From e76b485de3cb85978d5e5300f004a65b4f7ff968 Mon Sep 17 00:00:00 2001
From: Dave Waggoner <waggoner.dave@gmail.com>
Date: Fri, 21 Nov 2025 11:01:06 -0800
Subject: [PATCH] terminal: New settings for path hyperlink regexes (#40305)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Closes:
- #12338
- #40202

1. Adds two new settings which allow customizing the set of regexes used
to identify path hyperlinks in terminal
1. Fixes path hyperlinks for paths containing unicode emoji and
punctuation, for example, `mojo.🔥`
1. Fixes path hyperlinks for Windows verbatim paths, for example,
`\\?\C:\Over\here.rs`.
1. Improves path hyperlink performance, especially for terminals with a
lot of content
1. Replaces existing custom hard-coded default path hyperlink parsing
logic with a set of customizable default regexes

## New settings

(from default.json)

### terminal.path_hyperlink_regexes

Regexes used to identify paths for hyperlink navigation. Supports
optional named capture
groups `path`, `line`, `column`, and `link`. If none of these are
present, the entire match
is the hyperlink target. If `path` is present, it is the hyperlink
target, along with `line`
and `column` if present. `link` may be used to customize what text in
terminal is part of the
hyperlink. If `link` is not present, the text of the entire match is
used. If `line` and
`column` are not present, the default built-in line and column suffix
processing is used
which parses `line:column` and `(line,column)` variants. The default
value handles Python
diagnostics and common path, line, column syntaxes. This can be extended
or replaced to
handle specific scenarios. For example, to enable support for
hyperlinking paths which
contain spaces in rust output,
```
[
  "\\s+(-->|:::|at) (?<link>(?<path>.+?))(:$|$)",
  "\\s+(Compiling|Checking|Documenting) [^(]+\\((?<link>(?<path>.+))\\)"
],
```
could be used. Processing stops at the first regex with a match, even if
no link is
produced which is the case when the cursor is not over the hyperlinked
text. For best
performance it is recommended to order regexes from most common to least
common. For
readability and documentation, each regex may be an array of strings
which are collected
into one multi-line regex string for use in terminal path hyperlink
detection.

### terminal.path_hyperlink_timeout_ms
Timeout for hover and Cmd-click path hyperlink discovery in
milliseconds. Specifying a
timeout of `0` will disable path hyperlinking in terminal.

## Performance

This PR fixes terminal to only search the hovered line for hyperlinks
and adds a benchmark. Before this fix, hyperlink detection grows
linearly with terminal content, with this fix it is proportional only to
the hovered line. The gains come from replacing
`visible_regex_match_iter`, which searched all visible lines, with code
that only searches the line hovered on (including if the line is
wrapped).

Local benchmark timings (terminal with 500 lines of content):

||main|this PR|Δ|
|-|-|-:|-|
| cargo_hyperlink_benchmark | 1.4 ms | 13 µs | -99.0% |
| rust_hyperlink_benchmark | 1.2 ms | 11 µs | -99.1% |
| ls_hyperlink_benchmark | 1.3 ms | 7 µs |  -99.5% |

Release Notes:

- terminal: New settings to allow customizing the set of regexes used to
identify path hyperlinks in terminal
- terminal: Fixed terminal path hyperlinks for paths containing unicode
punctuation and emoji, e.g. mojo.🔥
- terminal: Fixed path hyperlinks for Windows verbatim paths, for
example, `\\?\C:\Over\here.rs`
- terminal: Improved terminal hyperlink performance, especially for
terminals with a lot of content visible
---
 Cargo.lock                                    |   4 +-
 assets/settings/default.json                  |  54 +-
 crates/project/src/terminals.rs               |   4 +
 .../settings/src/settings_content/terminal.rs |  42 +
 crates/settings/src/vscode_import.rs          |   2 +
 crates/terminal/Cargo.toml                    |   7 +-
 crates/terminal/src/terminal.rs               |  23 +-
 crates/terminal/src/terminal_hyperlinks.rs    | 854 ++++++++++++------
 crates/terminal/src/terminal_settings.rs      |  16 +-
 docs/src/configuring-zed.md                   |  47 +
 10 files changed, 768 insertions(+), 285 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 9917862e72..bfce6ab287 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -17072,16 +17072,17 @@ dependencies = [
  "alacritty_terminal",
  "anyhow",
  "collections",
+ "fancy-regex 0.14.0",
  "futures 0.3.31",
  "gpui",
  "itertools 0.14.0",
  "libc",
  "log",
  "rand 0.9.2",
- "regex",
  "release_channel",
  "schemars",
  "serde",
+ "serde_json",
  "settings",
  "smol",
  "sysinfo 0.37.2",
@@ -17091,6 +17092,7 @@ dependencies = [
  "url",
  "urlencoding",
  "util",
+ "util_macros",
  "windows 0.61.3",
 ]
 
diff --git a/assets/settings/default.json b/assets/settings/default.json
index 9a6146f75b..9b289bdf08 100644
--- a/assets/settings/default.json
+++ b/assets/settings/default.json
@@ -1585,7 +1585,59 @@
     //
     // Most terminal themes have APCA values of 40-70.
     // A value of 45 preserves colorful themes while ensuring legibility.
-    "minimum_contrast": 45
+    "minimum_contrast": 45,
+    // Regexes used to identify paths for hyperlink navigation. Supports optional named capture
+    // groups `path`, `line`, `column`, and `link`. If none of these are present, the entire match
+    // is the hyperlink target. If `path` is present, it is the hyperlink target, along with `line`
+    // and `column` if present. `link` may be used to customize what text in terminal is part of the
+    // hyperlink. If `link` is not present, the text of the entire match is used. If `line` and
+    // `column` are not present, the default built-in line and column suffix processing is used
+    // which parses `line:column` and `(line,column)` variants. The default value handles Python
+    // diagnostics and common path, line, column syntaxes. This can be extended or replaced to
+    // handle specific scenarios. For example, to enable support for hyperlinking paths which
+    // contain spaces in rust output,
+    //
+    // [
+    //   "\\s+(-->|:::|at) (?<link>(?<path>.+?))(:$|$)",
+    //   "\\s+(Compiling|Checking|Documenting) [^(]+\\((?<link>(?<path>.+))\\)"
+    // ],
+    //
+    // could be used. Processing stops at the first regex with a match, even if no link is
+    // produced which is the case when the cursor is not over the hyperlinked text. For best
+    // performance it is recommended to order regexes from most common to least common. For
+    // readability and documentation, each regex may be an array of strings which are collected
+    // into one multi-line regex string for use in terminal path hyperlink detection.
+    "path_hyperlink_regexes": [
+      // Python-style diagnostics
+      "File \"(?<path>[^\"]+)\", line (?<line>[0-9]+)",
+      // Common path syntax with optional line, column, description, trailing punctuation, or
+      // surrounding symbols or quotes
+      [
+        "(?x)",
+        "# optionally starts with 0-2 opening prefix symbols",
+        "[({\\[<]{0,2}",
+        "# which may be followed by an opening quote",
+        "(?<quote>[\"'`])?",
+        "# `path` is the shortest sequence of any non-space character",
+        "(?<link>(?<path>[^ ]+?",
+        "    # which may end with a line and optionally a column,",
+        "    (?<line_column>:+[0-9]+(:[0-9]+)?|:?\\([0-9]+([,:][0-9]+)?\\))?",
+        "))",
+        "# which must be followed by a matching quote",
+        "(?(<quote>)\\k<quote>)",
+        "# and optionally a single closing symbol",
+        "[)}\\]>]?",
+        "# if line/column matched, may be followed by a description",
+        "(?(<line_column>):[^ 0-9][^ ]*)?",
+        "# which may be followed by trailing punctuation",
+        "[.,:)}\\]>]*",
+        "# and always includes trailing whitespace or end of line",
+        "([ ]+|$)"
+      ]
+    ],
+    // Timeout for hover and Cmd-click path hyperlink discovery in milliseconds. Specifying a
+    // timeout of `0` will disable path hyperlinking in terminal.
+    "path_hyperlink_timeout_ms": 1
   },
   "code_actions_on_format": {},
   // Settings related to running tasks.
diff --git a/crates/project/src/terminals.rs b/crates/project/src/terminals.rs
index ef21c97f81..81172f5774 100644
--- a/crates/project/src/terminals.rs
+++ b/crates/project/src/terminals.rs
@@ -240,6 +240,8 @@ impl Project {
                         settings.cursor_shape,
                         settings.alternate_scroll,
                         settings.max_scroll_history_lines,
+                        settings.path_hyperlink_regexes,
+                        settings.path_hyperlink_timeout_ms,
                         is_via_remote,
                         cx.entity_id().as_u64(),
                         Some(completion_tx),
@@ -369,6 +371,8 @@ impl Project {
                         settings.cursor_shape,
                         settings.alternate_scroll,
                         settings.max_scroll_history_lines,
+                        settings.path_hyperlink_regexes,
+                        settings.path_hyperlink_timeout_ms,
                         is_via_remote,
                         cx.entity_id().as_u64(),
                         None,
diff --git a/crates/settings/src/settings_content/terminal.rs b/crates/settings/src/settings_content/terminal.rs
index c75b986bb8..c54ebe2d1c 100644
--- a/crates/settings/src/settings_content/terminal.rs
+++ b/crates/settings/src/settings_content/terminal.rs
@@ -29,6 +29,41 @@ pub struct ProjectTerminalSettingsContent {
     ///
     /// Default: on
     pub detect_venv: Option<VenvSettings>,
+    /// Regexes used to identify paths for hyperlink navigation.
+    ///
+    /// Default: [
+    ///   // Python-style diagnostics
+    ///   "File \"(?<path>[^\"]+)\", line (?<line>[0-9]+)",
+    ///   // Common path syntax with optional line, column, description, trailing punctuation, or
+    ///   // surrounding symbols or quotes
+    ///   [
+    ///     "(?x)",
+    ///     "# optionally starts with 0-2 opening prefix symbols",
+    ///     "[({\\[<]{0,2}",
+    ///     "# which may be followed by an opening quote",
+    ///     "(?<quote>[\"'`])?",
+    ///     "# `path` is the shortest sequence of any non-space character",
+    ///     "(?<link>(?<path>[^ ]+?",
+    ///     "    # which may end with a line and optionally a column,",
+    ///     "    (?<line_column>:+[0-9]+(:[0-9]+)?|:?\\([0-9]+([,:][0-9]+)?\\))?",
+    ///     "))",
+    ///     "# which must be followed by a matching quote",
+    ///     "(?(<quote>)\\k<quote>)",
+    ///     "# and optionally a single closing symbol",
+    ///     "[)}\\]>]?",
+    ///     "# if line/column matched, may be followed by a description",
+    ///     "(?(<line_column>):[^ 0-9][^ ]*)?",
+    ///     "# which may be followed by trailing punctuation",
+    ///     "[.,:)}\\]>]*",
+    ///     "# and always includes trailing whitespace or end of line",
+    ///     "([ ]+|$)"
+    ///   ]
+    /// ]
+    pub path_hyperlink_regexes: Option<Vec<PathHyperlinkRegex>>,
+    /// Timeout for hover and Cmd-click path hyperlink discovery in milliseconds.
+    ///
+    /// Default: 1
+    pub path_hyperlink_timeout_ms: Option<u64>,
 }
 
