Metadata model

Our metadata is contained in six highly related tables, which are defined using SQLAlchemy classes.

Relational Diagram

Name	Description
obs_column	OBSColumn
obs_column_table	OBSColumnTable
obs_column_to_column	OBSColumnToColumn
obs_column_tag	OBSColumnTag
obs_table	OBSTable
obs_tag	OBSTag

Manually generated entities

class tasks.meta.OBSColumn(**kwargs)

Describes the characteristics of data in a column, which can exist in multiple physical tables.

These should only be instantiated to generate the return value of ColumnsTask.columns(). Any other usage could have unexpected consequences.

id

The unique identifier for this column. Should be qualified by the module name of the class that created it. Is automatically generated by ColumnsTask if left unspecified, and automatically qualified by module name either way.

type

The type of this column – for example, Text, Numeric, Geometry, etc. This is used to generate a schema for any table using this column.

name

The human-readable name of this column. This is used in the catalog and API.

description

The human-readable description of this column. THis is used in the catalog.

weight

A numeric weight for this column. Higher weights are favored in certain rankings done by the API. Defaults to zero, which hides the column from catalog.

aggregate

How this column can be aggregated. For example, populations can be summed, so a population column should be sum. Should be left blank if it is not to aggregate.

tables

Iterable of all linked :class:`~.meta.OBSColumnTable`s, which could be traversed to find all tables with this column.

tags

Iterable of all linked :class:`~.meta.OBSColumnTag`s, which could be traversed to find all tags applying to this column.

targets

Dict of all related columns. Format is <OBSColumn>: <reltype>.

version

A version control number, used to determine whether the column and its metadata should be updated.

extra

Arbitrary additional information about this column stored as JSON.

catalog_lonlat()

Return tuple (longitude, latitude) for the catalog for this measurement.

denominators()

Return the OBSColumn denominator of this column.

geom_timespans()

Return a dict of geom columns and timespans that this measure is available for.

has_catalog_image()

Returns True if this column has a pre-generated image for the catalog.

has_children()

Returns True if this column has children, False otherwise.

has_denominators()

Returns True if this column has no denominator, False otherwise.

is_cartographic()

Returns True if this column is a geometry that can be used for cartography.

is_geomref()

Returns True if the column is a geomref, else Null

is_interpolation()

Returns True if this column is a geometry that can be used for interpolation.

license_tags()

Return license tags.

source_tags()

Return source tags.

summable()

Returns True if we can sum this column and calculate for arbitrary areas.

unit()

Return the OBSTag unit of this column.

class tasks.meta.OBSTag(**kwargs)

Tags permit arbitrary groupings of columns.

They should only be created as part of a tags() implementation.

id

The unique identifier for this table. Is always qualified by the module name of the TagsTask that created it, and should never be specified manually.

name

The name of this tag. This is exposed in the API and user interfaces.

type

The type of this tag. This is used to provide more information about what the tag means. Examples are section, subsection, license, unit, although any arbitrary type can be defined.

description

Description of this tag. This may be exposed in the API and catalog.

columns

An iterable of all the :class:`~.meta.OBSColumn`s tagged with this tag.

version

A version control number, used to determine whether the tag and its metadata should be updated.

Autogenerated entities

class tasks.meta.OBSColumnToColumn(**kwargs)

Relates one column to another. For example, a Text column may contain identifiers that are unique to geometries in another Geometry column. In that case, an OBSColumnToColumn object of reltype GEOM_REF should indicate the relationship, and make relational joins possible between tables that have both columns with those that only have one.

These should never be created manually. Their creation should be handled automatically from specifying OBSColumn.targets.

These are unique on (source_id, target_id).

source_id

ID of the linked source OBSColumn.

target_id

ID of the linked target OBSColumn.

reltype

required text specifying the relation type. Examples are GEOM_REF and ~.meta.DENOMINATOR.

source

The linked source OBSColumn.

target

The linked target OBSColumn.

class tasks.meta.OBSColumnTable(**kwargs)

Glues together OBSColumn and OBSTable. If this object exists, the related column should exist in the related table, and can be selected with colname.

Unique along both (column_id, table_id) and (table_id, colname).

These should never be created manually. Their creation and removal is handled automatically as part of targets.TableTarget.update_or_create_metadata().

column_id

ID of the linked OBSColumn.

table_id

ID of the linked OBSTable.

colname

Column name for selecting this column in a table.

column

The linked OBSColumn.

table

The linked OBSTable.

extra

Extra JSON information about the data. This could include statistics like max, min, average, etc.

class tasks.meta.OBSTable(**kwargs)

Describes a physical table in our database.

These should never be instantiated manually. They are automatically created by output(). The unique key is :attr:~.meta.OBSTable.id:.

id

The unique identifier for this table. Is always qualified by the module name of the class that created it.

columns

An iterable of all the OBSColumnTable instances contained in this table.

tablename

The automatically generated name of this table, which can be used directly in select statements. Of the format obs_<hash>.

timespan

An OBSTimespan instance containing information about the timespan this table applies to. Obtained from timespan().

the_geom

A simple geometry approximating the boundaries of the data in this table.

description

A description of the table. Not used.

version

A version control number, used to determine whether the table and its metadata should be updated.

geom_column()

Return the column geometry column for this table, if it has one.

Returns None if there is none.

geomref_column()

Return the geomref column for this table, if it has one.

Returns None if there is none.

class tasks.meta.OBSColumnTag(**kwargs)

Glues together OBSColumn and OBSTag. If this object exists, the related column should be tagged with the related tag.

Unique along (column_id, tag_id).

These should never be created manually. Their creation and removal is handled automatically as part of targets.TableTarget.update_or_create_metadata().

column_id

ID of the linked OBSColumn.

tag_id

ID of the linked OBSTag.

column

The linked OBSColumn.

tag

The linked OBSTag.