def value2json(value):
try:
scrubbed = scrub(value, scrub_number=float)
return text(_json_encoder(scrubbed))
except Exception as e:
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=text(repr(value)), cause=e)
raise e
def encode(self, value, pretty=False):
if pretty:
return pretty_json(value)
try:
scrubbed = scrub(value)
return text_type(self.encoder(scrubbed))
except Exception as e:
from mo_logs.exceptions import Except
from mo_logs import Log
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=text_type(repr(value)), cause=e)
raise e
def ujson_encode(value, pretty=False):
if pretty:
return pretty_json(value)
try:
scrubbed = scrub(value)
return ujson_dumps(scrubbed, ensure_ascii=False, sort_keys=True, escape_forward_slashes=False).decode('utf8')
except Exception as e:
from mo_logs.exceptions import Except
from mo_logs import Log
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=text_type(repr(value)), cause=e)
raise e
def encode(self, value, pretty=False):
if pretty:
return pretty_json(value)
try:
scrubbed = scrub(value)
return unicode(self.encoder.encode(scrubbed))
except Exception as e:
from mo_logs.exceptions import Except
from mo_logs import Log
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=_repr(value), cause=e)
raise e
def encode(self, value, pretty=False):
if pretty:
return pretty_json(value)
try:
with Timer("scrub", too_long=0.1):
scrubbed = scrub(value)
with Timer("encode", too_long=0.1):
return text_type(self.encoder(scrubbed))
except Exception as e:
from mo_logs.exceptions import Except
from mo_logs import Log
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=text_type(repr(value)), cause=e)
raise e
def update(self, command):
"""
EXPECTING command == {"set":term, "where":where}
THE set CLAUSE IS A DICT MAPPING NAMES TO VALUES
THE where CLAUSE IS AN ES FILTER
"""
command = wrap(command)
schema = self._es.get_properties()
# GET IDS OF DOCUMENTS
results = self._es.search({
"stored_fields": listwrap(schema._routing.path),
"query": {"bool": {
"filter": jx_expression(command.where).to_esfilter(Null)
}},
"size": 10000
})
# SCRIPT IS SAME FOR ALL (CAN ONLY HANDLE ASSIGNMENT TO CONSTANT)
scripts = FlatList()
for k, v in command.set.items():
if not is_variable_name(k):
Log.error("Only support simple paths for now")
if isinstance(v, Mapping) and v.doc:
scripts.append({"doc": v.doc})
else:
v = scrub(v)
scripts.append({"script": "ctx._source." + k + " = " + jx_expression(v).to_painless(schema).script(schema)})
if results.hits.hits:
updates = []
for h in results.hits.hits:
for s in scripts:
updates.append({"update": {"_id": h._id, "_routing": unwraplist(h.fields[literal_field(schema._routing.path)])}})
updates.append(s)
content = ("\n".join(convert.value2json(c) for c in updates) + "\n").encode('utf-8')
response = self._es.cluster.post(
self._es.path + "/_bulk",
data=content,
headers={"Content-Type": "application/json"},
timeout=self.settings.timeout,
params={"wait_for_active_shards": self.settings.wait_for_active_shards}
)
if response.errors:
Log.error("could not update: {{error}}", error=[e.error for i in response["items"] for e in i.values() if e.status not in (200, 201)])
def update(self, command):
"""
EXPECTING command == {"set":term, "where":where}
THE set CLAUSE IS A DICT MAPPING NAMES TO VALUES
THE where CLAUSE IS AN ES FILTER
"""
command = wrap(command)
schema = self.es.get_properties()
# GET IDS OF DOCUMENTS
results = self.es.search({
"fields": listwrap(schema._routing.path),
"query": {"filtered": {
"filter": jx_expression(command.where).to_esfilter(Null)
}},
"size": 10000
})
# SCRIPT IS SAME FOR ALL (CAN ONLY HANDLE ASSIGNMENT TO CONSTANT)
scripts = FlatList()
for k, v in command.set.items():
if not is_variable_name(k):
Log.error("Only support simple paths for now")
if isinstance(v, Mapping) and v.doc:
scripts.append({"doc": v.doc})
else:
v = scrub(v)
scripts.append({"script": "ctx._source." + k + " = " + jx_expression(v).to_es_script(schema).script(schema)})
if results.hits.hits:
