[ci] Add automated memory impact analysis for pull requests (#11242)

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: pre-commit-ci-lite[bot] <117423508+pre-commit-ci-lite[bot]@users.noreply.github.com>

.coveragerc

@@ -1,4 +1,5 @@
 [run]
 omit =
     esphome/components/*
+    esphome/analyze_memory/*
     tests/integration/*

.github/workflows/ci.yml

@@ -175,6 +175,7 @@ jobs:
       changed-components-with-tests: ${{ steps.determine.outputs.changed-components-with-tests }}
       directly-changed-components-with-tests: ${{ steps.determine.outputs.directly-changed-components-with-tests }}
       component-test-count: ${{ steps.determine.outputs.component-test-count }}
+      memory_impact: ${{ steps.determine.outputs.memory-impact }}
     steps:
       - name: Check out code from GitHub
         uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
@@ -204,6 +205,7 @@ jobs:
           echo "changed-components-with-tests=$(echo "$output" | jq -c '.changed_components_with_tests')" >> $GITHUB_OUTPUT
           echo "directly-changed-components-with-tests=$(echo "$output" | jq -c '.directly_changed_components_with_tests')" >> $GITHUB_OUTPUT
           echo "component-test-count=$(echo "$output" | jq -r '.component_test_count')" >> $GITHUB_OUTPUT
+          echo "memory-impact=$(echo "$output" | jq -c '.memory_impact')" >> $GITHUB_OUTPUT
 
   integration-tests:
     name: Run integration tests
@@ -521,6 +523,292 @@ jobs:
       - uses: pre-commit-ci/lite-action@5d6cc0eb514c891a40562a58a8e71576c5c7fb43 # v1.1.0
         if: always()
 
+  memory-impact-target-branch:
+    name: Build target branch for memory impact
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+      - determine-jobs
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.memory_impact).should_run == 'true'
+    outputs:
+      ram_usage: ${{ steps.extract.outputs.ram_usage }}
+      flash_usage: ${{ steps.extract.outputs.flash_usage }}
+      cache_hit: ${{ steps.cache-memory-analysis.outputs.cache-hit }}
+      skip: ${{ steps.check-script.outputs.skip }}
+    steps:
+      - name: Check out target branch
+        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+        with:
+          ref: ${{ github.base_ref }}
+
+      # Check if memory impact extraction script exists on target branch
+      # If not, skip the analysis (this handles older branches that don't have the feature)
+      - name: Check for memory impact script
+        id: check-script
+        run: |
+          if [ -f "script/ci_memory_impact_extract.py" ]; then
+            echo "skip=false" >> $GITHUB_OUTPUT
+          else
+            echo "skip=true" >> $GITHUB_OUTPUT
+            echo "::warning::ci_memory_impact_extract.py not found on target branch, skipping memory impact analysis"
+          fi
+
+      # All remaining steps only run if script exists
+      - name: Generate cache key
+        id: cache-key
+        if: steps.check-script.outputs.skip != 'true'
+        run: |
+          # Get the commit SHA of the target branch
+          target_sha=$(git rev-parse HEAD)
+
+          # Hash the build infrastructure files (all files that affect build/analysis)
+          infra_hash=$(cat \
+            script/test_build_components.py \
+            script/ci_memory_impact_extract.py \
+            script/analyze_component_buses.py \
+            script/merge_component_configs.py \
+            script/ci_helpers.py \
+            .github/workflows/ci.yml \
+            | sha256sum | cut -d' ' -f1)
+
+          # Get platform and components from job inputs
+          platform="${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}"
+          components='${{ toJSON(fromJSON(needs.determine-jobs.outputs.memory_impact).components) }}'
+          components_hash=$(echo "$components" | sha256sum | cut -d' ' -f1)
+
+          # Combine into cache key
+          cache_key="memory-analysis-target-${target_sha}-${infra_hash}-${platform}-${components_hash}"
+          echo "cache-key=${cache_key}" >> $GITHUB_OUTPUT
+          echo "Cache key: ${cache_key}"
+
+      - name: Restore cached memory analysis
+        id: cache-memory-analysis
+        if: steps.check-script.outputs.skip != 'true'
+        uses: actions/cache/restore@0057852bfaa89a56745cba8c7296529d2fc39830 # v4.3.0
+        with:
+          path: memory-analysis-target.json
+          key: ${{ steps.cache-key.outputs.cache-key }}
+
+      - name: Cache status
+        if: steps.check-script.outputs.skip != 'true'
+        run: |
+          if [ "${{ steps.cache-memory-analysis.outputs.cache-hit }}" == "true" ]; then
+            echo "✓ Cache hit! Using cached memory analysis results."
+            echo "  Skipping build step to save time."
+          else
+            echo "✗ Cache miss. Will build and analyze memory usage."
+          fi
+
+      - name: Restore Python
+        if: steps.check-script.outputs.skip != 'true' && steps.cache-memory-analysis.outputs.cache-hit != 'true'
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+
+      - name: Cache platformio
+        if: steps.check-script.outputs.skip != 'true' && steps.cache-memory-analysis.outputs.cache-hit != 'true'
+        uses: actions/cache/restore@0057852bfaa89a56745cba8c7296529d2fc39830 # v4.3.0
+        with:
+          path: ~/.platformio
+          key: platformio-memory-${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}-${{ hashFiles('platformio.ini') }}
+
+      - name: Build, compile, and analyze memory
+        if: steps.check-script.outputs.skip != 'true' && steps.cache-memory-analysis.outputs.cache-hit != 'true'
+        id: build
+        run: |
+          . venv/bin/activate
+          components='${{ toJSON(fromJSON(needs.determine-jobs.outputs.memory_impact).components) }}'
+          platform="${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}"
+
+          echo "Building with test_build_components.py for $platform with components:"
+          echo "$components" | jq -r '.[]' | sed 's/^/  - /'
+
+          # Use test_build_components.py which handles grouping automatically
+          # Pass components as comma-separated list
+          component_list=$(echo "$components" | jq -r 'join(",")')
+
+          echo "Compiling with test_build_components.py..."
+
+          # Run build and extract memory with auto-detection of build directory for detailed analysis
+          # Use tee to show output in CI while also piping to extraction script
+          python script/test_build_components.py \
+            -e compile \
+            -c "$component_list" \
+            -t "$platform" 2>&1 | \
+            tee /dev/stderr | \
+            python script/ci_memory_impact_extract.py \
+              --output-env \
+              --output-json memory-analysis-target.json
+
+      - name: Save memory analysis to cache
+        if: steps.check-script.outputs.skip != 'true' && steps.cache-memory-analysis.outputs.cache-hit != 'true' && steps.build.outcome == 'success'
+        uses: actions/cache/save@0057852bfaa89a56745cba8c7296529d2fc39830 # v4.3.0
+        with:
+          path: memory-analysis-target.json
+          key: ${{ steps.cache-key.outputs.cache-key }}
+
+      - name: Extract memory usage for outputs
+        id: extract
+        if: steps.check-script.outputs.skip != 'true'
+        run: |
+          if [ -f memory-analysis-target.json ]; then
+            ram=$(jq -r '.ram_bytes' memory-analysis-target.json)
+            flash=$(jq -r '.flash_bytes' memory-analysis-target.json)
+            echo "ram_usage=${ram}" >> $GITHUB_OUTPUT
+            echo "flash_usage=${flash}" >> $GITHUB_OUTPUT
+            echo "RAM: ${ram} bytes, Flash: ${flash} bytes"
+          else
+            echo "Error: memory-analysis-target.json not found"
+            exit 1
+          fi
+
+      - name: Upload memory analysis JSON
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
+        with:
+          name: memory-analysis-target
+          path: memory-analysis-target.json
+          if-no-files-found: warn
+          retention-days: 1
+
+  memory-impact-pr-branch:
+    name: Build PR branch for memory impact
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+      - determine-jobs
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.memory_impact).should_run == 'true'
+    outputs:
+      ram_usage: ${{ steps.extract.outputs.ram_usage }}
+      flash_usage: ${{ steps.extract.outputs.flash_usage }}
+    steps:
+      - name: Check out PR branch
+        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+      - name: Restore Python
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+      - name: Cache platformio
+        uses: actions/cache/restore@0057852bfaa89a56745cba8c7296529d2fc39830 # v4.3.0
+        with:
+          path: ~/.platformio
+          key: platformio-memory-${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}-${{ hashFiles('platformio.ini') }}
+      - name: Build, compile, and analyze memory
+        id: extract
+        run: |
+          . venv/bin/activate
+          components='${{ toJSON(fromJSON(needs.determine-jobs.outputs.memory_impact).components) }}'
+          platform="${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}"
+
+          echo "Building with test_build_components.py for $platform with components:"
+          echo "$components" | jq -r '.[]' | sed 's/^/  - /'
+
+          # Use test_build_components.py which handles grouping automatically
+          # Pass components as comma-separated list
+          component_list=$(echo "$components" | jq -r 'join(",")')
+
+          echo "Compiling with test_build_components.py..."
+
+          # Run build and extract memory with auto-detection of build directory for detailed analysis
+          # Use tee to show output in CI while also piping to extraction script
+          python script/test_build_components.py \
+            -e compile \
+            -c "$component_list" \
+            -t "$platform" 2>&1 | \
+            tee /dev/stderr | \
+            python script/ci_memory_impact_extract.py \
+              --output-env \
+              --output-json memory-analysis-pr.json
+      - name: Upload memory analysis JSON
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
+        with:
+          name: memory-analysis-pr
+          path: memory-analysis-pr.json
+          if-no-files-found: warn
+          retention-days: 1
+
+  memory-impact-comment:
+    name: Comment memory impact
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+      - determine-jobs
+      - memory-impact-target-branch
+      - memory-impact-pr-branch
+    if: github.event_name == 'pull_request' && fromJSON(needs.determine-jobs.outputs.memory_impact).should_run == 'true' && needs.memory-impact-target-branch.outputs.skip != 'true'
+    permissions:
+      contents: read
+      pull-requests: write
+    steps:
+      - name: Check out code
+        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+      - name: Restore Python
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+      - name: Download target analysis JSON
+        uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4.3.0
+        with:
+          name: memory-analysis-target
+          path: ./memory-analysis
+        continue-on-error: true
+      - name: Download PR analysis JSON
+        uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4.3.0
+        with:
+          name: memory-analysis-pr
+          path: ./memory-analysis
+        continue-on-error: true
+      - name: Post or update PR comment
+        env:
+          GH_TOKEN: ${{ github.token }}
+          COMPONENTS: ${{ toJSON(fromJSON(needs.determine-jobs.outputs.memory_impact).components) }}
+          PLATFORM: ${{ fromJSON(needs.determine-jobs.outputs.memory_impact).platform }}
+          TARGET_RAM: ${{ needs.memory-impact-target-branch.outputs.ram_usage }}
+          TARGET_FLASH: ${{ needs.memory-impact-target-branch.outputs.flash_usage }}
+          PR_RAM: ${{ needs.memory-impact-pr-branch.outputs.ram_usage }}
+          PR_FLASH: ${{ needs.memory-impact-pr-branch.outputs.flash_usage }}
+          TARGET_CACHE_HIT: ${{ needs.memory-impact-target-branch.outputs.cache_hit }}
+        run: |
+          . venv/bin/activate
+
+          # Check if analysis JSON files exist
+          target_json_arg=""
+          pr_json_arg=""
+
+          if [ -f ./memory-analysis/memory-analysis-target.json ]; then
+            echo "Found target analysis JSON"
+            target_json_arg="--target-json ./memory-analysis/memory-analysis-target.json"
+          else
+            echo "No target analysis JSON found"
+          fi
+
+          if [ -f ./memory-analysis/memory-analysis-pr.json ]; then
+            echo "Found PR analysis JSON"
+            pr_json_arg="--pr-json ./memory-analysis/memory-analysis-pr.json"
+          else
+            echo "No PR analysis JSON found"
+          fi
+
+          # Add cache flag if target was cached
+          cache_flag=""
+          if [ "$TARGET_CACHE_HIT" == "true" ]; then
+            cache_flag="--target-cache-hit"
+          fi
+
+          python script/ci_memory_impact_comment.py \
+            --pr-number "${{ github.event.pull_request.number }}" \
+            --components "$COMPONENTS" \
+            --platform "$PLATFORM" \
+            --target-ram "$TARGET_RAM" \
+            --target-flash "$TARGET_FLASH" \
+            --pr-ram "$PR_RAM" \
+            --pr-flash "$PR_FLASH" \
+            $target_json_arg \
+            $pr_json_arg \
+            $cache_flag
+
   ci-status:
     name: CI Status
     runs-on: ubuntu-24.04
@@ -535,6 +823,9 @@ jobs:
       - test-build-components-splitter
       - test-build-components-split
       - pre-commit-ci-lite
+      - memory-impact-target-branch
+      - memory-impact-pr-branch
+      - memory-impact-comment
     if: always()
     steps:
       - name: Success

