def pull_occurrences_from_ala(params, dest_url, context, import_multspecies_params):
# 1. set progress
set_progress('RUNNING', 'Download occurrence dataset from ala', None, context)
# 2. Download all the occurrence dataset in the params list
results = []
try:
item, results = download_occurrence_from_ala(params, context)
# This is the zip file path of the occurrence dataset
ala_csv = item.get('file').get('url').split('file://')[1]
# Add the number of occurrence records to the metadata
# TODO: This is a hack. Any better solution.
occurrence_csv_filename = os.path.join('data', 'ala_occurrence.csv')
if occurrence_csv_filename in item['filemetadata']:
# FIXME: copy all occurrence metadata to zip level, for backwards
# compatibility... this should go away after we fully support 'layered'
# occurrence zips.
for key in ('rows', 'headers', 'bounds'): # what about 'species' ?
if key in item['filemetadata'][occurrence_csv_filename]['metadata']:
item['filemetadata'][key] = item['filemetadata'][occurrence_csv_filename]['metadata'][key]
# move data file to destination and build data_url
src = build_source('file://{}'.format(ala_csv))
dst = build_destination(os.path.join(dest_url, os.path.basename(ala_csv)), app.conf.get('bccvl', {}))
item['file']['url'] = dst['url']
movelib.move(src, dst)
# tell importer about new dataset (import it)
set_progress("RUNNING", u"Import dataset '{0}' from ALA".format(item['title']), None, context)
cleanup_job = import_cleanup_job(dest_url, context)
import_job = import_ala_job([item], dest_url, context)
import_job.link_error(set_progress_job(
"FAILED", u"Import of dataset '{0}' from ALA failed".format(item['title']), None, context))
import_job.link_error(cleanup_job)
finish_job = set_progress_job("COMPLETED", u"ALA import '{}' complete".format(item['title']), None, context)
# Split multi-species dataset
if import_multspecies_params:
import_multispecies_job = import_multi_species_csv_job(item.get('file').get('url'),
import_multspecies_params['results_dir'],
import_multspecies_params['import_context'],
context)
import_multispecies_job.link_error(set_progress_job(
"FAILED", u"Split multi-species dataset '{0}' from ALA failed".format(item['title']), None, context))
import_multispecies_job.link_error(cleanup_job)
(import_job | import_multispecies_job | cleanup_job | finish_job).delay()
else:
(import_job | cleanup_job | finish_job).delay()
except Exception as e:
set_progress('FAILED', 'Download occurrence dataset from ALA: {}'.format(e), None, context)
import_cleanup(dest_url, context)
LOG.error('Download from %s to %s failed: %s', params, dest_url, e, exc_info=True)
finally:
for tmpdir in results:
if tmpdir and os.path.exists(tmpdir):
shutil.rmtree(tmpdir)
def pull_occurrences_from_gbif(lsid, dest_url, context):
# 1. set progress
set_progress('RUNNING', 'Download {0} from gbif'.format(lsid), None, context)
# 2. do move
src = None
dst = None
try:
tmpdir = tempfile.mkdtemp(prefix='gbif_download_')
src = build_source('gbif://gbif?lsid={}'.format(lsid))
dst = build_destination('file://{}'.format(tmpdir))
movelib.move(src, dst)
