letarette.sql

letarette.sql - SQL-based Letarette Document Manager

This is an all-SQL Document Manager for Letarette. To connect a SQL-based primary document storage to Letarette, only two queries need to be supplied. Check the example to see how basic they can be, given that the primary storage has a similar structure.

The following SQL drivers are supported:

• PostgreSQL
• MySQL
• SQLite
• MS SQL Server

Getting started

If you just want to try it out, there is a docker-compose config that launches a complete little Letarette system, fed by the database and queries from the example project.

Simply run docker-compose up to start the system.

When the system is up, you can run queries (and other commands) by running the lrcli inside the "letarette" container:

$ docker-compose exec letarette ./lrcli search -i docs
search>carrots -celery
Query executed in 0.000564746 seconds with status "found in index"
Returning 2 of 2 total hits, capped: false

[135] …and a carrot; boil until soft. When done, take out the…
[303] …pot, 1 carrot, 1 onion, thyme, bay leaf, salt and pepper, 2 cloves…

Basic setup

The letarette.sql service, lrsql, needs to know how to connect to the SQL database, and where to find the queries that provide documents to the Letarette indexer.

To make the service connect to a PostgreSQL source, and use default values for the query file locations (indexrequest.sql and documentrequest.sql):

$ export LRSQL_DB_DRIVER="postgres"
$ export LRSQL_DB_CONNECTION="postgres://user:password@localhost/testdb?sslmode=verify-full"
$ ./lrsql

Running lrsql with any command-line argument will print out available settings and their default values.

The two queries

The Letarette indexer update cycle has two separate steps, first it fetches an "interest list" of documents that are newer than the current index position (index request), and then it fetches the documents on that list (document request).

Timestamps are UTC - referenced UNIX epoch nanoseconds.

Index requests

The current index position is a combination of a document ID and the updated timestamp of that document. To handle the situation where the index position document has been updated since it was last fetched, index request queries need to follow a strict document ordering. This is best handled by sorting primarily on update timestamp and secondarily on document ID and handling the case where the update timestamp is unchanged separately. See indexrequest.sql from the example project.

The index request query gets three bound parameters: afterDocument (string), fromTimeNanos (int64) and documentLimit (uint16) and should return rows of two columns: id (string) and updatedNanos (int64). The bounds parameters are referred to by prefixing them with a colon:

where :afterDocument > 2

Document requests

Implementing the document request is even easier, since this only needs to retrieve all documents for a list of document IDs: documentrequest.sql. The single bound parameter wantedIDs will be replaced with a list of document IDs (string).

The document request query should return rows of id (string), updatedNanos (int64), title (string), txt (string) and alive (bool).

Building

letarette.sql uses build tags to control which drivers are built into the service binary. The build tags have the same names as the drivers they are enabling.

The current list of driver build tags is:

• postgres
• mysql
• sqlite3
• sqlserver

For example, to build a binary with "postgres" and "mysql" support:

$ go build -tags "postgres,mysql"

Service configuration

The letarette.sql service is configured by environment variables.