[core] Extract the shared idedata and size-summary helpers into build_helpers (#18661)

tests/script/test_determine_jobs.py

@@ -1120,7 +1120,13 @@ def test_should_run_esp32_platformio_with_branch() -> None:
         (["esphome/espidf/runner.py"], True),
         (["esphome/espidf/framework.py"], True),
         (["esphome/build_gen/espidf.py"], True),
-        # PlatformIO build gen and esp32 component are NOT IDF-infra triggers
+        # Shared native-build modules the IDF build imports -> trigger
+        (["esphome/build_helpers/idedata.py"], True),
+        (["esphome/platformio/library.py"], True),
+        (["esphome/framework_helpers.py"], True),
+        # PlatformIO build gen, its toolchain, and the esp32 component are
+        # NOT IDF-infra triggers
+        (["esphome/platformio/toolchain.py"], False),
         (["esphome/build_gen/platformio.py"], False),
         (["esphome/components/esp32/__init__.py"], False),
         (["README.md"], False),
@@ -1132,6 +1138,16 @@ def test_esp_idf_infra_changed(changed_files: list[str], expected: bool) -> None
     assert determine_jobs._esp_idf_infra_changed(changed_files) is expected
 
 
+def test_esp_idf_infra_trigger_paths_exist() -> None:
+    """A renamed or moved trigger module must fail here, not silently stop
+    forcing the esp32 IDF compile."""
+    repo_root = Path(__file__).resolve().parents[2]
+    for file in determine_jobs.ESP_IDF_INFRA_TRIGGER_FILES:
+        assert (repo_root / file).is_file(), f"trigger file {file} moved or renamed"
+    for prefix in determine_jobs.ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES:
+        assert (repo_root / prefix).is_dir(), f"trigger dir {prefix} moved or renamed"
+
+
 @pytest.mark.parametrize(
     ("changed_files", "expected_result"),
     [

tests/unit_tests/analyze_memory/test_build_artifacts.py

@@ -9,7 +9,7 @@ from esphome.analyze_memory.toolchain import (
     find_idedata_path,
     idedata_candidates,
 )
-from esphome.espidf.idedata import _cc_path_from_cxx
+from esphome.build_helpers.idedata import _cc_path_from_cxx
 from esphome.platformio.toolchain import IDEData
 
 

