ucx/array_list.c

branch
newapi
changeset 324
ce13a778654a
parent 253
087cc9216f28
--- a/ucx/array_list.c	Thu Oct 03 18:54:19 2024 +0200
+++ b/ucx/array_list.c	Sun Oct 06 12:00:31 2024 +0200
@@ -55,7 +55,7 @@
         size_t *size,
         size_t *capacity,
         size_t index,
-        void const *src,
+        const void *src,
         size_t elem_size,
         size_t elem_count,
         struct cx_array_reallocator_s *reallocator
@@ -120,6 +120,173 @@
     return CX_ARRAY_SUCCESS;
 }
 
+enum cx_array_result cx_array_insert_sorted(
+        void **target,
+        size_t *size,
+        size_t *capacity,
+        cx_compare_func cmp_func,
+        const void *sorted_data,
+        size_t elem_size,
+        size_t elem_count,
+        struct cx_array_reallocator_s *reallocator
+) {
+    // assert pointers
+    assert(target != NULL);
+    assert(size != NULL);
+    assert(capacity != NULL);
+    assert(cmp_func != NULL);
+    assert(sorted_data != NULL);
+    assert(reallocator != NULL);
+
+    // corner case
+    if (elem_count == 0) return 0;
+
+    // store some counts
+    size_t old_size = *size;
+    size_t needed_capacity = old_size + elem_count;
+
+    // if we need more than we have, try a reallocation
+    if (needed_capacity > *capacity) {
+        size_t new_capacity = needed_capacity - (needed_capacity % 16) + 16;
+        void *new_mem = reallocator->realloc(
+                *target, new_capacity, elem_size, reallocator
+        );
+        if (new_mem == NULL) {
+            // give it up right away, there is no contract
+            // that requires us to insert as much as we can
+            return CX_ARRAY_REALLOC_FAILED;
+        }
+        *target = new_mem;
+        *capacity = new_capacity;
+    }
+
+    // now we have guaranteed that we can insert everything
+    size_t new_size = old_size + elem_count;
+    *size = new_size;
+
+    // declare the source and destination indices/pointers
+    size_t si = 0, di = 0;
+    const char *src = sorted_data;
+    char *dest = *target;
+
+    // find the first insertion point
+    di = cx_array_binary_search_sup(dest, old_size, elem_size, src, cmp_func);
+    dest += di * elem_size;
+
+    // move the remaining elements in the array completely to the right
+    // we will call it the "buffer" for parked elements
+    size_t buf_size = old_size - di;
+    size_t bi = new_size - buf_size;
+    char *bptr = ((char *) *target) + bi * elem_size;
+    memmove(bptr, dest, buf_size * elem_size);
+
+    // while there are both source and buffered elements left,
+    // copy them interleaving
+    while (si < elem_count && bi < new_size) {
+        // determine how many source elements can be inserted
+        size_t copy_len, bytes_copied;
+        copy_len = cx_array_binary_search_sup(
+                src,
+                elem_count - si,
+                elem_size,
+                bptr,
+                cmp_func
+        );
+
+        // copy the source elements
+        bytes_copied = copy_len * elem_size;
+        memcpy(dest, src, bytes_copied);
+        dest += bytes_copied;
+        src += bytes_copied;
+        si += copy_len;
+
+        // when all source elements are in place, we are done
+        if (si >= elem_count) break;
+
+        // determine how many buffered elements need to be restored
+        copy_len = cx_array_binary_search_sup(
+                bptr,
+                new_size - bi,
+                elem_size,
+                src,
+                cmp_func
+        );
+
+        // restore the buffered elements
+        bytes_copied = copy_len * elem_size;
+        memmove(dest, bptr, bytes_copied);
+        dest += bytes_copied;
+        bptr += bytes_copied;
+        bi += copy_len;
+    }
+
+    // still source elements left? simply append them
+    if (si < elem_count) {
+        memcpy(dest, src, elem_size * (elem_count - si));
+    }
+
+    // still buffer elements left?
+    // don't worry, we already moved them to the correct place
+
+    return CX_ARRAY_SUCCESS;
+}
+
+size_t cx_array_binary_search_inf(
+        const void *arr,
+        size_t size,
+        size_t elem_size,
+        const void *elem,
+        cx_compare_func cmp_func
+) {
+    // special case: empty array
+    if (size == 0) return 0;
+
+    // declare a variable that will contain the compare results
+    int result;
+
+    // cast the array pointer to something we can use offsets with
+    const char *array = arr;
+
+    // check the first array element
+    result = cmp_func(elem, array);
+    if (result < 0) {
+        return size;
+    } else if (result == 0) {
+        return 0;
+    }
+
+    // check the last array element
+    result = cmp_func(elem, array + elem_size * (size - 1));
+    if (result >= 0) {
+        return size - 1;
+    }
+
+    // the element is now guaranteed to be somewhere in the list
+    // so start the binary search
+    size_t left_index = 1;
+    size_t right_index = size - 1;
+    size_t pivot_index;
+
+    while (left_index <= right_index) {
+        pivot_index = left_index + (right_index - left_index) / 2;
+        const char *arr_elem = array + pivot_index * elem_size;
+        result = cmp_func(elem, arr_elem);
+        if (result == 0) {
+            // found it!
+            return pivot_index;
+        } else if (result < 0) {
+            // element is smaller than pivot, continue search left
+            right_index = pivot_index - 1;
+        } else {
+            // element is larger than pivot, continue search right
+            left_index = pivot_index + 1;
+        }
+    }
+
+    // report the largest upper bound
+    return result < 0 ? (pivot_index - 1) : pivot_index;
+}
+
 #ifndef CX_ARRAY_SWAP_SBO_SIZE
 #define CX_ARRAY_SWAP_SBO_SIZE 128
 #endif
@@ -180,7 +347,7 @@
         struct cx_array_reallocator_s *alloc
 ) {
     // retrieve the pointer to the list allocator
-    CxAllocator const *al = alloc->ptr1;
+    const CxAllocator *al = alloc->ptr1;
 
