def contar_registros():
db = conectar('openlibrary')
for nome in db.collection_names():
if nome.startswith('system.'):
continue
colecao = getattr(db, nome)
print '{1:10,} {0}'.format(nome, colecao.count())
from mongo_util import conectar
from bson.code import Code
from pprint import pprint
db = conectar('openlibrary')
for res in db.editions.find({'contributions':{'$exists':1}}, {'contributions':1}):
pprint(res)
#!/coding: utf-8
from mongo_util import conectar
from bson.code import Code
from pprint import pprint
db = conectar("openlibrary")
for res in db.editions.find({"authors": {"$exists": 1}}, {"authors": 1}):
for author_fkey in res[u"authors"]:
author = db.authors.find_one({"_id": author_fkey[u"key"]}, {"name": 1})
if author is None:
print author["name"]
#!/coding: utf-8
from mongo_util import conectar
from bson.code import Code
db = conectar('openlibrary')
TYPE_CODES = {'object': 1, 'array': 2, 'string': 4, 'number': 8}
MAP = Code('''function () {
var type_codes = %r;
var type = '';
for (field_name in this) {
type = typeof this[field_name];
if (type === 'object') {
type = this[field_name] instanceof Array ? 'array' : 'object';
}
emit(field_name, type_codes[type])
}
}''' % TYPE_CODES)
REDUCE = Code('''function (key, values) {
var types = 0;
values.forEach(function(type_code) { types |= type_code; });
return types;
}''')
def count_fields(collection):
return db[collection].map_reduce(MAP, REDUCE, "field_types").find()
def contar_registros():
db = conectar('test')
print db.dup_test.count(), 'registros efetivamente salvos'
for r in db.dup_test.find():
pprint(r)