Refactor ansi stripping into nu-utils functions (#6966)

Allows use of slightly optimized variants that check if they have to use
the heavier vte parser. Tries to avoid unnnecessary allocations. Initial
performance characteristics proven out in #4378.

Also reduces boilerplate with right-ward drift.

Cargo.lock

@@ -2535,7 +2535,6 @@ dependencies = [
  "percent-encoding",
  "reedline",
  "rstest",
- "strip-ansi-escapes",
  "sysinfo",
  "thiserror",
 ]
@@ -2630,7 +2629,6 @@ dependencies = [
  "sha2",
  "shadow-rs",
  "sqlparser",
- "strip-ansi-escapes",
  "sysinfo",
  "terminal_size 0.2.1",
  "thiserror",
@@ -2657,7 +2655,6 @@ dependencies = [
  "nu-path",
  "nu-protocol",
  "nu-utils",
- "strip-ansi-escapes",
  "sysinfo",
 ]
 
@@ -2777,8 +2774,8 @@ dependencies = [
  "json_to_table",
  "nu-ansi-term",
  "nu-protocol",
+ "nu-utils",
  "serde_json",
- "strip-ansi-escapes",
  "tabled",
 ]
 
@@ -2786,7 +2783,7 @@ dependencies = [
 name = "nu-term-grid"
 version = "0.70.1"
 dependencies = [
- "strip-ansi-escapes",
+ "nu-utils",
  "unicode-width",
 ]
 
@@ -2811,6 +2808,7 @@ dependencies = [
  "crossterm_winapi",
  "lscolors",
  "num-format",
+ "strip-ansi-escapes",
  "sys-locale",
 ]
 

crates/nu-cli/Cargo.toml

@@ -32,7 +32,6 @@ lazy_static = "1.4.0"
 log = "0.4"
 miette = { version = "5.1.0", features = ["fancy-no-backtrace"] }
 percent-encoding = "2"
-strip-ansi-escapes = "0.1.1"
 sysinfo = "0.26.2"
 thiserror = "1.0.31"
 

crates/nu-cli/src/repl.rs

@@ -24,7 +24,6 @@ use std::{
     sync::atomic::Ordering,
     time::Instant,
 };
-use strip_ansi_escapes::strip;
 use sysinfo::SystemExt;
 
 // According to Daniel Imms @Tyriar, we need to do these this way:
@@ -140,15 +139,7 @@ pub fn evaluate_repl(
         if use_ansi {
             println!("{}", banner);
         } else {
-            let stripped_string = {
+            println!("{}", nu_utils::strip_ansi_string_likely(banner));
-                if let Ok(bytes) = strip(&banner) {
-                    String::from_utf8_lossy(&bytes).to_string()
-                } else {
-                    banner
-                }
-            };
-
-            println!("{}", stripped_string);
         }
     }
 

crates/nu-command/Cargo.toml

@@ -77,7 +77,6 @@ serde_yaml = "0.9.4"
 sha2 = "0.10.0"
 # Disable default features b/c the default features build Git (very slow to compile)
 shadow-rs = { version = "0.16.1", default-features = false }
-strip-ansi-escapes = "0.1.1"
 sysinfo = "0.26.2"
 terminal_size = "0.2.1"
 thiserror = "1.0.31"

crates/nu-command/src/core_commands/help.rs

@@ -363,15 +363,12 @@ pub fn highlight_search_string(
         }
     };
     // strip haystack to remove existing ansi style
-    let stripped_haystack: String = match strip_ansi_escapes::strip(haystack) {
+    let stripped_haystack = nu_utils::strip_ansi_likely(haystack);
-        Ok(i) => String::from_utf8(i).unwrap_or_else(|_| String::from(haystack)),
-        Err(_) => String::from(haystack),
-    };
     let mut last_match_end = 0;
     let style = Style::new().fg(White).on(Red);
     let mut highlighted = String::new();
 
-    for cap in regex.captures_iter(stripped_haystack.as_str()) {
+    for cap in regex.captures_iter(stripped_haystack.as_ref()) {
         match cap {
             Ok(capture) => {
                 let start = match capture.get(0) {

crates/nu-command/src/database/commands/into_sqlite.rs

@@ -248,14 +248,8 @@ fn nu_value_to_string(value: Value, separator: &str, config: &Config) -> String
         }
         Value::String { val, .. } => {
             // don't store ansi escape sequences in the database
-            let stripped = {
-                match strip_ansi_escapes::strip(&val) {
-                    Ok(item) => String::from_utf8(item).unwrap_or(val),
-                    Err(_) => val,
-                }
-            };
             // escape single quotes
-            stripped.replace('\'', "''")
+            nu_utils::strip_ansi_unlikely(&val).replace('\'', "''")
         }
         Value::List { vals: val, .. } => val
             .iter()

