def tuple(data, field_name):
"""
RETURN LIST OF TUPLES
"""
if isinstance(data, Cube):
Log.error("not supported yet")
if isinstance(data, FlatList):
Log.error("not supported yet")
if isinstance(field_name, Mapping) and "value" in field_name:
# SIMPLIFY {"value":value} AS STRING
field_name = field_name["value"]
# SIMPLE PYTHON ITERABLE ASSUMED
if isinstance(field_name, basestring):
if len(split_field(field_name)) == 1:
return [(d[field_name], ) for d in data]
else:
path = split_field(field_name)
output = []
flat_list._tuple1(data, path, 0, output)
return output
elif isinstance(field_name, list):
paths = [_select_a_field(f) for f in field_name]
output = FL()
_tuple((), unwrap(data), paths, 0, output)
return output
else:
paths = [_select_a_field(field_name)]
output = FL()
_tuple((), data, paths, 0, output)
return output
def select(self, selectList, fromPath, varName, sourceVar):
path = split_field(fromPath)
is_deep = len(path) > 1
heads = []
list = []
for s in selectList:
if is_deep:
if s.value and isKeyword(s.value):
shortForm = self._translate(s.value)
list.append("Value2Pipe(" + shortForm + ")\n")
else:
Log.error("do not know how to handle yet")
else:
if s.value and isKeyword(s.value):
list.append("Value2Pipe(getDocValue(" + value2MVEL(s.value) + "))\n")
elif s.value:
shortForm = self._translate(s.value)
list.append("Value2Pipe(" + shortForm + ")\n")
else:
code, decode = self.Parts2Term(s.domain)
heads.append(code.head)
list.append("Value2Pipe(" + code.body + ")\n")
if len(split_field(fromPath)) > 1:
output = 'if (' + varName + ' != "") ' + varName + '+="|";\n' + varName + '+=' + '+"|"+'.join(["Value2Pipe("+v+")\n" for v in list]) + ';\n'
else:
output = varName + ' = ' + '+"|"+'.join(["Value2Pipe("+v+")\n" for v in list]) + ';\n'
return Data(
head="".join(heads),
body=output
)
def _select_a_field(field):
if isinstance(field, basestring):
return wrap({"name": field, "value": split_field(field)})
elif isinstance(wrap(field).value, basestring):
field = wrap(field)
return wrap({"name": field.name, "value": split_field(field.value)})
else:
return wrap({"name": field.name, "value": field.value})
def select(self, fields):
if isinstance(fields, Mapping):
fields=fields.value
if isinstance(fields, basestring):
# RETURN LIST OF VALUES
if len(split_field(fields)) == 1:
if self.path[0] == fields:
return [d[1] for d in self.data]
else:
return [d[0][fields] for d in self.data]
else:
keys = split_field(fields)
depth = coalesce(MIN([i for i, (k, p) in enumerate(zip(keys, self.path)) if k != p]), len(self.path)) # LENGTH OF COMMON PREFIX
short_key = keys[depth:]
output = FlatList()
_select1((wrap(d[depth]) for d in self.data), short_key, 0, output)
return output
if isinstance(fields, list):
output = FlatList()
meta = []
for f in fields:
if hasattr(f.value, "__call__"):
meta.append((f.name, f.value))
else:
meta.append((f.name, functools.partial(lambda v, d: d[v], f.value)))
for row in self._values():
agg = Data()
for name, f in meta:
agg[name] = f(row)
output.append(agg)
return output
# meta = []
# for f in fields:
# keys = split_field(f.value)
# depth = coalesce(MIN([i for i, (k, p) in enumerate(zip(keys, self.path)) if k != p]), len(self.path)) # LENGTH OF COMMON PREFIX
# short_key = join_field(keys[depth:])
#
# meta.append((f.name, depth, short_key))
#
# for row in self._data:
# agg = Data()
# for name, depth, short_key in meta:
# if short_key:
# agg[name] = row[depth][short_key]
# else:
# agg[name] = row[depth]
# output.append(agg)
# return output
Log.error("multiselect over FlatList not supported")
def add_column(c, query_path):
c.last_updated = Date.now()
c.table = join_field([c.es_index]+split_field(query_path[0]))
with self.meta.columns.locker:
self._upsert_column(c)
for alias in meta.aliases:
c = copy(c)
c.table = join_field([alias]+split_field(query_path[0]))
self._upsert_column(c)
def defParent(name):
# DO NOT MAKE THE SAME PARENT TWICE
if name in parentVarNames:
return
parentVarNames.add(name)
if len(split_field(name)) == 1:
contextVariables.append("Map " + name + " = new HashMap();\n")
else:
defParent(join_field(split_field(name)[0:-1]))
contextVariables.append(name + " = new HashMap();\n")
def getFrameVariables(self, body):
contextVariables = []
columns = self.fromData.columns
parentVarNames = set() # ALL PARENTS OF VARIABLES WITH "." IN NAME
body = body.replace(".?", ".")
