Need to mark the string returned as UTF-8

[shrektan]

Hi @eddelbuettel , first of all, thanks for making this great package.

Occasionally, we need to return non-English characters from Bloomberg (as the example below). However, because the current version doesn't mark the encoding as UTF8 explicitly, it causes trouble on a Windows machine, where the native encoding varies.

This PR will fix this. I have manually compiled the PR version and tested it on the Bloomberg computer in the office and find no problems by far.

If there's more need to be done, please let me know.

Thanks.

Example

library(Rblpapi)
blpConnect()
out <- bdp(securities = c("BBG006YQMFQ5", "BBG005CMJQ71"),
           fields = c("NAME_CHINESE_SIMPLIFIED", "ISSUE_DT"),
           overrides = c("SECURITY_NAME_LANG" = "10"))
Encoding(out$NAME_CHINESE_SIMPLIFIED)
# before this PR, it says "unknown" and it causes trouble, because we have 
# to change the encoding by hand like `Encoding(out$NAME_CHINESE_SIMPLIFIED) <- "UTF-8"`
# in order to display the strings correctly.
out$ISSUE_DT
out$NAME_CHINESE_SIMPLIFIED

[shrektan]

Sorry, there's a bug... The NA Date value now returns "1970-01-01". I'll fix this and let you know...

[shrektan]

It's weird... If I compiled the package from the master branch (w/o this PR) on my computer, it still returns "1970-01-01" but the CRAN version returns the correct value "NA".

I have no clue at all.

So I think it's unrelated to this PR and the PR itself should be safe...

[shrektan]

I believe the "Date Issue" is caused by the PR #273

Rblpapi/src/bds.cpp

Lines 134 to 136 in 6e52cec

	case BLPAPI_DATATYPE_DATE:
	ans = Rcpp::DateVector(n);
	break;

The default value created by Rcpp::DateVector() is 1970-01-01 rather than NA

Rcpp::cppFunction("SEXP test() { 
SEXP out;
out = Rcpp::DateVector(5);
return out;
}")
test()
#> [1] "1970-01-01" "1970-01-01" "1970-01-01" "1970-01-01" "1970-01-01"

^{Created on 2018-10-25 by the reprex package (v0.2.1)}

[shrektan]

@eddelbuettel I address the above issue in another PR #279

And as usual, I tested it and works on my computer.

[eddelbuettel]

Yes, marking as utf-8 is a good idea. Having a similar issue in another project as it happens.

I cannot currently test. @johnlaing Any chance you could take a peek?

[johnlaing]

Not going to be able to get to this during the day, will try to have a look this evening.

[johnlaing]

@shrektan This is great, thanks for your contribution here and also on the date regression.

[eddelbuettel]

Seconded, and FWIW I just got a very similar "mark as UTF-8" PR for RcppTOML. Being where we are we something forget not that all encodings as as "baked in" as the one we use here.