feat(directory): add support for regexes in substitutions (#7145)

---------

Co-authored-by: David Knaack <davidkna@users.noreply.github.com>

.github/config-schema.json

@@ -367,7 +367,7 @@
       "default": {
         "truncation_length": 3,
         "truncate_to_repo": true,
-        "substitutions": {},
+        "substitutions": [],
         "fish_style_pwd_dir_length": 0,
         "use_logical_path": true,
         "format": "[$path]($style)[$read_only]($read_only_style) ",
@@ -2668,11 +2668,8 @@
           "default": true
         },
         "substitutions": {
-          "type": "object",
+          "$ref": "#/$defs/Either",
-          "additionalProperties": {
+          "default": []
-            "type": "string"
-          },
-          "default": {}
         },
         "fish_style_pwd_dir_length": {
           "type": "integer",
@@ -2736,6 +2733,42 @@
       },
       "additionalProperties": false
     },
+    "Either": {
+      "anyOf": [
+        {
+          "type": "array",
+          "items": {
+            "$ref": "#/$defs/SubstitutionConfig"
+          }
+        },
+        {
+          "type": "object",
+          "additionalProperties": {
+            "type": "string"
+          }
+        }
+      ]
+    },
+    "SubstitutionConfig": {
+      "type": "object",
+      "properties": {
+        "from": {
+          "type": "string"
+        },
+        "to": {
+          "type": "string"
+        },
+        "regex": {
+          "type": "boolean",
+          "default": false
+        }
+      },
+      "additionalProperties": false,
+      "required": [
+        "from",
+        "to"
+      ]
+    },
     "DirenvConfig": {
       "type": "object",
       "properties": {
@@ -5413,7 +5446,7 @@
       "type": "object",
       "properties": {
         "pixi_binary": {
-          "$ref": "#/$defs/Either",
+          "$ref": "#/$defs/Either2",
           "default": [
             "pixi"
           ]
@@ -5469,7 +5502,7 @@
       },
       "additionalProperties": false
     },
-    "Either": {
+    "Either2": {
       "anyOf": [
         {
           "type": "string"
@@ -5577,7 +5610,7 @@
           "default": "pyenv "
         },
         "python_binary": {
-          "$ref": "#/$defs/Either",
+          "$ref": "#/$defs/Either2",
           "default": [
             [
               "python"
@@ -6187,7 +6220,7 @@
           "default": "S "
         },
         "compiler": {
-          "$ref": "#/$defs/Either",
+          "$ref": "#/$defs/Either2",
           "default": [
             "solc"
           ]
@@ -6893,7 +6926,7 @@
           "default": ""
         },
         "when": {
-          "$ref": "#/$defs/Either2",
+          "$ref": "#/$defs/Either3",
           "default": false
         },
         "require_repo": {
@@ -6901,7 +6934,7 @@
           "default": false
         },
         "shell": {
-          "$ref": "#/$defs/Either",
+          "$ref": "#/$defs/Either2",
           "default": []
         },
         "description": {
@@ -6960,7 +6993,7 @@
       },
       "additionalProperties": false
     },
-    "Either2": {
+    "Either3": {
       "anyOf": [
         {
           "type": "boolean"

docs/config/README.md

@@ -1206,12 +1206,34 @@ it would have been `nixpkgs/pkgs`.
 
 | Advanced Option             | Default | Description                                                                                                                                                            |
 | --------------------------- | ------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| `substitutions`             |         | A table of substitutions to be made to the path.                                                                                                                       |
+| `substitutions`             |         | An Array or table of substitutions to be made to the path.                                                                                                             |
 | `fish_style_pwd_dir_length` | `0`     | The number of characters to use when applying fish shell pwd path logic.                                                                                               |
 | `use_logical_path`          | `true`  | If `true` render the logical path sourced from the shell via `PWD` or `--logical-path`. If `false` instead render the physical filesystem path with symlinks resolved. |
 
