def app():
app = create_app(debug=True)
app.testing = True
app.config["SQLALCHEMY_TRACK_MODIFICATIONS"] = True
with app.app_context():
#db.create_all()
yield app
def app(G):
app = create_app(debug=True)
app.testing = True
app.config["SQLALCHEMY_DATABASE_URI"] = "sqlite:////tmp/test.db"
app.config["SQLALCHEMY_TRACK_MODIFICATIONS"] = True
with app.app_context():
G.to_sql(app)
yield app
db.drop_all()
def to_sql(self, app=None, last_known_value_date: Optional[date] = None):
""" Inserts the model into the SQLite3 database associated with Delphi,
for use with the ICM REST API. """
from delphi.apps.rest_api import create_app, db
self.assign_uuids_to_nodes_and_edges()
icm_metadata = ICMMetadata(
id=self.id,
created=self.dateCreated,
estimatedNumberOfPrimitives=len(self.nodes) + len(self.edges),
createdByUser_id=1,
lastAccessedByUser_id=1,
lastUpdatedByUser_id=1,
)
if last_known_value_date is None:
today = date.today().isoformat()
else:
today = last_known_value_date.isoformat()
default_latent_var_value = 1.0
causal_primitives = []
nodeset = {n.split("/")[-1] for n in self.nodes}
simplified_labels = len(nodeset) == len(self)
for n in self.nodes(data=True):
n[1]["rv"] = LatentVar(n[0])
n[1]["update_function"] = self.default_update_function
rv = n[1]["rv"]
rv.dataset = [default_latent_var_value for _ in range(self.res)]
causal_variable = CausalVariable(
id=n[1]["id"],
model_id=self.id,
units="",
namespaces={},
auxiliaryProperties=[],
label=n[0].split("/")[-1].replace("_", " ").capitalize()
if simplified_labels else n[0],
description=n[0],
lastUpdated=today,
confidence=1.0,
lastKnownValue={
"active": "ACTIVE",
"trend": None,
"time": today,
"value": {
"baseType": "FloatValue",
"value": n[1]["rv"].dataset[0],
},
},
range={
"baseType": "FloatRange",
"range": {
"min": 0,
"max": 5,
"step": 0.1
},
},
)
causal_primitives.append(causal_variable)
max_mean_betas = max(
[abs(np.median(e[2]["βs"])) for e in self.edges(data=True)])
for e in self.edges(data=True):
causal_relationship = CausalRelationship(
id=e[2]["id"],
namespaces={},
source={
"id": self.nodes[e[0]]["id"],
"baseType": "CausalVariable",
},
target={
"id": self.nodes[e[1]]["id"],
"baseType": "CausalVariable",
},
model_id=self.id,
auxiliaryProperties=[],
lastUpdated=today,
types=["causal"],
description="",
confidence=np.mean(
[s.belief for s in e[2]["InfluenceStatements"]]),
label="",
strength=abs(np.median(e[2]["βs"]) / max_mean_betas),
reinforcement=(True if np.mean([
stmt.subj.delta.polarity * stmt.obj.delta.polarity
for stmt in e[2]["InfluenceStatements"]
]) > 0 else False),
)
causal_primitives.append(causal_relationship)
evidences = []
for edge in self.edges(data=True):
for stmt in edge[2]["InfluenceStatements"]:
for ev in stmt.evidence:
evidence = Evidence(
id=str(uuid4()),
causalrelationship_id=edge[2]["id"],
# TODO - make source and target appear in CauseEx HMI
description=(ev.text),
)
evidences.append(evidence)
if app is None:
app = create_app()
# Tag this model for displaying in the CauseWorks interface
self.tag_for_CX = True
with app.app_context():
db.create_all()
db.session.add(icm_metadata)
db.session.add(DelphiModel(id=self.id, model=self))
for causal_primitive in causal_primitives:
db.session.add(causal_primitive)
for evidence in evidences:
db.session.add(evidence)
db.session.commit()
def main():
app = create_app(debug=True)
app.config["SQLALCHEMY_DATABASE_URI"] = f"sqlite:///{db_path}"
app.run(host="127.0.0.1", port=5000)
def write_model_to_database(G):
G.assemble_transition_model_from_gradable_adjectives()
delphi_model = DelphiModel(id=G.id, model=G)
icm_metadata = ICMMetadata(
id=G.id,
created=date.today().isoformat(),
estimatedNumberOfPrimitives=len(G.nodes) + len(G.edges),
createdByUser_id=1,
lastAccessedByUser_id=1,
lastUpdatedByUser_id=1,
)
today = date.today().isoformat()
default_latent_var_value = 1.0
causal_primitives = []
for n in G.nodes(data=True):
n[1]["rv"] = LatentVar(n[0])
n[1]["update_function"] = G.default_update_function
rv = n[1]["rv"]
rv.dataset = [default_latent_var_value for _ in range(G.res)]
causal_variable = CausalVariable(
id=n[1]["id"],
model_id=G.id,
units="",
namespaces={},
auxiliaryProperties=[],
label=n[0],
description=f"Long description of {n[0]}.",
lastUpdated=today,
confidence=1.0,
lastKnownValue={
"active": "ACTIVE",
"trend": None,
"time": today,
"value": {
"baseType": "FloatValue",
"value": n[1]["rv"].dataset[0],
},
},
range={
"baseType": "FloatRange",
"range": {
"min": -2,
"max": 2,
"step": 0.1
},
},
)
causal_primitives.append(causal_variable)
max_evidences = max([
sum([len(s.evidence) for s in e[2]["InfluenceStatements"]])
for e in G.edges(data=True)
])
max_median_betas = max(
[abs(np.median(e[2]["betas"])) for e in G.edges(data=True)])
evidences = []
for e in G.edges(data=True):
causal_relationship = CausalRelationship(
id=e[2]["id"],
namespaces={},
source={
"id": G.nodes[e[0]]["id"],
"baseType": "CausalVariable"
},
target={
"id": G.nodes[e[1]]["id"],
"baseType": "CausalVariable"
},
model_id=G.id,
auxiliaryProperties=[],
lastUpdated=today,
types=["causal"],
description=f"{e[0]} influences {e[1]}.",
confidence=np.mean([s.belief
for s in e[2]["InfluenceStatements"]]),
label="influences",
strength=abs(np.median(e[2]["betas"]) / max_median_betas),
reinforcement=(True if np.median([
stmt.subj_delta["polarity"] * stmt.obj_delta["polarity"]
for stmt in e[2]["InfluenceStatements"]
]) > 0 else False),
)
causal_primitives.append(causal_relationship)
for stmt in e[2]["InfluenceStatements"]:
for ev in stmt.evidence:
evidence = Evidence(
id=str(uuid4()),
causalrelationship_id=e[2]["id"],
description=ev.text,
)
evidences.append(evidence)
app = create_app()
app.testing = True
app.config["SQLALCHEMY_DATABASE_URI"] = "sqlite:///delphi.db"
app.config["SQLALCHEMY_TRACK_MODIFICATIONS"] = True
with app.app_context():
db.create_all()
db.session.add(icm_metadata)
db.session.add(delphi_model)
for causal_primitive in causal_primitives:
db.session.add(causal_primitive)
for evidence in evidences:
db.session.add(evidence)
db.session.commit()