Merge branch 'master' of github.com:static-frame/arraymap

Author: flexatone

arraymap.c

@@ -145,11 +145,15 @@ char_get_end_p(char* p, Py_ssize_t dt_size) {
 }
 
 
+// This masks the input with INT64_MAX, which removes the MSB; we then cast to an int64; the range is now between 0 and INT64_MAX. We then use the MSB of the original value; if set, we negate the number, producing negative values for the upper half of the uint64 range. Note that we only need to check for hash -1 in this branch.
 static inline Py_hash_t
 uint_to_hash(npy_uint64 v) {
-    Py_hash_t hash = (Py_hash_t)(v >> 1); // half unsigned fits in signed
-    if (hash == -1) {
-        return -2;
+    Py_hash_t hash = (Py_hash_t)(v & INT64_MAX);
+    if (v >> 63) {
+        hash = -hash;
+        if (hash == -1) {
+            return -2;
+        }
     }
     return hash;
 }
@@ -1239,7 +1243,12 @@ insert_int(
         return -1;
     }
     if (self->table[table_pos].hash != -1) {
-        PyErr_SetObject(NonUniqueError, PyLong_FromSsize_t(key));
+        PyObject* er = PyLong_FromLongLong(key); // for error reporting
+        if (er == NULL) {
+            return -1;
+        }
+        PyErr_SetObject(NonUniqueError, er);
+        Py_DECREF(er);
         return -1;
     }
     self->table[table_pos].keys_pos = keys_pos;
@@ -1264,7 +1273,12 @@ insert_uint(
         return -1;
     }
     if (self->table[table_pos].hash != -1) {
-        PyErr_SetObject(NonUniqueError, PyLong_FromSsize_t(key));
+        PyObject* er = PyLong_FromUnsignedLongLong(key);
+        if (er == NULL) {
+            return -1;
+        }
+        PyErr_SetObject(NonUniqueError, er);
+        Py_DECREF(er);
         return -1;
     }
     self->table[table_pos].keys_pos = keys_pos;
@@ -1290,7 +1304,12 @@ insert_double(
         return -1;
     }
     if (self->table[table_pos].hash != -1) {
-        PyErr_SetObject(NonUniqueError, PyFloat_FromDouble(key));
+        PyObject* er = PyFloat_FromDouble(key);
+        if (er == NULL) {
+            return -1;
+        }
+        PyErr_SetObject(NonUniqueError, er);
+        Py_DECREF(er);
         return -1;
     }
     self->table[table_pos].keys_pos = keys_pos;
@@ -1316,8 +1335,12 @@ insert_unicode(
         return -1;
     }
     if (self->table[table_pos].hash != -1) {
-        PyErr_SetObject(NonUniqueError,
-            PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, key, key_size));
+        PyObject* er = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, key, key_size);
+        if (er == NULL) {
+            return -1;
+        }
+        PyErr_SetObject(NonUniqueError, er);
+        Py_DECREF(er);
         return -1;
     }
     self->table[table_pos].keys_pos = keys_pos;
@@ -1343,8 +1366,12 @@ insert_string(
         return -1;
     }
     if (self->table[table_pos].hash != -1) {
-        PyErr_SetObject(NonUniqueError,
-            PyBytes_FromStringAndSize(key, key_size));
+        PyObject* er = PyBytes_FromStringAndSize(key, key_size);
+        if (er == NULL) {
+            return -1;
+        }
+        PyErr_SetObject(NonUniqueError, er);
+        Py_DECREF(er);
         return -1;
     }
     self->table[table_pos].keys_pos = keys_pos;
@@ -1947,6 +1974,7 @@ fam_init(PyObject *self, PyObject *args, PyObject *kwargs)
     }
     return 0;
 error:
+    // assume all dynamic memory assigned to struct attrs that will be cleaned
     return -1;
 }
 

doc/articles/npy-opt.py

@@ -309,9 +309,20 @@ def get_array(size: int) -> np.ndarray:
         return array
 
 
+class FFUInt32(FixtureFactory):
+    NAME = "uint32"
+    SORT = 3
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        array = np.arange(INT_START, INT_START + size, dtype=np.uint32)
+        array.flags.writeable = False
+        return array
+
+
 class FFFloat64(FixtureFactory):
     NAME = "float64"
-    SORT = 3
+    SORT = 4
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
@@ -322,7 +333,7 @@ def get_array(size: int) -> np.ndarray:
 
 class FFFloat32(FixtureFactory):
     NAME = "float32"
-    SORT = 4
+    SORT = 5
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
@@ -331,37 +342,112 @@ def get_array(size: int) -> np.ndarray:
         return array
 
 
-class FFString(FixtureFactory):
-    NAME = "string"
+def get_string_array(size: int, char_count: int, kind: str) -> str:
+    fmt = f'-<{char_count}'
+    array = np.array(
+            [f'{hex(e) * (char_count // 8)}'.format(fmt) for e in range(INT_START, INT_START + size)],
+            dtype=f'{kind}{char_count}')
+    array.flags.writeable = False
+    return array
+
+class FFU8(FixtureFactory):
+    NAME = "U8"
     SORT = 6
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        array = np.array([hex(e) for e in range(size)])
-        array.flags.writeable = False
-        return array
-
+        return get_string_array(size, 8, 'U')
 
