From 8953e53a04a282fb3261e3be3da95aaebdee0c59 Mon Sep 17 00:00:00 2001
From: "J. Nick Koston" <nick@koston.org>
Date: Thu, 10 Jul 2025 23:54:57 -1000
Subject: [PATCH] CI: Centralize test determination logic to reduce unnecessary
 job runners (#9432)

---
 .github/workflows/ci.yml            | 105 +++++----
 script/determine-jobs.py            | 245 +++++++++++++++++++
 script/helpers.py                   | 126 +++++++++-
 script/list-components.py           |  27 ++-
 tests/script/test_determine_jobs.py | 352 ++++++++++++++++++++++++++++
 tests/script/test_helpers.py        | 165 +++++++++++++
 6 files changed, 963 insertions(+), 57 deletions(-)
 create mode 100755 script/determine-jobs.py
 create mode 100644 tests/script/test_determine_jobs.py

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 08d64243b9..503a50c5c2 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -66,6 +66,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.python-linters == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -87,6 +89,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.python-linters == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -108,6 +112,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.python-linters == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -129,6 +135,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.python-linters == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -232,11 +240,54 @@ jobs:
           path: venv
           key: ${{ runner.os }}-${{ steps.restore-python.outputs.python-version }}-venv-${{ needs.common.outputs.cache-key }}
 
+  determine-jobs:
+    name: Determine which jobs to run
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+    outputs:
+      integration-tests: ${{ steps.determine.outputs.integration-tests }}
+      clang-tidy: ${{ steps.determine.outputs.clang-tidy }}
+      clang-format: ${{ steps.determine.outputs.clang-format }}
+      python-linters: ${{ steps.determine.outputs.python-linters }}
+      changed-components: ${{ steps.determine.outputs.changed-components }}
+      component-test-count: ${{ steps.determine.outputs.component-test-count }}
+    steps:
+      - name: Check out code from GitHub
+        uses: actions/checkout@v4.2.2
+        with:
+          # Fetch enough history to find the merge base
+          fetch-depth: 2
+      - name: Restore Python
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+      - name: Determine which tests to run
+        id: determine
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          . venv/bin/activate
+          output=$(python script/determine-jobs.py)
+          echo "Test determination output:"
+          echo "$output" | jq
+
+          # Extract individual fields
+          echo "integration-tests=$(echo "$output" | jq -r '.integration_tests')" >> $GITHUB_OUTPUT
+          echo "clang-tidy=$(echo "$output" | jq -r '.clang_tidy')" >> $GITHUB_OUTPUT
+          echo "clang-format=$(echo "$output" | jq -r '.clang_format')" >> $GITHUB_OUTPUT
+          echo "python-linters=$(echo "$output" | jq -r '.python_linters')" >> $GITHUB_OUTPUT
+          echo "changed-components=$(echo "$output" | jq -c '.changed_components')" >> $GITHUB_OUTPUT
+          echo "component-test-count=$(echo "$output" | jq -r '.component_test_count')" >> $GITHUB_OUTPUT
+
   integration-tests:
     name: Run integration tests
     runs-on: ubuntu-latest
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.integration-tests == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -271,6 +322,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
+      - determine-jobs
+    if: needs.determine-jobs.outputs.clang-format == 'true'
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@v4.2.2
@@ -304,6 +357,8 @@ jobs:
       - pylint
       - pytest
       - pyupgrade
+      - determine-jobs
+    if: needs.determine-jobs.outputs.clang-tidy == 'true'
     env:
       GH_TOKEN: ${{ github.token }}
     strategy:
@@ -411,50 +466,18 @@ jobs:
         # yamllint disable-line rule:line-length
         if: always()
 
-  list-components:
-    runs-on: ubuntu-24.04
-    needs:
-      - common
-    if: github.event_name == 'pull_request'
-    env:
-      GH_TOKEN: ${{ github.token }}
-    outputs:
-      components: ${{ steps.list-components.outputs.components }}
-      count: ${{ steps.list-components.outputs.count }}
-    steps:
-      - name: Check out code from GitHub
-        uses: actions/checkout@v4.2.2
-      - name: Restore Python
-        uses: ./.github/actions/restore-python
-        with:
-          python-version: ${{ env.DEFAULT_PYTHON }}
-          cache-key: ${{ needs.common.outputs.cache-key }}
-      - name: Find changed components
-        id: list-components
-        run: |
-          . venv/bin/activate
-          components=$(script/list-components.py --changed)
-          output_components=$(echo "$components" | jq -R -s -c 'split("\n")[:-1] | map(select(length > 0))')
-          count=$(echo "$output_components" | jq length)
-
-          echo "components=$output_components" >> $GITHUB_OUTPUT
-          echo "count=$count" >> $GITHUB_OUTPUT
-
-          echo "$count Components:"
-          echo "$output_components" | jq
-
   test-build-components:
     name: Component test ${{ matrix.file }}
     runs-on: ubuntu-24.04
     needs:
       - common
-      - list-components
-    if: github.event_name == 'pull_request' && fromJSON(needs.list-components.outputs.count) > 0 && fromJSON(needs.list-components.outputs.count) < 100
+      - determine-jobs
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.component-test-count) > 0 && fromJSON(needs.determine-jobs.outputs.component-test-count) < 100
     strategy:
       fail-fast: false
       max-parallel: 2
       matrix:
-        file: ${{ fromJson(needs.list-components.outputs.components) }}
+        file: ${{ fromJson(needs.determine-jobs.outputs.changed-components) }}
     steps:
       - name: Install dependencies
         run: |
@@ -482,8 +505,8 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
-      - list-components
-    if: github.event_name == 'pull_request' && fromJSON(needs.list-components.outputs.count) >= 100
+      - determine-jobs
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.component-test-count) >= 100
     outputs:
       matrix: ${{ steps.split.outputs.components }}
     steps:
@@ -492,7 +515,7 @@ jobs:
       - name: Split components into 20 groups
         id: split
         run: |
-          components=$(echo '${{ needs.list-components.outputs.components }}' | jq -c '.[]' | shuf | jq -s -c '[_nwise(20) | join(" ")]')
+          components=$(echo '${{ needs.determine-jobs.outputs.changed-components }}' | jq -c '.[]' | shuf | jq -s -c '[_nwise(20) | join(" ")]')
           echo "components=$components" >> $GITHUB_OUTPUT
 
