Attempt 3

Author: kiri11

.idea/.gitignore

@@ -1,21 +1,27 @@
-"""Round 1 solution — byte-pair histogram."""
-
+import os
+import mmap
 import numpy as np
 
+_BIGRAM_BYTES = tuple(i.to_bytes(2, "big") for i in range(65536))
 
 def compute_histogram(path: str) -> dict[bytes, int]:
-    data = np.fromfile(path, dtype=np.uint8)
-    if len(data) < 2:
+    if os.path.getsize(path) < 2:
         return {}
 
-    bigrams = (data[:-1].astype(np.uint16) << 8) | data[1:]
+    with open(path, "rb") as f:
+        with mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ) as mm:
+            byte_values = np.frombuffer(mm, dtype=np.uint8)
 
-    counts = np.bincount(bigrams, minlength=65536)
+            bigrams = byte_values[:-1].astype(np.uint16)
+            bigrams <<= 8
+            bigrams |= byte_values[1:]
 
+            del byte_values
+
+    counts = np.bincount(bigrams, minlength=65536)
     valid_indices = np.nonzero(counts)[0]
-    valid_counts = counts[valid_indices]
 
     return {
-        int(idx).to_bytes(2, 'big'): int(count) 
-        for idx, count in zip(valid_indices, valid_counts)
+        _BIGRAM_BYTES[idx]: counts[idx].item()
+        for idx in valid_indices
     }