WIP: a bit working

examples/ex_4.c

@@ -62,6 +62,92 @@ shuffle(rnd_ctx_t *prng, int *array, size_t n)
   }
 }
 
+int
+compare_ints(const void *a, const void *b)
+{
+  return *(const int *)a - *(const int *)b;
+}
+
+void
+print_array(int *array, size_t l)
+{
+  int a[l];
+  memcpy(a, array, sizeof(int) * l);
+  qsort(a, l, sizeof(int), compare_ints);
+
+  printf("int a[] = {");
+  for (int i = 0; i < l; i++) {
+    printf("%d", a[i]);
+    if (i != l) {
+      printf(", ");
+    }
+  }
+  printf("};\n");
+}
+
+bool
+has_span(int *array, size_t n, int x, int l)
+{
+  if (n == 0 || l <= 0) {
+    return false;
+  }
+
+  int a[n];
+  memcpy(a, array, sizeof(int) * n);
+  qsort(a, n, sizeof(int), compare_ints);
+
+  // Iterate through the array to find a span starting at x of length l
+  for (size_t i = 0; i < n; i++) {
+    if (a[i] == x) {
+      // Check if the span can fit in the array
+      if (i + l - 1 < n && a[i + l - 1] == x + l - 1) {
+        return true; // Found the span
+      }
+    }
+  }
+
+  return false; // Span not found
+}
+
+void
+print_spans(int *array, size_t n)
+{
+  int a[n];
+  size_t start = 0;
+  size_t end = 0;
+
+  if (n == 0) {
+    printf("Array is empty\n");
+    return;
+  }
+
+  memcpy(a, array, sizeof(int) * n);
+  qsort(a, n, sizeof(int), compare_ints);
+
+  for (size_t i = 1; i < n; i++) {
+    if (a[i] == a[i - 1] + 1) {
+      end = i; // Extend the span
+    } else {
+      // Print the current span
+      if (start == end) {
+        printf("[%d] ", a[start]);
+      } else {
+        printf("[%d, %d] ", a[start], a[end]);
+      }
+      // Move to the next span
+      start = i;
+      end = i;
+    }
+  }
+
+  // Print the last span if needed
+  if (start == end) {
+    printf("[%d]\n", a[start]);
+  } else {
+    printf("[%d, %d]\n", a[start], a[end]);
+  }
+}
+
 bool
 was_set(size_t bit, const int array[])
 {
@@ -84,13 +170,18 @@ main(void)
   int array[TEST_ARRAY_SIZE];
 
   // disable buffering
+  setbuf(stdout, 0);
   setbuf(stderr, 0);
 
+  // int foo[] = {1, 2, 3, 5, 7, 8 ,9, 10, 21};
+  // print_spans(foo, sizeof(foo) / sizeof(foo[0]));
+  // exit(0);
+
   // start with a 1KiB buffer, TEST_ARRAY_SIZE bits
-  uint8_t *buf = calloc(TEST_ARRAY_SIZE, sizeof(uint8_t));
+  uint8_t *buf = calloc(TEST_ARRAY_SIZE * 1, sizeof(uint8_t));
 
   // create the sparse bitmap
-  sparsemap_t *map = sparsemap(buf, sizeof(uint8_t) * TEST_ARRAY_SIZE, 0);
+  sparsemap_t *map = sparsemap(buf, sizeof(uint8_t) * 1024 * 1, 0);
 
 #if 0
   for (i = 0; i < 8; i++) sparsemap_set(map, i, true);
@@ -117,9 +208,9 @@ main(void)
   __diag("rank was %lu between [0, 10]\n", rank);
   assert(rank == 5);
   sparsemap_clear(map);
-#endif
 
-  for (i = 2049; i < 2057; i++) sparsemap_set(map, i, true);
+  for (i = 2049; i < 2057; i++)
+    sparsemap_set(map, i, true);
   rank = sparsemap_rank(map, 2049, 2058);
   __diag("rank was %lu at offset 108\n", rank);
   assert(rank == 8);
@@ -137,6 +228,7 @@ main(void)
   // rank = sparsemap_span(map, 2048, 8);
   //__diag("span was found at %lu\n", rank);
   sparsemap_clear(map);
+#endif
 
