Reshape issues

[Davide Spano]

Dear all,
maybe this question will make smile several of you, but now I am facing some difficulties to resolve it.
I got a matrix with data (580+k) on three variables (2 string, 1 double) and I am trying to reshape it wide from long. Stata do not accept that the j(var) have names with spaces, but I am obliged to use that.
As in the do file I have also a variable (str5 "Code") that would allow to resolve this problem because it is uniquely linked to one (origin) of the two string variables that I need to use to perform the reshape, I am trying to replicate the variable (str5) to the second string variable (destination), but I need to respect the unique linkage with the new variable.

I am trying to perform a trivial foreach loop, but I cannot arrive to find the good sintax.

gen code_dest= Code

foreach i in origin {
    replace code_dest=... if ...
}

Hope someone could help me to find a plausible solution to that. Thanks.

[Andrew Musau]

See the function -strtoname()-

help strtoname()

ADDED IN EDIT: I may have misunderstood what you want. The above will allow you to transform destination to a name string that reshape can handle. If you mean that there is a 1:1 correspondence between code and origin, but you want to map code to destination, then

preserve
keep code origin
rename (code origin) (code2 destination)
contract code2 destination
tempfile code
keep code2 destination
save `code'
restore
merge m:1 destination using `code', nogen

[Fei Wang]

Another solution would be

gen code2 = code if origin == destination
bys destination (code2): replace code2 = code2[_N]

[Davide Spano]

Thanks you all, I resolved that! Now I have my 764x764 matrix of distances and I can perform the simulation I have in mind! Thanks a lot!
D.