# extract metadata and do other stuff....
set_progress('RUNNING', 'Extract metadata {0} from gbif'.format(lsid), None, context)
# open gbif_dateset.json
gbif_ds = json.load(open(os.path.join(tmpdir, 'gbif_dataset.json'), 'r'))
# collect files inside ds per datatype
files = dict(((f['dataset_type'], f) for f in gbif_ds['files']))
# read gbif metadata from attribution file
gbif_md = json.load(open(files['attribution']['url'], 'r'))
gbif_csv = files['occurrence']['url']
# build bccvl metadata:
bccvlmd = {
'genre': 'DataGenreSpeciesOccurrence',
'categories': ['occurrence'],
'species': {
'scientificName': gbif_md.get('scientificName', None),
'vernacularName': gbif_md.get('vernacularName', None),
'taxonID': gbif_md.get('key', None),
'rank': gbif_md.get('rank', None),
'genus': gbif_md.get('genus', None),
'genusGuid': gbif_md.get('genusKey', None),
'family': gbif_md.get('family', None),
'familyGuid': gbif_md.get('familyKey', None),
'order': gbif_md.get('order', None),
'orderGuid': gbif_md.get('orderKey', None),
'clazz': gbif_md.get('class', None),
'clazzGuid': gbif_md.get('classKey', None),
'phylum': gbif_md.get('phylum', None),
'phylumGuid': gbif_md.get('phylumKey', None),
'kingdom': gbif_md.get('kingdom', None),
'kingdomGuid': gbif_md.get('kingdomKey', None)
},
}
# build item to import
item = {
'title': gbif_ds['title'],
'description': gbif_ds['description'],
'file': {
'url': 'file://{}'.format(gbif_csv), # local file url
'contenttype': 'application/zip',
'filename': os.path.basename(gbif_csv)
},
'bccvlmetadata': bccvlmd,
'filemetadata': extract_metadata(gbif_csv, 'application/zip'),
}
# Add the number of occurrence records to the metadata
# TODO: This is a hack. Any better solution.
occurrence_csv_filename = os.path.join('data', 'gbif_occurrence.csv')
if occurrence_csv_filename in item['filemetadata']:
# FIXME: copy all occurrence metadata to zip level, for backwards
# compatibility... this should go away after we fully support 'layered'
# occurrence zips.
for key in ('rows', 'headers', 'bounds'): # what about 'species' ?
if key in item['filemetadata'][occurrence_csv_filename]['metadata']:
item['filemetadata'][key] = item['filemetadata'][occurrence_csv_filename]['metadata'][key]
# move data file to destination and build data_url
src = build_source('file://{}'.format(gbif_csv))
dst = build_destination(os.path.join(dest_url, os.path.basename(gbif_csv)), app.conf.get('bccvl', {}))
item['file']['url'] = dst['url']
movelib.move(src, dst)
# tell importer about new dataset (import it)
set_progress('RUNNING', 'Import gbif data {0}'.format(lsid), None, context)
cleanup_job = import_cleanup_job(dest_url, context)
import_job = import_ala_job([item], dest_url, context)
import_job.link_error(set_progress_job("FAILED", "Import of gbif data failed {0}".format(lsid), None, context))
import_job.link_error(cleanup_job)
finish_job = set_progress_job("COMPLETED", 'GBIF import {} complete'.format(lsid), None, context)
(import_job | cleanup_job | finish_job).delay()
except Exception as e:
set_progress('FAILED', 'Download {0} from gbif: {1}'.format(lsid, e), None, context)
import_cleanup(dest_url, context)
LOG.error('Download from %s to %s failed: %s', src, dest_url, e, exc_info=True)
finally:
if tmpdir and os.path.exists(tmpdir):
shutil.rmtree(tmpdir)
def run_script(wrapper, params, context):
# TODO: there are many little things that can fail here, and we
# need to communicate it properly back to the user.
# TODO: however, we can't really do anything in case sending
# messages doesn't work.
items = []
try:
errmsg = 'Fail to transfer/import data'
set_progress('RUNNING', 'Transferring data', None, context)
# create initial folder structure
create_workenv(params)
# transfer input files
transfer_inputs(params, context)
# create script
scriptname = create_scripts(params, context)
# run the script
errmsg = 'Fail to run experiement'
set_progress('RUNNING', 'Executing job', None, context)
scriptout = os.path.join(params['env']['outputdir'],
params['worker']['script']['name'] + 'out')
outfile = open(scriptout, 'w')
wrapsh = os.path.join(params['env']['scriptdir'], 'wrap.sh')
open(wrapsh, 'w').write(wrapper)
# zip up workenv if requested
if params['worker'].get('zipworkenv', False):
# make sure tmp is big enough
# TODO: add toolkit name to zip name ... workenv_bioclim.zip
zip_folder(os.path.join(params['env']['outputdir'], 'workenv.zip'),
params['env']['workdir'])
cmd = ["/bin/bash", "-l", "wrap.sh", scriptname]
LOG.info("Executing: %s", ' '.join(cmd))
proc = subprocess.Popen(cmd, cwd=params['env']['scriptdir'],
close_fds=True,
stdout=outfile, stderr=subprocess.STDOUT)
rpid, ret, rusage = os.wait4(proc.pid, 0)
