utils: Add remove_ctrl

Add a helper function to remove control characters / multi-spaces, and a
test.

2 files changed, 32 insertions, 0 deletions

diff --git a/lib/utils/mu-utils.cc b/lib/utils/mu-utils.cc
index 59ece80..d80dfb2 100644
--- a/lib/utils/mu-utils.cc
+++ b/lib/utils/mu-utils.cc
@@ -147,6 +147,22 @@ Mu::utf8_clean (const std::string& dirty)
 	clean.erase (clean.find_last_not_of(" ") + 1); // remove trailing space
 
 	return clean;
+std::string
+Mu::remove_ctrl (const std::string& str)
+{
+        char prev{'\0'};
+        std::string result;
+        result.reserve(str.length());
+
+        for (auto&& c: str) {
+                if (::iscntrl(c) || c == ' ') {
+                        if (prev != ' ')
+                                result += prev = ' ';
+                } else
+                        result += prev = c;
+        }
+
+        return result;
 }
 
 std::vector<std::string>
diff --git a/lib/utils/test-utils.cc b/lib/utils/test-utils.cc
index b1f666e..020135f 100644
--- a/lib/utils/test-utils.cc
+++ b/lib/utils/test-utils.cc
@@ -137,6 +137,22 @@ test_flatten ()
 }
 
 static void
+test_remove_ctrl ()
+{
+	CaseVec cases = {
+		{ "Foo\n\nbar", true,  "Foo bar" },
+		{ "",    false, "" },
+		{ "   ",    false, " " },
+		{ "Hello   World   ",    false, "Hello World " },
+		{ "Ångström", false,  "Ångström" },
+	};
+
+	test_cases (cases, [](auto s, auto f){ return remove_ctrl(s); });
+}
+
+
+
+static void
 test_clean ()
 {
 	CaseVec cases = {