def test_active_samples_on_flow_cell(
active_flow_cell_store: Store,
flowcell_name: str,
active_wipe_demultiplex_api: DeleteDemuxAPI,
):
"""Test if the function to find active samples works correctly"""
# GIVEN a flowcell with active samples related to it
store_: Store = active_flow_cell_store
samples_on_flow_cell: List[Sample] = (store_.query(Flowcell).filter(
Flowcell.name == flowcell_name).first().samples)
assert samples_on_flow_cell
for sample in samples_on_flow_cell:
active: bool = store_.active_sample(internal_id=sample.internal_id)
assert active
# WHEN checking for active samples on flowcell
active_wipe_demultiplex_api._set_samples_on_flow_cell()
active_samples_on_flow_cell: Optional[
List[str]] = active_wipe_demultiplex_api.active_samples_on_flow_cell()
# THEN there should be active samples found
assert any(sample.internal_id in active_samples_on_flow_cell
for sample in samples_on_flow_cell)
def demultiplex_flowcell(
context: CGConfig,
dry_run: bool,
flowcell_id: str,
bcl_converter: str,
):
"""Demultiplex a flowcell on slurm using CG
flowcell-id is the flowcell run directory name, e.g. '201203_A00689_0200_AHVKJCDRXX'
"""
LOG.info("Running cg demultiplex flowcell, using %s.", bcl_converter)
flowcell_directory: Path = Path(context.demultiplex.run_dir) / flowcell_id
demultiplex_api: DemultiplexingAPI = context.demultiplex_api
demultiplex_api.set_dry_run(dry_run=dry_run)
LOG.info(f"SETTING FLOWCELL ID TO {flowcell_id}")
LOG.info(f"SETTING OUT DIR TO {demultiplex_api.out_dir}")
try:
flowcell_obj = Flowcell(flowcell_path=flowcell_directory,
bcl_converter=bcl_converter)
except FlowcellError as e:
raise click.Abort from e
delete_demux_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=context,
demultiplex_base=demultiplex_api.out_dir,
dry_run=dry_run,
run_path=flowcell_directory,
)
delete_demux_api.delete_flow_cell(
cg_stats=True,
demultiplexing_dir=True,
run_dir=False,
housekeeper=True,
init_files=True,
status_db=False,
)
if not demultiplex_api.is_demultiplexing_possible(
flowcell=flowcell_obj) and not dry_run:
LOG.warning("Can not start demultiplexing!")
return
if not flowcell_obj.validate_sample_sheet():
LOG.warning(
"Malformed sample sheet. Run cg demultiplex samplesheet validate %s",
flowcell_obj.sample_sheet_path,
)
raise click.Abort
slurm_job_id: int = demultiplex_api.start_demultiplexing(
flowcell=flowcell_obj)
tb_api: TrailblazerAPI = context.trailblazer_api
demultiplex_api.add_to_trailblazer(tb_api=tb_api,
slurm_job_id=slurm_job_id,
flowcell=flowcell_obj)
def demultiplex_all(context: CGConfig, bcl_converter: str,
flowcells_directory: click.Path, dry_run: bool):
"""Demultiplex all flowcells that are ready under the flowcells_directory"""
LOG.info("Running cg demultiplex all, using %s.", bcl_converter)
if flowcells_directory:
flowcells_directory: Path = Path(str(flowcells_directory))
else:
flowcells_directory: Path = Path(context.demultiplex.run_dir)
demultiplex_api: DemultiplexingAPI = context.demultiplex_api
demultiplex_api.set_dry_run(dry_run=dry_run)
tb_api: TrailblazerAPI = context.trailblazer_api
LOG.info("Search for flowcells ready to demultiplex in %s",
flowcells_directory)
for sub_dir in flowcells_directory.iterdir():
if not sub_dir.is_dir():
continue
LOG.info("Found directory %s", sub_dir)
try:
flowcell_obj = Flowcell(flowcell_path=sub_dir,
bcl_converter=bcl_converter)
except FlowcellError:
continue
if not demultiplex_api.is_demultiplexing_possible(
flowcell=flowcell_obj) and not dry_run:
continue
if not flowcell_obj.validate_sample_sheet():
LOG.warning(
"Malformed sample sheet. Run cg demultiplex samplesheet validate %s",
flowcell_obj.sample_sheet_path,
)
continue