updates = []
for h in results.hits.hits:
for s in scripts:
updates.append({"update": {"_id": h._id, "_routing": unwraplist(h.fields[literal_field(schema._routing.path)])}})
updates.append(s)
content = ("\n".join(value2json(c) for c in updates) + "\n")
response = self.es.cluster.post(
self.es.path + "/_bulk",
data=content,
headers={"Content-Type": "application/json"},
timeout=self.settings.timeout,
params={"wait_for_active_shards": self.settings.wait_for_active_shards}
)
if response.errors:
Log.error("could not update: {{error}}", error=[e.error for i in response["items"] for e in i.values() if e.status not in (200, 201)])
def test_json(results, description, method, n):
output = []
for case in cases:
try:
data, count = globals()[case]
if "scrub" in description:
# SCRUB BEFORE SENDING TO C ROUTINE (NOT FAIR, BUT WE GET TO SEE HOW FAST ENCODING GOES)
data = unwrap(scrub(data))
try:
example = method(data)
if case == "HUGE":
example = "<too big to show>"
except Exception as e:
Log.warning(u"json encoding failure", cause=e)
example = "<CRASH>"
t0 = time.time()
try:
for i in range(n):
for i in range(count):
output.append(method(data))
duration = time.time() - t0
except Exception:
duration = time.time() - t0
summary = {
"description": description,
"interpreter": platform.python_implementation(),
"time": duration,
"type": case,
"num": n,
"count": count,
"length": len(output),
"result": example
}
Log.note(
u"using {{interpreter}}: {{description}} {{type}} x {{num}} x {{count}} = {{time}} result={{result}}",
**summary)
results.append(summary)
except Exception as e:
Log.warning(u"problem with encoding: {{message}}",
{"message": e.message}, e)
def encode(self, value, pretty=False):
if pretty:
return pretty_json(value)
try:
with Timer("scrub", too_long=0.1):
scrubbed = scrub(value)
param = {"size": 0}
with Timer("encode {{size}} characters", param=param, too_long=0.1):
output = text_type(self.encoder(scrubbed))
param["size"] = len(output)
return output
except Exception as e:
from mo_logs.exceptions import Except
from mo_logs import Log
e = Except.wrap(e)
Log.warning("problem serializing {{type}}", type=text_type(repr(value)), cause=e)
raise e
def query(self, query):
"""
:param query: JSON Query Expression, SET `format="container"` TO MAKE NEW TABLE OF RESULT
:return:
"""
if not startswith_field(query['from'], self.name):
Log.error("Expecting table, or some nested table")
frum, query['from'] = query['from'], self
query = QueryOp.wrap(query, self.columns)
# TYPE CONFLICTS MUST NOW BE RESOLVED DURING
# TYPE-SPECIFIC QUERY NORMALIZATION
# vars_ = query.vars(exclude_select=True)
# type_map = {
# v: c.es_column
# for v in vars_
# if v in self.columns and len([c for c in self.columns[v] if c.type != "nested"]) == 1
# for c in self.columns[v]
# if c.type != "nested"
# }
#
# sql_query = query.map(type_map)
query = query
new_table = "temp_" + unique_name()
if query.format == "container":
create_table = "CREATE TABLE " + quote_table(new_table) + " AS "
else:
create_table = ""
if query.groupby:
op, index_to_columns = self._groupby_op(query, frum)
command = create_table + op
elif query.edges or any(a != "none"
for a in listwrap(query.select).aggregate):
op, index_to_columns = self._edges_op(query, frum)
command = create_table + op
else:
op = self._set_op(query, frum)
return op
if query.sort:
command += "\nORDER BY " + ",\n".join(
"(" + sql[t] + ") IS NULL" +
(" DESC" if s.sort == -1 else "") + ",\n" + sql[t] +
(" DESC" if s.sort == -1 else "")
for s, sql in [(s, s.value.to_sql(self)[0].sql)
for s in query.sort] for t in "bns" if sql[t])
result = self.db.query(command)
column_names = query.edges.name + query.groupby.name + listwrap(
query.select).name
if query.format == "container":
output = QueryTable(new_table,
db=self.db,
uid=self.uid,
exists=True)
elif query.format == "cube" or (not query.format and query.edges):
if len(query.edges) == 0 and len(query.groupby) == 0:
data = {n: Data() for n in column_names}
for s in index_to_columns.values():