 #[with_fallible_options]
@@ -412,6 +447,13 @@ impl VenvSettings {
     }
 }
 
+#[derive(Clone, Debug, Serialize, Deserialize, PartialEq, JsonSchema, MergeFrom)]
+#[serde(untagged)]
+pub enum PathHyperlinkRegex {
+    SingleLine(String),
+    MultiLine(Vec<String>),
+}
+
 #[derive(
     Copy,
     Clone,
diff --git a/crates/settings/src/vscode_import.rs b/crates/settings/src/vscode_import.rs
index f5df817dcd..4d893011d4 100644
--- a/crates/settings/src/vscode_import.rs
+++ b/crates/settings/src/vscode_import.rs
@@ -772,6 +772,8 @@ impl VsCodeSettings {
             working_directory: None,
             env,
             detect_venv: None,
+            path_hyperlink_regexes: None,
+            path_hyperlink_timeout_ms: None,
         }
     }
 
diff --git a/crates/terminal/Cargo.toml b/crates/terminal/Cargo.toml
index 0dc7338e04..dac9db190d 100644
--- a/crates/terminal/Cargo.toml
+++ b/crates/terminal/Cargo.toml
@@ -25,6 +25,7 @@ anyhow.workspace = true
 collections.workspace = true
 futures.workspace = true
 gpui.workspace = true
+itertools.workspace = true
 libc.workspace = true
 log.workspace = true
 release_channel.workspace = true
@@ -37,9 +38,8 @@ task.workspace = true
 theme.workspace = true
 thiserror.workspace = true
 util.workspace = true
-regex.workspace = true
+fancy-regex.workspace = true
 urlencoding.workspace = true
-itertools.workspace = true
 
 [target.'cfg(windows)'.dependencies]
 windows.workspace = true
@@ -47,4 +47,7 @@ windows.workspace = true
 [dev-dependencies]
 gpui = { workspace = true, features = ["test-support"] }
 rand.workspace = true
+serde_json.workspace = true
+settings = { workspace = true, features = ["test-support"] }
 url.workspace = true
+util_macros.workspace = true
diff --git a/crates/terminal/src/terminal.rs b/crates/terminal/src/terminal.rs
index 3c71a7f0e1..69b6be5f24 100644
--- a/crates/terminal/src/terminal.rs
+++ b/crates/terminal/src/terminal.rs
@@ -374,7 +374,7 @@ impl TerminalBuilder {
             scroll_px: px(0.),
             next_link_id: 0,
             selection_phase: SelectionPhase::Ended,
-            hyperlink_regex_searches: RegexSearches::new(),
+            hyperlink_regex_searches: RegexSearches::default(),
             vi_mode_enabled: false,
             is_remote_terminal: false,
             last_mouse_move_time: Instant::now(),
@@ -388,6 +388,8 @@ impl TerminalBuilder {
                 cursor_shape,
                 alternate_scroll,
                 max_scroll_history_lines,
+                path_hyperlink_regexes: Vec::default(),
+                path_hyperlink_timeout_ms: 0,
                 window_id,
             },
             child_exited: None,
@@ -408,6 +410,8 @@ impl TerminalBuilder {
         cursor_shape: CursorShape,
         alternate_scroll: AlternateScroll,
         max_scroll_history_lines: Option<usize>,
+        path_hyperlink_regexes: Vec<String>,
+        path_hyperlink_timeout_ms: u64,
         is_remote_terminal: bool,
         window_id: u64,
         completion_tx: Option<Sender<Option<ExitStatus>>>,
@@ -592,7 +596,10 @@ impl TerminalBuilder {
                 scroll_px: px(0.),
                 next_link_id: 0,
                 selection_phase: SelectionPhase::Ended,
-                hyperlink_regex_searches: RegexSearches::new(),
+                hyperlink_regex_searches: RegexSearches::new(
+                    &path_hyperlink_regexes,
+                    path_hyperlink_timeout_ms,
+                ),
                 vi_mode_enabled: false,
                 is_remote_terminal,
                 last_mouse_move_time: Instant::now(),
@@ -606,6 +613,8 @@ impl TerminalBuilder {
                     cursor_shape,
                     alternate_scroll,
                     max_scroll_history_lines,
+                    path_hyperlink_regexes,
+                    path_hyperlink_timeout_ms,
                     window_id,
                 },
                 child_exited: None,
@@ -838,6 +847,8 @@ struct CopyTemplate {
     cursor_shape: CursorShape,
     alternate_scroll: AlternateScroll,
     max_scroll_history_lines: Option<usize>,
+    path_hyperlink_regexes: Vec<String>,
+    path_hyperlink_timeout_ms: u64,
     window_id: u64,
 }
 
@@ -2163,6 +2174,8 @@ impl Terminal {
             self.template.cursor_shape,
             self.template.alternate_scroll,
             self.template.max_scroll_history_lines,
+            self.template.path_hyperlink_regexes.clone(),
+            self.template.path_hyperlink_timeout_ms,
             self.is_remote_terminal,
             self.template.window_id,
             None,
@@ -2404,6 +2417,8 @@ mod tests {
                     CursorShape::default(),
                     AlternateScroll::On,
                     None,
+                    vec![],
+                    0,
                     false,
                     0,
                     Some(completion_tx),
@@ -2452,6 +2467,8 @@ mod tests {
                     CursorShape::default(),
                     AlternateScroll::On,
                     None,
+                    vec![],
+                    0,
                     false,
                     0,
                     Some(completion_tx),
@@ -2527,6 +2544,8 @@ mod tests {
                     CursorShape::default(),
                     AlternateScroll::On,
                     None,
+                    Vec::new(),
+                    0,
                     false,
                     0,
                     Some(completion_tx),
diff --git a/crates/terminal/src/terminal_hyperlinks.rs b/crates/terminal/src/terminal_hyperlinks.rs
index d3b5067420..94e8d1716f 100644
--- a/crates/terminal/src/terminal_hyperlinks.rs
+++ b/crates/terminal/src/terminal_hyperlinks.rs
@@ -2,45 +2,64 @@ use alacritty_terminal::{
     Term,
     event::EventListener,
     grid::Dimensions,
-    index::{Boundary, Column, Direction as AlacDirection, Line, Point as AlacPoint},
-    term::search::{Match, RegexIter, RegexSearch},
+    index::{Boundary, Column, Direction as AlacDirection, Point as AlacPoint},
+    term::{
+        cell::Flags,
+        search::{Match, RegexIter, RegexSearch},
+    },
+};
+use fancy_regex::Regex;
+use log::{info, warn};
+use std::{
+    ops::{Index, Range},
+    time::{Duration, Instant},
 };
-use regex::Regex;
-use std::{ops::Index, sync::LazyLock};
 
