nolta/Calendar.jl

customize how CalendarTime prints

milktrader opened this issue · 3 comments

milktrader commented

I have gotten the CalendarTime type into a DataFrame here,

julia> head(GLD, 3)
3x5 DataFrame:
                                Date  Open  High   Low Close   
[1,]    Nov 18, 2004 12:00:00 AM EST 44.43 44.49 44.07 44.38  
[2,]    Nov 19, 2004 12:00:00 AM EST 44.49 44.92 44.47 44.78 
[3,]    Nov 22, 2004 12:00:00 AM EST 44.75 44.97 44.74 44.95 

julia> typeof(GLD[:,1])
DataVec{CalendarTime}

julia> GLD[nrow(GLD),1] - GLD[1,1]
2948 days

But it would be nice to have a less verbose print or show of CalendarTime, so it looks like this contrived example:

julia> head(GLD,3)
3x5 DataFrame:
          Date      Open  High Low Close   
[1,]  2004-11-18  44.43 44.49 44.07 44.38  
[2,]  2004-11-19  44.49 44.92 44.47 44.78 
[3,]  2004-11-22  44.75 44.97 44.74 44.95
nolta commented

Hmm, i'm not sure what the best approach here is. Do you want to change the global default format? Or just for this particular DataFrame?

milktrader commented

Changing the global default might be asking too much, since my preferred format may not be right for things other than a time series.

Is the formatting handled at the parse function level or does it go lower-level? I'll pour over the code to see if I can figure this out. But I'd hope it would be as simple as writing a parse_timeseries function that applied different formatting.

Or maybe it's in the show method? Not sure.

okay, it's in show. I'll checkout a branch and see if I can hack this to do what I'd like ..

milktrader commented

Alright, the following is not contrived:

julia> head(GLD,3)
3x7 DataFrame:
              Date  Open  High   Low Close   Volume Adj Close
[1,]    2004-11-18 44.43 44.49 44.07 44.38  5991500     44.38
[2,]    2004-11-19 44.49 44.92 44.47 44.78 11655300     44.78
[3,]    2004-11-22 44.75 44.97 44.74 44.95 11996000     44.95


julia> GLD[nrow(GLD),1] - GLD[1,1]
2948 days

The hack involves the show method around line 189. I just commented out:

#s = ICU.format(_get_format(t.tz), t.millis)

And replaced it with:

s = format("yyyy-MM-dd", t)

I have this in my timeseries branch and it works for me. Not sure how this fits in with the overall goals of the package.

Since this is a working solution, I'll close the issue.

Regards.