analyze-code.yml

name: Analyze Repositories with PAT

on:
  schedule:
    - cron: "0 0 * * 0"
  workflow_dispatch:

permissions:
  contents: write

jobs:
  count-lines:
    # Keep this workflow copyable without running PAT analysis in this demo repo.
    if: github.repository != 'arhamkhnz/github-code-analyzer'
    runs-on: ubuntu-latest
    env:
      # Use cloc language names. Highlighted languages get their own README row.
      HIGHLIGHT_LANGS: "JavaScript,TypeScript,JSX,Vuejs Component,PHP,C#"
      # These languages are excluded from the cloc total.
      IGNORE_LANGS: "JSON,HTML,CSS,SCSS,Sass,Markdown,SVG,XML,YAML,TOML,CSV,Text,Properties"

    steps:
      - name: Checkout report repository
        uses: actions/checkout@v7
        with:
          persist-credentials: false

      - name: Install cloc and jq
        run: |
          sudo apt-get update
          sudo apt-get install -y cloc jq

      - name: Discover repositories accessible to the PAT
        env:
          GH_TOKEN: ${{ secrets.GH_PAT }}
        shell: bash
        run: |
          set -euo pipefail
          if [[ -z "$GH_TOKEN" ]]; then
            echo "Add a GH_PAT repository secret before running this workflow" >&2
            exit 1
          fi

          # The authenticated-user endpoint can include owned, collaborator,
          # public, and private repositories. Fetch every page, then skip forks.
          if ! gh api --paginate --slurp 'user/repos?per_page=100' > "$RUNNER_TEMP/pat-repos.json" 2>/dev/null; then
            echo "Could not list repositories. Check GH_PAT access." >&2
            exit 1
          fi
          jq -e 'type == "array" and all(.[]; type == "array")' "$RUNNER_TEMP/pat-repos.json" > /dev/null
          if ! jq -e 'any(.[][]; .private == true and .fork == false)' "$RUNNER_TEMP/pat-repos.json" > /dev/null; then
            echo "::warning::No private non-fork repositories were returned. Check the token's resource owner and repository access if you expected private repos."
          fi
          jq -r '.[][] | select(.fork == false) | .full_name' \
            "$RUNNER_TEMP/pat-repos.json" > "$RUNNER_TEMP/pat-source-repos.txt"

      - name: Clone default branches
        env:
          GH_TOKEN: ${{ secrets.GH_PAT }}
        shell: bash
        run: |
          set -euo pipefail
          gh auth setup-git --hostname github.com --force

          mapfile -t repos < "$RUNNER_TEMP/pat-source-repos.txt"
          clone_root="$RUNNER_TEMP/pat-code-repos"
          mkdir -p "$clone_root"

          for index in "${!repos[@]}"; do
            repo="${repos[$index]}"
            if ! git clone --quiet --depth 1 --single-branch \
              "https://github.com/$repo.git" "$clone_root/$index" > /dev/null 2>&1; then
              echo "A repository could not be cloned. Check GH_PAT access." >&2
              exit 1
            fi
          done

      - name: Count code lines
        shell: bash
        run: |
          set -euo pipefail
          mkdir -p output
          if [[ -s "$RUNNER_TEMP/pat-source-repos.txt" ]]; then
            if ! cloc "$RUNNER_TEMP/pat-code-repos" --json \
              --report-file=output/cloc-output.json \
              --exclude-dir=.git --exclude-lang="$IGNORE_LANGS" > /dev/null 2>&1; then
              echo "Could not count code lines; no stats were published." >&2
              exit 1
            fi
          else
            jq -n '{SUM: {code: 0}}' > output/cloc-output.json
          fi
          jq 'del(.header)' output/cloc-output.json > "$RUNNER_TEMP/pat-cloc-summary.json"
          mv "$RUNNER_TEMP/pat-cloc-summary.json" output/cloc-output.json

      - name: Update PAT section in README
        shell: python
        run: |
          import json
          import os
          import re
          from pathlib import Path

          report = json.loads(Path("output/cloc-output.json").read_text())
          total = report.get("SUM", {}).get("code")
          if not isinstance(total, int) or total < 0:
              raise ValueError("cloc did not produce a valid code total")

          languages = []
          for name, counts in report.items():
              if name in {"header", "SUM"}:
                  continue
              code = counts.get("code")
              if not isinstance(code, int) or code < 0:
                  raise ValueError(f"Invalid cloc count for {name}")
              if code:
                  languages.append((name, code))
          if sum(code for _, code in languages) != total:
              raise ValueError("Language counts do not match the cloc total")

          highlights = {name.strip() for name in os.environ["HIGHLIGHT_LANGS"].split(",")}
          named = sorted(
              ((name, code) for name, code in languages if name in highlights),
              key=lambda item: (-item[1], item[0]),
          )
          other = sum(code for name, code in languages if name not in highlights)
          width = max([12, *(len(name) for name, _ in named)])
          fence = chr(96) * 3
          rows = [fence, "[ LANGUAGES BREAKDOWN ]", ""]
          rows += [f"{name:<{width}} --> {code:,} lines" for name, code in named]
          if other:
              rows.append(f"{'Others':<{width}} --> {other:,} lines")
          rows += ["", f"[ TOTAL LINES OF CODE: {total:,} ]", fence]

          start = re.compile(r"^<!-- LANGUAGES BREAKDOWN START -->$", re.MULTILINE)
          end = re.compile(r"^<!-- LANGUAGES BREAKDOWN END -->$", re.MULTILINE)
          readme = Path("README.md")
          content = readme.read_text()
          starts = list(start.finditer(content))
          ends = list(end.finditer(content))
          if len(starts) != 1 or len(ends) != 1 or starts[0].end() >= ends[0].start():
              raise ValueError("README must contain exactly one ordered PAT stats marker pair")
          updated = (
              content[: starts[0].end()]
              + "\n"
              + "\n".join(rows)
              + "\n"
              + content[ends[0].start() :]
          )
          if updated != content:
              readme.write_text(updated)

      - name: Commit changed PAT stats
        env:
          GH_TOKEN: ${{ github.token }}
        shell: bash
        run: |
          set -euo pipefail
          git add README.md output/cloc-output.json
          if git diff --cached --quiet; then
            echo "PAT code statistics are unchanged"
            exit 0
          fi
          git config user.name "github-actions[bot]"
          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
          git commit -m "chore: update PAT repository code stats"
          gh auth setup-git --hostname github.com --force
          git push origin "HEAD:$GITHUB_REF"