benchmark_queries

utilities.benchmarking.benchmark_queries

Run a benchmark on the queries to the database.

This script will run a series of queries to the database and measure the time it takes. The queries are random and the number of records returned is also random. The idea is to measure the performance of the database when querying for different time ranges and different variables.

Attributes

days = end_date - start_date / timedelta(days=1) module-attribute

end = Measurement.objects.latest('time').time module-attribute

end_date = random_date(start_date, end) module-attribute

execution = [] module-attribute

queries = 500 module-attribute

records = Measurement.objects.filter(station=station, variable=variable, time__range=(start_date, end_date)) module-attribute

records_df = pd.DataFrame.from_records(records.values()) module-attribute

start = Measurement.objects.earliest('time').time module-attribute

start_date = random_date(start, end) module-attribute

station = random.choice(stations) module-attribute

stations = [Station.objects.get(pk=v) for v in set(Measurement.objects.values_list('station', flat=True))] module-attribute

tend = time.time() module-attribute

tstart = time.time() module-attribute

variable = random.choice(variables) module-attribute

variables = [Variable.objects.get(pk=v) for v in set(Measurement.objects.values_list('variable', flat=True))] module-attribute

years = end - start.days // 365 module-attribute

Classes

Measurement

Bases: MeasurementBase

Class to store the measurements and their validation status.

This class holds the value of a given variable and station at a specific time, as well as auxiliary information such as maximum and minimum values, depth and direction, for vector quantities. All of these have a raw version where a backup of the original data is kept, should this change at any point.

Flags to monitor its validation status, if the data is active (and therefore can be used for reporting) and if it has actually been used for that is also included.

Attributes:

Name	Type	Description
depth	int	Depth of the measurement.
direction	Decimal	Direction of the measurement, useful for vector quantities.
raw_value	Decimal	Original value of the measurement.
raw_maximum	Decimal	Original maximum value of the measurement.
raw_minimum	Decimal	Original minimum value of the measurement.
raw_direction	Decimal	Original direction of the measurement.
raw_depth	int	Original depth of the measurement.
is_validated	bool	Flag to indicate if the measurement has been validated.
is_active	bool	Flag to indicate if the measurement is active. An inactive measurement is not used for reporting

Attributes

overwritten: bool property

Indicates if any of the values associated to the entry have been overwritten.

Returns:

Name	Type	Description
bool	bool	True if any raw field is different to the corresponding standard field.

raws: tuple[str, ...] property

Return the raw fields of the measurement.

Returns:

Type	Description
tuple[str, ...]	tuple[str]: Tuple with the names of the raw fields of the measurement.

Functions

clean()

Check consistency of validation, reporting and backs-up values.

Source code in measurement\models.py

def clean(self) -> None:
    """Check consistency of validation, reporting and backs-up values."""
    # Check consistency of validation
    if not self.is_validated and not self.is_active:
        raise ValidationError("Only validated entries can be declared as inactive.")

    # Backup values to raws, if needed
    for r in self.raws:
        value = getattr(self, r.removeprefix("raw_"))
        if value and not getattr(self, r):
            setattr(self, r, value)

Station

Bases: PermissionsBase

Main representation of a station, including several metadata.

Attributes:

Name	Type	Description
visibility	str	Visibility level of the object, including an "internal" option.
station_id	int	Primary key.
station_code	str	Unique code for the station.
station_name	str	Brief description of the station.
station_type	StationType	Type of the station.
country	Country	Country where the station is located.
region	Region	Region within the Country where the station is located.
ecosystem	Ecosystem	Ecosystem associated with the station.
institution	Institution	Institutional partner responsible for the station.
place_basin	PlaceBasin	Place-Basin association.
station_state	bool	Is the station operational?
timezone	str	Timezone of the station.
station_latitude	Decimal	Latitude of the station, in degrees [-90 to 90].
station_longitude	Decimal	Longitude of the station, in degrees [-180 to 180].
station_altitude	int	Altitude of the station.
influence_km	Decimal	Area of influence in km2.
station_file	ImageField	Photography of the station.
station_external	bool	Is the station external?
variables	str	Comma-separated list of variables measured by the station.

Attributes

variables_list: list[str] property

Return the list of variables measured by the station.

Only variables with data in the database are returned.

Returns:

Type	Description
list[str]	list[str]: List of variables measured by the station.

Functions

__str__()

Return the station code.

