Adopt the shared ninja quoting helpers and typed install paths in the build spec

esphome/build_gen/arduino8266.py

@@ -17,14 +17,18 @@ from __future__ import annotations
 from dataclasses import dataclass, field
 import logging
 from pathlib import Path
-import re
 import subprocess
+from typing import TYPE_CHECKING
 
+from esphome.build_helpers.ninja import shell_token as _shell_token
 from esphome.components.esp8266 import build_surgery
 from esphome.core import CORE, EsphomeError
 from esphome.helpers import mkdir_p, write_file_if_changed
 from esphome.platformio.library import join_flag_args, split_flag_entry
 
+if TYPE_CHECKING:
+    from esphome.arduino8266.framework import InstalledPaths
+
 _LOGGER = logging.getLogger(__name__)
 
 # From platformio-build.py. The first entry is the default; with multiple SDK
@@ -234,38 +238,6 @@ def _resolve_build_config(defines: dict[str, str]) -> _BuildConfig:
     )
 
 
-def _quote_arg(tok: str) -> str:
-    """Wrap a token in double quotes with the Windows argv rule.
-
-    Same escaping rule as ``subprocess.list2cmdline``: a backslash run
-    doubles only immediately before a quote (or the closing quote), and the
-    quote itself is escaped. POSIX sh parses the result identically for
-    backslashes and quotes. ``$`` must already be doubled for ninja.
-    """
-    quoted = re.sub(r'(\\*)"', lambda m: m.group(1) * 2 + '\\"', tok)
-    quoted = re.sub(r"(\\+)\Z", lambda m: m.group(1) * 2, quoted)
-    return f'"{quoted}"'
-
-
-_NEEDS_QUOTE = re.compile(r'[\s"\']')
-
-
-def _shell_token(tok: str) -> str:
-    """Quote a lexed token only when needed; ``_q`` force-quotes paths.
-
-    Lexing strips the quoting a user wrote (``-DX="a b"`` becomes the single
-    token ``-DX=a b``); re-quote on the way out so the compiler receives the
-    same argv element SCons would pass under PlatformIO. After ninja
-    un-doubles ``$$``, sh still expands ``$VAR`` while CreateProcess passes
-    it literally -- the same divergence SCons-under-sh has, so this stays
-    PlatformIO parity.
-    """
-    tok = tok.replace("$", "$$")  # ninja would expand a bare $ to nothing
-    if not _NEEDS_QUOTE.search(tok):
-        return tok
-    return _quote_arg(tok)
-
-
 def _defines_flags(
     config: _BuildConfig, flash_mode: str, board: str, board_defines: tuple[str, ...]
 ) -> list[str]:
@@ -301,7 +273,7 @@ def _unflag_tokens() -> set[str]:
 
 
 def _project_flags(
-    unflags: set[str] | None = None,
+    unflags: set[str],
 ) -> tuple[list[str], list[str], list[Path], list[str]]:
     """Split the ESPHome build flags into compile, linker, -L, and -l lists.
 
@@ -310,8 +282,6 @@ def _project_flags(
     ``build_unflags`` matches individual tokens (``-Os`` inside ``-Os -g3``).
     Lexed tokens are re-quoted at emission via ``_shell_token``.
     """
-    if unflags is None:
-        unflags = _unflag_tokens()
     compile_flags: list[str] = []
     link_flags: list[str] = []
     lib_dirs: list[Path] = []
@@ -332,7 +302,7 @@ def _project_flags(
 
 
 def generate_ld_scripts(
-    paths: dict[str, Path], config: _BuildConfig, flash_ld_name: str
+    paths: InstalledPaths, config: _BuildConfig, flash_ld_name: str
 ) -> None:
     """Generate the common linker script (and testing-mode flash ld copy).
 
@@ -340,8 +310,8 @@ def generate_ld_scripts(
     ``eagle.app.v6.common.ld.h``, then applies ESPHome's surgeries: the wifi
     rate-table DRAM relocation, and enlarged memory segments in testing mode.
     """
-    framework = paths["framework_path"]
-    gcc = paths["toolchain_path"] / "bin" / "xtensa-lx106-elf-gcc"
+    framework = paths.framework
+    gcc = paths.toolchain / "bin" / "xtensa-lx106-elf-gcc"
     ld_dir = CORE.relative_pioenvs_path(CORE.name, "ld")
     mkdir_p(ld_dir)
 

