ci.yml

# Runs by itself on every push and pull request.
# Never calls the Instagram API, so it needs no token and costs nothing.
name: Tests (automatic, no API calls, free)

on:
  push:
  pull_request:
  workflow_dispatch:  # a Run workflow button, to re-run the tests without a commit

permissions:
  contents: read

jobs:
  tests:
    name: Run the tests and the code style check
    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1  # v7.0.1
      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97  # v7.0.0
        with:
          python-version: "3.12"
      - name: Install the package and its test tools
        run: pip install -e ".[dev]"
      - name: Check code style with ruff
        run: ruff check .
      - name: Run the tests
        run: pytest
      - name: README headings are well formed
        # A section rewrite once produced "## The data## The data" and shipped;
        # a heading with a second run of hashes mid-line is never intended.
        run: |
          if grep -nE '^#+ .*#{2,} ' README.md; then
            echo "::error::malformed heading in README.md"; exit 1
          fi

  # These install steps shipped broken once, because they were only ever run
  # from a working copy that already had the package. This job follows the
  # README on an empty machine and stops where the tool asks for a token, which
  # proves the clone, the install and the command without spending money.
  coldstart:
    name: Follow the README from scratch on an empty machine
    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1  # v7.0.1
      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97  # v7.0.0
        with:
          python-version: "3.12"
      - name: Copy the install steps out of the README and run them
        run: |
          set -o pipefail
          python - <<'PY' > "$RUNNER_TEMP/quickstart.sh"
          import os
          import pathlib
          import re
          import sys

          readme = pathlib.Path("README.md").read_text(encoding="utf-8")

          def section(title):
              return readme.split(f"## {title}", 1)[1].split("\n## ", 1)[0]

          def block(text, lang):
              return re.search(r"```" + lang + r"\n(.*?)```", text, re.S).group(1)

          quick = section("Quickstart")
          install, snippet = block(quick, "bash"), block(quick, "python")
          command = block(section("Or run it as a command"), "bash")
          for required, where in (("pip install brightdata-sdk", install),
                                  ("SyncBrightDataClient", snippet),
                                  ("linkedin-scraper", command)):
              if required not in where:
                  sys.exit(f"the README no longer contains {required!r} where the cold start expects it")

          steps = ["# --- Quickstart: the SDK path ---"]
          steps += [l for l in install.splitlines() if l.startswith("pip install")]
          pathlib.Path(os.environ["RUNNER_TEMP"], "quickstart_snippet.py").write_text(snippet, encoding="utf-8")
          steps += [
              "set +e",
              "python quickstart_snippet.py > out.txt 2> err.txt",
              "code=$?",
              "set -e",
              'test "$code" != "0" || { echo "the snippet ran without a token?"; cat out.txt; exit 1; }',
              'grep -qi token err.txt || { echo "no token message"; cat err.txt; exit 1; }',
              'echo "Quickstart: installs, and stops at the token, as documented."',
              "# --- Or run it as a command: the CLI path ---",
          ]
          for line in command.splitlines():
              if line.startswith("linkedin-scraper"):
                  continue  # run below, so the exit code can be checked
              if line.startswith("pip install"):
                  # The repository is public, so this installs the exact line a
                  # reader pastes. On a private fork the URL is unreachable, and
                  # the check falls back to the checkout on its own.
                  url = line.split("git+", 1)[1].strip() if "git+" in line else ""
                  line = "\n".join([
                      f'repo="{url}"',
                      'code=$(curl -s -o /dev/null -w "%{http_code}" "$repo" || echo 000)',
                      'if [ "$code" = "200" ]; then',
                      f'  echo "$repo is public: installing the exact README line"',
                      f'  {line}',
                      "else",
                      '  echo "NOTE: $repo returns HTTP $code, so it is not public yet."',
                      '  echo "Installing from this checkout instead."',
                      '  pip install "$GITHUB_WORKSPACE"',
                      "fi",
                  ])
              steps.append(line)
          steps += [
              "set +e",
              "linkedin-scraper satyanadella > out.txt 2> err.txt",
              "code=$?",
              "set -e",
              'test "$code" = "2" || { echo "expected exit 2, got $code"; cat err.txt; exit 1; }',
              'grep -qi token err.txt || { echo "no token message"; cat err.txt; exit 1; }',
              'echo "Success: a new user can follow both README paths from scratch."',
          ]
          print("\n".join(steps))
          PY
          echo "--- these commands came out of README.md ---"
          cat "$RUNNER_TEMP/quickstart.sh"
          echo "--- running them on this empty machine ---"
          cd "$RUNNER_TEMP" && bash -e "$RUNNER_TEMP/quickstart.sh"