tests/unit_tests/build_helpers/test_idedata.py

@@ -0,0 +1,678 @@
+"""Tests for esphome.build_helpers.idedata (compile_commands.json -> idedata)."""
+
+# pylint: disable=protected-access
+
+import json
+import logging
+import os
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from esphome.build_helpers import idedata
+from esphome.core import EsphomeError
+
+# An absolute, forward-slash (shlex-safe) path prefix valid on the host OS, so
+# tests exercise the same is-absolute / normalize behavior as a real compile DB
+# (a drive-qualified path on Windows, a leading slash elsewhere).
+ABS = "C:/" if os.name == "nt" else "/"
+
+
+def _entry(directory: str, file: str, command: str) -> dict:
+    return {"directory": directory, "file": file, "command": command}
+
+
+def test_parse_entry_extracts_fields() -> None:
+    """cxx_path, defines, includes and remaining flags are split apart."""
+    entry = _entry(
+        f"{ABS}build",
+        f"{ABS}build/src/esphome/core/application.cpp",
+        f"/tools/xtensa-esp32-elf-g++ -DUSE_ESP32 -DESPHOME_LOG_LEVEL=5 "
+        f"-I{ABS}inc/a -isystem {ABS}sys/b -std=gnu++20 -c app.cpp -o app.cpp.o",
+    )
+
+    cxx_path, defines, includes, cxx_flags = idedata.parse_entry(entry)
+
+    assert cxx_path == "/tools/xtensa-esp32-elf-g++"
+    assert "USE_ESP32" in defines
+    assert "ESPHOME_LOG_LEVEL=5" in defines
+    assert f"{ABS}inc/a" in includes
+    assert f"{ABS}sys/b" in includes
+    assert "-std=gnu++20" in cxx_flags
+    # input/output files and their flags are not treated as flags
+    assert "-c" not in cxx_flags
+    assert "-o" not in cxx_flags
+    assert "app.cpp" not in cxx_flags
+    assert "app.cpp.o" not in cxx_flags
+
+
+def test_parse_entry_space_separated_args() -> None:
+    """``-D X`` / ``-I path`` (separate arg) and ``-isystem<path>`` (joined)."""
+    entry = _entry(
+        f"{ABS}build",
+        f"{ABS}build/src/esphome/x.cpp",
+        f"g++ -D FOO=1 -I {ABS}inc/sep -isystem{ABS}sys/joined -c x.cpp",
+    )
+
+    _, defines, includes, _ = idedata.parse_entry(entry)
+
+    assert "FOO=1" in defines
+    assert f"{ABS}inc/sep" in includes
+    assert f"{ABS}sys/joined" in includes
+
+
+def test_parse_entry_resolves_relative_includes() -> None:
+    """Relative includes are resolved against the entry's ``directory``."""
+    directory = f"{ABS}build/proj"
+    entry = _entry(
+        directory,
+        f"{directory}/src/esphome/x.cpp",
+        "g++ -Iconfig -I../shared -isystem rel/sys -c x.cpp",
+    )
+
+    _, _, includes, _ = idedata.parse_entry(entry)
+
+    def resolved(rel: str) -> str:
+        # parse_entry emits forward slashes for consistency (normpath would
+        # yield backslashes on Windows).
+        return os.path.normpath(Path(directory) / rel).replace("\\", "/")
+
+    assert resolved("config") in includes
+    assert resolved("../shared") in includes  # ../ normalized away
+    assert resolved("rel/sys") in includes
+    # nothing is left relative
+    assert all(Path(inc).is_absolute() for inc in includes)
+
+
+def test_parse_entry_skips_dependency_flags() -> None:
+    """Dependency-generation flags (and their args) are dropped."""
+    entry = _entry(
+        "/build",
+        "/build/src/esphome/x.cpp",
+        "g++ -MD -MT x.cpp.o -MF x.cpp.o.d -c x.cpp -o x.cpp.o",
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    for tok in ("-MD", "-MT", "x.cpp.o", "-MF", "x.cpp.o.d", "-c", "-o", "x.cpp"):
+        assert tok not in cxx_flags
+
+
+def test_expand_response_files(tmp_path: Path) -> None:
+    """``@file`` arguments are inlined relative to the directory."""
+    rsp = tmp_path / "flags.rsp"
+    rsp.write_text("-DFROM_RSP -I/rsp/inc")
+
+    tokens = idedata._expand_response_files(
+        ["g++", f"@{rsp.name}", "-c", "x.cpp"], tmp_path
+    )
+
+    assert "-DFROM_RSP" in tokens
+    assert "-I/rsp/inc" in tokens
+    assert not any(t.startswith("@") for t in tokens)
+
+
+def test_expand_response_files_keeps_literal_when_missing(tmp_path: Path) -> None:
+    """An unreadable ``@file`` token is kept verbatim rather than dropped."""
+    tokens = idedata._expand_response_files(["g++", "@nope.rsp"], tmp_path)
+    assert "@nope.rsp" in tokens
+
+
+def test_pick_entry_prefers_esphome_tu() -> None:
+    """A ``/src/esphome/`` C++ TU is picked over other compile entries."""
+    entries = [
+        _entry("/b", "/b/managed_components/foo/foo.c", "gcc -c foo.c"),
+        _entry("/b", "/b/src/esphome/core/app.cpp", "g++ -c app.cpp"),
+    ]
+    assert idedata._pick_entry(entries)["file"].endswith("app.cpp")
+
+
+def test_pick_entry_falls_back_to_any_cxx_tu() -> None:
+    """With no ``/src/esphome/`` TU present, the first C++ entry is the fallback."""
+    entries = [
+        _entry("/b", "/b/managed_components/foo/foo.c", "gcc -c foo.c"),
+        _entry("/b", "/b/components/x/x.cpp", "g++ -c x.cpp"),
+    ]
+    assert idedata._pick_entry(entries)["file"].endswith("x.cpp")
+
+
+def test_is_esphome_src_handles_backslash_paths() -> None:
+    r"""The src marker must match Windows ``\src\esphome\`` paths too.
+
+    compile_commands ``file`` entries use the OS-native separator; if the
+    marker only matched forward slashes no source would match on Windows and
+    the build-include union would be silently empty.
+    """
+    assert idedata._is_esphome_src(r"C:\b\src\esphome\core\app.cpp")
+    assert idedata._is_esphome_src("/b/src/esphome/core/app.cpp")
+    # non-esphome and non-C++ still rejected regardless of separator
+    assert not idedata._is_esphome_src(r"C:\b\managed_components\x\x.cpp")
+    assert not idedata._is_esphome_src(r"C:\b\src\esphome\core\app.h")
+
+
+@pytest.mark.parametrize(
+    ("command", "launcher"),
+    [
+        ("", None),
+        # A command that is only the launcher strips to nothing
+        ("/usr/bin/ccache", "/usr/bin/ccache"),
+    ],
+)
+def test_parse_entry_empty_command_raises(command: str, launcher: str | None) -> None:
+    """A blank (or launcher-only) command fails with a named ValueError,
+    not an IndexError."""
+    entry = {"directory": "/b", "file": "/b/src/x.cpp", "command": command}
+    with pytest.raises(ValueError, match="empty compile command"):
+        idedata.parse_entry(entry, launcher)
+
+
+def test_idedata_from_build_empty_includes_raises(tmp_path: Path) -> None:
+    """A compile DB with no ESPHome TU is never usable idedata and must
+    not be cached (call sites downgrade the raise to a build warning)."""
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(
+        json.dumps(
+            [
+                _entry(
+                    f"{ABS}build",
+                    f"{ABS}build/other/lib.cpp",
+                    "/tools/g++ -c other/lib.cpp -o lib.o",
+                )
+            ]
+        )
+    )
+    with (
+        patch.object(idedata, "get_toolchain_includes", return_value=[]),
+        pytest.raises(EsphomeError, match="No ESPHome translation unit found"),
+    ):
+        idedata.idedata_from_build(compile_commands)
+
+
+def test_idedata_from_build_rsp_commands_never_dedupe(tmp_path: Path) -> None:
+    """Per-object response files strip to one shape while holding different
+    include sets; @-commands must tokenize per TU."""
+    entries = []
+    for name in ("a", "b"):
+        rsp = tmp_path / f"{name}.cpp.o.rsp"
+        rsp.write_text(f"-I{ABS}inc/{name}")
+        file = f"{ABS}build/src/esphome/core/{name}.cpp"
+        entries.append(
+            {
+                "directory": str(tmp_path),
+                "file": file,
+                "command": f"/tools/g++ @{rsp.name} -c {file} -o {name}.o",
+                "output": f"{name}.o",
+            }
+        )
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(json.dumps(entries))
+    with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+        data = idedata.idedata_from_build(compile_commands)
+    joined = " ".join(data["includes"]["build"])
+    assert "inc/a" in joined and "inc/b" in joined
+
+
+def test_idedata_from_build_dedupes_identical_command_shapes(
+    tmp_path: Path,
+) -> None:
+    """Translation units sharing one ninja rule (same command modulo
+    file/output) carry
+    identical includes, so only one per shape is tokenized; a differing
+    shape still contributes its includes."""
+
+    def _tu(name: str, inc: str) -> dict:
+        # ninja's compdb embeds the file and output strings verbatim
+        file = f"{ABS}build/src/esphome/core/{name}.cpp"
+        return _entry(
+            f"{ABS}build", file, f"/tools/g++ -I{ABS}inc/{inc} -c {file} -o {name}.o"
+        ) | {"output": f"{name}.o"}
+
+    entries = [_tu(name, "shared") for name in ("application", "component", "helpers")]
+    entries.append(_tu("extra", "extra"))
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(json.dumps(entries))
+    with (
+        patch.object(idedata, "get_toolchain_includes", return_value=[]),
+        patch.object(idedata, "parse_entry", wraps=idedata.parse_entry) as spy,
+    ):
+        data = idedata.idedata_from_build(compile_commands)
+    includes = set(data["includes"]["build"])
+    assert f"{ABS}inc/shared".replace("\\", "/") in {
+        i.replace("\\", "/") for i in includes
+    }
+    assert any("inc/extra" in i for i in includes)
+    # Representative + one distinct shape; the two same-shape duplicates
+    # are never tokenized
+    assert spy.call_count == 2
+
+
+def test_idedata_from_build(tmp_path: Path) -> None:
+    """Full transform: representative entry + include union + toolchain dirs."""
+    compile_commands = tmp_path / "compile_commands.json"
+    entries = [
+        _entry(
+            f"{ABS}b",
+            f"{ABS}b/src/esphome/core/app.cpp",
+            f"g++ -DUSE_ESP32 -I{ABS}inc/core -std=gnu++20 -c app.cpp -o app.cpp.o",
+        ),
+        _entry(
+            f"{ABS}b",
+            f"{ABS}b/src/esphome/sensor/s.cpp",
+            f"g++ -DUSE_ESP32 -I{ABS}inc/sensor -c s.cpp -o s.cpp.o",
+        ),
+        # non-esphome TU: its includes must not leak into the union
+        _entry(
+            f"{ABS}b",
+            f"{ABS}b/managed_components/x/x.c",
+            f"gcc -I{ABS}inc/managed -c x.c",
+        ),
+    ]
+    compile_commands.write_text(json.dumps(entries))
+
+    fake_proc = MagicMock(
+        returncode=0,
+        stderr=(
+            "ignored\n"
+            "#include <...> search starts here:\n"
+            " /tc/inc/c++\n"
+            " /tc/inc\n"
+            "End of search list.\n"
+            "more ignored\n"
+        ),
+    )
+    with patch.object(idedata.subprocess, "run", return_value=fake_proc):
+        data = idedata.idedata_from_build(compile_commands)
+
+    assert data["cxx_path"] == "g++"
+    assert "USE_ESP32" in data["defines"]
+    assert "-std=gnu++20" in data["cxx_flags"]
+    # include dirs unioned across all esphome TUs
+    assert f"{ABS}inc/core" in data["includes"]["build"]
+    assert f"{ABS}inc/sensor" in data["includes"]["build"]
+    # the non-esphome TU is excluded from the union
+    assert f"{ABS}inc/managed" not in data["includes"]["build"]
+    # toolchain search dirs parsed from the compiler's -v output
+    assert data["includes"]["toolchain"] == ["/tc/inc/c++", "/tc/inc"]
+
+
+def test_get_toolchain_includes_raises_on_probe_failure() -> None:
+    """A failed compiler probe is a hard error, not a silent empty list."""
+    fake_proc = MagicMock(returncode=1, stderr="xtensa-esp32-elf-g++: not found")
+    with (
+        patch.object(idedata.subprocess, "run", return_value=fake_proc),
+        pytest.raises(RuntimeError, match="builtin include dirs"),
+    ):
+        idedata.get_toolchain_includes("/bad/compiler")
+
+
+def test_get_toolchain_includes_raises_when_no_dirs_found() -> None:
+    """Markers present but no dirs (anomalous output) also raises."""
+    fake_proc = MagicMock(
+        returncode=0,
+        stderr="#include <...> search starts here:\nEnd of search list.\n",
+    )
+    with (
+        patch.object(idedata.subprocess, "run", return_value=fake_proc),
+        pytest.raises(RuntimeError, match="builtin include dirs"),
+    ):
+        idedata.get_toolchain_includes("/some/compiler")
+
+
+# ESP-IDF's compile_commands.json on Windows mixes literal backslash path
+# separators in the compiler path with shell ``\"`` quote-escaping in defines,
+# which only the real Windows argv parser handles. These exercise that path.
+@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
+def test_split_command_preserves_paths_and_unescapes_quotes() -> None:
+    r"""Backslash paths survive while ``\"`` define-quoting is unescaped."""
+    command = r"C:\esp\bin\riscv32-esp-elf-g++.exe -DVER=\"1.2.3\" -IC:/inc/a -c x.cpp"
+
+    tokens = idedata._split_command(command)
+
+    assert tokens[0] == r"C:\esp\bin\riscv32-esp-elf-g++.exe"
+    assert '-DVER="1.2.3"' in tokens
+    assert "-IC:/inc/a" in tokens
+
+