delete_demux_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=context,
demultiplex_base=demultiplex_api.out_dir,
dry_run=dry_run,
run_path=(flowcells_directory / sub_dir),
)
delete_demux_api.delete_flow_cell(
cg_stats=False,
demultiplexing_dir=True,
run_dir=False,
housekeeper=True,
init_files=False,
status_db=False,
)
slurm_job_id: int = demultiplex_api.start_demultiplexing(
flowcell=flowcell_obj)
demultiplex_api.add_to_trailblazer(tb_api=tb_api,
slurm_job_id=slurm_job_id,
flowcell=flowcell_obj)
def test_no_active_samples_on_flow_cell(
populated_wipe_demultiplex_api: DeleteDemuxAPI, flowcell_name: str):
"""Test if the function to find no active samples works correctly"""
# GIVEN a flowcell with no active samples related to it
store_: Store = populated_wipe_demultiplex_api.status_db
samples_on_flow_cell: List[Sample] = (store_.query(Flowcell).filter(
Flowcell.name == flowcell_name).first().samples)
assert samples_on_flow_cell
for sample in samples_on_flow_cell:
active: bool = store_.active_sample(internal_id=sample.internal_id)
assert not active
# WHEN checking for active samples on flowcell
populated_wipe_demultiplex_api._set_samples_on_flow_cell()
active_samples_on_flow_cell: Optional[List[
str]] = populated_wipe_demultiplex_api.active_samples_on_flow_cell()
# THEN the no samples on the flowcell should be found active
assert not active_samples_on_flow_cell
def fixture_active_wipe_demultiplex_api(
active_wipe_demux_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flowcell_full_name: str,
) -> DeleteDemuxAPI:
"""Yield an instantiated DeleteDemuxAPI with active samples on a flowcell"""
return DeleteDemuxAPI(
config=active_wipe_demux_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=False,
run_path=Path(flowcell_full_name),
)
def fixture_populated_wipe_demultiplex_api(
populated_wipe_demux_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
tmp_flow_cell_run_path: Path,
) -> DeleteDemuxAPI:
"""Yield an initialized populated DeleteDemuxAPI"""
return DeleteDemuxAPI(
config=populated_wipe_demux_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=False,
run_path=tmp_flow_cell_run_path,
)
def fixture_wipe_demultiplex_api(
cg_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flowcell_full_name: str,
stats_api: StatsAPI,
) -> DeleteDemuxAPI:
"""Yield an initialized DeleteDemuxAPI"""
cg_context.cg_stats_api_ = stats_api
return DeleteDemuxAPI(
config=cg_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=False,
run_path=Path(flowcell_full_name),
)
def test_delete_flow_cell_housekeeper_only_sample_level(
caplog,
cg_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flowcell_full_name: str,
populated_flow_cell_store: Store,
sample_level_housekeeper_api: HousekeeperAPI,
tmp_fastq_paths: List[Path],
):
"""Test function to remove fastqs from Housekeeper when there are only files on sample level
(not on flow cell name)
"""
caplog.set_level(logging.INFO)
cg_context.housekeeper_api_ = sample_level_housekeeper_api
cg_context.status_db_ = populated_flow_cell_store
# GIVEN a DeleteDemuxAPI with a HousekeeperAPI with no files with flow cell name as a tag
sample_level_files: List[Path] = tmp_fastq_paths
wipe_demultiplex_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=cg_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=False,
run_path=Path(flowcell_full_name),
)
wipe_demultiplex_api._set_samples_on_flow_cell()
# WHEN wiping files in Housekeeper
wipe_demultiplex_api.delete_flow_cell_housekeeper()
# THEN you should be notified that there are no files on flow cell names
assert (
f"Housekeeper: No files found with tag: {wipe_demultiplex_api.flow_cell_name}"
in caplog.text)
