Package 'adbcdrivermanager'

Title: 'Arrow' Database Connectivity ('ADBC') Driver Manager
Description: Provides a developer-facing interface to 'Arrow' Database Connectivity ('ADBC') for the purposes of driver development, driver testing, and building high-level database interfaces for users. 'ADBC' <https://arrow.apache.org/adbc/> is an API standard for database access libraries that uses 'Arrow' for result sets and query parameters.
Authors: Dewey Dunnington [aut, cre] , Apache Arrow [aut, cph], Apache Software Foundation [cph]
Maintainer: Dewey Dunnington <[email protected]>
License: Apache License (>= 2)
Version: 0.14.0
Built: 2024-11-04 02:47:33 UTC
Source: https://github.com/apache/arrow-adbc

Help Index


Connection methods

Description

Connection methods

Usage

adbc_connection_get_info(connection, info_codes = NULL)

adbc_connection_get_objects(
  connection,
  depth = 0L,
  catalog = NULL,
  db_schema = NULL,
  table_name = NULL,
  table_type = NULL,
  column_name = NULL
)

adbc_connection_get_table_schema(connection, catalog, db_schema, table_name)

adbc_connection_get_table_types(connection)

adbc_connection_read_partition(connection, serialized_partition)

adbc_connection_commit(connection)

adbc_connection_rollback(connection)

adbc_connection_cancel(connection)

adbc_connection_get_statistic_names(connection)

adbc_connection_get_statistics(
  connection,
  catalog,
  db_schema,
  table_name,
  approximate = FALSE
)

adbc_connection_quote_identifier(connection, value, ...)

adbc_connection_quote_string(connection, value, ...)

Arguments

connection

An adbc_connection

info_codes

A list of metadata codes to fetch, or NULL to fetch all. Valid values are documented in the adbc.h header.

depth

The level of nesting to display. If 0, display all levels. If 1, display only catalogs (i.e., catalog_schemas will be null). If 2, display only catalogs and schemas (i.e., db_schema_tables will be null). If 3, display only catalogs, schemas, and tables.

catalog

Only show tables in the given catalog. If NULL, do not filter by catalog. If an empty string, only show tables without a catalog. May be a search pattern.

db_schema

Only show tables in the given database schema. If NULL, do not filter by database schema. If an empty string, only show tables without a database schema. May be a search pattern.

table_name

Constrain an object or statistics query for a specific table. If NULL, do not filter by name. May be a search pattern.

table_type

Only show tables matching one of the given table types. If NULL, show tables of any type. Valid table types can be fetched from GetTableTypes. Terminate the list with a NULL entry.

column_name

Only show columns with the given name. If NULL, do not filter by name. May be a search pattern.

serialized_partition

The partition descriptor.

approximate

If FALSE, request exact values of statistics, else allow for best-effort, approximate, or cached values. The database may return approximate values regardless, as indicated in the result. Requesting exact values may be expensive or unsupported.

value

A string or identifier.

...

Driver-specific options. For the default method, these are named values that are converted to strings.

Value

  • adbc_connection_get_info(), adbc_connection_get_objects(), adbc_connection_get_table_types(), and adbc_connection_read_partition() return a nanoarrow_array_stream.

  • adbc_connection_get_table_schema() returns a nanoarrow_schena

  • adbc_connection_commit() and adbc_connection_rollback() return connection, invisibly.

Examples

db <- adbc_database_init(adbc_driver_void())
con <- adbc_connection_init(db)
# (not implemented by the void driver)
try(adbc_connection_get_info(con, 0))

Connections

Description

Connections

Usage

adbc_connection_init(database, ...)

adbc_connection_init_default(database, options = NULL, subclass = character())

adbc_connection_release(connection)

adbc_connection_set_options(connection, options)

adbc_connection_get_option(connection, option)

adbc_connection_get_option_bytes(connection, option)

adbc_connection_get_option_int(connection, option)

adbc_connection_get_option_double(connection, option)

Arguments

database

An adbc_database.