 const URL_REGEX: &str = r#"(ipfs:|ipns:|magnet:|mailto:|gemini://|gopher://|https://|http://|news:|file://|git://|ssh:|ftp://)[^\u{0000}-\u{001F}\u{007F}-\u{009F}<>"\s{-}\^⟨⟩`']+"#;
-// Optional suffix matches MSBuild diagnostic suffixes for path parsing in PathLikeWithPosition
-// https://learn.microsoft.com/en-us/visualstudio/msbuild/msbuild-diagnostic-format-for-tasks
-const WORD_REGEX: &str =
-    r#"[\$\+\w.\[\]:/\\@\-~()]+(?:\((?:\d+|\d+,\d+)\))|[\$\+\w.\[\]:/\\@\-~()]+"#;
-
-const PYTHON_FILE_LINE_REGEX: &str = r#"File "(?P<file>[^"]+)", line (?P<line>\d+)"#;
-
-static PYTHON_FILE_LINE_MATCHER: LazyLock<Regex> =
-    LazyLock::new(|| Regex::new(PYTHON_FILE_LINE_REGEX).unwrap());
-
-fn python_extract_path_and_line(input: &str) -> Option<(&str, u32)> {
-    if let Some(captures) = PYTHON_FILE_LINE_MATCHER.captures(input) {
-        let path_part = captures.name("file")?.as_str();
-
-        let line_number: u32 = captures.name("line")?.as_str().parse().ok()?;
-        return Some((path_part, line_number));
-    }
-    None
-}
+const WIDE_CHAR_SPACERS: Flags =
+    Flags::from_bits(Flags::LEADING_WIDE_CHAR_SPACER.bits() | Flags::WIDE_CHAR_SPACER.bits())
+        .unwrap();
 
 pub(super) struct RegexSearches {
     url_regex: RegexSearch,
-    word_regex: RegexSearch,
-    python_file_line_regex: RegexSearch,
+    path_hyperlink_regexes: Vec<Regex>,
+    path_hyperlink_timeout: Duration,
 }
 
-impl RegexSearches {
-    pub(super) fn new() -> Self {
+impl Default for RegexSearches {
+    fn default() -> Self {
         Self {
             url_regex: RegexSearch::new(URL_REGEX).unwrap(),
-            word_regex: RegexSearch::new(WORD_REGEX).unwrap(),
-            python_file_line_regex: RegexSearch::new(PYTHON_FILE_LINE_REGEX).unwrap(),
+            path_hyperlink_regexes: Vec::default(),
+            path_hyperlink_timeout: Duration::default(),
+        }
+    }
+}
+impl RegexSearches {
+    pub(super) fn new(
+        path_hyperlink_regexes: impl IntoIterator<Item: AsRef<str>>,
+        path_hyperlink_timeout_ms: u64,
+    ) -> Self {
+        Self {
+            url_regex: RegexSearch::new(URL_REGEX).unwrap(),
+            path_hyperlink_regexes: path_hyperlink_regexes
+                .into_iter()
+                .filter_map(|regex| {
+                    Regex::new(regex.as_ref())
+                        .inspect_err(|error| {
+                            warn!(
+                                concat!(
+                                    "Ignoring path hyperlink regex specified in ",
+                                    "`terminal.path_hyperlink_regexes`:\n\n\t{}\n\nError: {}",
+                                ),
+                                regex.as_ref(),
+                                error
+                            );
+                        })
+                        .ok()
+                })
+                .collect(),
+            path_hyperlink_timeout: Duration::from_millis(path_hyperlink_timeout_ms),
         }
     }
 }
@@ -77,76 +96,32 @@ pub(super) fn find_from_grid_point<T: EventListener>(
         let url_match = min_index..=max_index;
 
         Some((url, true, url_match))
-    } else if let Some(url_match) = regex_match_at(term, point, &mut regex_searches.url_regex) {
-        let url = term.bounds_to_string(*url_match.start(), *url_match.end());
-        let (sanitized_url, sanitized_match) = sanitize_url_punctuation(url, url_match, term);
-        Some((sanitized_url, true, sanitized_match))
-    } else if let Some(python_match) =
-        regex_match_at(term, point, &mut regex_searches.python_file_line_regex)
-    {
-        let matching_line = term.bounds_to_string(*python_match.start(), *python_match.end());
-        python_extract_path_and_line(&matching_line).map(|(file_path, line_number)| {
-            (format!("{file_path}:{line_number}"), false, python_match)
-        })
-    } else if let Some(word_match) = regex_match_at(term, point, &mut regex_searches.word_regex) {
-        let file_path = term.bounds_to_string(*word_match.start(), *word_match.end());
-
-        let (sanitized_match, sanitized_word) = 'sanitize: {
-            let mut word_match = word_match;
-            let mut file_path = file_path;
-
-            if is_path_surrounded_by_common_symbols(&file_path) {
-                word_match = Match::new(
-                    word_match.start().add(term, Boundary::Grid, 1),
-                    word_match.end().sub(term, Boundary::Grid, 1),
-                );
-                file_path = file_path[1..file_path.len() - 1].to_owned();
-            }
-
-            while file_path.ends_with(':') {
-                file_path.pop();
-                word_match = Match::new(
-                    *word_match.start(),
-                    word_match.end().sub(term, Boundary::Grid, 1),
-                );
-            }
-            let mut colon_count = 0;
-            for c in file_path.chars() {
-                if c == ':' {
-                    colon_count += 1;
-                }
-            }
-            // strip trailing comment after colon in case of
-            // file/at/path.rs:row:column:description or error message
-            // so that the file path is `file/at/path.rs:row:column`
-            if colon_count > 2 {
-                let last_index = file_path.rfind(':').unwrap();
-                let prev_is_digit = last_index > 0
-                    && file_path
-                        .chars()
-                        .nth(last_index - 1)
-                        .is_some_and(|c| c.is_ascii_digit());
-                let next_is_digit = last_index < file_path.len() - 1
-                    && file_path
-                        .chars()
-                        .nth(last_index + 1)
-                        .is_none_or(|c| c.is_ascii_digit());
-                if prev_is_digit && !next_is_digit {
-                    let stripped_len = file_path.len() - last_index;
-                    word_match = Match::new(
-                        *word_match.start(),
-                        word_match.end().sub(term, Boundary::Grid, stripped_len),
-                    );
-                    file_path = file_path[0..last_index].to_owned();
-                }
-            }
-
-            break 'sanitize (word_match, file_path);
-        };
-
-        Some((sanitized_word, false, sanitized_match))
     } else {
-        None
+        let (line_start, line_end) = (term.line_search_left(point), term.line_search_right(point));
+        if let Some((url, url_match)) = RegexIter::new(
+            line_start,
+            line_end,
+            AlacDirection::Right,
+            term,
+            &mut regex_searches.url_regex,
+        )
+        .find(|rm| rm.contains(&point))
+        .map(|url_match| {
+            let url = term.bounds_to_string(*url_match.start(), *url_match.end());
+            sanitize_url_punctuation(url, url_match, term)
+        }) {
+            Some((url, true, url_match))
+        } else {
+            path_match(
+                &term,
+                line_start,
+                line_end,
+                point,
+                &mut regex_searches.path_hyperlink_regexes,
+                regex_searches.path_hyperlink_timeout,
+            )
+            .map(|(path, path_match)| (path, false, path_match))
+        }
     };
 
     found_word.map(|(maybe_url_or_path, is_url, word_match)| {
@@ -222,58 +197,171 @@ fn sanitize_url_punctuation<T: EventListener>(
     }
 }
 
-fn is_path_surrounded_by_common_symbols(path: &str) -> bool {
-    // Avoid detecting `[]` or `()` strings as paths, surrounded by common symbols
-    path.len() > 2
-        // The rest of the brackets and various quotes cannot be matched by the [`WORD_REGEX`] hence not checked for.
-        && (path.starts_with('[') && path.ends_with(']')
-            || path.starts_with('(') && path.ends_with(')'))
-}
+fn path_match<T>(
+    term: &Term<T>,
+    line_start: AlacPoint,
+    line_end: AlacPoint,
+    hovered: AlacPoint,
+    path_hyperlink_regexes: &mut Vec<Regex>,
+    path_hyperlink_timeout: Duration,
+) -> Option<(String, Match)> {
+    if path_hyperlink_regexes.is_empty() || path_hyperlink_timeout.as_millis() == 0 {
+        return None;
+    }
 