   test-build-components-split:
@@ -500,9 +523,9 @@ jobs:
     runs-on: ubuntu-24.04
     needs:
       - common
-      - list-components
+      - determine-jobs
       - test-build-components-splitter
-    if: github.event_name == 'pull_request' && fromJSON(needs.list-components.outputs.count) >= 100
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.component-test-count) >= 100
     strategy:
       fail-fast: false
       max-parallel: 4
@@ -553,7 +576,7 @@ jobs:
       - integration-tests
       - pyupgrade
       - clang-tidy
-      - list-components
+      - determine-jobs
       - test-build-components
       - test-build-components-splitter
       - test-build-components-split
diff --git a/script/determine-jobs.py b/script/determine-jobs.py
new file mode 100755
index 0000000000..fc5c397c65
--- /dev/null
+++ b/script/determine-jobs.py
@@ -0,0 +1,245 @@
+#!/usr/bin/env python3
+"""Determine which CI jobs should run based on changed files.
+
+This script is a centralized way to determine which CI jobs need to run based on
+what files have changed. It outputs JSON with the following structure:
+
+{
+  "integration_tests": true/false,
+  "clang_tidy": true/false,
+  "clang_format": true/false,
+  "python_linters": true/false,
+  "changed_components": ["component1", "component2", ...],
+  "component_test_count": 5
+}
+
+The CI workflow uses this information to:
+- Skip or run integration tests
+- Skip or run clang-tidy (and whether to do a full scan)
+- Skip or run clang-format
+- Skip or run Python linters (ruff, flake8, pylint, pyupgrade)
+- Determine which components to test individually
+- Decide how to split component tests (if there are many)
+
+Usage:
+  python script/determine-jobs.py [-b BRANCH]
+
+Options:
+  -b, --branch BRANCH  Branch to compare against (default: dev)
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+from pathlib import Path
+import subprocess
+import sys
+from typing import Any
+
+from helpers import (
+    CPP_FILE_EXTENSIONS,
+    ESPHOME_COMPONENTS_PATH,
+    PYTHON_FILE_EXTENSIONS,
+    changed_files,
+    get_all_dependencies,
+    get_components_from_integration_fixtures,
+    parse_list_components_output,
+    root_path,
+)
+
+
+def should_run_integration_tests(branch: str | None = None) -> bool:
+    """Determine if integration tests should run based on changed files.
+
+    This function is used by the CI workflow to intelligently skip integration tests when they're
+    not needed, saving significant CI time and resources.
+
+    Integration tests will run when ANY of the following conditions are met:
+
+    1. Core C++ files changed (esphome/core/*)
+       - Any .cpp, .h, .tcc files in the core directory
+       - These files contain fundamental functionality used throughout ESPHome
+       - Examples: esphome/core/component.cpp, esphome/core/application.h
+
+    2. Core Python files changed (esphome/core/*.py)
+       - Only .py files in the esphome/core/ directory
+       - These are core Python files that affect the entire system
+       - Examples: esphome/core/config.py, esphome/core/__init__.py
+       - NOT included: esphome/*.py, esphome/dashboard/*.py, esphome/components/*/*.py
+
+    3. Integration test files changed
+       - Any file in tests/integration/ directory
+       - This includes test files themselves and fixture YAML files
+       - Examples: tests/integration/test_api.py, tests/integration/fixtures/api.yaml
+
+    4. Components used by integration tests (or their dependencies) changed
+       - The function parses all YAML files in tests/integration/fixtures/
+       - Extracts which components are used in integration tests
+       - Recursively finds all dependencies of those components
+       - If any of these components have changes, tests must run
+       - Example: If api.yaml uses 'sensor' and 'api' components, and 'api' depends on 'socket',
+         then changes to sensor/, api/, or socket/ components trigger tests
+
+    Args:
+        branch: Branch to compare against. If None, uses default.
+
+    Returns:
+        True if integration tests should run, False otherwise.
+    """
+    files = changed_files(branch)
+
+    # Check if any core files changed (esphome/core/*)
+    for file in files:
+        if file.startswith("esphome/core/"):
+            return True
+
+    # Check if any integration test files changed
+    if any("tests/integration" in file for file in files):
+        return True
+
+    # Get all components used in integration tests and their dependencies
+    fixture_components = get_components_from_integration_fixtures()
+    all_required_components = get_all_dependencies(fixture_components)
+
+    # Check if any required components changed
+    for file in files:
+        if file.startswith(ESPHOME_COMPONENTS_PATH):
+            parts = file.split("/")
+            if len(parts) >= 3:
+                component = parts[2]
+                if component in all_required_components:
+                    return True
+
+    return False
+
+
+def should_run_clang_tidy(branch: str | None = None) -> bool:
+    """Determine if clang-tidy should run based on changed files.
+