+@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
+def test_split_command_empty_returns_empty() -> None:
+    """An empty or blank command tokenizes to ``[]`` (e.g. an empty response file).
+
+    Guards against ``CommandLineToArgvW("")`` returning the current process name
+    instead of an empty list.
+    """
+    assert idedata._split_command("") == []
+    assert idedata._split_command("   ") == []
+
+
+@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
+def test_parse_entry_normalizes_windows_cxx_path() -> None:
+    """A backslash compiler path is emitted forward-slashed; define unescaped."""
+    entry = _entry(
+        r"C:\b",
+        r"C:\b\src\esphome\x.cpp",
+        r"C:\esp\bin\g++.exe -DVER=\"1.2.3\" -IC:/inc/a -c x.cpp",
+    )
+
+    cxx_path, defines, includes, _ = idedata.parse_entry(entry)
+
+    assert cxx_path == "C:/esp/bin/g++.exe"
+    assert "\\" not in cxx_path
+    assert 'VER="1.2.3"' in defines
+    assert "C:/inc/a" in includes
+
+
+def test_parse_entry_strips_launcher_prefix() -> None:
+    """A launcher-wrapped compile names the compiler second; the exact
+    configured launcher is stripped, not anything ccache-shaped."""
+    entry = _entry(
+        f"{ABS}build",
+        f"{ABS}build/src/esphome/core/application.cpp",
+        "/opt/homebrew/bin/ccache /tools/xtensa-lx106-elf-g++ -DUSE_ESP8266 "
+        "-c app.cpp -o app.cpp.o",
+    )
+    cxx_path, defines, _, _ = idedata.parse_entry(
+        entry, launcher="/opt/homebrew/bin/ccache"
+    )
+    assert cxx_path == "/tools/xtensa-lx106-elf-g++"
+    assert defines == ["USE_ESP8266"]
+
+
+def test_parse_entry_recovers_from_unconfigured_launcher(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A stale compile DB built with a launcher this run no longer configures
+    still yields the real compiler (the next token), not the launcher."""
+    entry = _entry(
+        f"{ABS}build",
+        f"{ABS}build/src/esphome/core/application.cpp",
+        "/opt/homebrew/bin/ccache /tools/xtensa-lx106-elf-g++ -c a.cpp -o a.o",
+    )
+    caplog.set_level(logging.DEBUG)
+    cxx_path, _, _, _ = idedata.parse_entry(entry)
+    assert cxx_path == "/tools/xtensa-lx106-elf-g++"
+    assert "Stripping unconfigured launcher" in caplog.text
+
+
+def test_parse_entry_rejects_launcher_without_program() -> None:
+    """A launcher followed only by flags is rejected in the parser itself,
+    so no caller can record ccache as the compiler."""
+    entry = _entry(
+        f"{ABS}build",
+        f"{ABS}build/src/esphome/core/application.cpp",
+        "/opt/homebrew/bin/ccache -c a.cpp -o a.o",
+    )
+    with pytest.raises(EsphomeError, match="compile database is unusable"):
+        idedata.parse_entry(entry)
+
+
+def _write_compile_commands(tmp_path: Path) -> Path:
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(
+        json.dumps(
+            [
+                _entry(
+                    f"{ABS}build",
+                    f"{ABS}build/src/esphome/core/application.cpp",
+                    "/tools/g++ -DUSE_ESP8266 -c app.cpp -o app.cpp.o",
+                )
+            ]
+        )
+    )
+    return compile_commands
+
+
+def test_load_or_build_idedata_missing_compile_db(tmp_path: Path) -> None:
+    assert (
+        idedata.load_or_build_idedata(
+            tmp_path / "compile_commands.json", tmp_path / "f.elf", tmp_path / "c.json"
+        )
+        is None
+    )
+
+
+def test_load_or_build_idedata_builds_and_caches(tmp_path: Path) -> None:
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "cache" / "test.json"
+    with patch.object(
+        idedata, "get_toolchain_includes", return_value=["/toolchain/include"]
+    ):
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "firmware.elf", cache
+        )
+    assert data["cc_path"] == "/tools/gcc"
+    assert data["prog_path"] == str(tmp_path / "firmware.elf")
+    assert json.loads(cache.read_text()) == data
+
+    # A fresh cache is served without re-parsing the compile DB
+    os.utime(cache, (compile_commands.stat().st_mtime + 10,) * 2)
+    with patch.object(idedata, "idedata_from_build") as mock_build:
+        assert (
+            idedata.load_or_build_idedata(
+                compile_commands, tmp_path / "firmware.elf", cache
+            )
+            == data
+        )
+    mock_build.assert_not_called()
+
+
+def test_load_or_build_idedata_rebuilds_bad_cache(tmp_path: Path) -> None:
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "cache.json"
+    for bad in ("not json", json.dumps({"no_cc_path": True})):
+        cache.write_text(bad)
+        os.utime(cache, (compile_commands.stat().st_mtime + 10,) * 2)
+        with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+            data = idedata.load_or_build_idedata(
+                compile_commands, tmp_path / "f.elf", cache
+            )
+        assert "cc_path" in data
+
+
+def test_load_or_build_idedata_rebuilds_when_compile_db_newer(tmp_path: Path) -> None:
+    """A compile DB newer than the cache forces regeneration."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "cache.json"
+    cache.write_text(json.dumps({"cc_path": "stale"}))
+    os.utime(compile_commands, (cache.stat().st_mtime + 10,) * 2)
+    with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "f.elf", cache
+        )
+    assert data["cc_path"] != "stale"
+
+
+def test_load_or_build_idedata_rebuilds_non_dict_cache(tmp_path: Path) -> None:
+    """Valid JSON that is not an object is regenerated, never handed out.
+
+    A bare string would otherwise pass the cc_path check by substring.
+    """
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "cache.json"
+    for bad in ('"cc_path is a string"', "[]", "42"):
+        cache.write_text(bad)
+        os.utime(cache, (compile_commands.stat().st_mtime + 10,) * 2)
+        with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+            data = idedata.load_or_build_idedata(
+                compile_commands, tmp_path / "f.elf", cache
+            )
+        assert isinstance(data, dict)
+        assert "cc_path" in data
+
+
+def test_is_launcher_matches_only_known_launchers() -> None:
+    """Compilers of any shape pass; only the closed launcher set matches."""
+    for token in ("/t/g++-13", "gcc-8.4.0", "clang++-17", "armcc", "icx", "cc"):
+        assert not idedata._is_launcher(token)
+    for token in ("/opt/homebrew/bin/ccache", "CCACHE.EXE", "distcc", "sccache"):
+        assert idedata._is_launcher(token)
+
+
+def test_load_or_build_idedata_corrupted_cache_is_logged(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A truncated cache is diagnosable, not a silent slow-build cause."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "c.json"
+    cache.write_text('{"cc_path": trunc')
+    os.utime(cache, (compile_commands.stat().st_mtime + 5,) * 2)
+    with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "f.elf", cache
+        )
+    assert data["cxx_path"] == "/tools/g++"
+    assert "Discarding unreadable idedata cache" in caplog.text
+
+
+def test_load_or_build_idedata_discards_unreadable_cache_file(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """An OSError on the cache read (permissions, I/O) regenerates like a
+    parse failure instead of aborting the consumer."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "c.json"
+    cache.write_text("{}")
+    os.utime(cache, (compile_commands.stat().st_mtime + 5,) * 2)
+    real_read_text = Path.read_text
+
+    def fail_cache_read(self: Path, *args: object, **kwargs: object) -> str:
+        # chmod(0) cannot revoke read access on Windows, so fault the read
+        # itself for a platform-independent OSError
+        if self == cache:
+            raise OSError("permission denied")
+        return real_read_text(self, *args, **kwargs)
+
+    with (
+        patch.object(idedata, "get_toolchain_includes", return_value=[]),
+        patch.object(Path, "read_text", fail_cache_read),
+    ):
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "f.elf", cache
+        )
+    assert data["cxx_path"] == "/tools/g++"
+    assert "Discarding unreadable idedata cache" in caplog.text
+
+
+def test_load_or_build_idedata_never_caches_a_launcher(tmp_path: Path) -> None:
+    """A compile DB naming a launcher as the compiler is rejected, never cached."""
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(
+        json.dumps(
+            [
+                _entry(
+                    f"{ABS}build",
+                    f"{ABS}build/src/esphome/core/application.cpp",
+                    "/opt/homebrew/bin/ccache -c app.cpp -o app.cpp.o",
+                )
+            ]
+        )
+    )
+    cache = tmp_path / "c.json"
+    # No probe patch needed: the launcher is rejected before the probe runs
+    with pytest.raises(EsphomeError, match="compile database is unusable"):
+        idedata.load_or_build_idedata(compile_commands, tmp_path / "f.elf", cache)
+    assert not cache.exists()
+
+
+@pytest.mark.parametrize(
+    "cached",
+    (
+        {"cc_path": "/x/gcc", "cxx_path": "/opt/homebrew/bin/ccache"},
+        {"cc_path": "/x/gcc", "cxx_path": "/tools/g++"},
+        {"cc_path": "/x/gcc", "cxx_path": "/tools/g++", "includes": {}},
+    ),
+    ids=("launcher-cxx", "no-includes", "no-build-list"),
+)
+def test_load_or_build_idedata_regenerates_invalid_cache(
+    tmp_path: Path, cached: dict
+) -> None:
+    """A cache written by an older version fails validation and regenerates."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "c.json"
+    cache.write_text(json.dumps(cached))
+    os.utime(cache, (compile_commands.stat().st_mtime + 5,) * 2)
+    with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "f.elf", cache
+        )
+    assert data["cxx_path"] == "/tools/g++"
+    assert "includes" in data
+
+
+def test_load_or_build_idedata_cache_hit_restamps_prog_path(tmp_path: Path) -> None:
+    """A served cache carries the current ELF path, not the one it was written with."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "c.json"
+    cache.write_text(
+        json.dumps(
+            {
+                "cc_path": "/tools/gcc",
+                "cxx_path": "/tools/g++",
+                "includes": {"build": [], "toolchain": []},
+                "prog_path": "/old/location/firmware.elf",
+            }
+        )
+    )
+    os.utime(cache, (compile_commands.stat().st_mtime + 5,) * 2)
+    data = idedata.load_or_build_idedata(
+        compile_commands, tmp_path / "firmware.elf", cache
+    )
+    assert data["prog_path"] == str(tmp_path / "firmware.elf")
+
+
+def test_idedata_from_build_non_list_compile_db_raises(tmp_path: Path) -> None:
+    """Valid JSON that is not a list raises by name, inside the best-effort tuple."""
+    compile_commands = tmp_path / "compile_commands.json"
+    for bad in ("{}", "null", '"text"', '["a", "b"]', "[1, 2]"):
+        compile_commands.write_text(bad)
+        with pytest.raises(EsphomeError, match="not a compile-command list"):
+            idedata.idedata_from_build(compile_commands)
+
+
+def test_idedata_from_build_same_file_rsp_commands_never_dedupe(
+    tmp_path: Path,
+) -> None:
+    """Two objects built from one source with different .rsp files keep both
+    include sets; the rsp sentinel keys on the output, not the source."""
+    file = f"{ABS}build/src/esphome/core/shared.cpp"
+    entries = []
+    for name in ("a", "b"):
+        rsp = tmp_path / f"{name}.o.rsp"
+        rsp.write_text(f"-I{ABS}inc/{name}")
+        entries.append(
+            {
+                "directory": str(tmp_path),
+                "file": file,
+                "command": f"/tools/g++ @{rsp.name} -c {file} -o {name}.o",
+                "output": f"{name}.o",
+            }
+        )
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(json.dumps(entries))
+    with patch.object(idedata, "get_toolchain_includes", return_value=[]):
+        data = idedata.idedata_from_build(compile_commands)
+    joined = " ".join(data["includes"]["build"])
+    assert "inc/a" in joined and "inc/b" in joined
+
+
+def test_load_or_build_idedata_cache_hit_skips_rebuild(tmp_path: Path) -> None:
+    """A valid cache newer than the compile DB is served without re-parsing."""
+    compile_commands = _write_compile_commands(tmp_path)
+    cache = tmp_path / "c.json"
+    cache.write_text(
+        json.dumps(
+            {
+                "cc_path": "/tools/gcc",
+                "cxx_path": "/tools/g++",
+                "includes": {"build": ["/inc"], "toolchain": []},
+                "cached": True,
+            }
+        )
+    )
+    os.utime(cache, (compile_commands.stat().st_mtime + 5,) * 2)
+    with patch.object(idedata, "idedata_from_build") as mock_build:
+        data = idedata.load_or_build_idedata(
+            compile_commands, tmp_path / "f.elf", cache
+        )
+    mock_build.assert_not_called()
+    assert data["cached"] is True

