def download_all_structures(db_path, record_type=GlycanRecordWithTaxon): # pragma: no cover
response = requests.get(u'http://www.glycome-db.org/http-services/getStructureDump.action?user=eurocarbdb')
response.raise_for_status()
handle = gzip.GzipFile(fileobj=StringIO(response.content))
xml = etree.parse(handle)
db = RecordDatabase(db_path, record_type=record_type)
misses = []
i = 0
for structure in xml.iterfind(".//structure"):
try:
glycomedb_id = int(structure.attrib['id'])
i += 1
glycoct_str = structure.find("sequence").text
taxa = [Taxon(t.attrib['ncbi'], None, None) for t in structure.iterfind(".//taxon")]
glycan = glycoct.loads(glycoct_str)
if (glycoct.loads(str(glycan)).mass() - glycan.mass()) > 0.00001:
raise Exception("Mass did not match on reparse")
record = record_type(glycan, taxa=taxa, id=glycomedb_id)
db.load_data(record, commit=False, set_id=False)
if i % 1000 == 0:
print(i, "Records parsed.")
except Exception as e:
misses.append((glycomedb_id, e))
print(glycomedb_id, e)
db.set_metadata("misses", misses)
db.commit()
return db
