Version v4.1.3

Author: Ezibenroc

pyroaring/croaring_version.pxi

@@ -1 +1 @@
-__croaring_version__ = "v4.1.6"
+__croaring_version__ = "v4.1.3"

pyroaring/roaring.c

@@ -1,5 +1,5 @@
 // !!! DO NOT EDIT - THIS IS AN AUTO-GENERATED FILE !!!
-// Created by amalgamation.sh on 2024-09-20T14:21:41Z
+// Created by amalgamation.sh on 2024-09-19T00:48:39Z
 
 /*
  * The CRoaring project is under a dual license (Apache/MIT).
@@ -22633,7 +22633,7 @@ roaring_bitmap_t *roaring_bitmap_portable_deserialize_frozen(const char *buf) {
 
 bool roaring_bitmap_to_bitset(const roaring_bitmap_t *r, bitset_t *bitset) {
     uint32_t max_value = roaring_bitmap_maximum(r);
-    size_t new_array_size = (size_t)(max_value / 64 + 1);
+    size_t new_array_size = (size_t)(((uint64_t)max_value + 63) / 64);
     bool resize_ok = bitset_resize(bitset, new_array_size, true);
     if (!resize_ok) {
         return false;
@@ -24641,13 +24641,10 @@ roaring64_bitmap_t *roaring64_bitmap_portable_deserialize_safe(
         memcpy(&high32, buf, sizeof(high32));
         buf += sizeof(high32);
         read_bytes += sizeof(high32);
-        // High 32 bits must be strictly increasing.
-        if (high32 <= previous_high32) {
+        if(high32 < previous_high32) {
             roaring64_bitmap_free(r);
             return NULL;
         }
-        previous_high32 = high32;
-
         // Read the 32-bit Roaring bitmaps representing the least significant
         // bits of a set of elements.
         size_t bitmap32_size = roaring_bitmap_portable_deserialize_size(
@@ -24666,24 +24663,6 @@ roaring64_bitmap_t *roaring64_bitmap_portable_deserialize_safe(
         buf += bitmap32_size;
         read_bytes += bitmap32_size;
 
-        // While we don't attempt to validate much, we must ensure that there
-        // is no duplication in the high 48 bits - inserting into the ART
-        // assumes (or UB) no duplicate keys. The top 32 bits must be unique
-        // because we check for strict increasing values of  high32, but we
-        // must also ensure the top 16 bits within each 32-bit bitmap are also
-        // at least unique (we ensure they're strictly increasing as well,
-        // which they must be for a _valid_ bitmap, since it's cheaper to check)
-        int32_t last_bitmap_key = -1;
-        for (int i = 0; i < bitmap32->high_low_container.size; i++) {
-            uint16_t key = bitmap32->high_low_container.keys[i];
-            if (key <= last_bitmap_key) {
-                roaring_bitmap_free(bitmap32);
-                roaring64_bitmap_free(r);
-                return NULL;
-            }
-            last_bitmap_key = key;
-        }
-
         // Insert all containers of the 32-bit bitmap into the 64-bit bitmap.
         move_from_roaring32_offset(r, bitmap32, high32);
         roaring_bitmap_free(bitmap32);

pyroaring/roaring.h

@@ -1,5 +1,5 @@
 // !!! DO NOT EDIT - THIS IS AN AUTO-GENERATED FILE !!!
-// Created by amalgamation.sh on 2024-09-20T14:21:41Z
+// Created by amalgamation.sh on 2024-09-19T00:48:39Z
 
 /*
  * The CRoaring project is under a dual license (Apache/MIT).
@@ -59,11 +59,11 @@
 // /include/roaring/roaring_version.h automatically generated by release.py, do not change by hand
 #ifndef ROARING_INCLUDE_ROARING_VERSION
 #define ROARING_INCLUDE_ROARING_VERSION
-#define ROARING_VERSION "4.1.6"
+#define ROARING_VERSION "4.1.3"
 enum {
     ROARING_VERSION_MAJOR = 4,
     ROARING_VERSION_MINOR = 1,
-    ROARING_VERSION_REVISION = 6
+    ROARING_VERSION_REVISION = 3
 };
 #endif // ROARING_INCLUDE_ROARING_VERSION
 // clang-format on/* end file include/roaring/roaring_version.h */
@@ -1679,10 +1679,6 @@ size_t roaring_bitmap_shrink_to_fit(roaring_bitmap_t *r);
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
  * compatible with little-endian systems.
- *
- * When serializing data to a file, we recommend that you also use
- * checksums so that, at deserialization, you can be confident
- * that you are recovering the correct data.
  */
 size_t roaring_bitmap_serialize(const roaring_bitmap_t *r, char *buf);
 