-/// Based on alacritty/src/display/hint.rs > regex_match_at
-/// Retrieve the match, if the specified point is inside the content matching the regex.
-fn regex_match_at<T>(term: &Term<T>, point: AlacPoint, regex: &mut RegexSearch) -> Option<Match> {
-    visible_regex_match_iter(term, regex).find(|rm| rm.contains(&point))
-}
+    let search_start_time = Instant::now();
 
-/// Copied from alacritty/src/display/hint.rs:
-/// Iterate over all visible regex matches.
-fn visible_regex_match_iter<'a, T>(
-    term: &'a Term<T>,
-    regex: &'a mut RegexSearch,
-) -> impl Iterator<Item = Match> + 'a {
-    const MAX_SEARCH_LINES: usize = 100;
+    let timed_out = || {
+        let elapsed_time = Instant::now().saturating_duration_since(search_start_time);
+        (elapsed_time > path_hyperlink_timeout)
+            .then_some((elapsed_time.as_millis(), path_hyperlink_timeout.as_millis()))
+    };
 
-    let viewport_start = Line(-(term.grid().display_offset() as i32));
-    let viewport_end = viewport_start + term.bottommost_line();
-    let mut start = term.line_search_left(AlacPoint::new(viewport_start, Column(0)));
-    let mut end = term.line_search_right(AlacPoint::new(viewport_end, Column(0)));
-    start.line = start.line.max(viewport_start - MAX_SEARCH_LINES);
-    end.line = end.line.min(viewport_end + MAX_SEARCH_LINES);
+    // This used to be: `let line = term.bounds_to_string(line_start, line_end)`, however, that
+    // api compresses tab characters into a single space, whereas we require a cell accurate
+    // string representation of the line. The below algorithm does this, but seems a bit odd.
+    // Maybe there is a clean api for doing this, but I couldn't find it.
+    let mut line = String::with_capacity(
+        (line_end.line.0 - line_start.line.0 + 1) as usize * term.grid().columns(),
+    );
+    line.push(term.grid()[line_start].c);
+    for cell in term.grid().iter_from(line_start) {
+        if cell.point > line_end {
+            break;
+        }
 
-    RegexIter::new(start, end, AlacDirection::Right, term, regex)
-        .skip_while(move |rm| rm.end().line < viewport_start)
-        .take_while(move |rm| rm.start().line <= viewport_end)
+        if !cell.flags.intersects(WIDE_CHAR_SPACERS) {
+            line.push(match cell.c {
+                '\t' => ' ',
+                c @ _ => c,
+            });
+        }
+    }
+    let line = line.trim_ascii_end();
+
+    let found_from_range = |path_range: Range<usize>,
+                            link_range: Range<usize>,
+                            position: Option<(u32, Option<u32>)>| {
+        let advance_point_by_str = |mut point: AlacPoint, s: &str| {
+            for _ in s.chars() {
+                point = term
+                    .expand_wide(point, AlacDirection::Right)
+                    .add(term, Boundary::Grid, 1);
+            }
+
+            // There does not appear to be an alacritty api that is
+            // "move to start of current wide char", so we have to do it ourselves.
+            let flags = term.grid().index(point).flags;
+            if flags.contains(Flags::LEADING_WIDE_CHAR_SPACER) {
+                AlacPoint::new(point.line + 1, Column(0))
+            } else if flags.contains(Flags::WIDE_CHAR_SPACER) {
+                AlacPoint::new(point.line, point.column - 1)
+            } else {
+                point
+            }
+        };
+
+        let link_start = advance_point_by_str(line_start, &line[..link_range.start]);
+        let link_end = advance_point_by_str(link_start, &line[link_range]);
+        let link_match = link_start
+            ..=term
+                .expand_wide(link_end, AlacDirection::Left)
+                .sub(term, Boundary::Grid, 1);
+
+        Some((
+            {
+                let mut path = line[path_range].to_string();
+                position.inspect(|(line, column)| {
+                    path += &format!(":{line}");
+                    column.inspect(|column| path += &format!(":{column}"));
+                });
+                path
+            },
+            link_match,
+        ))
+    };
+
+    for regex in path_hyperlink_regexes {
+        let mut path_found = false;
+
+        for captures in regex.captures_iter(&line) {
+            let captures = match captures {
+                Ok(captures) => captures,
+                Err(error) => {
+                    warn!("Error '{error}' searching for path hyperlinks in line: {line}");
+                    info!(
+                        "Skipping match from path hyperlinks with regex: {}",
+                        regex.as_str()
+                    );
+                    continue;
+                }
+            };
+
+            let match_range = captures.get(0).unwrap().range();
+            let (path_range, line_column) = if let Some(path) = captures.name("path") {
+                let parse = |name: &str| {
+                    captures
+                        .name(name)
+                        .and_then(|capture| capture.as_str().parse().ok())
+                };
+
+                (
+                    path.range(),
+                    parse("line").map(|line| (line, parse("column"))),
+                )
+            } else {
+                (match_range.clone(), None)
+            };
+            let link_range = captures
+                .name("link")
+                .map_or(match_range, |link| link.range());
+            let found = found_from_range(path_range, link_range, line_column);
+
+            if let Some(found) = found {
+                path_found = true;
+                if found.1.contains(&hovered) {
+                    return Some(found);
+                }
+            }
+        }
+
+        if path_found {
+            return None;
+        }
+
+        if let Some((timed_out_ms, timeout_ms)) = timed_out() {
+            warn!("Timed out processing path hyperlink regexes after {timed_out_ms}ms");
+            info!("{timeout_ms}ms time out specified in `terminal.path_hyperlink_timeout_ms`");
+            return None;
+        }
+    }
+
+    None
 }
 
 #[cfg(test)]
 mod tests {
+    use crate::terminal_settings::TerminalSettings;
+
     use super::*;
     use alacritty_terminal::{
         event::VoidListener,
-        index::{Boundary, Point as AlacPoint},
+        grid::Dimensions,
+        index::{Boundary, Column, Line, Point as AlacPoint},
         term::{Config, cell::Flags, test::TermSize},
         vte::ansi::Handler,
     };
-    use std::{cell::RefCell, ops::RangeInclusive, path::PathBuf};
+    use fancy_regex::Regex;
+    use settings::{self, Settings, SettingsContent};
+    use std::{cell::RefCell, ops::RangeInclusive, path::PathBuf, rc::Rc};
     use url::Url;
     use util::paths::PathWithPosition;
 
     fn re_test(re: &str, hay: &str, expected: Vec<&str>) {
-        let results: Vec<_> = regex::Regex::new(re)
+        let results: Vec<_> = Regex::new(re)
             .unwrap()
             .find_iter(hay)
-            .map(|m| m.as_str())
+            .map(|m| m.unwrap().as_str())
             .collect();
         assert_eq!(results, expected);
     }
@@ -376,78 +464,6 @@ mod tests {
         }
     }
 
-    #[test]
-    fn test_word_regex() {
-        re_test(
-            WORD_REGEX,
-            "hello, world! \"What\" is this?",
-            vec!["hello", "world", "What", "is", "this"],
-        );
-    }
-
-    #[test]
-    fn test_word_regex_with_linenum() {
-        // filename(line) and filename(line,col) as used in MSBuild output
-        // should be considered a single "word", even though comma is
-        // usually a word separator
-        re_test(WORD_REGEX, "a Main.cs(20) b", vec!["a", "Main.cs(20)", "b"]);
-        re_test(
-            WORD_REGEX,
-            "Main.cs(20,5) Error desc",
-            vec!["Main.cs(20,5)", "Error", "desc"],
-        );
-        // filename:line:col is a popular format for unix tools
-        re_test(
-            WORD_REGEX,
-            "a Main.cs:20:5 b",
-            vec!["a", "Main.cs:20:5", "b"],
-        );
-        // Some tools output "filename:line:col:message", which currently isn't
-        // handled correctly, but might be in the future
-        re_test(
-            WORD_REGEX,
-            "Main.cs:20:5:Error desc",
-            vec!["Main.cs:20:5:Error", "desc"],
-        );
-    }
-
-    #[test]
-    fn test_python_file_line_regex() {
-        re_test(
-            PYTHON_FILE_LINE_REGEX,
-            "hay File \"/zed/bad_py.py\", line 8 stack",
-            vec!["File \"/zed/bad_py.py\", line 8"],
-        );
-        re_test(PYTHON_FILE_LINE_REGEX, "unrelated", vec![]);
-    }
-
-    #[test]
-    fn test_python_file_line() {
-        let inputs: Vec<(&str, Option<(&str, u32)>)> = vec![
-            (
-                "File \"/zed/bad_py.py\", line 8",
-                Some(("/zed/bad_py.py", 8u32)),
-            ),
-            ("File \"path/to/zed/bad_py.py\"", None),
-            ("unrelated", None),
-            ("", None),
-        ];
-        let actual = inputs
-            .iter()
-            .map(|input| python_extract_path_and_line(input.0))
-            .collect::<Vec<_>>();
-        let expected = inputs.iter().map(|(_, output)| *output).collect::<Vec<_>>();
-        assert_eq!(actual, expected);
-    }
-
-    // We use custom columns in many tests to workaround this issue by ensuring a wrapped
-    // line never ends on a wide char:
-    //
-    // <https://github.com/alacritty/alacritty/issues/8586>
-    //
-    // This issue was recently fixed, as soon as we update to a version containing the fix we
-    // can remove all the custom columns from these tests.
-    //
     macro_rules! test_hyperlink {
         ($($lines:expr),+; $hyperlink_kind:ident) => { {
             use crate::terminal_hyperlinks::tests::line_cells_count;
@@ -458,21 +474,28 @@ mod tests {
                 test_lines.iter().copied()
                     .map(line_cells_count)
                     .fold((0, 0), |state, cells| (state.0 + cells, cmp::max(state.1, cells)));
-
-            test_hyperlink!(
+            let contains_tab_char = test_lines.iter().copied()
+                .map(str::chars).flatten().find(|&c| c == '\t');
+            let columns = if contains_tab_char.is_some() {
+                // This avoids tabs at end of lines causing whitespace-eating line wraps...
+                vec![longest_line_cells + 1]
+            } else {
                 // Alacritty has issues with 2 columns, use 3 as the minimum for now.
-                [3, longest_line_cells / 2, longest_line_cells + 1];
+                vec![3, longest_line_cells / 2, longest_line_cells + 1]
+            };
+            test_hyperlink!(
+                columns;
                 total_cells;
                 test_lines.iter().copied();
                 $hyperlink_kind
             )
         } };
 
