Is it possible to get available dates from a TimeDimension?

[romran-altic]

Hello, I am trying to get dimensions explanations with:
cl = structure.dimensions.get(dimension).local_representation.enumerated
sdmx.to_pandas(cl)
Is there any possible approach to get available dates from TimeDimension because code above returns empty.
Asking because I need to get min max date or available dates list from smdx metadata.
Thanks for your help :)

[khaeru]

Hi there!

This is a general SDMX question, and doesn't have much to do with this package. It's also hard to answer because you don't say which particular data provider and structures you are working with. For instance, where do you get structure and dimension from? This is necessary info because each data provider chooses different ways to use the SDMX standard to represent their data.

In general—again, these are all features of the SDMX standard or ways it is commonly used, nothing about this package:

• The Representation (of a dimension in a data structure) can be either enumerated or non-enumerated.
  • "Enumerated" means "only values appearing in a specific code list"—it seems like you're aware fo this.
  • "Non-enumerated" data has no associated code list. It can still have specific formats. These are specified using Facet and related classes. For example, a non-enumerated representation can be specified as "a string with format YYYY-MM", or something like that.
• Many data providers use enumerated representations for most dimensions, yet not for TIME_PERIOD or similar. This would imply there is no codelist that gives allowable values of that dimension.
• I think it is possible to use SDMX's constraints mechanisms to specify allowable (a list, or min/max) values of a non-enumerated dimension. But I say "I think" because this is not very common; I don't think most of the known data providers do this.

But again:

• The purpose of this package and its documentation is not to repeat the SDMX standards documents. I encourage you to read them, or some of the learning resources on the official SDMX website.
• Without saying specifically which (meta)data you are working with, it's not possible to say whether this package is working correctly, or if there is specific information you can retrieve. If you can provide a “minimal working example” then we can discuss this.

[romran-altic]

Hello, many thanks for your detailed answer. Currently I am trying to get PRC_HICP_CMON table from ESTAT source.
table_msg = self.sdmx_client.dataflow(table)
table_flow = table_msg.dataflow[table]
structure = table_flow.structure

As mentioned for TIME_PERIOD component with structure.dimensions.get('TIME_PERIOD').local_representation I could get Facet:

<Representation: None, [Facet(type=FacetType(is_sequence=None, min_length=None, max_length=None, min_value=None, max_value=None, start_value=None, end_value=None, interval=None, time_interval=None, decimals=None, pattern=None, start_time=None, end_time=None, sentinel_values=None), value=None, value_type=<FacetValueType.observationalTimePeriod: 18>)]>

I noticed value_type=<FacetValueType.observationalTimePeriod: 18>)]> my understanding that it contains 18 values. Could you know if it's possible to retrieve any data from this class?

[khaeru]

Thanks for the extra information. So indeed this is a typical DSD in which TIME_PERIOD is not enumerated by any code list.

I noticed value_type=<FacetValueType.observationalTimePeriod: 18>)]> my understanding that it contains 18 values.

No, that's not correct. FacetValueType is an ordinary Python Enum; you can see this in the documentation where it says "Bases: Enum". For instance:

>>> from enum import Enum
>>> Foo = Enum("Foo", "A B C")
>>> Foo.C
<Foo.C: 3>

The only thing we can infer from "18" is that "observationalTimePeriod" is the 18th entry in this list:

sdmx/sdmx/model/common.py

Lines 478 to 496 in 8d5e467

	#: SDMX FacetValueType.
	#:
	#: In the SDMX 2.0 IM, three diagrams in the spec show this enumeration containing
	#: 'gregorianYearMonth' but not 'gregorianYear' or 'gregorianMonth'. The table in
	#: §3.6.3.3 Representation Constructs does the opposite. One ESTAT query (via SGR) shows
	#: a real-world usage of 'gregorianYear'; while one query shows usage of
	#: 'gregorianYearMonth'; so all three are included.
	FacetValueType = Enum(
	"FacetValueType",
	"""string bigInteger integer long short decimal float double boolean uri count
	inclusiveValueRange alpha alphaNumeric numeric exclusiveValueRange incremental
	observationalTimePeriod standardTimePeriod basicTimePeriod gregorianTimePeriod
	gregorianYear gregorianMonth gregorianYearMonth gregorianDay reportingTimePeriod
	reportingYear reportingSemester reportingTrimester reportingQuarter reportingMonth
	reportingWeek reportingDay dateTime timesRange month monthDay day time duration
	keyValues identifiableReference dataSetReference """
	# SDMX 3.0 only
	"geospatialInformation",
	)

Again, this FacetValueType is directly copied from the SDMX standards. Please, click on "Section 2 — Information Model" and search for "FacetValueType"; you will find it described in Section 3.6 of that file.