From aca46285944193512ab8c5ad4d6a7c11527e1e18 Mon Sep 17 00:00:00 2001
From: Wojciech Kozlowski <wk@wojciechkozlowski.eu>
Date: Sun, 11 Feb 2024 16:54:48 +0100
Subject: [PATCH] Correct-ish implementation with binary search

---
 src/tui/app/selection.rs | 44 +++++++++++++++++++++++++++-------------
 1 file changed, 30 insertions(+), 14 deletions(-)

diff --git a/src/tui/app/selection.rs b/src/tui/app/selection.rs
index edc3a96..147cd50 100644
--- a/src/tui/app/selection.rs
+++ b/src/tui/app/selection.rs
@@ -192,26 +192,42 @@ impl ArtistSelection {
         }
     }
 
-    // FIXME: think about converting punctuation characters to some common character.
+    fn normalize_search_string(search: &str, lowercase: bool) -> String {
+        let normalized = if lowercase {
+            search.to_lowercase()
+        } else {
+            search.to_owned()
+        };
+
+        // Unlikely that this covers all possible strings, but it should at least cover strings
+        // relevant for music (at least in English). The list of characters handled is based on
+        // https://wiki.musicbrainz.org/User:Yurim/Punctuation_and_Special_Characters.
+        normalized
+            .replace("‐", "-") // U+2010 hyphen
+            .replace("‒", "-") // U+2012 figure dash
+            .replace("–", "-") // U+2013 en dash
+            .replace("—", "-") // U+2014 em dash
+            .replace("―", "-") // U+2015 horizontal bar
+            .replace("‘", "'") // U+2018
+            .replace("’", "'") // U+2019
+            .replace("“", "\"") // U+201C
+            .replace("”", "\"") // U+201D
+            .replace("…", "...") // U+2026
+            .replace("−", "-") // U+2212 minus sign
+    }
+
     fn incremental_search(&mut self, artists: &[Artist], artist_name: &str) {
         if let Some(index) = self.state.list.selected() {
             let case_sensitive = artist_name
                 .chars()
-                .any(|ch| !(ch.is_lowercase() || ch.is_whitespace() || ch.is_ascii_punctuation()));
+                .any(|ch| ch.is_alphabetic() && ch.is_uppercase());
+            let search_name = Self::normalize_search_string(artist_name, !case_sensitive);
             let slice = &artists[index..];
 
-            let result = if case_sensitive {
-                slice.binary_search_by(|probe| probe.get_sort_key().name.as_str().cmp(artist_name))
-            } else {
-                slice.binary_search_by(|probe| {
-                    probe
-                        .get_sort_key()
-                        .name
-                        .to_lowercase()
-                        .as_str()
-                        .cmp(artist_name)
-                })
-            };
+            let result = slice.binary_search_by(|probe| {
+                Self::normalize_search_string(&probe.get_sort_key().name, !case_sensitive)
+                    .cmp(&search_name)
+            });
 
             let new_index = match result {
                 Ok(slice_index) | Err(slice_index) => index + slice_index,