Changeset: 0ea25c43cfce for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=0ea25c43cfce
Modified Files:
monetdb5/extras/pyapi/pyapi.c
sql/backends/monet5/Tests/pyapi12.sql
sql/backends/monet5/Tests/pyapi12.stable.out
Branch: pyapi
Log Message:
Fix for returning NPY_OBJECT arrays when using PYTHON_MAP.
diffs (116 lines):
diff --git a/monetdb5/extras/pyapi/pyapi.c b/monetdb5/extras/pyapi/pyapi.c
--- a/monetdb5/extras/pyapi/pyapi.c
+++ b/monetdb5/extras/pyapi/pyapi.c
@@ -893,8 +893,6 @@ str PyAPIeval(MalBlkPtr mb, MalStkPtr st
PyObject * pColO = NULL;
// This is the PyReturn header information for the current return
value, we will fill this now
PyReturn *ret = &pyreturn_values[i];
- // This is the expected BAT result type (the type of BAT we have to
make)
- int bat_type = ATOMstorage(getColumnType(getArgType(mb,pci,i)));
ret->multidimensional = FALSE;
// There are three possibilities (we have ensured this right after
executing the Python call)
@@ -936,16 +934,10 @@ str PyAPIeval(MalBlkPtr mb, MalStkPtr st
}
else {
// If it isn't we need to convert pColO to the expected Numpy
Array type
- ret->numpy_array = NULL;
- (void) bat_type;
- //if (bat_type != TYPE_str) ret->numpy_array = (PyArrayObject*)
PyArray_FromAny(pColO, PyArray_DescrFromType(BatType_ToPyType(bat_type)), 1, 1,
NPY_ARRAY_CARRAY | NPY_ARRAY_FORCECAST, NULL);
+ ret->numpy_array = PyArray_FromAny(pColO, NULL, 1, 1,
NPY_ARRAY_CARRAY | NPY_ARRAY_FORCECAST, NULL);
if (ret->numpy_array == NULL) {
- // If this conversion fails, we will set the expected type to
NULL, this means it will automatically pick a type for us
- ret->numpy_array = PyArray_FromAny(pColO, NULL, 1, 1,
NPY_ARRAY_CARRAY | NPY_ARRAY_FORCECAST, NULL);
- if (ret->numpy_array == NULL) {
- msg = createException(MAL, "pyapi.eval", "Could not create
a Numpy array from the return type.\n");
- goto wrapup;
- }
+ msg = createException(MAL, "pyapi.eval", "Could not create a
Numpy array from the return type.\n");
+ goto wrapup;
}
ret->result_type =
PyArray_DESCR((PyArrayObject*)ret->numpy_array)->type_num; // We read the
result type from the resulting array
ret->memory_size =
PyArray_DESCR((PyArrayObject*)ret->numpy_array)->elsize;
@@ -992,8 +984,28 @@ str PyAPIeval(MalBlkPtr mb, MalStkPtr st
for (i = 0; i < pci->retc; i++)
{
PyReturn *ret = &pyreturn_values[i];
+ ReturnBatDescr *descr = &ptr[(process_id - 1) * pci->retc + i];
- ReturnBatDescr *descr = &ptr[(process_id - 1) * pci->retc + i];
+ if (ret->result_type == NPY_OBJECT) {
+ // We can't deal with NPY_OBJECT arrays, because these are
'arrays of pointers', so we can't just copy the content of the array into
shared memory
+ // So if we're dealing with a NPY_OBJECT array, we convert
them to a Numpy Array of type NPY_<TYPE> that corresponds with the desired BAT
type
+ // WARNING: Because we could be converting to a NPY_STRING or
NPY_UNICODE array (if the desired type is TYPE_str or TYPE_hge), this means
that memory usage can explode
+ // because NPY_STRING/NPY_UNICODE arrays are 2D string
arrays with fixed string length (so if there's one very large string the size
explodes quickly)
+ // if someone has some problem with memory size exploding
when using PYTHON_MAP but it being fine in regular PYTHON this is probably the
issue
+ int bat_type =
ATOMstorage(getColumnType(getArgType(mb,pci,i)));
+ PyObject *new_array = PyArray_FromAny(ret->numpy_array,
PyArray_DescrFromType(BatType_ToPyType(bat_type)), 1, 1, NPY_ARRAY_CARRAY |
NPY_ARRAY_FORCECAST, NULL);
+ if (new_array == NULL) {
+ msg = createException(MAL, "pyapi.eval", "Could not
convert the returned NPY_OBJECT array to the desired array of type %s.\n",
BatType_Format(bat_type));
+ goto wrapup;
+ }
+ Py_DECREF(ret->numpy_array); //do we really care about
cleaning this up, considering this only happens in a separate process that will
be exited soon anyway?
+ ret->numpy_array = new_array;
+ ret->result_type =
PyArray_DESCR((PyArrayObject*)ret->numpy_array)->type_num;
+ ret->memory_size =
PyArray_DESCR((PyArrayObject*)ret->numpy_array)->elsize;
+ ret->count = PyArray_DIMS((PyArrayObject*)ret->numpy_array)[0];
+ ret->array_data =
PyArray_DATA((PyArrayObject*)ret->numpy_array);
+ }
+
descr->npy_type = ret->result_type;
descr->element_size = ret->memory_size;
descr->bat_count = ret->count;
diff --git a/sql/backends/monet5/Tests/pyapi12.sql
b/sql/backends/monet5/Tests/pyapi12.sql
--- a/sql/backends/monet5/Tests/pyapi12.sql
+++ b/sql/backends/monet5/Tests/pyapi12.sql
@@ -3,12 +3,24 @@ START TRANSACTION;
CREATE TABLE rval(i integer);
INSERT INTO rval VALUES (1),(2),(3),(4),(-1),(0);
+# PYTHON_MAP test in WHERE
CREATE FUNCTION pyapi12(i integer,z integer) returns boolean language
PYTHON_MAP
{
return(numpy.greater(i,z))
};
SELECT * FROM rval WHERE pyapi12(i,2);
DROP FUNCTION pyapi12;
+
+
+# Return NPY_OBJECT test
+CREATE FUNCTION pyapi12(i integer,z integer) returns string language PYTHON_MAP
+{
+ return(numpy.array(['Hello'] * len(i), dtype=object))
+};
+SELECT pyapi12(i,2) FROM rval;
+DROP FUNCTION pyapi12;
+
+
DROP TABLE rval;
diff --git a/sql/backends/monet5/Tests/pyapi12.stable.out
b/sql/backends/monet5/Tests/pyapi12.stable.out
--- a/sql/backends/monet5/Tests/pyapi12.stable.out
+++ b/sql/backends/monet5/Tests/pyapi12.stable.out
@@ -74,6 +74,22 @@ Ready.
[ 3 ]
[ 4 ]
#DROP FUNCTION pyapi12;
+#CREATE FUNCTION pyapi12(i integer,z integer) returns string language P
+#{
+# return(numpy.array(['Hello', 'Hello'], dtype=object))
+#};
+#SELECT pyapi12(i,2) FROM rval;
+% sys.L # table_name
+% pyapi12_i # name
+% clob # type
+% 5 # length
+[ "Hello" ]
+[ "Hello" ]
+[ "Hello" ]
+[ "Hello" ]
+[ "Hello" ]
+[ "Hello" ]
+#DROP FUNCTION pyapi12;
#DROP TABLE rval;
#ROLLBACK;
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
