nacrecontextlayer/connector-sql repository overview

⁠Nacre connector — sql

Keeps the rows a SQL query returns in Nacre⁠ layers — Postgres, MySQL or MariaDB, chosen by the URL. A row is a document, one column is its external_id, a template over its columns picks the layer, and a row the query stopped returning is a document that leaves the index. A watermark column is the row's version where the table has one; otherwise the row's own hash is, computed from the result the query already returned, so a sweep never reads a row twice.

docker run -v sql-state:/state \
  -e NACRE_URL=https://nacre.example.com -e NACRE_TOKEN=... \
  -e SQL_URL=postgres://reader:[email protected]:5432/wiki \
  -e 'SQL_QUERY=SELECT id, kind, title, body, updated_at FROM pages WHERE published' \
  -e 'SQL_LAYER=${kind}' -e 'SQL_CONTENT=${body}' -e SQL_VERSION=updated_at \
  nacrecontextlayer/connector-sql:0.1.1

SQL_QUERY is the listing, every row streamed; SQL_ID, SQL_LAYER, SQL_TITLE and SQL_CONTENT are templates over the columns. SQL_VERSION names a watermark column and is the cheap path — a row edited without that column moving is not re-sent, which is the contract such a column already makes with everything that reads it. Rows are text only: a bytea makes the row a binary skip naming the column, because files live in object stores. The variables are in the connector's README⁠.

⁠Every connector

…same as git…

⁠The product

…same as git… Tags: {version}, amd64 and arm64. ghcr.io/nacre-work/connectors/sql is the canonical address; this repository is a mirror pushed by the same build at the same version.

⁠Links

nacre.work⁠ · All connectors⁠ · This connector's README⁠ · Nacre quickstart⁠ · Releases⁠