+    This function is used by the CI workflow to intelligently skip clang-tidy checks when they're
+    not needed, saving significant CI time and resources.
+
+    Clang-tidy will run when ANY of the following conditions are met:
+
+    1. Clang-tidy configuration changed
+       - The hash of .clang-tidy configuration file has changed
+       - The hash includes the .clang-tidy file, clang-tidy version from requirements_dev.txt,
+         and relevant platformio.ini sections
+       - When configuration changes, a full scan is needed to ensure all code complies
+         with the new rules
+       - Detected by script/clang_tidy_hash.py --check returning exit code 0
+
+    2. Any C++ source files changed
+       - Any file with C++ extensions: .cpp, .h, .hpp, .cc, .cxx, .c, .tcc
+       - Includes files anywhere in the repository, not just in esphome/
+       - This ensures all C++ code is checked, including tests, examples, etc.
+       - Examples: esphome/core/component.cpp, tests/custom/my_component.h
+
+    If the hash check fails for any reason, clang-tidy runs as a safety measure to ensure
+    code quality is maintained.
+
+    Args:
+        branch: Branch to compare against. If None, uses default.
+
+    Returns:
+        True if clang-tidy should run, False otherwise.
+    """
+    # First check if clang-tidy configuration changed (full scan needed)
+    try:
+        result = subprocess.run(
+            [os.path.join(root_path, "script", "clang_tidy_hash.py"), "--check"],
+            capture_output=True,
+            check=False,
+        )
+        # Exit 0 means hash changed (full scan needed)
+        if result.returncode == 0:
+            return True
+    except Exception:
+        # If hash check fails, run clang-tidy to be safe
+        return True
+
+    return _any_changed_file_endswith(branch, CPP_FILE_EXTENSIONS)
+
+
+def should_run_clang_format(branch: str | None = None) -> bool:
+    """Determine if clang-format should run based on changed files.
+
+    This function is used by the CI workflow to skip clang-format checks when no C++ files
+    have changed, saving CI time and resources.
+
+    Clang-format will run when any C++ source files have changed.
+
+    Args:
+        branch: Branch to compare against. If None, uses default.
+
+    Returns:
+        True if clang-format should run, False otherwise.
+    """
+    return _any_changed_file_endswith(branch, CPP_FILE_EXTENSIONS)
+
+
+def should_run_python_linters(branch: str | None = None) -> bool:
+    """Determine if Python linters (ruff, flake8, pylint, pyupgrade) should run based on changed files.
+
+    This function is used by the CI workflow to skip Python linting checks when no Python files
+    have changed, saving CI time and resources.
+
+    Python linters will run when any Python source files have changed.
+
+    Args:
+        branch: Branch to compare against. If None, uses default.
+
+    Returns:
+        True if Python linters should run, False otherwise.
+    """
+    return _any_changed_file_endswith(branch, PYTHON_FILE_EXTENSIONS)
+
+
+def _any_changed_file_endswith(branch: str | None, extensions: tuple[str, ...]) -> bool:
+    """Check if a changed file ends with any of the specified extensions."""
+    return any(file.endswith(extensions) for file in changed_files(branch))
+
+
+def main() -> None:
+    """Main function that determines which CI jobs to run."""
+    parser = argparse.ArgumentParser(
+        description="Determine which CI jobs should run based on changed files"
+    )
+    parser.add_argument(
+        "-b", "--branch", help="Branch to compare changed files against"
+    )
+    args = parser.parse_args()
+
+    # Determine what should run
+    run_integration = should_run_integration_tests(args.branch)
+    run_clang_tidy = should_run_clang_tidy(args.branch)
+    run_clang_format = should_run_clang_format(args.branch)
+    run_python_linters = should_run_python_linters(args.branch)
+
+    # Get changed components using list-components.py for exact compatibility
+    script_path = Path(__file__).parent / "list-components.py"
+    cmd = [sys.executable, str(script_path), "--changed"]
+    if args.branch:
+        cmd.extend(["-b", args.branch])
+
+    result = subprocess.run(cmd, capture_output=True, text=True, check=True)
+    changed_components = parse_list_components_output(result.stdout)
+
+    # Build output
+    output: dict[str, Any] = {
+        "integration_tests": run_integration,
+        "clang_tidy": run_clang_tidy,
+        "clang_format": run_clang_format,
+        "python_linters": run_python_linters,
+        "changed_components": changed_components,
+        "component_test_count": len(changed_components),
+    }
+
+    # Output as JSON
+    print(json.dumps(output))
+
+
+if __name__ == "__main__":
+    main()
diff --git a/script/helpers.py b/script/helpers.py
index 5dbc7a32cc..ff63bbc5b6 100644
--- a/script/helpers.py
+++ b/script/helpers.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+from functools import cache
 import json
 import os
 import os.path
@@ -7,6 +8,7 @@ from pathlib import Path
 import re
 import subprocess
 import time
+from typing import Any
 
