Update or append rows in a SeaTable table

seatable_update_rows updates existing rows (matched by their _id/row_id), returning TRUE on success. seatable_append_rows adds new rows, optionally straight into big data storage.

Usage

seatable_update_rows(
  df,
  table,
  base = NULL,
  con = default_connection(),
  append_allowed = TRUE,
  chunksize = 1000L,
  multi_select_cols = NULL,
  allow_new_options = FALSE
)

seatable_append_rows(
  df,
  table,
  base = NULL,
  con = default_connection(),
  bigdata = FALSE,
  chunksize = 1000L,
  multi_select_cols = NULL,
  allow_new_options = FALSE
)

Arguments

df

A data.frame of rows to write. For updates it must carry an _id or row_id column identifying each remote row.

table

Name of the table.

base

Optional base name or Base object; discovered from table when NULL.

con

A seatable_connection.

append_allowed

For seatable_update_rows, whether rows whose id is missing may be appended as new rows rather than raising an error.

chunksize

Split large requests into calls of at most this many rows.

multi_select_cols

Column names to treat as multiple-select (list-per-cell). The default NULL auto-detects them from column metadata.

allow_new_options

When a multiple-select value is not already an option for its column, whether to add it automatically (via seatable_add_select_options()) rather than erroring.

bigdata

For seatable_append_rows, whether to write straight into big data (archived) storage via REST rather than the ordinary grid.

Value

Logical, invisibly: TRUE on success. Failures normally abort with the SeaTable error message.

Details

SeaTable keeps a unique _id for every row, returned by seatable_query(). To update, keep that column (named _id or row_id) in df; to append you do not need it (and are warned if you supply one). Writes are chunked at chunksize rows because the API caps a single call at 1000.

Multiple-select columns need care: SeaTable expects a genuine list of option names per cell, and silently invents a new option from any bare string it is handed instead. Any column SeaTable reports as "multiple-select" is auto-detected and routed through a list-per-cell path; pass multi_select_cols to override detection. A scalar cell ("AB", or comma-joined "AB,CD") is accepted as shorthand and split on commas — symmetric with how such a cell reads back — or supply a list-column directly (I(list(c("AB", "CD")))), which also lets you write an option name that itself contains a comma. Every resulting value is checked against the column's existing vocabulary and rejected unless it is already an option; see allow_new_options and seatable_add_select_options().

Big data (bigdata = TRUE, append only): rows go through the add-archived-rows REST endpoint straight into archived storage, rather than the ordinary grid via the SDK. This is for bulk loads that would exceed the base's grid row cap; the base must have big data enabled or the gateway returns an error. Big data is an append destination, not a property of the table (a table can hold both grid and archived rows) — see seatable_archive_rows().

See also

seatable_query() to read rows back, seatable_archive_rows() to move rows into and out of big data storage.

Examples

if (FALSE) { # \dontrun{
rows <- seatable_query("SELECT _id, name FROM neurons LIMIT 2", con = con)
rows$name <- toupper(rows$name)
seatable_update_rows(rows, "neurons", con = con)

seatable_append_rows(data.frame(name = "new cell"), "neurons", con = con)
} # }