From 9c7feb2b1915e1855410c0e0b39796005f6bb69d Mon Sep 17 00:00:00 2001
From: Stefan Holderbach <sholderbach@users.noreply.github.com>
Date: Wed, 9 Feb 2022 00:43:32 +0100
Subject: [PATCH] Reduce table allocs: only strip ANSI if necessary (#4378)

For the width calculations for table layout the `strip_ansi` function
has to be called frequently. By checking for the ASCII control chars
(0x00 to 0x1f) except `\n` that are stripped by `strip_ansi_escapes` the number of
necessary allocations can be reduced significantly for the simple case
of text not containing ANSI escapes.

**Benchmark:**

```
nu -c "for i in 0..1000 { ls } | flatten | table"
```

**Allocation reduction**

Running on the nushell repo root as the directory, this change reduces the
allocation volume by approximately 400 MB

(Measured run via KDE heaptrack)
**Speed improvement to output**

Measured via `/usr/bin/time -v`

*before*

```
Command being timed: "./eager_nu -c for i in 0..1000 {ls} | flatten | table"
	User time (seconds): 0.87
	System time (seconds): 0.14
	Percent of CPU this job got: 87%
	Elapsed (wall clock) time (h:mm:ss or m:ss): 0:01.16
	Average shared text size (kbytes): 0
	Average unshared data size (kbytes): 0
	Average stack size (kbytes): 0
	Average total size (kbytes): 0
	Maximum resident set size (kbytes): 18888
	Average resident set size (kbytes): 0
	Major (requiring I/O) page faults: 0
	Minor (reclaiming a frame) page faults: 4809
	Voluntary context switches: 38
	Involuntary context switches: 14
	Swaps: 0
	File system inputs: 0
	File system outputs: 0
	Socket messages sent: 0
	Socket messages received: 0
	Signals delivered: 0
	Page size (bytes): 4096
	Exit status: 0
```

*after*

```
Command being timed: "./lazy_nu -c for i in 0..1000 {ls} | flatten | table"
	User time (seconds): 0.63
	System time (seconds): 0.14
	Percent of CPU this job got: 80%
	Elapsed (wall clock) time (h:mm:ss or m:ss): 0:00.97
	Average shared text size (kbytes): 0
	Average unshared data size (kbytes): 0
	Average stack size (kbytes): 0
	Average total size (kbytes): 0
	Maximum resident set size (kbytes): 18660
	Average resident set size (kbytes): 0
	Major (requiring I/O) page faults: 0
	Minor (reclaiming a frame) page faults: 5149
	Voluntary context switches: 24
	Involuntary context switches: 5
	Swaps: 0
	File system inputs: 0
	File system outputs: 0
	Socket messages sent: 0
	Socket messages received: 0
	Signals delivered: 0
	Page size (bytes): 4096
	Exit status: 0
```
---
 crates/nu-table/src/wrap.rs | 40 +++++++++++++++++++------------------
 1 file changed, 21 insertions(+), 19 deletions(-)

diff --git a/crates/nu-table/src/wrap.rs b/crates/nu-table/src/wrap.rs
index bd0e95275a..3180882b68 100644
--- a/crates/nu-table/src/wrap.rs
+++ b/crates/nu-table/src/wrap.rs
@@ -1,6 +1,7 @@
 use crate::table::TextStyle;
 use ansi_cut::AnsiCut;
 use nu_ansi_term::Style;
+use std::borrow::Cow;
 use std::collections::HashMap;
 use std::{fmt::Display, iter::Iterator};
 use unicode_width::{UnicodeWidthChar, UnicodeWidthStr};
@@ -53,24 +54,24 @@ impl Display for Line {
     }
 }
 
-fn strip_ansi(astring: &str) -> String {
-    if let Ok(bytes) = strip_ansi_escapes::strip(astring) {
-        String::from_utf8_lossy(&bytes).to_string()
-    } else {
-        astring.to_string()
-    }
-}
-
-fn unicode_width_strip_ansi(astring: &str) -> usize {
-    let stripped_string: String = {
-        if let Ok(bytes) = strip_ansi_escapes::strip(astring) {
-            String::from_utf8_lossy(&bytes).to_string()
-        } else {
-            astring.to_string()
+/// Removes ANSI escape codes and some ASCII control characters
+///
+/// Keeps `\n` removes `\r`, `\t` etc.
+///
+/// If parsing fails silently returns the input string
+fn strip_ansi(string: &str) -> Cow<str> {
+    // Check if any ascii control character except LF(0x0A = 10) is present,
+    // which will be stripped. Includes the primary start of ANSI sequences ESC
+    // (0x1B = decimal 27)
+    if string.bytes().any(|x| matches!(x, 0..=9 | 11..=31)) {
+        if let Ok(stripped) = strip_ansi_escapes::strip(string) {
+            if let Ok(new_string) = String::from_utf8(stripped) {
+                return Cow::Owned(new_string);
+            }
         }
-    };
-
-    UnicodeWidthStr::width(&stripped_string[..])
+    }
+    // Else case includes failures to parse!
+    Cow::Borrowed(string)
 }
 
 // fn special_width(astring: &str) -> usize {
@@ -108,9 +109,10 @@ pub fn split_sublines(input: &str) -> Vec<Vec<Subline>> {
                         // let c = strip_ansi(x).chars().count();
                         // let u = special_width(x);
                         // std::cmp::max(c, u)
+                        let stripped = strip_ansi(x);
 
-                        let c = strip_ansi(x).chars().count();
-                        let u = unicode_width_strip_ansi(x);
+                        let c = stripped.chars().count();
+                        let u = stripped.width();
                         std::cmp::max(c, u)
                     },
                 })