 import colorama
 
@@ -15,6 +17,34 @@ basepath = os.path.join(root_path, "esphome")
 temp_folder = os.path.join(root_path, ".temp")
 temp_header_file = os.path.join(temp_folder, "all-include.cpp")
 
+# C++ file extensions used for clang-tidy and clang-format checks
+CPP_FILE_EXTENSIONS = (".cpp", ".h", ".hpp", ".cc", ".cxx", ".c", ".tcc")
+
+# Python file extensions
+PYTHON_FILE_EXTENSIONS = (".py", ".pyi")
+
+# YAML file extensions
+YAML_FILE_EXTENSIONS = (".yaml", ".yml")
+
+# Component path prefix
+ESPHOME_COMPONENTS_PATH = "esphome/components/"
+
+
+def parse_list_components_output(output: str) -> list[str]:
+    """Parse the output from list-components.py script.
+
+    The script outputs one component name per line.
+
+    Args:
+        output: The stdout from list-components.py
+
+    Returns:
+        List of component names, or empty list if no output
+    """
+    if not output or not output.strip():
+        return []
+    return [c.strip() for c in output.strip().split("\n") if c.strip()]
+
 
 def styled(color: str | tuple[str, ...], msg: str, reset: bool = True) -> str:
     prefix = "".join(color) if isinstance(color, tuple) else color
@@ -96,6 +126,7 @@ def _get_pr_number_from_github_env() -> str | None:
     return None
 
 
+@cache
 def _get_changed_files_github_actions() -> list[str] | None:
     """Get changed files in GitHub Actions environment.
 
@@ -135,7 +166,7 @@ def changed_files(branch: str | None = None) -> list[str]:
             return github_files
 
     # Original implementation for local development
-    if branch is None:
+    if not branch:  # Treat None and empty string the same
         branch = "dev"
     check_remotes = ["upstream", "origin"]
     check_remotes.extend(splitlines_no_ends(get_output("git", "remote")))
@@ -183,7 +214,7 @@ def get_changed_components() -> list[str] | None:
     changed = changed_files()
     core_cpp_changed = any(
         f.startswith("esphome/core/")
-        and f.endswith((".cpp", ".h", ".hpp", ".cc", ".cxx", ".c"))
+        and f.endswith(CPP_FILE_EXTENSIONS[:-1])  # Exclude .tcc for core files
         for f in changed
     )
     if core_cpp_changed:
@@ -198,8 +229,7 @@ def get_changed_components() -> list[str] | None:
         result = subprocess.run(
             cmd, capture_output=True, text=True, check=True, close_fds=False
         )
-        components = [c.strip() for c in result.stdout.strip().split("\n") if c.strip()]
-        return components
+        return parse_list_components_output(result.stdout)
     except subprocess.CalledProcessError:
         # If the script fails, fall back to full scan
         print("Could not determine changed components - will run full clang-tidy scan")
@@ -249,7 +279,9 @@ def _filter_changed_ci(files: list[str]) -> list[str]:
         # Action: Check only the specific non-component files that changed
         changed = changed_files()
         files = [
-            f for f in files if f in changed and not f.startswith("esphome/components/")
+            f
+            for f in files
+            if f in changed and not f.startswith(ESPHOME_COMPONENTS_PATH)
         ]
         if not files:
             print("No files changed")
@@ -267,7 +299,7 @@ def _filter_changed_ci(files: list[str]) -> list[str]:
     # because changes in one file can affect other files in the same component.
     filtered_files = []
     for f in files:
-        if f.startswith("esphome/components/"):
+        if f.startswith(ESPHOME_COMPONENTS_PATH):
             # Check if file belongs to any of the changed components
             parts = f.split("/")
             if len(parts) >= 3 and parts[2] in component_set:
@@ -326,7 +358,7 @@ def git_ls_files(patterns: list[str] | None = None) -> dict[str, int]:
     return {s[3].strip(): int(s[0]) for s in lines}
 
 
-def load_idedata(environment):
+def load_idedata(environment: str) -> dict[str, Any]:
     start_time = time.time()
     print(f"Loading IDE data for environment '{environment}'...")
 
