Ev visibility

[irmatgit]

add ev as a end-use for the stride-viewer

[Copilot]

Pull request overview

This PR expands STRIDE’s projection/UI support by separating EV charging into its own end-use metric, improves scenario override dataset loading, and adds a new “custom demand component” capability that injects user-defined loads into the hourly energy projection.

Changes:

• Tag EV charging energy as a distinct end-use (metric='ev_charging') in Transportation when EV projection is enabled (dbt + test implementation).
• Add scenario-specific dataset substitution in make_mapped_datasets to correctly load scenario overrides registered as {scenario}__{table}.
• Introduce configurable custom demand components (new model/config fields, injection into energy_projection, and CLI commands), plus UI/CLI and regression test updates.

Reviewed changes

Copilot reviewed 12 out of 13 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
tests/test_project.py	Adds tests for recent-project registration and scenario dataset override substitution.
tests/test_energy_projection.py	Updates test-side projection logic to tag EV charging with a dedicated metric.
tests/test_custom_demand.py	Adds a new test suite covering custom demand component behaviors and CLI lifecycle.
tests/test_app_cache.py	Adds regression test ensuring Settings save doesn’t set the CLI override.
tests/test_api.py	Relaxes assertions to allow zero values (non-negative) for annual metrics.
src/stride/ui/settings/callbacks.py	Stops Settings save from setting the runtime max-projects override; persists config + evicts only.
src/stride/ui/app.py	Allows passing dynamic sector lists into fresh color manager creation.
src/stride/project.py	Injects custom demand rows into scenario energy projections (materializes view to table first).
src/stride/models.py	Adds CustomDemandComponent, scenario-level overrides, and project-level component configuration.
src/stride/dsgrid_integration.py	Substitutes scenario-specific dataset IDs when present to fix override loading.
src/stride/dbt/models/energy_projection_com_ind_tra_load_shapes.sql	Adds energy_source tagging and emits EV charging as ev_charging metric.
src/stride/cli/stride.py	Registers newly created projects as “recent” and adds custom-demand CLI commands.
.gitignore	Adds ignores for macOS metadata and equinor*.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

PR description mentions only “EV visibility / add ev as an end-use”, but this PR also introduces a full custom-demand component system + CLI commands and settings behavior changes. If these are intended, consider updating the PR description (or splitting into separate PRs) to make review/rollback safer and scope clearer.

[Copilot]

custom_demand_overrides uses Field(default={}), which is a shared mutable default and can leak overrides across Scenario instances. Use default_factory=dict instead (and similarly ensure values are copied when mutating).

[Copilot]

custom_demand_components is declared with Field(default=[]), which is a shared mutable default and can leak components across ProjectConfig instances. Use default_factory=list (and consider aligning other list defaults in this model similarly).

[Copilot]

The DELETE statement interpolates component.sector/component.metric directly into SQL. This will break for values containing quotes and is also SQL-injection prone since these fields come from user config/CLI. Use parameterized queries (DuckDB ? placeholders) rather than f-string interpolation here.

[Copilot]

Deleting by only (sector, metric) can wipe out baseline projection rows if a user chooses an existing sector/metric (e.g., Residential/heating), turning an “additive” component into destructive overwrite. Consider preventing collisions (validate against existing sector/metric pairs) or tagging injected rows with a component identifier so DELETE targets only the injected component.

Suggested change

	# Defensive DELETE: remove any existing rows for this custom component
	self._con.sql(
	f"DELETE FROM {scenario.name}.energy_projection "
	f"WHERE sector = '{component.sector}'"
	f" AND metric = '{component.metric}'"
	)
	# Prevent destructive overwrite of baseline rows by rejecting
	# custom components that collide with an existing sector/metric pair.
	existing_count = self._con.sql(
	f"SELECT COUNT(*) FROM {scenario.name}.energy_projection "
	f"WHERE sector = '{component.sector}'"
	f" AND metric = '{component.metric}'"
	).fetchone()[0]
	if existing_count > 0:
	msg = (
	f"Custom demand component '{component.name}' uses sector "
	f"'{component.sector}' and metric '{component.metric}', "
	f"which already exist in scenario '{scenario.name}' "
	f"energy_projection data. Choose a unique sector/metric "
	f"pair for additive custom demand components."
	)
	raise InvalidParameter(msg)

[Copilot]

Flat profile injection divides by a hard-coded 8760 hours, but energy_projection timestamps can be 8784 for leap weather years (and could differ for other calendars). This will cause annual totals to be wrong. Compute the per-year hour count from hourly_timestamps and divide by that instead.

[Copilot]

File-profile injection hard-requires exactly 8760 rows, but energy_projection can have a different number of hourly timestamps (e.g., 8784 for leap years). Validate the profile length against the per-year timestamp count you will join to (and/or support 8784), otherwise joins will silently drop/mismatch hours.

Suggested change

	if row_count != 8760:
	msg = (
	f"Custom profile file must have exactly 8760 rows "
	f"(got {row_count})"
	)
	raise InvalidParameter(msg)
	timestamp_counts = self._con.sql(
	f"""
	SELECT model_year, COUNT(DISTINCT timestamp) AS timestamp_count
	FROM {scenario_name}.energy_projection
	WHERE model_year IN ({model_years_str})
	GROUP BY model_year
	ORDER BY model_year
	"""
	).fetchall()
	if not timestamp_counts:
	msg = (
	"No hourly timestamps found in energy_projection for the "
	f"selected model years: {model_years_str}"
	)
	raise InvalidParameter(msg)
	expected_counts = {count for _, count in timestamp_counts}
	if len(expected_counts) != 1:
	counts_by_year = ", ".join(
	f"{model_year}: {count}" for model_year, count in timestamp_counts
	)
	msg = (
	"Custom profile file cannot be applied because the selected "
	"model years do not all have the same number of hourly "
	f"timestamps in energy_projection. Found: {counts_by_year}"
	)
	raise InvalidParameter(msg)
	expected_row_count = next(iter(expected_counts))
	if row_count != expected_row_count:
	msg = (
	"Custom profile file must have exactly one row per hourly "
	"timestamp in energy_projection for each selected model year "
	f"(expected {expected_row_count}, got {row_count})"
	)
	raise InvalidParameter(msg)

[Copilot]

_validate_load_shape_reference interpolates ref_value directly into SQL. A reference containing quotes will break the query, and since load_profile is user-provided this is also injection-prone. Use parameter binding for ref_value (and avoid string formatting values into SQL).

Suggested change

	count = self._con.sql(
	f"SELECT COUNT(*) FROM {scenario_name}.load_shapes_expanded "
	f"WHERE {dimension} = '{ref_value}' "
	f"AND model_year IN ({model_years_str})"
	count = self._con.execute(
	f"SELECT COUNT(*) FROM {scenario_name}.load_shapes_expanded "
	f"WHERE {dimension} = ? "
	f"AND model_year IN ({model_years_str})",
	[ref_value],

[irmatgit]

PR on the wrong repo..