From add18d4e351f582757781aa92f68eaa534ea8748 Mon Sep 17 00:00:00 2001 From: "J. Nick Koston" Date: Thu, 13 Aug 2026 20:06:48 -0500 Subject: [PATCH] [core] Partially revert "Hash entity keys from the raw name to fix collisions" (#18361) --- esphome/components/api/api_connection.cpp | 6 +- esphome/components/infrared/infrared.cpp | 8 +- esphome/components/mqtt/__init__.py | 63 --- esphome/components/prometheus/__init__.py | 6 - .../radio_frequency/radio_frequency.cpp | 8 +- .../template/text/template_text.cpp | 20 +- .../components/template/text/template_text.h | 15 +- esphome/core/application.h | 8 +- esphome/core/entity_base.cpp | 52 +-- esphome/core/entity_base.h | 80 ++-- esphome/core/entity_helpers.py | 190 ++++---- esphome/core/helpers.h | 29 +- esphome/core/preference_backend.h | 14 +- esphome/core/preferences.cpp | 25 - esphome/core/preferences.h | 12 - esphome/helpers.py | 15 +- tests/integration/entity_utils.py | 27 +- .../fixtures/fnv1_hash_object_id.yaml | 32 -- .../fixtures/multi_device_preferences.yaml | 21 +- .../fixtures/preference_key_migration.yaml | 35 -- tests/integration/host_prefs.py | 24 +- tests/integration/test_fnv1_hash_object_id.py | 4 - .../test_object_id_api_verification.py | 10 +- ...t_object_id_friendly_name_no_mac_suffix.py | 4 +- .../test_object_id_no_friendly_name.py | 6 +- .../test_preference_key_migration.py | 165 ------- tests/unit_tests/components/mqtt/__init__.py | 0 .../mqtt/test_object_id_conflicts.py | 239 ---------- tests/unit_tests/core/test_entity_helpers.py | 432 ++++++++++-------- .../object_id_conflict_mqtt.yaml | 22 - .../object_id_conflict_no_mqtt.yaml | 15 - .../test_preference_hash_stability.py | 34 +- 32 files changed, 489 insertions(+), 1132 deletions(-) delete mode 100644 esphome/core/preferences.cpp delete mode 100644 tests/integration/fixtures/preference_key_migration.yaml delete mode 100644 tests/integration/test_preference_key_migration.py delete mode 100644 tests/unit_tests/components/mqtt/__init__.py delete mode 100644 tests/unit_tests/components/mqtt/test_object_id_conflicts.py delete mode 100644 tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_mqtt.yaml delete mode 100644 tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_no_mqtt.yaml diff --git a/esphome/components/api/api_connection.cpp b/esphome/components/api/api_connection.cpp index d05f98d03b2..73b4f3e5bd4 100644 --- a/esphome/components/api/api_connection.cpp +++ b/esphome/components/api/api_connection.cpp @@ -448,7 +448,7 @@ void APIConnection::on_disconnect_response() { uint16_t APIConnection::fill_and_encode_entity_state(EntityBase *entity, StateResponseProtoMessage &msg, CalculateSizeFn size_fn, MessageEncodeFn encode_fn, APIConnection *conn, uint32_t remaining_size) { - msg.key = entity->get_entity_key(); + msg.key = entity->get_object_id_hash(); #ifdef USE_DEVICES msg.device_id = entity->get_device_id(); #endif @@ -459,7 +459,7 @@ uint16_t APIConnection::fill_and_encode_entity_info(EntityBase *entity, InfoResp CalculateSizeFn size_fn, MessageEncodeFn encode_fn, APIConnection *conn, uint32_t remaining_size) { // Set common fields that are shared by all entity types - msg.key = entity->get_entity_key(); + msg.key = entity->get_object_id_hash(); if (entity->has_own_name()) { msg.name = entity->get_name(); @@ -1149,7 +1149,7 @@ void APIConnection::try_send_camera_image_() { bool done = this->image_reader_->available() == to_send; CameraImageResponse msg; - msg.key = camera::Camera::instance()->get_entity_key(); + msg.key = camera::Camera::instance()->get_object_id_hash(); msg.set_data(this->image_reader_->peek_data_buffer(), to_send); msg.done = done; #ifdef USE_DEVICES diff --git a/esphome/components/infrared/infrared.cpp b/esphome/components/infrared/infrared.cpp index 288b1e5c407..9b97995a964 100644 --- a/esphome/components/infrared/infrared.cpp +++ b/esphome/components/infrared/infrared.cpp @@ -154,8 +154,12 @@ bool Infrared::on_receive(remote_base::RemoteReceiveData data) { // Forward received IR data to API server #if defined(USE_API) && defined(USE_IR_RF) if (api::global_api_server != nullptr) { - api::global_api_server->send_infrared_rf_receive_event(this->get_device_id_or_zero(), this->get_entity_key(), - &data.get_raw_data()); +#ifdef USE_DEVICES + uint32_t device_id = this->get_device_id(); +#else + uint32_t device_id = 0; +#endif + api::global_api_server->send_infrared_rf_receive_event(device_id, this->get_object_id_hash(), &data.get_raw_data()); } #endif return false; // Don't consume the event, allow other listeners to process it diff --git a/esphome/components/mqtt/__init__.py b/esphome/components/mqtt/__init__.py index 713969ab883..98ca23b60b1 100644 --- a/esphome/components/mqtt/__init__.py +++ b/esphome/components/mqtt/__init__.py @@ -63,7 +63,6 @@ from esphome.const import ( PlatformFramework, ) from esphome.core import CORE, CoroPriority, coroutine_with_priority -from esphome.core.entity_helpers import ObjectIdEntity, validate_no_object_id_conflicts from esphome.types import ConfigType DEPENDENCIES = ["network"] @@ -333,68 +332,6 @@ CONFIG_SCHEMA = cv.All( ) -# Platforms whose MQTT components subscribe to an object_id-derived command topic. -# Keep in sync with the platforms extending cv.MQTT_COMMAND_COMPONENT_SCHEMA, plus -# text, whose MQTT component subscribes a command topic that cannot be overridden. -_COMMAND_TOPIC_PLATFORMS = frozenset( - { - "alarm_control_panel", - "button", - "climate", - "cover", - "datetime", - "fan", - "light", - "lock", - "number", - "select", - "switch", - "text", - "update", - "valve", - } -) - - -# Platforms whose MQTT components derive extra sub-topics (position/command, -# mode/command, speed/command, ...) from the object_id, each with its own config -# key; custom state and command topics cannot exempt them from conflicting. -_SUB_TOPIC_PLATFORMS = frozenset({"climate", "cover", "fan", "valve"}) - - -def _topics_conflict(entities: list[ObjectIdEntity], config: ConfigType) -> bool: - """Check whether more than one entity actually uses an object_id-derived topic. - - An empty topic_prefix disables default topics entirely, custom state and - command topics avoid the default topics, and disabling discovery (globally - or per entity) avoids the discovery config topic. - """ - if config[CONF_TOPIC_PREFIX]: - platform = entities[0].platform - if platform in _SUB_TOPIC_PLATFORMS: - return True - if sum(CONF_STATE_TOPIC not in entity.config for entity in entities) > 1: - return True - if ( - platform in _COMMAND_TOPIC_PLATFORMS - and sum(CONF_COMMAND_TOPIC not in entity.config for entity in entities) > 1 - ): - return True - if not config[CONF_DISCOVERY]: - return False - discovery_entities = sum( - entity.config.get(CONF_DISCOVERY, True) for entity in entities - ) - return discovery_entities > 1 - - -FINAL_VALIDATE_SCHEMA = validate_no_object_id_conflicts( - "mqtt builds default topics and discovery topics from the entity object_id, " - "which is the name converted to ASCII", - conflict_filter=_topics_conflict, -) - - def exp_mqtt_message(config): if config is None: return cg.optional(cg.TemplateArguments(MQTTMessage)) diff --git a/esphome/components/prometheus/__init__.py b/esphome/components/prometheus/__init__.py index 0a69160fc11..cc1541ce804 100644 --- a/esphome/components/prometheus/__init__.py +++ b/esphome/components/prometheus/__init__.py @@ -3,7 +3,6 @@ from esphome.components import web_server_base from esphome.components.web_server_base import CONF_WEB_SERVER_BASE_ID import esphome.config_validation as cv from esphome.const import CONF_ID, CONF_INCLUDE_INTERNAL, CONF_NAME, CONF_RELABEL -from esphome.core.entity_helpers import validate_no_object_id_conflicts from esphome.cpp_types import EntityBase AUTO_LOAD = ["web_server_base"] @@ -36,11 +35,6 @@ CONFIG_SCHEMA = cv.Schema( }, ).extend(cv.COMPONENT_SCHEMA) -FINAL_VALIDATE_SCHEMA = validate_no_object_id_conflicts( - "prometheus builds metric labels from the entity object_id, " - "which is the name converted to ASCII" -) - async def to_code(config): paren = await cg.get_variable(config[CONF_WEB_SERVER_BASE_ID]) diff --git a/esphome/components/radio_frequency/radio_frequency.cpp b/esphome/components/radio_frequency/radio_frequency.cpp index fe6c6a9cb59..3e0a905737e 100644 --- a/esphome/components/radio_frequency/radio_frequency.cpp +++ b/esphome/components/radio_frequency/radio_frequency.cpp @@ -99,8 +99,12 @@ bool RadioFrequency::on_receive(remote_base::RemoteReceiveData data) { // Forward received RF data to API server #if defined(USE_API) && defined(USE_RADIO_FREQUENCY) if (api::global_api_server != nullptr) { - api::global_api_server->send_infrared_rf_receive_event(this->get_device_id_or_zero(), this->get_entity_key(), - &data.get_raw_data()); +#ifdef USE_DEVICES + uint32_t device_id = this->get_device_id(); +#else + uint32_t device_id = 0; +#endif + api::global_api_server->send_infrared_rf_receive_event(device_id, this->get_object_id_hash(), &data.get_raw_data()); } #endif return false; // Don't consume the event, allow other listeners to process it diff --git a/esphome/components/template/text/template_text.cpp b/esphome/components/template/text/template_text.cpp index ffe11cf229d..af134e6ed4a 100644 --- a/esphome/components/template/text/template_text.cpp +++ b/esphome/components/template/text/template_text.cpp @@ -20,14 +20,18 @@ void TemplateText::setup() { // Need std::string for pref_->setup() to fill from flash std::string value{this->initial_value_ != nullptr ? this->initial_value_ : ""}; - uint32_t extra = 0; - extra += this->traits.get_min_length() << 2; - extra += this->traits.get_max_length() << 4; - extra += fnv1_hash(this->traits.get_pattern_c_str()) << 6; - // TextSaver::setup() picks the key for the platform and migrates old data once - uint32_t key = this->preference_key_base_() + extra; - uint32_t old_key = this->old_preference_key_base_() + extra; - this->pref_->setup(key, old_key, value); + // For future hash migration: use migrate_entity_preference_() with: + // old_key = get_preference_hash() + extra + // new_key = get_preference_hash_v2() + extra + // See: https://github.com/esphome/backlog/issues/85 +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wdeprecated-declarations" + uint32_t key = this->get_preference_hash(); +#pragma GCC diagnostic pop + key += this->traits.get_min_length() << 2; + key += this->traits.get_max_length() << 4; + key += fnv1_hash(this->traits.get_pattern_c_str()) << 6; + this->pref_->setup(key, value); if (!value.empty()) this->publish_state(value); } diff --git a/esphome/components/template/text/template_text.h b/esphome/components/template/text/template_text.h index beeea4396a7..229a61d9b8e 100644 --- a/esphome/components/template/text/template_text.h +++ b/esphome/components/template/text/template_text.h @@ -14,9 +14,7 @@ class TemplateTextSaverBase { public: virtual bool save(const std::string &value) { return true; } - /// old_id is the pre-2026.8.0 preference key; data stored under it is moved to id once. - /// See: https://github.com/esphome/backlog/issues/85 - virtual void setup(uint32_t id, uint32_t old_id, std::string &value) {} + virtual void setup(uint32_t id, std::string &value) {} protected: ESPPreferenceObject pref_; @@ -47,16 +45,11 @@ template class TextSaver : public TemplateTextSaverBase { // Make the preference object. Fill the provided location with the saved data // If it is available, else leave it alone - void setup(uint32_t id, uint32_t old_id, std::string &value) override { - char temp[SZ + 1]; -#ifdef USE_PREFERENCE_KEY_LOOKUP + void setup(uint32_t id, std::string &value) override { this->pref_ = global_preferences->make_preference(id); - bool hasdata = migrate_preference(this->pref_, reinterpret_cast(temp), SZ + 1, old_id, id); -#else - // Slot-based backends keep the old key; it is only a validity tag on a positional slot - this->pref_ = global_preferences->make_preference(old_id); + + char temp[SZ + 