-        ([ $($columns:expr),+ ]; $total_cells:expr; $lines:expr; $hyperlink_kind:ident) => { {
+        ($columns:expr; $total_cells:expr; $lines:expr; $hyperlink_kind:ident) => { {
             use crate::terminal_hyperlinks::tests::{ test_hyperlink, HyperlinkKind };
 
             let source_location = format!("{}:{}", std::file!(), std::line!());
-            for columns in vec![ $($columns),+] {
+            for columns in $columns {
                 test_hyperlink(columns, $total_cells, $lines, HyperlinkKind::$hyperlink_kind,
                     &source_location);
             }
@@ -522,24 +545,80 @@ mod tests {
             test_path!("‹«/test/cool.rs»:«4»:«👉2»›:");
             test_path!("‹«/👉test/cool.rs»(«4»,«2»)›:");
             test_path!("‹«/test/cool.rs»(«4»,«2»👉)›:");
+            test_path!("‹«/👉test/cool.rs»:(«4»,«2»)›:");
+            test_path!("‹«/test/cool.rs»:(«4»,«2»👉)›:");
+            test_path!("‹«/👉test/cool.rs»:(«4»:«2»)›:");
+            test_path!("‹«/test/cool.rs»:(«4»:«2»👉)›:");
+            test_path!("/test/cool.rs:4:2👉:", "What is this?");
+            test_path!("/test/cool.rs(4,2)👉:", "What is this?");
 
             // path, line, column, and description
-            test_path!("‹«/test/cool.rs»:«4»:«2»›👉:Error!");
-            test_path!("‹«/test/cool.rs»:«4»:«2»›:👉Error!");
+            test_path!("/test/cool.rs:4:2👉:Error!");
+            test_path!("/test/cool.rs:4:2:👉Error!");
+            test_path!("‹«/test/co👉ol.rs»:«4»:«2»›:Error!");
             test_path!("‹«/test/co👉ol.rs»(«4»,«2»)›:Error!");
 
             // Cargo output
-            test_path!("    Compiling Cool 👉(‹«/test/Cool»›)");
+            test_path!("    Compiling Cool 👉(/test/Cool)");
             test_path!("    Compiling Cool (‹«/👉test/Cool»›)");
-            test_path!("    Compiling Cool (‹«/test/Cool»›👉)");
+            test_path!("    Compiling Cool (/test/Cool👉)");
 
             // Python
             test_path!("‹«awe👉some.py»›");
 
             test_path!("    ‹F👉ile \"«/awesome.py»\", line «42»›: Wat?");
-            test_path!("    ‹File \"«/awe👉some.py»\", line «42»›: Wat?");
+            test_path!("    ‹File \"«/awe👉some.py»\", line «42»›");
             test_path!("    ‹File \"«/awesome.py»👉\", line «42»›: Wat?");
-            test_path!("    ‹File \"«/awesome.py»\", line «4👉2»›: Wat?");
+            test_path!("    ‹File \"«/awesome.py»\", line «4👉2»›");
+        }
+
+        #[test]
+        fn simple_with_descriptions() {
+            // path, line, column and description
+            test_path!("‹«/👉test/cool.rs»:«4»:«2»›:例Desc例例例");
+            test_path!("‹«/test/cool.rs»:«4»:«👉2»›:例Desc例例例");
+            test_path!("/test/cool.rs:4:2:例Desc例👉例例");
+            test_path!("‹«/👉test/cool.rs»(«4»,«2»)›:例Desc例例例");
+            test_path!("‹«/test/cool.rs»(«4»👉,«2»)›:例Desc例例例");
+            test_path!("/test/cool.rs(4,2):例Desc例👉例例");
+
+            // path, line, column and description w/extra colons
+            test_path!("‹«/👉test/cool.rs»:«4»:«2»›::例Desc例例例");
+            test_path!("‹«/test/cool.rs»:«4»:«👉2»›::例Desc例例例");
+            test_path!("/test/cool.rs:4:2::例Desc例👉例例");
+            test_path!("‹«/👉test/cool.rs»(«4»,«2»)›::例Desc例例例");
+            test_path!("‹«/test/cool.rs»(«4»,«2»👉)›::例Desc例例例");
+            test_path!("/test/cool.rs(4,2)::例Desc例👉例例");
+        }
+
+        #[test]
+        fn multiple_same_line() {
+            test_path!("‹«/👉test/cool.rs»› /test/cool.rs");
+            test_path!("/test/cool.rs ‹«/👉test/cool.rs»›");
+
+            test_path!(
+                "‹«🦀 multiple_👉same_line 🦀» 🚣«4» 🏛️«2»›: 🦀 multiple_same_line 🦀 🚣4 🏛️2:"
+            );
+            test_path!(
+                "🦀 multiple_same_line 🦀 🚣4 🏛️2 ‹«🦀 multiple_👉same_line 🦀» 🚣«4» 🏛️«2»›:"
+            );
+
+            // ls output (tab separated)
+            test_path!(
+                "‹«Carg👉o.toml»›\t\texperiments\t\tnotebooks\t\trust-toolchain.toml\ttooling"
+            );
+            test_path!(
+                "Cargo.toml\t\t‹«exper👉iments»›\t\tnotebooks\t\trust-toolchain.toml\ttooling"
+            );
+            test_path!(
+                "Cargo.toml\t\texperiments\t\t‹«note👉books»›\t\trust-toolchain.toml\ttooling"
+            );
+            test_path!(
+                "Cargo.toml\t\texperiments\t\tnotebooks\t\t‹«rust-t👉oolchain.toml»›\ttooling"
+            );
+            test_path!(
+                "Cargo.toml\t\texperiments\t\tnotebooks\t\trust-toolchain.toml\t‹«too👉ling»›"
+            );
         }
 
         #[test]
@@ -555,6 +634,7 @@ mod tests {
             test_path!("‹«/test/co👉ol.rs»::«42»›");
             test_path!("‹«/test/co👉ol.rs»::«42»›:");
             test_path!("‹«/test/co👉ol.rs:4:2»(«1»,«618»)›");
+            test_path!("‹«/test/co👉ol.rs:4:2»(«1»,«618»)›:");
             test_path!("‹«/test/co👉ol.rs»(«1»,«618»)›::");
         }
 