crates/nu-command/src/filesystem/cd.rs

@@ -67,10 +67,7 @@ impl Command for Cd {
         let path_val = {
             if let Some(path) = path_val {
                 Some(Spanned {
-                    item: match strip_ansi_escapes::strip(&path.item) {
+                    item: nu_utils::strip_ansi_string_unlikely(path.item),
-                        Ok(item) => String::from_utf8(item).unwrap_or(path.item),
-                        Err(_) => path.item,
-                    },
                     span: path.span,
                 })
             } else {

crates/nu-command/src/filesystem/cp.rs

@@ -73,10 +73,7 @@ impl Command for Cp {
         let src: Spanned<String> = call.req(engine_state, stack, 0)?;
         let src = {
             Spanned {
-                item: match strip_ansi_escapes::strip(&src.item) {
+                item: nu_utils::strip_ansi_string_unlikely(src.item),
-                    Ok(item) => String::from_utf8(item).unwrap_or(src.item),
-                    Err(_) => src.item,
-                },
                 span: src.span,
             }
         };

crates/nu-command/src/filesystem/ls.rs

@@ -86,10 +86,7 @@ impl Command for Ls {
         let pattern_arg = {
             if let Some(path) = pattern_arg {
                 Some(Spanned {
-                    item: match strip_ansi_escapes::strip(&path.item) {
+                    item: nu_utils::strip_ansi_string_unlikely(path.item),
-                        Ok(item) => String::from_utf8(item).unwrap_or(path.item),
-                        Err(_) => path.item,
-                    },
                     span: path.span,
                 })
             } else {

crates/nu-command/src/filesystem/mv.rs

@@ -66,10 +66,7 @@ impl Command for Mv {
         let spanned_source: Spanned<String> = call.req(engine_state, stack, 0)?;
         let spanned_source = {
             Spanned {
-                item: match strip_ansi_escapes::strip(&spanned_source.item) {
+                item: nu_utils::strip_ansi_string_unlikely(spanned_source.item),
-                    Ok(item) => String::from_utf8(item).unwrap_or(spanned_source.item),
-                    Err(_) => spanned_source.item,
-                },
                 span: spanned_source.span,
             }
         };

crates/nu-command/src/filesystem/open.rs

@@ -53,10 +53,7 @@ impl Command for Open {
         let path = {
             if let Some(path_val) = path {
                 Some(Spanned {
-                    item: match strip_ansi_escapes::strip(&path_val.item) {
+                    item: nu_utils::strip_ansi_string_unlikely(path_val.item),
-                        Ok(item) => String::from_utf8(item).unwrap_or(path_val.item),
-                        Err(_) => path_val.item,
-                    },
                     span: path_val.span,
                 })
             } else {

crates/nu-command/src/filesystem/rm.rs

@@ -143,10 +143,7 @@ fn rm(
 
     for (idx, path) in targets.clone().into_iter().enumerate() {
         let corrected_path = Spanned {
-            item: match strip_ansi_escapes::strip(&path.item) {
+            item: nu_utils::strip_ansi_string_unlikely(path.item),
-                Ok(item) => String::from_utf8(item).unwrap_or(path.item),
-                Err(_) => path.item,
-            },
             span: path.span,
         };
         let _ = std::mem::replace(&mut targets[idx], corrected_path);

crates/nu-command/src/platform/ansi/strip.rs

@@ -3,7 +3,6 @@ use nu_protocol::{
     ast::Call, ast::CellPath, engine::Command, engine::EngineState, engine::Stack, Category,
     Example, PipelineData, ShellError, Signature, Span, SyntaxShape, Value,
 };
-use strip_ansi_escapes::strip;
 
 #[derive(Clone)]
 pub struct SubCommand;
@@ -79,15 +78,7 @@ fn operate(
 fn action(input: &Value, command_span: &Span) -> Value {
     match input {
         Value::String { val, span } => {
-            let stripped_string = {
+            Value::string(nu_utils::strip_ansi_likely(val).to_string(), *span)
-                if let Ok(bytes) = strip(&val) {
-                    String::from_utf8_lossy(&bytes).to_string()
-                } else {
-                    val.to_string()
-                }
-            };
-
-            Value::string(stripped_string, *span)
         }
         other => {
             let got = format!("value is {}, not string", other.get_type());

crates/nu-command/src/viewers/griddle.rs

@@ -1,5 +1,3 @@
-use std::borrow::Cow;
-
 // use super::icons::{icon_for_file, iconify_style_ansi_to_nu};
 use super::icons::icon_for_file;
 use lscolors::Style;
@@ -178,26 +176,6 @@ prints out the list properly."#
     }
 }
 