1]; bool hasdata = this->pref_.load(&temp); -#endif if (hasdata) { size_t len = static_cast(temp[0]); diff --git a/esphome/core/application.h b/esphome/core/application.h index a18a6b31c84..a12cdc4ac88 100644 --- a/esphome/core/application.h +++ b/esphome/core/application.h @@ -120,8 +120,8 @@ class Application { // NOLINTBEGIN(bugprone-macro-parentheses) #define ENTITY_TYPE_(type, singular, plural, count, upper) \ void register_##singular(type *obj) { this->plural##_.push_back(obj); } \ - void register_##singular(type *obj, const char *name, uint32_t entity_key, uint32_t entity_fields) { \ - obj->configure_entity_(name, entity_key, entity_fields); \ + void register_##singular(type *obj, const char *name, uint32_t object_id_hash, uint32_t entity_fields) { \ + obj->configure_entity_(name, object_id_hash, entity_fields); \ this->plural##_.push_back(obj); \ } #define ENTITY_CONTROLLER_TYPE_(type, singular, plural, count, upper, callback) \ @@ -329,7 +329,7 @@ class Application { #define GET_ENTITY_METHOD(entity_type, entity_name, entities_member) \ entity_type *get_##entity_name##_by_key(uint32_t key, uint32_t device_id, bool include_internal = false) { \ for (auto *obj : this->entities_member##_) { \ - if (obj->get_entity_key() == key && obj->get_device_id() == device_id && \ + if (obj->get_object_id_hash() == key && obj->get_device_id() == device_id && \ (include_internal || !obj->is_internal())) \ return obj; \ } \ @@ -340,7 +340,7 @@ class Application { #define GET_ENTITY_METHOD(entity_type, entity_name, entities_member) \ entity_type *get_##entity_name##_by_key(uint32_t key, bool include_internal = false) { \ for (auto *obj : this->entities_member##_) { \ - if (obj->get_entity_key() == key && (include_internal || !obj->is_internal())) \ + if (obj->get_object_id_hash() == key && (include_internal || !obj->is_internal())) \ return obj; \ } \ return nullptr; \ diff --git a/esphome/core/entity_base.cpp b/esphome/core/entity_base.cpp index 328de053022..fc6ac503b5c 100644 --- a/esphome/core/entity_base.cpp +++ b/esphome/core/entity_base.cpp @@ -8,7 +8,7 @@ namespace esphome { static const char *const TAG = "entity_base"; -void EntityBase::configure_entity_(const char *name, uint32_t entity_key, uint32_t entity_fields) { +void EntityBase::configure_entity_(const char *name, uint32_t object_id_hash, uint32_t entity_fields) { this->name_ = StringRef(name); if (this->name_.empty()) { #ifdef USE_DEVICES @@ -30,15 +30,15 @@ void EntityBase::configure_entity_(const char *name, uint32_t entity_key, uint32 } } this->flags_.has_own_name = false; - // Dynamic name - must calculate key at runtime - this->calc_entity_key_(); + // Dynamic name - must calculate hash at runtime + this->calc_object_id_(); } else { this->flags_.has_own_name = true; - // Static name - use pre-computed key if provided - if (entity_key != 0) { - this->entity_key_ = entity_key; + // Static name - use pre-computed hash if provided + if (object_id_hash != 0) { + this->object_id_hash_ = object_id_hash; } else { - this->calc_entity_key_(); + this->calc_object_id_(); } } // Unpack entity string table indices and flags from entity_fields. @@ -147,15 +147,9 @@ std::string EntityBase::get_icon() const { } #endif // !USE_ESP8266 -// Calculate the entity key directly from the raw name (no transformations) -void EntityBase::calc_entity_key_() { this->entity_key_ = fnv1_hash_bytes(this->name_.c_str(), this->name_.size()); } - -// Reconstruct the OLD (pre-2026.8.0) object_id-based hash for preference key compatibility. -// Named entities historically used the hash pre-computed by Python code generation, which -// sanitized per UTF-8 code point; entities without their own name computed the hash at -// runtime per byte. See https://github.com/esphome/backlog/issues/85 -uint32_t EntityBase::calc_old_object_id_hash_() const { - return fnv1_hash_object_id(this->name_.c_str(), this->name_.size(), this->flags_.has_own_name); +// Calculate Object ID Hash directly from name using snake_case + sanitize +void EntityBase::calc_object_id_() { + this->object_id_hash_ = fnv1_hash_object_id(this->name_.c_str(), this->name_.size()); } size_t EntityBase::write_object_id_to(char *buf, size_t buf_size) const { @@ -173,22 +167,16 @@ StringRef EntityBase::get_object_id_to(std::span buf) c } ESPPreferenceObject EntityBase::make_entity_preference_(size_t size, uint32_t version) { - // The old key hashed the sanitized object_id, so multiple entity names could collide on - // one key and overwrite each other's stored preferences; the new key hashes the raw name. - // See: https://github.com/esphome/backlog/issues/85 - uint32_t old_key = this->old_preference_key_base_() ^ version; -#ifdef USE_PREFERENCE_KEY_LOOKUP - uint32_t new_key = this->preference_key_base_() ^ version; - auto pref = global_preferences->make_preference(size, new_key); - // All in-tree entity preferences fit the stack buffer, so migration never hits the heap - SmallBufferWithHeapFallback<64> buffer(size); - migrate_preference(pref, buffer.get(), size, old_key, new_key); - return pref; -#else - // Slot-based backends keep the old key: it is only a validity tag on a positional slot, - // so collisions cannot corrupt data there and keeping it preserves stored state. - return global_preferences->make_preference(size, old_key); -#endif + // The key hashes the sanitized object_id, so multiple entity names can collide on one + // key and overwrite each other's stored preferences ("Living Room" and "living_room", + // or two UTF-8 names that both sanitize to underscores). Keys hashed from the raw name + // fix this, but they change the entity key API clients track, which the Home Assistant + // esphome integration cannot handle yet. See: https://github.com/esphome/backlog/issues/85 +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wdeprecated-declarations" + uint32_t key = this->get_preference_hash() ^ version; +#pragma GCC diagnostic pop + return global_preferences->make_preference(size, key); } #ifdef USE_ENTITY_ICON diff --git a/esphome/core/entity_base.h b/esphome/core/entity_base.h index 7f8e5f2630f..5f2e173d8d2 100644 --- a/esphome/core/entity_base.h +++ b/esphome/core/entity_base.h @@ -73,17 +73,8 @@ class EntityBase { // Get whether this Entity has its own name or it should use the device friendly_name. bool has_own_name() const { return this->flags_.has_own_name; } - // Get the unique key of this Entity: FNV-1 hash of the raw entity name. - // This is the key sent to API clients and used to route entity state. - uint32_t get_entity_key() const { return this->entity_key_; } - - /// Returns the LEGACY object_id hash, unchanged from previous releases, so existing - /// callers keep getting stable values (for example preference keys). This is no longer - /// the key sent to API clients; that is get_entity_key(). - ESPDEPRECATED("Use get_entity_key() for the entity key sent to API clients, or " - "make_entity_preference() for preference storage. Will be removed in 2027.1.0.", - "2026.8.0") - uint32_t get_object_id_hash() const { return this->calc_old_object_id_hash_(); } + // Get the unique Object ID of this Entity + uint32_t get_object_id_hash() const { return this->object_id_hash_; } /// Get object_id with zero heap allocation /// For static case: returns StringRef to internal storage (buffer unused) @@ -190,23 +181,39 @@ class EntityBase { // Set has_state - for components that need to manually set this void set_has_state(bool state) { this->flags_.has_state = state; } - /// Get this entity's device id, or 0 when devices are not compiled in (main device). - uint32_t get_device_id_or_zero() const { -#ifdef USE_DEVICES - return this->get_device_id(); -#else - return 0; -#endif - } - - /// Get the LEGACY preference key: FNV-1 hash of the sanitized object_id, XOR device_id. - /// Intentionally keeps the old algorithm so external callers that store preferences under - /// this key keep stable keys; make_entity_preference() migrates to the new raw-name key, - /// this method never will. + /** + * @brief Get a unique hash for storing preferences/settings for this entity. + * + * This method returns a hash that uniquely identifies the entity for the purpose of + * storing preferences (such as calibration, state, etc.). Unlike get_object_id_hash(), + * this hash also incorporates the device_id (if devices are enabled), ensuring uniqueness + * across multiple devices that may have entities with the same object_id. + * + * Use this method when storing or retrieving preferences/settings that should be unique + * per device-entity pair. Use get_object_id_hash() when you need a hash that identifies + * the entity regardless of the device it belongs to. + * + * For backward compatibility, if device_id is 0 (the main device), the hash is unchanged + * from previous versions, so existing single-device configurations will continue to work. + * + * @return uint32_t The unique hash for preferences, including device_id if available. + * @deprecated Use make_entity_preference() instead, or preferences won't be migrated. + * See https://github.com/esphome/backlog/issues/85 + */ ESPDEPRECATED("Use make_entity_preference() instead, or preferences won't be migrated. " "See https://github.com/esphome/backlog/issues/85. Will be removed in 2027.1.0.", - "2026.8.0") - uint32_t get_preference_hash() { return this->old_preference_key_base_(); } + "2026.7.0") + uint32_t get_preference_hash() { +#ifdef USE_DEVICES + // Combine object_id_hash with device_id to ensure uniqueness across devices + // Note: device_id is 0 for the main device, so XORing with 0 preserves the original hash + // This ensures backward compatibility for existing single-device configurations + return this->get_object_id_hash() ^ this->get_device_id(); +#else + // Without devices, just use object_id_hash as before + return this->get_object_id_hash(); +#endif + } /// Create a preference object for storing this entity's state/settings. /// @tparam T The type of data to store (must be trivially copyable) @@ -223,9 +230,9 @@ class EntityBase { // before push_back, so codegen can emit a single combined call per entity. friend class Application; - /// Combined entity setup from codegen: set name, entity key, entity string indices, and flags. + /// Combined entity setup from codegen: set name, object_id hash, entity string indices, and flags. /// Bit layout of entity_fields is defined by the ENTITY_FIELD_*_SHIFT constants above. - void configure_entity_(const char *name, uint32_t entity_key, uint32_t entity_fields); + void configure_entity_(const char *name, uint32_t object_id_hash, uint32_t entity_fields); #ifdef USE_DEVICES // Codegen-only setter — only accessible from setup() via friend declaration. @@ -233,24 +240,13 @@ class EntityBase { #endif /// Non-template helper for make_entity_preference() to avoid code bloat. - /// Migrates preferences from the old sanitized-object_id key to the raw-name key - /// on key-lookup platforms. See: https://github.com/esphome/backlog/issues/85 + /// When the preference hash algorithm changes, migration logic goes here. ESPPreferenceObject make_entity_preference_(size_t size, uint32_t version); - void calc_entity_key_(); - - /// Reconstruct the OLD (pre-2026.8.0) sanitized-object_id hash for preference keys. - uint32_t calc_old_object_id_hash_() const; - - /// Preference key base for this entity: raw-name entity key XOR device_id. - uint32_t preference_key_base_() const { return this->entity_key_ ^ this->get_device_id_or_zero(); } - - /// Legacy preference key base: sanitized-object_id hash XOR device_id. - /// Note: device_id is 0 for the main device, so XORing with 0 preserves the original hash. - uint32_t old_preference_key_base_() const { return this->calc_old_object_id_hash_() ^ this->get_device_id_or_zero(); } + void calc_object_id_(); StringRef name_; - uint32_t entity_key_{}; + uint32_t object_id_hash_{}; #ifdef USE_DEVICES Device *device_{}; #endif diff --git a/esphome/core/entity_helpers.py b/esphome/core/entity_helpers.py index 5060e32a2d8..54e2551cb4e 100644 --- a/esphome/core/entity_helpers.py +++ b/esphome/core/entity_helpers.py @@ -25,86 +25,25 @@ from esphome.core.config import ( from esphome.cpp_generator import MockObj, RawStatement, add, get_variable from esphome.cpp_types import App import esphome.final_validate as fv -from esphome.helpers import cpp_string_escape, fnv1_hash_name, sanitize, snake_case +from esphome.helpers import ( + cpp_string_escape, + fnv1_hash, + fnv1_hash_object_id, + sanitize, + snake_case, +) from esphome.types import ConfigType, EntityMetadata _LOGGER = logging.getLogger(__name__) DOMAIN = "entity_string_pool" -_OBJECT_ID_DOMAIN = "entity_object_ids" - - -@dataclass -class ObjectIdEntity: - """An entity tracked by the sanitized object_id its name resolves to.""" - - name: str - platform: str - config: ConfigType - - -def _get_object_id_registry() -> dict[tuple[str, str, str], list[ObjectIdEntity]]: - """(device_id, platform, sanitized object_id) -> entities resolving to it.""" - return CORE.data.setdefault(_OBJECT_ID_DOMAIN, {}) - - -def validate_no_object_id_conflicts( - reason: str, - conflict_filter: Callable[[list[ObjectIdEntity], ConfigType], bool] | None = None, -) -> Callable[[ConfigType], ConfigType]: - """Create a final-validate step that rejects entities with colliding object_ids. - - Entity keys are hashed from the raw name, so names that only differ in characters - lost during sanitizing (for example two UTF-8 names) validate fine in general. - Components that still address entities by the sanitized object_id string must - reject those configs until they are migrated to raw names. - - Args: - reason: One sentence stating what the component builds from the object_id, - e.g. "mqtt builds default topics from the entity object_id" - conflict_filter: Optional predicate receiving the colliding entities and the - component config; return False when the component is not affected - - Returns: - A validator function for use as (or within) FINAL_VALIDATE_SCHEMA - """ - - def validator(config: ConfigType) -> ConfigType: - # Skip in testing_mode, which is used for grouped component testing - if CORE.testing_mode: - return config - conflicts = { - key: entities - for key, entities in _get_object_id_registry().items() - if len(entities) > 1 - and (conflict_filter is None or conflict_filter(entities, config)) - } - if not conflicts: - return config - lines = [f"{reason}, so these entities would conflict:"] - lines.extend( - f" - {platform} entities " - + ", ".join(f"'{e.name}'" for e in entities) - + (f" on device '{device_id}'" if device_id else "") - + f" share the object_id '{object_id}'" - for (device_id, platform, object_id), entities in conflicts.items() - ) - lines.append( - "To fix: Add unique ASCII characters (e.g., '1', '2', or 'A', 'B') " - "to distinguish the names" - ) - raise cv.Invalid("\n".join(lines)) - - return validator - - # Private config keys for storing registered string indices _KEY_DC_IDX = "_entity_dc_idx" _KEY_UOM_IDX = "_entity_uom_idx" _KEY_ICON_IDX = "_entity_icon_idx" _KEY_ENTITY_NAME = "_entity_name" -_KEY_ENTITY_KEY = "_entity_key" +_KEY_OBJECT_ID_HASH = "_entity_object_id_hash" # Bit layout for entity_fields in configure_entity_(). # Keep in sync with ENTITY_FIELD_*_SHIFT constants in esphome/core/entity_base.h @@ -367,7 +306,7 @@ def finalize_entity_strings(var: MockObj, config: ConfigType) -> None: standalone ``var->configure_entity_(name, hash, packed)``. """ entity_name = config[_KEY_ENTITY_NAME] - entity_key = config[_KEY_ENTITY_KEY] + object_id_hash = config[_KEY_OBJECT_ID_HASH] dc_idx = config.get(_KEY_DC_IDX, 0) uom_idx = config.get(_KEY_UOM_IDX, 0) icon_idx = config.get(_KEY_ICON_IDX, 0) @@ -387,30 +326,57 @@ def finalize_entity_strings(var: MockObj, config: ConfigType) -> None: register_method = config.get(_KEY_REGISTER_METHOD) if register_method is not None: expr = getattr(App, f"register_{register_method}")( - var, entity_name, entity_key, packed + var, entity_name, object_id_hash, packed ) else: - expr = var.configure_entity_(entity_name, entity_key, packed) + expr = var.configure_entity_(entity_name, object_id_hash, packed) if comment: add(RawStatement(f"{expr}; // {comment}")) else: add(expr) -def get_base_entity_name( +def get_base_entity_object_id( name: str, friendly_name: str | None, device_name: str | None = None ) -> str: - """Return the base name whose hash becomes this entity's key on the device. + """Calculate the base object ID for an entity that will be set via set_object_id(). - Follows the name selection in C++ EntityBase::configure_entity_() (entity_base.cpp): - entity name, then sub-device name, then friendly name, then the device name. + This function calculates what object_id_c_str_ should be set to in C++. - This is a config-time approximation for duplicate checking: when - name_add_mac_suffix is enabled the device appends the MAC suffix at runtime, - which is unknown here and identical for every entity on the device, so - ignoring it cannot change whether two entities collide with each other. + The C++ EntityBase::write_object_id_to() (entity_base.cpp) works as: + - If !has_own_name && is_name_add_mac_suffix_enabled(): + return str_sanitize(str_snake_case(App.get_friendly_name())) // Dynamic + - Else: + return object_id_c_str_ ?? "" // What we set via set_object_id() + + Since we're calculating what to pass to set_object_id(), we always need to + generate the object_id the same way, regardless of name_add_mac_suffix setting. + + Args: + name: The entity name (empty string if no name) + friendly_name: The friendly name from CORE.friendly_name + device_name: The device name if entity is on a sub-device + + Returns: + The base object ID to use for duplicate checking and to pass to set_object_id() """ - return name or device_name or friendly_name or CORE.name + + if name: + # Entity has its own name (has_own_name will be true) + base_str = name + elif device_name: + # Entity has empty name and is on a sub-device + # C++ EntityBase::set_name() uses device->get_name() when device is set + base_str = device_name + elif friendly_name: + # Entity has empty name (has_own_name will be false) + # C++ uses App.get_friendly_name() which returns friendly_name or device name + base_str = friendly_name + else: + # Fallback to device name + base_str = CORE.name + + return sanitize(snake_case(base_str)) def setup_entity(var_or_platform, config=None, platform=None): @@ -469,15 +435,15 @@ async def _setup_entity_impl(var: MockObj, config: ConfigType, platform: str) -> device: MockObj = await get_variable(device_id_obj) add(var.set_device_(device)) - # Pre-compute entity name and entity key for configure_entity_() + # Pre-compute entity name and object_id hash for configure_entity_() # which is emitted later by finalize_entity_strings(). - # For named entities: pre-compute the key from the raw entity name - # For empty-name entities: pass 0, C++ calculates the key at runtime from - # device name, friendly_name, or app name + # For named entities: pre-compute hash from entity name + # For empty-name entities: pass 0, C++ calculates hash at runtime from + # device name, friendly_name, or app name (bug-for-bug compatibility) entity_name = config[CONF_NAME] - entity_key = fnv1_hash_name(entity_name) if entity_name else 0 + object_id_hash = fnv1_hash_object_id(entity_name) if entity_name else 0 config[_KEY_ENTITY_NAME] = entity_name - config[_KEY_ENTITY_KEY] = entity_key + config[_KEY_OBJECT_ID_HASH] = object_id_hash # Store flags for packing into configure_entity_() config[_KEY_DISABLED_BY_DEFAULT] = int(config[CONF_DISABLED_BY_DEFAULT]) if CONF_INTERNAL in config: @@ -590,13 +556,16 @@ def entity_duplicate_validator(platform: str) -> Callable[[ConfigType], ConfigTy # Use the device ID string directly for uniqueness device_id = device_id_obj.id - # Hash the same raw name the device hashes into the entity key at runtime. - # This handles empty names correctly by using device/friendly names. - base_name = get_base_entity_name(entity_name, CORE.friendly_name, device_name) - name_hash = fnv1_hash_name(base_name) + # Calculate what object_id will actually be used + # This handles empty names correctly by using device/friendly names + name_key = get_base_entity_object_id( + entity_name, CORE.friendly_name, device_name + ) - # Check for duplicates: two entities on the same device and platform must not - # share an entity key, since the key is what routes state to API clients + # Check for duplicates by the FNV-1 hash of the object_id, which is the entity + # key that routes state to API clients. This rejects names that sanitize to the + # same object_id, and also two different object_ids whose 32-bit hashes collide. + name_hash = fnv1_hash(name_key) unique_key = (device_id, platform, name_hash) if unique_key in CORE.unique_ids: # Get the existing entity metadata @@ -621,14 +590,26 @@ def entity_duplicate_validator(platform: str) -> Callable[[ConfigType], ConfigTy if existing_component != "unknown": conflict_msg += f" from component '{existing_component}'" - # Different names can only clash here through a genuine hash collision + # Distinguish names that sanitize to the same object_id from a genuine + # 32-bit hash collision between two different object_ids collision_msg = "" if entity_name != existing_name: - collision_msg = ( - f"\n The names '{entity_name}' and '{existing_name}' produce the" - f"\n same entity key hash ({name_hash:#010x})." - "\n To fix: Rename one of the entities" + existing_object_id = get_base_entity_object_id( + existing_name, CORE.friendly_name, existing_device or None ) + if existing_object_id == name_key: + collision_msg = ( + f"\n Original names: '{entity_name}' and '{existing_name}'" + f"\n Both convert to ASCII ID: '{name_key}'" + "\n To fix: Add unique ASCII characters (e.g., '1', '2', or 'A', 'B')" + "\n to distinguish them" + ) + else: + collision_msg = ( + f"\n The object_ids '{name_key}' and '{existing_object_id}'" + f"\n produce the same entity key hash ({name_hash:#010x})." + "\n To fix: Rename one of the entities" + ) # Skip duplicate entity name validation when testing_mode is enabled # This flag is used for grouped component testing @@ -640,19 +621,6 @@ def entity_duplicate_validator(platform: str) -> Callable[[ConfigType], ConfigTy f"{collision_msg}" ) - # Components that still address entities by the sanitized object_id reject - # colliding names in final validation via validate_no_object_id_conflicts(), - # so track every entity by the object_id its name resolves to. Scoped per - # device and platform to match the strictness configs had before entity keys - # moved to raw names: same-named entities on different sub-devices were - # already accepted then, internal entities were already skipped (above), and - # overlaps between platforms that share an MQTT component type (sensor and - # text_sensor both publish under "sensor") were already possible. - object_id = sanitize(snake_case(base_name)) - _get_object_id_registry().setdefault( - (device_id, platform, object_id), [] - ).append(ObjectIdEntity(base_name, platform, config)) - # Store metadata about this entity entity_metadata: EntityMetadata = { "name": entity_name, diff --git a/esphome/core/helpers.h b/esphome/core/helpers.h index d883ce146e9..994fa2c26a4 100644 --- a/esphome/core/helpers.h +++ b/esphome/core/helpers.h @@ -809,19 +809,6 @@ constexpr uint32_t FNV1_OFFSET_BASIS = 2166136261UL; /// FNV-1 32-bit prime constexpr uint32_t FNV1_PRIME = 16777619UL; -/// Calculate a FNV-1 hash over raw bytes with an explicit length. Unlike fnv1_hash(const char *), -/// each byte is hashed as an unsigned value, so results are platform-independent for bytes >= 0x80. -/// IMPORTANT: Must match Python fnv1_hash_name() in esphome/helpers.py, which hashes the UTF-8 -/// encoded bytes of the name. Used to compute entity keys from raw names. -inline uint32_t fnv1_hash_bytes(const char *str, size_t len) { - uint32_t hash = FNV1_OFFSET_BASIS; - for (size_t i = 0; i < len; i++) { - hash *= FNV1_PRIME; - hash ^= static_cast(str[i]); - } - return hash; -} - /// Extend a FNV-1 hash with an integer (hashes each byte). template constexpr uint32_t fnv1_hash_extend(uint32_t hash, T value) { using UnsignedT = std::make_unsigned_t; @@ -1026,20 +1013,12 @@ template inline char *str_sanitize_to(char (&buffer)[N], const char *s // str_sanitize moved to alloc_helpers.h - remove this comment before 2026.11.0 /// Calculate FNV-1 hash of a string while applying snake_case + sanitize transformations. -/// This is the LEGACY entity hash, kept only to reconstruct preference keys that existing -/// devices already have stored; see https://github.com/esphome/backlog/issues/85. -/// With per_code_point set, UTF-8 continuation bytes are skipped so each multi-byte character -/// contributes one underscore — this matches Python fnv1_hash_object_id() in esphome/helpers.py, -/// which produced the hash for named entities. The per-byte form (default) matches the old -/// runtime hash for entities without their own name. Do not change either behavior. -/// Known limitation: Python's lower() is Unicode aware, so the rare code points it maps to a -/// different number of characters or to ASCII (e.g. 