Week from Dec 14 to Dec 20

Introduction

During the twentieth week I finished working on the new implementation of the Storer class. The online repository was moved to the address “https://github.com/opencitations/oc_ocdm”, since this will be its new official location on the web. Other little changes to the codebase are documented below.

Report

New Storer implementation

The new Storer implementation is now ready to be used by oc_ocdm users. It doesn’t work anymore on a per-graph basis, as it now reasons upon GraphEntity instances. This means that now the Storer can correctly handle all the situations that may occur while working with [OCDM][ocdm-2.0.1]-compliant data: an entity can be created out of nothing, it can be removed or it can be updated (in overwrite or append mode), while before it could only be created or updated in append mode.

The entities can be stored singularly or as a result of a Storer operation applied to an entire GraphSet. In the case of entities which have to be updated, the modality of this operation (overwrite/append) is choosen based of the presence of a non-empty preexisting graph inside the entity. This means that:

• if the user wants to update an existing entity by adding/subtracting triples from it, then he/she is required to import the actual graph of that same entity and to pass it as an argument for the correct add_* method of GraphSet;
• if the user simply wants to add new triples to an entity (which can also be non existent), then he/she must call the correct add_* method of GraphSet without passing the preexistent graph as an input (even if he/she already possesses that rdflib.Graph instance).

The methods get_delete_query, get_insert_query and get_update_query of the ProvSet class were moved into a new query_utils.py module because they are needed also by the new Storer implementation.

Various changes

I added the LICENSE.md file to the project: the license choosen for this OpenCitations project is the ISC license. The pyproject.toml file was modified with updated metadata regarding the project.

In GraphEntity and GraphSet, the method apply_changes was renamed to commit_changes, since this new name better conveys its semantics (it should be called at the end of all the edits made on an entity so that new changes can be made, which is very similar to the “commit” operation of transactional languages such as SQL).