-/// 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);
-            }
-        }
-    }
-    // Else case includes failures to parse!
-    Cow::Borrowed(string)
-}
-
 fn create_grid_output(
     items: Vec<(usize, String, String)>,
     call: &Call,
@@ -232,7 +210,7 @@ fn create_grid_output(
         if header == "name" {
             if color_param {
                 if use_grid_icons {
-                    let no_ansi = strip_ansi(&value);
+                    let no_ansi = nu_utils::strip_ansi_unlikely(&value);
                     let path = std::path::Path::new(no_ansi.as_ref());
                     let icon = icon_for_file(path, call.head)?;
                     let ls_colors_style = ls_colors.style_for_path(path);

crates/nu-command/src/viewers/table.rs

@@ -1394,17 +1394,14 @@ fn render_path_name(
         return None;
     }
 
-    let stripped_path = match strip_ansi_escapes::strip(path) {
+    let stripped_path = nu_utils::strip_ansi_unlikely(path);
-        Ok(v) => String::from_utf8(v).unwrap_or_else(|_| path.to_owned()),
-        Err(_) => path.to_owned(),
-    };
 
-    let (style, has_metadata) = match std::fs::symlink_metadata(&stripped_path) {
+    let (style, has_metadata) = match std::fs::symlink_metadata(stripped_path.as_ref()) {
         Ok(metadata) => (
-            ls_colors.style_for_path_with_metadata(&stripped_path, Some(&metadata)),
+            ls_colors.style_for_path_with_metadata(stripped_path.as_ref(), Some(&metadata)),
             true,
         ),
-        Err(_) => (ls_colors.style_for_path(&stripped_path), false),
+        Err(_) => (ls_colors.style_for_path(stripped_path.as_ref()), false),
     };
 
     // clickable links don't work in remote SSH sessions
@@ -1416,9 +1413,9 @@ fn render_path_name(
         // .map(ToNuAnsiStyle::to_nu_ansi_style)
         .unwrap_or_default();
 
-    let full_path = PathBuf::from(&stripped_path)
+    let full_path = PathBuf::from(stripped_path.as_ref())
         .canonicalize()
-        .unwrap_or_else(|_| PathBuf::from(&stripped_path));
+        .unwrap_or_else(|_| PathBuf::from(stripped_path.as_ref()));
 
     let full_path_link = make_clickable_link(
         full_path.display().to_string(),

crates/nu-engine/Cargo.toml

@@ -15,7 +15,6 @@ nu-utils = { path = "../nu-utils", version = "0.70.1"  }
 
 chrono = { version="0.4.21", features=["serde"] }
 sysinfo = "0.26.2"
-strip-ansi-escapes = "0.1.1"
 
 [features]
 plugin = []

crates/nu-engine/src/documentation.rs

@@ -170,17 +170,11 @@ fn get_documentation(
 
     long_desc.push('\n');
 
-    let stripped_string = if config.no_color {
+    if config.no_color {
-        if let Ok(bytes) = strip_ansi_escapes::strip(&long_desc) {
+        nu_utils::strip_ansi_string_likely(long_desc)
-            String::from_utf8_lossy(&bytes).to_string()
-        } else {
-            long_desc
-        }
     } else {
         long_desc
-    };
+    }
-
-    stripped_string
 }
 
 // document shape helps showing more useful information

crates/nu-table/Cargo.toml

@@ -15,7 +15,7 @@ path = "src/main.rs"
 [dependencies]
 nu-ansi-term = "0.46.0"
 nu-protocol = { path = "../nu-protocol", version = "0.70.1" }
-strip-ansi-escapes = "0.1.1"
+nu-utils = { path = "../nu-utils", version = "0.70.1" }
 atty = "0.2.14"
 tabled = { version = "0.10.0", features = ["color"], default-features = false }
 json_to_table = { version = "0.2.0", features = ["color"] }

crates/nu-table/src/table.rs

@@ -181,10 +181,7 @@ fn print_table(table: tabled::Table<Data>, config: &Config) -> String {
     // the atty is for when people do ls from vim, there should be no coloring there
     if !config.use_ansi_coloring || !atty::is(atty::Stream::Stdout) {
         // Draw the table without ansi colors
-        match strip_ansi_escapes::strip(&output) {
+        nu_utils::strip_ansi_string_likely(output)
-            Ok(bytes) => String::from_utf8_lossy(&bytes).to_string(),
-            Err(_) => output, // we did our best; so return at least something
-        }
     } else {
         // Draw the table with ansi colors
         output

crates/nu-term-grid/Cargo.toml

@@ -14,4 +14,5 @@ path = "src/main.rs"
 