-class FFString4x(FixtureFactory):
-    NAME = "string 4x"
+class FFU16(FixtureFactory):
+    NAME = "U16"
     SORT = 7
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        array = np.array([hex(e) * 4 for e in range(size)])
-        array.flags.writeable = False
-        return array
+        return get_string_array(size, 16, 'U')
 
 
-class FFBytes(FixtureFactory):
-    NAME = "bytes"
+class FFU32(FixtureFactory):
+    NAME = "U32"
     SORT = 8
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        array = np.array([bytes(hex(e), encoding="utf-8") for e in range(size)])
-        array.flags.writeable = False
-        return array
+        return get_string_array(size, 32, 'U')
+
+
+class FFU64(FixtureFactory):
+    NAME = "U64"
+    SORT = 9
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 64, 'U')
+
+
+class FFU128(FixtureFactory):
+    NAME = "U128"
+    SORT = 10
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 128, 'U')
+
+
+class FFS8(FixtureFactory):
+    NAME = "S8"
+    SORT = 11
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 8, 'S')
+
+class FFS16(FixtureFactory):
+    NAME = "S16"
+    SORT = 12
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 16, 'S')
+
+
+class FFS32(FixtureFactory):
+    NAME = "S32"
+    SORT = 13
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 32, 'S')
+
+
+class FFS64(FixtureFactory):
+    NAME = "S64"
+    SORT = 14
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 64, 'S')
+
+class FFS128(FixtureFactory):
+    NAME = "S128"
+    SORT = 15
+
+    @staticmethod
+    def get_array(size: int) -> np.ndarray:
+        return get_string_array(size, 128, 'S')
+
+
+
+
+# class FFBytes(FixtureFactory):
+#     NAME = "bytes"
+#     SORT = 8
+
+#     @staticmethod
+#     def get_array(size: int) -> np.ndarray:
+#         array = np.array([bytes(hex(e), encoding="utf-8") for e in range(INT_START, INT_START + size)])
+#         array.flags.writeable = False
+#         return array
 
 
 class FFObject(FixtureFactory):
@@ -390,19 +476,28 @@ def get_versions() -> str:
 
 
 CLS_FF = (
-    FFInt32,
-    FFInt64,
-    FFUInt64,
-    FFFloat64,
-    FFString,
-    FFString4x,
-    FFBytes,
+    # FFInt32,
+    # FFInt64,
+    # FFUInt32,
+    # FFUInt64,
+    # FFFloat64,
+    FFU8,
+    FFU16,
+    FFU32,
+    FFU64,
+    FFU128,
+
+    FFS8,
+    FFS16,
+    FFS32,
+    FFS64,
+    FFS128,
     # FFObject,
 )
 FF_ORDER = [f.NAME for f in sorted(CLS_FF, key=lambda ff: ff.SORT)]
 
 # -------------------------------------------------------------------------------
-NUMBER = 2
+NUMBER = 20
 
 from itertools import product
 
@@ -452,24 +547,33 @@ def plot_performance(frame, suffix: str = ""):
             ax.set_title(title, fontsize=6)
             ax.set_box_aspect(0.8)
             time_max = fixture["time"].max()
-            ax.set_yticks([0, time_max * 0.5, time_max])
-            ax.set_yticklabels(
-                [
+            time_min = fixture["time"].min()
+            y_ticks = [0, time_min, time_max * 0.5, time_max]
+            y_labels = [
                     "",
+                    seconds_to_display(time_min),
                     seconds_to_display(time_max * 0.5),
                     seconds_to_display(time_max),
-                ],
-                fontsize=6,
-            )
+                ]
+            if time_min > time_max * 0.25:
+                # remove the min if it is greater than quarter
+                y_ticks.pop(1)
+                y_labels.pop(1)
+
+            ax.set_yticks(y_ticks)
+            ax.set_yticklabels(y_labels, fontsize=4)
             # ax.set_xticks(x, names_display, rotation='vertical')
             ax.tick_params(
                 axis="x",
-                which="both",
                 bottom=False,
-                top=False,
                 labelbottom=False,
             )
-
+            ax.tick_params(
+                axis="y",
+                length=2,
+                width=.5,
+                pad=1,
+            )
     fig.set_size_inches(9, 3)  # width, height
     fig.legend(post, names_display, loc="center right", fontsize=6)
     # horizontal, vertical
@@ -483,7 +587,7 @@ def plot_performance(frame, suffix: str = ""):
         right=0.85,
         top=0.80,
         wspace=1.0,  # width
-        hspace=0.2,
+        hspace=0.4,
     )
     # plt.rcParams.update({'font.size': 22})
     plt.savefig(fp, dpi=300)

test/test_unit.py

@@ -92,6 +92,23 @@ def test_fam_constructor_array_int_d():
         assert k in fam
 
 
+def test_fam_constructor_array_int_e():
+    # https://github.com/static-frame/arraymap/issues/12
+    a1 = np.array((0, 0, 1, 1, 2, 2), dtype=int)
+    a2 = a1[[0, 2, 4]]
+    a2.flags.writeable = False
+    fam1 = FrozenAutoMap(a2)
+    assert list(fam1) == [0, 1, 2]
+
+    d1 = {i: int(i) for i in a2}
+    fam2 = FrozenAutoMap(d1)
+    assert list(fam2) == [0, 1, 2]
+
+    d2 = {0: 0, 3: 1}
+    fam3 = FrozenAutoMap(d2)
+    assert list(fam3) == [0, 3]
+
+
 # ------------------------------------------------------------------------------