One RAG conversation with its ordered messages

Materialises a single retrieval-augmented chat conversation: the rag_conversations row plus its rag_messages, ordered by sort_index. The sources citations on assistant messages are parsed from JSON into a list-column (each element a data.frame of {chunk_id, text, score} rows); user messages carry NA. Timestamps become POSIXct.

Usage

entropia_conversation(con, id)

Arguments

con

A connection returned by entropia_connect().

id

A single conversation id (a value of rag_conversations.id).

Value

A list of class entropia_conversation.

Details

Returns a list of class entropia_conversation with two elements:

• conversation: a one-row tibble (id, title, created_at, updated_at).

• messages: a tibble with id, sort_index, role, content, sources, model and created_at, ordered by sort_index.

Examples

con <- entropia_connect(system.file("extdata", "entropia-example.sqlite",
  package = "entropiaR"
))
entropia_conversation(con, "bbbbbbbb-bbbb-4bbb-8bbb-bbbbbbbbbbb1")
#> EntropIA conversation: Consulta sobre la huelga
#>   created: 2026-01-15 12:12:00 UTC
#>   2 message(s):
#>   [0] user: ¿Que paso en la huelga?
#>   [1] assistant: Hubo una huelga general en 1920.
entropia_disconnect(con)