@@ -442,3 +474,83 @@ def get_usable_cpu_count() -> int:
     return (
         os.process_cpu_count() if hasattr(os, "process_cpu_count") else os.cpu_count()
     )
+
+
+def get_all_dependencies(component_names: set[str]) -> set[str]:
+    """Get all dependencies for a set of components.
+
+    Args:
+        component_names: Set of component names to get dependencies for
+
+    Returns:
+        Set of all components including dependencies and auto-loaded components
+    """
+    from esphome.const import KEY_CORE
+    from esphome.core import CORE
+    from esphome.loader import get_component
+
+    all_components: set[str] = set(component_names)
+
+    # Reset CORE to ensure clean state
+    CORE.reset()
+
+    # Set up fake config path for component loading
+    root = Path(__file__).parent.parent
+    CORE.config_path = str(root)
+    CORE.data[KEY_CORE] = {}
+
+    # Keep finding dependencies until no new ones are found
+    while True:
+        new_components: set[str] = set()
+
+        for comp_name in all_components:
+            comp = get_component(comp_name)
+            if not comp:
+                continue
+
+            # Add dependencies (extract component name before '.')
+            new_components.update(dep.split(".")[0] for dep in comp.dependencies)
+
+            # Add auto_load components
+            new_components.update(comp.auto_load)
+
+        # Check if we found any new components
+        new_components -= all_components
+        if not new_components:
+            break
+
+        all_components.update(new_components)
+
+    return all_components
+
+
+def get_components_from_integration_fixtures() -> set[str]:
+    """Extract all components used in integration test fixtures.
+
+    Returns:
+        Set of component names used in integration test fixtures
+    """
+    import yaml
+
+    components: set[str] = set()
+    fixtures_dir = Path(__file__).parent.parent / "tests" / "integration" / "fixtures"
+
+    for yaml_file in fixtures_dir.glob("*.yaml"):
+        with open(yaml_file) as f:
+            config: dict[str, any] | None = yaml.safe_load(f)
+            if not config:
+                continue
+
+            # Add all top-level component keys
+            components.update(config.keys())
+
+            # Add platform components (e.g., output.template)
+            for value in config.values():
+                if not isinstance(value, list):
+                    continue
+
+                for item in value:
+                    if isinstance(item, dict) and "platform" in item:
+                        components.add(item["platform"])
+
+    return components
diff --git a/script/list-components.py b/script/list-components.py
index 0afcaa0f9d..66212f44e7 100755
--- a/script/list-components.py
+++ b/script/list-components.py
@@ -20,6 +20,12 @@ def filter_component_files(str):
     return str.startswith("esphome/components/") | str.startswith("tests/components/")
 
 
+def get_all_component_files() -> list[str]:
+    """Get all component files from git."""
+    files = git_ls_files()
+    return list(filter(filter_component_files, files))
+
+
 def extract_component_names_array_from_files_array(files):
     components = []
     for file in files:
@@ -165,17 +171,20 @@ def main():
     if args.branch and not args.changed:
         parser.error("--branch requires --changed")
 
-    files = git_ls_files()
-    files = filter(filter_component_files, files)
-
     if args.changed:
-        if args.branch:
-            changed = changed_files(args.branch)
-        else:
-            changed = changed_files()
+        # When --changed is passed, only get the changed files
+        changed = changed_files(args.branch)
+
         # If any base test file(s) changed, there's no need to filter out components
-        if not any("tests/test_build_components" in file for file in changed):
-            files = [f for f in files if f in changed]
+        if any("tests/test_build_components" in file for file in changed):
+            # Need to get all component files
+            files = get_all_component_files()
+        else:
+            # Only look at changed component files
+            files = [f for f in changed if filter_component_files(f)]
+    else:
+        # Get all component files
+        files = get_all_component_files()
 
