diff --git a/esphome/core/config.py b/esphome/core/config.py index bf210876dfb..018e05f17b4 100644 --- a/esphome/core/config.py +++ b/esphome/core/config.py @@ -242,6 +242,10 @@ PROJECT_MAX_LENGTH = 127 # Max board/model string length (must fit in single-byte varint for proto encoding) BOARD_MAX_LENGTH = 127 +# Keep in sync with ESPHOME_COMMENT_SIZE_MAX in esphome/core/application.h +# (C++ side includes the null terminator). +COMMENT_MAX_LEN = 255 + AREA_SCHEMA = cv.Schema( { cv.GenerateID(CONF_ID): cv.declare_id(Area), @@ -275,7 +279,9 @@ CONFIG_SCHEMA = cv.All( cv.string_no_slash, cv.ByteLength(max=FRIENDLY_NAME_MAX_LEN) ), cv.Optional(CONF_AREA): validate_area_config, - cv.Optional(CONF_COMMENT): cv.All(cv.string, cv.Length(max=255)), + cv.Optional(CONF_COMMENT): cv.All( + cv.string, cv.ByteLength(max=COMMENT_MAX_LEN) + ), cv.Required(CONF_BUILD_PATH): cv.string, cv.Optional(CONF_PLATFORMIO_OPTIONS, default={}): cv.Schema( { diff --git a/esphome/writer.py b/esphome/writer.py index 787ecac6f6e..816c57a0bc1 100644 --- a/esphome/writer.py +++ b/esphome/writer.py @@ -22,7 +22,6 @@ from esphome.helpers import ( read_file, rmtree, walk_files, - write_file, write_file_if_changed, ) from esphome.storage_json import StorageJSON, storage_path @@ -171,6 +170,7 @@ VERSION_H_FORMAT = """\ DEFINES_H_TARGET = "esphome/core/defines.h" VERSION_H_TARGET = "esphome/core/version.h" BUILD_INFO_DATA_H_TARGET = "esphome/core/build_info_data.h" +BUILD_INFO_DATA_CPP_TARGET = "esphome/core/build_info_data.cpp" ENTITY_TYPES_H_TARGET = "esphome/core/entity_types.h" ESPHOME_README_TXT = """ THIS DIRECTORY IS AUTO-GENERATED, DO NOT MODIFY @@ -209,13 +209,22 @@ def copy_src_tree(): source_files_copy = source_files_map.copy() ignore_targets = [ - Path(x) for x in (DEFINES_H_TARGET, VERSION_H_TARGET, BUILD_INFO_DATA_H_TARGET) + Path(x) + for x in ( + DEFINES_H_TARGET, + VERSION_H_TARGET, + BUILD_INFO_DATA_H_TARGET, + BUILD_INFO_DATA_CPP_TARGET, + ) ] for t in ignore_targets: source_files_copy.pop(t, None) # Files to exclude from sources_changed tracking (generated files) - generated_files = {Path("esphome/core/build_info_data.h")} + generated_files = { + Path("esphome/core/build_info_data.h"), + Path("esphome/core/build_info_data.cpp"), + } sources_changed = False for fname in walk_files(CORE.relative_src_path("esphome")): @@ -268,12 +277,15 @@ def copy_src_tree(): build_info_data_h_path = CORE.relative_src_path( "esphome", "core", "build_info_data.h" ) + build_info_data_cpp_path = CORE.relative_src_path( + "esphome", "core", "build_info_data.cpp" + ) build_info_json_path = CORE.relative_build_path("build_info.json") config_hash, build_time, build_time_str, comment = get_build_info() # Defensively force a rebuild if the build_info files don't exist, or if # there was a config change which didn't actually cause a source change - if not build_info_data_h_path.exists(): + if not build_info_data_h_path.exists() or not build_info_data_cpp_path.exists(): sources_changed = True else: try: @@ -288,13 +300,19 @@ def copy_src_tree(): # Write build_info header and JSON metadata if sources_changed: - write_file( + # write_file_if_changed avoids bumping mtime on identical content, + # which is what makes the stable header actually isolate metadata churn. + write_file_if_changed( build_info_data_h_path, - generate_build_info_data_h( + generate_build_info_data_h(), + ) + write_file_if_changed( + build_info_data_cpp_path, + generate_build_info_data_cpp( config_hash, build_time, build_time_str, comment ), ) - write_file( + write_file_if_changed( build_info_json_path, json.dumps( { @@ -345,27 +363,60 @@ def get_build_info() -> tuple[int, int, str, str]: return config_hash, build_time, build_time_str, comment -def generate_build_info_data_h( - config_hash: int, build_time: int, build_time_str: str, comment: str -) -> str: - """Generate build_info_data.h header with