def _jsonify_collection_info(coll, dbname=None):
"""Private function to turn information about one collection into base
JSON """
if dbname is None:
dbname = coll.name
results = _get_db_records(coll)
json_db_data = {}
json_db_data['dbinfo'] = {}
json_db_data['dbinfo']['name'] = dbname
json_db_data['records'] = {}
json_db_data['fields'] = {}
lst = set()
for doc in results:
lst = lst | set(doc['_id'])
lst = list(lst)
lst.sort()
for doc in lst:
try:
rls = dbtools.get_sample_record(coll, str(doc))
try:
typedesc = id_object.get_description(rls[str(doc)])
except:
typedesc = 'Type cannot be identified (' \
+ str(type(rls[str(doc)])) + ')'
try:
strval = str(rls[str(doc)]).decode('unicode_escape').\
encode('ascii','ignore')
except:
strval = 'Record cannot be stringified'
except:
typedesc = 'Record cannot be found containing key'
strval = 'N/A'
lstr = len(strval)
strval = strval.replace('\n', ' ').replace('\r', '')
strval = '`' + strval[:100].strip() + '`'
if lstr > 100:
strval = strval + ' ...'
json_db_data['fields'][str(doc)] = {}
json_db_data['fields'][str(doc)]['type'] = typedesc
json_db_data['fields'][str(doc)]['example'] = strval
for recordid, doc in enumerate(results):
json_db_data['records'][recordid] = {}
json_db_data['records'][recordid]['count'] = int(doc['value'])
json_db_data['records'][recordid]['schema'] = doc['_id']
indices = coll.index_information()
json_db_data['indices'] = {}
for recordid, index in enumerate(indices):
json_db_data['indices'][recordid] = {}
json_db_data['indices'][recordid]['name'] = index
json_db_data['indices'][recordid]['keys'] = indices[index]['key']
return json_db_data
def _jsonify_record(name, record, parse_jsonb=False, inferred=False):
vals = {}
if isinstance(record, dict) and parse_jsonb:
for el in record:
merge_dicts(
vals,
_jsonify_record(name + "." + str(el),
record[el],
parse_jsonb=parse_jsonb,
inferred=True))
if isinstance(record, list) and parse_jsonb:
if len(record) > 0 and isinstance(record[0], dict):
merge_dicts(
vals,
_jsonify_record(name,
record[0],
parse_jsonb=parse_jsonb,
inferred=True))
try:
typedesc = id_object.get_description(record)
except:
typedesc = 'Type cannot be identified (' \
+ str(type(record)) + ')'
try:
strval = str(record).decode('unicode_escape').\
encode('ascii','ignore')
except:
strval = 'Record cannot be stringified'
lstr = len(strval)
strval = strval.replace('\n', ' ').replace('\r', '')
strval = '`' + strval[:100].strip() + '`'
if lstr > 100:
strval = strval + ' ...'
vals[str(name)] = {}
vals[str(name)]['type'] = typedesc
vals[str(name)]['example'] = strval
vals[str(name)]['inferred'] = inferred
return vals
def _jsonify_table_info(table, dbname = None):
"""Private function to turn information about one table into base
JSON """
# Needs to be rewritten for Postgres
raise NotImplementedError
if dbname is None:
dbname = table.search_table
results = _get_db_records(table)
json_db_data = {}
json_db_data['dbinfo'] ={}
json_db_data['dbinfo']['name'] = dbname
json_db_data['records'] = {}
json_db_data['fields'] = {}
lst=set()
for doc in results:
lst = lst | set(doc['_id'])
lst=list(lst)
lst.sort()
for doc in lst:
try:
rls = dbtools.get_sample_record(table, str(doc))
try:
typedesc = id_object.get_description(rls[str(doc)])
except:
typedesc = 'Type cannot be identified (' \
+ str(type(rls[str(doc)])) + ')'
try:
strval = str(rls[str(doc)]).decode('unicode_escape').\
encode('ascii','ignore')
except:
strval = 'Record cannot be stringified'
except:
typedesc = 'Record cannot be found containing key'
strval = 'N/A'
lstr = len(strval)
strval = strval.replace('\n',' ').replace('\r','')
strval = '`' + strval[:100].strip() + '`'
if lstr > 100:
strval = strval + ' ...'
json_db_data['fields'][str(doc)] = {}
json_db_data['fields'][str(doc)]['type'] = typedesc
json_db_data['fields'][str(doc)]['example'] = strval
for recordid, doc in enumerate(results):
json_db_data['records'][recordid] = {}
json_db_data['records'][recordid]['count'] = int(doc['value'])
json_db_data['records'][recordid]['schema'] = doc['_id']
indices = table.index_information()
json_db_data['indices'] = {}
for recordid, index in enumerate(indices):
json_db_data['indices'][recordid] = {}
json_db_data['indices'][recordid]['name'] = index
json_db_data['indices'][recordid]['keys'] = indices[index]['key']
return json_db_data