dm_zoom_to {dm}R Documentation

Mark table for manipulation


Zooming to a table of a dm allows for the use of many dplyr-verbs directly on this table, while retaining the context of the dm object.

dm_zoom_to() zooms to the given table.

dm_update_zoomed() overwrites the originally zoomed table with the manipulated table. The filter conditions for the zoomed table are added to the original filter conditions.

dm_insert_zoomed() adds a new table to the dm.

dm_discard_zoomed() discards the zoomed table and returns the dm as it was before zooming.

Please refer to vignette("tech-db-zoom", package = "dm") for a more detailed introduction.


dm_zoom_to(dm, table)

dm_insert_zoomed(dm, new_tbl_name = NULL, repair = "unique", quiet = FALSE)





A dm object.


A table in the dm.


Name of the new table.


Either a string or a function. If a string, it must be one of "check_unique", "minimal", "unique", or "universal". If a function, it is invoked with a vector of minimal names and must return minimal names, otherwise an error is thrown.

  • Minimal names are never NULL or NA. When an element doesn't have a name, its minimal name is an empty string.

  • Unique names are unique. A suffix is appended to duplicate names to make them unique.

  • Universal names are unique and syntactic, meaning that you can safely use the names as variables without causing a syntax error.

The "check_unique" option doesn't perform any name repair. Instead, an error is raised if the names don't suit the "unique" criteria.


By default, the user is informed of any renaming caused by repairing the names. This only concerns unique and universal repairing. Set quiet to TRUE to silence the messages.

Users can silence the name repair messages by setting the "rlib_name_repair_verbosity" global option to "quiet".


Whenever possible, the key relations of the original table are transferred to the resulting table when using dm_insert_zoomed() or dm_update_zoomed().

Functions from dplyr that are supported for a zoomed_dm: group_by(), summarise(), mutate(), transmute(), filter(), select(), rename() and ungroup(). You can use these functions just like you would with a normal table.

Calling filter() on a zoomed dm is different from calling dm_filter(): only with the latter, the filter expression is added to the list of table filters stored in the dm.

Furthermore, different join()-variants from dplyr are also supported, e.g. left_join() and semi_join(). (Support for nest_join() is planned.) The join-methods for zoomed_dm infer the columns to join by from the primary and foreign keys, and have an extra argument select that allows choosing the columns of the RHS table.

And – last but not least – also the tidyr-functions unite() and separate() are supported for zoomed_dm.


For dm_zoom_to(): A zoomed_dm object.

For dm_insert_zoomed(), dm_update_zoomed() and dm_discard_zoomed(): A dm object.


flights_zoomed <- dm_zoom_to(dm_nycflights13(), flights)


flights_zoomed_transformed <-
  flights_zoomed %>%
  mutate(am_pm_dep = ifelse(dep_time < 1200, "am", "pm")) %>%
  # `by`-argument of `left_join()` can be explicitly given
  # otherwise the key-relation is used
  left_join(airports) %>%
  select(year:dep_time, am_pm_dep, everything())


# replace table `flights` with the zoomed table
flights_zoomed_transformed %>%

# insert the zoomed table as a new table
flights_zoomed_transformed %>%
  dm_insert_zoomed("extended_flights") %>%

# discard the zoomed table
flights_zoomed_transformed %>%

[Package dm version 0.2.8 Index]