Error converting SDMX objects without time series structure to data.frame

[laakmann]

Some useful datasets on OECD.stats are available without the usual time series structure, i.e., one cannot filter the dataset by time and the data are not returned with time attributes. For instance, the Better Life Index data can be obtained for several recent years but this requires querying a different dataset for each year (not exactly ideal, but....).

I am able to successfully query the data and the XML contents appear to be consistent with what I'd expect to see (values, structure, etc), however when I attempt to convert the object into a data frame it errors out.

basic example with 2015 BLI dataset

# short simple query
obj = rsdmx::readSDMX('http://stats.oecd.org/restsdmx/sdmx.ashx/GetData/BLI2015/AUS+AUT+BEL.IW_HADI+IW_HNFW.L.TOT/all')

sdf = as.data.frame(obj)
### Error Message here:
# Error in getNodeSet(obsXML, paste("//ns:", timeElement, sep = ""), namespaces = ns)[[1]] : 
#  subscript out of bounds


### Traceback
# 11: parseObs(x)
# 10: FUN(X[[i]], ...)
# 9: lapply(obssXML, function(x) parseObs(x))
# 8: do.call("rbind.fill", lapply(obssXML, function(x) parseObs(x)))
# 7: parseSerie(x)
# 6: FUN(X[[i]], ...)
# 5: lapply(seriesXML, function(x) {
#   serie <- parseSerie(x)
# })
# 4: do.call("rbind.fill", lapply(seriesXML, function(x) {
#   serie <- parseSerie(x)
# }))
# 3: as.data.frame.SDMXGenericData(x, labels = labels)
# 2: as.data.frame.SDMXMessageGroup(obj)
# 1: as.data.frame(obj)

It is apparent this is occurring due to the missing time series attribute, however I need to spend more time studying the code, documentation, etc to fix it properly myself (i.e., without breaking other stuff). Perhaps I can take a crack at this later.

Thank you for the useful package!
Pete

[eblondel]

@laakmann thanks for this info. I've pushed a quick fix, as it didn't require a lot of work. Can you please re-test? Thanks

[laakmann]

Works for me. Thanks!

[eblondel]

Great! @laakmann If rsdmx is useful for you, do not forget to start it! :-)