tests/unit_tests/build_helpers/test_size_summary.py

@@ -0,0 +1,22 @@
+"""Tests for the shared PlatformIO-format size bar."""
+
+from __future__ import annotations
+
+import pytest
+
+from esphome.build_helpers.size_summary import format_bar, print_size_line
+
+
+def test_format_bar_zero_total() -> None:
+    """A zero total must not divide by zero."""
+    assert format_bar(0, 0) == "[          ]   0.0% (used 0 bytes from 0 bytes)"
+
+
+def test_print_size_line_label_padding(capsys: pytest.CaptureFixture[str]) -> None:
+    """The label column is exactly what ci_memory_impact_extract.py greps."""
+    print_size_line("RAM", 47932, 180736)
+    print_size_line("Flash", 888511, 1835008)
+    out = capsys.readouterr().out.splitlines()
+    assert out[0].startswith("RAM:   [")
+    assert out[1].startswith("Flash: [")
+    assert "26.5% (used 47932 bytes from 180736 bytes)" in out[0]

tests/unit_tests/test_core.py

@@ -6,9 +6,9 @@ from unittest.mock import patch
 
 from hypothesis import given
 import pytest
-from strategies import mac_addr_strings
 
 from esphome import const, core
+from tests.unit_tests.strategies import mac_addr_strings
 
 
 class TestHexInt:

tests/unit_tests/test_espidf_clang_tidy.py

@@ -1,10 +1,13 @@
 """Tests for esphome.espidf.clang_tidy tidy-project generation."""
 
