dict.table {container}

Combining Dict and data.table

Description

The dict.table is a combination of dict and data.table and basically can be considered a data.table with unique column names and an extended set of functions to add, extract and remove data columns with the goal to further facilitate code development using data.table. A dict.table object provides all dict and data.table functions and operators at the same time.

Usage

dict.table(...)

as.dict.table(x, ...)

## S3 method for class 'data.table'
as.dict.table(x, copy = TRUE, ...)

is.dict.table(x)

## S3 method for class 'dict.table'
rbind(x, ...)

## S3 method for class 'dict.table'
cbind(x, ...)

Arguments

...

elements put into the dict.table and/or additional arguments to be passed on.

x

any R object or a dict.table object.

copy

if TRUE creates a copy of the data.table object otherwise works on the passed object by reference.

Details

Methods that alter dict.table objects usually come in two versions providing either copy or reference semantics where the latter start with 'ref_' to note the reference semantic, for example, add() and ref_add().

• dict.table(...) initializes and returns a dict object.

• as.dict.table(x, ...) coerce x to a dict.table

• is.dict.table(x) check if x is a dict.table

• add(.x, ...) and ref_add(.x, ...) add columns to .x. If the column name already exists, an error is given.

• at(.x, ...) returns the columns at the given indices. Indices can be letters or numbers or both. All columns must exist.

• at2(x, index) returns the column at the given index or signals an error if not found.

• clear(x) and ref_clear(x) remove all elements from x.

• clone(x) create a copy of x.

• delete_at(.x, ...) and ref_delete_at(.x, ...) find and remove columns either by name or index (or both). If one or more columns don't exist, an error is signaled.

• discard_at(.x, ...) and ref_discard_at(.x, ...) find and remove columns either by name or index (or both). Invalid column indices are ignored.

• has(x, column) check if some column is in dict.table object.

• has_name(x, name) check if x has the given column name.

• is_empty(x) TRUE if object is empty otherwise FALSE

• peek_at(x, ..., .default = NULL) returns the columns at the given indices or (if not found) columns with the given default value.

• peek_at2(x, index, default = NULL) return column named index if it exist otherwise the given default value. If the default length does not match the number of rows, it is recycled accordingly and a warning is given, unless the default value has a length of 1, in which case recycling is done silently.

• ref_pop(.x, index) return element at given column index and remove the column from the dict.table object.

• rename(.x, old, new) and ref_rename(.x, old, new) rename one or more columns from old to new, respectively, by copy and in place (i.e. by reference).

• replace_at(.x, .., .add = FALSE) and ref_replace_at(.x, ..., .add = FALSE) replace values at given indices. If a given index is invalid, an error is signaled unless .add was set to TRUE.

• update(object, other) and ref_update(object, other) adds columns of other dict that are not yet in object and replaces the values at existing columns.

See Also

dict, data.table

Examples

# Some basic examples using some typical data.table and dict operations.
# The constructor can take the 'key' argument known from data.table():
require(data.table)
dit = dict.table(x = rep(c("b","a","c"), each = 3), y = c(1,3,6), key = "y")
print(dit)
setkey(dit, "x")                             # sort by 'x'
print(dit)
(add(dit, "v" = 1:9))                        # add column v = 1:9
dit[y > 5]
(ref_discard_at(dit, "x"))                   # discard column 'x'

try(at(dit, "x"))                            # index 'x' not found
try(replace_at(dit, x = 0))                  # cannot be replaced, if it does not exist

dit = replace_at(dit, x = 0, .add = TRUE)    # ok - re-adds column 'x' with all 0s
peek_at(dit, "x")                            # glance at column 'x'
has_name(dit, "x")                           # TRUE
ref_pop(dit, "x")                            # get column and remove it
has_name(dit, "x")                           # FALSE


# Copy and reference semantics when coercing *from* a data.table
dat = data.table(a = 1, b = 2)
dit = as.dict.table(dat)
is.dict.table(dit)                           # TRUE
is.dict.table(dat)                           # FALSE
ref_replace_at(dit, "a", 9)
dit[["a"]]                                   # 9
dat[["a"]]                                   # 1
dit.dat = as.dict.table(dat, copy = FALSE)   # init by reference
ref_replace_at(dit.dat, "a", 9)
dat[["a"]]                                   # 9
is.dict.table(dit.dat)                       # TRUE
is.dict.table(dat)                           # TRUE now as well!

