return type based on content of an array
Hi,
I'm going circles... ;) I read a string that contains an array of
unknown dimension like:
a = [1,2,3,4] or
a = [[1,2],[3,4]] or
a = [[[1,2],[3,4]],[[5,6],[7,8]]]
With that I want to perform specified operations e.g. also
provided on command line.
Because at compile time the dimension of the array is unkown I
did not find any possibility to do something like:
auto arr = func(a);
where 'auto func(string str)' should return either a 1D, 2D, 3D
array and arr is not locked in an if(){} scope.
Is there any way and What should I'm looking for?
Thanks, thorstein
Re: confused with data types
Thanks for all the insights :)
Re: confused with data types
On Sunday, 18 February 2018 at 12:51:04 UTC, thorstein wrote:
Thank you for the very informative answers showing different
gears in D!
However, there are still some details I'm struggling with:
Assume some calculations on a very big numeric array
'double[][][] arr'.
Now we could choose 1 out of 3 different implementations:
// Solution 1
foreach(row; arr)
{ foreach(col; row)
{ col[] *= skalar;
}
}
return arr;
// Solution 2
import std.array;
return array(arr.map!(b => array(b[].map!(c => array(c[].map!(d
=> d * skalar));
// Solution 3
import std.algorithm;
arr.each!(a => a[].each!(b => b[] *= skalar));
return arr;
Q#1:
Does the compiler optimizes all solutions equally strong or
does it prefer implementations like solution 1?
Q#2:
Solution 2 is a 1-liner but a bit harder to read. Why reducing
solution 3 to:
return arr.each!(a => a[].each!(b => b[] *= skalar));
gives a compile error? I do writeln() the function result.
Q#3:
If I can:
static import std.array;
return std.array.array(arr.map!(b => std.array.array(b[].map!(c
=>...
How would I apply a similar version with 'static import
std.algorithm' to solution 3?
static import std.algorithm;
std.algorithm.arr.each!(a => a[]... //does obviously not work
Thanks, thorstein
Sorry, Solution 2 should be:
import std.array;
return array(arr.map!(b => array(b[].map!(c => c[] *= skalar;
and is as readable as Solution 3. However, Q#2 still remains.
Re: confused with data types
Thank you for the very informative answers showing different
gears in D!
However, there are still some details I'm struggling with:
Assume some calculations on a very big numeric array
'double[][][] arr'.
Now we could choose 1 out of 3 different implementations:
// Solution 1
foreach(row; arr)
{ foreach(col; row)
{ col[] *= skalar;
}
}
return arr;
// Solution 2
import std.array;
return array(arr.map!(b => array(b[].map!(c => array(c[].map!(d
=> d * skalar));
// Solution 3
import std.algorithm;
arr.each!(a => a[].each!(b => b[] *= skalar));
return arr;
Q#1:
Does the compiler optimizes all solutions equally strong or does
it prefer implementations like solution 1?
Q#2:
Solution 2 is a 1-liner but a bit harder to read. Why reducing
solution 3 to:
return arr.each!(a => a[].each!(b => b[] *= skalar));
gives a compile error? I do writeln() the function result.
Q#3:
If I can:
static import std.array;
return std.array.array(arr.map!(b => std.array.array(b[].map!(c
=>...
How would I apply a similar version with 'static import
std.algorithm' to solution 3?
static import std.algorithm;
std.algorithm.arr.each!(a => a[]... //does obviously not work
Thanks, thorstein
confused with data types
Hello,
This was my goal:
-
public double[][] skalar_m_2d(double[][] array, double skalar)
{
return array.map!(b => b[].map!(c => c * skalar));
}
!!! But: return value is not double!
Type check for return value:
a = array.map!(b => b[].map!(c => c * skalar))
writeln(typeof(a).stringof); //-> MapResult!("__lambda4,
double[][])
a = array.map!(b => b[].map!(c => c * skalar))[1];
writeln(typeof(a).stringof); //-> MapResult!("__lambda2, double[])
How can I get the result as double[][] ???
