REGISTER NOW: Brightway Spring School! 🇨🇦 Québec City (Canada) 🗓️ 20.-24. May 2024


Module Contents#



A base class for SQLite backends.


tqdm_wrapper(iterable, is_test)



class bw2data.backends.base.SQLiteBackend(*args, **kwargs)[source]#

Bases: bw2data.data_store.ProcessedDataStore

Inheritance diagram of bw2data.backends.base.SQLiteBackend

A base class for SQLite backends.

Subclasses must support at least the following calls:

  • load()

  • write(data)

In addition, they should specify their backend with the backend attribute (a unicode string).

  • rename

  • copy

  • find_dependents

  • random

  • process

For new classes to be recognized by the DatabaseChooser, they need to be registered with the config object, e.g.:

config.backends['backend type string'] = BackendClass

Instantiation does not load any data. If this database is not yet registered in the metadata store, a warning is written to stdout.

The data schema for databases in voluptuous is:

exchange = {
        Required("input"): valid_tuple,
        Required("type"): basestring,
lci_dataset = {
    Optional("categories"): Any(list, tuple),
    Optional("location"): object,
    Optional("unit"): basestring,
    Optional("name"): basestring,
    Optional("type"): basestring,
    Optional("exchanges"): [exchange]
db_validator = Schema({valid_tuple: lci_dataset}, extra=True)
  • valid_tuple is a dataset identifier, like ("ecoinvent", "super strong steel")

  • uncertainty_fields are fields from an uncertainty dictionary.

Processing a Database actually produces two parameter arrays: one for the exchanges, which make up the technosphere and biosphere matrices, and a geomapping array which links activities to locations.


name (unicode string): Name of the database to manage.

property _searchable[source]#
backend = 'sqlite'[source]#
_efficient_write_dataset(index, key, ds, exchanges, activities)[source]#
_efficient_write_many_data(data, indices=True)[source]#
_get_queryset(random=False, filters=True)[source]#

Make a copy of the database.

Internal links within the database will be updated to match the new database name, i.e. ("old name", "some id") will be converted to ("new name", "some id") for all exchanges.

  • name (str): Name of the new database. Must not already exist.

delete(keep_params=False, warn=True, vacuum=True)[source]#

Delete all data from SQLite database and Whoosh index

delete_duplicate_exchanges(fields=['amount', 'type'])[source]#

Delete exchanges which are exact duplicates. Useful if you accidentally ran your input data notebook twice.

To determine uniqueness, we look at the exchange input and output nodes, and at the exchanges values for fields fields.

edges_to_dataframe(categorical: bool = True, formatters: List[Callable] | None = None) pandas.DataFrame[source]#

Return a pandas DataFrame with all database exchanges. Standard DataFrame columns are:

target_id: int, target_database: str, target_code: str, target_name: Optional[str], target_reference_product: Optional[str], target_location: Optional[str], target_unit: Optional[str], target_type: Optional[str] source_id: int, source_database: str, source_code: str, source_name: Optional[str], source_product: Optional[str], # Note different label source_location: Optional[str], source_unit: Optional[str], source_categories: Optional[str] # Tuple concatenated with “::” as in bw2io edge_amount: float, edge_type: str,

Target is the node consuming the edge, source is the node or flow being consumed. The terms target and source were chosen because they also work well for biosphere edges.


categorical will turn each string column in a pandas Categorical Series. This takes 1-2 extra seconds, but saves around 50% of the memory consumption.

formatters is a list of callables that modify each row. These functions must take the following keyword arguments, and use the Wurst internal data format:

  • node: The target node, as a dict

  • edge: The edge, including attributes of the source node

  • row: The current row dict being modified.

The functions in formatters don’t need to return anything, they modify row in place.

Returns a pandas DataFrame.

exchange_data_iterator(sql, dependents, flip=False)[source]#

Iterate over exchanges and format for bw_processing arrays.

dependents is a set of dependent database names.

flip means flip the numeric sign; see bw_processing docs.