     // use the list allocator to reallocate the memory
     return cxRealloc(al, array, capacity * elem_size);
@@ -191,49 +358,49 @@
 
     char *ptr = arl->data;
 
-    if (list->simple_destructor) {
-        for (size_t i = 0; i < list->size; i++) {
+    if (list->collection.simple_destructor) {
+        for (size_t i = 0; i < list->collection.size; i++) {
             cx_invoke_simple_destructor(list, ptr);
-            ptr += list->item_size;
+            ptr += list->collection.elem_size;
         }
     }
-    if (list->advanced_destructor) {
-        for (size_t i = 0; i < list->size; i++) {
+    if (list->collection.advanced_destructor) {
+        for (size_t i = 0; i < list->collection.size; i++) {
             cx_invoke_advanced_destructor(list, ptr);
-            ptr += list->item_size;
+            ptr += list->collection.elem_size;
         }
     }
 
-    cxFree(list->allocator, arl->data);
-    cxFree(list->allocator, list);
+    cxFree(list->collection.allocator, arl->data);
+    cxFree(list->collection.allocator, list);
 }
 
 static size_t cx_arl_insert_array(
         struct cx_list_s *list,
         size_t index,
-        void const *array,
+        const void *array,
         size_t n
 ) {
     // out of bounds and special case check
-    if (index > list->size || n == 0) return 0;
+    if (index > list->collection.size || n == 0) return 0;
 
     // get a correctly typed pointer to the list
     cx_array_list *arl = (cx_array_list *) list;
 
     // do we need to move some elements?
-    if (index < list->size) {
-        char const *first_to_move = (char const *) arl->data;
-        first_to_move += index * list->item_size;
-        size_t elems_to_move = list->size - index;
+    if (index < list->collection.size) {
+        const char *first_to_move = (const char *) arl->data;
+        first_to_move += index * list->collection.elem_size;
+        size_t elems_to_move = list->collection.size - index;
         size_t start_of_moved = index + n;
 
         if (CX_ARRAY_SUCCESS != cx_array_copy(
                 &arl->data,
-                &list->size,
+                &list->collection.size,
                 &arl->capacity,
                 start_of_moved,
                 first_to_move,
-                list->item_size,
+                list->collection.elem_size,
                 elems_to_move,
                 &arl->reallocator
         )) {
@@ -249,11 +416,36 @@
     // place the new elements
     if (CX_ARRAY_SUCCESS == cx_array_copy(
             &arl->data,
-            &list->size,
+            &list->collection.size,
             &arl->capacity,
             index,
             array,
-            list->item_size,
+            list->collection.elem_size,
+            n,
+            &arl->reallocator
+    )) {
+        return n;
+    } else {
+        // array list implementation is "all or nothing"
+        return 0;
+    }
+}
+
+static size_t cx_arl_insert_sorted(
+        struct cx_list_s *list,
+        const void *sorted_data,
+        size_t n
+) {
+    // get a correctly typed pointer to the list
+    cx_array_list *arl = (cx_array_list *) list;
+
+    if (CX_ARRAY_SUCCESS == cx_array_insert_sorted(
+            &arl->data,
+            &list->collection.size,
+            &arl->capacity,
+            list->collection.cmpfunc,
+            sorted_data,
+            list->collection.elem_size,
             n,
             &arl->reallocator
     )) {
@@ -267,31 +459,37 @@
 static int cx_arl_insert_element(
         struct cx_list_s *list,
         size_t index,
-        void const *element
+        const void *element
 ) {
     return 1 != cx_arl_insert_array(list, index, element, 1);
 }
 