Thanks, thorstein
conversion error related to array index
Hi,
I get the following compile error with this function below:
source\mod_data\matrices.d(66,12): Error: cannot implicitly
convert expression (j) of type ulong to uint
(66,12) is [j] in row[j]
Using uint as type for rows, cols (i.e. the indices) works. Is
ulong not allowed for array indices? I could not find the
respective information.
Thanks!
49 private double[][] matrix_uniform(ulong rows, ulong cols,
double lbound, double ubound, ulong precision)
50 {
51double[][] array;
52double[] rowT;
53auto gen = Random(unpredictableSeed);
54auto rndUniform = uniform(lbound, ubound, gen);
55
56foreach(ulong i; 0..cols)
57{
58 rowT ~= 0;
59}
60
61foreach(ulong i; 0..rows)
62{
63 foreach(ulong j; 0..cols)
64 {
65rndUniform = uniform(lbound, ubound, gen);
66rowT[j] = to!double(rndUniform) / 10^^precision;
67 }
68 array ~= rowT.dup;
69}
70 return array;
71 }
Re: how to use unknown size of array at compile time for further processing
Replace arrayT ~= rowT; with arrayT ~= rowT.dup; Also, you may want to look into ndslice package [1]. [1] https://github.com/libmir/mir-algorithm Best regards, Ilya Yaroshenko Works! Thanks Ilya. I'll take a look on the difference. Also soon will get bit familiar with the mir-algorithm... Best, Torsten
Re: how to use unknown size of array at compile time for further processing
Guyes, I obviously made a mistake. Forget my post!
Sorry, I'm still really confused with the results from my
function:
double[][] transp(double[][] array)
{
double[][] arrayT;
double[] rowT;
// initialize rowT
foreach(i; 0..array.length)
{
rowT ~= 0;
}
foreach(col; 0..array[0].length-3)
{
foreach(row; 0..array.length)
{
rowT[row] = array[row][col];
}
writeln("rowT 1: ",rowT);
writeln("arrayT 1: ", arrayT);
arrayT ~= rowT;
writeln("rowT 2: ",rowT);
writeln("arrayT 2: ", arrayT,"\n");
}
return arrayT;
}
Again, assuming my original array to be:
array =
[
[ 4, 3, 1, 5, 2],
[19, 34, 23, 100, 59],
[62, 1, 4, 0, 76],
[23, 6, 989, 98, 1],
[ 5, 87, 45, 62, 9],
[ 5, 87, 45, 62, 9]
];
... the results are strange. Before the appending (arrayT ~=
rowT;) happens for the second column of array arrayT was
miraculous modified:
rowT 1: [4, 19, 62, 23, 5, 5]
arrayT 1: []
rowT 2: [4, 19, 62, 23, 5, 5]
arrayT 2: [[4, 19, 62, 23, 5, 5]]
rowT 1: [3, 34, 1, 6, 87, 87]
arrayT 1: [[3, 34, 1, 6, 87, 87]]
rowT 2: [3, 34, 1, 6, 87, 87]
arrayT 2: [[3, 34, 1, 6, 87, 87], [3, 34, 1, 6, 87, 87]]
etc...
Is there any linkage between rowT and arrayT
Thanks again!
Re: how to use unknown size of array at compile time for further processing
On Sunday, 1 October 2017 at 13:53:57 UTC, thorstein wrote: Why is that Thanks! Guyes, I obviously made a mistake. Forget my post!
Re: how to use unknown size of array at compile time for further processing
They are not alternatives. They are the only way of doing
things.
Yes, sounds logic - static arrays require a size before
compilation.
