It can be useful to set up all the data sources you need and just draw from them when necessary. This upfront configuration with tbl_store() lets us define the methods for obtaining tabular data from mixed sources (e.g., database tables, tables generated from flat files, etc.) and provide names for these data preparation procedures. Then we have a convenient way to access the materialized tables with tbl_get(), or, the table-prep formulas with tbl_source(). Table-prep formulas can be as simple as getting a table from a location, or, it can involve as much mutation as is necessary (imagine procuring several mutated variations of the same source table, generating a table from multiple sources, or pre-filtering a database table according to the system time). Another nice aspect of organizing table-prep formulas in a single object is supplying it to the read_fn argument of create_agent() or create_informant() via $ notation (e.g, create_agent(read_fn = <tbl_store>$<name>)) or with tbl_source() (e.g., create_agent(read_fn = ~ tbl_source("<name>", <tbl_store>))).

tbl_store(..., .list = list2(...))

## Arguments

... Expressions that contain table-prep formulas and table names for data retrieval. Two-sided formulas (e.g,  ~ ) are to be used, where the left-hand side is a given name and the right-hand is the portion that is is used to obtain the table. Allows for the use of a list as an input alternative to ....

## Value

A tbl_store object that contains table-prep formulas.

## YAML

A pointblank table store can be written to YAML with yaml_write() and the resulting YAML can be used in several ways. The ideal scenario is to have pointblank agents and informants also in YAML form. This way the agent and informant can refer to the table store YAML (via tbl_source()), and, the processing of both agents and informants can be performed with yaml_agent_interrogate() and yaml_informant_incorporate(). With the following R code, a table store with two table-prep formulas is generated and written to YAML (if no filename is given then the YAML is written to "tbl_store.yml").

# R statement for generating the "tbl_store.yml" file
tbl_store(
tbl_duckdb ~ db_tbl(small_table, dbname = ":memory:", dbtype = "duckdb"),
sml_table_high ~ small_table %>% dplyr::filter(f == "high")
) %>%
yaml_write()

# YAML representation ("tbl_store.yml")
tbls:
tbl_duckdb: ~ db_tbl(small_table, dbname = ":memory:", dbtype = "duckdb")
sml_table_high: ~ small_table %>% dplyr::filter(f == "high")


This is useful when you want to get fresh pulls of prepared data from a source materialized in an R session (with the tbl_get() function. For example, the sml_table_high table can be obtained by using tbl_get("sml_table_high", "tbl_store.yml"). To get an agent to check this prepared data periodically, then the following example with tbl_source() will be useful:

# Generate agent that checks sml_table_high, write it to YAML
create_agent(
label = "An example that uses a table store.",
actions = action_levels(warn_at = 0.10)
) %>%
col_exists(vars(date, date_time)) %>%
write_yaml()

# YAML representation ("agent-sml_table_high.yml")
tbl_name: sml_table_high
label: An example that uses a table store.
actions:
warn_fraction: 0.1
locale: en
steps:
- col_exists:
columns: vars(date, date_time)


Now, whenever the sml_table_high table needs to be validated, it can be done with yaml_agent_interrogate() (e.g., yaml_agent_interrogate("agent-sml_table_high.yml")).

## Function ID

1-8

Other Planning and Prep: action_levels(), create_agent(), create_informant(), db_tbl(), draft_validation(), file_tbl(), scan_data(), tbl_get(), tbl_source(), validate_rmd()

## Examples

if (interactive()) {

# Define a tbl_store object by adding
# table-prep formulas inside the
# tbl_store() call
tbls <-
tbl_store(
small_table_duck ~ db_tbl(
table = small_table,
dbname = ":memory:",
dbtype = "duckdb"
),
~ db_tbl(
table = "rna",
dbname = "pfmegrnargs",
dbtype = "postgres",
host = "hh-pgsql-public.ebi.ac.uk",
port = 5432,
),
all_revenue ~ db_tbl(
table = file_tbl(
file = from_github(
file = "all_revenue_large.rds",
repo = "rich-iannone/intendo",
subdir = "data-large"
)
),
dbname = ":memory:",
dbtype = "duckdb"
),
sml_table ~ pointblank::small_table
)

# Once this object is available, you
# can check that the table of interest
# is produced to your specification with
# the tbl_get() function
tbl_get(
tbl = "small_table_duck",
store = tbls
)

# Another simpler way to get the same
# table materialized is by using $ to # get the entry of choice for tbl_get() tbls$small_table_duck %>% tbl_get()

# Creating an agent is easy when all
# table-prep formulas are encapsulated
# in a tbl_store object; use $ # notation to pass the appropriate # procedure for reading a table to the # read_fn argument agent_1 <- create_agent( read_fn = tbls$small_table_duck
)

# There are other ways to use the
# table store to assign a target table
# to an agent, like using the
# tbl_source() function
agent_2 <-
create_agent(
tbl = "small_table_duck",
store = tbls
)
)

# The table store can be moved to
# YAML with yaml_write and the
# tbl_source() call could then
# refer to that on-disk table store;
# let's do that YAML conversion
yaml_write(tbls)

# The above writes the tbl_store.yml
# file (by not providing a filename
# this default filename is chosen);
# next, modify the tbl_source()
# so that store refer to the YAML
# file
agent_3 <-
create_agent(