@@ -570,7 +650,58 @@ mod tests {
             test_path!("<‹«/test/co👉ol.rs»:«4»›>");
 
             test_path!("[\"‹«/test/co👉ol.rs»:«4»›\"]");
-            test_path!("'(‹«/test/co👉ol.rs»:«4»›)'");
+            test_path!("'‹«(/test/co👉ol.rs:4)»›'");
+
+            test_path!("\"‹«/test/co👉ol.rs»:«4»:«2»›\"");
+            test_path!("'‹«/test/co👉ol.rs»:«4»:«2»›'");
+            test_path!("`‹«/test/co👉ol.rs»:«4»:«2»›`");
+
+            test_path!("[‹«/test/co👉ol.rs»:«4»:«2»›]");
+            test_path!("(‹«/test/co👉ol.rs»:«4»:«2»›)");
+            test_path!("{‹«/test/co👉ol.rs»:«4»:«2»›}");
+            test_path!("<‹«/test/co👉ol.rs»:«4»:«2»›>");
+
+            test_path!("[\"‹«/test/co👉ol.rs»:«4»:«2»›\"]");
+
+            test_path!("\"‹«/test/co👉ol.rs»(«4»)›\"");
+            test_path!("'‹«/test/co👉ol.rs»(«4»)›'");
+            test_path!("`‹«/test/co👉ol.rs»(«4»)›`");
+
+            test_path!("[‹«/test/co👉ol.rs»(«4»)›]");
+            test_path!("(‹«/test/co👉ol.rs»(«4»)›)");
+            test_path!("{‹«/test/co👉ol.rs»(«4»)›}");
+            test_path!("<‹«/test/co👉ol.rs»(«4»)›>");
+
+            test_path!("[\"‹«/test/co👉ol.rs»(«4»)›\"]");
+
+            test_path!("\"‹«/test/co👉ol.rs»(«4»,«2»)›\"");
+            test_path!("'‹«/test/co👉ol.rs»(«4»,«2»)›'");
+            test_path!("`‹«/test/co👉ol.rs»(«4»,«2»)›`");
+
+            test_path!("[‹«/test/co👉ol.rs»(«4»,«2»)›]");
+            test_path!("(‹«/test/co👉ol.rs»(«4»,«2»)›)");
+            test_path!("{‹«/test/co👉ol.rs»(«4»,«2»)›}");
+            test_path!("<‹«/test/co👉ol.rs»(«4»,«2»)›>");
+
+            test_path!("[\"‹«/test/co👉ol.rs»(«4»,«2»)›\"]");
+
+            // Imbalanced
+            test_path!("([‹«/test/co👉ol.rs»:«4»›] was here...)");
+            test_path!("[Here's <‹«/test/co👉ol.rs»:«4»›>]");
+            test_path!("('‹«/test/co👉ol.rs»:«4»›' was here...)");
+            test_path!("[Here's `‹«/test/co👉ol.rs»:«4»›`]");
+        }
+
+        #[test]
+        fn trailing_punctuation() {
+            test_path!("‹«/test/co👉ol.rs»›:,..");
+            test_path!("/test/cool.rs:,👉..");
+            test_path!("‹«/test/co👉ol.rs»:«4»›:,");
+            test_path!("/test/cool.rs:4:👉,");
+            test_path!("[\"‹«/test/co👉ol.rs»:«4»›\"]:,");
+            test_path!("'‹«(/test/co👉ol.rs:4),,»›'..");
+            test_path!("('‹«/test/co👉ol.rs»:«4»›'::: was here...)");
+            test_path!("[Here's <‹«/test/co👉ol.rs»:«4»›>]::: ");
         }
 
         #[test]
@@ -585,6 +716,20 @@ mod tests {
             test_path!("    Compiling Cool (‹«/👉例/Cool»›)");
             test_path!("    Compiling Cool (‹«/例👈/Cool»›)");
 
+            test_path!("    Compiling Cool (‹«/👉例/Cool Spaces»›)");
+            test_path!("    Compiling Cool (‹«/例👈/Cool Spaces»›)");
+            test_path!("    Compiling Cool (‹«/👉例/Cool Spaces»:«4»:«2»›)");
+            test_path!("    Compiling Cool (‹«/例👈/Cool Spaces»(«4»,«2»)›)");
+
+            test_path!("    --> ‹«/👉例/Cool Spaces»›");
+            test_path!("    ::: ‹«/例👈/Cool Spaces»›");
+            test_path!("    --> ‹«/👉例/Cool Spaces»:«4»:«2»›");
+            test_path!("    ::: ‹«/例👈/Cool Spaces»(«4»,«2»)›");
+            test_path!("    panicked at ‹«/👉例/Cool Spaces»:«4»:«2»›:");
+            test_path!("    panicked at ‹«/例👈/Cool Spaces»(«4»,«2»)›:");
+            test_path!("    at ‹«/👉例/Cool Spaces»:«4»:«2»›");
+            test_path!("    at ‹«/例👈/Cool Spaces»(«4»,«2»)›");
+
             // Python
             test_path!("‹«👉例wesome.py»›");
             test_path!("‹«例👈wesome.py»›");
@@ -624,7 +769,14 @@ mod tests {
             }
 
             #[test]
-            #[should_panic(expected = "No hyperlink found")]
+            // <https://github.com/zed-industries/zed/issues/12338>
+            fn issue_12338_regex() {
+                // Issue #12338
+                test_path!(".rw-r--r--     0     staff 05-27 14:03 ‹«'test file 👉1.txt'»›");
+                test_path!(".rw-r--r--     0     staff 05-27 14:03 ‹«👉'test file 1.txt'»›");
+            }
+
+            #[test]
             // <https://github.com/zed-industries/zed/issues/12338>
             fn issue_12338() {
                 // Issue #12338
@@ -659,30 +811,48 @@ mod tests {
             }
 
             #[test]
-            #[cfg_attr(
-                not(target_os = "windows"),
-                should_panic(
-                    expected = "Path = «test/controllers/template_items_controller_test.rb», line = 20, at grid cells (0, 0)..=(17, 1)"
-                )
-            )]
-            #[cfg_attr(
-                target_os = "windows",
-                should_panic(
-                    expected = r#"Path = «test\\controllers\\template_items_controller_test.rb», line = 20, at grid cells (0, 0)..=(17, 1)"#
-                )
-            )]
+            // <https://github.com/zed-industries/zed/issues/40202>
+            fn issue_40202() {
+                // Elixir
+                test_path!("[‹«lib/blitz_apex_👉server/stats/aggregate_rank_stats.ex»:«35»›: BlitzApexServer.Stats.AggregateRankStats.update/2]
+                1 #=> 1");
+            }
+
+            #[test]
             // <https://github.com/zed-industries/zed/issues/28194>
-            //
-            // #28194 was closed, but the link includes the description part (":in" here), which
-            // seems wrong...
             fn issue_28194() {
                 test_path!(
                     "‹«test/c👉ontrollers/template_items_controller_test.rb»:«20»›:in 'block (2 levels) in <class:TemplateItemsControllerTest>'"
                 );
                 test_path!(
-                    "‹«test/controllers/template_items_controller_test.rb»:«19»›:i👉n 'block in <class:TemplateItemsControllerTest>'"
+                    "test/controllers/template_items_controller_test.rb:19:i👉n 'block in <class:TemplateItemsControllerTest>'"
                 );
             }
+
+            #[test]
+            #[cfg_attr(
+                not(target_os = "windows"),
+                should_panic(
+                    expected = "Path = «/test/cool.rs:4:NotDesc», at grid cells (0, 1)..=(7, 2)"
+                )
+            )]
+            #[cfg_attr(
+                target_os = "windows",
+                should_panic(
+                    expected = r#"Path = «C:\\test\\cool.rs:4:NotDesc», at grid cells (0, 1)..=(8, 1)"#
+                )
+            )]
+            // PathWithPosition::parse_str considers "/test/co👉ol.rs:4:NotDesc" invalid input, but
+            // still succeeds and truncates the part after the position. Ideally this would be
+            // parsed as the path "/test/co👉ol.rs:4:NotDesc" with no position.
+            fn path_with_position_parse_str() {
+                test_path!("`‹«/test/co👉ol.rs:4:NotDesc»›`");
+                test_path!("<‹«/test/co👉ol.rs:4:NotDesc»›>");
+
+                test_path!("'‹«(/test/co👉ol.rs:4:2)»›'");
+                test_path!("'‹«(/test/co👉ol.rs(4))»›'");
+                test_path!("'‹«(/test/co👉ol.rs(4,2))»›'");
+            }
         }
 
         /// Minor issues arguably not important enough to fix/workaround...
@@ -715,35 +885,38 @@ mod tests {
                 test_path!("‹«/👉test/cool.rs(1,618033988749)»›");
             }
 
-            #[test]
-            #[should_panic(expected = "Path = «»")]
-            fn colon_suffix_succeeds_in_finding_an_empty_maybe_path() {
-                test_path!("‹«/test/cool.rs»:«4»:«2»›👉:", "What is this?");
-                test_path!("‹«/test/cool.rs»(«4»,«2»)›👉:", "What is this?");
-            }
-
             #[test]
             #[cfg_attr(
                 not(target_os = "windows"),
-                should_panic(expected = "Path = «/test/cool.rs»")
+                should_panic(expected = "Path = «/te:st/co:ol.r:s:4:2::::::»")
             )]
             #[cfg_attr(
                 target_os = "windows",
-                should_panic(expected = r#"Path = «C:\\test\\cool.rs»"#)
+                should_panic(expected = r#"Path = «C:\\te:st\\co:ol.r:s:4:2::::::»"#)
             )]
             fn many_trailing_colons_should_be_parsed_as_part_of_the_path() {
-                test_path!("‹«/test/cool.rs:::👉:»›");
                 test_path!("‹«/te:st/👉co:ol.r:s:4:2::::::»›");
+                test_path!("/test/cool.rs:::👉:");
             }
         }
 