...

Driver-specific options. For the default method, these are named values that are converted to strings.

options

A named character() or list() whose values are converted to strings.

subclass

An extended class for an object so that drivers can specify finer-grained control over behaviour at the R level.

connection

An adbc_connection

option

A specific option name

Value

An object of class 'adbc_connection'

Examples

db <- adbc_database_init(adbc_driver_void())
adbc_connection_init(db)

Join the lifecycle of a unique parent to its child

Description

It is occasionally useful to return a connection, statement, or stream from a function that was created from a unique parent. These helpers tie the lifecycle of a unique parent object to its child such that the parent object is released predictably and immediately after the child. These functions will invalidate all references to the previous R object.

Usage

adbc_connection_join(connection, database)

adbc_statement_join(statement, connection)

Arguments

connection

A connection created with adbc_connection_init()

database

A database created with adbc_database_init()

statement

A statement created with adbc_statement_init()

Value

The input, invisibly.

Examples

# Use local_adbc to ensure prompt cleanup on error;
# use join functions to return a single object that manages
# the lifecycle of all three.
stmt <- local({
  db <- local_adbc(adbc_database_init(adbc_driver_log()))

  con <- local_adbc(adbc_connection_init(db))
  adbc_connection_join(con, db)

  stmt <- local_adbc(adbc_statement_init(con))
  adbc_statement_join(stmt, con)

  adbc_xptr_move(stmt)
})

# Everything is released immediately when the last object is released
adbc_statement_release(stmt)

Databases

Description

Databases

Usage

adbc_database_init(driver, ...)

adbc_database_init_default(driver, options = NULL, subclass = character())

adbc_database_release(database)

adbc_database_set_options(database, options)

adbc_database_get_option(database, option)

adbc_database_get_option_bytes(database, option)

adbc_database_get_option_int(database, option)

adbc_database_get_option_double(database, option)

Arguments

driver

An adbc_driver().

...

Driver-specific options. For the default method, these are named values that are converted to strings.

options

A named character() or list() whose values are converted to strings.

subclass

An extended class for an object so that drivers can specify finer-grained control over behaviour at the R level.

database

An adbc_database.

option

A specific option name

Value

An object of class adbc_database

Examples

adbc_database_init(adbc_driver_void())

Log calls to another driver

Description

Useful for debugging or ensuring that certain calls occur during initialization and/or cleanup. The current logging output should not be considered stable and may change in future releases.

Usage

adbc_driver_log()

Value

An object of class 'adbc_driver_log'

Examples

drv <- adbc_driver_log()
db <- adbc_database_init(drv, key = "value")
con <- adbc_connection_init(db, key = "value")
stmt <- adbc_statement_init(con, key = "value")
try(adbc_statement_execute_query(stmt))
adbc_statement_release(stmt)
adbc_connection_release(con)
adbc_database_release(db)

Monkey see, monkey do!

Description

A driver whose query results are set in advance.

Usage

adbc_driver_monkey()

Value

An object of class 'adbc_driver_monkey'

Examples

db <- adbc_database_init(adbc_driver_monkey())
con <- adbc_connection_init(db)
stmt <- adbc_statement_init(con, mtcars)
stream <- nanoarrow::nanoarrow_allocate_array_stream()
adbc_statement_execute_query(stmt, stream)
as.data.frame(stream$get_next())

Create ADBC drivers

Description

Creates the R object representation of an ADBC driver, which consists of a name and an initializer function with an optional subclass to control finer-grained behaviour at the R level.

Usage

adbc_driver_void()

adbc_driver(x, entrypoint = NULL, ..., subclass = character())

Arguments

x, entrypoint

An ADBC driver may be defined either as an init function or as an identifier with an entrypoint name. A driver init func must be an external pointer to a DL_FUNC with the type AdbcDriverInitFunc specified in the adbc.h header.

...

