def get(self, db, collection=None, id=None):
import random
from biothings.utils.mongo import get_src_conn
get_random = self.get_argument('random', None) != 'false'
size = int(self.get_argument('size', 10))
conn = get_src_conn()
if collection:
if collection == 'fs':
import gridfs
fs = gridfs.GridFS(conn[db])
out = fs.list()
else:
collection = conn[db][collection]
if id:
out = collection.find_one({"_id": id})
elif get_random:
cnt = collection.count()
num = random.randint(0, max(cnt - size, 0))
out = list(collection.find().skip(num).limit(size))
else:
out = list(collection.find().limit(size))
else:
#list all collection in this db
out = conn[db].collection_names()
def date_handler(obj):
return obj.isoformat() if hasattr(obj, 'isoformat') else obj
_json_data = json.dumps(out, default=date_handler)
self.set_header("Content-Type", "application/json; charset=UTF-8")
self.write(_json_data)
def prepare(self, state={}):
"""Sync uploader information with database (or given state dict)"""
if self.prepared:
return
if state:
# let's be explicit, _state takes what it wants
for k in self._state:
self._state[k] = state[k]
return
self._state["conn"] = get_src_conn()
self._state["db"] = self._state["conn"][self.__class__.__database__]
self._state["collection"] = self._state["db"][self.collection_name]
self._state["src_dump"] = self.prepare_src_dump()
self._state["src_master"] = get_src_master()
self._state["logger"], self.logfile = self.setup_log()
self.data_folder = self.src_doc.get("download", {}).get("data_folder") or \
self.src_doc.get("data_folder")
# flag ready
self.prepared = True
'uniprot',
'uniprot.uniprot_pdb',
# 'uniprot.uniprot_ipi', # IPI is now discontinued, last update is still in the db, but won't be updated.
'uniprot.uniprot_pir'
],
'pharmgkb': ['pharmgkb'],
'reporter': ['reporter'],
'ucsc': ['ucsc.ucsc_exons'],
'exac': ['exac.broadinstitute_exac'],
'cpdb': ['cpdb'],
'reagent': ['reagent'],
}
__sources__ = None # should be a list defined at runtime
conn = get_src_conn()
doc_register = {}
class GeneDocSourceMaster(dict):
'''A class to manage various genedoc data sources.'''
__collection__ = DATA_SRC_MASTER_COLLECTION
__database__ = DATA_SRC_DATABASE
use_dot_notation = True
use_schemaless = True
structure = {
'name': str,
'timestamp': datetime.datetime,
}
'uniprot',
'uniprot.uniprot_pdb',
# 'uniprot.uniprot_ipi', # IPI is now discontinued, last update is still in the db, but won't be updated.
'uniprot.uniprot_pir'
],
'pharmgkb': ['pharmgkb'],
'reporter': ['reporter'],
'ucsc': ['ucsc.ucsc_exons'],
'exac': ['exac.broadinstitute_exac'],
'cpdb': ['cpdb'],
'reagent': ['reagent'],
}
__sources__ = None # should be a list defined at runtime
conn = get_src_conn()
doc_register = {}
class GeneDocSourceMaster(dict):
'''A class to manage various genedoc data sources.'''
__collection__ = DATA_SRC_MASTER_COLLECTION
__database__ = DATA_SRC_DATABASE
use_dot_notation = True
use_schemaless = True
structure = {
'name': str,
'timestamp': datetime.datetime,
}