prioritize-unused-includes.yml

name: Prioritize Unused Chromium Includes

on:
  workflow_dispatch:
  schedule:
    - cron: '0 18 * * *'

permissions: {}

jobs:
  prioritized_by_added_size:
    name: Prioritized by added size
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          repository: dsanders11/chromium-include-cleanup
      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
        with:
          python-version: '3.12'
          cache: 'pip'
      - name: Install Dependencies
        run: pip install -r requirements.txt
      - name: Download Unused Includes Output
        run: |
          curl https://gist.githubusercontent.com/dsanders11/ccf3505dcbb35492cbadade1ceb663b6/raw/unused-includes.csv > unused-includes.csv
      - name: Download Include Analysis Output
        run: |
          curl https://commondatastorage.googleapis.com/chromium-browser-clang/include-analysis.js > include-analysis.js
      - name: Find Priority Unused Includes
        run: |
          python set_edge_weights.py unused-includes.csv include-analysis.js --config chromium > weighted-unused-includes.csv
          python filter_include_changes.py weighted-unused-includes.csv --config chromium --filter-third-party --no-filter-mojom-headers --weight-threshold 10000000 > priority-unused-includes.csv
          python set_edge_weights.py priority-unused-includes.csv include-analysis.js --config chromium --metric expanded_size > priority-unused-includes.expanded_size.csv
          python set_edge_weights.py priority-unused-includes.csv include-analysis.js --config chromium --metric prevalence > priority-unused-includes.prevalence.csv
          python set_edge_weights.py priority-unused-includes.csv include-analysis.js --config chromium --metric includer_size > priority-unused-includes.includer_size.csv
          python set_edge_weights.py priority-unused-includes.csv include-analysis.js --config chromium --metric centrality > priority-unused-includes.centrality.csv
      - uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0
        with:
          name: weighted-unused-includes
          path: |
            weighted-unused-includes.csv
            priority-unused-includes.csv
            priority-unused-includes.expanded_size.csv
            priority-unused-includes.prevalence.csv
            priority-unused-includes.includer_size.csv
            priority-unused-includes.centrality.csv
      - run: npm install @actions/cache
      - uses: actions/github-script@ed597411d8f924073f98dfc5c65a23a2325f34cd # v8.0.0
        with:
          script: |
            const fs = require('node:fs');
            const cache = await import('${{ github.workspace }}/node_modules/@actions/cache/lib/cache.js');

            const data = fs.readFileSync('./priority-unused-includes.csv', 'utf8').trim().split('\n');
            const expandedSizeData = fs.readFileSync('./priority-unused-includes.expanded_size.csv', 'utf8').trim().split('\n');
            const prevalenceData = fs.readFileSync('./priority-unused-includes.prevalence.csv', 'utf8').trim().split('\n');
            const includerSizeData = fs.readFileSync('./priority-unused-includes.includer_size.csv', 'utf8').trim().split('\n');
            const centralityData = fs.readFileSync('./priority-unused-includes.centrality.csv', 'utf8').trim().split('\n');
            let totalAddedSize = 0;

            const unusedIncludes = await Promise.all(data.map(async (line, idx) => {
              const [,, filename, include, added_size] = line.trim().split(',');
              const addedSize = parseInt(added_size);
              totalAddedSize += addedSize;

              // Check if this is known from a previous run
              const cacheKey = `prioritize-unused-chromium-${filename}-${include}`;
              const cacheHit =
                (await cache.restoreCache(['/dev/null'], cacheKey, undefined, {
                  lookupOnly: true,
                })) !== undefined;

              if (!cacheHit) {
                // Create a cache entry (only the name matters) to keep track of
                // includes we've seen from previous runs to mark them as stale
                await cache.saveCache(['/dev/null'], cacheKey);
              }

              return [
                `${filename} --> ${include.replace(/</g, '&lt;').replace(/>/g, '&gt;')}`,
                addedSize,
                parseFloat(prevalenceData[idx].trim().split(',')[4]),
                parseInt(expandedSizeData[idx].trim().split(',')[4]),
                parseInt(includerSizeData[idx].trim().split(',')[4]),
                parseFloat(centralityData[idx].trim().split(',')[4]),
                cacheHit,
              ];
            }));