for i, c in enumerate(columns):
j = body.find(c.name, 0)
while j >= 0:
s = j
j = body.find(c.name, s + 1)
test0 = body[s - 1: s + len(c.name) + 1:]
test3 = body[s - 8: s + len(c.name):]
if test0[:-1] == "\"" + c.name:
continue
if test3 == "_source." + c.name:
continue
def defParent(name):
# DO NOT MAKE THE SAME PARENT TWICE
if name in parentVarNames:
return
parentVarNames.add(name)
if len(split_field(name)) == 1:
contextVariables.append("Map " + name + " = new HashMap();\n")
else:
defParent(join_field(split_field(name)[0:-1]))
contextVariables.append(name + " = new HashMap();\n")
body = body.replace(c.name, "-"*len(c.name))
if self.isLean or c.useSource:
if len(split_field(c.name)) > 1:
defParent(join_field(split_field(c.name)[0:-1]))
contextVariables.append(c.name + " = getSourceValue(\"" + c.name + "\");\n")
else:
contextVariables.append(c.name + " = _source[\"" + c.name + "\"];\n")
else:
if len(split_field(c.name)) > 1:
defParent(join_field(split_field(c.name)[0:-1]))
contextVariables.append(c.name + " = getDocValue(\"" + c.name + "\");\n")
else:
contextVariables.append(c.name + " = getDocValue(\"" + c.name + "\");\n")
break
return "".join(contextVariables)
def parse_field(fieldname, data, depth):
"""
RETURN (first, rest) OF fieldname
"""
col = split_field(fieldname)
d = data
for i, c in enumerate(col):
try:
d = d[c]
except Exception, e:
Log.error("{{name}} does not exist", name=fieldname)
if isinstance(d, list) and len(col) > 1:
if len(primary_column) <= depth+i:
primary_nested.append(True)
primary_column.append(c)
primary_branch.append(d)
elif primary_nested[depth] and primary_column[depth+i] != c:
Log.error("only one branch of tree allowed")
else:
primary_nested[depth+i] = True
primary_column[depth+i] = c
primary_branch[depth+i] = d
return c, join_field(col[i+1:])
else:
if len(primary_column) <= depth+i:
primary_nested.append(False)
primary_column.append(c)
primary_branch.append([d])
def get_columns(self, table_name, column_name=None, force=False):
"""
RETURN METADATA COLUMNS
"""
try:
# LAST TIME WE GOT INFO FOR THIS TABLE
short_name = join_field(split_field(table_name)[0:1])
table = self.get_table(short_name)[0]
if not table:
table = Table(
name=short_name,
url=None,
query_path=None,
timestamp=Date.now()
)
with self.meta.tables.locker:
self.meta.tables.add(table)
self._get_columns(table=short_name)
elif force or table.timestamp == None or table.timestamp < Date.now() - MAX_COLUMN_METADATA_AGE:
table.timestamp = Date.now()
self._get_columns(table=short_name)
with self.meta.columns.locker:
columns = self.meta.columns.find(table_name, column_name)
if columns:
columns = jx.sort(columns, "name")
# AT LEAST WAIT FOR THE COLUMNS TO UPDATE
while len(self.todo) and not all(columns.get("last_updated")):
Log.note("waiting for columns to update {{columns|json}}", columns=[c.table+"."+c.es_column for c in columns if not c.last_updated])
Till(seconds=1).wait()
return columns
except Exception, e:
Log.error("Not expected", cause=e)
def getDomain(self, **kwargs):
# kwargs.depth IS MEANT TO REACH INTO SUB-PARTITIONS
kwargs = wrap(kwargs)
kwargs.depth = coalesce(kwargs.depth, len(self.fields)-1 if isinstance(self.fields, list) else None)
if not self.partitions and self.edges:
# USE EACH EDGE AS A PARTITION, BUT isFacet==True SO IT ALLOWS THE OVERLAP
partitions = [
{
"name": v.name,
"value": v.name,
"where": v.where,
"style": v.style,
"weight": v.weight # YO! WHAT DO WE *NOT* COPY?
}
for i, v in enumerate(self.edges)
if i < coalesce(self.limit, DEFAULT_QUERY_LIMIT) and v.where
]
self.isFacet = True
elif kwargs.depth == None: # ASSUME self.fields IS A dict
partitions = FlatList()
for i, part in enumerate(self.partitions):
if i >= coalesce(self.limit, DEFAULT_QUERY_LIMIT):
break
partitions.append({
"name":part.name,
"value":part.value,
"where":part.where,
"style":coalesce(part.style, part.parent.style),
"weight":part.weight # YO! WHAT DO WE *NOT* COPY?
})
elif kwargs.depth == 0:
partitions = [
{
"name":v.name,
"value":v.value,
"where":v.where,
"style":v.style,
"weight":v.weight # YO! WHAT DO WE *NOT* COPY?
}
for i, v in enumerate(self.partitions)
if i < coalesce(self.limit, DEFAULT_QUERY_LIMIT)]
elif kwargs.depth == 1:
partitions = FlatList()
rownum = 0
for i, part in enumerate(self.partitions):
if i >= coalesce(self.limit, DEFAULT_QUERY_LIMIT):
continue
rownum += 1
try:
for j, subpart in enumerate(part.partitions):
partitions.append({
"name":join_field(split_field(subpart.parent.name) + [subpart.name]),
"value":subpart.value,
"where":subpart.where,
"style":coalesce(subpart.style, subpart.parent.style),
"weight":subpart.weight # YO! WHAT DO WE *NOT* COPY?