esphome/__main__.py

@@ -466,7 +466,9 @@ def write_cpp_file() -> int:
 def compile_program(args: ArgsProtocol, config: ConfigType) -> int:
     from esphome import platformio_api
 
-    _LOGGER.info("Compiling app...")
+    # NOTE: "Build path:" format is parsed by script/ci_memory_impact_extract.py
+    # If you change this format, update the regex in that script as well
+    _LOGGER.info("Compiling app... Build path: %s", CORE.build_path)
     rc = platformio_api.run_compile(config, CORE.verbose)
     if rc != 0:
         return rc

esphome/analyze_memory/__main__.py

@@ -0,0 +1,6 @@
+"""Main entry point for running the memory analyzer as a module."""
+
+from .cli import main
+
+if __name__ == "__main__":
+    main()

esphome/analyze_memory/cli.py

@@ -0,0 +1,408 @@
+"""CLI interface for memory analysis with report generation."""
+
+from collections import defaultdict
+import sys
+
+from . import (
+    _COMPONENT_API,
+    _COMPONENT_CORE,
+    _COMPONENT_PREFIX_ESPHOME,
+    _COMPONENT_PREFIX_EXTERNAL,
+    MemoryAnalyzer,
+)
+
+
+class MemoryAnalyzerCLI(MemoryAnalyzer):
+    """Memory analyzer with CLI-specific report generation."""
+
+    # Column width constants
+    COL_COMPONENT: int = 29
+    COL_FLASH_TEXT: int = 14
+    COL_FLASH_DATA: int = 14
+    COL_RAM_DATA: int = 12
+    COL_RAM_BSS: int = 12
+    COL_TOTAL_FLASH: int = 15
+    COL_TOTAL_RAM: int = 12
+    COL_SEPARATOR: int = 3  # " | "
+
+    # Core analysis column widths
+    COL_CORE_SUBCATEGORY: int = 30
+    COL_CORE_SIZE: int = 12
+    COL_CORE_COUNT: int = 6
+    COL_CORE_PERCENT: int = 10
+
+    # Calculate table width once at class level
+    TABLE_WIDTH: int = (
+        COL_COMPONENT
+        + COL_SEPARATOR
+        + COL_FLASH_TEXT
+        + COL_SEPARATOR
+        + COL_FLASH_DATA
+        + COL_SEPARATOR
+        + COL_RAM_DATA
+        + COL_SEPARATOR
+        + COL_RAM_BSS
+        + COL_SEPARATOR
+        + COL_TOTAL_FLASH
+        + COL_SEPARATOR
+        + COL_TOTAL_RAM
+    )
+
+    @staticmethod
+    def _make_separator_line(*widths: int) -> str:
+        """Create a separator line with given column widths.
+
+        Args:
+            widths: Column widths to create separators for
+
+        Returns:
+            Separator line like "----+---------+-----"
+        """
+        return "-+-".join("-" * width for width in widths)
+
+    # Pre-computed separator lines
+    MAIN_TABLE_SEPARATOR: str = _make_separator_line(
+        COL_COMPONENT,
+        COL_FLASH_TEXT,
+        COL_FLASH_DATA,
+        COL_RAM_DATA,
+        COL_RAM_BSS,
+        COL_TOTAL_FLASH,
+        COL_TOTAL_RAM,
+    )
+
+    CORE_TABLE_SEPARATOR: str = _make_separator_line(
+        COL_CORE_SUBCATEGORY,
+        COL_CORE_SIZE,
+        COL_CORE_COUNT,
+        COL_CORE_PERCENT,
+    )
+
+    def generate_report(self, detailed: bool = False) -> str:
+        """Generate a formatted memory report."""
+        components = sorted(
+            self.components.items(), key=lambda x: x[1].flash_total, reverse=True
+        )
+
+        # Calculate totals
+        total_flash = sum(c.flash_total for _, c in components)
+        total_ram = sum(c.ram_total for _, c in components)
+
+        # Build report
+        lines: list[str] = []
+
+        lines.append("=" * self.TABLE_WIDTH)
+        lines.append("Component Memory Analysis".center(self.TABLE_WIDTH))
+        lines.append("=" * self.TABLE_WIDTH)
+        lines.append("")
+
+        # Main table - fixed column widths
+        lines.append(
+            f"{'Component':<{self.COL_COMPONENT}} | {'Flash (text)':>{self.COL_FLASH_TEXT}} | {'Flash (data)':>{self.COL_FLASH_DATA}} | {'RAM (data)':>{self.COL_RAM_DATA}} | {'RAM (bss)':>{self.COL_RAM_BSS}} | {'Total Flash':>{self.COL_TOTAL_FLASH}} | {'Total RAM':>{self.COL_TOTAL_RAM}}"
+        )
+        lines.append(self.MAIN_TABLE_SEPARATOR)
+
+        for name, mem in components:
+            if mem.flash_total > 0 or mem.ram_total > 0:
+                flash_rodata = mem.rodata_size + mem.data_size
+                lines.append(
+                    f"{name:<{self.COL_COMPONENT}} | {mem.text_size:>{self.COL_FLASH_TEXT - 2},} B | {flash_rodata:>{self.COL_FLASH_DATA - 2},} B | "
+                    f"{mem.data_size:>{self.COL_RAM_DATA - 2},} B | {mem.bss_size:>{self.COL_RAM_BSS - 2},} B | "
+                    f"{mem.flash_total:>{self.COL_TOTAL_FLASH - 2},} B | {mem.ram_total:>{self.COL_TOTAL_RAM - 2},} B"
+                )
+
+        lines.append(self.MAIN_TABLE_SEPARATOR)
+        lines.append(
+            f"{'TOTAL':<{self.COL_COMPONENT}} | {' ':>{self.COL_FLASH_TEXT}} | {' ':>{self.COL_FLASH_DATA}} | "
+            f"{' ':>{self.COL_RAM_DATA}} | {' ':>{self.COL_RAM_BSS}} | "
+            f"{total_flash:>{self.COL_TOTAL_FLASH - 2},} B | {total_ram:>{self.COL_TOTAL_RAM - 2},} B"
+        )
+
+        # Top consumers
+        lines.append("")
+        lines.append("Top Flash Consumers:")
+        for i, (name, mem) in enumerate(components[:25]):
+            if mem.flash_total > 0:
+                percentage = (
+                    (mem.flash_total / total_flash * 100) if total_flash > 0 else 0
+                )
+                lines.append(
+                    f"{i + 1}. {name} ({mem.flash_total:,} B) - {percentage:.1f}% of analyzed flash"
+                )
+
+        lines.append("")
+        lines.append("Top RAM Consumers:")
+        ram_components = sorted(components, key=lambda x: x[1].ram_total, reverse=True)
+        for i, (name, mem) in enumerate(ram_components[:25]):
+            if mem.ram_total > 0:
+                percentage = (mem.ram_total / total_ram * 100) if total_ram > 0 else 0
+                lines.append(
+                    f"{i + 1}. {name} ({mem.ram_total:,} B) - {percentage:.1f}% of analyzed RAM"
+                )
+
+        lines.append("")
+        lines.append(
+            "Note: This analysis covers symbols in the ELF file. Some runtime allocations may not be included."
+        )
+        lines.append("=" * self.TABLE_WIDTH)
+
+        # Add ESPHome core detailed analysis if there are core symbols
+        if self._esphome_core_symbols:
+            lines.append("")
+            lines.append("=" * self.TABLE_WIDTH)
+            lines.append(
+                f"{_COMPONENT_CORE} Detailed Analysis".center(self.TABLE_WIDTH)
+            )
+            lines.append("=" * self.TABLE_WIDTH)
+            lines.append("")
+
+            # Group core symbols by subcategory
+            core_subcategories: dict[str, list[tuple[str, str, int]]] = defaultdict(
+                list
+            )
+
+            for symbol, demangled, size in self._esphome_core_symbols:
+                # Categorize based on demangled name patterns
+                subcategory = self._categorize_esphome_core_symbol(demangled)
+                core_subcategories[subcategory].append((symbol, demangled, size))
+
+            # Sort subcategories by total size
+            sorted_subcategories = sorted(
+                [
+                    (name, symbols, sum(s[2] for s in symbols))
+                    for name, symbols in core_subcategories.items()
+                ],
+                key=lambda x: x[2],
+                reverse=True,
+            )
+
+            lines.append(
+                f"{'Subcategory':<{self.COL_CORE_SUBCATEGORY}} | {'Size':>{self.COL_CORE_SIZE}} | "
+                f"{'Count':>{self.COL_CORE_COUNT}} | {'% of Core':>{self.COL_CORE_PERCENT}}"
+            )
+            lines.append(self.CORE_TABLE_SEPARATOR)
+
+            core_total = sum(size for _, _, size in self._esphome_core_symbols)
+
+            for subcategory, symbols, total_size in sorted_subcategories:
+                percentage = (total_size / core_total * 100) if core_total > 0 else 0
+                lines.append(
+                    f"{subcategory:<{self.COL_CORE_SUBCATEGORY}} | {total_size:>{self.COL_CORE_SIZE - 2},} B | "
+                    f"{len(symbols):>{self.COL_CORE_COUNT}} | {percentage:>{self.COL_CORE_PERCENT - 1}.1f}%"
+                )
+
+            # Top 15 largest core symbols
+            lines.append("")