-        #[cfg(target_os = "windows")]
         mod windows {
             // Lots of fun to be had with long file paths (verbatim) and UNC paths on Windows.
             // See <https://learn.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation>
             // See <https://users.rust-lang.org/t/understanding-windows-paths/58583>
             // See <https://github.com/rust-lang/cargo/issues/13919>
 
+            #[test]
+            fn default_prompts() {
+                // Windows command prompt
+                test_path!(r#"‹«C:\Users\someone\👉test»›>"#);
+                test_path!(r#"C:\Users\someone\test👉>"#);
+
+                // Windows PowerShell
+                test_path!(r#"PS ‹«C:\Users\someone\👉test\cool.rs»›>"#);
+                test_path!(r#"PS C:\Users\someone\test\cool.rs👉>"#);
+            }
+
             #[test]
             fn unc() {
                 test_path!(r#"‹«\\server\share\👉test\cool.rs»›"#);
@@ -752,24 +925,116 @@ mod tests {
 
             mod issues {
                 #[test]
-                #[should_panic(
-                    expected = r#"Path = «C:\\test\\cool.rs», at grid cells (0, 0)..=(6, 0)"#
-                )]
                 fn issue_verbatim() {
                     test_path!(r#"‹«\\?\C:\👉test\cool.rs»›"#);
                     test_path!(r#"‹«\\?\C:\test\cool👉.rs»›"#);
                 }
 
                 #[test]
-                #[should_panic(
-                    expected = r#"Path = «\\\\server\\share\\test\\cool.rs», at grid cells (0, 0)..=(10, 2)"#
-                )]
                 fn issue_verbatim_unc() {
                     test_path!(r#"‹«\\?\UNC\server\share\👉test\cool.rs»›"#);
                     test_path!(r#"‹«\\?\UNC\server\share\test\cool👉.rs»›"#);
                 }
             }
         }
+
+        mod perf {
+            use super::super::*;
+            use crate::TerminalSettings;
+            use alacritty_terminal::{
+                event::VoidListener,
+                grid::Dimensions,
+                index::{Column, Point as AlacPoint},
+                term::test::mock_term,
+                term::{Term, search::Match},
+            };
+            use settings::{self, Settings, SettingsContent};
+            use std::{cell::RefCell, rc::Rc};
+            use util_macros::perf;
+
+            fn build_test_term(line: &str) -> (Term<VoidListener>, AlacPoint) {
+                let content = line.repeat(500);
+                let term = mock_term(&content);
+                let point = AlacPoint::new(
+                    term.grid().bottommost_line() - 1,
+                    Column(term.grid().last_column().0 / 2),
+                );
+
+                (term, point)
+            }
+
+            #[perf]
+            pub fn cargo_hyperlink_benchmark() {
+                const LINE: &str = "    Compiling terminal v0.1.0 (/Hyperlinks/Bench/Source/zed-hyperlinks/crates/terminal)\r\n";
+                thread_local! {
+                    static TEST_TERM_AND_POINT: (Term<VoidListener>, AlacPoint) =
+                        build_test_term(LINE);
+                }
+                TEST_TERM_AND_POINT.with(|(term, point)| {
+                    assert!(
+                        find_from_grid_point_bench(term, *point).is_some(),
+                        "Hyperlink should have been found"
+                    );
+                });
+            }
+
+            #[perf]
+            pub fn rust_hyperlink_benchmark() {
+                const LINE: &str = "    --> /Hyperlinks/Bench/Source/zed-hyperlinks/crates/terminal/terminal.rs:1000:42\r\n";
+                thread_local! {
+                    static TEST_TERM_AND_POINT: (Term<VoidListener>, AlacPoint) =
+                        build_test_term(LINE);
+                }
+                TEST_TERM_AND_POINT.with(|(term, point)| {
+                    assert!(
+                        find_from_grid_point_bench(term, *point).is_some(),
+                        "Hyperlink should have been found"
+                    );
+                });
+            }
+
+            #[perf]
+            pub fn ls_hyperlink_benchmark() {
+                const LINE: &str = "Cargo.toml        experiments        notebooks        rust-toolchain.toml    tooling\r\n";
+                thread_local! {
+                    static TEST_TERM_AND_POINT: (Term<VoidListener>, AlacPoint) =
+                        build_test_term(LINE);
+                }
+                TEST_TERM_AND_POINT.with(|(term, point)| {
+                    assert!(
+                        find_from_grid_point_bench(term, *point).is_some(),
+                        "Hyperlink should have been found"
+                    );
+                });
+            }
+
+            pub fn find_from_grid_point_bench(
+                term: &Term<VoidListener>,
+                point: AlacPoint,
+            ) -> Option<(String, bool, Match)> {
+                const PATH_HYPERLINK_TIMEOUT_MS: u64 = 1000;
+
+                thread_local! {
+                    static TEST_REGEX_SEARCHES: RefCell<RegexSearches> =
+                        RefCell::new({
+                            let default_settings_content: Rc<SettingsContent> =
+                                settings::parse_json_with_comments(&settings::default_settings())
+                                    .unwrap();
+                            let default_terminal_settings =
+                                TerminalSettings::from_settings(&default_settings_content);
+
+                            RegexSearches::new(
+                                &default_terminal_settings.path_hyperlink_regexes,
+                                PATH_HYPERLINK_TIMEOUT_MS
+                            )
+                        });
+                }
+
+                TEST_REGEX_SEARCHES.with(|regex_searches| {
+                    find_from_grid_point(&term, point, &mut regex_searches.borrow_mut())
+                })
+            }
+        }
     }
 
     mod file_iri {
@@ -821,11 +1086,12 @@ mod tests {
                 }
 
                 // See https://en.wikipedia.org/wiki/File_URI_scheme
+                // https://github.com/zed-industries/zed/issues/39189
                 #[test]
                 #[should_panic(
                     expected = r#"Path = «C:\\test\\cool\\index.rs», at grid cells (0, 0)..=(9, 1)"#
                 )]
-                fn issue_absolute_file_iri() {
+                fn issue_39189() {
                     test_file_iri!("file:///C:/test/cool/index.rs");
                     test_file_iri!("file:///C:/test/cool/");
                 }
@@ -981,7 +1247,7 @@ mod tests {
             let mut point = cursor.point;
 
             if !cursor.input_needs_wrap {
-                point.column -= 1;
+                point = point.sub(term, Boundary::Grid, 1);
             }
 
             if grid.index(point).flags.contains(Flags::WIDE_CHAR_SPACER) {
@@ -1007,6 +1273,13 @@ mod tests {
             }
         }
 
+        fn process_input(term: &mut Term<VoidListener>, c: char) {
+            match c {
+                '\t' => term.put_tab(1),
+                c @ _ => term.input(c),
+            }
+        }
+
         let mut hovered_grid_point: Option<AlacPoint> = None;
         let mut hyperlink_match = AlacPoint::default()..=AlacPoint::default();
         let mut iri_or_path = String::default();
@@ -1098,9 +1371,9 @@ mod tests {
                             term.input('C');
                             prev_input_point = prev_input_point_from_term(&term);
                             term.input(':');
-                            term.input(c);
+                            process_input(&mut term, c);
                         } else {
-                            term.input(c);
+                            process_input(&mut term, c);
                             prev_input_point = prev_input_point_from_term(&term);
                         }
 
@@ -1130,15 +1403,6 @@ mod tests {
             iri_or_path = path.to_string_lossy().into_owned();
         }
 
-        if cfg!(windows) {
-            // Handle verbatim and UNC paths for Windows
-            if let Some(stripped) = iri_or_path.strip_prefix(r#"\\?\UNC\"#) {
-                iri_or_path = format!(r#"\\{stripped}"#);
-            } else if let Some(stripped) = iri_or_path.strip_prefix(r#"\\?\"#) {
-                iri_or_path = stripped.to_string();
-            }
-        }
-
         let hovered_grid_point = hovered_grid_point.expect("Missing hovered point (👉 or 👈)");
         let hovered_char = term.grid().index(hovered_grid_point).c;
         (
@@ -1161,6 +1425,7 @@ mod tests {
             match c {
                 // Fullwidth unicode characters used in tests
                 '例' | '🏃' | '🦀' | '🔥' => 2,
+                '\t' => 8, // it's really 0-8, use the max always
                 _ => 1,
             }
         }
@@ -1283,11 +1548,9 @@ mod tests {
             let mut marker_header_row = String::new();
             for index in 0..self.term.columns() {
                 let remainder = index % 10;
-                first_header_row.push_str(
-                    &(index > 0 && remainder == 0)
-                        .then_some((index / 10).to_string())
-                        .unwrap_or(" ".into()),
-                );
+                if index > 0 && remainder == 0 {
+                    first_header_row.push_str(&format!("{:>10}", (index / 10)));
+                }
                 second_header_row += &remainder.to_string();
                 if index == self.expected_hyperlink.hovered_grid_point.column.0 {
                     marker_header_row.push('↓');
@@ -1296,16 +1559,20 @@ mod tests {
                 }
             }
 
