Add sanitize_identifier() utility function

2 files changed, 34 insertions, 0 deletions

diff --git a/libbutl/utility.ixx b/libbutl/utility.ixx
index 04fb161..8f37242 100644
--- a/libbutl/utility.ixx
+++ b/libbutl/utility.ixx
@@ -4,6 +4,7 @@
 
 #ifndef __cpp_lib_modules_ts
 #include <cstdlib> // getenv()
+#include <algorithm>
 #endif
 
 namespace butl
@@ -192,6 +193,30 @@ namespace butl
     return e - b;
   }
 
+  inline std::string&
+  sanitize_identifier (std::string& s)
+  {
+    std::for_each (s.begin (), s.end (), [] (char& c)
+                   {
+                     if (!alnum (c) && c != '_')
+                       c = '_';
+                   });
+    return s;
+  }
+
+  inline std::string
+  sanitize_identifier (std::string&& s)
+  {
+    sanitize_identifier (s);
+    return std::move (s);
+  }
+
+  inline std::string
+  sanitize_identifier (const std::string& s)
+  {
+    return sanitize_identifier (std::string (s));
+  }
+
   inline bool
   eof (std::istream& is)
   {
diff --git a/libbutl/utility.mxx b/libbutl/utility.mxx
index 8085e79..03fb89e 100644
--- a/libbutl/utility.mxx
+++ b/libbutl/utility.mxx
@@ -186,6 +186,15 @@ LIBBUTL_MODEXPORT namespace butl
   next_word (const std::string&, std::size_t n, std::size_t& b, std::size_t& e,
              char d1 = ' ', char d2 = '\0');
 
+  // Sanitize a string to only contain characters valid in an identifier
+  // (ASCII alphanumeric plus `_`) replacing all others with `_`.
+  //
+  // Note that it doesn't make sure the first character is not a digit.
+  //
+  std::string& sanitize_identifier (std::string&);
+  std::string sanitize_identifier (std::string&&);
+  std::string sanitize_identifier (const std::string&);
+
   // If an input stream is in a failed state, then return true if this is
   // because of the eof and throw istream::failure otherwise. If the stream
   // is not in a failed state, return false. This helper function is normally