chenglou/pretext · error · Error

No corpora found

Error message

No corpora found

What it means

Thrown in init() after await loadSources() returns: if the corpus metadata array is empty, there is nothing to display and the page cannot proceed. Unlike the other corpus errors this one is fully expected to be caught — init wraps it in try/catch and calls setError(message), so it renders as an error status in the UI rather than crashing the script. It distinguishes 'sources fetch returned valid JSON but zero entries' from a fetch failure (which would throw earlier).

Source

Thrown at pages/corpus.ts:1129

slider.addEventListener('input', () => {
  setWidth(Number.parseInt(slider.value, 10))
})

select.addEventListener('change', () => {
  navigateToCorpus(select.value)
})

window.__CORPUS_REPORT__ = withRequestId({ status: 'error', message: 'Pending initial layout' })
stats.textContent = 'Loading...'
clearNavigationReport()
publishNavigationPhase('loading', requestId)

async function init(): Promise<void> {
  try {
    corpusList = await loadSources()
    if (corpusList.length === 0) {
      throw new Error('No corpora found')
    }

    const selected = corpusList.find(meta => meta.id === requestedCorpusId) ?? corpusList[0]!
    await loadCorpus(selected)
  } catch (error) {
    const message = error instanceof Error ? error.message : String(error)
    setError(message)
  }
}

void init()

View on GitHub (pinned to ac49b09b7d)

Solutions

1. Inspect the sources JSON that loadSources() fetches and confirm it contains the expected corpus entries.
2. Ensure the page is served from the correct working directory where the sources manifest is populated.
3. If you intentionally removed corpora, re-add at least one entry so the page has a default selection.
4. Check the network tab for the sources fetch response body to rule out a proxy truncating it to [].

Example fix

// before
const sourcesData = await fetch('sources.json').then(r => r.json())
return sourcesData as CorpusMeta[]

// after
const sourcesData = await fetch('sources.json').then(r => r.json())
if (!Array.isArray(sourcesData) || sourcesData.length === 0) {
  throw new Error('Sources manifest is missing or empty')
}
return sourcesData as CorpusMeta[]

// Validate the fetched sources before returning.
async function loadSources(): Promise<CorpusMeta[]> {
  const data = await fetch('/sources.json').then(r => r.json())
  if (!Array.isArray(data) || data.length === 0) {
    throw new Error('Sources manifest is missing or empty')
  }
  return data
}

function isNonEmptyCorpusList(value: unknown): value is CorpusMeta[] {
  return Array.isArray(value) && value.length > 0 &&
    value.every(item => typeof item === 'object' && item !== null && 'id' in item)
}

// init() already wraps this; the error is shown via setError.
try {
  corpusList = await loadSources()
  if (corpusList.length === 0) throw new Error('No corpora found')
  await loadCorpus(corpusList[0]!)
} catch (error) {
  setError(error instanceof Error ? error.message : String(error))
}

When it happens

Trigger: loadSources() successfully fetches and parses the sources JSON, but the parsed array is empty (length 0). This happens when the sources manifest is an empty array, or when a build step stripped all entries, or when a filter removed everything.

Common situations: The sources JSON was replaced with [] during a refactor; a CI build generated an empty manifest; the dev server is serving a stale/empty manifest from a different branch; a conditional inclusion removed all corpora for the current build configuration.

Related errors

• Invalid widths parameter: ${raw}
• Failed to measure ${currentMeta?.id ?? 'corpus'} @ ${width}
• No bundled text import for corpus ${meta.id}
• Invalid ACCURACY_CHECK_PORT: ${requestedPortRaw}
• Invalid value for --${name}: ${raw}

AI-assisted analysis of chenglou/pretext@ac49b09b7d (2026-08-12). Data as JSON: /api/errors/59ef1505c9b699b9. Report an issue: GitHub.