def filter(table, predicates):
"""
Select rows from table based on boolean expressions
Parameters
----------
predicates : boolean array expressions, or list thereof
Returns
-------
filtered_expr : TableExpr
"""
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
predicates = util.promote_list(predicates)
predicates = [ir.bind_expr(table, x) for x in predicates]
resolved_predicates = []
for pred in predicates:
if isinstance(pred, ir.AnalyticExpr):
pred = pred.to_filter()
resolved_predicates.append(pred)
op = _L.apply_filter(table, resolved_predicates)
return TableExpr(op)
def _clean_predicates(self, predicates):
import ibis.expr.analysis as L
result = []
if not isinstance(predicates, (list, tuple)):
predicates = [predicates]
for pred in predicates:
if isinstance(pred, tuple):
if len(pred) != 2:
raise com.ExpressionError('Join key tuple must be '
'length 2')
lk, rk = pred
lk = self.left._ensure_expr(lk)
rk = self.right._ensure_expr(rk)
pred = lk == rk
else:
pred = L.substitute_parents(pred, past_projection=False)
if not isinstance(pred, ir.BooleanArray):
raise com.ExpressionError('Join predicate must be comparison')
preds = L.unwrap_ands(pred)
result.extend(preds)
return result
def join(left, right, predicates=(), how='inner'):
"""
Perform a relational join between two tables. Does not resolve resulting
table schema.
Parameters
----------
left : TableExpr
right : TableExpr
predicates : join expression(s)
how : string, default 'inner'
- 'inner': inner join
- 'left': left join
- 'outer': full outer join
- 'semi' or 'left_semi': left semi join
- 'anti': anti join
Returns
-------
joined : TableExpr
Note, schema is not materialized yet
"""
klass = _join_classes[how.lower()]
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
op = klass(left, right, predicates)
return TableExpr(op)
def join(left, right, predicates=(), how="inner"):
"""
Perform a relational join between two tables. Does not resolve resulting
table schema.
Parameters
----------
left : TableExpr
right : TableExpr
predicates : join expression(s)
how : string, default 'inner'
- 'inner': inner join
- 'left': left join
- 'outer': full outer join
- 'semi' or 'left_semi': left semi join
- 'anti': anti join
Returns
-------
joined : TableExpr
Note, schema is not materialized yet
"""
klass = _join_classes[how.lower()]
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
op = klass(left, right, predicates)
return TableExpr(op)
def filter(table, predicates):
"""
Select rows from table based on boolean expressions
Parameters
----------
predicates : boolean array expressions, or list thereof
Returns
-------
filtered_expr : TableExpr
"""
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
predicates = util.promote_list(predicates)
predicates = [ir.bind_expr(table, x) for x in predicates]
resolved_predicates = []
for pred in predicates:
if isinstance(pred, ir.AnalyticExpr):
pred = pred.to_filter()
resolved_predicates.append(pred)
op = _L.apply_filter(table, resolved_predicates)
return TableExpr(op)
def _clean_predicates(self, predicates):
import ibis.expr.analysis as L
result = []
if not isinstance(predicates, (list, tuple)):
predicates = [predicates]
for pred in predicates:
if isinstance(pred, tuple):
if len(pred) != 2:
raise com.ExpressionError('Join key tuple must be '
'length 2')
lk, rk = pred
lk = self.left._ensure_expr(lk)
rk = self.right._ensure_expr(rk)
pred = lk == rk
else:
pred = L.substitute_parents(pred, past_projection=False)
if not isinstance(pred, ir.BooleanArray):
raise com.ExpressionError('Join predicate must be comparison')
preds = L.unwrap_ands(pred)
result.extend(preds)
return result
def _resolve_predicates(table, predicates):
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
predicates = util.promote_list(predicates)
predicates = [ir.bind_expr(table, x) for x in predicates]
resolved_predicates = []
for pred in predicates:
if isinstance(pred, ir.AnalyticExpr):
pred = pred.to_filter()
resolved_predicates.append(pred)
return resolved_predicates
def _visit(self, expr):
node = expr.op()
for arg in node.flat_args():
if isinstance(arg, ir.TableExpr):
self._visit_table(arg)
elif isinstance(arg, ir.BooleanColumn):
for sub_expr in L.unwrap_ands(arg):
self.predicates.append(sub_expr)
self._visit(sub_expr)
elif isinstance(arg, ir.Expr):
self._visit(arg)
else:
continue
def _visit(self, expr):
node = expr.op()
for arg in node.flat_args():
if isinstance(arg, ir.TableExpr):
self._visit_table(arg)
elif isinstance(arg, ir.BooleanArray):
for sub_expr in L.unwrap_ands(arg):
self.predicates.append(sub_expr)
self._visit(sub_expr)
elif isinstance(arg, ir.Expr):
self._visit(arg)
else:
continue
def filter(table, predicates):
"""
Select rows from table based on boolean expressions
Parameters
----------
predicates : boolean array expressions, or list thereof
Returns
-------
filtered_expr : TableExpr
"""
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
op = _L.apply_filter(table, predicates)
return TableExpr(op)
def filter(table, predicates):
"""
Select rows from table based on boolean expressions
Parameters
----------
predicates : boolean array expressions, or list thereof
Returns
-------
filtered_expr : TableExpr
"""
if isinstance(predicates, Expr):
predicates = _L.unwrap_ands(predicates)
op = _L.apply_filter(table, predicates)
return TableExpr(op)