 static int cx_arl_insert_iter(
-        struct cx_mut_iterator_s *iter,
-        void const *elem,
+        struct cx_iterator_s *iter,
+        const void *elem,
         int prepend
 ) {
-    struct cx_list_s *list = iter->src_handle;
-    if (iter->index < list->size) {
+    struct cx_list_s *list = iter->src_handle.m;
+    if (iter->index < list->collection.size) {
         int result = cx_arl_insert_element(
                 list,
                 iter->index + 1 - prepend,
                 elem
         );
-        if (result == 0 && prepend != 0) {
-            iter->index++;
-            iter->elem_handle = ((char *) iter->elem_handle) + list->item_size;
+        if (result == 0) {
+            iter->elem_count++;
+            if (prepend != 0) {
+                iter->index++;
+                iter->elem_handle = ((char *) iter->elem_handle) + list->collection.elem_size;
+            }
         }
         return result;
     } else {
-        int result = cx_arl_insert_element(list, list->size, elem);
-        iter->index = list->size;
+        int result = cx_arl_insert_element(list, list->collection.size, elem);
+        if (result == 0) {
+            iter->elem_count++;
+            iter->index = list->collection.size;
+        }
         return result;
     }
 }
@@ -303,28 +501,28 @@
     cx_array_list *arl = (cx_array_list *) list;
 
     // out-of-bounds check
-    if (index >= list->size) {
+    if (index >= list->collection.size) {
         return 1;
     }
 
     // content destruction
-    cx_invoke_destructor(list, ((char *) arl->data) + index * list->item_size);
+    cx_invoke_destructor(list, ((char *) arl->data) + index * list->collection.elem_size);
 
     // short-circuit removal of last element
-    if (index == list->size - 1) {
-        list->size--;
+    if (index == list->collection.size - 1) {
+        list->collection.size--;
         return 0;
     }
 
     // just move the elements starting at index to the left
     int result = cx_array_copy(
             &arl->data,
-            &list->size,
+            &list->collection.size,
             &arl->capacity,
             index,
-            ((char *) arl->data) + (index + 1) * list->item_size,
-            list->item_size,
-            list->size - index - 1,
+            ((char *) arl->data) + (index + 1) * list->collection.elem_size,
+            list->collection.elem_size,
+            list->collection.size - index - 1,
             &arl->reallocator
     );
 
@@ -332,32 +530,32 @@
     assert(result == 0);
 
     // decrease the size
-    list->size--;
+    list->collection.size--;
 
     return 0;
 }
 
 static void cx_arl_clear(struct cx_list_s *list) {
-    if (list->size == 0) return;
+    if (list->collection.size == 0) return;
 
     cx_array_list *arl = (cx_array_list *) list;
     char *ptr = arl->data;
 
-    if (list->simple_destructor) {
-        for (size_t i = 0; i < list->size; i++) {
+    if (list->collection.simple_destructor) {
+        for (size_t i = 0; i < list->collection.size; i++) {
             cx_invoke_simple_destructor(list, ptr);
-            ptr += list->item_size;
+            ptr += list->collection.elem_size;
         }
     }
-    if (list->advanced_destructor) {
-        for (size_t i = 0; i < list->size; i++) {
+    if (list->collection.advanced_destructor) {
+        for (size_t i = 0; i < list->collection.size; i++) {
             cx_invoke_advanced_destructor(list, ptr);
-            ptr += list->item_size;
+            ptr += list->collection.elem_size;
         }
     }
 
