def upload_to_gdocs(fcdir, credentials_file=None, gdocs_folder=None):
output_data = {
'stdout': StringIO(),
'stderr': StringIO(),
'debug': StringIO()
}
if not os.path.exists(fcdir):
LOG.error("The run folder, {} does not exist!".format(
os.path.basename(fcdir)))
return output_data
credentials = google.get_credentials(credentials_file)
if credentials is None:
LOG.error("Could not parse the Google Docs credentials")
return output_data
metrics = collect_metrics(fcdir)
samples = _format_samples(metrics)
ssheet_name = _demultiplex_spreadsheet(metrics['RunInfo'].get(
'Date', None))
ssheet = SpreadSheet(credentials, ssheet_name)
ssheet.move_to_folder(gdocs_folder)
run_id = metrics['RunInfo']['Id'].split("_")
wsheet_name = "_".join([run_id[0], run_id[-1]])
# Write the metrics for the entire flowcell
write_flowcell_metrics(samples, ssheet, wsheet_name)
# Write project-centered metrics
projects = list(set([sample.get('Project name', '')
for sample in samples]))
for project in projects:
if project in ['Undetermined_indices', '']:
continue
project_samples = [
sample for sample in samples
if sample.get('Project name', '') == project
]
# Insert the run name as description
for sample in project_samples:
sample['Description'] = wsheet_name
ssheet_name = "{}_sequencing_results".format(project)
ssheet = SpreadSheet(credentials, ssheet_name)
ssheet.move_to_folder(gdocs_folder)
# Truncate the summary worksheet so that it won't show the wrong information in case upload fails
write_flowcell_metrics([], ssheet, "Summary")
write_flowcell_metrics(project_samples, ssheet, wsheet_name)
# Create the summary over all worksheets in the project
summary_samples = summarize_project(ssheet)
write_flowcell_metrics(summary_samples, ssheet, "Summary")
return output_data
def upload_to_gdocs(log, fcdir, credentials_file=None, gdocs_folder=None):
output_data = {'stdout':StringIO(), 'stderr':StringIO(), 'debug':StringIO()}
if not os.path.exists(fcdir):
log.error("The run folder, {} does not exist!".format(os.path.basename(fcdir)))
return output_data
credentials = google.get_credentials(credentials_file)
if credentials is None:
log.error("Could not parse the Google Docs credentials")
return output_data
metrics = collect_metrics(fcdir, log)
samples = _format_samples(metrics)
ssheet_name = _demultiplex_spreadsheet(metrics['RunInfo'].get('Date',None))
ssheet = SpreadSheet(credentials,ssheet_name)
if ssheet.new_doc:
ssheet.move_to_folder(gdocs_folder)
run_id = metrics['RunInfo']['Id'].split("_")
wsheet_name = "_".join([run_id[0],run_id[-1]])
# Write the metrics for the entire flowcell
write_flowcell_metrics(samples, ssheet, wsheet_name)
# Write project-centered metrics
projects = list(set([sample.get('Project name','') for sample in samples]))
for project in projects:
if project in ['Undetermined_indices','']:
continue
project_samples = [sample for sample in samples if sample.get('Project name','') == project]
# Insert the run name as description
for sample in project_samples:
sample['Description'] = wsheet_name
ssheet_name = "{}_sequencing_results".format(project)
ssheet = SpreadSheet(credentials,ssheet_name)
if ssheet.new_doc:
ssheet.move_to_folder(gdocs_folder)
# Truncate the summary worksheet so that it won't show the wrong information in case upload fails
write_flowcell_metrics([], ssheet, "Summary")
project_samples = summarize_project(log, ssheet,{wsheet_name: project_samples})
write_flowcell_metrics(project_samples, ssheet, wsheet_name)
# Create the summary over all worksheets in the project
summary_samples = summarize_project(log, ssheet)
write_flowcell_metrics(summary_samples, ssheet, "Summary")
return output_data