config hash, build time, and comment.""" - # cpp_string_escape returns '"escaped"', slice off the quotes since template has them - escaped_comment = cpp_string_escape(comment)[1:-1] - # +1 for null terminator - comment_size = len(comment) + 1 - return f"""#pragma once -// Auto-generated build_info data -#define ESPHOME_CONFIG_HASH 0x{config_hash:08x}U // NOLINT -#define ESPHOME_BUILD_TIME {build_time} // NOLINT -#define ESPHOME_COMMENT_SIZE {comment_size} // NOLINT +def generate_build_info_data_h() -> str: + """Generate stable declarations for build info provided by generated C++.""" + return """#pragma once +// Auto-generated build_info declarations +#include +#include +#include #ifdef USE_ESP8266 #include -static const char ESPHOME_BUILD_TIME_STR[] PROGMEM = "{build_time_str}"; -static const char ESPHOME_COMMENT_STR[] PROGMEM = "{escaped_comment}"; -#else -static const char ESPHOME_BUILD_TIME_STR[] = "{build_time_str}"; -static const char ESPHOME_COMMENT_STR[] = "{escaped_comment}"; #endif + +namespace esphome { +extern const uint32_t ESPHOME_CONFIG_HASH; +extern const time_t ESPHOME_BUILD_TIME; +extern const size_t ESPHOME_COMMENT_SIZE; +#ifdef USE_ESP8266 +extern const char ESPHOME_BUILD_TIME_STR[] PROGMEM; +extern const char ESPHOME_COMMENT_STR[] PROGMEM; +#else +extern const char ESPHOME_BUILD_TIME_STR[]; +extern const char ESPHOME_COMMENT_STR[]; +#endif +} // namespace esphome +""" + + +def generate_build_info_data_cpp( + config_hash: int, build_time: int, build_time_str: str, comment: str +) -> str: + """Generate build_info_data.cpp with config hash, build time, and comment.""" + from esphome.core.config import COMMENT_MAX_LEN + + # Defense-in-depth clamp; errors="ignore" drops a partial trailing UTF-8 + # sequence so the literal never decodes to a truncated codepoint. + encoded = comment.encode("utf-8")[:COMMENT_MAX_LEN] + comment = encoded.decode("utf-8", errors="ignore") + # cpp_string_escape wraps in quotes; strip them since the template has them. + escaped_comment = cpp_string_escape(comment)[1:-1] + comment_size = len(comment.encode("utf-8")) + 1 # +1 for NUL + return f"""// Auto-generated build_info data +#include "esphome/core/build_info_data.h" + +namespace esphome {{ +const uint32_t ESPHOME_CONFIG_HASH = 0x{config_hash:08x}U; // NOLINT +const time_t ESPHOME_BUILD_TIME = {build_time}; // NOLINT +const size_t ESPHOME_COMMENT_SIZE = {comment_size}; // NOLINT +#ifdef USE_ESP8266 +const char ESPHOME_BUILD_TIME_STR[] PROGMEM = "{build_time_str}"; +const char ESPHOME_COMMENT_STR[] PROGMEM = "{escaped_comment}"; +#else +const char ESPHOME_BUILD_TIME_STR[] = "{build_time_str}"; +const char ESPHOME_COMMENT_STR[] = "{escaped_comment}"; +#endif +}} // namespace esphome """ diff --git a/tests/unit_tests/test_writer.py b/tests/unit_tests/test_writer.py index 9c4a491e100..251913b0209 100644 --- a/tests/unit_tests/test_writer.py +++ b/tests/unit_tests/test_writer.py @@ -7,6 +7,7 @@ from datetime import datetime import json import os from pathlib import Path +import re import stat from typing import Any from unittest.mock import MagicMock, patch @@ -32,6 +33,7 @@ from esphome.writer import ( clean_build, clean_cmake_cache, copy_src_tree, + generate_build_info_data_cpp, generate_build_info_data_h, get_build_info, storage_should_clean, @@ -1617,49 +1619,62 @@ def test_get_build_info_build_time_str_format( def test_generate_build_info_data_h_format() -> None: """Test generate_build_info_data_h produces correct header content.""" - config_hash = 0x12345678 - build_time = 1700000000 - build_time_str = "2023-11-14 22:13:20 +0000" - comment = "Test comment" - - result = generate_build_info_data_h( - config_hash, build_time, build_time_str, comment - ) + result = generate_build_info_data_h() assert "#pragma once" in result - assert "#define ESPHOME_CONFIG_HASH 0x12345678U" in result - assert "#define ESPHOME_BUILD_TIME 1700000000" in