'İ', the Kelvin sign) reconstruct wrong; -/// such names skip migration once and fall back to their defaults. -inline uint32_t fnv1_hash_object_id(const char *str, size_t len, bool per_code_point = false) { +/// This computes object_id hashes directly from names without creating an intermediate buffer. +/// IMPORTANT: Must match Python fnv1_hash_object_id() in esphome/helpers.py. +/// If you modify this function, update the Python version and tests in both places. +inline uint32_t fnv1_hash_object_id(const char *str, size_t len) { uint32_t hash = FNV1_OFFSET_BASIS; for (size_t i = 0; i < len; i++) { - if (per_code_point && (static_cast(str[i]) & 0xC0) == 0x80) - continue; // UTF-8 continuation byte, already counted via its lead byte hash *= FNV1_PRIME; // Apply snake_case (space->underscore, uppercase->lowercase) then sanitize hash ^= static_cast(to_sanitized_char(to_snake_case_char(str[i]))); diff --git a/esphome/core/preference_backend.h b/esphome/core/preference_backend.h index 0622376fcae..5df0804bdd1 100644 --- a/esphome/core/preference_backend.h +++ b/esphome/core/preference_backend.h @@ -24,9 +24,10 @@ #endif // Key-lookup preference backends find stored data by key; their platforms add the -// USE_PREFERENCE_KEY_LOOKUP define from Python codegen, which enables preference key -// migration. Slot-based backends (ESP8266, RP2040) instead allocate a storage slot for -// every make_preference() call and use the key only as a validity tag on that slot; +// USE_PREFERENCE_KEY_LOOKUP define from Python codegen, which enables one-shot reads +// of stored data by key (the primitive preference key migrations need). Slot-based +// backends (ESP8266, RP2040) instead allocate a storage slot for every +// make_preference() call and use the key only as a validity tag on that slot; // migration is not possible there, and key collisions cannot corrupt data. namespace esphome { @@ -104,10 +105,9 @@ concept PreferencesContract = requires(T prefs, size_t len, uint32_t type, bool }; // Key-lookup platforms additionally provide load_from_key(), a one-shot read -// of a stored preference by key that migrate_preference() relies on; see the -// key-lookup note at the top of this file. Not part of PreferencesContract, -// so it is asserted in preferences.h only where USE_PREFERENCE_KEY_LOOKUP -// is set. +// of a stored preference by key; see the key-lookup note at the top of this +// file. Not part of PreferencesContract, so it is asserted in preferences.h +// only where USE_PREFERENCE_KEY_LOOKUP is set. template concept PreferencesKeyLookupContract = requires(T prefs, uint32_t type, uint8_t *data, size_t len) { { prefs.load_from_key(type, data, len) } -> std::same_as; diff --git a/esphome/core/preferences.cpp b/esphome/core/preferences.cpp deleted file mode 100644 index 8508647255a..00000000000 --- a/esphome/core/preferences.cpp +++ /dev/null @@ -1,25 +0,0 @@ -#include "esphome/core/preferences.h" -#include "esphome/core/log.h" -#include - -namespace esphome { - -#ifdef USE_PREFERENCE_KEY_LOOKUP -static const char *const TAG = "preferences"; - -bool migrate_preference(ESPPreferenceObject &new_pref, uint8_t *scratch, size_t size, uint32_t old_key, - uint32_t new_key) { - if (new_pref.load(scratch, size)) - return true; // Current data present - never overwrite newer data with the old copy - // One-shot read by key: no backend is allocated for the old key, so boots with - // nothing to migrate (for example fresh installs) cost no heap - if (old_key == new_key || !global_preferences->load_from_key(old_key, scratch, size)) - return false; // No data stored under the old key, nothing to migrate - if (!new_pref.save(scratch, size)) { - ESP_LOGW(TAG, "Pref migration %" PRIx32 " -> %" PRIx32 " failed", old_key, new_key); - } - return true; -} -#endif // USE_PREFERENCE_KEY_LOOKUP - -} // namespace esphome diff --git a/esphome/core/preferences.h b/esphome/core/preferences.h index cfeddebda75..ed23dfae560 100644 --- a/esphome/core/preferences.h +++ b/esphome/core/preferences.h @@ -56,17 +56,5 @@ namespace esphome { static_assert(PreferencesKeyLookupContract, "This platform emits USE_PREFERENCE_KEY_LOOKUP but its preferences manager does not provide " "load_from_key() (esphome/core/preference_backend.h)"); - -/// Copy preference data stored under old_key into new_pref (created for new_key) if the keys -/// differ and new_pref has no data yet. scratch must hold at least size bytes. -/// Returns true when scratch holds the entity's current data (loaded or just migrated). -/// The old entry is intentionally left in place so a firmware downgrade still finds its data. -/// If saving under the new key fails, callers that consume scratch (like TextSaver) still get -/// valid data for this boot, callers that reload from the preference fall back to their -/// defaults, and the migration simply runs again on the next boot. -/// Only available on key-lookup preference backends; slot-based backends keep their old -/// keys instead. See: https://github.com/esphome/backlog/issues/85 -bool migrate_preference(ESPPreferenceObject &new_pref, uint8_t *scratch, size_t size, uint32_t old_key, - uint32_t new_key); } // namespace esphome #endif // USE_PREFERENCE_KEY_LOOKUP diff --git a/esphome/helpers.py b/esphome/helpers.py index 27311091649..9b2a461ccd7 100644 --- a/esphome/helpers.py +++ b/esphome/helpers.py @@ -91,13 +91,8 @@ def fnv1a_32bit_hash(string: str) -> int: def fnv1_hash_object_id(name: str) -> int: """Compute FNV-1 hash of name with snake_case + sanitize transformations. - IMPORTANT: Must produce same result as C++ fnv1_hash_object_id() in helpers.h - with per_code_point set. This is the OLD entity hash; it computes preference - keys that existing devices already have stored (see - https://github.com/esphome/backlog/issues/85) and is also still used for live - keys derived from config IDs (see the motion component's calibration key). - Note: lower() here is Unicode aware while the C++ reconstruction is not; see - the known limitation note on the C++ function. + IMPORTANT: Must produce same result as C++ fnv1_hash_object_id() in helpers.h. + If you modify this function, update the C++ version and tests in both places. """ return fnv1_hash(sanitize(snake_case(name))) @@ -105,9 +100,9 @@ def fnv1_hash_object_id(name: str) -> int: def fnv1_hash_name(name: str) -> int: """Compute FNV-1 hash of the raw entity name (UTF-8 bytes, no transformations). - IMPORTANT: Must produce same result as C++ fnv1_hash_bytes() in helpers.h, - which hashes the name bytes as stored on the device. - Used for pre-computing entity keys at code generation time. + 2026.8 beta firmware stored preferences under keys derived from this hash; + a future key migration must reconstruct those keys to recover that data + (see https://github.com/esphome/backlog/issues/85). """ return _fnv1_hash(name.encode("utf-8")) diff --git a/tests/integration/entity_utils.py b/tests/integration/entity_utils.py index 95f6a0321ee..7596983ee23 100644 --- a/tests/integration/entity_utils.py +++ b/tests/integration/entity_utils.py @@ -8,7 +8,7 @@ from __future__ import annotations from typing import TYPE_CHECKING -from esphome.helpers import fnv1_hash_name, sanitize, snake_case +from esphome.helpers import fnv1_hash_object_id, sanitize, snake_case if TYPE_CHECKING: from aioesphomeapi import DeviceInfo, EntityInfo @@ -25,16 +25,15 @@ def infer_name_add_mac_suffix(device_info: DeviceInfo) -> bool: return device_info.name.endswith(f"-{mac_suffix}") -def _resolve_entity_name( +def _get_name_for_object_id( entity: EntityInfo, device_info: DeviceInfo, device_id_to_name: dict[int, str], ) -> str: - """Resolve the effective name for an entity. + """Get the name used for object_id computation. This is the algorithm that aioesphomeapi will use to determine which - name to use for computing object_id client-side from API data; the same - name is what the device hashes into the entity key. + name to use for computing object_id client-side from API data. Args: entity: The entity to get name for @@ -73,27 +72,27 @@ def compute_entity_object_id( Returns: The computed object_id string """ - name = _resolve_entity_name(entity, device_info, device_id_to_name) - return compute_object_id(name) + name_for_id = _get_name_for_object_id(entity, device_info, device_id_to_name) + return compute_object_id(name_for_id) -def compute_entity_key( +def compute_entity_hash( entity: EntityInfo, device_info: DeviceInfo, device_id_to_name: dict[int, str], ) -> int: - """Compute expected entity key for an entity. + """Compute expected object_id hash for an entity. Args: - entity: The entity to compute the key for + entity: The entity to compute hash for device_info: Device info from the API device_id_to_name: Mapping of device_id to device name for sub-devices Returns: - The computed FNV-1 hash of the raw name + The computed FNV-1 hash """ - name = _resolve_entity_name(entity, device_info, device_id_to_name) - return fnv1_hash_name(name) + name_for_id = _get_name_for_object_id(entity, device_info, device_id_to_name) + return fnv1_hash_object_id(name_for_id) def verify_entity_object_id( @@ -119,7 +118,7 @@ def verify_entity_object_id( f"expected '{expected_object_id}', got '{entity.object_id}'" ) - expected_hash = compute_entity_key(entity, device_info, device_id_to_name) + expected_hash = compute_entity_hash(entity, device_info, device_id_to_name) assert entity.key == expected_hash, ( f"hash mismatch for entity '{entity.name}': " f"expected {expected_hash:#x}, got {entity.key:#x}" diff --git a/tests/integration/fixtures/fnv1_hash_object_id.yaml b/tests/integration/fixtures/fnv1_hash_object_id.yaml index d4511bb8c6f..2097b2fbf9c 100644 --- a/tests/integration/fixtures/fnv1_hash_object_id.yaml +++ b/tests/integration/fixtures/fnv1_hash_object_id.yaml @@ -71,38 +71,6 @@ esphome: ESP_LOGE("FNV1_OID", "empty FAILED: 0x%08x != 0x811c9dc5", hash_empty); } - // Raw name hash: matches Python fnv1_hash_name("My Sensor Name") - uint32_t hash_raw = esphome::fnv1_hash_bytes("My Sensor Name", 14); - if (hash_raw == 0x8cec6fb0) { - ESP_LOGI("FNV1_OID", "raw PASSED"); - } else { - ESP_LOGE("FNV1_OID", "raw FAILED: 0x%08x != 0x8cec6fb0", hash_raw); - } - - // Raw name hash over UTF-8 bytes: matches Python fnv1_hash_name("Température") - uint32_t hash_raw_utf8 = esphome::fnv1_hash_bytes("Temp\xc3\xa9rature", 12); - if (hash_raw_utf8 == 0x531a74aa) { - ESP_LOGI("FNV1_OID", "raw_utf8 PASSED"); - } else { - ESP_LOGE("FNV1_OID", "raw_utf8 FAILED: 0x%08x != 0x531a74aa", hash_raw_utf8); - } - - // Old-key UTF-8 variant: matches Python