fix(hyprland/language): split activelayout on the last comma outside parens

include/util/string.hpp

@@ -1,7 +1,10 @@
 #pragma once
 
+#include <algorithm>
 #include <iostream>
 #include <string>
+#include <string_view>
+#include <vector>
 
 const std::string WHITESPACE = " \n\r\t\f\v";
 
@@ -31,6 +34,24 @@ inline std::string toLower(const std::string& str) {
   return result;
 }
 
+// Finds the last occurrence of `c` that is not nested inside parentheses,
+// scanning from the end. Unbalanced brackets are tolerated: a stray '(' never
+// pushes the depth below zero, and a stray ')' only shields what precedes it.
+// Returns std::string::npos when there is no such occurrence.
+inline size_t rfindOutsideParens(std::string_view s, char c) {
+  int depth = 0;
+  for (size_t i = s.size(); i-- > 0;) {
+    if (s[i] == ')') {
+      ++depth;
+    } else if (s[i] == '(') {
+      if (depth > 0) --depth;
+    } else if (s[i] == c && depth == 0) {
+      return i;
+    }
+  }
+  return std::string::npos;
+}
+
 inline std::vector<std::string> split(std::string_view s, std::string_view delimiter,
                                       int max_splits = -1) {
   std::vector<std::string> result;

src/modules/hyprland/language.cpp

@@ -126,17 +126,15 @@ void Language::onEvent(const std::string& ev) {
     spdlog::warn("hyprland language received malformed event payload: {}", ev);
     return;
   }
-  // Last comma before variants parenthesis, eg:
+  // Both halves of "<keyboard>,<layout>" may contain commas of their own: the
+  // keyboard in its vendor string, the layout inside its variant parentheses, eg
   // activelayout>>micro-star-int'l-co.,-ltd.-msi-gk50-elite-gaming-keyboard,English (US, intl.,
   // with dead keys)
-  std::string beforeParenthesis;
-  auto parenthesisPos = payload.find_last_of('(');
-  if (parenthesisPos == std::string::npos) {
-    beforeParenthesis = payload;
-  } else {
-    beforeParenthesis = payload.substr(0, parenthesisPos);
-  }
-  const auto layoutSeparator = beforeParenthesis.find_last_of(',');
+  // The separator is therefore the last comma outside parentheses. Splitting on
+  // the last '(' instead would assume every bracket belongs to the layout, but
+  // keyboard names carry them too, eg "ite-tech.-inc.-ite-device(8910)-keyboard",
+  // which left no comma to find and dropped the event entirely (#4586).
+  const auto layoutSeparator = rfindOutsideParens(payload, ',');
   if (layoutSeparator == std::string::npos) {
     spdlog::warn("hyprland language received malformed layout payload: {}", ev);
     return;

test/utils/meson.build

@@ -14,6 +14,7 @@ test_src = files(
     'JsonParser.cpp',
     'SafeSignal.cpp',
     'format.cpp',
+    'string.cpp',
     'sleeper_thread.cpp',
     'command.cpp',
     'command_line_stream.cpp',

test/utils/string.cpp

@@ -0,0 +1,64 @@
+#include "util/string.hpp"
+
+#if __has_include(<catch2/catch_test_macros.hpp>)
+#include <catch2/catch_test_macros.hpp>
+#else
+#include <catch2/catch.hpp>
+#endif
+
+// The layout half of a hyprland "activelayout" payload, split the way
+// Language::onEvent splits it.
+static std::string layoutOf(const std::string& payload) {
+  const auto sep = rfindOutsideParens(payload, ',');
+  if (sep == std::string::npos) return "";
+  return payload.substr(sep + 1);
+}
+
+TEST_CASE("rfindOutsideParens", "[string]") {
+  SECTION("returns npos when the character is absent") {
+    REQUIRE(rfindOutsideParens("abc", ',') == std::string::npos);
+    REQUIRE(rfindOutsideParens("", ',') == std::string::npos);
+  }
+
+  SECTION("finds the last match when no parentheses are present") {
+    REQUIRE(rfindOutsideParens("a,b,c", ',') == 3);
+  }
+
+  SECTION("ignores matches nested inside parentheses") {
+    REQUIRE(rfindOutsideParens("a,b(c,d)", ',') == 1);
+    REQUIRE(rfindOutsideParens("a(b,c)d(e,f)", ',') == std::string::npos);
+  }
+
+  SECTION("handles nested parentheses") { REQUIRE(rfindOutsideParens("a,b(c(d,e),f)", ',') == 1); }
+
+  SECTION("tolerates unbalanced parentheses") {
+    // A stray '(' left of the match must not shield it.
+    REQUIRE(rfindOutsideParens("a(b,c", ',') == 3);
+    // A stray ')' must not drive the depth negative and swallow later matches.
+    REQUIRE(rfindOutsideParens("a)b,c", ',') == 3);
+  }
+}
+
+// Regression test for https://github.com/Alexays/Waybar/issues/4586
+TEST_CASE("hyprland activelayout payload split", "[string]") {
+  SECTION("keyboard name containing brackets, layout without") {
+    // Splitting on the last '(' picked the keyboard's own bracket, leaving no
+    // comma behind it, so the event was discarded and the layout never updated.
+    REQUIRE(layoutOf("ite-tech.-inc.-ite-device(8910)-keyboard,Russian") == "Russian");
+  }
+
+  SECTION("layout containing commas inside its variant parentheses") {
+    REQUIRE(layoutOf("micro-star-int'l-co.,-ltd.-msi-gk50-elite-gaming-keyboard,"
+                     "English (US, intl., with dead keys)") ==
+            "English (US, intl., with dead keys)");
+  }
+
+  SECTION("brackets in the keyboard name and commas in the layout variant") {
+    REQUIRE(layoutOf("ite-tech.-inc.-ite-device(8910)-keyboard,English (US, intl.)") ==
+            "English (US, intl.)");
+  }
+
+  SECTION("plain keyboard name and layout") {
+    REQUIRE(layoutOf("at-translated-set-2-keyboard,English (US)") == "English (US)");
+  }
+}