result - assert "#define ESPHOME_COMMENT_SIZE 13" in result # len("Test comment") + 1 - assert 'ESPHOME_BUILD_TIME_STR[] = "2023-11-14 22:13:20 +0000"' in result - assert 'ESPHOME_COMMENT_STR[] = "Test comment"' in result + assert "extern const uint32_t ESPHOME_CONFIG_HASH;" in result + assert "extern const time_t ESPHOME_BUILD_TIME;" in result + assert "extern const size_t ESPHOME_COMMENT_SIZE;" in result + assert "extern const char ESPHOME_BUILD_TIME_STR[]" in result + assert "extern const char ESPHOME_COMMENT_STR[]" in result def test_generate_build_info_data_h_esp8266_progmem() -> None: """Test generate_build_info_data_h includes PROGMEM for ESP8266.""" - result = generate_build_info_data_h(0xABCDEF01, 1700000000, "test", "comment") + result = generate_build_info_data_h() # Should have ESP8266 PROGMEM conditional assert "#ifdef USE_ESP8266" in result assert "#include " in result assert "PROGMEM" in result - # Both build time and comment should have PROGMEM versions + + +def test_generate_build_info_data_cpp_format() -> None: + """Test generate_build_info_data_cpp produces correct data definitions.""" + result = generate_build_info_data_cpp( + 0x12345678, 1700000000, "2023-11-14 22:13:20 +0000", "Test comment" + ) + + assert '#include "esphome/core/build_info_data.h"' in result + assert "const uint32_t ESPHOME_CONFIG_HASH = 0x12345678U;" in result + assert "const time_t ESPHOME_BUILD_TIME = 1700000000;" in result + assert "const size_t ESPHOME_COMMENT_SIZE = 13;" in result + assert 'ESPHOME_BUILD_TIME_STR[] = "2023-11-14 22:13:20 +0000"' in result + assert 'ESPHOME_COMMENT_STR[] = "Test comment"' in result + + +def test_generate_build_info_data_cpp_esp8266_progmem() -> None: + """Test generate_build_info_data_cpp includes PROGMEM definitions.""" + result = generate_build_info_data_cpp(0xABCDEF01, 1700000000, "test", "comment") + + assert "#ifdef USE_ESP8266" in result assert 'ESPHOME_BUILD_TIME_STR[] PROGMEM = "test"' in result assert 'ESPHOME_COMMENT_STR[] PROGMEM = "comment"' in result -def test_generate_build_info_data_h_hash_formatting() -> None: - """Test generate_build_info_data_h formats hash with leading zeros.""" +def test_generate_build_info_data_cpp_hash_formatting() -> None: + """Test generate_build_info_data_cpp formats hash with leading zeros.""" # Test with small hash value that needs leading zeros - result = generate_build_info_data_h(0x00000001, 0, "test", "") - assert "#define ESPHOME_CONFIG_HASH 0x00000001U" in result + result = generate_build_info_data_cpp(0x00000001, 0, "test", "") + assert "const uint32_t ESPHOME_CONFIG_HASH = 0x00000001U;" in result # Test with larger hash value - result = generate_build_info_data_h(0xFFFFFFFF, 0, "test", "") - assert "#define ESPHOME_CONFIG_HASH 0xffffffffU" in result + result = generate_build_info_data_cpp(0xFFFFFFFF, 0, "test", "") + assert "const uint32_t ESPHOME_CONFIG_HASH = 0xffffffffU;" in result -def test_generate_build_info_data_h_comment_escaping() -> None: - r"""Test generate_build_info_data_h properly escapes special characters in comment. +def test_generate_build_info_data_cpp_comment_escaping() -> None: + r"""Test generate_build_info_data_cpp properly escapes special characters in comment. Uses cpp_string_escape which outputs octal escapes for special characters: - backslash (ASCII 92) -> \134 @@ -1667,26 +1682,52 @@ def test_generate_build_info_data_h_comment_escaping() -> None: - newline (ASCII 10) -> \012 """ # Test backslash escaping (ASCII 92 = octal 134) - result = generate_build_info_data_h(0, 0, "test", "backslash\\here") + result = generate_build_info_data_cpp(0, 0, "test", "backslash\\here") assert 'ESPHOME_COMMENT_STR[] = "backslash\\134here"' in result # Test quote escaping (ASCII 34 = octal 042) - result = generate_build_info_data_h(0, 0, "test", 'has "quotes"') + result = generate_build_info_data_cpp(0, 0, "test", 'has "quotes"') assert 