+            lines.append(f"Top 15 Largest {_COMPONENT_CORE} Symbols:")
+            sorted_core_symbols = sorted(
+                self._esphome_core_symbols, key=lambda x: x[2], reverse=True
+            )
+
+            for i, (symbol, demangled, size) in enumerate(sorted_core_symbols[:15]):
+                lines.append(f"{i + 1}. {demangled} ({size:,} B)")
+
+            lines.append("=" * self.TABLE_WIDTH)
+
+        # Add detailed analysis for top ESPHome and external components
+        esphome_components = [
+            (name, mem)
+            for name, mem in components
+            if name.startswith(_COMPONENT_PREFIX_ESPHOME) and name != _COMPONENT_CORE
+        ]
+        external_components = [
+            (name, mem)
+            for name, mem in components
+            if name.startswith(_COMPONENT_PREFIX_EXTERNAL)
+        ]
+
+        top_esphome_components = sorted(
+            esphome_components, key=lambda x: x[1].flash_total, reverse=True
+        )[:30]
+
+        # Include all external components (they're usually important)
+        top_external_components = sorted(
+            external_components, key=lambda x: x[1].flash_total, reverse=True
+        )
+
+        # Check if API component exists and ensure it's included
+        api_component = None
+        for name, mem in components:
+            if name == _COMPONENT_API:
+                api_component = (name, mem)
+                break
+
+        # Combine all components to analyze: top ESPHome + all external + API if not already included
+        components_to_analyze = list(top_esphome_components) + list(
+            top_external_components
+        )
+        if api_component and api_component not in components_to_analyze:
+            components_to_analyze.append(api_component)
+
+        if components_to_analyze:
+            for comp_name, comp_mem in components_to_analyze:
+                if not (comp_symbols := self._component_symbols.get(comp_name, [])):
+                    continue
+                lines.append("")
+                lines.append("=" * self.TABLE_WIDTH)
+                lines.append(f"{comp_name} Detailed Analysis".center(self.TABLE_WIDTH))
+                lines.append("=" * self.TABLE_WIDTH)
+                lines.append("")
+
+                # Sort symbols by size
+                sorted_symbols = sorted(comp_symbols, key=lambda x: x[2], reverse=True)
+
+                lines.append(f"Total symbols: {len(sorted_symbols)}")
+                lines.append(f"Total size: {comp_mem.flash_total:,} B")
+                lines.append("")
+
+                # Show all symbols > 100 bytes for better visibility
+                large_symbols = [
+                    (sym, dem, size) for sym, dem, size in sorted_symbols if size > 100
+                ]
+
+                lines.append(
+                    f"{comp_name} Symbols > 100 B ({len(large_symbols)} symbols):"
+                )
+                for i, (symbol, demangled, size) in enumerate(large_symbols):
+                    lines.append(f"{i + 1}. {demangled} ({size:,} B)")
+
+                lines.append("=" * self.TABLE_WIDTH)
+
+        return "\n".join(lines)
+
+    def dump_uncategorized_symbols(self, output_file: str | None = None) -> None:
+        """Dump uncategorized symbols for analysis."""
+        # Sort by size descending
+        sorted_symbols = sorted(
+            self._uncategorized_symbols, key=lambda x: x[2], reverse=True
+        )
+
+        lines = ["Uncategorized Symbols Analysis", "=" * 80]
+        lines.append(f"Total uncategorized symbols: {len(sorted_symbols)}")
+        lines.append(
+            f"Total uncategorized size: {sum(s[2] for s in sorted_symbols):,} bytes"
+        )
+        lines.append("")
+        lines.append(f"{'Size':>10} | {'Symbol':<60} | Demangled")
+        lines.append("-" * 10 + "-+-" + "-" * 60 + "-+-" + "-" * 40)
+
+        for symbol, demangled, size in sorted_symbols[:100]:  # Top 100
+            demangled_display = (
+                demangled[:100] if symbol != demangled else "[not demangled]"
+            )
+            lines.append(f"{size:>10,} | {symbol[:60]:<60} | {demangled_display}")
+
+        if len(sorted_symbols) > 100:
+            lines.append(f"\n... and {len(sorted_symbols) - 100} more symbols")
+
+        content = "\n".join(lines)
+
+        if output_file:
+            with open(output_file, "w", encoding="utf-8") as f:
+                f.write(content)
+        else:
+            print(content)
+
+
+def analyze_elf(
+    elf_path: str,
+    objdump_path: str | None = None,
+    readelf_path: str | None = None,
+    detailed: bool = False,
+    external_components: set[str] | None = None,
+) -> str:
+    """Analyze an ELF file and return a memory report."""
+    analyzer = MemoryAnalyzerCLI(
+        elf_path, objdump_path, readelf_path, external_components
+    )
+    analyzer.analyze()
+    return analyzer.generate_report(detailed)
+
+
+def main():
+    """CLI entrypoint for memory analysis."""
+    if len(sys.argv) < 2:
+        print("Usage: python -m esphome.analyze_memory <build_directory>")
+        print("\nAnalyze memory usage from an ESPHome build directory.")
+        print("The build directory should contain firmware.elf and idedata will be")
+        print("loaded from ~/.esphome/.internal/idedata/<device>.json")
+        print("\nExamples:")
+        print("  python -m esphome.analyze_memory ~/.esphome/build/my-device")
+        print("  python -m esphome.analyze_memory .esphome/build/my-device")
+        print("  python -m esphome.analyze_memory my-device  # Short form")
+        sys.exit(1)
+
+    build_dir = sys.argv[1]
+
+    # Load build directory
+    import json
+    from pathlib import Path
+
+    from esphome.platformio_api import IDEData
+
+    build_path = Path(build_dir)
+
+    # If no path separator in name, assume it's a device name
+    if "/" not in build_dir and not build_path.is_dir():
+        # Try current directory first
+        cwd_path = Path.cwd() / ".esphome" / "build" / build_dir
+        if cwd_path.is_dir():
+            build_path = cwd_path
+            print(f"Using build directory: {build_path}", file=sys.stderr)
+        else:
+            # Fall back to home directory
+            build_path = Path.home() / ".esphome" / "build" / build_dir
+            print(f"Using build directory: {build_path}", file=sys.stderr)
+
+    if not build_path.is_dir():
+        print(f"Error: {build_path} is not a directory", file=sys.stderr)
+        sys.exit(1)
+
+    # Find firmware.elf
+    elf_file = None
+    for elf_candidate in [
+        build_path / "firmware.elf",
+        build_path / ".pioenvs" / build_path.name / "firmware.elf",
+    ]:
+        if elf_candidate.exists():
+            elf_file = str(elf_candidate)
+            break
+
+    if not elf_file:
+        print(f"Error: firmware.elf not found in {build_dir}", file=sys.stderr)
+        sys.exit(1)
+
+    # Find idedata.json - check current directory first, then home
+    device_name = build_path.name
+    idedata_candidates = [
+        Path.cwd() / ".esphome" / "idedata" / f"{device_name}.json",
+        Path.home() / ".esphome" / "idedata" / f"{device_name}.json",
+    ]
+
+    idedata = None
+    for idedata_path in idedata_candidates:
+        if not idedata_path.exists():
+            continue
+        try:
+            with open(idedata_path, encoding="utf-8") as f:
+                raw_data = json.load(f)
+            idedata = IDEData(raw_data)
+            print(f"Loaded idedata from: {idedata_path}", file=sys.stderr)
+            break
+        except (json.JSONDecodeError, OSError) as e:
+            print(f"Warning: Failed to load idedata: {e}", file=sys.stderr)
+
+    if not idedata:
+        print(
+            f"Warning: idedata not found (searched {idedata_candidates[0]} and {idedata_candidates[1]})",
+            file=sys.stderr,
+        )
+
+    analyzer = MemoryAnalyzerCLI(elf_file, idedata=idedata)
+    analyzer.analyze()
+    report = analyzer.generate_report()
+    print(report)
+
+
+if __name__ == "__main__":
+    main()