data[s.push_name][s.push_child] = unwrap(
s.pull(result.data[0]))
return Data(data=unwrap(data), meta={"format": "cube"})
if not result.data:
edges = []
dims = []
for i, e in enumerate(query.edges + query.groupby):
allowNulls = coalesce(e.allowNulls, True)
if e.domain.type == "set" and e.domain.partitions:
domain = SimpleSetDomain(
partitions=e.domain.partitions.name)
elif e.domain.type == "range":
domain = e.domain
elif isinstance(e.value, TupleOp):
pulls = jx.sort([
c for c in index_to_columns.values()
if c.push_name == e.name
], "push_child").pull
parts = [
tuple(p(d) for p in pulls) for d in result.data
]
domain = SimpleSetDomain(
partitions=jx.sort(set(parts)))
else:
domain = SimpleSetDomain(partitions=[])
dims.append(1 if allowNulls else 0)
edges.append(
Data(name=e.name, allowNulls=allowNulls,
domain=domain))
zeros = [
0 if s.aggregate == "count"
and index_to_columns[si].push_child == "." else Data
for si, s in enumerate(listwrap(query.select))
]
data = {
s.name: Matrix(dims=dims, zeros=zeros[si])
for si, s in enumerate(listwrap(query.select))
}
if isinstance(query.select, list):
select = [{"name": s.name} for s in query.select]
else:
select = {"name": query.select.name}
return Data(meta={"format": "cube"},
edges=edges,
select=select,
data={k: v.cube
for k, v in data.items()})
columns = None
edges = []
dims = []
for g in query.groupby:
g.is_groupby = True
for i, e in enumerate(query.edges + query.groupby):
allowNulls = coalesce(e.allowNulls, True)
if e.domain.type == "set" and e.domain.partitions:
domain = SimpleSetDomain(
partitions=e.domain.partitions.name)
elif e.domain.type == "range":
domain = e.domain
elif e.domain.type == "time":
domain = wrap(mo_json.scrub(e.domain))
elif e.domain.type == "duration":
domain = wrap(mo_json.scrub(e.domain))
elif isinstance(e.value, TupleOp):
pulls = jx.sort([
c for c in index_to_columns.values()
if c.push_name == e.name
], "push_child").pull
parts = [tuple(p(d) for p in pulls) for d in result.data]
domain = SimpleSetDomain(partitions=jx.sort(set(parts)))
else:
if not columns:
columns = zip(*result.data)
parts = set(columns[i])
if e.is_groupby and None in parts:
allowNulls = True
parts -= {None}
domain = SimpleSetDomain(partitions=jx.sort(parts))
dims.append(len(domain.partitions) + (1 if allowNulls else 0))
edges.append(
Data(name=e.name, allowNulls=allowNulls, domain=domain))
zeros = [
0 if s.aggregate == "count"
and index_to_columns[si].push_child == "." else Data
for si, s in enumerate(listwrap(query.select))
]
data_cubes = {
s.name: Matrix(dims=dims, zeros=zeros[si])
for si, s in enumerate(listwrap(query.select))
}
r2c = index_to_coordinate(
dims) # WORKS BECAUSE THE DATABASE SORTED THE EDGES TO CONFORM
for rownum, row in enumerate(result.data):
coord = r2c(rownum)
for i, s in enumerate(index_to_columns.values()):
if s.is_edge:
continue
if s.push_child == ".":
data_cubes[s.push_name][coord] = s.pull(row)
else:
data_cubes[s.push_name][coord][s.push_child] = s.pull(
row)
if isinstance(query.select, list):
select = [{"name": s.name} for s in query.select]
else:
select = {"name": query.select.name}
return Data(meta={"format": "cube"},
edges=edges,
select=select,
data={k: v.cube
for k, v in data_cubes.items()})
elif query.format == "table" or (not query.format and query.groupby):
data = []
for d in result.data:
row = [None for _ in column_names]
for s in index_to_columns.values():
if s.push_child == ".":
row[s.push_column] = s.pull(d)
elif s.num_push_columns:
tuple_value = row[s.push_column]
if tuple_value == None:
tuple_value = row[
s.push_column] = [None] * s.num_push_columns
tuple_value[s.push_child] = s.pull(d)
elif row[s.push_column] == None:
row[s.push_column] = Data()
row[s.push_column][s.push_child] = s.pull(d)
else:
row[s.push_column][s.push_child] = s.pull(d)
data.append(tuple(unwrap(r) for r in row))
output = Data(meta={"format": "table"},
header=column_names,
data=data)
elif query.format == "list" or (not query.edges and not query.groupby):