However, I tried another variation with a totally unexpected
result:
double[][] transp3(double[][] array)
{
double[][] arrayT;
double[] rowT;
// initialize rowT
foreach(i; 0..array.length)
{
rowT ~= 0;
}
foreach(col; 0..array[0].length)
{
foreach(row; 0..array.length)
{
rowT[row] = array[row][col];
}
arrayT ~= rowT;
writeln("rowT: ",rowT);
writeln("arrayT: ", arrayT);
}
return arrayT;
}
Assuming my original array to be:
array =
[
[ 4, 3, 1, 5, 2],
[19, 34, 23, 100, 59],
[62, 1, 4, 0, 76],
[23, 6, 989, 98, 1],
[ 5, 87, 45, 62, 9],
[ 5, 87, 45, 62, 9]
];
...then the printed result is unexpectedly wrong. The transposed
columns (rowT) are correct, but arrayT ~= rowT doesn't only
append rowT to arrayT. It overrides also the previous rows with
the new transposed column:
rowT: [4, 19, 62, 23, 5, 5]
arrayT: [[4, 19, 62, 23, 5, 5]]
rowT: [3, 34, 1, 6, 87, 87]
arrayT: [[3, 34, 1, 6, 87, 87], [3, 34, 1, 6, 87, 87]]
etc...
Why is that
Thanks!
how to use unknown size of array at compile time for further processing
Hi,
assumed, I need the size of a dynamic array for further
processing, which is unknown at compile time. Below are my
example, which doesn't work and two alternatives.
Alternative1: may create large rowT-arrays depending on original
array size.
Alternative2: overrides rowT after exiting the inner loop (which
may be costly?)
How could I implement something like the first approach?
Thanks!
// Not working example of transposing a matrix
double[][] transp(double[][] array)
{
double[][] arrayT;
double[array[0].length] rowT;
foreach(col; 0..array[0].length)
{
foreach(row; array)
{
rowT[col]= row[col];
}
arrayT ~= rowT;
}
return arrayT;
}
// Alternative1
double[][] arrayT;
double[] rowT;
foreach(col; 0..array[0].length)
{
foreach(row; array)
{
rowT ~= row[col];
}
arrayT ~= rowT[$-6..$];
}
// Alternative2
double[][] arrayT;
foreach(col; 0..array[0].length)
{
double[] rowT;
foreach(row; array)
{
rowT ~= row[col];
}
arrayT ~= rowT;
}
Re: questions about dub
Ups, somehow overread the last sentence of tourge :) Thanks for the detailed insights!
Re: questions about dub
Thanks to all, I got it! I created a new dub package and copied my code there. Compiles. So I guess I get the rest working as well. Still will have to figure out later the procedure to do the same for a VisualD project, if it is possible. Thorstein
questions about dub
Hi, I have questions regarding the usage of 'dub'. I'm learning D under Win7. I have installed VisualD for the community edition of Visual Studio and got some file i/o working. Next I would like to continue with the mir-tools for matrix manipulation. I understood that I have to build them first using dub. But I didn't succeed: C:\..\AppData\Roaming\dub>dub fetch mir-algorithm Fetching mir-algorithm 0.1.1... Please note that you need to use `dub run ` or add it to dependencies of your package to actually use/run it. dub does not do actual installation of packages outside of its own ecosystem. C:\..\AppData\Roaming\dub>dub run mir-algorithm Building package mir-algorithm in C:\..\AppData\Roaming\dub\packages\mir-algorithm-0.1.1\mir-algorithm\ Fetching mir-internal 0.0.5 (getting selected version)... Main package must have a binary target type, not sourceLibrary. Cannot build. Thats where I stuck. Beside my specific problem of how to start with the mir-tools I wonder how and for what purpose 'dub' is applied when building projects in connection with Visual Studio? Or is it just a more light-weight command line build tool? Thanks for shedding some light! Thorstein
Re: code D'ish enough? - ignore previous post with same subject
On Sunday, 26 February 2017 at 21:50:38 UTC, Jordan Wilson wrote:
auto readNumMatCsv2 (string filePath, string ndv, string
new_ndv){
double[][] p_numArray;
try {
auto lines = File(filePath,"r").byLine;
lines.popFront; // get read of header
p_numArray = lines.map!(a => a.replace (ndv,new_ndv)
.splitter (",")
.map!(a => a.to!double)
.array)
.array;
} catch (Exception e){
e.msg.writeln; // this replaces "Could not read file.
Quit here."
}
return p_numArray;
}
It took me quite a while to get the whole usage of range stuff
like "map" and "filter" etc., but I think it's worth the effort.