esphome/analyze_memory/helpers.py

@@ -0,0 +1,121 @@
+"""Helper functions for memory analysis."""
+
+from functools import cache
+from pathlib import Path
+
+from .const import SECTION_MAPPING
+
+# Import namespace constant from parent module
+# Note: This would create a circular import if done at module level,
+# so we'll define it locally here as well
+_NAMESPACE_ESPHOME = "esphome::"
+
+
+# Get the list of actual ESPHome components by scanning the components directory
+@cache
+def get_esphome_components():
+    """Get set of actual ESPHome components from the components directory."""
+    # Find the components directory relative to this file
+    # Go up two levels from analyze_memory/helpers.py to esphome/
+    current_dir = Path(__file__).parent.parent
+    components_dir = current_dir / "components"
+
+    if not components_dir.exists() or not components_dir.is_dir():
+        return frozenset()
+
+    return frozenset(
+        item.name
+        for item in components_dir.iterdir()
+        if item.is_dir()
+        and not item.name.startswith(".")
+        and not item.name.startswith("__")
+    )
+
+
+@cache
+def get_component_class_patterns(component_name: str) -> list[str]:
+    """Generate component class name patterns for symbol matching.
+
+    Args:
+        component_name: The component name (e.g., "ota", "wifi", "api")
+
+    Returns:
+        List of pattern strings to match against demangled symbols
+    """
+    component_upper = component_name.upper()
+    component_camel = component_name.replace("_", "").title()
+    return [
+        f"{_NAMESPACE_ESPHOME}{component_upper}Component",  # e.g., esphome::OTAComponent
+        f"{_NAMESPACE_ESPHOME}ESPHome{component_upper}Component",  # e.g., esphome::ESPHomeOTAComponent
+        f"{_NAMESPACE_ESPHOME}{component_camel}Component",  # e.g., esphome::OtaComponent
+        f"{_NAMESPACE_ESPHOME}ESPHome{component_camel}Component",  # e.g., esphome::ESPHomeOtaComponent
+    ]
+
+
+def map_section_name(raw_section: str) -> str | None:
+    """Map raw section name to standard section.
+
+    Args:
+        raw_section: Raw section name from ELF file (e.g., ".iram0.text", ".rodata.str1.1")
+
+    Returns:
+        Standard section name (".text", ".rodata", ".data", ".bss") or None
+    """
+    for standard_section, patterns in SECTION_MAPPING.items():
+        if any(pattern in raw_section for pattern in patterns):
+            return standard_section
+    return None
+
+
+def parse_symbol_line(line: str) -> tuple[str, str, int, str] | None:
+    """Parse a single symbol line from objdump output.
+
+    Args:
+        line: Line from objdump -t output
+
+    Returns:
+        Tuple of (section, name, size, address) or None if not a valid symbol.
+        Format: address l/g w/d F/O section size name
+        Example: 40084870 l     F .iram0.text    00000000 _xt_user_exc
+    """
+    parts = line.split()
+    if len(parts) < 5:
+        return None
+
+    try:
+        # Validate and extract address
+        address = parts[0]
+        int(address, 16)
+    except ValueError:
+        return None
+
+    # Look for F (function) or O (object) flag
+    if "F" not in parts and "O" not in parts:
+        return None
+
+    # Find section, size, and name
+    for i, part in enumerate(parts):
+        if not part.startswith("."):
+            continue
+
+        section = map_section_name(part)
+        if not section:
+            break
+
+        # Need at least size field after section
+        if i + 1 >= len(parts):
+            break
+
+        try:
+            size = int(parts[i + 1], 16)
+        except ValueError:
+            break
+
+        # Need symbol name and non-zero size
+        if i + 2 >= len(parts) or size == 0:
+            break
+
+        name = " ".join(parts[i + 2 :])
+        return (section, name, size, address)
+
+    return None

esphome/platformio_api.py

@@ -374,3 +374,23 @@ class IDEData:
             return f"{self.cc_path[:-7]}addr2line.exe"
 
         return f"{self.cc_path[:-3]}addr2line"
+
+    @property
+    def objdump_path(self) -> str:
+        # replace gcc at end with objdump
+        path = self.cc_path
+        return (
+            f"{path[:-7]}objdump.exe"
+            if path.endswith(".exe")
+            else f"{path[:-3]}objdump"
+        )
+
+    @property
+    def readelf_path(self) -> str:
+        # replace gcc at end with readelf
+        path = self.cc_path
+        return (
+            f"{path[:-7]}readelf.exe"
+            if path.endswith(".exe")
+            else f"{path[:-3]}readelf"
+        )

script/analyze_component_buses.py

@@ -34,6 +34,8 @@ from typing import Any
 # Add esphome to path
 sys.path.insert(0, str(Path(__file__).parent.parent))
 
+from helpers import BASE_BUS_COMPONENTS
+
 from esphome import yaml_util
 from esphome.config_helpers import Extend, Remove
 
@@ -67,18 +69,6 @@ NO_BUSES_SIGNATURE = "no_buses"
 # Isolated components have unique signatures and cannot be merged with others
 ISOLATED_SIGNATURE_PREFIX = "isolated_"
 
-# Base bus components - these ARE the bus implementations and should not
-# be flagged as needing migration since they are the platform/base components
-BASE_BUS_COMPONENTS = {
-    "i2c",
-    "spi",
-    "uart",
-    "modbus",
-    "canbus",
-    "remote_transmitter",
-    "remote_receiver",
-}
-
 # Components that must be tested in isolation (not grouped or batched with others)
 # These have known build issues that prevent grouping
 # NOTE: This should be kept in sync with both test_build_components and split_components_for_ci.py

script/ci_helpers.py

@@ -0,0 +1,23 @@
+"""Common helper functions for CI scripts."""
+
+from __future__ import annotations
+
+import os
+
+
+def write_github_output(outputs: dict[str, str | int]) -> None:
+    """Write multiple outputs to GITHUB_OUTPUT or stdout.
+
+    When running in GitHub Actions, writes to the GITHUB_OUTPUT file.
+    When running locally, writes to stdout for debugging.
+
+    Args:
+        outputs: Dictionary of key-value pairs to write
+    """
+    github_output = os.environ.get("GITHUB_OUTPUT")
+    if github_output:
+        with open(github_output, "a", encoding="utf-8") as f:
+            f.writelines(f"{key}={value}\n" for key, value in outputs.items())
+    else:
+        for key, value in outputs.items():
+            print(f"{key}={value}")