-            result += &format!("\n      [{}]\n", first_header_row);
+            let remainder = (self.term.columns() - 1) % 10;
+            if remainder != 0 {
+                first_header_row.push_str(&" ".repeat(remainder));
+            }
+
+            result += &format!("\n      [ {}]\n", first_header_row);
             result += &format!("      [{}]\n", second_header_row);
             result += &format!("       {}", marker_header_row);
 
-            let spacers: Flags = Flags::LEADING_WIDE_CHAR_SPACER | Flags::WIDE_CHAR_SPACER;
             for cell in self
                 .term
                 .renderable_content()
                 .display_iter
-                .filter(|cell| !cell.flags.intersects(spacers))
+                .filter(|cell| !cell.flags.intersects(WIDE_CHAR_SPACERS))
             {
                 if cell.point.column.0 == 0 {
                     let prefix =
@@ -1317,7 +1584,10 @@ mod tests {
                     result += &format!("\n{prefix}[{:>3}] ", cell.point.line.to_string());
                 }
 
-                result.push(cell.c);
+                match cell.c {
+                    '\t' => result.push(' '),
+                    c @ _ => result.push(c),
+                }
             }
 
             result
@@ -1331,8 +1601,34 @@ mod tests {
         hyperlink_kind: HyperlinkKind,
         source_location: &str,
     ) {
+        const CARGO_DIR_REGEX: &str =
+            r#"\s+(Compiling|Checking|Documenting) [^(]+\((?<link>(?<path>.+))\)"#;
+        const RUST_DIAGNOSTIC_REGEX: &str = r#"\s+(-->|:::|at) (?<link>(?<path>.+?))(:$|$)"#;
+        const ISSUE_12338_REGEX: &str =
+            r#"[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2} (?<link>(?<path>.+))"#;
+        const MULTIPLE_SAME_LINE_REGEX: &str =
+            r#"(?<link>(?<path>🦀 multiple_same_line 🦀) 🚣(?<line>[0-9]+) 🏛(?<column>[0-9]+)):"#;
+        const PATH_HYPERLINK_TIMEOUT_MS: u64 = 1000;
+
         thread_local! {
-            static TEST_REGEX_SEARCHES: RefCell<RegexSearches> = RefCell::new(RegexSearches::new());
+            static TEST_REGEX_SEARCHES: RefCell<RegexSearches> =
+                RefCell::new({
+                    let default_settings_content: Rc<SettingsContent> =
+                        settings::parse_json_with_comments(&settings::default_settings()).unwrap();
+                    let default_terminal_settings = TerminalSettings::from_settings(&default_settings_content);
+
+                    RegexSearches::new([
+                        RUST_DIAGNOSTIC_REGEX,
+                        CARGO_DIR_REGEX,
+                        ISSUE_12338_REGEX,
+                        MULTIPLE_SAME_LINE_REGEX,
+                    ]
+                        .into_iter()
+                        .chain(default_terminal_settings.path_hyperlink_regexes
+                            .iter()
+                            .map(AsRef::as_ref)),
+                    PATH_HYPERLINK_TIMEOUT_MS)
+                });
         }
 
         let term_size = TermSize::new(columns, total_cells / columns + 2);
@@ -1357,12 +1653,16 @@ mod tests {
             Some((hyperlink_word, true, hyperlink_match)) => {
                 check_hyperlink_match.check_iri_and_match(hyperlink_word, &hyperlink_match);
             }
-            _ => {
-                assert!(
-                    false,
-                    "No hyperlink found\n     at {source_location}:\n{}",
-                    check_hyperlink_match.format_renderable_content()
-                )
+            None => {
+                if expected_hyperlink.hyperlink_match.start()
+                    != expected_hyperlink.hyperlink_match.end()
+                {
+                    assert!(
+                        false,
+                        "No hyperlink found\n     at {source_location}:\n{}",
+                        check_hyperlink_match.format_renderable_content()
+                    )
+                }
             }
         }
     }
diff --git a/crates/terminal/src/terminal_settings.rs b/crates/terminal/src/terminal_settings.rs
index 0c6f03832c..3b3070c6f6 100644
--- a/crates/terminal/src/terminal_settings.rs
+++ b/crates/terminal/src/terminal_settings.rs
@@ -9,8 +9,8 @@ use serde::{Deserialize, Serialize};
 pub use settings::AlternateScroll;
 
 use settings::{
-    RegisterSetting, ShowScrollbar, TerminalBlink, TerminalDockPosition, TerminalLineHeight,
-    VenvSettings, WorkingDirectory, merge_from::MergeFrom,
+    PathHyperlinkRegex, RegisterSetting, ShowScrollbar, TerminalBlink, TerminalDockPosition,
+    TerminalLineHeight, VenvSettings, WorkingDirectory, merge_from::MergeFrom,
 };
 use task::Shell;
 use theme::FontFamilyName;
@@ -47,6 +47,8 @@ pub struct TerminalSettings {
     pub toolbar: Toolbar,
     pub scrollbar: ScrollbarSettings,
     pub minimum_contrast: f32,
+    pub path_hyperlink_regexes: Vec<String>,
+    pub path_hyperlink_timeout_ms: u64,
 }
 
 #[derive(Copy, Clone, Debug, Serialize, Deserialize, JsonSchema, PartialEq, Eq)]
@@ -116,6 +118,16 @@ impl settings::Settings for TerminalSettings {
                 show: user_content.scrollbar.unwrap().show,
             },
             minimum_contrast: user_content.minimum_contrast.unwrap(),
+            path_hyperlink_regexes: project_content
+                .path_hyperlink_regexes
+                .unwrap()
+                .into_iter()
+                .map(|regex| match regex {
+                    PathHyperlinkRegex::SingleLine(regex) => regex,
+                    PathHyperlinkRegex::MultiLine(regex) => regex.join("\n"),
+                })
+                .collect(),
+            path_hyperlink_timeout_ms: project_content.path_hyperlink_timeout_ms.unwrap(),
         }
     }
 }
diff --git a/docs/src/configuring-zed.md b/docs/src/configuring-zed.md
index a3e24506c4..6edcafb3d8 100644
--- a/docs/src/configuring-zed.md
+++ b/docs/src/configuring-zed.md
@@ -4115,6 +4115,53 @@ Example command to set the title: `echo -e "\e]2;New Title\007";`
 }
 ```
 
+### Terminal: Path Hyperlink Regexes
+
+- Description: Regexes used to identify path hyperlinks. The regexes can be specified in two forms - a single regex string, or an array of strings (which will be collected into a single multi-line regex string).
+- Setting: `path_hyperlink_regexes`
+- Default:
+
+```json [settings]
+{
+  "terminal": {
+    "path_hyperlink_regexes": [
+      // Python-style diagnostics
+      "File \"(?<path>[^\"]+)\", line (?<line>[0-9]+)",
+      // Common path syntax with optional line, column, description, trailing punctuation, or
+      // surrounding symbols or quotes
+      [
+        "(?x)",
+        "# optionally starts with 0-2 opening prefix symbols",
+        "[({\\[<]{0,2}",
+        "# which may be followed by an opening quote",
+        "(?<quote>[\"'`])?",
+        "# `path` is the shortest sequence of any non-space character",
+        "(?<link>(?<path>[^ ]+?",
+        "    # which may end with a line and optionally a column,",
+        "    (?<line_column>:+[0-9]+(:[0-9]+)?|:?\\([0-9]+([,:][0-9]+)?\\))?",
+        "))",
+        "# which must be followed by a matching quote",
+        "(?(<quote>)\\k<quote>)",
+        "# and optionally a single closing symbol",
+        "[)}\\]>]?",
+        "# if line/column matched, may be followed by a description",
+        "(?(<line_column>):[^ 0-9][^ ]*)?",
+        "# which may be followed by trailing punctuation",
+        "[.,:)}\\]>]*",
+        "# and always includes trailing whitespace or end of line",
+        "([ ]+|$)"
+      ]
+    ]
+  }
+}
+```
+
+### Terminal: Path Hyperlink Timeout (ms)
+
+- Description: Maximum time to search for a path hyperlink. When set to 0, path hyperlinks are disabled.
+- Setting: `path_hyperlink_timeout_ms`
+- Default: `1`
+
 ## REPL
 
 - Description: Repl settings.