@@ -1746,10 +1742,7 @@ roaring_bitmap_t *roaring_bitmap_portable_deserialize(const char *buf);
  * https://github.com/RoaringBitmap/RoaringFormatSpec
  *
  * The function itself is safe in the sense that it will not cause buffer
- * overflows: it will not read beyond the scope of the provided buffer
- * (buf,maxbytes).
- *
- * However, for correct operations, it is assumed that the bitmap
+ * overflows. However, for correct operations, it is assumed that the bitmap
  * read was once serialized from a valid bitmap (i.e., it follows the format
  * specification). If you provided an incorrect input (garbage), then the bitmap
  * read may not be in a valid state and following operations may not lead to
@@ -1759,10 +1752,8 @@ roaring_bitmap_t *roaring_bitmap_portable_deserialize(const char *buf);
  * but not for random inputs.
  *
  * You may use roaring_bitmap_internal_validate to check the validity of the
- * bitmap prior to using it.
- *
- * We recommend that you use checksums to check that serialized data corresponds
- * to a serialized bitmap.
+ * bitmap prior to using it. You may also use other strategies to check for
+ * corrupted inputs (e.g., checksums).
  *
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
@@ -1824,10 +1815,6 @@ size_t roaring_bitmap_portable_size_in_bytes(const roaring_bitmap_t *r);
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
  * compatible with little-endian systems.
- *
- * When serializing data to a file, we recommend that you also use
- * checksums so that, at deserialization, you can be confident
- * that you are recovering the correct data.
  */
 size_t roaring_bitmap_portable_serialize(const roaring_bitmap_t *r, char *buf);
 
@@ -1862,10 +1849,6 @@ size_t roaring_bitmap_frozen_size_in_bytes(const roaring_bitmap_t *r);
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
  * compatible with little-endian systems.
- *
- * When serializing data to a file, we recommend that you also use
- * checksums so that, at deserialization, you can be confident
- * that you are recovering the correct data.
  */
 void roaring_bitmap_frozen_serialize(const roaring_bitmap_t *r, char *buf);
 
@@ -2826,10 +2809,6 @@ size_t roaring64_bitmap_portable_size_in_bytes(const roaring64_bitmap_t *r);
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
  * compatible with little-endian systems.
- *
- * When serializing data to a file, we recommend that you also use
- * checksums so that, at deserialization, you can be confident
- * that you are recovering the correct data.
  */
 size_t roaring64_bitmap_portable_serialize(const roaring64_bitmap_t *r,
                                            char *buf);
@@ -2844,17 +2823,14 @@ size_t roaring64_bitmap_portable_deserialize_size(const char *buf,
                                                   size_t maxbytes);
 
 /**
- * Read a bitmap from a serialized buffer (reading up to maxbytes).
+ * Read a bitmap from a serialized buffer safely (reading up to maxbytes).
  * In case of failure, NULL is returned.
  *
  * This is meant to be compatible with other languages
  * https://github.com/RoaringBitmap/RoaringFormatSpec#extension-for-64-bit-implementations
  *
  * The function itself is safe in the sense that it will not cause buffer
- * overflows: it will not read beyond the scope of the provided buffer
- * (buf,maxbytes).
- *
- * However, for correct operations, it is assumed that the bitmap
+ * overflows. However, for correct operations, it is assumed that the bitmap
  * read was once serialized from a valid bitmap (i.e., it follows the format
  * specification). If you provided an incorrect input (garbage), then the bitmap
  * read may not be in a valid state and following operations may not lead to
@@ -2863,12 +2839,6 @@ size_t roaring64_bitmap_portable_deserialize_size(const char *buf,
  * order. This is is guaranteed to happen when serializing an existing bitmap,
  * but not for random inputs.
  *
- * You may use roaring64_bitmap_internal_validate to check the validity of the
- * bitmap prior to using it.
- *
- * We recommend that you use checksums to check that serialized data corresponds
- * to a serialized bitmap.
- *
  * This function is endian-sensitive. If you have a big-endian system (e.g., a
  * mainframe IBM s390x), the data format is going to be big-endian and not
  * compatible with little-endian systems.