   // seed the PRNG
 #ifdef SEED
@@ -146,13 +238,26 @@ main(void)
 #endif
 
   for (i = 0; i < TEST_ARRAY_SIZE; i++) {
-    array[i] = (int)__random(&prng) % TEST_ARRAY_SIZE + 1;
+    uint32_t r = __random(&prng);
+    array[i] = (int)r % (4 * TEST_ARRAY_SIZE);
     if (array[i] < 0) {
       i--;
     }
+    for (int j = 0; j < i; j++) {
+      if (array[j] == array[i]) {
+        i--;
+      }
+    }
+  }
+  // create a span of at least 8 for testing between 141 and 153
+  int j = 143;
+  for (i = 0; i < 8; i++) {
+    uint32_t r = __random(&prng) % TEST_ARRAY_SIZE;
+    array[r] = j++;
   }
   // randomize setting the bits on
   shuffle(&prng, array, TEST_ARRAY_SIZE);
+  print_spans(array, TEST_ARRAY_SIZE);
 
   // set all the bits on in a random order
   for (i = 0; i < TEST_ARRAY_SIZE; i++) {
@@ -161,18 +266,23 @@ main(void)
     assert(sparsemap_is_set(map, array[i]) == true);
   }
 
-  size_t len  = 6;
+//  for (size_t len = 1; len < TEST_ARRAY_SIZE; len++) {
+  for (size_t len = 2; len <= 2; len++) {
+    __diag("================> %lu\n", len);
     size_t l = sparsemap_span(map, 0, len);
-  __diag("found span of %d at %lu starting from 0\n", len, l);
+    __diag("found span of %lu at %lu\n", len, l);
+    __diag("is_span(%lu, %lu) == %s\n", l, len,
+      has_span(array, TEST_ARRAY_SIZE, l, len) ? "yes" : "no");
     for (i = (int)l; i < l + len; i++) {
       bool set = sparsemap_is_set(map, i);
       if (set) {
-      __diag("verified %lu was set\n", i);
+        __diag("verified %d was set\n", i);
       } else {
-      __diag("darn, %lu was not really set, %s\n", i,
+        __diag("darn, %d was not really set, %s\n", i,
           was_set(i, array) ? "but we thought it was" : "because it wasn't");
       }
     }
+  }
 
   return 0;
 }

src/sparsemap.c

@@ -384,7 +384,7 @@ __sm_chunk_map_set(__sm_chunk_t *map, size_t idx, bool value, size_t *pos,
  * n'th bit was found in this __sm_chunk_t, or to the new, reduced value of |n|
  */
 static size_t
-__sm_chunk_map_select(__sm_chunk_t *map, ssize_t n, ssize_t *pnew_n)
+__sm_chunk_map_select(__sm_chunk_t *map, size_t n, ssize_t *pnew_n)
 {
   size_t ret = 0;
   register uint8_t *p;
@@ -432,7 +432,7 @@ __sm_chunk_map_select(__sm_chunk_t *map, ssize_t n, ssize_t *pnew_n)
     }
   }
 
-  *pnew_n = n;
+  *pnew_n = (ssize_t)n;
   return (ret);
 }
 
@@ -440,7 +440,7 @@ __sm_chunk_map_select(__sm_chunk_t *map, ssize_t n, ssize_t *pnew_n)
  * Counts the set bits in the range [start, idx].
  */
 static size_t
-__sm_chunk_map_rank(__sm_chunk_t *map, size_t start, size_t idx)
+__sm_chunk_map_rank(__sm_chunk_t *map, size_t *loc, size_t idx)
 {
   size_t ret = 0;
 
