How to export seed from Emdash instance deployed on Cloudflare?

[segmentationfaulter]

Is it possible to export seed file from instance of Emdash deployed at Cloudflare using D1 as database? If yes, how?

[marcusbellamyshaw-cell]

emdash export-seed exists (npx emdash export-seed [--with-content], see the CLI reference) but its DB connector only supports file: SQLite URLs — the libsql: scheme throws "not yet implemented" and D1 isn't supported at all in the CLI. So there's no direct path from a live Cloudflare D1 deployment straight into seed.json today.

Two workarounds:

1. wrangler d1 export --remote <db-name> — gives you a raw SQL dump (not emdash's seed format, but it's your data).
2. Pull a local copy of the D1 data into a local sqlite file, then run emdash export-seed against that.

Might be worth filing a feature request for direct D1 support in export-seed.

[segmentationfaulter]

I tried exporting my d1 database and got this error:

✘ [ERROR] D1 Export error: cannot export databases with Virtual Tables (fts5)

Why do I see this error and how to get around it?

[marcusbellamyshaw-cell]

@segmentationfaulter that error is a known upstream Cloudflare limitation, not an emdash bug: cloudflare/workers-sdk#6305 and #9519 (both still open). wrangler d1 export --remote refuses to export the whole database if it contains any fts5 virtual table. Root cause: SQLite's native .dump of an fts5 table needs PRAGMA writable_schema=ON to write the shadow-table rows directly into sqlite_schema, and D1 doesn't support that pragma — so it bails on the entire export rather than skipping just the fts5 table. The --table filter doesn't help either; that check runs server-side against the whole DB before any table filtering is applied.

Good news: emdash's fts5 tables are fully derived and safe to drop from the export. Each searchable collection gets a virtual table named _emdash_fts_<slug> (plus SQLite's auto-generated shadow tables _emdash_fts_<slug>_data, _idx, _docsize, _config, _content), all built from the real content table ec_<slug>. On next deploy/first search request, FTSManager.verifyAndRepairAll() detects the missing FTS table and rebuilds it automatically — no manual reindex step needed.

Workaround — export everything except the fts5 tables:

# 1. list real tables, excluding fts5 + its shadow tables
npx wrangler d1 execute <db-name> --remote --json \
  --command "SELECT name FROM sqlite_master WHERE type='table' AND name NOT LIKE '_emdash_fts_%'"

# 2. export only those tables (repeat --table per name from step 1)
npx wrangler d1 export <db-name> --remote --output=dump.sql \
  --table=_emdash_collections --table=ec_posts --table=ec_pages ...

Load dump.sql into a local sqlite file, then run emdash export-seed against that local file (per the workaround above) to get your seed.json.