 `substitutions` allows you to define arbitrary replacements for literal strings that occur in the path, for example long network
-prefixes or development directories of Java. Note that this will disable the fish style PWD.
+prefixes or development directories of Java. Note that this will disable the fish style PWD. It takes an array of the following
+key/value pairs:
+
+| Value   | Type    | Description                              |
+| ------- | ------- | ---------------------------------------- |
+| `from`  | String  | The value to substitute                  |
+| `to`    | String  | The replacement for that value, if found |
+| `regex` | Boolean | (Optional) Whether `from` is a regex     |
+
+By using `regex = true`, you can use [Rust's regular expressions](https://docs.rs/regex/latest/regex/#syntax) in `from`.
+For instance you can replace every slash except the first with the following:
+
+```toml
+substitutions = [
+  { from = "^/", to = "<root>/", regex = true },
+  { from = "/", to = " | " },
+  { from = "^<root>", to = "/", regex = true },
+]
+```
+
+This will replace `/var/log` to `/ | var | log`.
+
+The old syntax still works, although it doesn't support regular expressions:
 
 ```toml
 [directory.substitutions]

src/configs/directory.rs

@@ -1,7 +1,20 @@
+use crate::config::Either;
 use indexmap::IndexMap;
-
 use serde::{Deserialize, Serialize};
 
+#[derive(Clone, Deserialize, Serialize)]
+#[cfg_attr(
+    feature = "config-schema",
+    derive(schemars::JsonSchema),
+    schemars(deny_unknown_fields)
+)]
+pub struct SubstitutionConfig<'a> {
+    pub from: String,
+    pub to: &'a str,
+    #[serde(default)]
+    pub regex: bool,
+}
+
 #[derive(Clone, Deserialize, Serialize)]
 #[cfg_attr(
     feature = "config-schema",
@@ -12,7 +25,7 @@ use serde::{Deserialize, Serialize};
 pub struct DirectoryConfig<'a> {
     pub truncation_length: i64,
     pub truncate_to_repo: bool,
-    pub substitutions: IndexMap<String, &'a str>,
+    pub substitutions: Either<Vec<SubstitutionConfig<'a>>, IndexMap<String, &'a str>>,
     pub fish_style_pwd_dir_length: i64,
     pub use_logical_path: bool,
     pub format: &'a str,
@@ -28,6 +41,15 @@ pub struct DirectoryConfig<'a> {
     pub use_os_path_sep: bool,
 }
 
+impl<'a> DirectoryConfig<'a> {
+    pub fn substitutions_empty(&self) -> bool {
+        match &self.substitutions {
+            Either::First(vec) => vec.is_empty(),
+            Either::Second(table) => table.is_empty(),
+        }
+    }
+}
+
 impl Default for DirectoryConfig<'_> {
     fn default() -> Self {
         Self {
@@ -35,7 +57,7 @@ impl Default for DirectoryConfig<'_> {
             truncate_to_repo: true,
             fish_style_pwd_dir_length: 0,
             use_logical_path: true,
-            substitutions: IndexMap::new(),
+            substitutions: Either::First(vec![]),
             format: "[$path]($style)[$read_only]($read_only_style) ",
             repo_root_format: "[$before_root_path]($before_repo_root_style)[$repo_root]($repo_root_style)[$path]($style)[$read_only]($read_only_style) ",
             style: "cyan bold",

src/modules/directory.rs

@@ -5,6 +5,7 @@ use super::utils::directory_win as directory_utils;
 use super::utils::path::PathExt as SPathExt;
 use indexmap::IndexMap;
 use path_slash::{PathBufExt, PathExt};
+use regex::{Error, Regex};
 use std::borrow::Cow;
 use std::iter::FromIterator;
 use std::path::{Path, PathBuf};
@@ -13,8 +14,8 @@ use unicode_segmentation::UnicodeSegmentation;
 use super::{Context, Module};
 
 use super::utils::directory::truncate;
-use crate::config::ModuleConfig;
+use crate::config::{Either, ModuleConfig};
-use crate::configs::directory::DirectoryConfig;
+use crate::configs::directory::{DirectoryConfig, SubstitutionConfig};
 use crate::formatter::StringFormatter;
 
 /// Creates a module with the current logical or physical directory
@@ -74,7 +75,13 @@ pub fn module<'a>(context: &'a Context) -> Option<Module<'a>> {
     let dir_string = remove_extended_path_prefix(dir_string);
 
     // Apply path substitutions
-    let dir_string = substitute_path(dir_string, &config.substitutions);
+    let dir_string = match substitute_path(dir_string.clone(), &config.substitutions) {
+        Ok(result) => result,
+        Err(err) => {
+            log::warn!("Invalid regex in directory substitutions: {err}");
+            dir_string
+        }
+    };
 
     // Truncate the dir string to the maximum number of path components
     let dir_string =
@@ -88,7 +95,7 @@ pub fn module<'a>(context: &'a Context) -> Option<Module<'a>> {
     let prefix = if is_truncated {
         // Substitutions could have changed the prefix, so don't allow them and
         // fish-style path contraction together
-        if config.fish_style_pwd_dir_length > 0 && config.substitutions.is_empty() {
+        if config.fish_style_pwd_dir_length > 0 && config.substitutions_empty() {
             // If user is using fish style path, we need to add the segment first
             let contracted_home_dir = contract_path(display_dir, &home_dir, config.home_symbol);
             to_fish_style(
