Data Sources

Query supports many different types of data sources, and you can often mix and match different source types in one query. This section describes all the currently supported data source types.

DataFrame

DataFrames are probably the most common data source in Query. They are implemented as an Enumerable data source type, and can therefore be combined with any other Enuermable data source type within one query. The range variable in a query that has a DataFrame as its source is a NamedTuple that has fields for each column of the DataFrame. The implementation of DataFrame sources gets around all problems of type stability that are sometimes associated with the DataFrames package.

Example

using Query, DataFrames

df = DataFrame(name=["John", "Sally", "Kirk"], age=[23., 42., 59.], children=[3,5,2])

x = @from i in df begin
    @select i
    @collect DataFrame
end

println(x)

# output

3×3 DataFrames.DataFrame
│ Row │ name    │ age  │ children │
├─────┼─────────┼──────┼──────────┤
│ 1   │ "John"  │ 23.0 │ 3        │
│ 2   │ "Sally" │ 42.0 │ 5        │
│ 3   │ "Kirk"  │ 59.0 │ 2        │

TypedTable

The TypedTables package provides an alternative implementation of a DataFrame-like data structure. Support for TypedTable data sources works in the same way as normal DataFrame sources, i.e. columns are represented as fields of NamedTuples. TypedTable sources are implemented as Enumerable data source and can therefore be combined with any other Enumerable data source in a single query.

Example

using Query, DataFrames, TypedTables

tt = @Table(name=["John", "Sally", "Kirk"], age=[23., 42., 59.], children=[3,5,2])

x = @from i in tt begin
    @select i
    @collect DataFrame
end

println(x)

# output

3×3 DataFrames.DataFrame
│ Row │ name    │ age  │ children │
├─────┼─────────┼──────┼──────────┤
│ 1   │ "John"  │ 23.0 │ 3        │
│ 2   │ "Sally" │ 42.0 │ 5        │
│ 3   │ "Kirk"  │ 59.0 │ 2        │

Arrays

Any array can be a data source for a query. The range variables are of the element type of the array and the elements are iterated in the order of the standard iterator of the array. Array sources are implemented as Enumerable data sources and can therefore be combined with any other Enumerable data source in a single query.

Example

using Query, DataFrames, NamedTuples

immutable Person
    Name::String
    Friends::Vector{String}
end

source = Array(Person,0)
push!(source, Person("John", ["Sally", "Miles", "Frank"]))
push!(source, Person("Sally", ["Don", "Martin"]))

result = @from i in source begin
         @where length(i.Friends) > 2
         @select {i.Name, Friendcount=length(i.Friends)}
         @collect
end

println(result)

# output

NamedTuples._NT_NameFriendcount{String,Int64}[(Name => John, Friendcount => 3)]

DataStream

Any DataStream source can be a source in a query. This includes CSV.jl, Feather.jl and SQLite.jl sources (these are currenlty tested as part of Query.jl). Individual rows of these sources are represented as NamedTuple elements that have fields for all the columns of the data source. DataStreams sources are implemented as Enumerable data sources and can therefore be combined with any other Enumerable data source in a single query.

Example

This example reads a CSV file:

using Query, DataStreams, CSV

q = @from i in CSV.Source(joinpath(Pkg.dir("Query"),"example", "data.csv")) begin
    @where i.Children > 2
    @select i.Name
    @collect
end

println(q)

# output

Nullable{String}["John","Kirk"]

This example reads a Feather file:

using Query, DataStreams, Feather

q = @from i in Feather.Source(joinpath(Pkg.dir("Feather"),"test", "data", "airquality.feather")) begin
    @where i.Day==2
    @select i.Month
    @collect
end

println(q)

# output

WARNING: This Feather file is old and will not be readable beyond the 0.3.0 release
Int32[5,6,7,8,9]

IndexedTables

NDSparse data sources can be a source in a query. Individual rows are represented as a NamedTuple with two fields. The index field holds the index data for this row. If the source has named columns, the type of the index field is a NamedTuple, where the fieldnames correspond to the names of the index columns. If the source doesn't use named columns, the type of the index field is a regular tuple. The second field is named value and holds the value of the row in the original source. NDSparse sources are implemented as Enumerable data sources and can therefore be combined with any other Enumerable data source in a single query.

Example

using Query, IndexedTables

source_ndsparsearray = NDSparse(Columns(city = [fill("New York",3); fill("Boston",3)], date = repmat(Date(2016,7,6):Date(2016,7,8), 2)), [91,89,91,95,83,76])

q = @from i in source_ndsparsearray begin
    @where i.index.city=="New York"
    @select i.value
    @collect
end

println(q)

# output

[91,89,91]

Any iterable type

Any data source type that implements the standard julia iterator protocoll (i.e. a start, next and done method) can be a query data source. Iterable data sources are implemented as Enumerable data sources and can therefore be combined with any other Enumerable data source in a single query.

Example

[TODO]