script/ci_memory_impact_extract.py

@@ -0,0 +1,281 @@
+#!/usr/bin/env python3
+"""Extract memory usage statistics from ESPHome build output.
+
+This script parses the PlatformIO build output to extract RAM and flash
+usage statistics for a compiled component. It's used by the CI workflow to
+compare memory usage between branches.
+
+The script reads compile output from stdin and looks for the standard
+PlatformIO output format:
+    RAM:   [====      ]  36.1% (used 29548 bytes from 81920 bytes)
+    Flash: [===       ]  34.0% (used 348511 bytes from 1023984 bytes)
+
+Optionally performs detailed memory analysis if a build directory is provided.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+import re
+import sys
+
+# Add esphome to path
+sys.path.insert(0, str(Path(__file__).parent.parent))
+
+# pylint: disable=wrong-import-position
+from esphome.analyze_memory import MemoryAnalyzer
+from esphome.platformio_api import IDEData
+from script.ci_helpers import write_github_output
+
+# Regex patterns for extracting memory usage from PlatformIO output
+_RAM_PATTERN = re.compile(r"RAM:\s+\[.*?\]\s+\d+\.\d+%\s+\(used\s+(\d+)\s+bytes")
+_FLASH_PATTERN = re.compile(r"Flash:\s+\[.*?\]\s+\d+\.\d+%\s+\(used\s+(\d+)\s+bytes")
+_BUILD_PATH_PATTERN = re.compile(r"Build path: (.+)")
+
+
+def extract_from_compile_output(
+    output_text: str,
+) -> tuple[int | None, int | None, str | None]:
+    """Extract memory usage and build directory from PlatformIO compile output.
+
+    Supports multiple builds (for component groups or isolated components).
+    When test_build_components.py creates multiple builds, this sums the
+    memory usage across all builds.
+
+    Looks for lines like:
+        RAM:   [====      ]  36.1% (used 29548 bytes from 81920 bytes)
+        Flash: [===       ]  34.0% (used 348511 bytes from 1023984 bytes)
+
+    Also extracts build directory from lines like:
+        INFO Compiling app... Build path: /path/to/build
+
+    Args:
+        output_text: Compile output text (may contain multiple builds)
+
+    Returns:
+        Tuple of (total_ram_bytes, total_flash_bytes, build_dir) or (None, None, None) if not found
+    """
+    # Find all RAM and Flash matches (may be multiple builds)
+    ram_matches = _RAM_PATTERN.findall(output_text)
+    flash_matches = _FLASH_PATTERN.findall(output_text)
+
+    if not ram_matches or not flash_matches:
+        return None, None, None
+
+    # Sum all builds (handles multiple component groups)
+    total_ram = sum(int(match) for match in ram_matches)
+    total_flash = sum(int(match) for match in flash_matches)
+
+    # Extract build directory from ESPHome's explicit build path output
+    # Look for: INFO Compiling app... Build path: /path/to/build
+    # Note: Multiple builds reuse the same build path (each overwrites the previous)
+    build_dir = None
+    if match := _BUILD_PATH_PATTERN.search(output_text):
+        build_dir = match.group(1).strip()
+
+    return total_ram, total_flash, build_dir
+
+
+def run_detailed_analysis(build_dir: str) -> dict | None:
+    """Run detailed memory analysis on build directory.
+
+    Args:
+        build_dir: Path to ESPHome build directory
+
+    Returns:
+        Dictionary with analysis results or None if analysis fails
+    """
+    build_path = Path(build_dir)
+    if not build_path.exists():
+        print(f"Build directory not found: {build_dir}", file=sys.stderr)
+        return None
+
+    # Find firmware.elf
+    elf_path = None
+    for elf_candidate in [
+        build_path / "firmware.elf",
+        build_path / ".pioenvs" / build_path.name / "firmware.elf",
+    ]:
+        if elf_candidate.exists():
+            elf_path = str(elf_candidate)
+            break
+
+    if not elf_path:
+        print(f"firmware.elf not found in {build_dir}", file=sys.stderr)
+        return None
+
+    # Find idedata.json - check multiple locations
+    device_name = build_path.name
+    idedata_candidates = [
+        # In .pioenvs for test builds
+        build_path / ".pioenvs" / device_name / "idedata.json",
+        # In .esphome/idedata for regular builds
+        Path.home() / ".esphome" / "idedata" / f"{device_name}.json",
+        # Check parent directories for .esphome/idedata (for test_build_components)
+        build_path.parent.parent.parent / "idedata" / f"{device_name}.json",
+    ]
+
+    idedata = None
+    for idedata_path in idedata_candidates:
+        if not idedata_path.exists():
+            continue
+        try:
+            with open(idedata_path, encoding="utf-8") as f:
+                raw_data = json.load(f)
+            idedata = IDEData(raw_data)
+            print(f"Loaded idedata from: {idedata_path}", file=sys.stderr)
+            break
+        except (json.JSONDecodeError, OSError) as e:
+            print(
+                f"Warning: Failed to load idedata from {idedata_path}: {e}",
+                file=sys.stderr,
+            )
+
+    analyzer = MemoryAnalyzer(elf_path, idedata=idedata)
+    components = analyzer.analyze()
+
+    # Convert to JSON-serializable format
+    result = {
+        "components": {
+            name: {
+                "text": mem.text_size,
+                "rodata": mem.rodata_size,
+                "data": mem.data_size,
+                "bss": mem.bss_size,
+                "flash_total": mem.flash_total,
+                "ram_total": mem.ram_total,
+                "symbol_count": mem.symbol_count,
+            }
+            for name, mem in components.items()
+        },
+        "symbols": {},
+    }
+
+    # Build symbol map
+    for section in analyzer.sections.values():
+        for symbol_name, size, _ in section.symbols:
+            if size > 0:
+                demangled = analyzer._demangle_symbol(symbol_name)
+                result["symbols"][demangled] = size
+
+    return result
+
+
+def main() -> int:
+    """Main entry point."""
+    parser = argparse.ArgumentParser(
+        description="Extract memory usage from ESPHome build output"
+    )
+    parser.add_argument(
+        "--output-env",
+        action="store_true",
+        help="Output to GITHUB_OUTPUT environment file",
+    )
+    parser.add_argument(
+        "--build-dir",
+        help="Optional build directory for detailed memory analysis (overrides auto-detection)",
+    )
+    parser.add_argument(
+        "--output-json",
+        help="Optional path to save detailed analysis JSON",
+    )
+    parser.add_argument(
+        "--output-build-dir",
+        help="Optional path to write the detected build directory",
+    )
+
+    args = parser.parse_args()
+
+    # Read compile output from stdin
+    compile_output = sys.stdin.read()
+
+    # Extract memory usage and build directory
+    ram_bytes, flash_bytes, detected_build_dir = extract_from_compile_output(
+        compile_output
+    )
+
+    if ram_bytes is None or flash_bytes is None:
+        print("Failed to extract memory usage from compile output", file=sys.stderr)
+        print("Expected lines like:", file=sys.stderr)
+        print(
+            "  RAM:   [====      ]  36.1% (used 29548 bytes from 81920 bytes)",
+            file=sys.stderr,
+        )
+        print(
+            "  Flash: [===       ]  34.0% (used 348511 bytes from 1023984 bytes)",
+            file=sys.stderr,
+        )
+        return 1
+
+    # Count how many builds were found
+    num_builds = len(_RAM_PATTERN.findall(compile_output))
+
+    if num_builds > 1:
+        print(
+            f"Found {num_builds} builds - summing memory usage across all builds",
+            file=sys.stderr,
+        )
+        print(
+            "WARNING: Detailed analysis will only cover the last build",
+            file=sys.stderr,
+        )
+
+    print(f"Total RAM: {ram_bytes} bytes", file=sys.stderr)
+    print(f"Total Flash: {flash_bytes} bytes", file=sys.stderr)
+
+    # Determine which build directory to use (explicit arg overrides auto-detection)
+    build_dir = args.build_dir or detected_build_dir
+
+    if detected_build_dir:
+        print(f"Detected build directory: {detected_build_dir}", file=sys.stderr)
+        if num_builds > 1:
+            print(
+                f"  (using last of {num_builds} builds for detailed analysis)",
+                file=sys.stderr,
+            )
+
+    # Write build directory to file if requested
+    if args.output_build_dir and build_dir:
+        build_dir_path = Path(args.output_build_dir)
+        build_dir_path.parent.mkdir(parents=True, exist_ok=True)
+        build_dir_path.write_text(build_dir)
+        print(f"Wrote build directory to {args.output_build_dir}", file=sys.stderr)
+
+    # Run detailed analysis if build directory available
+    detailed_analysis = None
+    if build_dir:
+        print(f"Running detailed analysis on {build_dir}", file=sys.stderr)
+        detailed_analysis = run_detailed_analysis(build_dir)
+
+    # Save JSON output if requested
+    if args.output_json:
+        output_data = {
+            "ram_bytes": ram_bytes,
+            "flash_bytes": flash_bytes,
+            "detailed_analysis": detailed_analysis,
+        }
+
+        output_path = Path(args.output_json)
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+        with open(output_path, "w", encoding="utf-8") as f:
+            json.dump(output_data, f, indent=2)
+        print(f"Saved analysis to {args.output_json}", file=sys.stderr)
+
+    if args.output_env:
+        # Output to GitHub Actions
+        write_github_output(
+            {
+                "ram_usage": ram_bytes,
+                "flash_usage": flash_bytes,
+            }
+        )
+    else:
+        print(f"{ram_bytes},{flash_bytes}")
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

script/determine-jobs.py

@@ -10,7 +10,13 @@ what files have changed. It outputs JSON with the following structure:
   "clang_format": true/false,
   "python_linters": true/false,
   "changed_components": ["component1", "component2", ...],
-  "component_test_count": 5
+  "component_test_count": 5,
+  "memory_impact": {
+    "should_run": "true/false",
+    "components": ["component1", "component2", ...],
+    "platform": "esp32-idf",
+    "use_merged_config": "true"
+  }
 }
 
 The CI workflow uses this information to:
@@ -20,6 +26,7 @@ The CI workflow uses this information to:
 - 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)
+- Run memory impact analysis whenever there are changed components (merged config), and also for core-only changes
 
 Usage:
   python script/determine-jobs.py [-b BRANCH]
@@ -31,6 +38,8 @@ Options:
 from __future__ import annotations
 
 import argparse
+from collections import Counter
+from enum import StrEnum
 from functools import cache
 import json
 import os
@@ -40,16 +49,47 @@ import sys
 from typing import Any
 
 from helpers import (
+    BASE_BUS_COMPONENTS,
     CPP_FILE_EXTENSIONS,
-    ESPHOME_COMPONENTS_PATH,
     PYTHON_FILE_EXTENSIONS,
     changed_files,
     get_all_dependencies,
+    get_component_from_path,
+    get_component_test_files,
     get_components_from_integration_fixtures,
+    parse_test_filename,
     root_path,
 )
 
 
+class Platform(StrEnum):
+    """Platform identifiers for memory impact analysis."""
+
+    ESP8266_ARD = "esp8266-ard"
+    ESP32_IDF = "esp32-idf"
+    ESP32_C3_IDF = "esp32-c3-idf"
+    ESP32_C6_IDF = "esp32-c6-idf"
+    ESP32_S2_IDF = "esp32-s2-idf"
+    ESP32_S3_IDF = "esp32-s3-idf"
+
+
+# Memory impact analysis constants
+MEMORY_IMPACT_FALLBACK_COMPONENT = "api"  # Representative component for core changes
+MEMORY_IMPACT_FALLBACK_PLATFORM = Platform.ESP32_IDF  # Most representative platform
+
+# Platform preference order for memory impact analysis
+# Prefer newer platforms first as they represent the future of ESPHome
+# ESP8266 is most constrained but many new features don't support it
+MEMORY_IMPACT_PLATFORM_PREFERENCE = [
+    Platform.ESP32_C6_IDF,  # ESP32-C6 IDF (newest, supports Thread/Zigbee)
+    Platform.ESP8266_ARD,  # ESP8266 Arduino (most memory constrained - best for impact analysis)
+    Platform.ESP32_IDF,  # ESP32 IDF platform (primary ESP32 platform, most representative)
+    Platform.ESP32_C3_IDF,  # ESP32-C3 IDF
+    Platform.ESP32_S2_IDF,  # ESP32-S2 IDF
+    Platform.ESP32_S3_IDF,  # ESP32-S3 IDF
+]
+
+
 def should_run_integration_tests(branch: str | None = None) -> bool:
     """Determine if integration tests should run based on changed files.
 