Uses raw sqlite3 to retrieve data for ~2x speed boost.

abstract filepath_intermediate()[source]#
find_dependents(data=None, ignore=None)[source]#

Get sorted list of direct dependent databases (databases linked from exchanges).

  • data (dict, optional): Inventory data

  • ignore (list): List of database names to ignore


List of database names


Recursively get list of all dependent databases.


A set of database names

get(code=None, **kwargs)[source]#
graph_technosphere(filename=None, **kwargs)[source]#
load(*args, **kwargs)[source]#

Load the intermediate data for this object.


The intermediate data.

new_activity(code, **kwargs)[source]#
new_node(code, **kwargs)[source]#
nodes_to_dataframe(columns: List[str] | None = None, return_sorted: bool = True) pandas.DataFrame[source]#

Return a pandas DataFrame with all database nodes. Uses the provided node attributes by default, such as name, unit, location.

By default, returns a DataFrame sorted by name, reference product, location, and unit. Set return_sorted to False to skip sorting.

Specify columns to get custom columns. You will need to write your own function to get more customization, there are endless possibilities here.

Returns a pandas DataFrame.


Create structured arrays for the technosphere and biosphere matrices.

Uses bw_processing for array creation and metadata serialization.

Also creates a geomapping array, linking activities to locations. Used for regionalized calculations.

Use a raw SQLite3 cursor instead of Peewee for a ~2 times speed advantage.


Search through the database.

random(filters=True, true_random=False)[source]#

True random requires loading and sorting data in SQLite, and can be resource-intensive.

register(write_empty=True, **kwargs)[source]#

Register a database with the metadata store.

Databases must be registered before data can be written.

Writing data automatically sets the following metadata:
  • depends: Names of the databases that this database references, e.g. “biosphere”

  • number: Number of processes in this database.

  • format (str, optional): Format that the database was converted from, e.g. “Ecospold”

relabel_data(data, new_name)[source]#

Relabel database keys and exchanges.

In a database which internally refer to the same database, update to new database name new_name.

Needed to copy a database completely or cut out a section of a database.

For example:

data = {
    ("old and boring", 1):
        {"exchanges": [
            {"input": ("old and boring", 42),
            "amount": 1.0},
    ("old and boring", 2):
        {"exchanges": [
            {"input": ("old and boring", 1),
            "amount": 4.0}
print(relabel_database(data, "shiny new"))
>> {
    ("shiny new", 1):
        {"exchanges": [
            {"input": ("old and boring", 42),
            "amount": 1.0},
    ("shiny new", 2):
        {"exchanges": [
            {"input": ("shiny new", 1),
            "amount": 4.0}

In the example, the exchange to ("old and boring", 42) does not change, as this is not part of the updated data.

  • data (dict): The data to modify

  • new_name (str): The name of the modified database


The modified data


Rename a database. Modifies exchanges to link to new name. Deregisters old database.

  • name (str): New name.


New Database object.

search(string, **kwargs)[source]#

Search this database for string.

The searcher include the following fields:

  • name

  • comment

  • categories

  • location

  • reference product

string can include wild cards, e.g. "trans*".

By default, the name field is given the most weight. The full weighting set is called the boost dictionary, and the default weights are:

    "name": 5,
    "comment": 1,
    "product": 3,
    "categories": 2,
    "location": 3

Optional keyword arguments:

  • limit: Number of results to return.

  • boosts: Dictionary of field names and numeric boosts - see default boost values above. New values must be in the same format, but with different weights.

  • filter: Dictionary of criteria that search results must meet, e.g. {'categories': 'air'}. Keys must be one of the above fields.

  • mask: Dictionary of criteria that exclude search results. Same format as filter.

  • facet: Field to facet results. Must be one of name, product, categories, location, or database.

  • proxy: Return Activity proxies instead of raw Whoosh documents. Default is True.

Returns a list of Activity datasets.


Set geocollections attribute for databases which don’t currently have it.

write(data, process=True, searchable=True)[source]#

Write data to database.

data must be a dictionary of the form:

    ('database name', 'dataset code'): {dataset}

Writing a database will first deletes all existing data.

bw2data.backends.base.tqdm_wrapper(iterable, is_test)[source]#