Source code in station\models.py

def __str__(self) -> str:
    """Return the station code."""
    return str(self.station_code)

get_absolute_url()

Return the absolute url of the station.

Source code in station\models.py

def get_absolute_url(self) -> str:
    """Return the absolute url of the station."""
    return reverse("station:station_detail", kwargs={"pk": self.pk})

set_object_permissions()

Set object-level permissions.

This method is called by the save method of the model to set the object-level permissions based on the visibility level of the object. In addition to the standard permissions for the station, the view_measurements permission is set which controls who can view the measurements associated to the station.

Source code in station\models.py

def set_object_permissions(self) -> None:
    """Set object-level permissions.

    This method is called by the save method of the model to set the object-level
    permissions based on the visibility level of the object. In addition to the
    standard permissions for the station, the view_measurements permission is set
    which controls who can view the measurements associated to the station.
    """
    super().set_object_permissions()

    standard_group = Group.objects.get(name="Standard")
    anonymous_user = get_anonymous_user()

    # Assign view_measurements permission based on permissions level
    if self.visibility == "public":
        assign_perm("view_measurements", standard_group, self)
        assign_perm("view_measurements", anonymous_user, self)
        if self.owner:
            remove_perm("view_measurements", self.owner, self)
    elif self.visibility == "internal":
        assign_perm("view_measurements", standard_group, self)
        remove_perm("view_measurements", anonymous_user, self)
        if self.owner:
            remove_perm("view_measurements", self.owner, self)
    elif self.visibility == "private":
        remove_perm("view_measurements", standard_group, self)
        remove_perm("view_measurements", anonymous_user, self)
        if self.owner:
            assign_perm("view_measurements", self.owner, self)

Variable

Bases: PermissionsBase

A variable with a physical meaning.

Such as precipitation, wind speed, wind direction, soil moisture, including the associated unit. It also includes metadata to help identify what is a reasonable value for the data, to flag outliers and to help with the validation process.

The nature of the variable can be one of the following:

• sum: Cumulative value over a period of time.
• average: Average value over a period of time.
• value: One-off value.

Attributes:

Name	Type	Description
variable_id	AutoField	Primary key.
variable_code	CharField	Code of the variable, eg. airtemperature.
name	CharField	Human-readable name of the variable, eg. Air temperature.
unit	ForeignKey	Unit of the variable.
maximum	DecimalField	Maximum value allowed for the variable.
minimum	DecimalField	Minimum value allowed for the variable.
diff_error	DecimalField	If two sequential values in the time-series data of this variable differ by more than this value, the validation process can mark this with an error flag.
outlier_limit	DecimalField	The statistical deviation for defining outliers, in times the standard deviation (sigma).
null_limit	DecimalField	The max % of null values (missing, caused by e.g. equipment malfunction) allowed for hourly, daily, monthly data. Cumulative values are not deemed trustworthy if the number of missing values in a given period is greater than the null_limit.
nature	CharField	Nature of the variable, eg. if it represents a one-off value, the average over a period of time or the cumulative value over a period

Attributes

is_cumulative: bool property

Return True if the nature of the variable is sum.

Functions

__str__()

Return the string representation of the object.

Source code in variable\models.py

def __str__(self) -> str:
    """Return the string representation of the object."""
    return str(self.name)

clean()

Validate the model fields.

Source code in variable\models.py

def clean(self) -> None:
    """Validate the model fields."""
    if self.maximum < self.minimum:
        raise ValidationError(
            {
                "maximum": "The maximum value must be greater than the minimum "
                "value."
            }
        )
    if not self.variable_code.isidentifier():
        raise ValidationError(
            {
                "variable_code": "The variable code must be a valid Python "
                "identifier. Only letters, numbers and underscores are allowed, and"
                " it cannot start with a number."
            }
        )
    return super().clean()

get_absolute_url()

Get the absolute URL of the object.

Source code in variable\models.py

def get_absolute_url(self) -> str:
    """Get the absolute URL of the object."""
    return reverse("variable:variable_detail", kwargs={"pk": self.pk})

Functions

random_date(start_, end_)

This function will return a random datetime between two datetime objects.

Source code in utilities\benchmarking\benchmark_queries.py

def random_date(start_: datetime, end_: datetime) -> datetime:
    """This function will return a random datetime between two datetime
    objects.
    """
    delta = end_ - start_
    int_delta = (delta.days * 24 * 60 * 60) + delta.seconds
    random_second = random.randrange(int_delta)
    return start_ + timedelta(seconds=random_second)