     for c in get_components(files, args.changed):
         print(c)
diff --git a/tests/script/test_determine_jobs.py b/tests/script/test_determine_jobs.py
new file mode 100644
index 0000000000..4aaaadd80a
--- /dev/null
+++ b/tests/script/test_determine_jobs.py
@@ -0,0 +1,352 @@
+"""Unit tests for script/determine-jobs.py module."""
+
+from collections.abc import Generator
+import importlib.util
+import json
+import os
+import subprocess
+import sys
+from unittest.mock import Mock, patch
+
+import pytest
+
+# Add the script directory to Python path so we can import the module
+script_dir = os.path.abspath(
+    os.path.join(os.path.dirname(__file__), "..", "..", "script")
+)
+sys.path.insert(0, script_dir)
+
+spec = importlib.util.spec_from_file_location(
+    "determine_jobs", os.path.join(script_dir, "determine-jobs.py")
+)
+determine_jobs = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(determine_jobs)
+
+
+@pytest.fixture
+def mock_should_run_integration_tests() -> Generator[Mock, None, None]:
+    """Mock should_run_integration_tests from helpers."""
+    with patch.object(determine_jobs, "should_run_integration_tests") as mock:
+        yield mock
+
+
+@pytest.fixture
+def mock_should_run_clang_tidy() -> Generator[Mock, None, None]:
+    """Mock should_run_clang_tidy from helpers."""
+    with patch.object(determine_jobs, "should_run_clang_tidy") as mock:
+        yield mock
+
+
+@pytest.fixture
+def mock_should_run_clang_format() -> Generator[Mock, None, None]:
+    """Mock should_run_clang_format from helpers."""
+    with patch.object(determine_jobs, "should_run_clang_format") as mock:
+        yield mock
+
+
+@pytest.fixture
+def mock_should_run_python_linters() -> Generator[Mock, None, None]:
+    """Mock should_run_python_linters from helpers."""
+    with patch.object(determine_jobs, "should_run_python_linters") as mock:
+        yield mock
+
+
+@pytest.fixture
+def mock_subprocess_run() -> Generator[Mock, None, None]:
+    """Mock subprocess.run for list-components.py calls."""
+    with patch.object(determine_jobs.subprocess, "run") as mock:
+        yield mock
+
+
+def test_main_all_tests_should_run(
+    mock_should_run_integration_tests: Mock,
+    mock_should_run_clang_tidy: Mock,
+    mock_should_run_clang_format: Mock,
+    mock_should_run_python_linters: Mock,
+    mock_subprocess_run: Mock,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Test when all tests should run."""
+    mock_should_run_integration_tests.return_value = True
+    mock_should_run_clang_tidy.return_value = True
+    mock_should_run_clang_format.return_value = True
+    mock_should_run_python_linters.return_value = True
+
+    # Mock list-components.py output
+    mock_result = Mock()
+    mock_result.stdout = "wifi\napi\nsensor\n"
+    mock_subprocess_run.return_value = mock_result
+
+    # Run main function with mocked argv
+    with patch("sys.argv", ["determine-jobs.py"]):
+        determine_jobs.main()
+
+    # Check output
+    captured = capsys.readouterr()
+    output = json.loads(captured.out)
+
+    assert output["integration_tests"] is True
+    assert output["clang_tidy"] is True
+    assert output["clang_format"] is True
+    assert output["python_linters"] is True
+    assert output["changed_components"] == ["wifi", "api", "sensor"]
+    assert output["component_test_count"] == 3
+
+
+def test_main_no_tests_should_run(
+    mock_should_run_integration_tests: Mock,
+    mock_should_run_clang_tidy: Mock,
+    mock_should_run_clang_format: Mock,
+    mock_should_run_python_linters: Mock,
+    mock_subprocess_run: Mock,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Test when no tests should run."""
+    mock_should_run_integration_tests.return_value = False
+    mock_should_run_clang_tidy.return_value = False
+    mock_should_run_clang_format.return_value = False
+    mock_should_run_python_linters.return_value = False
+
+    # Mock empty list-components.py output
+    mock_result = Mock()
+    mock_result.stdout = ""
+    mock_subprocess_run.return_value = mock_result
+
+    # Run main function with mocked argv
+    with patch("sys.argv", ["determine-jobs.py"]):
+        determine_jobs.main()
+
+    # Check output
+    captured = capsys.readouterr()
+    output = json.loads(captured.out)
+
+    assert output["integration_tests"] is False
+    assert output["clang_tidy"] is False
+    assert output["clang_format"] is False
+    assert output["python_linters"] is False
+    assert output["changed_components"] == []
+    assert output["component_test_count"] == 0
+
+
+def test_main_list_components_fails(
+    mock_should_run_integration_tests: Mock,
+    mock_should_run_clang_tidy: Mock,
+    mock_should_run_clang_format: Mock,
+    mock_should_run_python_linters: Mock,
+    mock_subprocess_run: Mock,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Test when list-components.py fails."""
+    mock_should_run_integration_tests.return_value = True
+    mock_should_run_clang_tidy.return_value = True
+    mock_should_run_clang_format.return_value = True
+    mock_should_run_python_linters.return_value = True
+
+    # Mock list-components.py failure
+    mock_subprocess_run.side_effect = subprocess.CalledProcessError(1, "cmd")
+
+    # Run main function with mocked argv - should raise
+    with patch("sys.argv", ["determine-jobs.py"]):
+        with pytest.raises(subprocess.CalledProcessError):
+            determine_jobs.main()
+
+
+def test_main_with_branch_argument(
+    mock_should_run_integration_tests: Mock,
+    mock_should_run_clang_tidy: Mock,
+    mock_should_run_clang_format: Mock,
+    mock_should_run_python_linters: Mock,
+    mock_subprocess_run: Mock,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Test with branch argument."""
+    mock_should_run_integration_tests.return_value = False
+    mock_should_run_clang_tidy.return_value = True
+    mock_should_run_clang_format.return_value = False
+    mock_should_run_python_linters.return_value = True
+
+    # Mock list-components.py output
+    mock_result = Mock()
+    mock_result.stdout = "mqtt\n"
+    mock_subprocess_run.return_value = mock_result
+
+    with patch("sys.argv", ["script.py", "-b", "main"]):
+        determine_jobs.main()
+
+    # Check that functions were called with branch
+    mock_should_run_integration_tests.assert_called_once_with("main")
+    mock_should_run_clang_tidy.assert_called_once_with("main")
+    mock_should_run_clang_format.assert_called_once_with("main")
+    mock_should_run_python_linters.assert_called_once_with("main")
+
+    # Check that list-components.py was called with branch
+    mock_subprocess_run.assert_called_once()
+    call_args = mock_subprocess_run.call_args[0][0]
+    assert "--changed" in call_args
+    assert "-b" in call_args
+    assert "main" in call_args