# AND you should be notified that there were fastq files removed on sample level
for file in sample_level_files:
assert f"{file.as_posix()} deleted" in caplog.text
def test_delete_flow_cell_housekeeper_flowcell_name(
caplog,
cg_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flow_cell_name_housekeeper_api: HousekeeperAPI,
flowcell_full_name: str,
populated_flow_cell_store: Store,
tmp_fastq_paths: List[Path],
tmp_sample_sheet_path: Path,
):
"""Test function to remove files from Housekeeper using flow cell name as a tag"""
caplog.set_level(logging.INFO)
cg_context.housekeeper_api_ = flow_cell_name_housekeeper_api
cg_context.status_db_ = populated_flow_cell_store
# GIVEN
fastq_files: List[Path] = tmp_fastq_paths
sample_sheet_file: Path = tmp_sample_sheet_path
wipe_demultiplex_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=cg_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=False,
run_path=Path(flowcell_full_name),
)
wipe_demultiplex_api._set_samples_on_flow_cell()
# WHEN
wipe_demultiplex_api.delete_flow_cell_housekeeper()
# THEN
assert (
f"Housekeeper: No files found with tag: {wipe_demultiplex_api.flow_cell_name}"
not in caplog.text)
assert f"Deleted {sample_sheet_file.as_posix()} from housekeeper" in caplog.text
for fastq_file in fastq_files:
assert f"{fastq_file.as_posix()} deleted" in caplog.text
def test_set_dry_run_delete_demux_api(
caplog,
cg_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flowcell_full_name: str,
stats_api: StatsAPI,
):
"""Test to test function to set the API to run in dry run mode"""
caplog.set_level(logging.DEBUG)
cg_context.cg_stats_api_ = stats_api
# WHEN setting the dry_run mode on a DeleteDemuxAPI
wipe_demultiplex_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=cg_context,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=True,
run_path=flowcell_full_name,
)
# THEN the dry run parameter should be set to True and it should be logged
assert wipe_demultiplex_api.dry_run
assert f"DeleteDemuxAPI: Setting dry run mode to True" in caplog.text
def test_initiate_delete_demux_api(
caplog,
cg_context: CGConfig,
demultiplexed_flowcells_working_directory: Path,
flowcell_full_name: str,
):
"""Test to initialize the DeleteDemuxAPI"""
caplog.set_level(logging.DEBUG)
# GIVEN a correct config
config = cg_context
# WHEN initializing the DeleteDemuxAPI
DeleteDemuxAPI(
config=config,
demultiplex_base=demultiplexed_flowcells_working_directory,
dry_run=True,
run_path=flowcell_full_name,
)
# THEN the API should be correctly initialized
assert "DeleteDemuxAPI: API initiated" in caplog.text
def delete_flow_cell(
context: CGConfig,
dry_run: bool,
demultiplexing_dir: str,
cg_stats: bool,
demultiplex_base: bool,
housekeeper: bool,
init_files: bool,
run_dir: bool,
run_directory: str,
status_db: bool,
yes: bool,
):
"""Delete a flowcell. If --status-db is passed then all other options will be treated as True"""
demultiplex_base: Path = Path(demultiplex_base)
run_path: Path = Path(run_directory)
wipe_demux_api: DeleteDemuxAPI = DeleteDemuxAPI(
config=context,
demultiplex_base=demultiplex_base,
dry_run=dry_run,
run_path=run_path)
if yes or click.confirm(
f"Are you sure you want to delete the flow cell from the following databases:\n"
f"cg-stats={True if status_db else cg_stats}\nDemultiplexing-dir={True if status_db else demultiplexing_dir}\n"
f"Housekeeper={True if status_db else housekeeper}\nInit_files={True if status_db else init_files}\n"
f"Run-dir={True if status_db else run_dir}\nStatusdb={status_db}"):
wipe_demux_api.delete_flow_cell(
cg_stats=cg_stats,
demultiplexing_dir=demultiplexing_dir,
housekeeper=housekeeper,
init_files=init_files,
run_dir=run_dir,
status_db=status_db,
)