fnv1_hash_object_id("Température") - uint32_t hash_old_utf8 = esphome::fnv1_hash_object_id("Temp\xc3\xa9rature", 12, true); - if (hash_old_utf8 == 0x965698f3) { - ESP_LOGI("FNV1_OID", "old_utf8 PASSED"); - } else { - ESP_LOGE("FNV1_OID", "old_utf8 FAILED: 0x%08x != 0x965698f3", hash_old_utf8); - } - - // Old-key UTF-8 variant with multi-byte only name: Python fnv1_hash_object_id("温度") - uint32_t hash_old_cjk = esphome::fnv1_hash_object_id("\xe6\xb8\xa9\xe5\xba\xa6", 6, true); - if (hash_old_cjk == 0x3276cb9f) { - ESP_LOGI("FNV1_OID", "old_cjk PASSED"); - } else { - ESP_LOGE("FNV1_OID", "old_cjk FAILED: 0x%08x != 0x3276cb9f", hash_old_cjk); - } - host: api: logger: diff --git a/tests/integration/fixtures/multi_device_preferences.yaml b/tests/integration/fixtures/multi_device_preferences.yaml index 582add90a8e..01e4394559b 100644 --- a/tests/integration/fixtures/multi_device_preferences.yaml +++ b/tests/integration/fixtures/multi_device_preferences.yaml @@ -156,17 +156,10 @@ button: ESP_LOGI("test", "Device A Mode: %s", id(mode_device_a).current_option().c_str()); ESP_LOGI("test", "Device B Mode: %s", id(mode_device_b).current_option().c_str()); ESP_LOGI("test", "Main Mode: %s", id(mode_main).current_option().c_str()); - // Log preference key bases for entities that actually store preferences. - // This is the key base make_entity_preference() uses: entity key XOR device id. - ESP_LOGI("test", "Device A Switch Pref Hash: %u", - id(light_device_a).get_entity_key() ^ id(light_device_a).get_device_id_or_zero()); - ESP_LOGI("test", "Device B Switch Pref Hash: %u", - id(light_device_b).get_entity_key() ^ id(light_device_b).get_device_id_or_zero()); - ESP_LOGI("test", "Main Switch Pref Hash: %u", - id(light_main).get_entity_key() ^ id(light_main).get_device_id_or_zero()); - ESP_LOGI("test", "Device A Number Pref Hash: %u", - id(setpoint_device_a).get_entity_key() ^ id(setpoint_device_a).get_device_id_or_zero()); - ESP_LOGI("test", "Device B Number Pref Hash: %u", - id(setpoint_device_b).get_entity_key() ^ id(setpoint_device_b).get_device_id_or_zero()); - ESP_LOGI("test", "Main Number Pref Hash: %u", - id(setpoint_main).get_entity_key() ^ id(setpoint_main).get_device_id_or_zero()); + // Log preference hashes for entities that actually store preferences + ESP_LOGI("test", "Device A Switch Pref Hash: %u", id(light_device_a).get_preference_hash()); + ESP_LOGI("test", "Device B Switch Pref Hash: %u", id(light_device_b).get_preference_hash()); + ESP_LOGI("test", "Main Switch Pref Hash: %u", id(light_main).get_preference_hash()); + ESP_LOGI("test", "Device A Number Pref Hash: %u", id(setpoint_device_a).get_preference_hash()); + ESP_LOGI("test", "Device B Number Pref Hash: %u", id(setpoint_device_b).get_preference_hash()); + ESP_LOGI("test", "Main Number Pref Hash: %u", id(setpoint_main).get_preference_hash()); diff --git a/tests/integration/fixtures/preference_key_migration.yaml b/tests/integration/fixtures/preference_key_migration.yaml deleted file mode 100644 index a9b01fc2d22..00000000000 --- a/tests/integration/fixtures/preference_key_migration.yaml +++ /dev/null @@ -1,35 +0,0 @@ -esphome: - name: host-pref-key-migration - -host: -api: -logger: - -switch: - - platform: template - id: test_switch_restore - name: Test Switch - optimistic: true - restore_mode: RESTORE_DEFAULT_OFF - -number: - - platform: template - id: test_number_restore - name: Test Number - optimistic: true - restore_value: true - initial_value: 1.0 - min_value: 0 - max_value: 100 - step: 0.5 - -text: - - platform: template - id: test_text_restore - name: Test Text - mode: text - optimistic: true - restore_value: true - initial_value: fallback - min_length: 0 - max_length: 20 diff --git a/tests/integration/host_prefs.py b/tests/integration/host_prefs.py index c7f21d8a010..f835bee3bc6 100644 --- a/tests/integration/host_prefs.py +++ b/tests/integration/host_prefs.py @@ -25,25 +25,15 @@ def clear_host_prefs(device_name: str) -> None: host_prefs_path(device_name).unlink(missing_ok=True) -def write_host_prefs(device_name: str, entries: dict[int, bytes]) -> Path: - """Write preference entries, replacing the file's contents. - - Returns the path that was written. - """ - payload = b"" - for key, data in entries.items(): - if len(data) > 255: - raise ValueError(f"Preference data too long: {len(data)} bytes (max 255)") - payload += struct.pack(" Path: """Write a single preference entry, replacing the file's contents. Returns the path that was written. """ - return write_host_prefs(device_name, {key: data}) + if len(data) > 255: + raise ValueError(f"Preference data too long: {len(data)} bytes (max 255)") + path = host_prefs_path(device_name) + path.parent.mkdir(parents=True, exist_ok=True) + payload = struct.pack(" None: diff --git a/tests/integration/test_object_id_api_verification.py b/tests/integration/test_object_id_api_verification.py index 8dafb37c64d..c8603e06829 100644 --- a/tests/integration/test_object_id_api_verification.py +++ b/tests/integration/test_object_id_api_verification.py @@ -2,8 +2,8 @@ This test verifies a three-way match between: 1. C++ object_id generation (get_object_id_to using to_sanitized_char/to_snake_case_char) -2. C++ entity key generation (fnv1_hash of the raw name in helpers.h) -3. Python computation (sanitize/snake_case and fnv1_hash_name in helpers.py) +2. C++ hash generation (fnv1_hash_object_id in helpers.h) +3. Python computation (sanitize/snake_case in helpers.py, fnv1_hash_object_id) The API response contains C++ computed values, so verifying API == Python implicitly verifies C++ == Python == API for both object_id and hash. @@ -25,7 +25,7 @@ from __future__ import annotations import pytest -from esphome.helpers import fnv1_hash_name +from esphome.helpers import fnv1_hash_object_id from .entity_utils import compute_object_id, verify_all_entities from .types import APIClientConnectedFactory, RunCompiledFunction @@ -123,7 +123,7 @@ async def test_object_id_api_verification( ) # Verify hash can be computed from the name - hash_from_name = fnv1_hash_name(entity_name) + hash_from_name = fnv1_hash_object_id(entity_name) assert hash_from_name == entity.key, ( f"Entity '{entity_name}': hash mismatch. " f"Python hash {hash_from_name:#x}, API key {entity.key:#x}" @@ -164,7 +164,7 @@ async def test_object_id_api_verification( ) # Verify hash matches - expected_hash = fnv1_hash_name(expected_name) + expected_hash = fnv1_hash_object_id(expected_name) assert entity.key == expected_hash, ( f"Empty-name entity (device_id={entity.device_id}): hash mismatch. " f"API key: {entity.key:#x}, expected: {expected_hash:#x}" diff --git a/tests/integration/test_object_id_friendly_name_no_mac_suffix.py b/tests/integration/test_object_id_friendly_name_no_mac_suffix.py index b58593f2ef0..7199a2b3719 100644 --- a/tests/integration/test_object_id_friendly_name_no_mac_suffix.py +++ b/tests/integration/test_object_id_friendly_name_no_mac_suffix.py @@ -11,7 +11,7 @@ from __future__ import annotations import pytest -from esphome.helpers import fnv1_hash_name +from esphome.helpers import fnv1_hash_object_id from .entity_utils import ( compute_object_id, @@ -62,7 +62,7 @@ async def test_object_id_friendly_name_no_mac_suffix( ) # Hash should match friendly_name - expected_hash = fnv1_hash_name("My Friendly Device") + expected_hash = fnv1_hash_object_id("My Friendly Device") assert entity.key == expected_hash, ( f"Expected hash {expected_hash:#x}, got {entity.key:#x}" ) diff --git a/tests/integration/test_object_id_no_friendly_name.py b/tests/integration/test_object_id_no_friendly_name.py index 45b5f730a6d..b548f02fde2 100644 --- a/tests/integration/test_object_id_no_friendly_name.py +++ b/tests/integration/test_object_id_no_friendly_name.py @@ -17,7 +17,7 @@ from __future__ import annotations import pytest -from esphome.helpers import fnv1_hash_name +from esphome.helpers import fnv1_hash_object_id from .entity_utils import compute_object_id, verify_all_entities from .types import APIClientConnectedFactory, RunCompiledFunction @@ -96,7 +96,7 @@ async def test_object_id_no_friendly_name_no_mac_suffix( OLD behavior: - is_object_id_dynamic_() returned false (mac suffix not enabled) - Used object_id_c_str_ which was pre-computed in Python - - Python used get_base_entity_name() with fallback to CORE.name + - Python used get_base_entity_object_id() with fallback to CORE.name Result: object_id = sanitize(snake_case(device_name)) """ @@ -126,7 +126,7 @@ async def test_object_id_no_friendly_name_no_mac_suffix( ) # Hash should match device name - expected_hash = fnv1_hash_name("test-device") + expected_hash = fnv1_hash_object_id("test-device") assert entity.key == expected_hash, ( f"Expected hash {expected_hash:#x}, got {entity.key:#x}" ) diff --git a/tests/integration/test_preference_key_migration.py b/tests/integration/test_preference_key_migration.py deleted file mode 100644 index e7f699bb124..00000000000 --- a/tests/integration/test_preference_key_migration.py +++ /dev/null @@ -1,165 +0,0 @@ -"""Integration test for entity preference key migration. - -Entity keys are now the FNV-1 hash of the raw name instead of the sanitized -object_id (https://github.com/esphome/backlog/issues/85). On key-lookup -preference backends, make_entity_preference() must move data stored under the -old key to the new key, so devices keep their restored state after upgrading. - -This test seeds the host preferences file the way a pre-migration firmware -would have written it and verifies: -1. Data stored under the OLD key is restored (migration happened, no data loss) -2. Data already stored under the NEW key is never overwritten by old data -""" - -from __future__ import annotations - -import socket -import struct - -from aioesphomeapi import ( - NumberInfo, - NumberState, - SwitchInfo, - SwitchState, - TextInfo, - TextState, -) -import pytest - -from esphome.helpers import fnv1_hash, fnv1_hash_name, fnv1_hash_object_id - -from .conftest import run_binary_and_wait_for_port, wait_and_connect_api_client -from .host_prefs import clear_host_prefs, write_host_prefs -from .state_utils import InitialStateHelper, require_entity -from .types import CompileFunction, ConfigWriter - -DEVICE_NAME = "host-pref-key-migration" - -# The pre-migration preference key was the sanitized object_id hash; the new -# key is the raw-name hash. All entities are on the main device (device_id 0) -# and their preferences use no version salt, so the key is just the hash. -SWITCH_OLD_KEY = fnv1_hash_object_id("Test Switch") -SWITCH_NEW_KEY = fnv1_hash_name("Test Switch") -NUMBER_OLD_KEY = fnv1_hash_object_id("Test Number") -NUMBER_NEW_KEY = fnv1_hash_name("Test Number") - -# template_text salts its key with the length limits and pattern hash; this must -# match TemplateText::setup() in template_text.cpp (min_length 0, max_length 20, -# no pattern configured) -TEXT_KEY_EXTRA = (0 << 2) + (20 << 4) + (fnv1_hash("") << 6) -TEXT_OLD_KEY = (fnv1_hash_object_id("Test Text") + TEXT_KEY_EXTRA) & 0xFFFFFFFF -TEXT_NEW_KEY = (fnv1_hash_name("Test Text") + TEXT_KEY_EXTRA) & 0xFFFFFFFF - -# TextSaver<20> stores a length-prefixed buffer of max_length + 1 bytes -TEXT_MAX_LENGTH = 20 - - -def text_pref_payload(value: str) -> bytes: - """Build the length-prefixed buffer TextSaver stores for a value.""" - data = value.encode("utf-8") - assert len(data) <= TEXT_MAX_LENGTH - return bytes([len(data)]) + data + b"\x00" * (TEXT_MAX_LENGTH - len(data)) - - -@pytest.mark.asyncio -async def test_preference_key_migration( - yaml_config: str, - write_yaml_config: ConfigWriter, - compile_esphome: CompileFunction, - reserved_tcp_port: tuple[int, socket.socket], -) -> None: - """Test that preferences stored under the old key survive the upgrade.""" - port, port_socket = reserved_tcp_port - - assert SWITCH_OLD_KEY != SWITCH_NEW_KEY - assert NUMBER_OLD_KEY != NUMBER_NEW_KEY - assert TEXT_OLD_KEY != TEXT_NEW_KEY - - # Write and compile once - config_path = await