1000)) |
+| HUMANIZE_SIZE((1024 * 1024)) |
 +-------------------------+
-| 1 MB                    |
+| 1 MiB                    |
 +-------------------------+
 ```

tests/suites/0_stateless/02_function/02_0055_function_strings_humanize.result

@@ -1,14 +1,15 @@
-1 KB
-1 MB
-1 GB
-1 TB
-1 PB
-1 EB
-1 ZB
-1 YB
-1000 YB
--1 KB
-0 B
+1.00 KiB
+1.20 KiB
+1.00 MiB
+1.00 GiB
+1.00 TiB
+1.00 PiB
+1.00 EiB
+1.00 ZiB
+1.00 YiB
+1024.00 YiB
+-1.00 KiB
+0.00 B
 NULL
 1
 1 thousand

tests/suites/0_stateless/02_function/02_0055_function_strings_humanize.sql

@@ -1,13 +1,14 @@
-SELECT HUMANIZE_SIZE(1000);
-SELECT HUMANIZE_SIZE(1000000);
-SELECT HUMANIZE_SIZE(1000000000);
-SELECT HUMANIZE_SIZE(1000000000000);
-SELECT HUMANIZE_SIZE(1000000000000000);
-SELECT HUMANIZE_SIZE(1000000000000000000);
-SELECT HUMANIZE_SIZE(1000000000000000000000);
-SELECT HUMANIZE_SIZE(1000000000000000000000000);
-SELECT HUMANIZE_SIZE(1000000000000000000000000000);
-SELECT HUMANIZE_SIZE(-1000);
+SELECT HUMANIZE_SIZE(1024);
+SELECT HUMANIZE_SIZE(1229);
+SELECT HUMANIZE_SIZE(POW(2, 20));
+SELECT HUMANIZE_SIZE(POW(2, 30));
+SELECT HUMANIZE_SIZE(POW(2, 40));
+SELECT HUMANIZE_SIZE(POW(2, 50));
+SELECT HUMANIZE_SIZE(POW(2, 60));
+SELECT HUMANIZE_SIZE(POW(2, 70));
+SELECT HUMANIZE_SIZE(POW(2, 80));
+SELECT HUMANIZE_SIZE(POW(2, 90));
+SELECT HUMANIZE_SIZE(-1024);
 SELECT HUMANIZE_SIZE(0);
 SELECT HUMANIZE_SIZE(NULL);
 SELECT HUMANIZE_NUMBER(1);