@@ -105,12 +145,9 @@ def should_run_integration_tests(branch: str | None = None) -> bool:
 
     # Check if any required components changed
     for file in files:
-        if file.startswith(ESPHOME_COMPONENTS_PATH):
+        component = get_component_from_path(file)
-            parts = file.split("/")
+        if component and component in all_required_components:
-            if len(parts) >= 3:
+            return True
-                component = parts[2]
-                if component in all_required_components:
-                    return True
 
     return False
 
@@ -224,10 +261,136 @@ def _component_has_tests(component: str) -> bool:
     Returns:
         True if the component has test YAML files
     """
-    tests_dir = Path(root_path) / "tests" / "components" / component
+    return bool(get_component_test_files(component))
-    if not tests_dir.exists():
+
-        return False
+
-    return any(tests_dir.glob("test.*.yaml"))
+def detect_memory_impact_config(
+    branch: str | None = None,
+) -> dict[str, Any]:
+    """Determine memory impact analysis configuration.
+
+    Always runs memory impact analysis when there are changed components,
+    building a merged configuration with all changed components (like
+    test_build_components.py does) to get comprehensive memory analysis.
+
+    Args:
+        branch: Branch to compare against
+
+    Returns:
+        Dictionary with memory impact analysis parameters:
+        - should_run: "true" or "false"
+        - components: list of component names to analyze
+        - platform: platform name for the merged build
+        - use_merged_config: "true" (always use merged config)
+    """
+
+    # Get actually changed files (not dependencies)
+    files = changed_files(branch)
+
+    # Find all changed components (excluding core and base bus components)
+    changed_component_set: set[str] = set()
+    has_core_changes = False
+
+    for file in files:
+        component = get_component_from_path(file)
+        if component:
+            # Skip base bus components as they're used across many builds
+            if component not in BASE_BUS_COMPONENTS:
+                changed_component_set.add(component)
+        elif file.startswith("esphome/"):
+            # Core ESPHome files changed (not component-specific)
+            has_core_changes = True
+
+    # If no components changed but core changed, test representative component
+    force_fallback_platform = False
+    if not changed_component_set and has_core_changes:
+        print(
+            f"Memory impact: No components changed, but core files changed. "
+            f"Testing {MEMORY_IMPACT_FALLBACK_COMPONENT} component on {MEMORY_IMPACT_FALLBACK_PLATFORM}.",
+            file=sys.stderr,
+        )
+        changed_component_set.add(MEMORY_IMPACT_FALLBACK_COMPONENT)
+        force_fallback_platform = True  # Use fallback platform (most representative)
+    elif not changed_component_set:
+        # No components and no core changes
+        return {"should_run": "false"}
+
+    # Find components that have tests and collect their supported platforms
+    components_with_tests: list[str] = []
+    component_platforms_map: dict[
+        str, set[Platform]
+    ] = {}  # Track which platforms each component supports
+
+    for component in sorted(changed_component_set):
+        # Look for test files on preferred platforms
+        test_files = get_component_test_files(component)
+        if not test_files:
+            continue
+
+        # Check if component has tests for any preferred platform
+        available_platforms = [
+            platform
+            for test_file in test_files
+            if (platform := parse_test_filename(test_file)[1]) != "all"
+            and platform in MEMORY_IMPACT_PLATFORM_PREFERENCE
+        ]
+
+        if not available_platforms:
+            continue
+
+        component_platforms_map[component] = set(available_platforms)
+        components_with_tests.append(component)
+
+    # If no components have tests, don't run memory impact
+    if not components_with_tests:
+        return {"should_run": "false"}
+
+    # Find common platforms supported by ALL components
+    # This ensures we can build all components together in a merged config
+    common_platforms = set(MEMORY_IMPACT_PLATFORM_PREFERENCE)
+    for component, platforms in component_platforms_map.items():
+        common_platforms &= platforms
+
+    # Select the most preferred platform from the common set
+    # Exception: for core changes, use fallback platform (most representative of codebase)
+    if force_fallback_platform:
+        platform = MEMORY_IMPACT_FALLBACK_PLATFORM
+    elif common_platforms:
+        # Pick the most preferred platform that all components support
+        platform = min(common_platforms, key=MEMORY_IMPACT_PLATFORM_PREFERENCE.index)
+    else:
+        # No common platform - pick the most commonly supported platform
+        # This allows testing components individually even if they can't be merged
+        # Count how many components support each platform
+        platform_counts = Counter(
+            p for platforms in component_platforms_map.values() for p in platforms
+        )
+        # Pick the platform supported by most components, preferring earlier in MEMORY_IMPACT_PLATFORM_PREFERENCE
+        platform = max(
+            platform_counts.keys(),
+            key=lambda p: (
+                platform_counts[p],
+                -MEMORY_IMPACT_PLATFORM_PREFERENCE.index(p),
+            ),
+        )
+
+    # Debug output
+    print("Memory impact analysis:", file=sys.stderr)
+    print(f"  Changed components: {sorted(changed_component_set)}", file=sys.stderr)
+    print(f"  Components with tests: {components_with_tests}", file=sys.stderr)
+    print(
+        f"  Component platforms: {dict(sorted(component_platforms_map.items()))}",
+        file=sys.stderr,
+    )
+    print(f"  Common platforms: {sorted(common_platforms)}", file=sys.stderr)
+    print(f"  Selected platform: {platform}", file=sys.stderr)
+
+    return {
+        "should_run": "true",
+        "components": components_with_tests,
+        "platform": platform,
+        "use_merged_config": "true",
+    }
 
 
 def main() -> None:
@@ -279,6 +442,9 @@ def main() -> None:
         if component not in directly_changed_components
     ]
 
+    # Detect components for memory impact analysis (merged config)
+    memory_impact = detect_memory_impact_config(args.branch)
+
     # Build output
     output: dict[str, Any] = {
         "integration_tests": run_integration,
@@ -292,6 +458,7 @@ def main() -> None:
         "component_test_count": len(changed_components_with_tests),
         "directly_changed_count": len(directly_changed_with_tests),
         "dependency_only_count": len(dependency_only_components),
+        "memory_impact": memory_impact,
     }
 
     # Output as JSON

script/helpers.py

@@ -29,6 +29,18 @@ YAML_FILE_EXTENSIONS = (".yaml", ".yml")
 # Component path prefix
 ESPHOME_COMPONENTS_PATH = "esphome/components/"
 
+# Base bus components - these ARE the bus implementations and should not
+# be flagged as needing migration since they are the platform/base components
+BASE_BUS_COMPONENTS = {
+    "i2c",
+    "spi",
+    "uart",
+    "modbus",
+    "canbus",
+    "remote_transmitter",
+    "remote_receiver",
+}
+
 
 def parse_list_components_output(output: str) -> list[str]:
     """Parse the output from list-components.py script.
@@ -46,6 +58,65 @@ def parse_list_components_output(output: str) -> list[str]:
     return [c.strip() for c in output.strip().split("\n") if c.strip()]
 
 
+def parse_test_filename(test_file: Path) -> tuple[str, str]:
+    """Parse test filename to extract test name and platform.
+
+    Test files follow the naming pattern: test.<platform>.yaml or test-<variant>.<platform>.yaml
+
+    Args:
+        test_file: Path to test file
+
+    Returns:
+        Tuple of (test_name, platform)
+    """
+    parts = test_file.stem.split(".")
+    if len(parts) == 2:
+        return parts[0], parts[1]  # test, platform
+    return parts[0], "all"
+
+
+def get_component_from_path(file_path: str) -> str | None:
+    """Extract component name from a file path.
+
+    Args:
+        file_path: Path to a file (e.g., "esphome/components/wifi/wifi.cpp")
+
+    Returns:
+        Component name if path is in components directory, None otherwise
+    """
+    if not file_path.startswith(ESPHOME_COMPONENTS_PATH):
+        return None
+    parts = file_path.split("/")
+    if len(parts) >= 3:
+        return parts[2]
+    return None
+
+
+def get_component_test_files(
+    component: str, *, all_variants: bool = False
+) -> list[Path]:
+    """Get test files for a component.
+
+    Args:
+        component: Component name (e.g., "wifi")
+        all_variants: If True, returns all test files including variants (test-*.yaml).
+                     If False, returns only base test files (test.*.yaml).
+                     Default is False.
+
+    Returns:
+        List of test file paths for the component, or empty list if none exist
+    """
+    tests_dir = Path(root_path) / "tests" / "components" / component
+    if not tests_dir.exists():
+        return []
+
+    if all_variants:
+        # Match both test.*.yaml and test-*.yaml patterns
+        return list(tests_dir.glob("test[.-]*.yaml"))
+    # Match only test.*.yaml (base tests)
+    return list(tests_dir.glob("test.*.yaml"))
+
+
 def styled(color: str | tuple[str, ...], msg: str, reset: bool = True) -> str:
     prefix = "".join(color) if isinstance(color, tuple) else color
     suffix = colorama.Style.RESET_ALL if reset else ""
@@ -314,11 +385,9 @@ 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_PATH):
+        component = get_component_from_path(f)
-            # Check if file belongs to any of the changed components
+        if component and component in component_set:
-            parts = f.split("/")
+            filtered_files.append(f)
-            if len(parts) >= 3 and parts[2] in component_set:
-                filtered_files.append(f)
 
     return filtered_files
 

script/list-components.py

@@ -4,7 +4,7 @@ from collections.abc import Callable
 from pathlib import Path
 import sys
 
-from helpers import changed_files, git_ls_files
+from helpers import changed_files, get_component_from_path, git_ls_files
 
 from esphome.const import (
     KEY_CORE,
@@ -30,11 +30,9 @@ def get_all_component_files() -> list[str]:
 def extract_component_names_array_from_files_array(files):
     components = []
     for file in files:
-        file_parts = file.split("/")
+        component_name = get_component_from_path(file)
-        if len(file_parts) >= 4:
+        if component_name and component_name not in components:
-            component_name = file_parts[2]
+            components.append(component_name)
-            if component_name not in components:
-                components.append(component_name)
     return components
 
 

script/split_components_for_ci.py

@@ -28,6 +28,7 @@ from script.analyze_component_buses import (
     create_grouping_signature,
     merge_compatible_bus_groups,
 )
+from script.helpers import get_component_test_files
 
 # Weighting for batch creation
 # Isolated components can't be grouped/merged, so they count as 10x
@@ -45,17 +46,12 @@ def has_test_files(component_name: str, tests_dir: Path) -> bool:
 
     Args:
         component_name: Name of the component
-        tests_dir: Path to tests/components directory
+        tests_dir: Path to tests/components directory (unused, kept for compatibility)
 
