Merge pull request #137 from Dr-Emann/2_0_for_real

Add some tests which will fail until we get a 4.0.1 of CRoaring. Don't merge until we update again once CRoaring is fixed.

Also, include the matching CRoaring version in the readme and some clarification on versioning, this will hopefully help clarify somewhat the questions raised in #136. Link to the rendered added section: https://github.com/Dr-Emann/croaring-rs/blob/2_0_for_real/README.md#croaring-version

Author: Dr-Emann

.gitignore

@@ -1,3 +1,5 @@
 target
 .DS_Store
-.idea
+.idea
+README.md.tmp
+Cargo.toml.tmp

Cargo.lock

@@ -15,7 +15,7 @@ test:
 croaring_release_url_base = https://github.com/RoaringBitmap/CRoaring/releases/download
 
 # Fetch the c source amalgamation from a tagged CRoaring release (like `make version=0.9.3 update_croaring`)
-update_croaring: download_croaring bindgen
+update_croaring: download_croaring bindgen update_readme_croaring_version update_croaring_sys_version
 
 download_croaring:
 	rm -f '$(croaring_source)/roaring.c' '$(croaring_source)/roaring.h' '$(croaring_source)/roaring.hh'
@@ -35,6 +35,19 @@ bindgen:
 			-o bindgen_bundled_version.rs \
 			roaring.h
 
+
+# sed -i is a GNU extension, so we use a temporary file explicitly
+update_readme_croaring_version:
+	@echo "Updating README.md with CRoaring version $(version)"
+	@sed -r -e 's_\[CRoaring version `[0-9]+\.[0-9]+\.[0-9]+`\]\([^\)]+\)_[CRoaring version `$(version)`](https://github.com/RoaringBitmap/CRoaring/releases/tag/v$(version))_' README.md > README.md.tmp
+	@mv README.md.tmp README.md
+
+# We don't always want to update the version of croaring-sys dependency in croaring, but we always want to update croaring-sys
+update_croaring_sys_version:
+	@echo "Updating croaring-sys version in Cargo.toml to $(version)"
+	@sed -r -e 's/^version = ".*"/version = "$(version)"/' croaring-sys/Cargo.toml > croaring-sys/Cargo.toml.tmp
+	@mv croaring-sys/Cargo.toml.tmp croaring-sys/Cargo.toml
+
 # Build a c program to (re)generate the example serialized files for testing
 test_serialization_files:
 	cd croaring/tests/data/ && \

Makefile

@@ -124,3 +124,10 @@ cargo bench
 ### Documentation
 
 Current documentation is available at https://docs.rs/croaring/latest/croaring/
+
+## CRoaring Version
+
+This crate uses [CRoaring version `4.1.0`](https://github.com/RoaringBitmap/CRoaring/releases/tag/v4.1.0).
+The version of this crate does not necessarily match the version of CRoaring: the major version of the crate is only
+incremented when there are breaking changes in the Rust API: It is possible (and has happened) that breaking changes
+in the CRoaring C API do not necessitate a major version bump in this crate.