+
+    # Check output
+    captured = capsys.readouterr()
+    output = json.loads(captured.out)
+
+    assert output["integration_tests"] is False
+    assert output["clang_tidy"] is True
+    assert output["clang_format"] is False
+    assert output["python_linters"] is True
+    assert output["changed_components"] == ["mqtt"]
+    assert output["component_test_count"] == 1
+
+
+def test_should_run_integration_tests(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Test should_run_integration_tests function."""
+    # Core C++ files trigger tests
+    with patch.object(
+        determine_jobs, "changed_files", return_value=["esphome/core/component.cpp"]
+    ):
+        result = determine_jobs.should_run_integration_tests()
+        assert result is True
+
+    # Core Python files trigger tests
+    with patch.object(
+        determine_jobs, "changed_files", return_value=["esphome/core/config.py"]
+    ):
+        result = determine_jobs.should_run_integration_tests()
+        assert result is True
+
+    # Python files directly in esphome/ do NOT trigger tests
+    with patch.object(
+        determine_jobs, "changed_files", return_value=["esphome/config.py"]
+    ):
+        result = determine_jobs.should_run_integration_tests()
+        assert result is False
+
+    # Python files in subdirectories (not core) do NOT trigger tests
+    with patch.object(
+        determine_jobs,
+        "changed_files",
+        return_value=["esphome/dashboard/web_server.py"],
+    ):
+        result = determine_jobs.should_run_integration_tests()
+        assert result is False
+
+
+def test_should_run_integration_tests_with_branch() -> None:
+    """Test should_run_integration_tests with branch argument."""
+    with patch.object(determine_jobs, "changed_files") as mock_changed:
+        mock_changed.return_value = []
+        determine_jobs.should_run_integration_tests("release")
+        mock_changed.assert_called_once_with("release")
+
+
+def test_should_run_integration_tests_component_dependency() -> None:
+    """Test that integration tests run when components used in fixtures change."""
+    with patch.object(
+        determine_jobs, "changed_files", return_value=["esphome/components/api/api.cpp"]
+    ):
+        with patch.object(
+            determine_jobs, "get_components_from_integration_fixtures"
+        ) as mock_fixtures:
+            mock_fixtures.return_value = {"api", "sensor"}
+            with patch.object(determine_jobs, "get_all_dependencies") as mock_deps:
+                mock_deps.return_value = {"api", "sensor", "network"}
+                result = determine_jobs.should_run_integration_tests()
+                assert result is True
+
+
+@pytest.mark.parametrize(
+    ("check_returncode", "changed_files", "expected_result"),
+    [
+        (0, [], True),  # Hash changed - need full scan
+        (1, ["esphome/core.cpp"], True),  # C++ file changed
+        (1, ["README.md"], False),  # No C++ files changed
+    ],
+)
+def test_should_run_clang_tidy(
+    check_returncode: int,
+    changed_files: list[str],
+    expected_result: bool,
+) -> None:
+    """Test should_run_clang_tidy function."""
+    with patch.object(determine_jobs, "changed_files", return_value=changed_files):
+        # Test with hash check returning specific code
+        with patch("subprocess.run") as mock_run:
+            mock_run.return_value = Mock(returncode=check_returncode)
+            result = determine_jobs.should_run_clang_tidy()
+            assert result == expected_result
+
+        # Test with hash check failing (exception)
+        if check_returncode != 0:
+            with patch("subprocess.run", side_effect=Exception("Failed")):
+                result = determine_jobs.should_run_clang_tidy()
+            assert result is True  # Fail safe - run clang-tidy
+
+
+def test_should_run_clang_tidy_with_branch() -> None:
+    """Test should_run_clang_tidy with branch argument."""
+    with patch.object(determine_jobs, "changed_files") as mock_changed:
+        mock_changed.return_value = []
+        with patch("subprocess.run") as mock_run:
+            mock_run.return_value = Mock(returncode=1)  # Hash unchanged
+            determine_jobs.should_run_clang_tidy("release")
+            mock_changed.assert_called_once_with("release")
+
+
+@pytest.mark.parametrize(
+    ("changed_files", "expected_result"),
+    [
+        (["esphome/core.py"], True),
+        (["script/test.py"], True),
+        (["esphome/test.pyi"], True),  # .pyi files should trigger
+        (["README.md"], False),
+        ([], False),
+    ],
+)
+def test_should_run_python_linters(
+    changed_files: list[str], expected_result: bool
+) -> None:
+    """Test should_run_python_linters function."""
+    with patch.object(determine_jobs, "changed_files", return_value=changed_files):
+        result = determine_jobs.should_run_python_linters()
+        assert result == expected_result
+
+
+def test_should_run_python_linters_with_branch() -> None:
+    """Test should_run_python_linters with branch argument."""
+    with patch.object(determine_jobs, "changed_files") as mock_changed:
+        mock_changed.return_value = []
+        determine_jobs.should_run_python_linters("release")
+        mock_changed.assert_called_once_with("release")
+
+
+@pytest.mark.parametrize(
+    ("changed_files", "expected_result"),
+    [
+        (["esphome/core.cpp"], True),
+        (["esphome/core.h"], True),
+        (["test.hpp"], True),
+        (["test.cc"], True),
+        (["test.cxx"], True),
+        (["test.c"], True),
+        (["test.tcc"], True),
+        (["README.md"], False),
+        ([], False),
+    ],
+)
+def test_should_run_clang_format(
+    changed_files: list[str], expected_result: bool
+) -> None:
+    """Test should_run_clang_format function."""
+    with patch.object(determine_jobs, "changed_files", return_value=changed_files):
+        result = determine_jobs.should_run_clang_format()
+        assert result == expected_result
+
+
+def test_should_run_clang_format_with_branch() -> None:
+    """Test should_run_clang_format with branch argument."""
+    with patch.object(determine_jobs, "changed_files") as mock_changed:
+        mock_changed.return_value = []
+        determine_jobs.should_run_clang_format("release")
+        mock_changed.assert_called_once_with("release")
diff --git a/tests/script/test_helpers.py b/tests/script/test_helpers.py
index bbebdd79c8..d0db08e6f7 100644
--- a/tests/script/test_helpers.py
+++ b/tests/script/test_helpers.py
@@ -27,6 +27,7 @@ _filter_changed_ci = helpers._filter_changed_ci
 _filter_changed_local = helpers._filter_changed_local
 build_all_include = helpers.build_all_include
 print_file_list = helpers.print_file_list
+get_all_dependencies = helpers.get_all_dependencies
 