+import json
 import os
 from pathlib import Path
+from unittest.mock import patch
 
 import pytest
 
+from esphome.espidf import clang_tidy
 from esphome.espidf.clang_tidy import _Settings, _setup_core, _write_tidy_project
 
 REPO_ROOT = Path(__file__).resolve().parents[2]
@@ -64,3 +67,35 @@ def test_setup_core_sets_arduino_env(
     _setup_core(tmp_path / "proj", _settings(target_framework=target_framework))
 
     assert os.environ["ESPHOME_ARDUINO_COMPONENT"] == expected
+
+
+def test_idedata_from_tidy_project(tmp_path) -> None:
+    """The tidy TU's compile entry is assembled into consumer-shaped idedata."""
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(
+        json.dumps(
+            [
+                {
+                    "directory": str(tmp_path),
+                    "file": str(tmp_path / "main" / "tidy.cpp"),
+                    "command": "/tc/xtensa-esp32-elf-g++ -DUSE_ESP32 "
+                    f"-I{tmp_path}/inc -c main/tidy.cpp -o tidy.o",
+                }
+            ]
+        )
+    )
+    with patch(
+        "esphome.espidf.clang_tidy.get_toolchain_includes", return_value=["/tc/inc"]
+    ):
+        data = clang_tidy._idedata_from_tidy_project(compile_commands)
+    assert data["cxx_path"] == "/tc/xtensa-esp32-elf-g++"
+    assert data["defines"] == ["USE_ESP32"]
+    assert data["includes"]["toolchain"] == ["/tc/inc"]
+    assert any(inc.endswith("/inc") for inc in data["includes"]["build"])
+
+
+def test_idedata_from_tidy_project_missing_tu_raises(tmp_path) -> None:
+    compile_commands = tmp_path / "compile_commands.json"
+    compile_commands.write_text(json.dumps([]))
+    with pytest.raises(RuntimeError, match="tidy.cpp not found"):
+        clang_tidy._idedata_from_tidy_project(compile_commands)