write_yaml_config(yaml_config) - binary_path = await compile_esphome(config_path) - - # Release the reserved port so the binary can bind to it - port_socket.close() - - async def boot_and_get_initial_states() -> tuple[ - SwitchState, NumberState, TextState - ]: - """Boot the binary and return the restored entity states.""" - async with ( - run_binary_and_wait_for_port(binary_path, "127.0.0.1", port), - wait_and_connect_api_client(port=port) as client, - ): - device_info = await client.device_info() - assert device_info.name == DEVICE_NAME - - entities, _ = await client.list_entities_services() - switch_entity = require_entity( - entities, "test_switch", SwitchInfo, "Test Switch" - ) - number_entity = require_entity( - entities, "test_number", NumberInfo, "Test Number" - ) - text_entity = require_entity(entities, "test_text", TextInfo, "Test Text") - - initial_state_helper = InitialStateHelper(entities) - client.subscribe_states( - initial_state_helper.on_state_wrapper(lambda s: None) - ) - await initial_state_helper.wait_for_initial_states() - - switch_state = initial_state_helper.initial_states[switch_entity.key] - number_state = initial_state_helper.initial_states[number_entity.key] - text_state = initial_state_helper.initial_states[text_entity.key] - assert isinstance(switch_state, SwitchState) - assert isinstance(number_state, NumberState) - assert isinstance(text_state, TextState) - return switch_state, number_state, text_state - - try: - # --- Run 1: only OLD keys present, as written by pre-migration firmware. - # The restored states prove the data was migrated to the new keys. - write_host_prefs( - DEVICE_NAME, - { - SWITCH_OLD_KEY: b"\x01", # bool: switch was ON - NUMBER_OLD_KEY: struct.pack(" None: - """Verify _COMMAND_TOPIC_PLATFORMS matches the MQTT components that subscribe. - - Drift silently reintroduces shared subscribe topics, so this derives the set - from the C++ components that actually call subscribe(); that also catches - platforms like text that subscribe a command topic without exposing a - command_topic key in their schema. - """ - expected: set[str] = set() - for path in (COMPONENTS_DIR / "mqtt").glob("mqtt_*.cpp"): - if path.stem in _NON_ENTITY_MQTT_SOURCES: - continue - if "this->subscribe" not in path.read_text(encoding="utf-8"): - continue - stem = path.stem.removeprefix("mqtt_") - expected.add("datetime" if stem in _DATETIME_STEMS else stem) - assert expected == _COMMAND_TOPIC_PLATFORMS - - -def test_sub_topic_platforms_in_sync() -> None: - """Verify _SUB_TOPIC_PLATFORMS matches the MQTT components with sub-topics. - - Platforms whose MQTT headers use MQTT_COMPONENT_CUSTOM_TOPIC derive extra - topics such as position/command from the object_id. - """ - expected = { - path.stem.removeprefix("mqtt_") - for path in (COMPONENTS_DIR / "mqtt").glob("mqtt_*.h") - if path.stem != "mqtt_component" - and "MQTT_COMPONENT_CUSTOM_TOPIC" in path.read_text(encoding="utf-8") - } - assert expected == _SUB_TOPIC_PLATFORMS - - -def test_conflict_filter_exempts_custom_topics() -> None: - """Test that custom state topics with discovery off avoid the conflict.""" - validator = entity_duplicate_validator("sensor") - # Both entities have custom state topics and discovery disabled per entity, - # so no object_id-derived MQTT topic is used - validator( - { - CONF_NAME: "Датчик открытия", - CONF_STATE_TOPIC: "custom/topic/a", - CONF_DISCOVERY: False, - } - ) - validator( - { - CONF_NAME: "Датчик закрытия", - CONF_STATE_TOPIC: "custom/topic/b", - CONF_DISCOVERY: False, - } - ) - - component_validator = validate_no_object_id_conflicts( - REASON, conflict_filter=_topics_conflict - ) - config: dict = {CONF_DISCOVERY: True, CONF_TOPIC_PREFIX: "test-device"} - assert component_validator(config) is config - - # Without the filter the same conflicts are fatal - with pytest.raises(Invalid, match=r"mqtt builds default topics"): - validate_no_object_id_conflicts(REASON)({}) - - -def test_conflict_on_default_command_topic() -> None: - """Test that commandable platforms conflict through their default command topic. - - Custom state topics with discovery off are not enough for platforms that also - subscribe to an object_id-derived command topic. - """ - validator = entity_duplicate_validator("switch") - validator( - { - CONF_NAME: "Датчик открытия", - CONF_STATE_TOPIC: "custom/topic/a", - CONF_DISCOVERY: False, - } - ) - validator( - { - CONF_NAME: "Датчик закрытия", - CONF_STATE_TOPIC: "custom/topic/b", - CONF_DISCOVERY: False, - } - ) - - component_validator = validate_no_object_id_conflicts( - REASON, conflict_filter=_topics_conflict - ) - mqtt_config: dict = {CONF_DISCOVERY: True, CONF_TOPIC_PREFIX: "test-device"} - # Both switches share the default command topic: rejected - with pytest.raises(Invalid, match=r"mqtt builds default topics"): - component_validator(mqtt_config) - - # With custom command topics as well, nothing derives from the object_id - CORE.reset() - validator = entity_duplicate_validator("switch") - validator( - { - CONF_NAME: "Датчик открытия", - CONF_STATE_TOPIC: "custom/topic/a", - CONF_COMMAND_TOPIC: "custom/cmd/a", - CONF_DISCOVERY: False, - } - ) - validator( - { - CONF_NAME: "Датчик закрытия", - CONF_STATE_TOPIC: "custom/topic/b", - CONF_COMMAND_TOPIC: "custom/cmd/b", - CONF_DISCOVERY: False, - } - ) - assert component_validator(mqtt_config) is mqtt_config - - -def test_conflict_on_sub_topic_platforms() -> None: - """Test that platforms with extra object_id sub-topics always conflict. - - Covers derive topics like position/command from the object_id through their - own config keys, so custom state and command topics cannot exempt them. - """ - validator = entity_duplicate_validator("cover") - validator( - { - CONF_NAME: "Датчик открытия", - CONF_STATE_TOPIC: "custom/topic/a", - CONF_COMMAND_TOPIC: "custom/cmd/a", - CONF_DISCOVERY: False, - } - ) - validator( - { - CONF_NAME: "Датчик закрытия", - CONF_STATE_TOPIC: "custom/topic/b", - CONF_COMMAND_TOPIC: "custom/cmd/b", - CONF_DISCOVERY: False, - } - ) - - component_validator = validate_no_object_id_conflicts( - REASON, conflict_filter=_topics_conflict - ) - with pytest.raises(Invalid, match=r"mqtt builds default topics"): - component_validator({CONF_DISCOVERY: True, CONF_TOPIC_PREFIX: "test-device"}) - - -def test_no_conflict_on_disjoint_default_topics() -> None: - """Test that entities whose default topics are disjoint do not conflict. - - One entity uses only the default command topic and the other only the default - state topic, so they never share a topic. - """ - validator = entity_duplicate_validator("switch") - validator( - { - CONF_NAME: "Датчик открытия", - CONF_STATE_TOPIC: "custom/topic/a", - CONF_DISCOVERY: False, - } - ) - validator( - { - CONF_NAME: "Датчик закрытия", - CONF_COMMAND_TOPIC: "custom/cmd/b", - CONF_DISCOVERY: False, - } - ) - - component_validator = validate_no_object_id_conflicts( - REASON, conflict_filter=_topics_conflict - ) - config: dict = {CONF_DISCOVERY: True, CONF_TOPIC_PREFIX: "test-device"} - assert component_validator(config) is config - - -def test_no_conflict_on_empty_topic_prefix() -> None: - """Test that an empty topic_prefix disables the default topic conflict. - - With topic_prefix set to null no default topics exist at runtime, so entities - without custom state topics cannot conflict; only discovery still matters. - """ - validator = entity_duplicate_validator("sensor") - validator({CONF_NAME: "Датчик открытия"}) - validator({CONF_NAME: "Датчик закрытия"}) - - component_validator = validate_no_object_id_conflicts( - REASON, conflict_filter=_topics_conflict - ) - # No default topics and no discovery: valid - config: dict = {CONF_DISCOVERY: False, CONF_TOPIC_PREFIX: ""} - assert component_validator(config) is config - - # Discovery still uses object_id-derived config topics: rejected - with pytest.raises(Invalid, match=r"mqtt builds default topics"): - component_validator({CONF_DISCOVERY: True, CONF_TOPIC_PREFIX: ""}) diff --git a/tests/unit_tests/core/test_entity_helpers.py b/tests/unit_tests/core/test_entity_helpers.py index 64400c4fd4a..53035ad7136 100644 --- a/tests/unit_tests/core/test_entity_helpers.py +++ b/tests/unit_tests/core/test_entity_helpers.py @@ -1,4 +1,4 @@ -"""Tests for entity helpers: name selection, entity key hashing, duplicate checks.""" +"""Test get_base_entity_object_id function matches C++ behavior.""" from collections.abc import Callable, Generator from pathlib import Path @@ -25,17 +25,16 @@ from esphome.core.entity_helpers import ( _setup_entity_impl, entity_duplicate_validator, finalize_entity_strings, - get_base_entity_name, + get_base_entity_object_id, register_device_class, register_icon, register_unit_of_measurement, setup_device_class, setup_entity, setup_unit_of_measurement, - validate_no_object_id_conflicts, ) from esphome.cpp_generator import MockObj -from esphome.helpers import fnv1_hash_name, sanitize, snake_case +from esphome.helpers import fnv1_hash, sanitize, snake_case from .common import load_config_from_fixture @@ -58,26 +57,206 @@ def restore_core_state() -> Generator[None, None, None]: CORE.friendly_name = original_friendly_name -def test_get_base_entity_name_priority_order() -> None: +def test_with_entity_name() -> None: + """Test when entity has its own name - should use entity name.""" + # Simple name + assert get_base_entity_object_id("Temperature Sensor", None) == "temperature_sensor" + assert ( + get_base_entity_object_id("Temperature Sensor", "Device Name") + == "temperature_sensor" + ) + # Even with device name, entity name takes precedence + assert ( + get_base_entity_object_id("Temperature Sensor", "Device Name", "Sub Device") + == "temperature_sensor" + ) + + # Name with special characters + assert ( + get_base_entity_object_id("Temp!@#$%^&*()Sensor", None) + == "temp__________sensor" + ) + assert get_base_entity_object_id("Temp-Sensor_123", None) == "temp-sensor_123" + + # Already snake_case + assert get_base_entity_object_id("temperature_sensor", None) == "temperature_sensor" + + # Mixed case + assert get_base_entity_object_id("TemperatureSensor", None) == "temperaturesensor" + assert get_base_entity_object_id("TEMPERATURE SENSOR", None) == "temperature_sensor" + + +def test_empty_name_with_device_name() -> None: + """Test when entity has empty name and is on a sub-device - should use device name.""" + # C++ behavior: when has_own_name is false and device is set, uses device->get_name() + assert ( + get_base_entity_object_id("", "Friendly Device", "Sub Device 1") + == "sub_device_1" + ) + assert ( + get_base_entity_object_id("", "Kitchen Controller", "controller_1") + == "controller_1" + ) + assert get_base_entity_object_id("", None, "Test-Device_123") == "test-device_123" + + +def test_empty_name_with_friendly_name() -> None: + """Test when entity has empty name and no device - should use friendly name.""" + # C++ behavior: when has_own_name is false, uses App.get_friendly_name() + assert get_base_entity_object_id("", "Friendly Device") == "friendly_device" + assert get_base_entity_object_id("", "Kitchen Controller") == "kitchen_controller" + assert get_base_entity_object_id("", "Test-Device_123") == "test-device_123" + + # Special characters in friendly name + assert get_base_entity_object_id("", "Device!@#$%") == "device_____" + + +def test_empty_name_no_friendly_name() -> None: + """Test when entity has empty name and no friendly name - should use device name.""" + # Test with CORE.name set + CORE.name = "device-name" + assert get_base_entity_object_id("", None) == "device-name" + + CORE.name = "Test Device" + assert get_base_entity_object_id("", None) == "test_device" + + +def test_edge_cases() -> None: + """Test edge cases.""" + # Only spaces + assert get_base_entity_object_id(" ", None) == "___" + + # Unicode characters (should be replaced) + assert get_base_entity_object_id("Température", None) == "temp_rature" + assert get_base_entity_object_id("测试", None) == "__" + + # Empty string with empty friendly name (empty friendly name is treated as None) + # Falls back to CORE.name + CORE.name = "device" + assert get_base_entity_object_id("", "") == "device" + + # Very long name (should work fine) + long_name = "a" * 100 + " " + "b" * 100 + expected = "a" * 100 + "_" + "b" * 100 + assert get_base_entity_object_id(long_name, None) == expected + + +@pytest.mark.parametrize( + ("name", "expected"), + [ + ("Temperature Sensor", "temperature_sensor"), + ("Living Room Light", "living_room_light"), + ("Test-Device_123", "test-device_123"), + ("Special!