This looks like more friendly readable compact code, where should
be the goal. Thanks for your insights!
Re: code D'ish enough? - ignore previous post with same subject
I really appriciate your comments and thoughts! On Sunday, 26 February 2017 at 21:02:52 UTC, ag0aep6g wrote: On Sunday, 26 February 2017 at 19:34:33 UTC, thorstein wrote: * "no-data-value"? No-data-values in data sets are common at least in geosciences: raster images, spatial simulation outputs etc. Not all cells have evaluable information and need to be filtered by a specific very high or very low numeric value. And different software may require different no-data-values. * It's not obvious to me what "Mat" means in "readNumMatCsv". means 'read numeric matrix from csv-like files'
code D'ish enough? - ignore previous post with same subject
Hi,
sorry for posting again, but I used a keyboard combination that
accidently send my post before it was done.
Coming more or less from Python I just started with D. Not a real
programmer, just automating things and looking for a neat
compiled language.
Just to learn, I wrote a function to read CSV-like files (I know
D has its own routine). Since I'm still a bit overwhelmed by the
many complex language features, I'm curious what could I change
to make my code as D'ish as possible?
Thank you for any suggestion,
Thorstein
// Reads CSV-like files with only numeric values in each column
// new_ndv replaces ndv, which is the original no-data-value
double[][]* readNumMatCsv(char[] filePath, int numHeaderLines,
char[] ndv, char[] new_ndv)
{ double[][]* p_numArray;
double[][] numArray;
char[] line;
string noCont = "File content not usable. Quit here.";
string noFile = "Could not read file. Quit here.";
string nonNum = "Found a non-numeric value in data matrix. Quit
here.";
Regex!char re = regex(r"(\n$)");
if(exists(filePath))
{ File f = File(filePath, "r");
if((line = f.readln().dup).length > 0)
{ while (!f.eof())
// 1st replace ndv with new_ndv, 2nd remove all \n, 3rd
check id size of read line >0
{ if((line = replaceAll(f.readln().dup.replace(ndv,
new_ndv), re, "")).length > 0)
// check if all elements of splitted line are numeric
{ foreach(i;split(line,","))
{ if(isNumeric(i) == false)
// otherwise return pointer to empty array
{ writeln(nonNum);
return p_numArray;
}
}
// convert characters to double
if(split(line,",").length > 0)
{ numArray ~= to!(double[])(split(line,","));
}
}
}
// pass reference to pointer
p_numArray =
// first line empty -> return pointer to empty array
} else { writeln(noCont); }
// file could not be find
} else { writeln(noFile); }
return p_numArray;
}
code D'ish enough?
Hi,
coming more or less from Python I just started with D. Not a real
programmer, just automating things and looking for a neat
compiled language.
Just to learn, I wrote a function to read CSV-like files (I know
D has its own routine). Since I'm still a bit overwhelmed by the
many complex language features, I'm curious what could I change
to make my code as D'ish as possible?
Thank you for any suggestion,
Thorstein
// Reads CSV-like files with only numeric values in each column
// new_ndv replaces ndv, which is the original no-data-value
double[][]* readNumMatCsv(char[] filePath, char[] ndv, char[]
new_ndv)
{ double[][]* p_numArray;
double[][] numArray;
char[] line;
string noCont = "File content not usable. Quit here.";
string noFile = "Could not read file. Quit here.";
string nonNum = "Found a non-numeric value in data matrix. Quit
here.";
Regex!char re = regex(r"(\n$)");
if(exists(filePath))
{ File f = File(filePath, "r");
if((line = f.readln().dup).length > 0)
{ while (!f.eof())
{ if((line = replaceAll(f.readln().dup.replace(ndv,
new_ndv), re, "")).length > 0)
{ foreach(i;split(line,","))
{ if(isNumeric(i) == false)
{ writeln(nonNum);
return p_numArray;
}
}
if(split(line,",").length > 0)
{ numArray ~= to!(double[])(split(line,","));
}
}
}
p_numArray =
} else { writeln(noCont); }
} else { writeln(noFile); }
return p_numArray;
}