# Coerce from dict
d = dict(a = 1, b = 1:3)
as.dict.table(d)

dit = dict.table(a = 1:2, b = 1:2)
rbind(dit, dit)

# rbind ...
dit = dict.table(a = 1:2, b = 1:2)
rbind(dit, dit)

# ... can be mixed with data.tables
dat = data.table(a = 3:4, b = 3:4)
rbind(dit, dat)  # yields a dict.table
rbind(dat, dit)  # yields a data.table

# cbind ...
dit = dict.table(a = 1:2, b = 1:2)
dit2 = dict.table(c = 3:4, d = 5:6)
cbind(dit, dit2)

# ... can be mixed with data.tables
dat = data.table(x = 3:4, y = 3:4)
cbind(dit, dat)

dit = dict.table(a = 1:3)
add(dit, b = 3:1, d = 4:6)

try(add(dit, a = 7:9))  # column 'a' already exists

dit = dict.table(a = 1:3, b = 4:6)
at(dit, "a")
at(dit, 2)
at(dit, "a", 2)
try(at(dit, "x"))     # index 'x' not found
try(at(dit, 1:3))     # index 3 exceeds length of dict.table

dit = dict.table(a = 1:3, b = 4:6)
at2(dit, 1)
at2(dit, "a")
at2(dit, 2)
try(at2(dit, "x"))     # index 'x' not found
try(at2(dit, 5))       # index 5 exceeds length of dict.table

dit = dict.table(a = 1, b = 2)
clear(dit)
dit
ref_clear(dit)
dit

d = dict.table(a = 1:2, b = 3:4)
d2 = clone(d)
ref_clear(d)
print(d2)

(dit = as.dict.table(head(sleep)))
delete_at(dit, "ID")
delete_at(dit, "ID", 1)

try({
delete_at(dit, "foo")   # Column 'foo' not in dict.table
})

dit = as.dict.table(head(sleep))
discard_at(dit, "ID")
discard_at(dit, "ID", 1)
discard_at(dit, "foo")  # ignored

dit = dict.table(a = 1:3, b = as.list(4:6))
has(dit, 1:3)            # TRUE
has(dit, 4:6)            # FALSE
has(dit, as.list(4:6))   # TRUE

dit = dict.table(a = 1, b = 2)
has_name(dit, "a")    # TRUE
has_name(dit, "x")    # FALSE

d = dict.table(a = 1:4, b = 4:1)
is_empty(d)
is_empty(clear(d))

dit = dict.table(a = 1:3, b = 4:6)
peek_at(dit, "a")
peek_at(dit, 1)
peek_at(dit, 3)
peek_at(dit, "x")
peek_at(dit, "x", .default = 0)
peek_at(dit, "a", "x", .default = 0)

dit = dict.table(a = 1:3, b = 4:6)
peek_at2(dit, "a")
peek_at2(dit, 1)
peek_at2(dit, 3)
peek_at2(dit, 3, default = 9)
peek_at2(dit, "x")
peek_at2(dit, "x", default = 0)

dit = dict.table(a = 1:3, b = 4:6)
ref_pop(dit, "a")
ref_pop(dit, 1)

try({
ref_pop(dit, "x")  # index 'x' not found
})

dit = dict.table(a = 1, b = 2, c = 3)
rename(dit, c("a", "b"), c("a1", "y"))
print(dit)
ref_rename(dit, c("a", "b"), c("a1", "y"))
print(dit)

dit = dict.table(a = 1:3)
replace_at(dit, "a", 3:1)

try({
replace_at(dit, "b", 4:6)               # column 'b' not in dict.table
})
replace_at(dit, "b", 4:6, .add = TRUE)  # ok, adds column

# Update parts of tables (second overwrites columns of the first)
dit1 = dict.table(a = 1:2, b = 3:4)
dit2 = dict.table(         b = 5:6, c = 8:9)
update(dit1, dit2)
update(dit2, dit1)