README.md

@@ -5,6 +5,14 @@ pub const ROARING_VERSION_MAJOR: _bindgen_ty_1 = 4;
 pub const ROARING_VERSION_MINOR: _bindgen_ty_1 = 0;
 pub const ROARING_VERSION_REVISION: _bindgen_ty_1 = 0;
 pub type _bindgen_ty_1 = ::core::ffi::c_uint;
+extern "C" {
+    #[doc = " result might be undefined when input_num is zero"]
+    pub fn roaring_trailing_zeroes(input_num: ::core::ffi::c_ulonglong) -> ::core::ffi::c_int;
+}
+extern "C" {
+    #[doc = " result might be undefined when input_num is zero"]
+    pub fn roaring_leading_zeroes(input_num: ::core::ffi::c_ulonglong) -> ::core::ffi::c_int;
+}
 #[doc = " Roaring arrays are array-based key-value pairs having containers as values\n and 16-bit integer keys. A roaring bitmap  might be implemented as such."]
 #[repr(C)]
 #[derive(Debug, Copy, Clone)]
@@ -73,14 +81,6 @@ pub struct roaring_container_iterator_s {
 }
 #[doc = " Roaring-internal type used to iterate within a roaring container."]
 pub type roaring_container_iterator_t = roaring_container_iterator_s;
-extern "C" {
-    #[doc = " result might be undefined when input_num is zero"]
-    pub fn roaring_trailing_zeroes(input_num: ::core::ffi::c_ulonglong) -> ::core::ffi::c_int;
-}
-extern "C" {
-    #[doc = " result might be undefined when input_num is zero"]
-    pub fn roaring_leading_zeroes(input_num: ::core::ffi::c_ulonglong) -> ::core::ffi::c_int;
-}
 #[repr(C)]
 #[derive(Debug, Copy, Clone)]
 pub struct bitset_s {
@@ -925,6 +925,10 @@ extern "C" {
     #[doc = " Remove all values in range [min, max]."]
     pub fn roaring64_bitmap_remove_range_closed(r: *mut roaring64_bitmap_t, min: u64, max: u64);
 }
+extern "C" {
+    #[doc = " Empties the bitmap."]
+    pub fn roaring64_bitmap_clear(r: *mut roaring64_bitmap_t);
+}
 extern "C" {
     #[doc = " Returns true if the provided value is present."]
     pub fn roaring64_bitmap_contains(r: *const roaring64_bitmap_t, val: u64) -> bool;
@@ -977,6 +981,14 @@ extern "C" {
         max: u64,
     ) -> u64;
 }
+extern "C" {
+    #[doc = " Returns the number of elements in the range [min, max]"]
+    pub fn roaring64_bitmap_range_closed_cardinality(
+        r: *const roaring64_bitmap_t,
+        min: u64,
+        max: u64,
+    ) -> u64;
+}
 extern "C" {
     #[doc = " Returns true if the bitmap is empty (cardinality is zero)."]
     pub fn roaring64_bitmap_is_empty(r: *const roaring64_bitmap_t) -> bool;

croaring-sys/CRoaring/bindgen_bundled_version.rs

@@ -1,5 +1,5 @@
 // !!! DO NOT EDIT - THIS IS AN AUTO-GENERATED FILE !!!
-// Created by amalgamation.sh on 2024-05-13T21:29:25Z
+// Created by amalgamation.sh on 2024-07-03T21:30:32Z
 
 /*
  * The CRoaring project is under a dual license (Apache/MIT).
@@ -10770,7 +10770,7 @@ static bool art_node_iterator_lower_bound(const art_node_t *node,
 }
 
 art_iterator_t art_init_iterator(const art_t *art, bool first) {
-    art_iterator_t iterator = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+    art_iterator_t iterator = CROARING_ZERO_INITIALIZER;
     if (art->root == NULL) {
         return iterator;
     }
@@ -10793,8 +10793,11 @@ bool art_iterator_lower_bound(art_iterator_t *iterator,
         // a valid key. Start from the root.
         iterator->frame = 0;
         iterator->depth = 0;
-        return art_node_iterator_lower_bound(art_iterator_node(iterator),
-                                             iterator, key);
+        art_node_t *root = art_iterator_node(iterator);
+        if (root == NULL) {
+            return false;
+        }
+        return art_node_iterator_lower_bound(root, iterator, key);
     }
     int compare_result =
         art_compare_prefix(iterator->key, 0, key, 0, ART_KEY_BYTES);
@@ -10827,15 +10830,15 @@ bool art_iterator_lower_bound(art_iterator_t *iterator,
 }
 
 art_iterator_t art_lower_bound(const art_t *art, const art_key_chunk_t *key) {
-    art_iterator_t iterator = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+    art_iterator_t iterator = CROARING_ZERO_INITIALIZER;
     if (art->root != NULL) {
         art_node_iterator_lower_bound(art->root, &iterator, key);
     }
     return iterator;
 }
 