})
except Exception, e:
Log.error("", e)
def setValues(expression, constants):
if not constants:
return expression
constants = constants.copy()
# EXPAND ALL CONSTANTS TO PRIMITIVE VALUES (MVEL CAN ONLY ACCEPT PRIMITIVE VALUES)
for c in constants:
value = c.value
n = c.name
if len(split_field(n)) >= 3:
continue # DO NOT GO TOO DEEP
if isinstance(value, list):
continue # DO NOT MESS WITH ARRAYS
if isinstance(value, Mapping):
for k, v in value.items():
constants.append({"name": n + "." + k, "value": v})
for c in reverse(constants):# REVERSE ORDER, SO LONGER NAMES ARE TESTED FIRST
s = 0
while True:
s = expression.find(c.name, s)
if s == -1:
break
if re.match(r"\w", expression[s - 1]):
break
if re.match(r"\w", expression[s + len(c.name)]):
break
v = value2MVEL(c.value)
expression = expression[:s:] + "" + v + expression[:s + len(c.name):]
return expression
def _get(v, k, d):
for p in split_field(k):
try:
v = v.get(p)
if v is None:
return d
except Exception:
v = [vv.get(p) for vv in v]
return v
def _get_schema_from_list(frum, columns, prefix, nested_path, name_to_column):
"""
SCAN THE LIST FOR COLUMN TYPES
"""
for d in frum:
row_type = _type_to_name[d.__class__]
if row_type != "object":
full_name = join_field(prefix)
column = name_to_column.get(full_name)
if not column:
column = Column(
name=full_name,
table=".",
es_column=full_name,
es_index=".",
type="undefined",
nested_path=nested_path
)
columns[full_name] = column
column.type = _merge_type[column.type][row_type]
else:
for name, value in d.items():
full_name = join_field(prefix + [name])
column = name_to_column.get(full_name)
if not column:
column = Column(
name=full_name,
table=".",
es_column=full_name,
es_index=".",
type="undefined",
nested_path=nested_path
)
columns[full_name] = column
if isinstance(value, list):
if len(value)==0:
this_type = "undefined"
elif len(value)==1:
this_type = _type_to_name[value[0].__class__]
else:
this_type = _type_to_name[value[0].__class__]
if this_type == "object":
this_type = "nested"
else:
this_type = _type_to_name[value.__class__]
new_type = _merge_type[column.type][this_type]
column.type = new_type
if this_type == "object":
_get_schema_from_list([value], columns, prefix + [name], nested_path, name_to_column)
elif this_type == "nested":
np = listwrap(nested_path)
newpath = unwraplist([join_field(split_field(np[0])+[name])]+np)
_get_schema_from_list(value, columns, prefix + [name], newpath, name_to_column)
def is_deepop(es, query):
if query.edges or query.groupby:
return False
if all(s.aggregate not in (None, "none") for s in listwrap(query.select)):
return False
if len(split_field(query.frum.name)) > 1:
return True
# ASSUME IT IS NESTED IF WE ARE ASKING FOR NESTED COLUMNS
# vars_ = query_get_all_vars(query)
# columns = query.frum.get_columns()
# if any(c for c in columns if len(c.nested_path) != 1 and c.name in vars_):
# return True
return False
def wrap_from(frum, schema=None):
"""
:param frum:
:param schema:
:return:
"""
if not _containers:
_delayed_imports()
frum = wrap(frum)
if isinstance(frum, basestring):
if not _containers.config.default.settings:
Log.error("expecting pyLibrary.queries.query.config.default.settings to contain default elasticsearch connection info")
type_ = None
index = frum
if frum.startswith("meta."):
if frum == "meta.columns":
return _meta.singlton.meta.columns
elif frum == "meta.tables":
return _meta.singlton.meta.tables
else:
Log.error("{{name}} not a recognized table", name=frum)
else:
type_ = _containers.config.default.type
index = join_field(split_field(frum)[:1:])
settings = set_default(
{
"index": index,
"name": frum
},
_containers.config.default.settings
)
settings.type = None
return _containers.type2container[type_](settings)
elif isinstance(frum, Mapping) and frum.type and _containers.type2container[frum.type]:
# TODO: Ensure the frum.name is set, so we capture the deep queries
if not frum.type:
Log.error("Expecting from clause to have a 'type' property")
return _containers.type2container[frum.type](frum.settings)
elif isinstance(frum, Mapping) and (frum["from"] or isinstance(frum["from"], (list, set))):
from pyLibrary.queries.query import QueryOp
return QueryOp.wrap(frum, schema=schema)
elif isinstance(frum, (list, set)):
return _ListContainer("test_list", frum)
else:
return frum
def is_deep(query):
select = listwrap(query.select)
if len(select) > 1:
return False
if aggregates[select[0].aggregate] not in ("none", "count"):
return False
if len(query.edges)<=1:
return False
isDeep = len(split_field(query["from"].name)) > 1 # LOOKING INTO NESTED WILL REQUIRE A SCRIPT
if not isDeep:
return False # BETTER TO USE TERM QUERY
return True
def is_setop(query):
select = listwrap(query.select)
if not query.edges:
isDeep = len(split_field(query.frum.name)) > 1 # LOOKING INTO NESTED WILL REQUIRE A SCRIPT
simpleAgg = AND([s.aggregate in ("count", "none") for s in select]) # CONVERTING esfilter DEFINED PARTS WILL REQUIRE SCRIPT
# NO EDGES IMPLIES SIMPLER QUERIES: EITHER A SET OPERATION, OR RETURN SINGLE AGGREGATE
if simpleAgg or isDeep:
return True
else:
isSmooth = AND((e.domain.type in domains.ALGEBRAIC and e.domain.interval == "none") for e in query.edges)
if isSmooth:
return True
return False
def is_fieldop(query):
# THESE SMOOTH EDGES REQUIRE ALL DATA (SETOP)
select = listwrap(query.select)
if not query.edges:
isDeep = len(split_field(query.frum.name)) > 1 # LOOKING INTO NESTED WILL REQUIRE A SCRIPT
isSimple = AND(s.value != None and (s.value == "*" or is_keyword(s.value)) for s in select)
noAgg = AND(s.aggregate == "none" for s in select)
if not isDeep and isSimple and noAgg:
return True
else:
isSmooth = AND((e.domain.type in domains.ALGEBRAIC and e.domain.interval == "none") for e in query.edges)
if isSmooth:
return True
return False
def set(constants):
"""
REACH INTO THE MODULES AND OBJECTS TO SET CONSTANTS.
THINK OF THIS AS PRIMITIVE DEPENDENCY INJECTION FOR MODULES.
USEFUL FOR SETTING DEBUG FLAGS.
