Changelog

Source: NEWS.md

seatabler 0.2.2

• Declare the minimum nat.python version in DESCRIPTION (nat.python (>= 0.2.0)). Since 0.2.0 seatabler has called nat.python::check_module() / simple_python(), which only exist in nat.python 0.2.0; an older nat.python installed cleanly and then failed at call time (#8).

seatabler 0.2.1

• Restored the schema-aware result coercion that was dropped when flytable_query became seatable_query: nat.python::pandas2df() handles only the generic pandas -> R conversion (and leaves multiple-select cells as list-columns, having no SeaTable schema), so flytable2df() / flytable_fix_coltypes() / flytable_parse_date() are ported over as internal st_coerce_df() and applied on every read path. This collapses multiple-select columns to character vectors and restores SeaTable column types (numbers, checkboxes, dates).
• seatable_query() and seatable_list_rows() gain a collapse_lists argument (default TRUE, comma-joining; FALSE keeps list-columns; a string sets the separator); it also passes through seatable_list_selected().
• Restored the memoisation of seatable_columns() (schema cached for an hour, invalidated after schema mutations; cached = FALSE forces a refresh), which had been dropped, so paginated reads no longer refetch the schema per page. It now also carries the per-column data metadata, so date columns parse with their declared format and seatable_select_options() reads option names from the cached schema rather than a fresh get_metadata() call.

seatabler 0.2.0

First tagged release. A server-agnostic R client for SeaTable, factored out of the seatable code embedded in fafbseg (and independently in bancr).

Connections

• First-class seatable_connection() objects carrying URL, token environment variable name (never the token value) and workspace, with default_connection() / set_default_connection() / with_connection() for a resolvable session default. Wrapper packages bind one connection so end users never pass URL/token/workspace triples.

Core functionality

• Authentication by API token: seatable_module(), seatable_login(), seatable_generate_token().
• Base/workspace/table resolution: seatable_base(), seatable_workspaces(), seatable_workspace_id(), seatable_alltables().
• Querying and reading: seatable_query() (paginated), seatable_list_rows(), seatable_list_selected() (generic, no ID coercion).
• Writing: seatable_update_rows(), seatable_append_rows(), seatable_delete_rows(), with big-data batching.
• Columns and schema: seatable_columns(), seatable_add_column(s)(), seatable_delete_column(), seatable_select_options() / seatable_add_select_options() for multi-select columns.
• Big-data archive/unarchive: seatable_archive_rows(), seatable_unarchive_rows().
• Snapshot listing: seatable_snapshots().
• REST/JWT transport (seatable_rest(), seatable_base_rest()) behind the big-data and snapshot helpers, using httr2 against the api-gateway.

Python

• Python environment management and pandas→R conversion are delegated to nat.python (check_module() / simple_python() / pandas2df()) rather than fafbseg, avoiding a dependency cycle. Provision in one step with nat.python::simple_python("minimal", pkgs = "seatable_api").