-    memset(arl->data, 0, list->size * list->item_size);
-    list->size = 0;
+    memset(arl->data, 0, list->collection.size * list->collection.elem_size);
+    list->collection.size = 0;
 }
 
 static int cx_arl_swap(
@@ -365,20 +563,20 @@
         size_t i,
         size_t j
 ) {
-    if (i >= list->size || j >= list->size) return 1;
+    if (i >= list->collection.size || j >= list->collection.size) return 1;
     cx_array_list *arl = (cx_array_list *) list;
-    cx_array_swap(arl->data, list->item_size, i, j);
+    cx_array_swap(arl->data, list->collection.elem_size, i, j);
     return 0;
 }
 
 static void *cx_arl_at(
-        struct cx_list_s const *list,
+        const struct cx_list_s *list,
         size_t index
 ) {
-    if (index < list->size) {
-        cx_array_list const *arl = (cx_array_list const *) list;
+    if (index < list->collection.size) {
+        const cx_array_list *arl = (const cx_array_list *) list;
         char *space = arl->data;
-        return space + index * list->item_size;
+        return space + index * list->collection.elem_size;
     } else {
         return NULL;
     }
@@ -386,15 +584,15 @@
 
 static ssize_t cx_arl_find_remove(
         struct cx_list_s *list,
-        void const *elem,
+        const void *elem,
         bool remove
 ) {
-    assert(list->cmpfunc != NULL);
-    assert(list->size < SIZE_MAX / 2);
-    char *cur = ((cx_array_list const *) list)->data;
+    assert(list->collection.cmpfunc != NULL);
+    assert(list->collection.size < SIZE_MAX / 2);
+    char *cur = ((const cx_array_list *) list)->data;
 
-    for (ssize_t i = 0; i < (ssize_t) list->size; i++) {
-        if (0 == list->cmpfunc(elem, cur)) {
+    for (ssize_t i = 0; i < (ssize_t) list->collection.size; i++) {
+        if (0 == list->collection.cmpfunc(elem, cur)) {
             if (remove) {
                 if (0 == cx_arl_remove(list, i)) {
                     return i;
@@ -405,117 +603,106 @@
                 return i;
             }
         }
-        cur += list->item_size;
+        cur += list->collection.elem_size;
     }
 
     return -1;
 }
 
 static void cx_arl_sort(struct cx_list_s *list) {
-    assert(list->cmpfunc != NULL);
+    assert(list->collection.cmpfunc != NULL);
     qsort(((cx_array_list *) list)->data,
-          list->size,
-          list->item_size,
-          list->cmpfunc
+          list->collection.size,
+          list->collection.elem_size,
+          list->collection.cmpfunc
     );
 }
 
 static int cx_arl_compare(
-        struct cx_list_s const *list,
-        struct cx_list_s const *other
+        const struct cx_list_s *list,
+        const struct cx_list_s *other
 ) {
-    assert(list->cmpfunc != NULL);
-    if (list->size == other->size) {
-        char const *left = ((cx_array_list const *) list)->data;
-        char const *right = ((cx_array_list const *) other)->data;
-        for (size_t i = 0; i < list->size; i++) {
-            int d = list->cmpfunc(left, right);
+    assert(list->collection.cmpfunc != NULL);
+    if (list->collection.size == other->collection.size) {
+        const char *left = ((const cx_array_list *) list)->data;
+        const char *right = ((const cx_array_list *) other)->data;
+        for (size_t i = 0; i < list->collection.size; i++) {
+            int d = list->collection.cmpfunc(left, right);
             if (d != 0) {
                 return d;
             }
-            left += list->item_size;
-            right += other->item_size;
+            left += list->collection.elem_size;
+            right += other->collection.elem_size;
         }
         return 0;
     } else {
-        return list->size < other->size ? -1 : 1;
+        return list->collection.size < other->collection.size ? -1 : 1;
     }
 }
 
 static void cx_arl_reverse(struct cx_list_s *list) {
-    if (list->size < 2) return;
-    void *data = ((cx_array_list const *) list)->data;
-    size_t half = list->size / 2;
+    if (list->collection.size < 2) return;
+    void *data = ((const cx_array_list *) list)->data;
+    size_t half = list->collection.size / 2;
     for (size_t i = 0; i < half; i++) {
-        cx_array_swap(data, list->item_size, i, list->size - 1 - i);
+        cx_array_swap(data, list->collection.elem_size, i, list->collection.size - 1 - i);
     }
 }
 
-static bool cx_arl_iter_valid(void const *it) {
-    struct cx_iterator_s const *iter = it;
-    struct cx_list_s const *list = iter->src_handle;
-    return iter->index < list->size;
+static bool cx_arl_iter_valid(const void *it) {
+    const struct cx_iterator_s *iter = it;
+    const struct cx_list_s *list = iter->src_handle.c;
+    return iter->index < list->collection.size;
 }
 