"""
if not constants:
return
constants = wrap(constants)
for k, new_value in constants.leaves():
errors = []
try:
old_value = pyDots.set_attr(sys.modules, k, new_value)
continue
except Exception, e:
errors.append(e)
# ONE MODULE IS MISSING, THE CALLING MODULE
try:
caller_globals = sys._getframe(1).f_globals
caller_file = caller_globals["__file__"]
if not caller_file.endswith(".py"):
raise Exception("do not know how to handle non-python caller")
caller_module = caller_file[:-3].replace("/", ".")
path = split_field(k)
for i, p in enumerate(path):
if i == 0:
continue
prefix = join_field(path[:1])
name = join_field(path[i:])
if caller_module.endswith(prefix):
old_value = pyDots.set_attr(caller_globals, name, new_value)
if DEBUG:
from pyLibrary.debugs.logs import Log
Log.note("Changed {{module}}[{{attribute}}] from {{old_value}} to {{new_value}}",
module= prefix,
attribute= name,
old_value= old_value,
new_value= new_value)
break
except Exception, e:
errors.append[e]
def new_instance(type, frum, schema=None):
"""
Factory!
"""
if not type2container:
_delayed_imports()
if isinstance(frum, Container):
return frum
elif isinstance(frum, _Cube):
return frum
elif isinstance(frum, _Query):
return _run(frum)
elif isinstance(frum, (list, set, GeneratorType)):
return _ListContainer(frum)
elif isinstance(frum, basestring):
# USE DEFAULT STORAGE TO FIND Container
if not config.default.settings:
Log.error("expecting pyLibrary.queries.query.config.default.settings to contain default elasticsearch connection info")
settings = set_default(
{
"index": join_field(split_field(frum)[:1:]),
"name": frum,
},
config.default.settings
)
settings.type = None # WE DO NOT WANT TO INFLUENCE THE TYPE BECAUSE NONE IS IN THE frum STRING ANYWAY
return type2container["elasticsearch"](settings)
elif isinstance(frum, Mapping):
frum = wrap(frum)
if frum.type and type2container[frum.type]:
return type2container[frum.type](frum.settings)
elif frum["from"]:
frum = copy(frum)
frum["from"] = Container(frum["from"])
return _Query.wrap(frum)
else:
Log.error("Do not know how to handle {{frum|json}}", frum=frum)
else:
Log.error("Do not know how to handle {{type}}", type=frum.__class__.__name__)
def compile_expression(self, expression, constants=None):
# EXPAND EXPRESSION WITH ANY CONSTANTS
expression = setValues(expression, constants)
fromPath = self.fromData.name # FIRST NAME IS THE INDEX
indexName = join_field(split_field(fromPath)[:1:])
context = self.getFrameVariables(expression)
if context == "":
return addFunctions(expression).head+expression
func = UID()
code = addFunctions(context+expression)
output = code.head + \
'var ' + func + ' = function(' + indexName + '){\n' + \
context + \
expression + ";\n" + \
'};\n' + \
func + '(_source)\n'
return Compiled(output)
def select(data, field_name):
"""
return list with values from field_name
"""
if isinstance(data, Cube):
return data._select(_normalize_selects(field_name))
if isinstance(data, FL):
return data.select(field_name)
if isinstance(data, UniqueIndex):
data = data._data.values() # THE SELECT ROUTINE REQUIRES dicts, NOT Data WHILE ITERATING
if isinstance(data, Mapping):
return select_one(data, field_name)
if isinstance(field_name, Mapping):
field_name = wrap(field_name)
if field_name.value in ["*", "."]:
return data
if field_name.value:
# SIMPLIFY {"value":value} AS STRING
field_name = field_name.value
# SIMPLE PYTHON ITERABLE ASSUMED
if isinstance(field_name, basestring):
path = split_field(field_name)
if len(path) == 1:
return FL([d[field_name] for d in data])
else:
output = FL()
flat_list._select1(data, path, 0, output)
return output
elif isinstance(field_name, list):
keys = [_select_a_field(wrap(f)) for f in field_name]
return _select(Data(), unwrap(data), keys, 0)
else:
keys = [_select_a_field(field_name)]
return _select(Data(), unwrap(data), keys, 0)
def frum(self, fromPath, sourceVar, loopVariablePrefix):
"""
indexName NAME USED TO REFER TO HIGH LEVEL DOCUMENT
loopVariablePrefix PREFIX FOR LOOP VARIABLES
"""
loopCode = "if (<PATH> != null){ for(<VAR> : <PATH>){\n<CODE>\n}}\n"
self.prefixMap = []
code = "<CODE>"
path = split_field(fromPath)
# ADD LOCAL VARIABLES
columns = INDEX_CACHE[path[0]].columns
for i, c in enumerate(columns):
if c.name.find("\\.") >= 0:
self.prefixMap.insert(0, {
"path": c.name,
"variable": "get(" + sourceVar + ", \"" + c.name.replace("\\.", ".") + "\")"
})
else:
self.prefixMap.insert(0, {
"path": c.name,
"variable": sourceVar + ".?" + c.name
})
# ADD LOOP VARIABLES
currPath = []
# self.prefixMap.insert(0, {"path": path[0], "variable": path[0]})
for i, step in enumerate(path[1::]):
loopVariable = loopVariablePrefix + str(i)
currPath.append(step)
pathi = ".".join(currPath)
shortPath = self._translate(pathi)
self.prefixMap.insert(0, {"path": pathi, "variable": loopVariable})
loop = loopCode.replace("<VAR>", loopVariable).replace("<PATH>", shortPath)
code = code.replace("<CODE>", loop)
return code
def es_query_template(path):
"""
RETURN TEMPLATE AND PATH-TO-FILTER AS A 2-TUPLE
:param path:
:return:
"""
sub_path = split_field(path)[1:]
if sub_path:
f0 = {}
f1 = {}
output = wrap({
"filter": {"and": [
f0,
{"nested": {
"path": join_field(sub_path),
"filter": f1,
"inner_hits": {"size": 100000}
