Merge pull request #18 from static-frame/15/get-end

Refinements to `char_get_end_p()`

Author: flexatone

arraymap.c

@@ -122,26 +122,26 @@ typedef enum ViewKind{
 } ViewKind;
 
 
-// NOTE: would like to use strchr(str, '\0') instead of this routine, but some buffers might not have a null terminator and stread by full to the the dt_size.
+// Return the end pointer, or the pointer to the location after the last valid character. The end pointer minus the start pointer is the number of characters. For an empty string, all characters are NULL, and the start pointer and end pointer should be equal. NOTE: would like to use strchr(str, '\0') instead of this routine, but some buffers might not have a null terminator and stread by full to the the dt_size.
 static inline Py_UCS4*
 ucs4_get_end_p(Py_UCS4* p_start, Py_ssize_t dt_size) {
-    Py_UCS4* p;
-    for (p = p_start + dt_size - 1; p >= p_start; p--) {
+    for (Py_UCS4* p = p_start + dt_size - 1; p >= p_start; p--) {
         if (*p != '\0') {
-            return p + 1; // return 1 more than the first non-null from the right
+            return p + 1; // 1 after first non-null
         }
     }
-    return p; // p is equal to p_start
+    return p_start;
 }
 
 
 static inline char*
-char_get_end_p(char* p, Py_ssize_t dt_size) {
-    char* p_end = p + dt_size;
-    while (p < p_end && *p != '\0') {
-        p++;
+char_get_end_p(char* p_start, Py_ssize_t dt_size) {
+    for (char* p = p_start + dt_size - 1; p >= p_start; p--) {
+        if (*p != '\0') {
+            return p + 1; // 1 after first non-null
+        }
     }
-    return p;
+    return p_start;
 }
 
 

doc/articles/npy-opt.py

@@ -343,28 +343,34 @@ def get_array(size: int) -> np.ndarray:
 
 
 def get_string_array(size: int, char_count: int, kind: str) -> str:
-    fmt = f'-<{char_count}'
+    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}')
+        [
+            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:
-        return get_string_array(size, 8, 'U')
+        return get_string_array(size, 8, "U")
+
 
 class FFU16(FixtureFactory):
     NAME = "U16"
     SORT = 7
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 16, 'U')
+        return get_string_array(size, 16, "U")
 
 
 class FFU32(FixtureFactory):
@@ -373,7 +379,7 @@ class FFU32(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 32, 'U')
+        return get_string_array(size, 32, "U")
 
 
 class FFU64(FixtureFactory):
@@ -382,7 +388,7 @@ class FFU64(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 64, 'U')
+        return get_string_array(size, 64, "U")
 
 
 class FFU128(FixtureFactory):
@@ -391,7 +397,7 @@ class FFU128(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 128, 'U')
+        return get_string_array(size, 128, "U")
 
 
 class FFS8(FixtureFactory):
@@ -400,15 +406,16 @@ class FFS8(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 8, 'S')
+        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')
+        return get_string_array(size, 16, "S")
 
 
 class FFS32(FixtureFactory):
@@ -417,7 +424,7 @@ class FFS32(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 32, 'S')
+        return get_string_array(size, 32, "S")
 
 
 class FFS64(FixtureFactory):
@@ -426,17 +433,16 @@ class FFS64(FixtureFactory):
 
     @staticmethod
     def get_array(size: int) -> np.ndarray:
-        return get_string_array(size, 64, 'S')
+        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')
-
-
+        return get_string_array(size, 128, "S")
 
 
 # class FFBytes(FixtureFactory):
@@ -486,7 +492,6 @@ def get_versions() -> str:
     FFU32,
     FFU64,
     FFU128,
-
     FFS8,
     FFS16,
     FFS32,
@@ -550,11 +555,11 @@ def plot_performance(frame, suffix: str = ""):
             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),
-                ]
+                "",
+                seconds_to_display(time_min),
+                seconds_to_display(time_max * 0.5),
+                seconds_to_display(time_max),
+            ]
             if time_min > time_max * 0.25:
                 # remove the min if it is greater than quarter
                 y_ticks.pop(1)
@@ -571,7 +576,7 @@ def plot_performance(frame, suffix: str = ""):
             ax.tick_params(
                 axis="y",
                 length=2,
-                width=.5,
+                width=0.5,
                 pad=1,
             )
     fig.set_size_inches(9, 3)  # width, height

test/test_unit.py

@@ -234,6 +234,20 @@ def test_fam_copy_array_bytes_a():
 # ------------------------------------------------------------------------------
 
 
+def test_fam_array_bytes_get_a():
+    a1 = np.array((b"", b"  ", b"   ", b"    "))
+    a1.flags.writeable = False
+    fam = FrozenAutoMap(a1)
+
+    assert fam.get(b"") == 0
+    assert fam.get(b" ") == None
+    assert fam.get(b"   ") == 2
+    assert fam.get(b"    ") == 3
+
+
+# ------------------------------------------------------------------------------
+
+
 def test_fam_array_len_a():
     a1 = np.array((10, 20, 30, 40), dtype=np.int64)
     a1.flags.writeable = False
@@ -577,6 +591,17 @@ def test_fam_array_unicode_get_a():
     assert fam.get("cccc") is None
 
 
+def test_fam_array_unicode_get_b():
+    a1 = np.array(("", "  ", "   ", "    "))
+    a1.flags.writeable = False
+    fam = FrozenAutoMap(a1)
+
+    assert fam.get("") == 0
+    assert fam.get(" ") == None
+    assert fam.get("   ") == 2
+    assert fam.get("    ") == 3
+
+
 # ------------------------------------------------------------------------------