@#Chars", "special___chars"), + ("UPPERCASE NAME", "uppercase_name"), + ("lowercase name", "lowercase_name"), + ("Mixed Case Name", "mixed_case_name"), + (" Spaces ", "___spaces___"), + ], +) +def test_matches_cpp_helpers(name: str, expected: str) -> None: + """Test that the logic matches using snake_case and sanitize directly.""" + # For non-empty names, verify our function produces same result as direct snake_case + sanitize + assert get_base_entity_object_id(name, None) == sanitize(snake_case(name)) + assert get_base_entity_object_id(name, None) == expected + + +def test_empty_name_fallback() -> None: + """Test empty name handling which falls back to friendly_name or CORE.name.""" + # Empty name is handled specially - it doesn't just use sanitize(snake_case("")) + # Instead it falls back to friendly_name or CORE.name + assert sanitize(snake_case("")) == "" # Direct conversion gives empty string + # But our function returns a fallback + CORE.name = "device" + assert get_base_entity_object_id("", None) == "device" # Uses device name + + +def test_name_add_mac_suffix_behavior() -> None: + """Test behavior related to name_add_mac_suffix. + + In C++, an entity's object_id is computed from its name_ via + write_object_id_to() (sanitized snake_case). When an entity has no name, + configure_entity_() sets name_ from the friendly name, with the MAC suffix + appended when name_add_mac_suffix is enabled. Our function always returns + the same result since we're calculating the base for duplicate tracking. + """ + # The function should always return the same result regardless of + # name_add_mac_suffix setting, as we're calculating the base object_id + assert get_base_entity_object_id("", "Test Device") == "test_device" + assert get_base_entity_object_id("Entity Name", "Test Device") == "entity_name" + + +def test_priority_order() -> None: """Test the priority order: entity name > device name > friendly name > CORE.name.""" CORE.name = "core-device" - # 1. Entity name has highest priority and is used as-is, no transformations + # 1. Entity name has highest priority assert ( - get_base_entity_name("Entity Name", "Friendly Name", "Device Name") - == "Entity Name" + get_base_entity_object_id("Entity Name", "Friendly Name", "Device Name") + == "entity_name" ) - assert get_base_entity_name("Température", None) == "Température" # 2. Device name is next priority (when entity name is empty) - assert get_base_entity_name("", "Friendly Name", "Device Name") == "Device Name" + assert ( + get_base_entity_object_id("", "Friendly Name", "Device Name") == "device_name" + ) # 3. Friendly name is next (when entity and device names are empty) - assert get_base_entity_name("", "Friendly Name", None) == "Friendly Name" + assert get_base_entity_object_id("", "Friendly Name", None) == "friendly_name" - # 4. CORE.name is last resort; an empty friendly name falls through to it - assert get_base_entity_name("", None, None) == "core-device" - assert get_base_entity_name("", "") == "core-device" + # 4. CORE.name is last resort + assert get_base_entity_object_id("", None, None) == "core-device" + + +@pytest.mark.parametrize( + ("name", "friendly_name", "device_name", "expected"), + [ + # name, friendly_name, device_name, expected + ("Living Room Light", None, None, "living_room_light"), + ("", "Kitchen Controller", None, "kitchen_controller"), + ( + "", + "ESP32 Device", + "controller_1", + "controller_1", + ), # Device name takes precedence + ("GPIO2 Button", None, None, "gpio2_button"), + ("WiFi Signal", "My Device", None, "wifi_signal"), + ("", None, "esp32_node", "esp32_node"), + ("Front Door Sensor", "Home Assistant", "door_controller", "front_door_sensor"), + ], +) +def test_real_world_examples( + name: str, friendly_name: str | None, device_name: str | None, expected: str +) -> None: + """Test real-world entity naming scenarios.""" + result = get_base_entity_object_id(name, friendly_name, device_name) + assert result == expected + + +def test_issue_6953_scenarios() -> None: + """Test specific scenarios from issue #6953.""" + # Scenario 1: Multiple empty names on main device with name_add_mac_suffix + # The Python code calculates the base, C++ might append MAC suffix dynamically + CORE.name = "device-name" + CORE.friendly_name = "Friendly Device" + + # All empty names should resolve to same base + assert get_base_entity_object_id("", CORE.friendly_name) == "friendly_device" + assert get_base_entity_object_id("", CORE.friendly_name) == "friendly_device" + assert get_base_entity_object_id("", CORE.friendly_name) == "friendly_device" + + # Scenario 2: Empty names on sub-devices + assert ( + get_base_entity_object_id("", "Main Device", "controller_1") == "controller_1" + ) + assert ( + get_base_entity_object_id("", "Main Device", "controller_2") == "controller_2" + ) + + # Scenario 3: xyz duplicates + assert get_base_entity_object_id("xyz", None) == "xyz" + assert get_base_entity_object_id("xyz", "Device") == "xyz" # Tests for setup_entity function @@ -336,10 +515,9 @@ def test_entity_duplicate_validator() -> None: config1 = {CONF_NAME: "Temperature"} validated1 = validator(config1) assert validated1 == config1 - temperature_key = ("", "sensor", fnv1_hash_name("Temperature")) - assert temperature_key in CORE.unique_ids + assert ("", "sensor", fnv1_hash("temperature")) in CORE.unique_ids # Check metadata was stored - metadata = CORE.unique_ids[temperature_key] + metadata = CORE.unique_ids[("", "sensor", fnv1_hash("temperature"))] assert metadata["name"] == "Temperature" assert metadata["platform"] == "sensor" @@ -347,9 +525,8 @@ def test_entity_duplicate_validator() -> None: config2 = {CONF_NAME: "Humidity"} validated2 = validator(config2) assert validated2 == config2 - humidity_key = ("", "sensor", fnv1_hash_name("Humidity")) - assert humidity_key in CORE.unique_ids - metadata2 = CORE.unique_ids[humidity_key] + assert ("", "sensor", fnv1_hash("humidity")) in CORE.unique_ids + metadata2 = CORE.unique_ids[("", "sensor", fnv1_hash("humidity"))] assert metadata2["name"] == "Humidity" # Duplicate entity should fail @@ -360,6 +537,34 @@ def test_entity_duplicate_validator() -> None: validator(config3) +def test_entity_duplicate_validator_hash_collision() -> None: + """Test that two different object_ids with the same FNV-1 hash are rejected.""" + # Brute-forced FNV-1 32-bit collision pair; both object_ids hash to 0xe95747e4 + name_a = "Sensor aooxzi" + name_b = "Sensor baraia" + object_id_a = sanitize(snake_case(name_a)) + object_id_b = sanitize(snake_case(name_b)) + assert object_id_a != object_id_b + assert fnv1_hash(object_id_a) == fnv1_hash(object_id_b) + + validator = entity_duplicate_validator("sensor") + + config1 = {CONF_NAME: name_a} + validated1 = validator(config1) + assert validated1 == config1 + + config2 = {CONF_NAME: name_b} + with pytest.raises( + Invalid, + match=re.compile( + r"Duplicate sensor entity with name 'Sensor baraia' found.*" + r"produce the same entity key hash \(0xe95747e4\)", + re.DOTALL, + ), + ): + validator(config2) + + def test_entity_duplicate_validator_with_devices() -> None: """Test entity_duplicate_validator with devices.""" # Create validator for sensor platform @@ -370,19 +575,18 @@ def test_entity_duplicate_validator_with_devices() -> None: device2 = ID("device2", type="Device") # Same name on different devices should pass - name_hash = fnv1_hash_name("Temperature") config1 = {CONF_NAME: "Temperature", CONF_DEVICE_ID: device1} validated1 = validator(config1) assert validated1 == config1 - assert ("device1", "sensor", name_hash) in CORE.unique_ids - metadata1 = CORE.unique_ids[("device1", "sensor", name_hash)] + assert ("device1", "sensor", fnv1_hash("temperature")) in CORE.unique_ids + metadata1 = CORE.unique_ids[("device1", "sensor", fnv1_hash("temperature"))] assert metadata1["device_id"] == "device1" config2 = {CONF_NAME: "Temperature", CONF_DEVICE_ID: device2} validated2 = validator(config2) assert validated2 == config2 - assert ("device2", "sensor", name_hash) in CORE.unique_ids - metadata2 = CORE.unique_ids[("device2", "sensor", name_hash)] + assert ("device2", "sensor", fnv1_hash("temperature")) in CORE.unique_ids + metadata2 = CORE.unique_ids[("device2", "sensor", fnv1_hash("temperature"))] assert metadata2["device_id"] == "device2" # Duplicate on same device should fail @@ -434,33 +638,6 @@ def test_entity_different_platforms_yaml_validation( assert result is not None -def test_object_id_conflict_mqtt_yaml_validation( - yaml_file: Callable[[str], str], capsys: pytest.CaptureFixture[str] -) -> None: - """Test that names sanitizing to the same object_id fail when mqtt is configured.""" - result = load_config_from_fixture( - yaml_file, "object_id_conflict_mqtt.yaml", FIXTURES_DIR - ) - assert result is None - - captured = capsys.readouterr() - assert ( - "mqtt builds default topics and discovery topics from the entity object_id" - in captured.out - ) - - -def test_object_id_conflict_without_mqtt_yaml_validation( - yaml_file: Callable[[str], str], -) -> None: - """Test that names sanitizing to the same object_id pass without mqtt/prometheus.""" - result = load_config_from_fixture( - yaml_file, "object_id_conflict_no_mqtt.yaml", FIXTURES_DIR - ) - # This should succeed - assert result is not None - - def test_entity_duplicate_validator_error_message() -> None: """Test that duplicate entity error messages include helpful metadata.""" # Create validator for sensor platform @@ -519,8 +696,7 @@ def test_entity_duplicate_validator_internal_entities() -> None: validated1 = validator(config1) assert validated1 == config1 # New format includes device_id (empty string for main device) - temperature_key = ("", "sensor", fnv1_hash_name("Temperature")) - assert temperature_key in CORE.unique_ids + assert ("", "sensor", fnv1_hash("temperature")) in CORE.unique_ids # Internal entity with same name should pass (not added to unique_ids) config2 = {CONF_NAME: "Temperature", CONF_INTERNAL: True} @@ -528,7 +704,9 @@ def test_entity_duplicate_validator_internal_entities() -> None: assert validated2 == config2 # Internal entity should not be added to unique_ids # Count how many times the key appears (should still be 1) - count = sum(1 for k in CORE.unique_ids if k == temperature_key) + count = sum( + 1 for k in CORE.unique_ids if k == ("", "sensor", fnv1_hash("temperature")) + ) assert count == 1 # Another internal entity with same name should also pass @@ -536,7 +714,9 @@ def test_entity_duplicate_validator_internal_entities() -> None: validated3 = validator(config3) assert validated3 == config3 # Still only one entry in unique_ids (from the non-internal entity) - count = sum(1 for k in CORE.unique_ids if k == temperature_key) + count = sum( + 1 for k in CORE.unique_ids if k == ("", "sensor", fnv1_hash("temperature")) + ) assert count == 1 # Non-internal entity with same name should fail @@ -564,148 +744,30 @@ def test_empty_or_null_device_id_on_entity() -> None: def test_entity_duplicate_validator_non_ascii_names() -> None: - """Test that distinct non-ASCII names no longer collide. - - These names used to be rejected because both sanitize to only underscores; - the entity key now hashes the raw name so they stay distinct. - """ + """Test that non-ASCII names show helpful error messages.""" # Create validator for binary_sensor platform validator = entity_duplicate_validator("binary_sensor") - # Both Russian sensors should pass even though they sanitize identically + # First Russian sensor should pass config1 = {CONF_NAME: "Датчик открытия основного крана"} validated1 = validator(config1) assert validated1 == config1 + # Second Russian sensor with different text but same ASCII conversion should fail config2 = {CONF_NAME: "Датчик закрытия основного крана"} - validated2 = validator(config2) - assert validated2 == config2 - - # An exact duplicate still fails - config3 = {CONF_NAME: "Датчик открытия основного крана"} - with pytest.raises( - Invalid, - match=r"Duplicate binary_sensor entity with name 'Датчик открытия основного крана' found", - ): - validator(config3) - - -def test_entity_duplicate_validator_hash_collision() -> None: - """Test that two different names with the same FNV-1 hash are rejected.""" - # Brute-forced FNV-1 32-bit collision pair; both hash to 0x0ee5ff7b - name_a = "Sensor m2CZ" - name_b = "Sensor qCaa" - assert name_a != name_b - assert fnv1_hash_name(name_a) == fnv1_hash_name(name_b) - - validator = entity_duplicate_validator("sensor") - - config1 = {CONF_NAME: name_a} - validated1 = validator(config1) - assert validated1 == config1 - - config2 = {CONF_NAME: name_b} with pytest.raises( Invalid, match=re.compile( - rf"Duplicate sensor entity with name '{name_b}' found.