tests/unit_tests/test_espidf_idedata.py

@@ -1,264 +0,0 @@
-"""Tests for esphome.espidf.idedata (compile_commands.json -> idedata)."""
-
-# pylint: disable=protected-access
-
-import json
-import os
-from pathlib import Path
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-from esphome.espidf import idedata
-
-# An absolute, forward-slash (shlex-safe) path prefix valid on the host OS, so
-# tests exercise the same is-absolute / normalize behavior as a real compile DB
-# (a drive-qualified path on Windows, a leading slash elsewhere).
-ABS = "C:/" if os.name == "nt" else "/"
-
-
-def _entry(directory: str, file: str, command: str) -> dict:
-    return {"directory": directory, "file": file, "command": command}
-
-
-def test_parse_entry_extracts_fields() -> None:
-    """cxx_path, defines, includes and remaining flags are split apart."""
-    entry = _entry(
-        f"{ABS}build",
-        f"{ABS}build/src/esphome/core/application.cpp",
-        f"/tools/xtensa-esp32-elf-g++ -DUSE_ESP32 -DESPHOME_LOG_LEVEL=5 "
-        f"-I{ABS}inc/a -isystem {ABS}sys/b -std=gnu++20 -c app.cpp -o app.cpp.o",
-    )
-
-    cxx_path, defines, includes, cxx_flags = idedata._parse_entry(entry)
-
-    assert cxx_path == "/tools/xtensa-esp32-elf-g++"
-    assert "USE_ESP32" in defines
-    assert "ESPHOME_LOG_LEVEL=5" in defines
-    assert f"{ABS}inc/a" in includes
-    assert f"{ABS}sys/b" in includes
-    assert "-std=gnu++20" in cxx_flags
-    # input/output files and their flags are not treated as flags
-    assert "-c" not in cxx_flags
-    assert "-o" not in cxx_flags
-    assert "app.cpp" not in cxx_flags
-    assert "app.cpp.o" not in cxx_flags
-
-
-def test_parse_entry_space_separated_args() -> None:
-    """``-D X`` / ``-I path`` (separate arg) and ``-isystem<path>`` (joined)."""
-    entry = _entry(
-        f"{ABS}build",
-        f"{ABS}build/src/esphome/x.cpp",
-        f"g++ -D FOO=1 -I {ABS}inc/sep -isystem{ABS}sys/joined -c x.cpp",
-    )
-
-    _, defines, includes, _ = idedata._parse_entry(entry)
-
-    assert "FOO=1" in defines
-    assert f"{ABS}inc/sep" in includes
-    assert f"{ABS}sys/joined" in includes
-
-
-def test_parse_entry_resolves_relative_includes() -> None:
-    """Relative includes are resolved against the entry's ``directory``."""
-    directory = f"{ABS}build/proj"
-    entry = _entry(
-        directory,
-        f"{directory}/src/esphome/x.cpp",
-        "g++ -Iconfig -I../shared -isystem rel/sys -c x.cpp",
-    )
-
-    _, _, includes, _ = idedata._parse_entry(entry)
-
-    def resolved(rel: str) -> str:
-        # _parse_entry emits forward slashes for consistency (normpath would
-        # yield backslashes on Windows).
-        return os.path.normpath(Path(directory) / rel).replace("\\", "/")
-
-    assert resolved("config") in includes
-    assert resolved("../shared") in includes  # ../ normalized away
-    assert resolved("rel/sys") in includes
-    # nothing is left relative
-    assert all(Path(inc).is_absolute() for inc in includes)
-
-
-def test_parse_entry_skips_dependency_flags() -> None:
-    """Dependency-generation flags (and their args) are dropped."""
-    entry = _entry(
-        "/build",
-        "/build/src/esphome/x.cpp",
-        "g++ -MD -MT x.cpp.o -MF x.cpp.o.d -c x.cpp -o x.cpp.o",
-    )
-
-    _, _, _, cxx_flags = idedata._parse_entry(entry)
-
-    for tok in ("-MD", "-MT", "x.cpp.o", "-MF", "x.cpp.o.d", "-c", "-o", "x.cpp"):
-        assert tok not in cxx_flags
-
-
-def test_expand_response_files(tmp_path: Path) -> None:
-    """``@file`` arguments are inlined relative to the directory."""
-    rsp = tmp_path / "flags.rsp"
-    rsp.write_text("-DFROM_RSP -I/rsp/inc")
-
-    tokens = idedata._expand_response_files(
-        ["g++", f"@{rsp.name}", "-c", "x.cpp"], tmp_path
-    )
-
-    assert "-DFROM_RSP" in tokens
-    assert "-I/rsp/inc" in tokens
-    assert not any(t.startswith("@") for t in tokens)
-
-
-def test_expand_response_files_keeps_literal_when_missing(tmp_path: Path) -> None:
-    """An unreadable ``@file`` token is kept verbatim rather than dropped."""
-    tokens = idedata._expand_response_files(["g++", "@nope.rsp"], tmp_path)
-    assert "@nope.rsp" in tokens
-
-
-def test_pick_entry_prefers_esphome_tu() -> None:
-    """A ``/src/esphome/`` C++ TU is picked over other compile entries."""
-    entries = [
-        _entry("/b", "/b/managed_components/foo/foo.c", "gcc -c foo.c"),
-        _entry("/b", "/b/src/esphome/core/app.cpp", "g++ -c app.cpp"),
-    ]
-    assert idedata._pick_entry(entries)["file"].endswith("app.cpp")
-
-
-def test_pick_entry_falls_back_to_any_cxx_tu() -> None:
-    """With no ``/src/esphome/`` TU present, the first C++ entry is the fallback."""
-    entries = [
-        _entry("/b", "/b/managed_components/foo/foo.c", "gcc -c foo.c"),
-        _entry("/b", "/b/components/x/x.cpp", "g++ -c x.cpp"),
-    ]
-    assert idedata._pick_entry(entries)["file"].endswith("x.cpp")
-
-
-def test_is_esphome_src_handles_backslash_paths() -> None:
-    r"""The src marker must match Windows ``\src\esphome\`` paths too.
-
-    compile_commands ``file`` entries use the OS-native separator; if the
-    marker only matched forward slashes no source would match on Windows and
-    the build-include union would be silently empty.
-    """
-    assert idedata._is_esphome_src(r"C:\b\src\esphome\core\app.cpp")
-    assert idedata._is_esphome_src("/b/src/esphome/core/app.cpp")
-    # non-esphome and non-C++ still rejected regardless of separator
-    assert not idedata._is_esphome_src(r"C:\b\managed_components\x\x.cpp")
-    assert not idedata._is_esphome_src(r"C:\b\src\esphome\core\app.h")
-
-
-def test_idedata_from_build(tmp_path: Path) -> None:
-    """Full transform: representative entry + include union + toolchain dirs."""
-    compile_commands = tmp_path / "compile_commands.json"
-    entries = [
-        _entry(
-            f"{ABS}b",
-            f"{ABS}b/src/esphome/core/app.cpp",
-            f"g++ -DUSE_ESP32 -I{ABS}inc/core -std=gnu++20 -c app.cpp -o app.cpp.o",
-        ),
-        _entry(
-            f"{ABS}b",
-            f"{ABS}b/src/esphome/sensor/s.cpp",
-            f"g++ -DUSE_ESP32 -I{ABS}inc/sensor -c s.cpp -o s.cpp.o",
-        ),
-        # non-esphome TU: its includes must not leak into the union
-        _entry(
-            f"{ABS}b",
-            f"{ABS}b/managed_components/x/x.c",
-            f"gcc -I{ABS}inc/managed -c x.c",
-        ),
-    ]
-    compile_commands.write_text(json.dumps(entries))
-
-    fake_proc = MagicMock(
-        returncode=0,
-        stderr=(
-            "ignored\n"
-            "#include <...> search starts here:\n"
-            " /tc/inc/c++\n"
-            " /tc/inc\n"
-            "End of search list.\n"
-            "more ignored\n"
-        ),
-    )
-    with patch.object(idedata.subprocess, "run", return_value=fake_proc):
-        data = idedata.idedata_from_build(compile_commands)
-
-    assert data["cxx_path"] == "g++"
-    assert "USE_ESP32" in data["defines"]
-    assert "-std=gnu++20" in data["cxx_flags"]
-    # include dirs unioned across all esphome TUs
-    assert f"{ABS}inc/core" in data["includes"]["build"]
-    assert f"{ABS}inc/sensor" in data["includes"]["build"]
-    # the non-esphome TU is excluded from the union
-    assert f"{ABS}inc/managed" not in data["includes"]["build"]
-    # toolchain search dirs parsed from the compiler's -v output
-    assert data["includes"]["toolchain"] == ["/tc/inc/c++", "/tc/inc"]
-
-
-def test_get_toolchain_includes_raises_on_probe_failure() -> None:
-    """A failed compiler probe is a hard error, not a silent empty list."""
-    fake_proc = MagicMock(returncode=1, stderr="xtensa-esp32-elf-g++: not found")
-    with (
-        patch.object(idedata.subprocess, "run", return_value=fake_proc),
-        pytest.raises(RuntimeError, match="builtin include dirs"),
-    ):
-        idedata._get_toolchain_includes("/bad/compiler")
-
-
-def test_get_toolchain_includes_raises_when_no_dirs_found() -> None:
-    """Markers present but no dirs (anomalous output) also raises."""
-    fake_proc = MagicMock(
-        returncode=0,
-        stderr="#include <...> search starts here:\nEnd of search list.\n",
-    )
-    with (
-        patch.object(idedata.subprocess, "run", return_value=fake_proc),
-        pytest.raises(RuntimeError, match="builtin include dirs"),
-    ):
-        idedata._get_toolchain_includes("/some/compiler")
-
-
-# ESP-IDF's compile_commands.json on Windows mixes literal backslash path
-# separators in the compiler path with shell ``\"`` quote-escaping in defines,
-# which only the real Windows argv parser handles. These exercise that path.
-@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
-def test_split_command_preserves_paths_and_unescapes_quotes() -> None:
-    r"""Backslash paths survive while ``\"`` define-quoting is unescaped."""
-    command = r"C:\esp\bin\riscv32-esp-elf-g++.exe -DVER=\"1.2.3\" -IC:/inc/a -c x.cpp"
-
-    tokens = idedata._split_command(command)
-
-    assert tokens[0] == r"C:\esp\bin\riscv32-esp-elf-g++.exe"
-    assert '-DVER="1.2.3"' in tokens
-    assert "-IC:/inc/a" in tokens
-
-
-@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
-def test_split_command_empty_returns_empty() -> None:
-    """An empty or blank command tokenizes to ``[]`` (e.g. an empty response file).
-
-    Guards against ``CommandLineToArgvW("")`` returning the current process name
-    instead of an empty list.
-    """
-    assert idedata._split_command("") == []
-    assert idedata._split_command("   ") == []
-
-
-@pytest.mark.skipif(os.name != "nt", reason="Windows argv tokenization")
-def test_parse_entry_normalizes_windows_cxx_path() -> None:
-    """A backslash compiler path is emitted forward-slashed; define unescaped."""
-    entry = _entry(
-        r"C:\b",
-        r"C:\b\src\esphome\x.cpp",
-        r"C:\esp\bin\g++.exe -DVER=\"1.2.3\" -IC:/inc/a -c x.cpp",
-    )
-
-    cxx_path, defines, includes, _ = idedata._parse_entry(entry)
-
-    assert cxx_path == "C:/esp/bin/g++.exe"
-    assert "\\" not in cxx_path
-    assert 'VER="1.2.3"' in defines
-    assert "C:/inc/a" in includes

tests/unit_tests/test_espidf_toolchain.py

@@ -140,7 +140,7 @@ def test_get_idedata_generates_and_caches(setup_core: Path) -> None:
     compile_commands.write_text("[]")
 
     with patch(
-        "esphome.espidf.idedata.idedata_from_build",
+        "esphome.build_helpers.idedata.idedata_from_build",
         return_value={"cxx_path": "g++"},
     ) as mock_transform:
         result = toolchain.get_idedata()
@@ -151,114 +151,6 @@ def test_get_idedata_generates_and_caches(setup_core: Path) -> None:
     assert json.loads(cache.read_text()) == {"cxx_path": "g++", "prog_path": prog_path}
 