     Returns:
         True if the component has test.*.yaml files
     """
-    component_dir = tests_dir / component_name
+    return bool(get_component_test_files(component_name))
-    if not component_dir.exists() or not component_dir.is_dir():
-        return False
-
-    # Check for test.*.yaml files
-    return any(component_dir.glob("test.*.yaml"))
 
 
 def create_intelligent_batches(

script/templates/ci_memory_impact_comment_template.j2

@@ -0,0 +1,27 @@
+{{ comment_marker }}
+## Memory Impact Analysis
+
+**Components:** {{ components_str }}
+**Platform:** `{{ platform }}`
+
+| Metric | Target Branch | This PR | Change |
+|--------|--------------|---------|--------|
+| **RAM** | {{ target_ram }} | {{ pr_ram }} | {{ ram_change }} |
+| **Flash** | {{ target_flash }} | {{ pr_flash }} | {{ flash_change }} |
+{% if component_breakdown %}
+{{ component_breakdown }}
+{% endif %}
+{% if symbol_changes %}
+{{ symbol_changes }}
+{% endif %}
+{%- if target_cache_hit %}
+
+> ⚡ Target branch analysis was loaded from cache (build skipped for faster CI).
+{%- endif %}
+
+---
+> **Note:** This analysis measures **static RAM and Flash usage** only (compile-time allocation).
+> **Dynamic memory (heap)** cannot be measured automatically.
+> **⚠️ You must test this PR on a real device** to measure free heap and ensure no runtime memory issues.
+
+*This analysis runs automatically when components change. Memory usage is measured from {{ config_note }}.*

script/templates/ci_memory_impact_component_breakdown.j2

@@ -0,0 +1,15 @@
+
+<details open>
+<summary>📊 Component Memory Breakdown</summary>
+
+| Component | Target Flash | PR Flash | Change |
+|-----------|--------------|----------|--------|
+{% for comp, target_flash, pr_flash, delta in changed_components[:max_rows] -%}
+{% set threshold = component_change_threshold if comp.startswith("[esphome]") else none -%}
+| `{{ comp }}` | {{ target_flash|format_bytes }} | {{ pr_flash|format_bytes }} | {{ format_change(target_flash, pr_flash, threshold=threshold) }} |
+{% endfor -%}
+{% if changed_components|length > max_rows -%}
+| ... | ... | ... | *({{ changed_components|length - max_rows }} more components not shown)* |
+{% endif -%}
+
+</details>

script/templates/ci_memory_impact_macros.j2

@@ -0,0 +1,8 @@
+{#- Macro for formatting symbol names in tables -#}
+{%- macro format_symbol(symbol, max_length, truncate_length) -%}
+{%- if symbol|length <= max_length -%}
+`{{ symbol }}`
+{%- else -%}
+<details><summary><code>{{ symbol[:truncate_length] }}...</code></summary><code>{{ symbol }}</code></details>
+{%- endif -%}
+{%- endmacro -%}

script/templates/ci_memory_impact_symbol_changes.j2

@@ -0,0 +1,51 @@
+{%- from 'ci_memory_impact_macros.j2' import format_symbol -%}
+
+<details>
+<summary>🔍 Symbol-Level Changes (click to expand)</summary>
+
+{% if changed_symbols %}
+
+### Changed Symbols
+
+| Symbol | Target Size | PR Size | Change |
+|--------|-------------|---------|--------|
+{% for symbol, target_size, pr_size, delta in changed_symbols[:max_changed_rows] -%}
+| {{ format_symbol(symbol, symbol_max_length, symbol_truncate_length) }} | {{ target_size|format_bytes }} | {{ pr_size|format_bytes }} | {{ format_change(target_size, pr_size) }} |
+{% endfor -%}
+{% if changed_symbols|length > max_changed_rows -%}
+| ... | ... | ... | *({{ changed_symbols|length - max_changed_rows }} more changed symbols not shown)* |
+{% endif -%}
+
+{% endif %}
+{% if new_symbols %}
+
+### New Symbols (top {{ max_new_rows }})
+
+| Symbol | Size |
+|--------|------|
+{% for symbol, size in new_symbols[:max_new_rows] -%}
+| {{ format_symbol(symbol, symbol_max_length, symbol_truncate_length) }} | {{ size|format_bytes }} |
+{% endfor -%}
+{% if new_symbols|length > max_new_rows -%}
+{% set total_new_size = new_symbols|sum(attribute=1) -%}
+| *{{ new_symbols|length - max_new_rows }} more new symbols...* | *Total: {{ total_new_size|format_bytes }}* |
+{% endif -%}
+
+{% endif %}
+{% if removed_symbols %}
+
+### Removed Symbols (top {{ max_removed_rows }})
+
+| Symbol | Size |
+|--------|------|
+{% for symbol, size in removed_symbols[:max_removed_rows] -%}
+| {{ format_symbol(symbol, symbol_max_length, symbol_truncate_length) }} | {{ size|format_bytes }} |
+{% endfor -%}
+{% if removed_symbols|length > max_removed_rows -%}
+{% set total_removed_size = removed_symbols|sum(attribute=1) -%}
+| *{{ removed_symbols|length - max_removed_rows }} more removed symbols...* | *Total: {{ total_removed_size|format_bytes }}* |
+{% endif -%}
+
+{% endif %}
+
+</details>

script/test_build_components.py

@@ -39,6 +39,7 @@ from script.analyze_component_buses import (
     merge_compatible_bus_groups,
     uses_local_file_references,
 )
+from script.helpers import get_component_test_files
 from script.merge_component_configs import merge_component_configs
 
 
@@ -82,13 +83,14 @@ def show_disk_space_if_ci(esphome_command: str) -> None:
 
 
 def find_component_tests(
-    components_dir: Path, component_pattern: str = "*"
+    components_dir: Path, component_pattern: str = "*", base_only: bool = False
 ) -> dict[str, list[Path]]:
     """Find all component test files.
 
     Args:
         components_dir: Path to tests/components directory
         component_pattern: Glob pattern for component names
+        base_only: If True, only find base test files (test.*.yaml), not variant files (test-*.yaml)
 
     Returns:
         Dictionary mapping component name to list of test files
@@ -99,9 +101,10 @@ def find_component_tests(
         if not comp_dir.is_dir():
             continue
 
-        # Find test files matching test.*.yaml or test-*.yaml patterns
+        # Get test files using helper function
-        for test_file in comp_dir.glob("test[.-]*.yaml"):
+        test_files = get_component_test_files(comp_dir.name, all_variants=not base_only)
-            component_tests[comp_dir.name].append(test_file)
+        if test_files:
+            component_tests[comp_dir.name] = test_files
 
     return dict(component_tests)
 
@@ -931,6 +934,7 @@ def test_components(
     continue_on_fail: bool,
     enable_grouping: bool = True,
     isolated_components: set[str] | None = None,
+    base_only: bool = False,
 ) -> int:
     """Test components with optional intelligent grouping.
 
@@ -944,6 +948,7 @@ def test_components(
             These are tested WITHOUT --testing-mode to enable full validation
             (pin conflicts, etc). This is used in CI for directly changed components
             to catch issues that would be missed with --testing-mode.
+        base_only: If True, only test base test files (test.*.yaml), not variant files (test-*.yaml)
 
     Returns:
         Exit code (0 for success, 1 for failure)
@@ -961,7 +966,7 @@ def test_components(
     # Find all component tests
     all_tests = {}
     for pattern in component_patterns:
-        all_tests.update(find_component_tests(tests_dir, pattern))
+        all_tests.update(find_component_tests(tests_dir, pattern, base_only))
 
     if not all_tests:
         print(f"No components found matching: {component_patterns}")
@@ -1122,6 +1127,11 @@ def main() -> int:
         "These are tested WITHOUT --testing-mode to enable full validation. "
         "Used in CI for directly changed components to catch pin conflicts and other issues.",
     )
+    parser.add_argument(
+        "--base-only",
+        action="store_true",
+        help="Only test base test files (test.*.yaml), not variant files (test-*.yaml)",
+    )
 
     args = parser.parse_args()
 
@@ -1140,6 +1150,7 @@ def main() -> int:
         continue_on_fail=args.continue_on_fail,
         enable_grouping=not args.no_grouping,
         isolated_components=isolated_components,
+        base_only=args.base_only,
     )
 
 

tests/script/test_determine_jobs.py

@@ -17,6 +17,9 @@ script_dir = os.path.abspath(
 )
 sys.path.insert(0, script_dir)
 
+# Import helpers module for patching
+import helpers  # noqa: E402
+
 spec = importlib.util.spec_from_file_location(
     "determine_jobs", os.path.join(script_dir, "determine-jobs.py")
 )
@@ -59,15 +62,29 @@ def mock_subprocess_run() -> Generator[Mock, None, None]:
         yield mock
 
 
+@pytest.fixture
+def mock_changed_files() -> Generator[Mock, None, None]:
+    """Mock changed_files for memory impact detection."""
+    with patch.object(determine_jobs, "changed_files") as mock:
+        # Default to empty list
+        mock.return_value = []
+        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,
+    mock_changed_files: Mock,
     capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     """Test when all tests should run."""
+    # Ensure we're not in GITHUB_ACTIONS mode for this test
+    monkeypatch.delenv("GITHUB_ACTIONS", raising=False)
+
     mock_should_run_integration_tests.return_value = True
     mock_should_run_clang_tidy.return_value = True
     mock_should_run_clang_format.return_value = True
@@ -100,6 +117,9 @@ def test_main_all_tests_should_run(
     assert output["component_test_count"] == len(
         output["changed_components_with_tests"]
     )
+    # memory_impact should be present
+    assert "memory_impact" in output
+    assert output["memory_impact"]["should_run"] == "false"  # No files changed
 
 
 def test_main_no_tests_should_run(
@@ -108,9 +128,14 @@ def test_main_no_tests_should_run(
     mock_should_run_clang_format: Mock,
     mock_should_run_python_linters: Mock,
     mock_subprocess_run: Mock,
+    mock_changed_files: Mock,
     capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     """Test when no tests should run."""
+    # Ensure we're not in GITHUB_ACTIONS mode for this test
+    monkeypatch.delenv("GITHUB_ACTIONS", raising=False)
+
     mock_should_run_integration_tests.return_value = False
     mock_should_run_clang_tidy.return_value = False
     mock_should_run_clang_format.return_value = False
@@ -136,6 +161,9 @@ def test_main_no_tests_should_run(
     assert output["changed_components"] == []
     assert output["changed_components_with_tests"] == []
     assert output["component_test_count"] == 0
+    # memory_impact should be present
+    assert "memory_impact" in output
+    assert output["memory_impact"]["should_run"] == "false"
 
