1. Get a Google Cloud account

   • Go to cloud.google.com and sign in with your Google account.

   • If you’ve never used Google Cloud before, enable billing (you can set a monthly budget to control spend).

   • The free tier gives you 1 TB of BigQuery queries per month — plenty for this script if you run it weekly.

2. Open BigQuery

   • Navigate to console.cloud.google.com/bigquery.

   • At the top, select or create a new project (e.g., software-house-trends).

3. Create a dataset for your results

   • In the left panel, click your project name → Create dataset.

   • Name it gh_trending.

   • Set Location to US (GitHub Archive is stored in the US).

   • Click Create dataset.

4. Open the SQL editor

   • Click Compose new query.

   • Copy and paste the provided SQL script into the editor.

5. Set a safe cost limit

   • In the query editor, click Query settings → Advanced options.

   • Set Maximum bytes billed to something like 10 GB.

   • This prevents accidental large scans.

6. Run the query

   • Click Run.

   • Wait a few seconds — the results table will appear below the editor.

7. Save or export the results

   • Click Save Results in the top right of the results table.

   • Choose CSV, JSON, or Google Sheets if you want to share or analyze it further.

8. (Optional) Save as a view for reuse

   • Above the query, click Save → Save View.

   • Store it in the gh_trending dataset so you can run it again without copying SQL each time.

9. (Optional) Schedule to run automatically

   • With the query open, click Schedule.

   • Choose how often to refresh (daily, weekly).

   • Set the destination table in gh_trending so it overwrites with fresh data each run.

   • This gives you an always-up-to-date trending list.

10. Review regularly

• Go to the dataset → click the saved view or table → view results.

• You can download or share with your team anytime.

-- Top repos by UNIQUE STARRERS in the last 3 months,
-- plus unique issue openers and total issues opened (same window).

DECLARE m0 STRING DEFAULT FORMAT_DATE('%Y%m', DATE_TRUNC(CURRENT_DATE(), MONTH));
DECLARE m1 STRING DEFAULT FORMAT_DATE('%Y%m', DATE_TRUNC(DATE_SUB(CURRENT_DATE(), INTERVAL 1 MONTH), MONTH));
DECLARE m2 STRING DEFAULT FORMAT_DATE('%Y%m', DATE_TRUNC(DATE_SUB(CURRENT_DATE(), INTERVAL 2 MONTH), MONTH));

-- Keep the base CTE column-pruned to help BigQuery read fewer bytes.
WITH base AS (
  SELECT
    repo.name                      AS repo_name,
    actor.login                    AS actor_login,
    type,
    JSON_VALUE(payload, '$.action') AS action,
    created_at
  FROM `githubarchive.month.*`
  WHERE _TABLE_SUFFIX IN UNNEST([m0, m1, m2])
    AND created_at >= TIMESTAMP(DATETIME_SUB(CURRENT_DATETIME(), INTERVAL 3 MONTH))
    AND created_at <  CURRENT_TIMESTAMP()
),
stars AS (
  SELECT
    repo_name,
    COUNT(DISTINCT actor_login) AS unique_starrers
  FROM base
  WHERE type = 'WatchEvent'
    AND action = 'started'
    AND actor_login IS NOT NULL
    AND NOT ENDS_WITH(actor_login, '[bot]')
  GROUP BY repo_name
),
issues AS (
  SELECT
    repo_name,
    COUNT(DISTINCT actor_login) AS unique_issue_contributors,
    COUNT(*)                    AS issues_opened
  FROM base
  WHERE type = 'IssuesEvent'
    AND action = 'opened'
    AND actor_login IS NOT NULL
    AND NOT ENDS_WITH(actor_login, '[bot]')
  GROUP BY repo_name
)
SELECT
  COALESCE(stars.repo_name, issues.repo_name) AS repo_name,
  IFNULL(unique_starrers, 0)            AS unique_starrers,
  IFNULL(unique_issue_contributors, 0)  AS unique_issue_contributors,
  IFNULL(issues_opened, 0)              AS issues_opened
FROM stars
FULL JOIN issues USING (repo_name)
ORDER BY unique_starrers DESC, unique_issue_contributors DESC
LIMIT 1000;