@@ -453,46 +453,56 @@ __sm_chunk_map_rank(__sm_chunk_t *map, size_t start, size_t idx)
       }
       if (flags == SM_PAYLOAD_ZEROS) {
         if (idx > SM_BITS_PER_VECTOR) {
-          if (start > SM_BITS_PER_VECTOR) {
+          if (*loc > SM_BITS_PER_VECTOR) {
-            start -= SM_BITS_PER_VECTOR;
+            *loc = *loc - SM_BITS_PER_VECTOR;
           } else {
-            idx -= SM_BITS_PER_VECTOR - start;
+            idx -= SM_BITS_PER_VECTOR - *loc;
-            start = 0;
+            *loc = 0;
           }
         } else {
           return (ret);
         }
       } else if (flags == SM_PAYLOAD_ONES) {
         if (idx > SM_BITS_PER_VECTOR) {
-          if (start > SM_BITS_PER_VECTOR) {
+          if (*loc > SM_BITS_PER_VECTOR) {
-            start -= SM_BITS_PER_VECTOR;
+            *loc = *loc - SM_BITS_PER_VECTOR;
           } else {
-            idx -= SM_BITS_PER_VECTOR - start;
+            idx -= SM_BITS_PER_VECTOR - *loc;
-            if (start == 0) {
+            if (*loc == 0) {
               ret += SM_BITS_PER_VECTOR;
             }
-            start = 0;
+            *loc = 0;
           }
         } else {
           return (ret + idx);
         }
       } else if (flags == SM_PAYLOAD_MIXED) {
         if (idx > SM_BITS_PER_VECTOR) {
-          if (start > SM_BITS_PER_VECTOR) {
+          if (*loc > SM_BITS_PER_VECTOR) {
-            start -= SM_BITS_PER_VECTOR;
+            *loc = *loc - SM_BITS_PER_VECTOR;
           } else {
-            idx -= SM_BITS_PER_VECTOR - start;
+            idx -= SM_BITS_PER_VECTOR - *loc;
-            if (start == 0) {
+            if (*loc == 0) {
               ret += popcountll((uint64_t)map->m_data[1 +
                 __sm_chunk_map_get_position(map,
                   i * SM_FLAGS_PER_INDEX_BYTE + j)]);
             }
-            start = 0;
+            *loc = 0;
           }
         } else {
           sm_bitvec_t w = map->m_data[1 +
             __sm_chunk_map_get_position(map, i * SM_FLAGS_PER_INDEX_BYTE + j)];
-          for (size_t k = 0; k < idx; k++) {
+          size_t ks = 0;
+          if (*loc > 0) {
+            if (*loc > idx) {
+              ks = idx;
+              *loc = *loc - idx;
+            } else {
+              ks += *loc;
+              *loc = 0;
+            }
+          }
+          for (size_t k = ks; k < idx; k++) {
             if (w & ((sm_bitvec_t)1 << k)) {
               ret++;
             }
@@ -1149,7 +1159,6 @@ sparsemap_select(sparsemap_t *map, size_t n)
   assert(sparsemap_get_size(map) >= SM_SIZEOF_OVERHEAD);
   size_t result = 0;
   size_t count = __sm_get_chunk_map_count(map);
-
   uint8_t *p = __sm_get_chunk_map_data(map, 0);
 
   for (size_t i = 0; i < count; i++) {
@@ -1158,7 +1167,7 @@ sparsemap_select(sparsemap_t *map, size_t n)
     __sm_chunk_t chunk;
     __sm_chunk_map_init(&chunk, p);
 
-    ssize_t new_n = n;
+    ssize_t new_n = (ssize_t)n;
     size_t index = __sm_chunk_map_select(&chunk, n, &new_n);
     if (new_n == -1) {
       return (result + index);
@@ -1200,7 +1209,7 @@ sparsemap_rank(sparsemap_t *map, size_t loc, size_t idx)
       continue;
     }
 
-    result += __sm_chunk_map_rank(&chunk, loc, idx - start);
+    result += __sm_chunk_map_rank(&chunk, &loc, idx - start);
     p += __sm_chunk_map_get_size(&chunk);
   }
   return (result);
@@ -1214,17 +1223,21 @@ sparsemap_rank(sparsemap_t *map, size_t loc, size_t idx)
 size_t
 sparsemap_span(sparsemap_t *map, size_t loc, size_t len)
 {
-  size_t nth, count, size = 1024;
+  size_t offset, nth = 0, count = 0;
 
+  offset = sparsemap_select(map, 0);
   do {
-    nth = sparsemap_select(map, nth);
+    count = sparsemap_rank(map, offset, offset + len);
-    count = sparsemap_rank(map, nth, nth + len);
     if (count == len) {
-      return nth;
+      return offset;
     } else {
-      nth += count;
+      count = len;
+      while(--count && sparsemap_is_set(map, offset)) {
+          nth++;
       }
-  } while (1); // TODO... ? until what?
+    }
+    offset = sparsemap_select(map, nth);
+  } while (1); // TODO...until what?
 
-  return size;
+  return 0;
 }