            const addTable = (includes) => {
              core.summary.addTable([
                [
                  { data: 'Include Edge', header: true },
                  { data: 'Added Size', header: true },
                  { data: 'Prevalence', header: true },
                  { data: 'Expanded Size', header: true },
                  { data: 'Includer Size', header: true },
                  { data: 'Centrality', header: true },
                ],
                // Sort by added size, then convert it back to string or it won't render
                ...includes
                  .sort((a, b) => b[1] - a[1])
                  .map(([edge, added_size, prevalence, expandedSize, includerSize, centrality]) => [
                    edge,
                    added_size.toLocaleString(),
                    `${prevalence.toFixed(2)}%`,
                    expandedSize.toLocaleString(),
                    includerSize.toLocaleString(),
                    centrality.toFixed(5),
                  ]),
              ]);
            }

            if (unusedIncludes.length > 0) {
              core.summary.addHeading('✂️ Priority Unused Chromium Includes');
              core.summary.addRaw(`\n> [!NOTE]\n> These are suggestions from \`clangd\` and are known to contain false positives.\n\n`);
              core.summary.addRaw(`Found ${unusedIncludes.length} unused includes over 10 MB of added size for a total of ${totalAddedSize.toLocaleString()} bytes of added size`);

              const newlySeen = unusedIncludes.filter(([, , , , , , cacheHit]) => !cacheHit)
              if (newlySeen.length > 0) {
                core.summary.addHeading('Not Seen Before', '2');
                addTable(newlySeen);
                core.summary.addHeading('All Unused Includes', '2');
              } else {
                core.summary.addBreak();
              }

              addTable(unusedIncludes);
            } else {
              core.summary.addRaw('🎉 No priority unused Chromium includes');
            }

            await core.summary.write();

  prioritized_by_prevalence:
    name: Prioritized by prevalence
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          repository: dsanders11/chromium-include-cleanup
      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
        with:
          python-version: '3.12'
          cache: 'pip'
      - name: Install Dependencies
        run: pip install -r requirements.txt
      - name: Download Unused Includes Output
        run: |
          curl https://gist.githubusercontent.com/dsanders11/ccf3505dcbb35492cbadade1ceb663b6/raw/unused-includes.csv > unused-includes.csv
      - name: Download Include Analysis Output
        run: |
          curl https://commondatastorage.googleapis.com/chromium-browser-clang/include-analysis.js > include-analysis.js
      - name: Find Priority Unused Includes
        run: |
          python set_edge_weights.py unused-includes.csv include-analysis.js --config chromium --metric prevalence > weighted-unused-includes.csv
          python filter_include_changes.py weighted-unused-includes.csv --config chromium --filter-third-party --no-filter-mojom-headers --weight-threshold 5 > priority-unused-includes.prevalence.csv
          python set_edge_weights.py priority-unused-includes.prevalence.csv include-analysis.js --config chromium --metric expanded_size > priority-unused-includes.expanded_size.csv
          python set_edge_weights.py priority-unused-includes.prevalence.csv include-analysis.js --config chromium --metric input_size > priority-unused-includes.input_size.csv
          python set_edge_weights.py priority-unused-includes.prevalence.csv include-analysis.js --config chromium --metric includer_size > priority-unused-includes.includer_size.csv
          python set_edge_weights.py priority-unused-includes.prevalence.csv include-analysis.js --config chromium --metric centrality > priority-unused-includes.centrality.csv
      - run: npm install @actions/cache
      - uses: actions/github-script@ed597411d8f924073f98dfc5c65a23a2325f34cd # v8.0.0
        with:
          script: |
            const fs = require('node:fs');
            const cache = await import('${{ github.workspace }}/node_modules/@actions/cache/lib/cache.js');

            const addedSizeData = fs.readFileSync('./priority-unused-includes.input_size.csv', 'utf8').trim().split('\n');
            const expandedSizeData = fs.readFileSync('./priority-unused-includes.expanded_size.csv', 'utf8').trim().split('\n');
            const prevalenceData = fs.readFileSync('./priority-unused-includes.prevalence.csv', 'utf8').trim().split('\n');
            const includerSizeData = fs.readFileSync('./priority-unused-includes.includer_size.csv', 'utf8').trim().split('\n');
            const centralityData = fs.readFileSync('./priority-unused-includes.centrality.csv', 'utf8').trim().split('\n');

            const unusedIncludes = await Promise.all(prevalenceData.map(async (line, idx) => {
              const [,, filename, include, prevalence] = line.trim().split(',');