}}
]},
"from": 0,
"size": 0,
"sort": []
})
return output, wrap([f0, f1])
else:
f0 = {}
output = wrap({
"query": {"filtered": {
"filter": f0
}},
"from": 0,
"size": 0,
"sort": []
})
return output, wrap([f0])
def es_aggsop(es, frum, query):
select = wrap([s.copy() for s in listwrap(query.select)])
es_column_map = {c.name: unwraplist(c.es_column) for c in frum.schema.all_columns}
es_query = Data()
new_select = Data() #MAP FROM canonical_name (USED FOR NAMES IN QUERY) TO SELECT MAPPING
formula = []
for s in select:
if s.aggregate == "count" and isinstance(s.value, Variable) and s.value.var == ".":
s.pull = "doc_count"
elif isinstance(s.value, Variable):
if s.value.var == ".":
if frum.typed:
# STATISITCAL AGGS IMPLY $value, WHILE OTHERS CAN BE ANYTHING
if s.aggregate in NON_STATISTICAL_AGGS:
#TODO: HANDLE BOTH $value AND $objects TO COUNT
Log.error("do not know how to handle")
else:
s.value.var = "$value"
new_select["$value"] += [s]
else:
if s.aggregate in NON_STATISTICAL_AGGS:
#TODO: WE SHOULD BE ABLE TO COUNT, BUT WE MUST *OR* ALL LEAF VALUES TO DO IT
Log.error("do not know how to handle")
else:
Log.error('Not expecting ES to have a value at "." which {{agg}} can be applied', agg=s.aggregate)
elif s.aggregate == "count":
s.value = s.value.map(es_column_map)
new_select["count_"+literal_field(s.value.var)] += [s]
else:
s.value = s.value.map(es_column_map)
new_select[literal_field(s.value.var)] += [s]
else:
formula.append(s)
for canonical_name, many in new_select.items():
representative = many[0]
if representative.value.var == ".":
Log.error("do not know how to handle")
else:
field_name = representative.value.var
# canonical_name=literal_field(many[0].name)
for s in many:
if s.aggregate == "count":
es_query.aggs[literal_field(canonical_name)].value_count.field = field_name
s.pull = literal_field(canonical_name) + ".value"
elif s.aggregate == "median":
# ES USES DIFFERENT METHOD FOR PERCENTILES
key = literal_field(canonical_name + " percentile")
es_query.aggs[key].percentiles.field = field_name
es_query.aggs[key].percentiles.percents += [50]
s.pull = key + ".values.50\.0"
elif s.aggregate == "percentile":
# ES USES DIFFERENT METHOD FOR PERCENTILES
key = literal_field(canonical_name + " percentile")
if isinstance(s.percentile, basestring) or s.percetile < 0 or 1 < s.percentile:
Log.error("Expecting percentile to be a float from 0.0 to 1.0")
percent = Math.round(s.percentile * 100, decimal=6)
es_query.aggs[key].percentiles.field = field_name
es_query.aggs[key].percentiles.percents += [percent]
s.pull = key + ".values." + literal_field(unicode(percent))
elif s.aggregate == "cardinality":
# ES USES DIFFERENT METHOD FOR CARDINALITY
key = literal_field(canonical_name + " cardinality")
es_query.aggs[key].cardinality.field = field_name
s.pull = key + ".value"
elif s.aggregate == "stats":
# REGULAR STATS
stats_name = literal_field(canonical_name)
es_query.aggs[stats_name].extended_stats.field = field_name
# GET MEDIAN TOO!
median_name = literal_field(canonical_name + " percentile")
es_query.aggs[median_name].percentiles.field = field_name
es_query.aggs[median_name].percentiles.percents += [50]
s.pull = {
"count": stats_name + ".count",
"sum": stats_name + ".sum",
"min": stats_name + ".min",
"max": stats_name + ".max",
"avg": stats_name + ".avg",
"sos": stats_name + ".sum_of_squares",
"std": stats_name + ".std_deviation",
"var": stats_name + ".variance",
"median": median_name + ".values.50\.0"
}
elif s.aggregate == "union":
# USE TERMS AGGREGATE TO SIMULATE union
stats_name = literal_field(canonical_name)
es_query.aggs[stats_name].terms.field = field_name
es_query.aggs[stats_name].terms.size = Math.min(s.limit, MAX_LIMIT)
s.pull = stats_name + ".buckets.key"
else:
# PULL VALUE OUT OF THE stats AGGREGATE
es_query.aggs[literal_field(canonical_name)].extended_stats.field = field_name
s.pull = literal_field(canonical_name) + "." + aggregates1_4[s.aggregate]
for i, s in enumerate(formula):
canonical_name = literal_field(s.name)
abs_value = s.value.map(es_column_map)
if s.aggregate == "count":
es_query.aggs[literal_field(canonical_name)].value_count.script = abs_value.to_ruby()
s.pull = literal_field(canonical_name) + ".value"
elif s.aggregate == "median":
# ES USES DIFFERENT METHOD FOR PERCENTILES THAN FOR STATS AND COUNT
key = literal_field(canonical_name + " percentile")
es_query.aggs[key].percentiles.script = abs_value.to_ruby()
es_query.aggs[key].percentiles.percents += [50]
s.pull = key + ".values.50\.0"
elif s.aggregate == "percentile":
# ES USES DIFFERENT METHOD FOR PERCENTILES THAN FOR STATS AND COUNT
key = literal_field(canonical_name + " percentile")
percent = Math.round(s.percentile * 100, decimal=6)
es_query.aggs[key].percentiles.script = abs_value.to_ruby()
es_query.aggs[key].percentiles.percents += [percent]
s.pull = key + ".values." + literal_field(unicode(percent))
elif s.aggregate == "cardinality":
# ES USES DIFFERENT METHOD FOR CARDINALITY
key = canonical_name + " cardinality"
es_query.aggs[key].cardinality.script = abs_value.to_ruby()
s.pull = key + ".value"
elif s.aggregate == "stats":
# REGULAR STATS
stats_name = literal_field(canonical_name)
es_query.aggs[stats_name].extended_stats.script = abs_value.to_ruby()