if not query.edges and not query.groupby and any(
listwrap(query.select).aggregate):
if isinstance(query.select, list):
data = Data()
for c in index_to_columns.values():
if c.push_child == ".":
data[c.push_name] = c.pull(result.data[0])
else:
data[c.push_name][c.push_child] = c.pull(
result.data[0])
output = Data(meta={"format": "value"}, data=data)
else:
data = Data()
for s in index_to_columns.values():
data[s.push_child] = s.pull(result.data[0])
output = Data(meta={"format": "value"}, data=unwrap(data))
else:
data = []
for rownum in result.data:
row = Data()
for c in index_to_columns.values():
if c.push_child == ".":
row[c.push_name] = c.pull(rownum)
elif c.num_push_columns:
tuple_value = row[c.push_name]
if not tuple_value:
tuple_value = row[
c.push_name] = [None] * c.num_push_columns
tuple_value[c.push_child] = c.pull(rownum)
else:
row[c.push_name][c.push_child] = c.pull(rownum)
data.append(row)
output = Data(meta={"format": "list"}, data=data)
else:
Log.error("unknown format {{format}}", format=query.format)
return output
def pretty_json(value):
try:
if value is False:
return "false"
elif value is True:
return "true"
elif isinstance(value, Mapping):
try:
items = sort_using_key(list(value.items()), lambda r: r[0])
values = [
encode_basestring(k) + PRETTY_COLON +
indent(pretty_json(v)).strip() for k, v in items
if v != None
]
if not values:
return "{}"
elif len(values) == 1:
return "{" + values[0] + "}"
else:
return "{\n" + INDENT + (",\n" +
INDENT).join(values) + "\n}"
except Exception as e:
from mo_logs import Log
from mo_math import OR
if OR(not isinstance(k, text_type) for k in value.keys()):
Log.error("JSON must have string keys: {{keys}}:",
keys=[k for k in value.keys()],
cause=e)
Log.error("problem making dict pretty: keys={{keys}}:",
keys=[k for k in value.keys()],
cause=e)
elif value in (None, Null):
return "null"
elif isinstance(value, (text_type, binary_type)):
if isinstance(value, binary_type):
value = utf82unicode(value)
try:
return quote(value)
except Exception as e:
from mo_logs import Log
try:
Log.note(
"try explicit convert of string with length {{length}}",
length=len(value))
acc = [QUOTE]
for c in value:
try:
try:
c2 = ESCAPE_DCT[c]
except Exception:
c2 = c
c3 = text_type(c2)
acc.append(c3)
except BaseException:
pass
# Log.warning("odd character {{ord}} found in string. Ignored.", ord= ord(c)}, cause=g)
acc.append(QUOTE)
output = u"".join(acc)
Log.note("return value of length {{length}}",
length=len(output))
return output
except BaseException as f:
Log.warning("can not even explicit convert {{type}}",
type=f.__class__.__name__,
cause=f)
return "null"
elif isinstance(value, list):
if not value:
return "[]"
if ARRAY_MAX_COLUMNS == 1:
return "[\n" + ",\n".join(
[indent(pretty_json(v)) for v in value]) + "\n]"
if len(value) == 1:
j = pretty_json(value[0])
if j.find("\n") >= 0:
return "[\n" + indent(j) + "\n]"
else:
return "[" + j + "]"
js = [pretty_json(v) for v in value]
max_len = max(*[len(j) for j in js])
if max_len <= ARRAY_ITEM_MAX_LENGTH and max(
*[j.find("\n") for j in js]) == -1:
# ALL TINY VALUES
num_columns = max(
1,
min(
ARRAY_MAX_COLUMNS,
int(
floor((ARRAY_ROW_LENGTH + 2.0) /
float(max_len +
2))))) # +2 TO COMPENSATE FOR COMMAS
if len(js) <= num_columns: # DO NOT ADD \n IF ONLY ONE ROW
return "[" + PRETTY_COMMA.join(js) + "]"
if num_columns == 1: # DO NOT rjust IF THERE IS ONLY ONE COLUMN
return "[\n" + ",\n".join(
[indent(pretty_json(v)) for v in value]) + "\n]"
content = ",\n".join(
PRETTY_COMMA.join(
j.rjust(max_len) for j in js[r:r + num_columns])
for r in xrange(0, len(js), num_columns))
return "[\n" + indent(content) + "\n]"
pretty_list = js
output = ["[\n"]
for i, p in enumerate(pretty_list):
try:
if i > 0:
output.append(",\n")
output.append(indent(p))
except Exception:
from mo_logs import Log
Log.warning(
"problem concatenating string of length {{len1}} and {{len2}}",
len1=len("".join(output)),
len2=len(p))
output.append("\n]")
try:
return "".join(output)
except Exception as e:
from mo_logs import Log
Log.error("not expected", cause=e)
elif hasattr(value, '__data__'):
d = value.__data__()
return pretty_json(d)
elif hasattr(value, '__json__'):
j = value.__json__()
if j == None:
return " null " # TODO: FIND OUT WHAT CAUSES THIS
return pretty_json(json_decoder(j))
elif scrub(value) is None:
return "null"
elif hasattr(value, '__iter__'):
return pretty_json(list(value))
elif hasattr(value, '__call__'):
return "null"
else:
try:
if int(value) == value:
return text_type(int(value))
except Exception:
pass
try:
if float(value) == value:
return text_type(float(value))
except Exception:
pass
return pypy_json_encode(value)
except Exception as e:
problem_serializing(value, e)
def query(self, query):
"""
:param query: JSON Query Expression, SET `format="container"` TO MAKE NEW TABLE OF RESULT
:return:
"""
if not startswith_field(query['from'], self.sf.fact):
Log.error("Expecting table, or some nested table")
frum, query['from'] = query['from'], self
table = self.sf.tables[relative_field(frum, self.sf.fact)]
schema = table.schema
query = QueryOp.wrap(query, table=table, schema=schema)
new_table = "temp_" + unique_name()
if query.format == "container":
create_table = "CREATE TABLE " + quote_column(new_table) + " AS "
else:
create_table = ""
if query.groupby and query.format != "cube":
op, index_to_columns = self._groupby_op(query, frum)
command = create_table + op
elif query.groupby:
query.edges, query.groupby = query.groupby, query.edges
op, index_to_columns = self._edges_op(query, frum)
command = create_table + op
query.edges, query.groupby = query.groupby, query.edges
elif query.edges or any(a != "none" for a in listwrap(query.select).aggregate):
op, index_to_columns = self._edges_op(query, frum)
command = create_table + op
else:
op = self._set_op(query, frum)
return op
result = self.db.query(command)
if query.format == "container":
output = QueryTable(new_table, db=self.db, uid=self.uid, exists=True)
elif query.format == "cube" or (not query.format and query.edges):
column_names = [None] * (max(c.push_column for c in index_to_columns.values()) + 1)
for c in index_to_columns.values():
column_names[c.push_column] = c.push_column_name
if len(query.edges) == 0 and len(query.groupby) == 0:
data = {n: Data() for n in column_names}
for s in index_to_columns.values():
data[s.push_name][s.push_child] = unwrap(s.pull(result.data[0]))
if isinstance(query.select, list):
select = [{"name": s.name} for s in query.select]
else:
select = {"name": query.select.name}
return Data(
data=unwrap(data),
select=select,
meta={"format": "cube"}
)
if not result.data:
edges = []
dims = []
for i, e in enumerate(query.edges + query.groupby):
allowNulls = coalesce(e.allowNulls, True)
if e.domain.type == "set" and e.domain.partitions:
domain = SimpleSetDomain(partitions=e.domain.partitions.name)
elif e.domain.type == "range":
domain = e.domain
elif isinstance(e.value, TupleOp):
pulls = jx.sort([c for c in index_to_columns.values() if c.push_name == e.name],
"push_child").pull
parts = [tuple(p(d) for p in pulls) for d in result.data]
domain = SimpleSetDomain(partitions=jx.sort(set(parts)))
else:
domain = SimpleSetDomain(partitions=[])
dims.append(1 if allowNulls else 0)
edges.append(Data(
name=e.name,
allowNulls=allowNulls,
domain=domain
))
data = {}
for si, s in enumerate(listwrap(query.select)):
if s.aggregate == "count":
data[s.name] = Matrix(dims=dims, zeros=0)
else:
data[s.name] = Matrix(dims=dims)
if isinstance(query.select, list):
select = [{"name": s.name} for s in query.select]
else:
select = {"name": query.select.name}
return Data(
meta={"format": "cube"},
edges=edges,
select=select,
data={k: v.cube for k, v in data.items()}
)
columns = None
edges = []
dims = []
for g in query.groupby:
g.is_groupby = True
for i, e in enumerate(query.edges + query.groupby):
allowNulls = coalesce(e.allowNulls, True)
if e.domain.type == "set" and e.domain.partitions:
domain = SimpleSetDomain(partitions=e.domain.partitions.name)
elif e.domain.type == "range":