-static void *cx_arl_iter_current(void const *it) {
-    struct cx_iterator_s const *iter = it;
+static void *cx_arl_iter_current(const void *it) {
+    const struct cx_iterator_s *iter = it;
     return iter->elem_handle;
 }
 
 static void cx_arl_iter_next(void *it) {
-    struct cx_iterator_base_s *itbase = it;
-    if (itbase->remove) {
-        struct cx_mut_iterator_s *iter = it;
-        itbase->remove = false;
-        cx_arl_remove(iter->src_handle, iter->index);
+    struct cx_iterator_s *iter = it;
+    if (iter->base.remove) {
+        iter->base.remove = false;
+        cx_arl_remove(iter->src_handle.m, iter->index);
     } else {
-        struct cx_iterator_s *iter = it;
         iter->index++;
         iter->elem_handle =
                 ((char *) iter->elem_handle)
-                + ((struct cx_list_s const *) iter->src_handle)->item_size;
+                + ((const struct cx_list_s *) iter->src_handle.c)->collection.elem_size;
     }
 }
 
 static void cx_arl_iter_prev(void *it) {
-    struct cx_iterator_base_s *itbase = it;
-    struct cx_mut_iterator_s *iter = it;
-    cx_array_list *const list = iter->src_handle;
-    if (itbase->remove) {
-        itbase->remove = false;
-        cx_arl_remove(iter->src_handle, iter->index);
+    struct cx_iterator_s *iter = it;
+    const cx_array_list *list = iter->src_handle.c;
+    if (iter->base.remove) {
+        iter->base.remove = false;
+        cx_arl_remove(iter->src_handle.m, iter->index);
     }
     iter->index--;
-    if (iter->index < list->base.size) {
+    if (iter->index < list->base.collection.size) {
         iter->elem_handle = ((char *) list->data)
-                            + iter->index * list->base.item_size;
+                            + iter->index * list->base.collection.elem_size;
     }
 }
 
-static bool cx_arl_iter_flag_rm(void *it) {
-    struct cx_iterator_base_s *iter = it;
-    if (iter->mutating) {
-        iter->remove = true;
-        return true;
-    } else {
-        return false;
-    }
-}
 
 static struct cx_iterator_s cx_arl_iterator(
-        struct cx_list_s const *list,
+        const struct cx_list_s *list,
         size_t index,
         bool backwards
 ) {
     struct cx_iterator_s iter;
 
     iter.index = index;
-    iter.src_handle = list;
+    iter.src_handle.c = list;
     iter.elem_handle = cx_arl_at(list, index);
+    iter.elem_size = list->collection.elem_size;
+    iter.elem_count = list->collection.size;
     iter.base.valid = cx_arl_iter_valid;
     iter.base.current = cx_arl_iter_current;
     iter.base.next = backwards ? cx_arl_iter_prev : cx_arl_iter_next;
-    iter.base.flag_removal = cx_arl_iter_flag_rm;
     iter.base.remove = false;
     iter.base.mutating = false;
 
@@ -526,6 +713,7 @@
         cx_arl_destructor,
         cx_arl_insert_element,
         cx_arl_insert_array,
+        cx_arl_insert_sorted,
         cx_arl_insert_iter,
         cx_arl_remove,
         cx_arl_clear,
@@ -539,9 +727,9 @@
 };
 
 CxList *cxArrayListCreate(
-        CxAllocator const *allocator,
+        const CxAllocator *allocator,
         cx_compare_func comparator,
-        size_t item_size,
+        size_t elem_size,
         size_t initial_capacity
 ) {
     if (allocator == NULL) {
@@ -552,20 +740,20 @@
     if (list == NULL) return NULL;
 
     list->base.cl = &cx_array_list_class;
-    list->base.allocator = allocator;
+    list->base.collection.allocator = allocator;
     list->capacity = initial_capacity;
 
-    if (item_size > 0) {
-        list->base.item_size = item_size;
-        list->base.cmpfunc = comparator;
+    if (elem_size > 0) {
+        list->base.collection.elem_size = elem_size;
+        list->base.collection.cmpfunc = comparator;
     } else {
-        item_size = sizeof(void *);
-        list->base.cmpfunc = comparator == NULL ? cx_cmp_ptr : comparator;
+        elem_size = sizeof(void *);
+        list->base.collection.cmpfunc = comparator == NULL ? cx_cmp_ptr : comparator;
         cxListStorePointers((CxList *) list);
     }
 
-    // allocate the array after the real item_size is known
-    list->data = cxCalloc(allocator, initial_capacity, item_size);
+    // allocate the array after the real elem_size is known
+    list->data = cxCalloc(allocator, initial_capacity, elem_size);
     if (list->data == NULL) {
         cxFree(allocator, list);
         return NULL;

mercurial