# GET MEDIAN TOO!
median_name = literal_field(canonical_name + " percentile")
es_query.aggs[median_name].percentiles.script = abs_value.to_ruby()
es_query.aggs[median_name].percentiles.percents += [50]
s.pull = {
"count": stats_name + ".count",
"sum": stats_name + ".sum",
"min": stats_name + ".min",
"max": stats_name + ".max",
"avg": stats_name + ".avg",
"sos": stats_name + ".sum_of_squares",
"std": stats_name + ".std_deviation",
"var": stats_name + ".variance",
"median": median_name + ".values.50\.0"
}
elif s.aggregate=="union":
# USE TERMS AGGREGATE TO SIMULATE union
stats_name = literal_field(canonical_name)
es_query.aggs[stats_name].terms.script_field = abs_value.to_ruby()
s.pull = stats_name + ".buckets.key"
else:
# PULL VALUE OUT OF THE stats AGGREGATE
s.pull = canonical_name + "." + aggregates1_4[s.aggregate]
es_query.aggs[canonical_name].extended_stats.script = abs_value.to_ruby()
decoders = get_decoders_by_depth(query)
start = 0
vars_ = query.where.vars()
#<TERRIBLE SECTION> THIS IS WHERE WE WEAVE THE where CLAUSE WITH nested
split_where = split_expression_by_depth(query.where, schema=frum, map_=es_column_map)
if len(split_field(frum.name)) > 1:
if any(split_where[2::]):
Log.error("Where clause is too deep")
for d in decoders[1]:
es_query = d.append_query(es_query, start)
start += d.num_columns
if split_where[1]:
#TODO: INCLUDE FILTERS ON EDGES
filter_ = simplify_esfilter(AndOp("and", split_where[1]).to_esfilter())
es_query = Data(
aggs={"_filter": set_default({"filter": filter_}, es_query)}
)
es_query = wrap({
"aggs": {"_nested": set_default(
{
"nested": {
"path": frum.query_path
}
},
es_query
)}
})
else:
if any(split_where[1::]):
Log.error("Where clause is too deep")
for d in decoders[0]:
es_query = d.append_query(es_query, start)
start += d.num_columns
if split_where[0]:
#TODO: INCLUDE FILTERS ON EDGES
filter = simplify_esfilter(AndOp("and", split_where[0]).to_esfilter())
es_query = Data(
aggs={"_filter": set_default({"filter": filter}, es_query)}
)
# </TERRIBLE SECTION>
if not es_query:
es_query = wrap({"query": {"match_all": {}}})
es_query.size = 0
with Timer("ES query time") as es_duration:
result = es09.util.post(es, es_query, query.limit)
try:
format_time = Timer("formatting")
with format_time:
decoders = [d for ds in decoders for d in ds]
result.aggregations.doc_count = coalesce(result.aggregations.doc_count, result.hits.total) # IT APPEARS THE OLD doc_count IS GONE
formatter, groupby_formatter, aggop_formatter, mime_type = format_dispatch[query.format]
if query.edges:
output = formatter(decoders, result.aggregations, start, query, select)
elif query.groupby:
output = groupby_formatter(decoders, result.aggregations, start, query, select)
else:
output = aggop_formatter(decoders, result.aggregations, start, query, select)
output.meta.timing.formatting = format_time.duration
output.meta.timing.es_search = es_duration.duration
output.meta.content_type = mime_type
output.meta.es_query = es_query
return output
except Exception, e:
if query.format not in format_dispatch:
Log.error("Format {{format|quote}} not supported yet", format=query.format, cause=e)
Log.error("Some problem", e)
def __init__(self, dim, parent, jx):
dim = wrap(dim)
self.name = dim.name
self.parent = coalesce(parent)
self.full_name = join_field(split_field(self.parent.full_name)+[self.name])
self.edges = None # FOR NOW
dot.set_default(self, dim)
self.where = dim.where
self.type = coalesce(dim.type, "set")
self.limit = coalesce(dim.limit, DEFAULT_QUERY_LIMIT)
self.index = coalesce(dim.index, coalesce(parent, Null).index, jx.settings.index)
if not self.index:
Log.error("Expecting an index name")
# ALLOW ACCESS TO SUB-PART BY NAME (IF ONLY THERE IS NO NAME COLLISION)
self.edges = Data()
for e in listwrap(dim.edges):
new_e = Dimension(e, self, jx)
self.edges[new_e.full_name] = new_e
self.partitions = wrap(coalesce(dim.partitions, []))
parse_partition(self)
fields = coalesce(dim.field, dim.fields)
if not fields:
return # NO FIELDS TO SEARCH
elif isinstance(fields, Mapping):
self.fields = wrap(fields)
edges = wrap([{"name": k, "value": v, "allowNulls": False} for k, v in self.fields.items()])
else:
self.fields = listwrap(fields)
edges = wrap([{"name": f, "value": f, "index": i, "allowNulls": False} for i, f in enumerate(self.fields)])
if dim.partitions:
return # ALREADY HAVE PARTS
if self.type not in KNOWN - ALGEBRAIC:
return # PARTS OR TOO FUZZY (OR TOO NUMEROUS) TO FETCH
jx.get_columns()
with Timer("Get parts of {{name}}", {"name": self.name}):