domain = e.domain
elif e.domain.type == "time":
domain = wrap(mo_json.scrub(e.domain))
elif e.domain.type == "duration":
domain = wrap(mo_json.scrub(e.domain))
elif isinstance(e.value, TupleOp):
pulls = jx.sort([c for c in index_to_columns.values() if c.push_name == e.name], "push_child").pull
parts = [tuple(p(d) for p in pulls) for d in result.data]
domain = SimpleSetDomain(partitions=jx.sort(set(parts)))
else:
if not columns:
columns = zip(*result.data)
parts = set(columns[i])
if e.is_groupby and None in parts:
allowNulls = True
parts -= {None}
if query.sort[i].sort == -1:
domain = SimpleSetDomain(partitions=wrap(sorted(parts, reverse=True)))
else:
domain = SimpleSetDomain(partitions=jx.sort(parts))
dims.append(len(domain.partitions) + (1 if allowNulls else 0))
edges.append(Data(
name=e.name,
allowNulls=allowNulls,
domain=domain
))
data_cubes = {}
for si, s in enumerate(listwrap(query.select)):
if s.aggregate == "count":
data_cubes[s.name] = Matrix(dims=dims, zeros=0)
else:
data_cubes[s.name] = Matrix(dims=dims)
r2c = index_to_coordinate(dims) # WORKS BECAUSE THE DATABASE SORTED THE EDGES TO CONFORM
for rownum, row in enumerate(result.data):
coord = r2c(rownum)
for i, s in enumerate(index_to_columns.values()):
if s.is_edge:
continue
if s.push_child == ".":
data_cubes[s.push_name][coord] = s.pull(row)
else:
data_cubes[s.push_name][coord][s.push_child] = s.pull(row)
if query.select == None:
select = Null
elif isinstance(query.select, list):
select = [{"name": s.name} for s in query.select]
else:
select = {"name": query.select.name}
return Data(
meta={"format": "cube"},
edges=edges,
select=select,
data={k: v.cube for k, v in data_cubes.items()}
)
elif query.format == "table" or (not query.format and query.groupby):
column_names = [None] * (max(c.push_column for c in index_to_columns.values()) + 1)
for c in index_to_columns.values():
column_names[c.push_column] = c.push_column_name
data = []
for d in result.data:
row = [None for _ in column_names]
for s in index_to_columns.values():
if s.push_child == ".":
row[s.push_column] = s.pull(d)
elif s.num_push_columns:
tuple_value = row[s.push_column]
if tuple_value == None:
tuple_value = row[s.push_column] = [None] * s.num_push_columns
tuple_value[s.push_child] = s.pull(d)
elif row[s.push_column] == None:
row[s.push_column] = Data()
row[s.push_column][s.push_child] = s.pull(d)
else:
row[s.push_column][s.push_child] = s.pull(d)
data.append(tuple(unwrap(r) for r in row))
output = Data(
meta={"format": "table"},
header=column_names,
data=data
)
elif query.format == "list" or (not query.edges and not query.groupby):
if not query.edges and not query.groupby and any(listwrap(query.select).aggregate):
if isinstance(query.select, list):
data = Data()
for c in index_to_columns.values():
if c.push_child == ".":
if data[c.push_name] == None:
data[c.push_name] = c.pull(result.data[0])
elif isinstance(data[c.push_name], list):
data[c.push_name].append(c.pull(result.data[0]))
else:
data[c.push_name] = [data[c.push_name], c.pull(result.data[0])]
else:
data[c.push_name][c.push_child] = c.pull(result.data[0])
output = Data(
meta={"format": "value"},
data=data
)
else:
data = Data()
for s in index_to_columns.values():
if not data[s.push_child]:
data[s.push_child] = s.pull(result.data[0])
else:
data[s.push_child] += [s.pull(result.data[0])]
output = Data(
meta={"format": "value"},
data=unwrap(data)
)
else:
data = []
for rownum in result.data:
row = Data()
for c in index_to_columns.values():
if c.push_child == ".":
row[c.push_name] = c.pull(rownum)
elif c.num_push_columns:
tuple_value = row[c.push_name]
if not tuple_value:
tuple_value = row[c.push_name] = [None] * c.num_push_columns
tuple_value[c.push_child] = c.pull(rownum)
else:
row[c.push_name][c.push_child] = c.pull(rownum)
data.append(row)
output = Data(
meta={"format": "list"},
data=data
)
else:
Log.error("unknown format {{format}}", format=query.format)
return output
def pretty_json(value):
try:
if value is False:
return "false"
elif value is True:
return "true"
elif is_data(value):
try:
items = sort_using_key(value.items(), lambda r: r[0])
values = [encode_basestring(k) + PRETTY_COLON + pretty_json(v) for k, v in items if v != None]
if not values:
return "{}"
elif len(values) == 1:
return "{" + values[0] + "}"
else:
return "{\n" + ",\n".join(indent(v) for v in values) + "\n}"
except Exception as e:
from mo_logs import Log
from mo_math import OR
if OR(not is_text(k) for k in value.keys()):
Log.error(
"JSON must have string keys: {{keys}}:",
keys=[k for k in value.keys()],
cause=e
)
Log.error(
"problem making dict pretty: keys={{keys}}:",
keys=[k for k in value.keys()],
cause=e
)
elif value in (None, Null):
return "null"
elif value.__class__ in (binary_type, text_type):
if is_binary(value):
value = utf82unicode(value)
try:
return quote(value)
except Exception as e:
from mo_logs import Log
try:
Log.note("try explicit convert of string with length {{length}}", length=len(value))
acc = [QUOTE]
for c in value:
try:
try:
c2 = ESCAPE_DCT[c]
except Exception:
c2 = c
c3 = text_type(c2)
acc.append(c3)
except BaseException:
pass
# Log.warning("odd character {{ord}} found in string. Ignored.", ord= ord(c)}, cause=g)
acc.append(QUOTE)
output = u"".join(acc)
Log.note("return value of length {{length}}", length=len(output))
return output
except BaseException as f:
Log.warning("can not convert {{type}} to json", type=f.__class__.__name__, cause=f)
return "null"
elif is_list(value):
if not value:
return "[]"
if ARRAY_MAX_COLUMNS == 1:
return "[\n" + ",\n".join([indent(pretty_json(v)) for v in value]) + "\n]"
if len(value) == 1:
j = pretty_json(value[0])
if j.find("\n") >= 0:
return "[\n" + indent(j) + "\n]"
else:
return "[" + j + "]"
js = [pretty_json(v) for v in value]
max_len = max(*[len(j) for j in js])
if max_len <= ARRAY_ITEM_MAX_LENGTH and max(*[j.find("\n") for j in js]) == -1:
# ALL TINY VALUES
num_columns = max(1, min(ARRAY_MAX_COLUMNS, int(floor((ARRAY_ROW_LENGTH + 2.0) / float(max_len + 2))))) # +2 TO COMPENSATE FOR COMMAS
if len(js) <= num_columns: # DO NOT ADD \n IF ONLY ONE ROW
return "[" + PRETTY_COMMA.join(js) + "]"
if num_columns == 1: # DO NOT rjust IF THERE IS ONLY ONE COLUMN
return "[\n" + ",\n".join([indent(pretty_json(v)) for v in value]) + "\n]"
content = ",\n".join(
PRETTY_COMMA.join(j.rjust(max_len) for j in js[r:r + num_columns])
for r in xrange(0, len(js), num_columns)
)
return "[\n" + indent(content) + "\n]"
pretty_list = js
output = ["[\n"]
for i, p in enumerate(pretty_list):
try:
if i > 0:
output.append(",\n")
output.append(indent(p))
except Exception:
from mo_logs import Log
Log.warning("problem concatenating string of length {{len1}} and {{len2}}",
len1=len("".join(output)),
len2=len(p)
)
output.append("\n]")
try:
return "".join(output)
except Exception as e:
from mo_logs import Log
Log.error("not expected", cause=e)
elif hasattr(value, '__data__'):
d = value.__data__()
return pretty_json(d)
elif hasattr(value, '__json__'):
j = value.__json__()
if j == None:
return " null " # TODO: FIND OUT WHAT CAUSES THIS
return pretty_json(json_decoder(j))
elif scrub(value) is None:
return "null"
elif hasattr(value, '__iter__'):
return pretty_json(list(value))
elif hasattr(value, '__call__'):
return "null"
else:
try:
if int(value) == value:
return text_type(int(value))
except Exception:
pass
try:
if float(value) == value:
return text_type(float(value))
except Exception:
pass
return pypy_json_encode(value)
except Exception as e:
problem_serializing(value, e)
def process(
sig_id,
since,
source,
destination,
):
"""
:param sig_id: The performance hash
:param since: Only data after this date
:param show:
:param show_limit:
:param show_old:
:param show_distribution:
:return:
"""
if not isinstance(sig_id, int):
Log.error("expecting id")
# GET SIGNATURE DETAILS
sig = get_signature(source, sig_id)
# GET SIGNATURE DETAILS
pushes = get_dataum(source, sig_id, since, LIMIT)
pushes = jx.sort(
[{
"value": median(rows.value),
"runs": rows,