Further key/value parameters to store with the (R-level) driver object.

subclass

An optional subclass for finer-grained control of behaviour at the R level.

Value

An object of class 'adbc_driver'

Examples

adbc_driver_void()

Get extended error information from an array stream

Description

Get extended error information from an array stream

Usage

adbc_error_from_array_stream(stream)

Arguments

stream

A nanoarrow_array_stream

Value

NULL if stream was not created by a driver that supports extended error information or a list whose first element is the status code and second element is the adbc_error object. The acbc_error must not be accessed if stream is explicitly released.

Examples

db <- adbc_database_init(adbc_driver_monkey())
con <- adbc_connection_init(db)
stmt <- adbc_statement_init(con, mtcars)
stream <- nanoarrow::nanoarrow_allocate_array_stream()
adbc_statement_execute_query(stmt, stream)
adbc_error_from_array_stream(stream)

Statements

Description

Statements

Usage

adbc_statement_init(connection, ...)

adbc_statement_init_default(connection, options = NULL, subclass = character())

adbc_statement_release(statement)

adbc_statement_set_options(statement, options)

adbc_statement_get_option(statement, option)

adbc_statement_get_option_bytes(statement, option)

adbc_statement_get_option_int(statement, option)

adbc_statement_get_option_double(statement, option)

Arguments

connection

An adbc_connection

...

Driver-specific options. For the default method, these are named values that are converted to strings.

options

A named character() or list() whose values are converted to strings.

subclass

An extended class for an object so that drivers can specify finer-grained control over behaviour at the R level.

statement

An adbc_statement

option

A specific option name

Value

An object of class 'adbc_statement'

Examples

db <- adbc_database_init(adbc_driver_void())
con <- adbc_connection_init(db)
adbc_statement_init(con)

Statement methods

Description

Statement methods

Usage

adbc_statement_set_sql_query(statement, query)

adbc_statement_set_substrait_plan(statement, plan)

adbc_statement_prepare(statement)

adbc_statement_get_parameter_schema(statement)

adbc_statement_bind(statement, values, schema = NULL)

adbc_statement_bind_stream(statement, stream, schema = NULL)

adbc_statement_execute_query(
  statement,
  stream = NULL,
  stream_join_parent = FALSE
)

adbc_statement_execute_schema(statement)

adbc_statement_cancel(statement)

Arguments

statement

An adbc_statement

query

An SQL query as a string

plan

A raw vector representation of a serialized Substrait plan.

values

A nanoarrow_array or object that can be coerced to one.

schema

A nanoarrow_schema or object that can be coerced to one.

stream

A nanoarrow_array_stream or object that can be coerced to one.

stream_join_parent

Use TRUE to invalidate statement and tie its lifecycle to stream.

Value

  • adbc_statement_set_sql_query(), adbc_statement_set_substrait_plan(), adbc_statement_prepare(), adbc_statement_bind(), adbc_statement_bind_stream(), and adbc_statement_execute_query() return statement, invisibly.

  • adbc_statement_get_parameter_schema() returns a nanoarrow_schema.

Examples

db <- adbc_database_init(adbc_driver_void())
con <- adbc_connection_init(db)
stmt <- adbc_statement_init(con)
# (not implemented by the void driver)
try(adbc_statement_set_sql_query(stmt, "some query"))

Low-level pointer details

Description

  • adbc_xptr_move() allocates a fresh R object and moves all values pointed to by x into it. The original R object is invalidated by zeroing its content. This is useful when returning from a function where lifecycle helpers were used to manage the original object.

  • adbc_xptr_is_valid() provides a means by which to test for an invalidated pointer.

Usage

adbc_xptr_move(x, check_child_count = TRUE)

adbc_xptr_is_valid(x)

Arguments

x

An 'adbc_database', 'adbc_connection', 'adbc_statement', or 'nanoarrow_array_stream'

check_child_count

Ensures that x has a zero child count before performing the move. This should almost always be TRUE.