# TODO: should we write this as json file and send as result back
# or just send rusage with finished message?
usage = get_rusage(rusage)
# TODO: check whether ret and proc.returncode are the same
# move results back
errmsg = 'Fail to transfer results back'
set_progress('RUNNING', 'Transferring outputs', usage, context)
# TODO: maybe redesign this?
# transfer only uploads to destination and stores new url somewhere
# and we do metadata extraction and item creation afterwards (here)?
items = transfer_outputs(params, context)
# we are done here, hand over to result importer
# build a chain of the remaining tasks
start_import = set_progress_job(
'RUNNING', 'Import results', None, context)
cleanup_job = import_cleanup_job(
params['result']['results_dir'], context)
import_job = import_result_job(items, params['result'][
'results_dir'], context)
import_job.link_error(set_progress_job(
'FAILED', 'Result import failed', None, context))
import_job.link_error(cleanup_job)
if ret != 0:
errmsg = 'Script execution failed with exit code {0}'.format(ret)
finish_job = set_progress_job('FAILED', errmsg, None, context)
else:
finish_job = set_progress_job(
'COMPLETED', 'Task succeeded', None, context)
(start_import | import_job | cleanup_job | finish_job).delay()
except Exception as e:
# TODO: capture stacktrace
# need to start import to get import cleaned up
# Log error message with stacktrace.
#:( exposes internals, ugly hash, complicated with admin only access
#-> certainly need to get rid of exception in message.
# test exceptions:
# ... upload file, replace with something else (unzip error)
# ... delete file and rerun experiment (donwload error)
# ... create file/folder error? (can't write log)
# ... how to simulate fault? (download error)
# log error message with exception and traceback
LOG.error(errmsg, exc_info=True)
start_import = set_progress_job(
'RUNNING', 'Import results', None, context)
import_job = import_result_job(items, params['result'][
'results_dir'], context)
import_job.link_error(set_progress_job(
'FAILED', 'Result import failed', None, context))
finish_job = set_progress_job('FAILED', errmsg, None, context)
(start_import | import_job | finish_job).delay()
raise
finally:
# TODO: check if dir exists
path = params['env'].get('workdir', None)
if path and os.path.exists(path):
shutil.rmtree(path)
def import_multi_species_csv(url, results_dir, import_context, context):
# url .... source file
# results_dir ... folder to place split files into
# context ... the context with user and orig dataset
try:
set_progress('RUNNING', 'Split {0}'.format(url), None, context)
# step 1: update main dataset metadata
tmpdir = tempfile.mkdtemp()
userid = context.get('user', {}).get('id')
settings = app.conf.get('bccvl', {})
src = build_source(url, userid, settings)
dst = build_destination('file://{}'.format(tmpdir), settings)
movelib.move(src, dst)
# Get the downloaded filename
tmpfile = glob.glob(os.path.join(tmpdir, '*'))[0]
# Extract occurrence file from downloaded file
mimetype, enc = mimetypes.guess_type(tmpfile)
if mimetype == 'application/zip':
src_occ_data = os.path.join('data', 'ala_occurrence.csv')
with zipfile.ZipFile(tmpfile, 'r') as zipf:
occfile = os.path.join(tmpdir, src_occ_data)
zipf.extract(src_occ_data, tmpdir)
item = {
'filemetadata': extract_metadata(tmpfile, 'application/zip')
}
occmd = item['filemetadata'].get(src_occ_data, {}).get('metadata', {})
else:
# csv file
item = {
'filemetadata': extract_metadata(tmpfile, "text/csv")
}
occfile = tmpfile
occmd = item['filemetadata']
# Check that there are lon and lat columns
# if upload is of type csv, we validate column names as well
if ('headers' not in occmd