"push": {
"time": unwrap(t)["push.time"]
},
} for t, rows in jx.groupby(pushes, "push.time")
if t["push\\.time"] > since],
"push.time",
)
values = list(pushes.value)
title = "-".join(
map(
str,
[
sig.framework,
sig.suite,
sig.test,
sig.platform,
sig.repository,
],
))
Log.note("With {{title}}", title=title)
if len(values) > LIMIT:
Log.alert(
"Too many values for {{title}} ({at least {num}}), choosing last {{limit}}",
title=title,
num=len(values),
limit=LIMIT,
)
values = values[-LIMIT:]
with Timer("find segments"):
new_segments, new_diffs = find_segments(values, sig.alert_change_type,
sig.alert_threshold)
if len(new_segments) == 1:
overall_dev_status = None
overall_dev_score = None
last_mean = None
last_std = None
last_dev_status = None
last_dev_score = None
relative_noise = None
else:
# NOISE OF LAST SEGMENT
s, e = new_segments[-2], new_segments[-1]
last_segment = np.array(values[s:e])
trimmed_segment = last_segment
last_mean = np.mean(trimmed_segment)
last_std = np.std(trimmed_segment)
last_dev_status, last_dev_score = deviance(trimmed_segment)
relative_noise = last_std / last_mean
# FOR EACH SEGMENT, NORMALIZE MEAN AND VARIANCE
normalized = []
for s, e in jx.pairs(new_segments):
data = np.array(values[s:e])
norm = (data + last_mean - np.mean(data)) * last_std / np.std(data)
normalized.extend(norm)
overall_dev_status, overall_dev_score = deviance(normalized)
Log.note(
"\n\tdeviance = {{deviance}}\n\tnoise={{std}}\n\tpushes={{pushes}}\n\tsegments={{num_segments}}",
title=title,
deviance=(overall_dev_status, overall_dev_score),
std=relative_noise,
pushes=len(values),
num_segments=len(new_segments) - 1,
)
destination.add(
Data(
id=sig_id,
title=title,
num_pushes=len(values),
num_segments=len(new_segments) - 1,
relative_noise=relative_noise,
overall_dev_status=overall_dev_status,
overall_dev_score=overall_dev_score,
last_mean=last_mean,
last_std=last_std,
last_dev_status=last_dev_status,
last_dev_score=last_dev_score,
last_updated=Date.now(),
values=values,
)
| scrub(sig))
def json_encode(value):
"""
FOR PUTTING JSON INTO DATABASE (sort_keys=True)
dicts CAN BE USED AS KEYS
"""
return text(utf8_json_encoder(mo_json.scrub(value)))
output.append("\n]")
try:
return "".join(output)
except Exception as e:
from mo_logs import Log
Log.error("not expected", cause=e)
elif hasattr(value, '__data__'):
d = value.__data__()
return pretty_json(d)
elif hasattr(value, '__json__'):
j = value.__json__()
if j == None:
return " null " # TODO: FIND OUT WHAT CAUSES THIS
return pretty_json(json_decoder(j))
elif scrub(value) is None:
return "null"
elif hasattr(value, '__iter__'):
return pretty_json(list(value))
elif hasattr(value, '__call__'):
return "null"
else:
try:
if int(value) == value:
return str(int(value))
except Exception:
pass
try:
if float(value) == value:
return str(float(value))
output.append("\n]")
try:
return "".join(output)
except Exception as e:
from mo_logs import Log
Log.error("not expected", cause=e)
elif hasattr(value, '__data__'):
d = value.__data__()
return pretty_json(d)
elif hasattr(value, '__json__'):
j = value.__json__()
if j == None:
return " null " # TODO: FIND OUT WHAT CAUSES THIS
return pretty_json(json_decoder(j))
elif scrub(value) is None:
return "null"
elif hasattr(value, '__iter__'):
return pretty_json(list(value))
elif hasattr(value, '__call__'):
return "null"
else:
try:
if int(value) == value:
return str(int(value))
except Exception:
pass
try:
if float(value) == value:
return str(float(value))
def json_encode(value):
"""
FOR PUTTING JSON INTO DATABASE (sort_keys=True)
dicts CAN BE USED AS KEYS
"""
return text_type(utf8_json_encoder(mo_json.scrub(value)))
def json_encode(value):
"""
FOR PUTTING JSON INTO DATABASE (sort_keys=True)
dicts CAN BE USED AS KEYS
"""
return unicode(json_encoder.encode(mo_json.scrub(value)))
def test_minus_inf(self):
test = float("-inf")
output = value2json(test)
expecting = cPythonJSONEncoder().encode(mo_json.scrub(test))
self.assertEqual(output, expecting, "expecting " + expecting)