 art_iterator_t art_upper_bound(const art_t *art, const art_key_chunk_t *key) {
-    art_iterator_t iterator = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+    art_iterator_t iterator = CROARING_ZERO_INITIALIZER;
     if (art->root != NULL) {
         if (art_node_iterator_lower_bound(art->root, &iterator, key) &&
             art_compare_keys(iterator.key, key) == 0) {
@@ -19469,7 +19472,7 @@ roaring_bitmap_t *roaring_bitmap_of(size_t n_args, ...) {
     // todo: could be greatly optimized but we do not expect this call to ever
     // include long lists
     roaring_bitmap_t *answer = roaring_bitmap_create();
-    roaring_bulk_context_t context = {0, 0, 0, 0};
+    roaring_bulk_context_t context = CROARING_ZERO_INITIALIZER;
     va_list ap;
     va_start(ap, n_args);
     for (size_t i = 0; i < n_args; i++) {
@@ -20811,7 +20814,7 @@ roaring_bitmap_t *roaring_bitmap_deserialize(const void *buf) {
         if (bitmap == NULL) {
             return NULL;
         }
-        roaring_bulk_context_t context = {0, 0, 0, 0};
+        roaring_bulk_context_t context = CROARING_ZERO_INITIALIZER;
         for (uint32_t i = 0; i < card; i++) {
             // elems may not be aligned, read with memcpy
             uint32_t elem;
@@ -20854,7 +20857,7 @@ roaring_bitmap_t *roaring_bitmap_deserialize_safe(const void *buf,
         if (bitmap == NULL) {
             return NULL;
         }
-        roaring_bulk_context_t context = {0, 0, 0, 0};
+        roaring_bulk_context_t context = CROARING_ZERO_INITIALIZER;
         for (uint32_t i = 0; i < card; i++) {
             // elems may not be aligned, read with memcpy
             uint32_t elem;
@@ -22780,6 +22783,9 @@ roaring64_bitmap_t *roaring64_bitmap_create(void) {
 }
 
 void roaring64_bitmap_free(roaring64_bitmap_t *r) {
+    if (!r) {
+        return;
+    }
     art_iterator_t it = art_init_iterator(&r->art, /*first=*/true);
     while (it.value != NULL) {
         leaf_t *leaf = (leaf_t *)it.value;
@@ -22856,7 +22862,7 @@ roaring64_bitmap_t *roaring64_bitmap_of_ptr(size_t n_args,
 
 roaring64_bitmap_t *roaring64_bitmap_of(size_t n_args, ...) {
     roaring64_bitmap_t *r = roaring64_bitmap_create();
-    roaring64_bulk_context_t context = {0, 0, 0, 0, 0, 0, 0};
+    roaring64_bulk_context_t context = CROARING_ZERO_INITIALIZER;
     va_list ap;
     va_start(ap, n_args);
     for (size_t i = 0; i < n_args; i++) {
@@ -22949,7 +22955,7 @@ void roaring64_bitmap_add_many(roaring64_bitmap_t *r, size_t n_args,
         return;
     }
     const uint64_t *end = vals + n_args;
-    roaring64_bulk_context_t context = {0, 0, 0, 0, 0, 0, 0};
+    roaring64_bulk_context_t context = CROARING_ZERO_INITIALIZER;
     for (const uint64_t *current_val = vals; current_val != end;
          current_val++) {
         roaring64_bitmap_add_bulk(r, &context, *current_val);
@@ -23273,7 +23279,7 @@ void roaring64_bitmap_remove_many(roaring64_bitmap_t *r, size_t n_args,
         return;
     }
     const uint64_t *end = vals + n_args;
-    roaring64_bulk_context_t context = {0, 0, 0, 0, 0, 0, 0};
+    roaring64_bulk_context_t context = CROARING_ZERO_INITIALIZER;
     for (const uint64_t *current_val = vals; current_val != end;
          current_val++) {
         roaring64_bitmap_remove_bulk(r, &context, *current_val);
@@ -23339,6 +23345,10 @@ void roaring64_bitmap_remove_range_closed(roaring64_bitmap_t *r, uint64_t min,
     remove_range_closed_at(art, max_high48, 0, max_low16);
 }
 
+void roaring64_bitmap_clear(roaring64_bitmap_t *r) {
+    roaring64_bitmap_remove_range_closed(r, 0, UINT64_MAX);
+}
+
 uint64_t roaring64_bitmap_get_cardinality(const roaring64_bitmap_t *r) {
     art_iterator_t it = art_init_iterator(&r->art, /*first=*/true);
     uint64_t cardinality = 0;
@@ -23356,7 +23366,17 @@ uint64_t roaring64_bitmap_range_cardinality(const roaring64_bitmap_t *r,
     if (min >= max) {
         return 0;
     }
-    max--;  // A closed range is easier to work with.
+    // Convert to a closed range
+    // No underflow here: passing the above condition implies min < max, so
+    // there is a number less than max
+    return roaring64_bitmap_range_closed_cardinality(r, min, max - 1);
+}
+
+uint64_t roaring64_bitmap_range_closed_cardinality(const roaring64_bitmap_t *r,
+                                                   uint64_t min, uint64_t max) {
+    if (min > max) {
+        return 0;
+    }
 
     uint64_t cardinality = 0;
     uint8_t min_high48[ART_KEY_BYTES];