FlexMeasures / flexmeasures / 37803379655

87%

Build Type

push

github

Committed by web-flow

Commit Message

Let the service record what a data generator returns (#2689)

* Consolidate report computation and persistence into one shared service function

Give reporting one function that computes and persists (compute_and_save_report in flexmeasures/data/services/reporting.py), called by both the synchronous CLI (with persist=False for --dry-run) and run_report_job. The shared step counts persistable values with the existing _count_persistable_values semantics, validates all automation output sensors before any write, and commits once. Adds regression tests for sync persistence, dry runs, background-job persistence, multiple output sensors, refused automation outputs, persistence failures and CLI output/export behaviour.

Signed-off-by: Balogun Feranmi <bferanmi806@gmail.com>

* Roll back failed report persistence and pin guard semantics with tests

compute_and_save_report now rolls back and re-raises if saving or committing fails, so a failed run leaves nothing pending or committed on either path. Failure tests perform a real first-output save and assert emptiness without manual rollback. New tests pin that an empty permitted-output set rejects everything before any save while None disables the guard. Also reword the changelog entry (permission checks guard the worker only) and drop the unused results binding in run_report_job.

Signed-off-by: Balogun Feranmi <bferanmi806@gmail.com>

* Count the beliefs a report saved, not the ones it computed

`_count_persistable_values` leaves out NaN values but still counts beliefs
that `save_to_db` skips as unchanged, so a report rerun on the same inputs
claimed to have saved every value again. Count with `save_to_db_and_count`
(#2644), as plugin automations already do.

A truthful count of 0 would then have fallen into the warning meant for a
report that computed nothing, which blames inputs that do not align on source
and belief time. So the worker now tells the two apart: values computed but all
on record already is an... (continued)

Coverage Stats

94 of 96 new or added lines in 6 files covered. (97.92%)

2 existing lines in 1 file now uncovered.

21254 of 24560 relevant lines covered (86.54%)

0.87 hits per line