Handling Data

This section will take you through functions that will clean up your data just after querying it, as well as functions to convert your data into more friendly formats.

Cleaning Queried Data

After querying data (Querying Clients), you'll have a vector of dictionaries. Because of the nature of GraphQL, each of these dictionaries might actually be nested dictionaries. Depending on your query and application, this nesting may prove to be a headache. When your nested dictionaries have unique keys, you can flatten the dictionary to be nicer to handle. Remember, you can apply this function to a vector of dictionaries by using the broadcast operator!

julia> d = Dict("a" => 1, "b" => Dict("c" => Dict("d" => 4)))
julia> flatten(d)
Dict("a" => 1, "b.c.d" => 4)

flatten(d::Dict)

Given a dictionary or vector, you may also want to insert a key-value pair or element respectively if it isn't already set.

julia> d = Dict("a" => 1, "b" => 2)
julia> d = setdefault!(d, "c", 3)
Dict("a" => 1, "b" => 2, "c" => 3)
julia> d = setdefault!(d, "c", 4)
Dict("a" => 1, "b" => 2, "c" => 3)

julia> d = ["a", "b"]
julia> d = setdefault!(d, "c")
["a", "b", "c"]
julia> d = setdefault!(d, "c")
["a", "b", "c"]

setdefault!(d::Dict, k, v)

Reformatting Queried Data

A vector of dictionaries is not the nicest data format to work with. To this end, we provide helper functions to put the data into nicer formats.

TypedTable

In Julia, the Tables.jl interface is quite powerful. Hence, you may find it useful to format your data as a TypedTable. Note that these tables are immutable. They're more efficient if you're not going to be copying your data around for mutation since the compiler can infer column names and types. However, if you plan on mutating your table, use FlexTable instead. Once your data is a TypedTable, you can use packages like SplitApplyCombine.jl to manipulate your data. A typical workflow using the default client from beginning to TypedTable might look like

julia> qvalue = "indexers"
julia> qargs = Dict("first" => 1)
julia> qfields = ["id"]
julia> query(qvalue, qargs, qfields) .|> flatten |> table
Table with 1 column and 1 row:
     id
   ┌───────────────────────────────────────────
 1 │ 0x011bdfea664ece919d895d174f57331460056236

table(d::AbstractVector{D}) where {D<:Dict}
table(d::CSV.File)
table(d::NamedTuple)

FlexTable

This is similar to TypedTable above, except it supports mutation.

flextable(d::AbstractVector{D}) where {D<:Dict}
flextable(d::CSV.File)
flextable(d::NamedTuple)