 [dependencies]
 unicode-width = "0.1.9"
-strip-ansi-escapes = "0.1.1"
+
+nu-utils = { path = "../nu-utils", version = "0.70.1"  }

crates/nu-term-grid/src/grid.rs

@@ -91,35 +91,13 @@
 //! [`fit_into_width`]: ./struct.Grid.html#method.fit_into_width
 //! [`GridOptions`]: ./struct.GridOptions.html
 
-use std::borrow::Cow;
 use std::cmp::max;
 use std::fmt;
 use std::iter::repeat;
-use strip_ansi_escapes;
 use unicode_width::UnicodeWidthStr;
 
-/// 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);
-            }
-        }
-    }
-    // Else case includes failures to parse!
-    Cow::Borrowed(string)
-}
-
 fn unicode_width_strip_ansi(astring: &str) -> usize {
-    strip_ansi(astring).width()
+    nu_utils::strip_ansi_unlikely(astring).width()
 }
 
 /// Alignment indicate on which side the content should stick if some filling

crates/nu-utils/Cargo.toml

@@ -16,6 +16,7 @@ path = "src/main.rs"
 lscolors = { version = "0.12.0", features = ["crossterm"], default-features = false }
 num-format = { version = "0.4.3" }
 sys-locale = "0.2.1"
+strip-ansi-escapes = "0.1.1"
 
 [target.'cfg(windows)'.dependencies]
 crossterm_winapi = "0.9.0"

crates/nu-utils/src/deansi.rs

@@ -0,0 +1,91 @@
+use std::borrow::Cow;
+
+/// Removes ANSI escape codes and some ASCII control characters
+///
+/// Optimized for strings that rarely contain ANSI control chars.
+/// Uses fast search to avoid reallocations.
+///
+/// Keeps `\n` removes `\r`, `\t` etc.
+///
+/// If parsing fails silently returns the input string
+pub fn strip_ansi_unlikely(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);
+            }
+        }
+    }
+    // Else case includes failures to parse!
+    Cow::Borrowed(string)
+}
+
+/// Removes ANSI escape codes and some ASCII control characters
+///
+/// Optimized for strings that likely contain ANSI control chars.
+///
+/// Keeps `\n` removes `\r`, `\t` etc.
+///
+/// If parsing fails silently returns the input string
+pub fn strip_ansi_likely(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 let Ok(stripped) = strip_ansi_escapes::strip(string) {
+        if let Ok(new_string) = String::from_utf8(stripped) {
+            return Cow::Owned(new_string);
+        }
+    }
+    // Else case includes failures to parse!
+    Cow::Borrowed(string)
+}
+
+/// Removes ANSI escape codes and some ASCII control characters
+///
+/// Optimized for strings that rarely contain ANSI control chars.
+/// Uses fast search to avoid reallocations.
+///
+/// Keeps `\n` removes `\r`, `\t` etc.
+///
+/// If parsing fails silently returns the input string
+pub fn strip_ansi_string_unlikely(string: String) -> String {
+    // 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
+        .as_str()
+        .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 new_string;
+            }
+        }
+    }
+    // Else case includes failures to parse!
+    string
+}
+
+/// Removes ANSI escape codes and some ASCII control characters
+///
+/// Optimized for strings that likely contain ANSI control chars.
+///
+/// Keeps `\n` removes `\r`, `\t` etc.
+///
+/// If parsing fails silently returns the input string
+pub fn strip_ansi_string_likely(string: String) -> String {
+    // 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 let Ok(stripped) = strip_ansi_escapes::strip(&string) {
+        if let Ok(new_string) = String::from_utf8(stripped) {
+            return new_string;
+        }
+    }
+    // Else case includes failures to parse!
+    string
+}

crates/nu-utils/src/lib.rs

@@ -1,3 +1,4 @@
+mod deansi;
 pub mod locale;
 pub mod utils;
 
@@ -6,3 +7,7 @@ pub use utils::{
     enable_vt_processing, get_default_config, get_default_env, get_ls_colors,
     stderr_write_all_and_flush, stdout_write_all_and_flush,
 };
+
+pub use deansi::{
+    strip_ansi_likely, strip_ansi_string_likely, strip_ansi_string_unlikely, strip_ansi_unlikely,
+};