autofit.Aggregator#

class autofit.Aggregator(session: ~sqlalchemy.orm.session.Session, filename: ~typing.Optional[str] = None, predicate: ~autofit.database.query.query.abstract.AbstractQuery = <NullPredicate SELECT id FROM fit>, offset=0, limit=None, order_bys=None, top_level_only=True)[source]#

Bases: AbstractAggregator

Query results from an intermediary SQLite database.

Results can be scraped from a directory structure and stored in the database.

Parameters:

session – A session for communicating with the database.
filename – The path to the database file. If None, the database is in memory.
predicate – A predicate to filter the results by.
offset – The number of results to skip
limit – The maximum number of results to return
order_bys – A list of attributes to order the results by
top_level_only – If True, only return the top level fits

Methods

`add_directory`	Recursively search a directory for autofit results and add them to this database.
`child_values`	Retrieve the value associated with each fit with the given parameter name
`from_database`	Create an instance from a sqlite database file.
`grid_searches`	Filter to only grid searches and return an aggregator with grid search specific functionality.
`map`
`order_by`	Order the results by a given attribute of the search.
`query`	Apply a query on the model.
`values`	Retrieve the value associated with each fit with the given parameter name

Attributes

`fits`	Lazily query the database for a list of Fit objects that match the aggregator's predicate.
`info`	Query info associated with the fit in the info dictionary
`model`	Facilitates query construction.
`search`	An object facilitating queries on fit attributes such as: name unique_tag path_prefix is_complete is_grid_search

order_by(item: Attribute, reverse=False) → Aggregator[source]#

Order the results by a given attribute of the search. Can be applied multiple times with the first application taking precedence.

Parameters:

item – An attribute of the search
reverse – If True reverse the results

Return type:

An aggregator with ordering applied

Examples

aggregator = aggregator.order_by(: aggregator.search.unique_tag

)

property search: FitQuery#: An object facilitating queries on fit attributes such as: name unique_tag path_prefix is_complete is_grid_search

property info#: Query info associated with the fit in the info dictionary

property fits: List[Fit]#: Lazily query the database for a list of Fit objects that match the aggregator’s predicate.

property model: Query#

Facilitates query construction. If the Fit class has an attribute with the given name then a predicate is generated based on that attribute. Otherwise the query is assumed to apply to the best fit instance.

Return type:: A query

__call__(predicate) → Aggregator[source]#: Concise query syntax

query(predicate: AbstractQuery) → Aggregator[source]#

Apply a query on the model.

Parameters:: predicate – A predicate constructed to express which models should be included.
Return type:: A list of objects that match the predicate

Examples

>>>
>>> aggregator = Aggregator.from_database(
>>>     "my_database.sqlite"
>>> )
>>>
>>> lens = aggregator.galaxies.lens
>>>
>>> aggregator.filter((lens.bulge == SersicCore) & (lens.disk == Sersic))
>>> aggregator.filter((lens.bulge == SersicCore) | (lens.disk == Sersic))

add_directory(directory: str, auto_commit=True, reference: Optional[dict] = None, completed_only: bool = False)[source]#

Recursively search a directory for autofit results and add them to this database.

Any pickles found in the pickles file are implicitly added to the fit object.

Warning

If a directory is added twice then that will result in duplicate entries in the database.

Parameters:

auto_commit – If True the session is committed writing the new objects to the database
directory – A directory containing autofit results embedded in a file structure
reference – A dictionary mapping the names of objects in the model to their class path.
completed_only – If true only searches that have completed are added

classmethod from_database(filename: str, completed_only: bool = False, top_level_only: bool = True) → Aggregator[source]#

Create an instance from a sqlite database file.

If no file exists then one is created with the schema of the database.

Parameters:

completed_only – If True only completed fits are returned
filename – The name of the database file.
top_level_only – If True only top level fits are returned

Return type:

An aggregator connected to the database specified by the file.

grid_searches() → GridSearchAggregator[source]#

Filter to only grid searches and return an aggregator with grid search specific functionality.

Grid searches are initially implicitly ordered by their id