Clean-up string sanitation helpers (#2660)

esphome/components/mqtt/mqtt_component.cpp

@@ -17,7 +17,7 @@ static const char *const TAG = "mqtt.component";
 void MQTTComponent::set_retain(bool retain) { this->retain_ = retain; }
 
 std::string MQTTComponent::get_discovery_topic_(const MQTTDiscoveryInfo &discovery_info) const {
-  std::string sanitized_name = sanitize_string_allowlist(App.get_name(), HOSTNAME_CHARACTER_ALLOWLIST);
+  std::string sanitized_name = str_sanitize(App.get_name());
   return discovery_info.prefix + "/" + this->component_type() + "/" + sanitized_name + "/" +
          this->get_default_object_id_() + "/config";
 }
@@ -136,7 +136,7 @@ bool MQTTComponent::is_discovery_enabled() const {
 }
 
 std::string MQTTComponent::get_default_object_id_() const {
-  return sanitize_string_allowlist(to_lowercase_underscore(this->friendly_name()), HOSTNAME_CHARACTER_ALLOWLIST);
+  return str_sanitize(str_snake_case(this->friendly_name()));
 }
 
 void MQTTComponent::subscribe(const std::string &topic, mqtt_callback_t callback, uint8_t qos) {

esphome/core/entity_base.cpp

@@ -35,7 +35,7 @@ const std::string &EntityBase::get_object_id() { return this->object_id_; }
 
 // Calculate Object ID Hash from Entity Name
 void EntityBase::calc_object_id_() {
-  this->object_id_ = sanitize_string_allowlist(to_lowercase_underscore(this->name_), HOSTNAME_CHARACTER_ALLOWLIST);
+  this->object_id_ = str_sanitize(str_snake_case(this->name_));
   // FNV-1 hash
   this->object_id_hash_ = fnv1_hash(this->object_id_);
 }

esphome/core/helpers.cpp

@@ -128,31 +128,6 @@ float gamma_uncorrect(float value, float gamma) {
   return powf(value, 1 / gamma);
 }
 
-std::string to_lowercase_underscore(std::string s) {
-  std::transform(s.begin(), s.end(), s.begin(), ::tolower);
-  std::replace(s.begin(), s.end(), ' ', '_');
-  return s;
-}
-
-std::string sanitize_string_allowlist(const std::string &s, const std::string &allowlist) {
-  std::string out(s);
-  out.erase(std::remove_if(out.begin(), out.end(),
-                           [&allowlist](const char &c) { return allowlist.find(c) == std::string::npos; }),
-            out.end());
-  return out;
-}
-
-std::string sanitize_hostname(const std::string &hostname) {
-  std::string s = sanitize_string_allowlist(hostname, HOSTNAME_CHARACTER_ALLOWLIST);
-  return truncate_string(s, 63);
-}
-
-std::string truncate_string(const std::string &s, size_t length) {
-  if (s.length() > length)
-    return s.substr(0, length);
-  return s;
-}
-
 std::string value_accuracy_to_string(float value, int8_t accuracy_decimals) {
   if (accuracy_decimals < 0) {
     auto multiplier = powf(10.0f, accuracy_decimals);
@@ -191,8 +166,6 @@ ParseOnOffState parse_on_off(const char *str, const char *on, const char *off) {
   return PARSE_NONE;
 }
 
-const char *const HOSTNAME_CHARACTER_ALLOWLIST = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-_";
-
 uint8_t crc8(uint8_t *data, uint8_t len) {
   uint8_t crc = 0;
 
@@ -481,4 +454,24 @@ IRAM_ATTR InterruptLock::InterruptLock() { portDISABLE_INTERRUPTS(); }
 IRAM_ATTR InterruptLock::~InterruptLock() { portENABLE_INTERRUPTS(); }
 #endif
 
+// ---------------------------------------------------------------------------------------------------------------------
+
+std::string str_truncate(const std::string &str, size_t length) {
+  return str.length() > length ? str.substr(0, length) : str;
+}
+std::string str_snake_case(const std::string &str) {
+  std::string result;
+  result.resize(str.length());
+  std::transform(str.begin(), str.end(), result.begin(), ::tolower);
+  std::replace(result.begin(), result.end(), ' ', '_');
+  return result;
+}
+std::string str_sanitize(const std::string &str) {
+  std::string out;
+  std::copy_if(str.begin(), str.end(), std::back_inserter(out), [](const char &c) {
+    return c == '-' || c == '_' || (c >= '0' && c <= '9') || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
+  });
+  return out;
+}
+
 }  // namespace esphome

esphome/core/helpers.h

@@ -25,9 +25,6 @@
 
 namespace esphome {
 
-/// The characters that are allowed in a hostname.
-extern const char *const HOSTNAME_CHARACTER_ALLOWLIST;
-
 /// Read the raw MAC address into the provided byte array (6 bytes).
 void get_mac_address_raw(uint8_t *mac);
 
@@ -55,14 +52,6 @@ std::string to_string(double val);
 std::string to_string(long double val);
 optional<int> parse_hex(const std::string &str, size_t start, size_t length);
 optional<int> parse_hex(char chr);
-/// Sanitize the hostname by removing characters that are not in the allowlist and truncating it to 63 chars.
-std::string sanitize_hostname(const std::string &hostname);
-
-/// Truncate a string to a specific length
-std::string truncate_string(const std::string &s, size_t length);
-
-/// Convert the string to lowercase_underscore.
-std::string to_lowercase_underscore(std::string s);
 
 /// Compare string a to string b (ignoring case) and return whether they are equal.
 bool str_equals_case_insensitive(const std::string &a, const std::string &b);
@@ -145,9 +134,6 @@ std::string uint64_to_string(uint64_t num);
 /// Convert a uint32_t to a hex string
 std::string uint32_to_string(uint32_t num);
 
-/// Sanitizes the input string with the allowlist.
-std::string sanitize_string_allowlist(const std::string &s, const std::string &allowlist);
-
 uint8_t reverse_bits_8(uint8_t x);
 uint16_t reverse_bits_16(uint16_t x);
 uint32_t reverse_bits_32(uint32_t x);
@@ -331,6 +317,20 @@ template<typename T, enable_if_t<std::is_unsigned<T>::value, int> = 0> constexpr
 
 ///@}
 
+/// @name Strings
+///@{
+
+/// Truncate a string to a specific length.
+std::string str_truncate(const std::string &str, size_t length);
+
+/// Convert the string to snake case (lowercase with underscores).
+std::string str_snake_case(const std::string &str);
+
+/// Sanitizes the input string by removing all characters but alphanumerics, dashes and underscores.
+std::string str_sanitize(const std::string &str);
+
+///@}
+
 /// @name Parsing & formatting
 ///@{