Value

  • adbc_xptr_move(): A freshly-allocated R object identical to x

  • adbc_xptr_is_valid(): Returns FALSE if the ADBC object pointed to by x has been invalidated.

Examples

db <- adbc_database_init(adbc_driver_void())
adbc_xptr_is_valid(db)
db_new <- adbc_xptr_move(db)
adbc_xptr_is_valid(db)
adbc_xptr_is_valid(db_new)

Read, write, and execute on ADBC connections

Description

These are convenience methods useful for testing connections. Note that S3 dispatch is always on db_or_con (i.e., drivers may provide their own implementations).

Usage

read_adbc(db_or_con, query, ..., bind = NULL)

execute_adbc(db_or_con, query, ..., bind = NULL)

write_adbc(
  tbl,
  db_or_con,
  target_table,
  ...,
  mode = c("default", "create", "append"),
  temporary = FALSE
)

Arguments

db_or_con

An adbc_database or adbc_connection. If a database, a connection will be opened. For read_adbc(), this connection will be closed when the resulting stream has been released.

query

An SQL query

...

Passed to S3 methods.

bind

A data.frame, nanoarrow_array, or nanoarrow_array_stream of bind parameters or NULL to skip the bind/prepare step.

tbl

A data.frame, nanoarrow_array, or nanoarrow_array_stream.

target_table

A target table name to which tbl should be written.

mode

One of "create", "append", or "default" (error if the schema is not compatible or append otherwise).

temporary

Use TRUE to create a table as a temporary table.

Value

Examples

# On a database, connections are opened and closed
db <- adbc_database_init(adbc_driver_log())
try(read_adbc(db, "some sql"))
try(execute_adbc(db, "some sql"))
try(write_adbc(mtcars, db, "some_table"))

# Also works on a connection
con <- adbc_connection_init(db)
try(read_adbc(con, "some sql"))
try(execute_adbc(con, "some sql"))
try(write_adbc(mtcars, con, "some_table"))

Cleanup helpers

Description

Managing the lifecycle of databases, connections, and statements can be complex and error-prone. The R objects that wrap the underlying ADBC pointers will perform cleanup in the correct order if you rely on garbage collection (i.e., do nothing and let the objects go out of scope); however it is good practice to explicitly clean up these objects. These helpers are designed to make explicit and predictable cleanup easy to accomplish.

Usage

with_adbc(x, code)

local_adbc(x, .local_envir = parent.frame())

Arguments

x

An ADBC database, ADBC connection, ADBC statement, or nanoarrow_array_stream returned from calls to an ADBC function.

code

Code to execute before cleaning up the input.

.local_envir

The execution environment whose scope should be tied to the input.

Details

Note that you can use adbc_connection_join() and adbc_statement_join() to tie the lifecycle of the parent object to that of the child object. These functions mark any previous references to the parent object as released so you can still use local and with helpers to manage the parent object before it is joined. Use stream_join_parent = TRUE in adbc_statement_execute_query() to tie the lifecycle of a statement to the output stream.

Value

  • with_adbc() returns the result of code

  • local_adbc() returns the input, invisibly.

Examples

# Using with_adbc():
with_adbc(db <- adbc_database_init(adbc_driver_void()), {
  with_adbc(con <- adbc_connection_init(db), {
    with_adbc(stmt <- adbc_statement_init(con), {
      # adbc_statement_set_sql_query(stmt, "SELECT * FROM foofy")
      # adbc_statement_execute_query(stmt)
      "some result"
    })
  })
})

# Using local_adbc_*() (works best within a function, test, or local())
local({
  db <- local_adbc(adbc_database_init(adbc_driver_void()))
  con <- local_adbc(adbc_connection_init(db))
  stmt <- local_adbc(adbc_statement_init(con))
  # adbc_statement_set_sql_query(stmt, "SELECT * FROM foofy")
  # adbc_statement_execute_query(stmt)
  "some result"
})