@@ -291,12 +298,35 @@ fn real_path<P: AsRef<Path>>(path: P) -> PathBuf {
 ///
 /// Given a list of (from, to) pairs, this will perform the string
 /// substitutions, in order, on the path. Any non-pair of strings is ignored.
-fn substitute_path(dir_string: String, substitutions: &IndexMap<String, &str>) -> String {
+fn substitute_path(
+    dir_string: String,
+    substitutions: &Either<Vec<SubstitutionConfig>, IndexMap<String, &str>>,
+) -> Result<String, Error> {
+    let substitutions: &Vec<SubstitutionConfig> = match substitutions {
+        Either::First(vec) => vec,
+        Either::Second(table) => &table
+            .iter()
+            .map(|(from, to)| SubstitutionConfig {
+                from: String::from(from),
+                to,
+                regex: false,
+            })
+            .collect(),
+    };
+
     let mut substituted_dir = dir_string;
-    for substitution_pair in substitutions {
+
-        substituted_dir = substituted_dir.replace(substitution_pair.0, substitution_pair.1);
+    for substitution in substitutions {
+        if substitution.regex {
+            let re = Regex::new(substitution.from.as_str())?;
+            substituted_dir = re
+                .replace(substituted_dir.as_str(), substitution.to)
+                .to_string()
+        } else {
+            substituted_dir = substituted_dir.replace(substitution.from.as_str(), substitution.to)
         }
-    substituted_dir
+    }
+    Ok(substituted_dir)
 }
 
 /// Takes part before contracted path and replaces it with fish style path
@@ -350,6 +380,7 @@ fn before_root_dir<'a>(path: &'a str, repo: &'a str) -> &'a str {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use crate::config::Either::First;
     use crate::test::ModuleRenderer;
     use crate::utils::create_command;
     use crate::utils::home_dir;
@@ -443,11 +474,19 @@ mod tests {
     #[test]
     fn substitute_prefix_and_middle() {
         let full_path = "/absolute/path/foo/bar/baz";
-        let mut substitutions = IndexMap::new();
+        let substitutions = First(vec![
-        substitutions.insert("/absolute/path".to_string(), "");
+            SubstitutionConfig {
-        substitutions.insert("/bar/".to_string(), "/");
+                from: "/absolute/path".to_string(),
-
+                to: "",
-        let output = substitute_path(full_path.to_string(), &substitutions);
+                regex: false,
+            },
+            SubstitutionConfig {
+                from: "/bar/".to_string(),
+                to: "/",
+                regex: false,
+            },
+        ]);
+        let output = substitute_path(full_path.to_string(), &substitutions).unwrap();
         assert_eq!(output, "/foo/baz");
     }
 
@@ -691,6 +730,73 @@ mod tests {
         assert_eq!(expected, actual);
     }
 
+    #[test]
+    fn regex_substitution() {
+        let actual = ModuleRenderer::new("directory")
+            .path("/var/log")
+            .config(toml::toml! {
+                [directory]
+                format = "[$path]($style)"
+                substitutions = [
+                    { from = "~/Documents", to = "docs"},
+                    { from = "^/", to = "<root>/", regex = true},
+                    { from = "/", to = " | "},
+                    { from = "^<root>", to = "/", regex = true},
+                ]
+            })
+            .collect();
+        let expected = Some(format!(
+            "{}",
+            Color::Cyan.bold().paint(convert_path_sep("/ | var | log"))
+        ));
+
+        assert_eq!(expected, actual);
+
+        let actual = ModuleRenderer::new("directory")
+            .path("~/Documents/var/log")
+            .config(toml::toml! {
+                [directory]
+                format = "[$path]($style)"
+                substitutions = [
+                    { from = "~/Documents", to = "docs"},
+                    { from = "^/", to = "<root>/", regex = true},
+                    { from = "/", to = " | "},
+                    { from = "^<root>", to = "/", regex = true},
+                ]
+            })
+            .collect();
+        let expected = Some(format!(
+            "{}",
+            Color::Cyan
+                .bold()
+                .paint(convert_path_sep("docs | var | log"))
+        ));
+
+        assert_eq!(expected, actual);
+    }
+
+    #[test]
+    fn bad_regex_substitution_leaves_path_untouched() {
+        let path = "/var/log";
+        let actual = ModuleRenderer::new("directory")
+            .path(path)
+            .config(toml::toml! {
+                [directory]
+                format = "[$path]($style)"
+                substitutions = [
+                    { from = "~/Documents", to = "docs"},
+                    { from = "(^/", to = "<home>/", regex = true},
+                ]
+            })
+            .collect();
+        let expected = Some(format!(
+            "{}",
+            Color::Cyan.bold().paint(convert_path_sep(path))
+        ));
+
+        assert_eq!(expected, actual);
+    }
+
     #[test]
     fn directory_in_home() -> io::Result<()> {
         let (tmp_dir, name) = make_known_tempdir(home_dir().unwrap().as_path())?;