              // Check if this is known from a previous run
              const cacheKey = `prioritize-unused-chromium-${filename}-${include}-prevalence`;
              const cacheHit =
                (await cache.restoreCache(['/dev/null'], cacheKey, undefined, {
                  lookupOnly: true,
                })) !== undefined;

              if (!cacheHit) {
                // Create a cache entry (only the name matters) to keep track of
                // includes we've seen from previous runs to mark them as stale
                await cache.saveCache(['/dev/null'], cacheKey);
              }

              return [
                `${filename} --> ${include.replace(/</g, '&lt;').replace(/>/g, '&gt;')}`,
                parseFloat(prevalence),
                parseInt(addedSizeData[idx].trim().split(',')[4]),
                parseInt(expandedSizeData[idx].trim().split(',')[4]),
                parseInt(includerSizeData[idx].trim().split(',')[4]),
                parseFloat(centralityData[idx].trim().split(',')[4]),
                cacheHit,
              ];
            }));

            const addTable = (includes) => {
              core.summary.addTable([
                [
                  { data: 'Include Edge', header: true },
                  { data: 'Prevalence', header: true },
                  { data: 'Added Size', header: true },
                  { data: 'Expanded Size', header: true },
                  { data: 'Includer Size', header: true },
                  { data: 'Centrality', header: true },
                ],
                // Sort by prevalence, then convert it back to string or it won't render
                ...includes
                  .sort((a, b) => b[1] - a[1])
                  .map(([edge, prevalence, added_size, expandedSize, includerSize, centrality]) => [
                    edge,
                    `${prevalence.toFixed(2)}%`,
                    added_size.toLocaleString(),
                    expandedSize.toLocaleString(),
                    includerSize.toLocaleString(),
                    centrality.toFixed(5),
                  ]),
              ]);
            }

            if (unusedIncludes.length > 0) {
              core.summary.addHeading('✂️ Priority Unused Chromium Includes');
              core.summary.addRaw(`\n> [!NOTE]\n> These are suggestions from \`clangd\` and are known to contain false positives.\n\n`);
              core.summary.addRaw(`Found ${unusedIncludes.length} unused includes with over 5% prevalence`);

              const newlySeen = unusedIncludes.filter(([, , , , , , cacheHit]) => !cacheHit)
              if (newlySeen.length > 0) {
                core.summary.addHeading('Not Seen Before', '2');
                addTable(newlySeen);
                core.summary.addHeading('All Unused Includes', '2');
              } else {
                core.summary.addBreak();
              }

              addTable(unusedIncludes);
            } else {
              core.summary.addRaw('🎉 No priority unused Chromium includes');
            }

            await core.summary.write();

  prioritized_by_centrality:
    name: Prioritized by centrality
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          repository: dsanders11/chromium-include-cleanup
      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
        with:
          python-version: '3.12'
          cache: 'pip'
      - name: Install Dependencies
        run: pip install -r requirements.txt
      - name: Download Unused Includes Output
        run: |
          curl https://gist.githubusercontent.com/dsanders11/ccf3505dcbb35492cbadade1ceb663b6/raw/unused-includes.csv > unused-includes.csv
      - name: Download Include Analysis Output
        run: |
          curl https://commondatastorage.googleapis.com/chromium-browser-clang/include-analysis.js > include-analysis.js
      - name: Find Priority Unused Includes
        run: |
          python set_edge_weights.py unused-includes.csv include-analysis.js --config chromium --metric centrality > weighted-unused-includes.csv
          python filter_include_changes.py weighted-unused-includes.csv --config chromium --filter-third-party --no-filter-mojom-headers --weight-threshold 0.005 > priority-unused-includes.centrality.csv
          python set_edge_weights.py priority-unused-includes.centrality.csv include-analysis.js --config chromium --metric expanded_size > priority-unused-includes.expanded_size.csv
          python set_edge_weights.py priority-unused-includes.centrality.csv include-analysis.js --config chromium --metric input_size > priority-unused-includes.input_size.csv
          python set_edge_weights.py priority-unused-includes.centrality.csv include-analysis.js --config chromium --metric includer_size > priority-unused-includes.includer_size.csv
          python set_edge_weights.py priority-unused-includes.centrality.csv include-analysis.js --config chromium --metric prevalence > priority-unused-includes.prevalence.csv
      - run: npm install @actions/cache
      - uses: actions/github-script@ed597411d8f924073f98dfc5c65a23a2325f34cd # v8.0.0
        with:
          script: |
            const fs = require('node:fs');
            const cache = await import('${{ github.workspace }}/node_modules/@actions/cache/lib/cache.js');