parts = jx.query({
"from": self.index,
"select": {"name": "count", "aggregate": "count"},
"edges": edges,
"where": self.where,
"limit": self.limit
})
Log.note("{{name}} has {{num}} parts", name= self.name, num= len(parts))
d = parts.edges[0].domain
if dim.path:
if len(edges) > 1:
Log.error("Not supported yet")
# EACH TERM RETURNED IS A PATH INTO A PARTITION TREE
temp = Data(partitions=[])
for i, count in enumerate(parts):
a = dim.path(d.getEnd(d.partitions[i]))
if not isinstance(a, list):
Log.error("The path function on " + dim.name + " must return an ARRAY of parts")
addParts(
temp,
dim.path(d.getEnd(d.partitions[i])),
count,
0
)
self.value = coalesce(dim.value, "name")
self.partitions = temp.partitions
elif isinstance(fields, Mapping):
self.value = "name" # USE THE "name" ATTRIBUTE OF PARTS
partitions = FlatList()
for g, p in parts.groupby(edges):
if p:
partitions.append({
"value": g,
"where": {"and": [
{"term": {e.value: g[e.name]}}
for e in edges
]},
"count": int(p)
})
self.partitions = partitions
elif len(edges) == 1:
self.value = "name" # USE THE "name" ATTRIBUTE OF PARTS
# SIMPLE LIST OF PARTS RETURNED, BE SURE TO INTERRELATE THEM
self.partitions = wrap([
{
"name": str(d.partitions[i].name), # CONVERT TO STRING
"value": d.getEnd(d.partitions[i]),
"where": {"term": {edges[0].value: d.partitions[i].value}},
"count": count
}
for i, count in enumerate(parts)
])
self.order = {p.value: i for i, p in enumerate(self.partitions)}
elif len(edges) == 2:
self.value = "name" # USE THE "name" ATTRIBUTE OF PARTS
d2 = parts.edges[1].domain
# SIMPLE LIST OF PARTS RETURNED, BE SURE TO INTERRELATE THEM
array = parts.data.values()[0].cube # DIG DEEP INTO RESULT (ASSUME SINGLE VALUE CUBE, WITH NULL AT END)
def edges2value(*values):
if isinstance(fields, Mapping):
output = Data()
for e, v in zip(edges, values):
output[e.name] = v
return output
else:
return tuple(values)
self.partitions = wrap([
{
"name": str(d.partitions[i].name), # CONVERT TO STRING
"value": d.getEnd(d.partitions[i]),
"where": {"term": {edges[0].value: d.partitions[i].value}},
"count": SUM(subcube),
"partitions": [
{
"name": str(d2.partitions[j].name), # CONVERT TO STRING
"value": edges2value(d.getEnd(d.partitions[i]), d2.getEnd(d2.partitions[j])),
"where": {"and": [
{"term": {edges[0].value: d.partitions[i].value}},
{"term": {edges[1].value: d2.partitions[j].value}}
]},
"count": count2
}
for j, count2 in enumerate(subcube)
if count2 > 0 # ONLY INCLUDE PROPERTIES THAT EXIST
]
}
for i, subcube in enumerate(array)
])
else:
Log.error("Not supported")
parse_partition(self) # RELATE THE PARTS TO THE PARENTS
def extract_rows(es, es_query, query):
is_list = isinstance(query.select, list)
selects = wrap([s.copy() for s in listwrap(query.select)])
new_select = FlatList()
columns = query.frum.get_columns()
leaf_columns = set(c.name for c in columns if c.type not in STRUCT and (c.nested_path[0] == "." or c.es_column == c.nested_path))
nested_columns = set(c.name for c in columns if len(c.nested_path) != 1)
i = 0
source = "fields"
for select in selects:
# IF THERE IS A *, THEN INSERT THE EXTRA COLUMNS
if isinstance(select.value, LeavesOp):
term = select.value.term
if isinstance(term, Variable):
if term.var == ".":
es_query.fields = None
source = "_source"
net_columns = leaf_columns - set(selects.name)
for n in net_columns:
new_select.append({
"name": n,
"value": Variable(n),
"put": {"name": n, "index": i, "child": "."}
})
i += 1
else:
parent = term.var + "."
prefix = len(parent)
for c in leaf_columns:
if c.startswith(parent):
if es_query.fields is not None:
es_query.fields.append(c)
new_select.append({
"name": select.name + "." + c[prefix:],
"value": Variable(c),
"put": {"name": select.name + "." + c[prefix:], "index": i, "child": "."}
})
i += 1
elif isinstance(select.value, Variable):
if select.value.var == ".":
es_query.fields = None
source = "_source"
new_select.append({
"name": select.name,
"value": select.value,
"put": {"name": select.name, "index": i, "child": "."}
})
i += 1
elif select.value.var == "_id":
new_select.append({
"name": select.name,
"value": select.value,
"pull": "_id",
"put": {"name": select.name, "index": i, "child": "."}
})
i += 1
elif select.value.var in nested_columns or [c for c in nested_columns if c.startswith(select.value.var+".")]:
es_query.fields = None
source = "_source"
new_select.append({
"name": select.name,
"value": select.value,
"put": {"name": select.name, "index": i, "child": "."}
})
i += 1
else:
parent = select.value.var + "."