 
-def test_get_idedata_uses_cache_when_valid(setup_core: Path) -> None:
-    """A cache at least as new as the compile DB is reused without regenerating."""
-    compile_commands, cache = _setup_build(setup_core)
-    compile_commands.parent.mkdir(parents=True, exist_ok=True)
-    compile_commands.write_text("[]")
-    cache.parent.mkdir(parents=True, exist_ok=True)
-    cache.write_text('{"cc_path": "cached-gcc", "cxx_path": "cached"}')
-    cc_mtime = compile_commands.stat().st_mtime
-    os.utime(cache, (cc_mtime + 1, cc_mtime + 1))
-
-    with patch("esphome.espidf.idedata.idedata_from_build") as mock_transform:
-        result = toolchain.get_idedata()
-
-    mock_transform.assert_not_called()
-    assert result == {"cc_path": "cached-gcc", "cxx_path": "cached"}
-
-
-def test_get_idedata_regenerates_cache_without_cc_path(setup_core: Path) -> None:
-    """A cache predating cc_path is rebuilt even though it is newer.
-
-    Such a cache stays newer than the compile DB forever, so consumers that
-    derive the binutils paths from cc_path would keep failing on it.
-    """
-    compile_commands, cache = _setup_build(setup_core)
-    compile_commands.parent.mkdir(parents=True, exist_ok=True)
-    compile_commands.write_text("[]")
-    cache.parent.mkdir(parents=True, exist_ok=True)
-    cache.write_text('{"cxx_path": "cached"}')
-    cc_mtime = compile_commands.stat().st_mtime
-    os.utime(cache, (cc_mtime + 1, cc_mtime + 1))
-
-    with patch(
-        "esphome.espidf.idedata.idedata_from_build",
-        return_value={"cc_path": "gcc", "cxx_path": "g++"},
-    ) as mock_transform:
-        result = toolchain.get_idedata()
-
-    mock_transform.assert_called_once()
-    assert result["cc_path"] == "gcc"
-
-
-def test_get_idedata_regenerates_when_compile_commands_newer(setup_core: Path) -> None:
-    """A compile DB newer than the cache forces regeneration."""
-    compile_commands, cache = _setup_build(setup_core)
-    cache.parent.mkdir(parents=True, exist_ok=True)
-    cache.write_text('{"cxx_path": "stale"}')
-    compile_commands.parent.mkdir(parents=True, exist_ok=True)
-    compile_commands.write_text("[]")
-    cache_mtime = cache.stat().st_mtime
-    os.utime(compile_commands, (cache_mtime + 1, cache_mtime + 1))
-
-    with patch(
-        "esphome.espidf.idedata.idedata_from_build",
-        return_value={"cxx_path": "fresh"},
-    ) as mock_transform:
-        result = toolchain.get_idedata()
-
-    mock_transform.assert_called_once()
-    assert result == {"cxx_path": "fresh", "prog_path": str(toolchain.get_elf_path())}
-
-
-@pytest.mark.parametrize("cached", ['"cc_path is a string"', "[]", "42"])
-def test_get_idedata_regenerates_on_non_dict_cache(
-    setup_core: Path, cached: str
-) -> None:
-    """A newer cache holding valid JSON that is not an object is regenerated.
-
-    A bare string would otherwise pass the cc_path check by substring and be
-    handed to consumers expecting a dict.
-    """
-    compile_commands, cache = _setup_build(setup_core)
-    compile_commands.parent.mkdir(parents=True, exist_ok=True)
-    compile_commands.write_text("[]")
-    cache.parent.mkdir(parents=True, exist_ok=True)
-    cache.write_text(cached)
-    cc_mtime = compile_commands.stat().st_mtime
-    os.utime(cache, (cc_mtime + 1, cc_mtime + 1))
-
-    with patch(
-        "esphome.espidf.idedata.idedata_from_build",
-        return_value={"cc_path": "gcc", "cxx_path": "g++"},
-    ) as mock_transform:
-        result = toolchain.get_idedata()
-
-    mock_transform.assert_called_once()
-    assert isinstance(result, dict)
-
-
-def test_get_idedata_regenerates_on_corrupted_cache(setup_core: Path) -> None:
-    """An unparseable (but newer) cache falls back to regeneration."""
-    compile_commands, cache = _setup_build(setup_core)
-    compile_commands.parent.mkdir(parents=True, exist_ok=True)
-    compile_commands.write_text("[]")
-    cache.parent.mkdir(parents=True, exist_ok=True)
-    cache.write_text("{not json")
-    cc_mtime = compile_commands.stat().st_mtime
-    os.utime(cache, (cc_mtime + 1, cc_mtime + 1))
-
-    with patch(
-        "esphome.espidf.idedata.idedata_from_build",
-        return_value={"cxx_path": "regen"},
-    ) as mock_transform:
-        result = toolchain.get_idedata()
-
-    mock_transform.assert_called_once()
-    assert result == {"cxx_path": "regen", "prog_path": str(toolchain.get_elf_path())}
-
-
 def test_get_idedata_prog_path_points_at_firmware_elf(setup_core: Path) -> None:
     """The idedata exposes prog_path (the ELF) so consumers like build-action
     can locate firmware.factory.bin / firmware.ota.bin as its siblings."""
@@ -267,7 +159,7 @@ def test_get_idedata_prog_path_points_at_firmware_elf(setup_core: Path) -> None:
     compile_commands.write_text("[]")
 
     with patch(
-        "esphome.espidf.idedata.idedata_from_build",
+        "esphome.build_helpers.idedata.idedata_from_build",
         return_value={"cxx_path": "g++"},
     ):
         result = toolchain.get_idedata()

tests/unit_tests/test_main.py

@@ -17,6 +17,7 @@ from unittest.mock import AsyncMock, MagicMock, Mock, patch
 
 import pytest
 from pytest import CaptureFixture
+import serial
 from zeroconf import ServiceStateChange
 
 from esphome import __main__ as main, yaml_util
@@ -26,6 +27,7 @@ from esphome.__main__ import (
     _make_crystal_freq_callback,
     _redact_with_legacy_fallback,
     _resolve_network_devices,
+    _should_subscribe_states,
     _split_network_devices,
     _unresolved_default_error,
     _validate_bootloader_binary,
@@ -69,19 +71,21 @@ from esphome.__main__ import (
 )
 from esphome.address_cache import AddressCache
 from esphome.bundle import BUNDLE_EXTENSION, BundleFile, BundleResult
-from esphome.components import esp32, esp8266
+from esphome.components import esp32, esp8266, mqtt
 from esphome.components.esp32 import (
     KEY_ESP32,
     KEY_VARIANT,
     VARIANT_ESP32,
     get_esp32_variant,
 )
+from esphome.config import Config
 from esphome.const import (
     CONF_API,
     CONF_AUTH,
     CONF_BAUD_RATE,
     CONF_BROKER,
     CONF_DISABLED,
+    CONF_DISCOVER_IP,
     CONF_ESPHOME,
     CONF_LEVEL,
     CONF_LOG,
@@ -103,6 +107,7 @@ from esphome.const import (
     CONF_WEB_SERVER,
     CONF_WIFI,
     KEY_CORE,
+    KEY_TARGET_FRAMEWORK,
     KEY_TARGET_PLATFORM,
     PLATFORM_BK72XX,
     PLATFORM_ESP32,
@@ -567,8 +572,6 @@ def test_command_config__no_defaults_dumps_user_snapshot(
 ) -> None:
     """``--no-defaults`` dumps ``config.user_config`` instead of the
     validated config, so schema defaults don't leak into the output."""
-    from esphome.config import Config
-
     setup_core(tmp_path=tmp_path, config={"esphome": {"name": "test"}})
     args = MockArgs()
     args.show_secrets = True
@@ -621,8 +624,6 @@ def test_command_config__no_defaults_skips_strip_default_ids(
 ) -> None:
     """When ``--no-defaults`` is set, ``strip_default_ids`` isn't run --
     the user snapshot is already free of schema-injected IDs."""
-    from esphome.config import Config
-
     setup_core(tmp_path=tmp_path, config={"esphome": {"name": "test"}})
     args = MockArgs()
     args.show_secrets = True
@@ -3440,9 +3441,6 @@ def test_get_port_type() -> None:
 
 def test_mqtt_reexports_discover_ip() -> None:
     """The old import path must keep working for external code."""
-    from esphome.components import mqtt
-    from esphome.const import CONF_DISCOVER_IP
-
     assert mqtt.CONF_DISCOVER_IP is CONF_DISCOVER_IP
 
