blockr.dm

Relational data: bring in a whole data model (many linked tables) as one source, pull tables out of it, filter across the relationships. For data that lives in a database rather than a single frame.

Example data model

new_dm_example_block() · input

Load a pre-built dm from example datasets

Argument	Description
dataset	Character. ID of the dm example dataset to load. One of: "bi_star_schema" (retail star schema), "safetydata_adam" / "pharmaverseadam" / "bms_adam" (CDISC ADaM clinical dm with adsl, adae, ...), "pharmaversesdtm" (raw ...

Import multiple tables

new_dm_read_block() · input

Read a whole directory, ZIP archive or Excel workbook at once: one table per file or sheet, linked as a dm

Argument	Description
path	Character. Path to a file or directory to read. Supported: .xlsx / .xls (one sheet per table), .zip of per-table files, or a directory containing per-table CSV / Parquet / Feather files.
selected_tables	Optional character vector restricting which tables to load. null or unset loads every table found at path.
args	Named list of reader options applied uniformly to every member of the container; each member's format picks the parameters it understands (e.g. list(sep = ";") reads every CSV in the folder as semicolon-separated an...

Add keys

new_dm_add_keys_block() · structured

Add primary and foreign key relationships to dm

Argument	Description
pk_table	Character. Name of the table to receive the primary key.
pk_column	Character. Column name on pk_table to mark as the primary key.
fk_tables	Character vector of table names that should get a foreign key pointing to pk_table.
fk_column	Character. Column name on each fk_tables entry to mark as the foreign key. Must match pk_column in meaning, and usually in name.

CDISC dm

new_cdisc_dm_block() · structured

Set CDISC keys (USUBJID PK/FK) and optionally deduplicate subject columns

Argument	Description
set_keys	Logical. TRUE (default) sets USUBJID as primary key on ADSL and foreign key on all other tables. Required for any cascading dm filter to work.
dedup_cols	Logical. TRUE removes duplicated subject-level columns (AGE, SEX, ARM, etc.) from non-ADSL tables so they live only on ADSL, which avoids column-name collisions in downstream joins. FALSE (the default) leaves every ta...

Create dm

new_dm_block() · structured

Combine data frames into a dm (data model) object

Argument	Description
infer_keys	Logical. TRUE (default) auto-infers primary and foreign key relationships from matching column names. FALSE leaves the dm key-less (use dm_add_keys_block to set keys explicitly).

Crossfilter

new_crossfilter_block() · structured

Client-side crossfilter for data frames and dm objects using crossfilter2.js

Argument	Description
active_dims	Per-table active filter columns. Object: table name -> array of column names that are eligible for filtering. For a single data frame input (not a dm), use ".tbl" as the table name.
filters	Per-table categorical filters. Object: table name -> {column -> array of kept values}. Values not listed are excluded.
subgroup	A second column the board is split by, under pinned: tables nest it under the group, charts facet by it. Same eligible columns as pinned, minus the pinned one. Null for none.
groups	Group definitions for the pinned column and the subgroup. Object: column name -> {columns: array of {name, members: array of levels, custom: boolean}, in table order}. A column with an empty name is one level on its o...
range_filters	Per-table numeric/date range filters. Object: table name -> {column -> [min, max]}.
measure	Aggregation measure as a "table.column" string, or null for row counts. Used in linked summary views.
agg_func	Aggregation function applied to measure. One of "sum" or "mean". Only used when measure is set.
featured	Columns worth showing up front: they appear as one-click chips above the filter cards and supply the choices for pinned. Array of column names, resolved to whichever table carries them.
pinned	Column kept in an always-open card at the top of the block, picked through a select in that card's header. Must be one of featured and live on the parent table. Null for no pinned card.

dm Temporal join

new_dm_temporal_join_block() · structured

Temporal join between two dm tables, filtering by time window

Argument	Description
left_table	Character. Name of the "anchor" table in the dm.
left_date	Character. Date/datetime column on left_table that defines the anchor time (e.g. treatment start).
right_table	Character. Name of the "event" table in the dm being joined in.
right_date	Character. Date/datetime column on right_table whose offset from left_date is checked against the window.
window_days	Integer. Size of the time window in days (default 7).
direction	Window direction relative to left_date. One of "after" (default, keep events within window_days AFTER anchor), "before", or "around" (two-sided window).

Drill filter

new_drill_filter_block() · structured

The value filter a drill-down click writes into. One per board, beside blockr.viz's control bridge extension; senders find it by class and it resolves a claimed column to a dm table itself.

Filter dm

new_dm_filter_block() · structured

Filter dm by condition in any table, cascading to related tables via FKs

Argument	Description
table	Character. Name of the table within the dm where conditions are applied. Rows removed from this table cascade to related tables via the dm's FK relationships.
state	List with conditions and operator. Matches the state format of blockr.dplyr::new_filter_block - condition types values, numeric, and expr, combined with & or |.

Flatten dm

new_dm_flatten_block() · structured

Flatten dm into a single data frame by joining

Argument	Description
start_table	Character. Table to flatten from — becomes the left side of every join. Output row count equals this table's row count (for left joins).
include_tables	Character vector. Tables to join into start_table. Empty vector flattens the full dm by following all related tables.
join_type	Join type to use throughout. One of "left" (default), "inner", "full", "right".
recursive	Logical. TRUE (default) follows multi-hop relationships through the FK graph. FALSE joins only direct neighbors of start_table.

Pull table

new_dm_pull_block() · structured

Extract a single table from dm as a data frame

Argument	Description
table	Character. Name of the table to extract as a data frame.

Select tables

new_dm_select_block() · structured

Select a subset of tables to keep in a dm

Argument	Description
tables	Character vector of table names to keep in the dm. Tables not listed are dropped. Relationships to dropped tables are removed.

Temporal join

new_temporal_join_block() · structured

Join two tables and filter by time window between date columns

Argument	Description
by	Character vector of column names to match on as the join key (e.g. subject ID). Rows with matching key AND date-within-window are joined.
left_date	Character. Date column on the LEFT (x) table that defines the anchor time.
right_date	Character. Date column on the RIGHT (y) table whose offset from left_date is checked against the window.
window_days	Integer. Window size in days. Default 7.
direction	Window direction relative to left_date. One of "after" (default), "before", "around".

Value filter

new_value_filter_block() · structured

Minimal value filter for a data frame or dm. Columns hidden behind the gear; per-column single vs multi select. dm input cascades via FKs.

Write dm

new_dm_write_block() · output

Write dm object to Excel, ZIP, or directory

Argument	Description
directory	Character. Output directory path.
filename	Character. Base filename (no extension). Extension is derived from format. For Excel, one file; for CSV / Parquet, one file per table.
format	Output format. One of "excel", "csv", or "parquet". Default "excel".
auto_write	Logical. TRUE writes on every upstream update; FALSE (default) writes only on explicit action.