            const addedSizeData = fs.readFileSync('./priority-unused-includes.input_size.csv', 'utf8').trim().split('\n');
            const expandedSizeData = fs.readFileSync('./priority-unused-includes.expanded_size.csv', 'utf8').trim().split('\n');
            const prevalenceData = fs.readFileSync('./priority-unused-includes.prevalence.csv', 'utf8').trim().split('\n');
            const includerSizeData = fs.readFileSync('./priority-unused-includes.includer_size.csv', 'utf8').trim().split('\n');
            const centralityData = fs.readFileSync('./priority-unused-includes.centrality.csv', 'utf8').trim().split('\n');

            const unusedIncludes = await Promise.all(centralityData.map(async (line, idx) => {
              const [,, filename, include, centrality] = line.trim().split(',');

              // Check if this is known from a previous run
              const cacheKey = `prioritize-unused-chromium-${filename}-${include}-centrality`;
              const cacheHit =
                (await cache.restoreCache(['/dev/null'], cacheKey, undefined, {
                  lookupOnly: true,
                })) !== undefined;

              if (!cacheHit) {
                // Create a cache entry (only the name matters) to keep track of
                // includes we've seen from previous runs to mark them as stale
                await cache.saveCache(['/dev/null'], cacheKey);
              }

              return [
                `${filename} --> ${include.replace(/</g, '&lt;').replace(/>/g, '&gt;')}`,
                parseFloat(centrality),
                parseInt(addedSizeData[idx].trim().split(',')[4]),
                parseFloat(prevalenceData[idx].trim().split(',')[4]),
                parseInt(expandedSizeData[idx].trim().split(',')[4]),
                parseInt(includerSizeData[idx].trim().split(',')[4]),
                cacheHit,
              ];
            }));

            const addTable = (includes) => {
              core.summary.addTable([
                [
                  { data: 'Include Edge', header: true },
                  { data: 'Centrality', header: true },
                  { data: 'Added Size', header: true },
                  { data: 'Prevalence', header: true },
                  { data: 'Expanded Size', header: true },
                  { data: 'Includer Size', header: true },
                ],
                // Sort by centrality
                ...includes
                  .sort((a, b) => b[1] - a[1])
                  .map(([edge, centrality, added_size, prevalence, expandedSize, includerSize]) => [
                    edge,
                    centrality.toFixed(5),
                    added_size.toLocaleString(),
                    `${prevalence.toFixed(2)}%`,
                    expandedSize.toLocaleString(),
                    includerSize.toLocaleString(),
                  ]),
              ]);
            }

            if (unusedIncludes.length > 0) {
              core.summary.addHeading('✂️ Priority Unused Chromium Includes');
              core.summary.addRaw(`\n> [!NOTE]\n> These are suggestions from \`clangd\` and are known to contain false positives.\n\n`);
              core.summary.addRaw(`Found ${unusedIncludes.length} unused includes with >= 0.005 centrality`);

              const newlySeen = unusedIncludes.filter(([, , , , , , cacheHit]) => !cacheHit)
              if (newlySeen.length > 0) {
                core.summary.addHeading('Not Seen Before', '2');
                addTable(newlySeen);
                core.summary.addHeading('All Unused Includes', '2');
              } else {
                core.summary.addBreak();
              }

              addTable(unusedIncludes);
            } else {
              core.summary.addRaw('🎉 No priority unused Chromium includes');
            }

            await core.summary.write();