'ESPHOME_COMMENT_STR[] = "has \\042quotes\\042"' in result # Test newline escaping (ASCII 10 = octal 012) - result = generate_build_info_data_h(0, 0, "test", "line1\nline2") + result = generate_build_info_data_cpp(0, 0, "test", "line1\nline2") assert 'ESPHOME_COMMENT_STR[] = "line1\\012line2"' in result -def test_generate_build_info_data_h_empty_comment() -> None: - """Test generate_build_info_data_h handles empty comment.""" - result = generate_build_info_data_h(0, 0, "test", "") +def test_generate_build_info_data_cpp_empty_comment() -> None: + """Test generate_build_info_data_cpp handles empty comment.""" + result = generate_build_info_data_cpp(0, 0, "test", "") - assert "#define ESPHOME_COMMENT_SIZE 1" in result # Just null terminator + assert "const size_t ESPHOME_COMMENT_SIZE = 1;" in result # Just null terminator assert 'ESPHOME_COMMENT_STR[] = ""' in result +def test_generate_build_info_data_cpp_comment_size_counts_utf8_bytes() -> None: + """Comment size is in encoded UTF-8 bytes, not characters.""" + # "héllo" = 6 UTF-8 bytes + NUL. + result = generate_build_info_data_cpp(0, 0, "test", "héllo") + assert "const size_t ESPHOME_COMMENT_SIZE = 7;" in result + + +def test_generate_build_info_data_cpp_comment_clamped_to_buffer() -> None: + """Generator clamps at byte level and never truncates mid-codepoint.""" + # 100 thermometer-with-VS-16 sequences = 700 bytes, past the 256 buffer. + result = generate_build_info_data_cpp(0, 0, "test", "🌡️" * 100) + + match = re.search(r"ESPHOME_COMMENT_SIZE = (\d+);", result) + assert match is not None + size = int(match.group(1)) + assert 1 < size <= 256 + + lit_match = re.search(r'ESPHOME_COMMENT_STR\[\] = "([^"]*)"', result) + assert lit_match is not None + raw = re.sub( + r"\\([0-7]{3})", lambda m: chr(int(m.group(1), 8)), lit_match.group(1) + ).encode("latin-1") + raw.decode("utf-8") # raises if truncation left a partial UTF-8 sequence + assert len(raw) == size - 1 + + @patch("esphome.writer.CORE") @patch("esphome.writer.iter_components") @patch("esphome.writer.walk_files") @@ -1760,15 +1801,21 @@ def test_copy_src_tree_writes_build_info_files( ): copy_src_tree() - # Verify build_info_data.h was written + # Verify build_info_data.h declarations and build_info_data.cpp values were written build_info_h_path = esphome_core_path / "build_info_data.h" assert build_info_h_path.exists() build_info_h_content = build_info_h_path.read_text() - assert "#define ESPHOME_CONFIG_HASH 0xdeadbeefU" in build_info_h_content - assert "#define ESPHOME_BUILD_TIME" in build_info_h_content + assert "extern const uint32_t ESPHOME_CONFIG_HASH;" in build_info_h_content assert "ESPHOME_BUILD_TIME_STR" in build_info_h_content - assert "#define ESPHOME_COMMENT_SIZE" in build_info_h_content + assert "extern const size_t ESPHOME_COMMENT_SIZE;" in build_info_h_content assert "ESPHOME_COMMENT_STR" in build_info_h_content + build_info_cpp_path = esphome_core_path / "build_info_data.cpp" + assert build_info_cpp_path.exists() + build_info_cpp_content = build_info_cpp_path.read_text() + assert "const uint32_t ESPHOME_CONFIG_HASH = 0xdeadbeefU;" in build_info_cpp_content + assert "const time_t ESPHOME_BUILD_TIME" in build_info_cpp_content + assert "const size_t ESPHOME_COMMENT_SIZE" in build_info_cpp_content + assert "ESPHOME_COMMENT_STR" in build_info_cpp_content # Verify build_info.json was written build_info_json_path = build_path / "build_info.json" @@ -1835,7 +1882,9 @@ def test_copy_src_tree_detects_config_hash_change( # Verify build_info files were updated due to config_hash change assert build_info_h_path.exists() - new_content = build_info_h_path.read_text() + build_info_cpp_path = esphome_core_path / "build_info_data.cpp" + assert build_info_cpp_path.exists() + new_content = build_info_cpp_path.read_text() assert "0xdeadbeef" in new_content.lower() new_json = json.loads(build_info_json_path.read_text())