prefix = len(parent)
net_columns = [c for c in leaf_columns if c.startswith(parent)]
if not net_columns:
# LEAF
if es_query.fields is not None:
es_query.fields.append(select.value.var)
new_select.append({
"name": select.name,
"value": select.value,
"put": {"name": select.name, "index": i, "child": "."}
})
else:
# LEAVES OF OBJECT
for n in net_columns:
if es_query.fields is not None:
es_query.fields.append(n)
new_select.append({
"name": select.name,
"value": Variable(n),
"put": {"name": select.name, "index": i, "child": n[prefix:]}
})
i += 1
else:
es_query.script_fields[literal_field(select.name)] = {"script": select.value.to_ruby()}
new_select.append({
"name": select.name,
"pull": "fields." + literal_field(select.name),
"put": {"name": select.name, "index": i, "child": "."}
})
i += 1
for n in new_select:
if n.pull:
continue
if source == "_source":
n.pull = join_field(["_source"] + split_field(n.value.var))
elif isinstance(n.value, Variable):
n.pull = "fields." + literal_field(n.value.var)
else:
Log.error("Do not know what to do")
with Timer("call to ES") as call_timer:
data = es09.util.post(es, es_query, query.limit)
T = data.hits.hits
try:
formatter, groupby_formatter, mime_type = format_dispatch[query.format]
output = formatter(T, new_select, query)
output.meta.timing.es = call_timer.duration
output.meta.content_type = mime_type
output.meta.es_query = es_query
return output
except Exception, e:
Log.error("problem formatting", e)
def query_path(self):
return join_field(split_field(self.name)[1:])
def Parts2Term(self, domain):
"""
TERMS ARE ALWAYS ESCAPED SO THEY CAN BE COMPOUNDED WITH PIPE (|)
CONVERT AN ARRAY OF PARTS{name, esfilter} TO AN MVEL EXPRESSION
RETURN expression, function PAIR, WHERE
expression - MVEL EXPRESSION
function - TAKES RESULT OF expression AND RETURNS PART
"""
fields = domain.dimension.fields
term = []
if len(split_field(self.fromData.name)) == 1 and fields:
if isinstance(fields, Mapping):
# CONVERT UNORDERED FIELD DEFS
jx_fields, es_fields = zip(*[(k, fields[k]) for k in sorted(fields.keys())])
else:
jx_fields, es_fields = zip(*[(i, e) for i, e in enumerate(fields)])
# NO LOOPS BECAUSE QUERY IS SHALLOW
# DOMAIN IS FROM A DIMENSION, USE IT'S FIELD DEFS TO PULL
if len(es_fields) == 1:
def fromTerm(term):
return domain.getPartByKey(term)
return Data(
head="",
body='getDocValue('+convert.string2quote(domain.dimension.fields[0])+')'
), fromTerm
else:
def fromTerm(term):
terms = [convert.pipe2value(t) for t in convert.pipe2value(term).split("|")]
candidate = dict(zip(jx_fields, terms))
for p in domain.partitions:
for k, t in candidate.items():
if p.value[k] != t:
break
else:
return p
if domain.type in ["uid", "default"]:
part = {"value": candidate}
domain.partitions.append(part)
return part
else:
return Null
for f in es_fields:
term.append('Value2Pipe(getDocValue('+convert.string2quote(f)+'))')
return Data(
head="",
body='Value2Pipe('+('+"|"+'.join(term))+')'
), fromTerm
else:
for v in domain.partitions:
term.append("if (" + _where(v.esfilter, lambda x: self._translate(x)) + ") " + value2MVEL(domain.getKey(v)) + "; else ")
term.append(value2MVEL(domain.getKey(domain.NULL)))
func_name = "_temp"+UID()
return self.register_function("+\"|\"+".join(term))
def es_setop(es, mvel, query):
FromES = es09.util.build_es_query(query)
select = listwrap(query.select)
isDeep = len(split_field(query.frum.name)) > 1 # LOOKING INTO NESTED WILL REQUIRE A SCRIPT
isComplex = OR([s.value == None and s.aggregate not in ("count", "none") for s in select]) # CONVERTING esfilter DEFINED PARTS WILL REQUIRE SCRIPT
if not isDeep and not isComplex:
if len(select) == 1 and not select[0].value or select[0].value == "*":
FromES = wrap({
"query": {"filtered": {
"query": {"match_all": {}},
"filter": simplify_esfilter(jx_expression(query.where).to_esfilter())
}},
"sort": query.sort,
"size": 1
})
elif all(isinstance(v, Variable) for v in select.value):
FromES = wrap({
"query": {"filtered": {
"query": {"match_all": {}},
"filter": simplify_esfilter(query.where.to_esfilter())
}},
"fields": select.value,
"sort": query.sort,
"size": coalesce(query.limit, 200000)
})
elif not isDeep:
simple_query = query.copy()
simple_query.where = TRUE_FILTER # THE FACET FILTER IS FASTER
FromES.facets.mvel = {
"terms": {
"script_field": mvel.code(simple_query),
"size": coalesce(simple_query.limit, 200000)
},
"facet_filter": simplify_esfilter(jx_expression(query.where).to_esfilter())
}
else:
FromES.facets.mvel = {
"terms": {
"script_field": mvel.code(query),
"size": coalesce(query.limit, 200000)
},
"facet_filter": simplify_esfilter(jx_expression(query.where).to_esfilter())
}
data = es09.util.post(es, FromES, query.limit)
if len(select) == 1 and not select[0].value or select[0].value == "*":
# SPECIAL CASE FOR SINGLE COUNT
cube = wrap(data).hits.hits._source
elif isinstance(select[0].value, Variable):
# SPECIAL CASE FOR SINGLE TERM
cube = wrap(data).hits.hits.fields
else:
data_list = unpack_terms(data.facets.mvel, select)
if not data_list:
cube = Cube(select, [], {s.name: Matrix.wrap([]) for s in select})
else:
output = zip(*data_list)
cube = Cube(select, [], {s.name: Matrix(list=output[i]) for i, s in enumerate(select)})
return Data(
meta={"esquery": FromES},
data=cube
)