or 'lat' not in occmd['headers']
or 'lon' not in occmd['headers']):
raise Exception("Missing 'lat'/'lon' column")
set_progress('RUNNING',
'Import metadata for {0}'.format(url),
None, context)
import_md_job = import_file_metadata_job([item], url, context)
import_md_job.link_error(set_progress_job(
"FAILED", "Metadata update failed for {0}".format(url),
None, context))
# step 2: split csv file and create sub datasets
# start reading csv file and create new datasets which will be
# linked up with dataset collection item
# FIXME: large csv files should be streamed to seperate files (not read
# into ram like here)
f = io.open(occfile, 'r', encoding='utf-8', errors='ignore')
csvreader = UnicodeCSVReader(f)
headers = csvreader.next()
if 'species' not in headers:
raise Exception('missing species column')
speciesidx = headers.index('species')
# create dict with all data .... species column used as key, and rest
# is just added
data = {}
for row in csvreader:
if not row:
continue
species = row[speciesidx]
if species not in data:
# create new entry for species
fname = u'{0}.csv'.format(species).replace(
u'/', u'_').encode('idna')
# TODO: make sure fname contains only legal filename characters
fpath = os.path.join(tmpdir, fname)
file = io.open(fpath, 'wb')
fwriter = UnicodeCSVWriter(file)
fwriter.writerow(headers)
data[species] = {
'file': file,
'writer': fwriter,
'path': fpath,
'name': fname
}
data[species]['writer'].writerow(row)
# ok we have got all data and everything in separate files
# close all files
for species in data:
data[species]['file'].close()
del data[species]['file']
del data[species]['writer']
# extract metadata
for species in data:
data[species]['filemetadata'] = extract_metadata(
data[species]['path'],
'text/csv'
)
# send files to destination
for species in data:
src = build_source('file://{}'.format(data[species]['path']))
dst = build_destination(os.path.join(results_dir,
data[species]['name']),
app.conf.get('bccvl', {}))
data[species]['url'] = dst['url']
movelib.move(src, dst)
# all files uploaded .... send import jobs
set_progress('RUNNING', 'Create datasets for {0}'.format(
url), None, context)
items = []
for species in data:
# build item
item = {
'title': u'{0} occurrences'.format(species),
'description': '',
'file': {
'url': data[species]['url'],
'filename': data[species]['name'],
'contenttype': 'text/csv',
},
'bccvlmetadata': {
'genre': 'DataGenreSpeciesOccurrence',
'categories': ['occurrence'],
'species': {
'scientificName': species,
}
},
'filemetadata': data[species]['filemetadata'],
'_partof': {
# add back reference to orig dataset
# TODO: shouldn't use absolute path here
'path': context['context']
}
}
items.append(item)
# start import process
start_import = set_progress_job(
'RUNNING', 'Import results', None, context)
# What is results_dir being used for?
import_job = import_result_job(items, results_dir, import_context)
cleanup_job = import_cleanup_job(results_dir, context)
import_job.link_error(set_progress_job(
'FAILED', 'Multi species import failed', None, context))
import_job.link_error(cleanup_job)
finish_job = set_progress_job(
'COMPLETED', 'Task succeeded', None, context)
(start_import | import_md_job | import_job |
cleanup_job | finish_job).delay()
# FIXME: missing stuff...
# need to set multi species collection to finished at some stage
except Exception as e:
set_progress('FAILED',
'Error while splitting Multi Species CSV {}: {}'.format(
url, e),
None, context)
LOG.error('Multi species split for %s faild: %s', url, e, exc_info=True)
finally:
if tmpdir and os.path.exists(tmpdir):
shutil.rmtree(tmpdir)