 
 @pytest.mark.parametrize(
@@ -154,6 +155,14 @@ def test_github_actions_push_event(monkeypatch: MonkeyPatch) -> None:
         assert result == expected_files
 
 
+@pytest.fixture(autouse=True)
+def clear_caches():
+    """Clear function caches before each test."""
+    # Clear the cache for _get_changed_files_github_actions
+    _get_changed_files_github_actions.cache_clear()
+    yield
+
+
 def test_get_changed_files_github_actions_pull_request(
     monkeypatch: MonkeyPatch,
 ) -> None:
@@ -847,3 +856,159 @@ def test_print_file_list_default_title(capsys: pytest.CaptureFixture[str]) -> No
 
     assert "Files:" in captured.out
     assert "    test.cpp" in captured.out
+
+
+@pytest.mark.parametrize(
+    ("component_configs", "initial_components", "expected_components"),
+    [
+        # No dependencies
+        (
+            {"sensor": ([], [])},  # (dependencies, auto_load)
+            {"sensor"},
+            {"sensor"},
+        ),
+        # Simple dependencies
+        (
+            {
+                "sensor": (["esp32"], []),
+                "esp32": ([], []),
+            },
+            {"sensor"},
+            {"sensor", "esp32"},
+        ),
+        # Auto-load components
+        (
+            {
+                "light": ([], ["output", "power_supply"]),
+                "output": ([], []),
+                "power_supply": ([], []),
+            },
+            {"light"},
+            {"light", "output", "power_supply"},
+        ),
+        # Transitive dependencies
+        (
+            {
+                "comp_a": (["comp_b"], []),
+                "comp_b": (["comp_c"], []),
+                "comp_c": ([], []),
+            },
+            {"comp_a"},
+            {"comp_a", "comp_b", "comp_c"},
+        ),
+        # Dependencies with dots (sensor.base)
+        (
+            {
+                "my_comp": (["sensor.base", "binary_sensor.base"], []),
+                "sensor": ([], []),
+                "binary_sensor": ([], []),
+            },
+            {"my_comp"},
+            {"my_comp", "sensor", "binary_sensor"},
+        ),
+        # Circular dependencies (should not cause infinite loop)
+        (
+            {
+                "comp_a": (["comp_b"], []),
+                "comp_b": (["comp_a"], []),
+            },
+            {"comp_a"},
+            {"comp_a", "comp_b"},
+        ),
+    ],
+)
+def test_get_all_dependencies(
+    component_configs: dict[str, tuple[list[str], list[str]]],
+    initial_components: set[str],
+    expected_components: set[str],
+) -> None:
+    """Test dependency resolution for components."""
+    with patch("esphome.loader.get_component") as mock_get_component:
+
+        def get_component_side_effect(name: str):
+            if name in component_configs:
+                deps, auto_load = component_configs[name]
+                comp = Mock()
+                comp.dependencies = deps
+                comp.auto_load = auto_load
+                return comp
+            return None
+
+        mock_get_component.side_effect = get_component_side_effect
+
+        result = helpers.get_all_dependencies(initial_components)
+
+        assert result == expected_components
+
+
+def test_get_all_dependencies_handles_missing_components() -> None:
+    """Test handling of components that can't be loaded."""
+    with patch("esphome.loader.get_component") as mock_get_component:
+        # First component exists, its dependency doesn't
+        comp = Mock()
+        comp.dependencies = ["missing_comp"]
+        comp.auto_load = []
+
+        mock_get_component.side_effect = (
+            lambda name: comp if name == "existing" else None
+        )
+
+        result = helpers.get_all_dependencies({"existing", "nonexistent"})
+
+        # Should still include all components, even if some can't be loaded
+        assert result == {"existing", "nonexistent", "missing_comp"}
+
+
+def test_get_all_dependencies_empty_set() -> None:
+    """Test with empty initial component set."""
+    result = helpers.get_all_dependencies(set())
+    assert result == set()
+
+
+def test_get_components_from_integration_fixtures() -> None:
+    """Test extraction of components from fixture YAML files."""
+    yaml_content = {
+        "sensor": [{"platform": "template", "name": "test"}],
+        "binary_sensor": [{"platform": "gpio", "pin": 5}],
+        "esphome": {"name": "test"},
+        "api": {},
+    }
+    expected_components = {
+        "sensor",
+        "binary_sensor",
+        "esphome",
+        "api",
+        "template",
+        "gpio",
+    }
+
+    mock_yaml_file = Mock()
+
+    with (
+        patch("pathlib.Path.glob") as mock_glob,
+        patch("builtins.open", create=True),
+        patch("yaml.safe_load", return_value=yaml_content),
+    ):
+        mock_glob.return_value = [mock_yaml_file]
+
+        components = helpers.get_components_from_integration_fixtures()
+
+        assert components == expected_components
+
+
+@pytest.mark.parametrize(
+    "output,expected",
+    [
+        ("wifi\napi\nsensor\n", ["wifi", "api", "sensor"]),
+        ("wifi\n", ["wifi"]),
+        ("", []),
+        ("  \n  \n", []),
+        ("\n\n", []),
+        ("  wifi  \n  api  \n", ["wifi", "api"]),
+        ("wifi\n\napi\n\nsensor", ["wifi", "api", "sensor"]),
+    ],
+)
+def test_parse_list_components_output(output: str, expected: list[str]) -> None:
+    """Test parse_list_components_output function."""
+    result = helpers.parse_list_components_output(output)
+    assert result == expected