问题
I produce a number of large arrays in Julia using a script file. Printing out the whole array is cumbersome but I'd like to check the first few rows make sense.
I know in the REPL there's printing which is limited by the screen size e.g.
julia> zeros(1000,10)
1000×10 Array{Float64,2}:
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
⋮ ⋮
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
But I can't find any print/show function in base Julia which mimics this for scripts, say only printing out the first 10 rows of an array or something like R's head (I would have expected showcompact to do something like this).
Is there an analogous function to R's head in Julia or do I have to write my own.
回答1:
As I mentioned in a comment, the way to do this in v0.5 is to use an IOContext.
A very simple way to limit the data is to pass the limit=true parameter to the IOContext;
julia> show(IOContext(STDOUT, limit=true), v)
[0.147959 0.414018 … 0.595528 0.852563; 0.32679 0.824953 … 0.432143 0.036279; … ; 0.877398 0.661854 … 0.197207 0.15596; 0.0522946 0.508075 … 0.835359 0.705987]
But this still does not print out the way the REPL does; that's because show with two arguments uses a single-line display. To use a multiline display, pass text/plain as the second argument to show (a MIME type):
julia> show(IOContext(STDOUT, limit=true), "text/plain", v)
100×100 Array{Float64,2}:
0.147959 0.414018 0.0282934 … 0.816132 0.595528 0.852563
0.32679 0.824953 0.0582351 0.822526 0.432143 0.036279
0.754989 0.724317 0.533966 0.987273 0.931932 0.973622
0.547866 0.282694 0.0295411 0.75929 0.886218 0.0813057
0.0626663 0.111795 0.625083 0.439983 0.562143 0.669046
0.712093 0.469622 0.377298 … 0.298224 0.31853 0.376066
0.774625 0.754328 0.756725 0.61113 0.76566 0.999292
0.917846 0.308363 0.489246 0.715311 0.175302 0.124059
0.310922 0.140575 0.20635 0.0280192 0.683004 0.168129
0.753361 0.755103 0.831806 0.118009 0.122374 0.281476
⋮ ⋱
0.420264 0.7614 0.748408 0.330983 0.0776789 0.309464
0.984379 0.851735 0.595121 0.534982 0.255317 0.743713
0.814505 0.765941 0.71852 0.730677 0.477631 0.0360992
0.910384 0.0747604 0.490685 0.0904559 0.0756424 0.313898
0.628416 0.0790874 0.401488 … 0.523521 0.397249 0.58112
0.578361 0.336352 0.261118 0.838256 0.387374 0.451647
0.66724 0.586342 0.378968 0.602694 0.450686 0.901279
0.877398 0.661854 0.685156 0.658952 0.197207 0.15596
0.0522946 0.508075 0.244423 0.95935 0.835359 0.705987
You can of course change how many rows are shown by passing in displaysize to the IOContext:
julia> show(IOContext(STDOUT, limit=true, displaysize=(10,10)), "text/plain", v)
100×100 Array{Float64,2}:
0.147959 … 0.852563
0.32679 0.036279
0.754989 0.973622
⋮ ⋱
0.877398 0.15596
0.0522946 0.705987
Overall, IOContext is very flexible. See its documentation for more details.
回答2:
You could easily define a head function yourself. For one and two dimensions, this is pretty straightforward:
torange(n::Integer, m) = 1:min(n, m)
torange(c::Colon, m) = (:)
function head(a::AbstractArray{TypeVar(:T), 1}, n = 10)
view(a, torange(n,size(a,1)))
end
function head(a::AbstractArray{TypeVar(:T), 2}, n1 = 10, n2 = 10)
view(a, torange(n1, size(a,1)), torange(n2, size(a,2)))
end
The torange method allows to use a : to return the full length in the corresponding dimension. For example
head(zeros(10, 10), 5, :)
5×10 SubArray{Float64,2,Array{Float64,2},Tuple{UnitRange{Int64},Colon},false}:
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
For more than two dimensions, I have chosen to repeat the last argument beginning from the third:
function head{Na, Nn}(a::AbstractArray{TypeVar(:T), Na}, n1 = 10, n2 = 10, ns::Vararg{TypeVar(:T), Nn} = 2)
nend = last(ns)
view(a, torange(n1, size(a,1)), torange(n2, size(a,2)), (torange(ns[i], size(a,i+2)) for i = 1:Nn)..., (torange(nend, size(a, i)) for i in Nn+3:Na)...)
end
For example:
head(rand(10, 10, 5, 5), 3, 3, 2) # the last two is the default value and can be omitted
3×3×2×2 SubArray{Float64,4,Array{Float64,4},Tuple{UnitRange{Int64},UnitRange{Int64},UnitRange{Int64},UnitRange{Int64}},false}:
[:, :, 1, 1] =
0.384724 0.7328 0.585211
0.738284 0.95145 0.362914
0.43928 0.94307 0.758541
[:, :, 2, 1] =
0.78603 0.588877 0.677201
0.559547 0.800559 0.488433
0.993593 0.691884 0.236595
[:, :, 1, 2] =
0.25732 0.90491 0.323905
0.300924 0.703919 0.813316
0.040522 0.776142 0.624097
[:, :, 2, 2] =
0.746677 0.153574 0.155539
0.991624 0.90167 0.0880094
0.0423263 0.0153597 0.0608328
Note that the n-dimensional version is not type-stable, but that shouldn't matter for printing.
回答3:
We can get the same output as REPL by using display() (is this what you are looking for...?) Also, the "head" and "tail" parts can be printed by using array sections, e.g.,
disp( x ) = ( display(x) ; println() ; println() )
A = diagm( [ i for i=1:100 ] )
disp( A )
disp( A[ 1:5, : ] ) # head
disp( A[ end-4:end, : ] ) # tail
B = [ i for i=1:100 ]
disp( B )
disp( B[ 1:5 ] ) # head
disp( B[ end-4:end ] ) # tail
$ julia test.jl
100x100 Array{Int64,2}:
1 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0
0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 6 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 9 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 10 0 0 0 0 0 0 0 0 0 0 0
⋮ ⋮ ⋮ ⋱ ⋮
0 0 0 0 0 0 0 0 0 0 0 0 92 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 93 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 94 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 95 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 96 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 97 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 98 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 99 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100
5x100 Array{Int64,2}:
1 0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 0 0 0 0 0 0 0
0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
5x100 Array{Int64,2}:
0 0 0 0 0 0 0 0 0 0 0 0 0 … 0 0 0 0 0 96 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 97 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 98 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 99 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100
100-element Array{Int64,1}:
1
2
3
4
5
6
7
8
9
10
⋮
92
93
94
95
96
97
98
99
100
5-element Array{Int64,1}:
1
2
3
4
5
5-element Array{Int64,1}:
96
97
98
99
100
来源：https://stackoverflow.com/questions/40788316/julia-limited-printing-of-large-arrays