Repository Charts

BookOrbit renders its README charts in the repository instead of depending on external chart or badge services.

Storage

The generated-charts branch contains the generated files:

star-history.json: validated daily aggregate star counts and source metadata
star-history.svg: the self-contained star history chart
translation-progress.svg: the self-contained Crowdin translation chart

The workflow creates normal commits on this branch. It does not force-push, so previous chart data remains recoverable without adding automated commits to main.

The bootstrap snapshot at scripts/star-history/bootstrap.json contains the initial exact aggregate history. It is used only when no published series exists. Raw stargazer records and GitHub usernames are never stored.

Refresh behavior

The Repo Charts workflow runs daily and can also be dispatched manually. It restores the published series, regenerates both charts, and publishes the changed files in one branch commit. Nothing is pushed when both charts are byte-identical to the published copies.

Each generator runs independently. If one fails, the other still publishes and the failed chart keeps its previously published copy, while the workflow run is marked as failed.

Star history

The star history generator obtains the current repository star count, replaces any existing point for the current UTC date, and renders the SVG.

The built-in workflow token is sufficient for daily snapshots. An optional STAR_HISTORY_TOKEN repository secret belonging to a collaborator enables exact-history refreshes from GitHub's timestamped stargazer endpoint. Exact refreshes replace the aggregate curve and correct any drift from previous snapshots. If the endpoint is unavailable, the workflow retains the stored history and falls back to the current count.

The generator refuses malformed series, data for a different repository, future points, invalid API responses, and exact histories beyond GitHub's pagination limit.

Translation progress

The translation generator reads public Crowdin project statistics and renders every language the project exposes, so no code change is needed when a language is added. English is pinned first at 100% as the source language and is excluded from the average, though it renders like any other completed row.

Every language is one horizontal meter row, so the card grows in height as languages are added rather than switching layout. The label column is measured from the widest language name and long names are truncated, which keeps the bar from being squeezed out. Bars carry three progress tiers, listed in a legend on the card: complete at 100%, in progress at 70% or above, and needs work below that. Three is the ceiling; a fourth tier would need an amber that sits at deuteranopia delta-E 1.4 against the orange, which no threshold tuning can separate.

Local commands

Run the focused test suites:

pnpm run test:charts

Generate the star history chart from a stored series and the current count:

GITHUB_TOKEN=... node scripts/star-history/generate.mjs \
  --repo bookorbit/bookorbit \
  --series star-history.json \
  --bootstrap scripts/star-history/bootstrap.json \
  --out star-history.svg

Set STAR_HISTORY_TOKEN as well to attempt a complete exact-history refresh.

Generate the translation chart from live Crowdin statistics:

node scripts/crowdin-stats/generate.mjs --out translation-progress.svg

3.4 KiB Raw Permalink Blame History