Aranami (荒波, “rough waves”) keeps Wikipedia project reports up to date and helps editors find articles to improve. Its main focus is WikiProject Video games on Chinese Wikipedia.
The dashboard brings together main-page highlights, upcoming quality content, popular articles, maintenance worklists, and report update results. It supports English, Simplified Chinese, and Traditional Chinese. The About page introduces the tool to general readers.
These are UTC trigger times, not promises that an edit will be necessary. A successful check can finish with no update needed.
| Report | Frequency and UTC trigger | Result |
|---|---|---|
| New-article recommendations (DYK) | Hourly, HH:00 |
Completed DYK entries, current nominees, assessments, and cumulative statistics. |
| New pages | Daily, 00:07 |
Keyword-selected new pages and pages rewritten from redirects, with assessments to help editors review them. |
| Article assessments | Hourly, HH:31 |
Six assessment and review lists: Bplus, BPAN, A, AL, ACC, and PPR. |
| Pageview rankings | Daily cycle from 18:00; normal first attempt 18:59, retries at HH:59 |
Daily, weekly, monthly, seasonal, yearly, and monthly task-force rankings. |
| English Wikipedia comparisons | Daily, 23:24 |
Important and high-quality English articles, their Chinese counterparts, and relevant changes. |
| Database reports | Daily, 00:00 |
Delegated refresh requests for subscribed PexBot reports within the configured project scope. |
New pages skips further automatic work for the current UTC day after the complete retained window through yesterday is confirmed. Missing retained dates are filled when needed.
Pageviews uses an 18:00 UTC reporting boundary, not midnight. It fetches available history once and saves each missing daily report in order through the latest eligible complete day. It retries incomplete or unavailable data hourly and waits for the next cycle once current. Confirmed completion survives server restarts.
English comparisons and database reports refresh current state on their daily schedules. Database reports already edited by PexBot that UTC day are not requested again. Manual runs revalidate destinations rather than bypassing checks.
The website's five datasets refresh independently, by default 24 hours after each successful collection. Failed refreshes retain the last valid data and retry hourly. Website visits are read-only.
One Python 3.15 backend hub fetches source data, parses source pages,
calculates changes, and manages stored results, caches, logs, and status.
The wiki updater formats prepared data and replaces only the configured
<!-- aranami ... --> ranges on selected destinations, preserving surrounding
text.
Complete report tables and metadata are retained before publication, including review targets and changes used in edit summaries. A separate read-only Flask website renders saved backend data as complete HTML pages, served by Waitress.
Logs are ordinary, readable English diagnostics showing what data was collected and which wiki pages were checked or edited. Structured status is stored separately: routine and per-page starts, finishes, durations, outcomes, and confirmed edit counts. The website displays these saved observations. The dashboard advances running times locally every second and synchronizes routine/page statuses every five seconds, keeping status popups in place. Execution times use whole seconds, with two decimal places for subsecond results. While a routine is running, its card shows the current execution and available progress rather than a next-execution row. The log viewer includes a compact server-state display, uptime, and source version.
Choose the guide for your role:
- Terminal user: run reports, inspect data, and create previews.
- System administrator / maintainer: upload source files to Toolforge, promote deployments, stop, start, restart, and recover.
- Application developer: local setup, contribution workflow, tests, and release checks.
- System-level developer: component boundaries, data contracts, publication safety, scheduling, and engineering rules.
See the documentation index and change history for the full reference.
Aranami runs only on Toolforge; the local repository is for development, testing, and preparing deployments. Runtime files stay outside the uploaded source tree. The administrator guide covers the runtime layout, validated deployment, and recovery workflow.