 
 def test_main_list_components_fails(
@@ -169,9 +197,14 @@ def test_main_with_branch_argument(
     mock_should_run_clang_format: Mock,
     mock_should_run_python_linters: Mock,
     mock_subprocess_run: Mock,
+    mock_changed_files: Mock,
     capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     """Test with branch argument."""
+    # Ensure we're not in GITHUB_ACTIONS mode for this test
+    monkeypatch.delenv("GITHUB_ACTIONS", raising=False)
+
     mock_should_run_integration_tests.return_value = False
     mock_should_run_clang_tidy.return_value = True
     mock_should_run_clang_format.return_value = False
@@ -216,6 +249,9 @@ def test_main_with_branch_argument(
     assert output["component_test_count"] == len(
         output["changed_components_with_tests"]
     )
+    # memory_impact should be present
+    assert "memory_impact" in output
+    assert output["memory_impact"]["should_run"] == "false"
 
 
 def test_should_run_integration_tests(
@@ -403,10 +439,15 @@ def test_main_filters_components_without_tests(
     mock_should_run_clang_format: Mock,
     mock_should_run_python_linters: Mock,
     mock_subprocess_run: Mock,
+    mock_changed_files: Mock,
     capsys: pytest.CaptureFixture[str],
     tmp_path: Path,
+    monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     """Test that components without test files are filtered out."""
+    # Ensure we're not in GITHUB_ACTIONS mode for this test
+    monkeypatch.delenv("GITHUB_ACTIONS", raising=False)
+
     mock_should_run_integration_tests.return_value = False
     mock_should_run_clang_tidy.return_value = False
     mock_should_run_clang_format.return_value = False
@@ -440,9 +481,10 @@ def test_main_filters_components_without_tests(
     airthings_dir = tests_dir / "airthings_ble"
     airthings_dir.mkdir(parents=True)
 
-    # Mock root_path to use tmp_path
+    # Mock root_path to use tmp_path (need to patch both determine_jobs and helpers)
     with (
         patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
         patch("sys.argv", ["determine-jobs.py"]),
     ):
         # Clear the cache since we're mocking root_path
@@ -459,3 +501,188 @@ def test_main_filters_components_without_tests(
     assert set(output["changed_components_with_tests"]) == {"wifi", "sensor"}
     # component_test_count should be based on components with tests
     assert output["component_test_count"] == 2
+    # memory_impact should be present
+    assert "memory_impact" in output
+    assert output["memory_impact"]["should_run"] == "false"
+
+
+# Tests for detect_memory_impact_config function
+
+
+def test_detect_memory_impact_config_with_common_platform(tmp_path: Path) -> None:
+    """Test memory impact detection when components share a common platform."""
+    # Create test directory structure
+    tests_dir = tmp_path / "tests" / "components"
+
+    # wifi component with esp32-idf test
+    wifi_dir = tests_dir / "wifi"
+    wifi_dir.mkdir(parents=True)
+    (wifi_dir / "test.esp32-idf.yaml").write_text("test: wifi")
+
+    # api component with esp32-idf test
+    api_dir = tests_dir / "api"
+    api_dir.mkdir(parents=True)
+    (api_dir / "test.esp32-idf.yaml").write_text("test: api")
+
+    # Mock changed_files to return wifi and api component changes
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = [
+            "esphome/components/wifi/wifi.cpp",
+            "esphome/components/api/api.cpp",
+        ]
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    assert result["should_run"] == "true"
+    assert set(result["components"]) == {"wifi", "api"}
+    assert result["platform"] == "esp32-idf"  # Common platform
+    assert result["use_merged_config"] == "true"
+
+
+def test_detect_memory_impact_config_core_only_changes(tmp_path: Path) -> None:
+    """Test memory impact detection with core-only changes (no component changes)."""
+    # Create test directory structure with fallback component
+    tests_dir = tmp_path / "tests" / "components"
+
+    # api component (fallback component) with esp32-idf test
+    api_dir = tests_dir / "api"
+    api_dir.mkdir(parents=True)
+    (api_dir / "test.esp32-idf.yaml").write_text("test: api")
+
+    # Mock changed_files to return only core files (no component files)
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = [
+            "esphome/core/application.cpp",
+            "esphome/core/component.h",
+        ]
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    assert result["should_run"] == "true"
+    assert result["components"] == ["api"]  # Fallback component
+    assert result["platform"] == "esp32-idf"  # Fallback platform
+    assert result["use_merged_config"] == "true"
+
+
+def test_detect_memory_impact_config_no_common_platform(tmp_path: Path) -> None:
+    """Test memory impact detection when components have no common platform."""
+    # Create test directory structure
+    tests_dir = tmp_path / "tests" / "components"
+
+    # wifi component only has esp32-idf test
+    wifi_dir = tests_dir / "wifi"
+    wifi_dir.mkdir(parents=True)
+    (wifi_dir / "test.esp32-idf.yaml").write_text("test: wifi")
+
+    # logger component only has esp8266-ard test
+    logger_dir = tests_dir / "logger"
+    logger_dir.mkdir(parents=True)
+    (logger_dir / "test.esp8266-ard.yaml").write_text("test: logger")
+
+    # Mock changed_files to return both components
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = [
+            "esphome/components/wifi/wifi.cpp",
+            "esphome/components/logger/logger.cpp",
+        ]
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    # Should pick the most frequently supported platform
+    assert result["should_run"] == "true"
+    assert set(result["components"]) == {"wifi", "logger"}
+    # When no common platform, picks most commonly supported
+    # esp8266-ard is preferred over esp32-idf in the preference list
+    assert result["platform"] in ["esp32-idf", "esp8266-ard"]
+    assert result["use_merged_config"] == "true"
+
+
+def test_detect_memory_impact_config_no_changes(tmp_path: Path) -> None:
+    """Test memory impact detection when no files changed."""
+    # Mock changed_files to return empty list
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = []
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    assert result["should_run"] == "false"
+
+
+def test_detect_memory_impact_config_no_components_with_tests(tmp_path: Path) -> None:
+    """Test memory impact detection when changed components have no tests."""
+    # Create test directory structure
+    tests_dir = tmp_path / "tests" / "components"
+
+    # Create component directory but no test files
+    custom_component_dir = tests_dir / "my_custom_component"
+    custom_component_dir.mkdir(parents=True)
+
+    # Mock changed_files to return component without tests
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = [
+            "esphome/components/my_custom_component/component.cpp",
+        ]
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    assert result["should_run"] == "false"
+
+
+def test_detect_memory_impact_config_skips_base_bus_components(tmp_path: Path) -> None:
+    """Test that base bus components (i2c, spi, uart) are skipped."""
+    # Create test directory structure
+    tests_dir = tmp_path / "tests" / "components"
+
+    # i2c component (should be skipped as it's a base bus component)
+    i2c_dir = tests_dir / "i2c"
+    i2c_dir.mkdir(parents=True)
+    (i2c_dir / "test.esp32-idf.yaml").write_text("test: i2c")
+
+    # wifi component (should not be skipped)
+    wifi_dir = tests_dir / "wifi"
+    wifi_dir.mkdir(parents=True)
+    (wifi_dir / "test.esp32-idf.yaml").write_text("test: wifi")
+
+    # Mock changed_files to return both i2c and wifi
+    with (
+        patch.object(determine_jobs, "root_path", str(tmp_path)),
+        patch.object(helpers, "root_path", str(tmp_path)),
+        patch.object(determine_jobs, "changed_files") as mock_changed_files,
+    ):
+        mock_changed_files.return_value = [
+            "esphome/components/i2c/i2c.cpp",
+            "esphome/components/wifi/wifi.cpp",
+        ]
+        determine_jobs._component_has_tests.cache_clear()
+
+        result = determine_jobs.detect_memory_impact_config()
+
+    # Should only include wifi, not i2c
+    assert result["should_run"] == "true"
+    assert result["components"] == ["wifi"]
+    assert "i2c" not in result["components"]

tests/unit_tests/test_platformio_api.py

@@ -387,6 +387,42 @@ def test_idedata_addr2line_path_unix(setup_core: Path) -> None:
     assert result == "/usr/bin/addr2line"
 
 
+def test_idedata_objdump_path_windows(setup_core: Path) -> None:
+    """Test IDEData.objdump_path on Windows."""
+    raw_data = {"prog_path": "/path/to/firmware.elf", "cc_path": "C:\\tools\\gcc.exe"}
+    idedata = platformio_api.IDEData(raw_data)
+
+    result = idedata.objdump_path
+    assert result == "C:\\tools\\objdump.exe"
+
+
+def test_idedata_objdump_path_unix(setup_core: Path) -> None:
+    """Test IDEData.objdump_path on Unix."""
+    raw_data = {"prog_path": "/path/to/firmware.elf", "cc_path": "/usr/bin/gcc"}
+    idedata = platformio_api.IDEData(raw_data)
+
+    result = idedata.objdump_path
+    assert result == "/usr/bin/objdump"
+
+
+def test_idedata_readelf_path_windows(setup_core: Path) -> None:
+    """Test IDEData.readelf_path on Windows."""
+    raw_data = {"prog_path": "/path/to/firmware.elf", "cc_path": "C:\\tools\\gcc.exe"}
+    idedata = platformio_api.IDEData(raw_data)
+
+    result = idedata.readelf_path
+    assert result == "C:\\tools\\readelf.exe"
+
+
+def test_idedata_readelf_path_unix(setup_core: Path) -> None:
+    """Test IDEData.readelf_path on Unix."""
+    raw_data = {"prog_path": "/path/to/firmware.elf", "cc_path": "/usr/bin/gcc"}
+    idedata = platformio_api.IDEData(raw_data)
+
+    result = idedata.readelf_path
+    assert result == "/usr/bin/readelf"
+
+
 def test_patch_structhash(setup_core: Path) -> None:
     """Test patch_structhash monkey patches platformio functions."""
     # Create simple namespace objects to act as modules