tests/unit_tests/build_gen/test_arduino8266.py

@@ -16,6 +16,7 @@ from unittest.mock import MagicMock, patch
 
 import pytest
 
+from esphome.arduino8266.framework import InstalledPaths
 from esphome.build_gen import arduino8266
 from esphome.build_gen.arduino8266 import (
     _defines_flags,
@@ -171,11 +172,7 @@ def _make_framework(tmp_path: Path) -> dict[str, Path]:
     toolchain = tmp_path / "toolchain"
     (toolchain / "bin").mkdir(parents=True)
     (toolchain / "include").mkdir()
-    return {
-        "framework_path": framework,
-        "toolchain_path": toolchain,
-        "ninja_path": Path("ninja"),
-    }
+    return InstalledPaths(framework=framework, toolchain=toolchain, ninja=Path("ninja"))
 
 
 @pytest.mark.parametrize(
@@ -261,7 +258,7 @@ SECTIONS
 """
 
 
-def _run_generate_ld_scripts(paths: dict[str, Path]) -> Path:
+def _run_generate_ld_scripts(paths: InstalledPaths) -> Path:
 
     config = _resolve_build_config(_flag_defines())
     arduino8266.generate_ld_scripts(paths, config, "eagle.flash.4m.ld")
@@ -313,10 +310,11 @@ def test_generate_ld_scripts_failure(tmp_path: Path) -> None:
 def test_generate_ld_scripts_testing_mode(tmp_path: Path) -> None:
 
     paths = _make_framework(tmp_path)
-    (paths["framework_path"] / "tools" / "sdk" / "ld" / "eagle.flash.4m.ld").write_text(
+    (paths.framework / "tools" / "sdk" / "ld" / "eagle.flash.4m.ld").write_text(
+        # Real flash ld scripts carry no iram1_0_seg (that lives in the
+        # generated common ld); the surgery rejects one it was not asked about
         "MEMORY\n{\n"
         "  dram0_0_seg :    org = 0x3FFE8000, len = 0x14000\n"
-        "  iram1_0_seg :    org = 0x40100000, len = 0x8000\n"
         "  irom0_0_seg :    org = 0x40201010, len = 0xfeff0\n"
         "}\n"
     )
@@ -342,7 +340,9 @@ def test_project_flags_trailing_bare_linker_flag_warns(
     caplog: pytest.LogCaptureFixture,
 ) -> None:
     _set_flags("-l")
-    compile_flags, link_flags, lib_dirs, libs = arduino8266._project_flags()
+    compile_flags, link_flags, lib_dirs, libs = arduino8266._project_flags(
+        arduino8266._unflag_tokens()
+    )
     assert "Ignoring trailing '-l'" in caplog.text
     assert not libs
     assert not lib_dirs
@@ -352,7 +352,9 @@ def test_project_flags_trailing_bare_linker_flag_warns(
 
 def test_project_flags_lexed_entry_scatters_non_linker_tokens() -> None:
     _set_flags("-L /d -Wl,-Map=m stray")
-    compile_flags, link_flags, lib_dirs, libs = arduino8266._project_flags()
+    compile_flags, link_flags, lib_dirs, libs = arduino8266._project_flags(
+        arduino8266._unflag_tokens()
+    )
     assert lib_dirs == [Path("/d")]
     assert link_flags == ["-Wl,-Map=m"]
     assert "stray" in compile_flags
@@ -372,7 +374,9 @@ def test_flag_defines_lexes_multi_token_entries() -> None:
 def test_project_flags_lexes_every_entry() -> None:
     """A linker flag anywhere in an entry reaches the link line (PIO parity)."""
     _set_flags("-DFOO=1 -lbar")
-    compile_flags, _link, _dirs, libs = arduino8266._project_flags()
+    compile_flags, _link, _dirs, libs = arduino8266._project_flags(
+        arduino8266._unflag_tokens()
+    )
     assert libs == ["bar"]
     assert "-DFOO=1" in compile_flags
 
@@ -391,7 +395,9 @@ def test_project_flags_unflags_match_tokens() -> None:
 def test_project_flags_requotes_lexed_defines() -> None:
     """A quoted spaced value stays one compiler argument after lex/emit."""
     _set_flags('-DGREETING="hello world"')
-    compile_flags, _link, _dirs, _libs = arduino8266._project_flags()
+    compile_flags, _link, _dirs, _libs = arduino8266._project_flags(
+        arduino8266._unflag_tokens()
+    )
     # shlex folds the quotes (as PIO's ParseFlags does); _shell_token
     # re-quotes the spaced token so the shell passes one argv element
     assert compile_flags == ['"-DGREETING=hello world"']