*" - rf"The names '{name_b}' and '{name_a}' produce the.*" - r"same entity key hash \(0x0ee5ff7b\).*" - r"To fix: Rename one of the entities", + r"Duplicate binary_sensor entity with name 'Датчик закрытия основного крана' found.*" + r"Original names: 'Датчик закрытия основного крана' and 'Датчик открытия основного крана'.*" + r"Both convert to ASCII ID: '_______________________________'.*" + r"To fix: Add unique ASCII characters \(e\.g\., '1', '2', or 'A', 'B'\)", re.DOTALL, ), ): validator(config2) -def test_object_id_conflicts_rejected_by_component_validator() -> None: - """Test that object_id conflicts pass entity validation but fail for mqtt/prometheus.""" - validator = entity_duplicate_validator("sensor") - - # Both names validate fine in general (distinct raw names, distinct keys) - validator({CONF_NAME: "Датчик открытия"}) - validator({CONF_NAME: "Датчик закрытия"}) - - # A component that addresses entities by object_id must reject the config - component_validator = validate_no_object_id_conflicts( - "mqtt builds default topics from the entity object_id" - ) - with pytest.raises( - Invalid, - match=re.compile( - r"mqtt builds default topics from the entity object_id.*" - r"sensor entities 'Датчик открытия', 'Датчик закрытия' " - r"share the object_id '_______________'.*" - r"To fix: Add unique ASCII characters", - re.DOTALL, - ), - ): - component_validator({}) - - -def test_object_id_conflicts_skipped_in_testing_mode() -> None: - """Test that testing_mode skips the conflict check, as used for grouped testing.""" - validator = entity_duplicate_validator("sensor") - validator({CONF_NAME: "Датчик открытия"}) - validator({CONF_NAME: "Датчик закрытия"}) - - component_validator = validate_no_object_id_conflicts( - "mqtt builds default topics from the entity object_id" - ) - CORE.testing_mode = True - try: - config: dict = {} - assert component_validator(config) is config - finally: - CORE.testing_mode = False - - -def test_object_id_conflicts_none_recorded() -> None: - """Test that distinct object_ids produce no conflicts.""" - validator = entity_duplicate_validator("sensor") - validator({CONF_NAME: "Temperature"}) - validator({CONF_NAME: "Humidity"}) - - component_validator = validate_no_object_id_conflicts( - "mqtt builds default topics from the entity object_id" - ) - config: dict = {} - assert component_validator(config) is config - - -def test_object_id_conflicts_device_scoped() -> None: - """Test that the object_id conflict check is scoped per device. - - Same-named entities on different sub-devices were accepted before entity keys - moved to raw names, so the check keeps that scope; conflicts within one device - are still reported with the device named in the message. - """ - validator = entity_duplicate_validator("sensor") - validator({CONF_NAME: "Temperature", CONF_DEVICE_ID: ID("device1", type="Device")}) - validator({CONF_NAME: "Temperature", CONF_DEVICE_ID: ID("device2", type="Device")}) - - component_validator = validate_no_object_id_conflicts( - "prometheus builds metric labels from the entity object_id" - ) - config: dict = {} - assert component_validator(config) is config - - # Two names sanitizing identically on the same sub-device still conflict - validator( - {CONF_NAME: "Датчик открытия", CONF_DEVICE_ID: ID("device1", type="Device")} - ) - validator( - {CONF_NAME: "Датчик закрытия", CONF_DEVICE_ID: ID("device1", type="Device")} - ) - with pytest.raises( - Invalid, - match=re.compile( - r"prometheus builds metric labels.*on device 'device1'", re.DOTALL - ), - ): - component_validator({}) - - def test_entity_duplicate_validator_same_name_no_enhanced_message() -> None: """Test that identical names don't show the enhanced message.""" # Create validator for sensor platform @@ -763,7 +825,7 @@ async def test_setup_entity_empty_name_with_device( # For empty-name entities, Python stores hash 0 - C++ calculates hash at runtime assert config.get("_entity_name") == "" - assert config.get("_entity_key") == 0 + assert config.get("_entity_object_id_hash") == 0 @pytest.mark.asyncio @@ -792,7 +854,7 @@ async def test_setup_entity_empty_name_with_mac_suffix( # For empty-name entities, Python stores hash 0 - C++ calculates hash at runtime assert config.get("_entity_name") == "" - assert config.get("_entity_key") == 0 + assert config.get("_entity_object_id_hash") == 0 @pytest.mark.asyncio @@ -822,7 +884,7 @@ async def test_setup_entity_empty_name_with_mac_suffix_no_friendly_name( # For empty-name entities, Python stores hash 0 - C++ calculates hash at runtime assert config.get("_entity_name") == "" - assert config.get("_entity_key") == 0 + assert config.get("_entity_object_id_hash") == 0 @pytest.mark.asyncio @@ -853,7 +915,7 @@ async def test_setup_entity_empty_name_no_mac_suffix_no_friendly_name( # For empty-name entities, Python stores hash 0 - C++ calculates hash at runtime assert config.get("_entity_name") == "" - assert config.get("_entity_key") == 0 + assert config.get("_entity_object_id_hash") == 0 def test_register_string_overflow() -> None: diff --git a/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_mqtt.yaml b/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_mqtt.yaml deleted file mode 100644 index 4a6f56f4730..00000000000 --- a/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_mqtt.yaml +++ /dev/null @@ -1,22 +0,0 @@ -esphome: - name: test-object-id-conflict - -esp32: - board: esp32dev - -wifi: - ssid: MySSID - password: password1 - -mqtt: - broker: test.mosquitto.org - -sensor: - # Distinct raw names are fine in general, but both sanitize to the same - # object_id, which MQTT still uses to build default topics - should fail - - platform: template - name: "Датчик открытия" - lambda: return 21.0; - - platform: template - name: "Датчик закрытия" - lambda: return 22.0; diff --git a/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_no_mqtt.yaml b/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_no_mqtt.yaml deleted file mode 100644 index c0fbd5cbbaa..00000000000 --- a/tests/unit_tests/fixtures/core/entity_helpers/object_id_conflict_no_mqtt.yaml +++ /dev/null @@ -1,15 +0,0 @@ -esphome: - name: test-object-id-ok - -esp32: - board: esp32dev - -sensor: - # Distinct raw names that sanitize to the same object_id are allowed when no - # component addresses entities by object_id (no mqtt or prometheus configured) - - platform: template - name: "Датчик открытия" - lambda: return 21.0; - - platform: template - name: "Датчик закрытия" - lambda: return 22.0; diff --git a/tests/unit_tests/test_preference_hash_stability.py b/tests/unit_tests/test_preference_hash_stability.py index d3e5fac36a1..d8506afae7b 100644 --- a/tests/unit_tests/test_preference_hash_stability.py +++ b/tests/unit_tests/test_preference_hash_stability.py @@ -5,11 +5,11 @@ users to lose stored preferences (calibration values, restore states, etc.) on firmware upgrades, or break entity state routing to API clients. Two algorithms are locked here (see https://github.com/esphome/backlog/issues/85): -1. `fnv1_hash_object_id(name)` - the LEGACY hash (snake_case + sanitize, then FNV-1). - Existing devices have preferences stored under keys derived from it; slot-based - backends (ESP8266, RP2040) keep using it, and key-lookup backends migrate FROM it. -2. `fnv1_hash_name(name)` - the entity key (FNV-1 over the raw UTF-8 name bytes). - Sent to API clients and used as the preference key base on key-lookup backends. +1. `fnv1_hash_object_id(name)` - the object_id hash (snake_case + sanitize, then FNV-1). + The entity key sent to API clients and the base of every stored preference key. +2. `fnv1_hash_name(name)` - FNV-1 over the raw UTF-8 name bytes. 2026.8 beta + firmware stored preferences under keys derived from it; a future key migration + must reconstruct those keys to recover that data. DO NOT CHANGE THE EXPECTED VALUES - if tests fail after modifying a hash algorithm, the change breaks backward compatibility and will cause data loss. @@ -124,8 +124,9 @@ def test_entity_object_id_hash_stability( """Verify fnv1_hash_object_id produces stable hashes for entity names. CRITICAL: These expected values MUST NOT CHANGE. Existing devices have - preferences stored under keys derived from this legacy hash; changing it - breaks the old-to-new key migration and loses stored preferences. + preferences stored under keys derived from this hash, and it is the entity + key sent to API clients; changing it loses stored preferences and breaks + entity state routing. """ actual = fnv1_hash_object_id(entity_name) assert actual == expected_object_id_hash, ( @@ -144,9 +145,8 @@ def compute_legacy_preference_key( ) -> int: """Compute the legacy preference key: (object_id_hash ^ device_id) ^ version. - This is the key existing devices have data stored under. Slot-based backends - (ESP8266, RP2040) still use it directly; key-lookup backends compute it as the - migration source in EntityBase::make_entity_preference_() (entity_base.cpp). + This is the key EntityBase::make_entity_preference_() (entity_base.cpp) + stores every entity preference under. """ object_id_hash = fnv1_hash_object_id(entity_name) preference_hash = object_id_hash ^ device_id @@ -179,8 +179,8 @@ def test_legacy_preference_key_computation( ) -> None: """Verify legacy preference key computation matches expected values. - This test ensures the formula doesn't change, which would break both slot-based - preference storage and the migration source keys on key-lookup backends. + This test ensures the formula doesn't change, which would lose stored + preferences on every platform. """ actual_key = compute_legacy_preference_key(entity_name, version, device_id) @@ -215,12 +215,12 @@ def test_legacy_preference_key_computation( ], ) def test_entity_key_hash_stability(entity_name: str, expected_key: int) -> None: - """Verify fnv1_hash_name produces stable entity keys. + """Verify fnv1_hash_name produces stable raw-name hashes. - CRITICAL: These expected values MUST NOT CHANGE. The entity key is sent to - API clients and is the new preference key base; changing the algorithm - would break state routing and lose stored preferences. - Must match C++ fnv1_hash_bytes() in esphome/core/helpers.h. + CRITICAL: These expected values MUST NOT CHANGE. 2026.8 beta firmware stored + preferences under keys derived from this hash; a future key migration must + reconstruct those keys, and changing the algorithm would strand that data. + Matched C++ fnv1_hash_bytes() (2026.8 beta), which the unrevert restores. """ actual = fnv1_hash_name(entity_name) assert actual == expected_key, (