 
@@ -5909,8 +5907,6 @@ class MockSerial:
             chunk = self.chunks[self.chunk_index]
             if chunk is MOCK_SERIAL_END:
                 # Sentinel means we're done - simulate port closed
-                import serial
-
                 raise serial.SerialException("Port closed")
             # Respect the requested size and keep any remaining bytes
             if size <= 0:
@@ -5924,8 +5920,6 @@ class MockSerial:
                 # Entire chunk consumed; advance to the next one
                 self.chunk_index += 1
             return data  # type: ignore[return-value]
-        import serial
-
         raise serial.SerialException("Port closed")
 
 
@@ -6784,8 +6778,6 @@ def test_parse_args_argcomplete_only_runs_when_completing() -> None:
 
 def test_should_subscribe_states_default() -> None:
     """Test that states are shown by default when nothing is set."""
-    from esphome.__main__ import _should_subscribe_states
-
     args = parse_args(["esphome", "logs", "device.yaml"])
     with patch.dict(os.environ, {}, clear=False):
         os.environ.pop("ESPHOME_LOG_STATES", None)
@@ -6794,8 +6786,6 @@ def test_should_subscribe_states_default() -> None:
 
 def test_should_subscribe_states_env_suppresses() -> None:
     """Test that ESPHOME_LOG_STATES=false suppresses states by default."""
-    from esphome.__main__ import _should_subscribe_states
-
     args = parse_args(["esphome", "logs", "device.yaml"])
     with patch.dict(os.environ, {"ESPHOME_LOG_STATES": "false"}):
         assert _should_subscribe_states(args) is False
@@ -6803,8 +6793,6 @@ def test_should_subscribe_states_env_suppresses() -> None:
 
 def test_should_subscribe_states_env_enables() -> None:
     """Test that ESPHOME_LOG_STATES=true enables states by default."""
-    from esphome.__main__ import _should_subscribe_states
-
     args = parse_args(["esphome", "logs", "device.yaml"])
     with patch.dict(os.environ, {"ESPHOME_LOG_STATES": "true"}):
         assert _should_subscribe_states(args) is True
@@ -6812,8 +6800,6 @@ def test_should_subscribe_states_env_enables() -> None:
 
 def test_should_subscribe_states_flag_overrides_env() -> None:
     """Test that --states overrides ESPHOME_LOG_STATES=false."""
-    from esphome.__main__ import _should_subscribe_states
-
     args = parse_args(["esphome", "logs", "--states", "device.yaml"])
     with patch.dict(os.environ, {"ESPHOME_LOG_STATES": "false"}):
         assert _should_subscribe_states(args) is True
@@ -6821,8 +6807,6 @@ def test_should_subscribe_states_flag_overrides_env() -> None:
 
 def test_should_subscribe_states_no_flag_overrides_env() -> None:
     """Test that --no-states overrides ESPHOME_LOG_STATES=true."""
-    from esphome.__main__ import _should_subscribe_states
-
     args = parse_args(["esphome", "logs", "--no-states", "device.yaml"])
     with patch.dict(os.environ, {"ESPHOME_LOG_STATES": "true"}):
         assert _should_subscribe_states(args) is False
@@ -7135,6 +7119,82 @@ def test_warn_source_tree_mismatch_falls_back_when_stat_fails(
     assert not caplog.text
 
 
+@pytest.mark.parametrize(
+    "error",
+    [
+        FileNotFoundError("no such compiler"),
+        RuntimeError("Could not query builtin include dirs"),
+        ValueError("no C++ translation unit found"),
+        KeyError("command"),
+        None,  # replaced with EsphomeError inside
+    ],
+)
+def test_compile_program_espidf_idedata_failure_does_not_fail_build(
+    error: Exception,
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A post-compile idedata error is a warning: the firmware already built."""
+    if error is None:
+        error = EsphomeError("compile database is unusable")
+    CORE.toolchain = Toolchain.ESP_IDF
+    CORE.data[KEY_CORE] = {
+        KEY_TARGET_PLATFORM: "esp32",
+        KEY_TARGET_FRAMEWORK: "esp-idf",
+    }
+    with (
+        patch("esphome.espidf.toolchain.run_compile", return_value=0),
+        patch("esphome.espidf.toolchain.create_factory_bin"),
+        patch("esphome.espidf.toolchain.create_ota_bin"),
+        patch("esphome.espidf.toolchain.create_elf_copy"),
+        patch("esphome.espidf.toolchain.get_idedata", side_effect=error),
+        patch("esphome.__main__._check_and_emit_build_info"),
+    ):
+        assert compile_program(MagicMock(), {}) == 0
+    assert "Could not generate idedata" in caplog.text
+
+
+def test_compile_program_espidf_idedata_success_is_silent(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """The healthy path: idedata generated, nothing to warn about."""
+    CORE.toolchain = Toolchain.ESP_IDF
+    CORE.data[KEY_CORE] = {
+        KEY_TARGET_PLATFORM: "esp32",
+        KEY_TARGET_FRAMEWORK: "esp-idf",
+    }
+    with (
+        patch("esphome.espidf.toolchain.run_compile", return_value=0),
+        patch("esphome.espidf.toolchain.create_factory_bin"),
+        patch("esphome.espidf.toolchain.create_ota_bin"),
+        patch("esphome.espidf.toolchain.create_elf_copy"),
+        patch("esphome.espidf.toolchain.get_idedata", return_value={"cc_path": "x"}),
+        patch("esphome.__main__._check_and_emit_build_info"),
+    ):
+        assert compile_program(MagicMock(), {}) == 0
+    assert "idedata" not in caplog.text
+
+
+def test_compile_program_espidf_idedata_none_warns(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A silent None from the post-compile idedata refresh is made visible."""
+    CORE.toolchain = Toolchain.ESP_IDF
+    CORE.data[KEY_CORE] = {
+        KEY_TARGET_PLATFORM: "esp32",
+        KEY_TARGET_FRAMEWORK: "esp-idf",
+    }
+    with (
+        patch("esphome.espidf.toolchain.run_compile", return_value=0),
+        patch("esphome.espidf.toolchain.create_factory_bin"),
+        patch("esphome.espidf.toolchain.create_ota_bin"),
+        patch("esphome.espidf.toolchain.create_elf_copy"),
+        patch("esphome.espidf.toolchain.get_idedata", return_value=None),
+        patch("esphome.__main__._check_and_emit_build_info"),
+    ):
+        assert compile_program(MagicMock(), {}) == 0
+    assert "No idedata was generated" in caplog.text
+
+
 @pytest.mark.asyncio
 async def test_wrap_to_code_comment_is_insertion_order_independent() -> None:
     """The config comment dumps with sorted keys: voluptuous fills schema

tests/unit_tests/test_size_summary.py

@@ -126,3 +126,20 @@ def test_print_summary_handles_no_memory_types(
     size_json = _write_size_json(tmp_path, {"image_size": 0})
     print_summary(size_json, partitions_csv=None)
     assert capsys.readouterr().out == ""
+
+
+def test_print_summary_flash_line(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """A partition table with an app row yields the Flash line in the exact
+    padded shape script/ci_memory_impact_extract.py greps."""
+    size_json = _write_size_json(tmp_path, _esp32_size_data())
+    partitions = tmp_path / "partitions.csv"
+    partitions.write_text(
+        "# name, type, subtype, offset, size, flags\n"
+        "app0, app, ota_0, 0x10000, 0x1C0000,\n"
+    )
+    print_summary(size_json, partitions)
+    out = capsys.readouterr().out
+    assert "Flash: " in out
+    assert "(used 827455 bytes from 1835008 bytes)" in out