// Functions that operate on arrays

#include <ctype.h>
#include <err.h>
#include <gc.h>
#include <gc/cord.h>
#include <math.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdlib.h>
#include <sys/param.h>

#include "array.h"
#include "functions.h"
#include "halfsiphash.h"
#include "integers.h"
#include "table.h"
#include "types.h"
#include "util.h"

static inline int64_t get_padded_item_size(const TypeInfo *info)
{
    int64_t size = info->ArrayInfo.item->size;
    if (info->ArrayInfo.item->align > 1 && size % info->ArrayInfo.item->align)
        size += info->ArrayInfo.item->align - (size % info->ArrayInfo.item->align); // padding
    return size;
}

// Replace the array's .data pointer with a new pointer to a copy of the
// data that is compacted and has a stride of exactly `padded_item_size`
public void Array$compact(array_t *arr, int64_t padded_item_size)
{
    void *copy = NULL;
    if (arr->length > 0) {
        copy = arr->atomic ? GC_MALLOC_ATOMIC(arr->length * padded_item_size) : GC_MALLOC(arr->length * padded_item_size);
        if ((int64_t)arr->stride == padded_item_size) {
            memcpy(copy, arr->data, arr->length * padded_item_size);
        } else {
            for (int64_t i = 0; i < arr->length; i++)
                memcpy(copy + i*padded_item_size, arr->data + arr->stride*i, padded_item_size);
        }
    }
    *arr = (array_t){
        .data=copy,
        .length=arr->length,
        .stride=padded_item_size,
        .atomic=arr->atomic,
    };
}

public void Array$insert(array_t *arr, const void *item, Int_t int_index, int64_t padded_item_size)
{
    int64_t index = Int_to_Int64(int_index, false);
    if (index <= 0) index = arr->length + index + 1;

    if (index < 1) index = 1;
    else if (index > (int64_t)arr->length + 1)
        fail("Invalid insertion index %ld for an array with length %ld", index, arr->length);

    if (!arr->data) {
        arr->free = 4;
        arr->data = arr->atomic ? GC_MALLOC_ATOMIC(arr->free * padded_item_size) : GC_MALLOC(arr->free * padded_item_size);
        arr->stride = padded_item_size;
    } else if (arr->free < 1 || arr->data_refcount != 0 || (int64_t)arr->stride != padded_item_size) {
        arr->free = MIN(ARRAY_MAX_FREE_ENTRIES, MAX(8, arr->length/4));
        void *copy = arr->atomic ? GC_MALLOC_ATOMIC((arr->length + arr->free) * padded_item_size) : GC_MALLOC((arr->length + arr->free) * padded_item_size);
        for (int64_t i = 0; i < index-1; i++)
            memcpy(copy + i*padded_item_size, arr->data + arr->stride*i, padded_item_size);
        for (int64_t i = index-1; i < (int64_t)arr->length; i++)
            memcpy(copy + (i+1)*padded_item_size, arr->data + arr->stride*i, padded_item_size);
        arr->data = copy;
        arr->data_refcount = 0;
        arr->stride = padded_item_size;
    } else {
        if (index != arr->length+1)
            memmove(
                arr->data + index*padded_item_size,
                arr->data + (index-1)*padded_item_size,
                (arr->length - index + 1)*padded_item_size);
    }
    assert(arr->free > 0);
    --arr->free;
    ++arr->length;
    memcpy((void*)arr->data + (index-1)*padded_item_size, item, padded_item_size);
}

public void Array$insert_all(array_t *arr, array_t to_insert, Int_t int_index, int64_t padded_item_size)
{
    int64_t index = Int_to_Int64(int_index, false);
    if (to_insert.length == 0)
        return;

    if (!arr->data) {
        *arr = to_insert;
        ARRAY_INCREF(*arr);
        return;
    }

    if (index < 1) index = arr->length + index + 1;

    if (index < 1) index = 1;
    else if (index > (int64_t)arr->length + 1)
        fail("Invalid insertion index %ld for an array with length %ld", index, arr->length);

    if ((int64_t)arr->free >= (int64_t)to_insert.length // Adequate free space
        && arr->data_refcount == 0 // Not aliased memory
        && (int64_t)arr->stride == padded_item_size) { // Contiguous array
        // If we can fit this within the array's preallocated free space, do that:
        arr->free -= to_insert.length;
        arr->length += to_insert.length;
        if (index != arr->length+1)
            memmove((void*)arr->data + index*padded_item_size,
                    arr->data + (index-1)*padded_item_size,
                    (arr->length - index + to_insert.length-1)*padded_item_size);
        for (int64_t i = 0; i < to_insert.length; i++)
            memcpy((void*)arr->data + (index-1 + i)*padded_item_size,
                   to_insert.data + i*to_insert.stride, padded_item_size);
    } else {
        // Otherwise, allocate a new chunk of memory for the array and populate it:
        int64_t new_len = arr->length + to_insert.length;
        arr->free = MIN(ARRAY_MAX_FREE_ENTRIES, MAX(8, new_len/4));
        void *data = arr->atomic ? GC_MALLOC_ATOMIC((new_len + arr->free) * padded_item_size)
            : GC_MALLOC((new_len + arr->free) * padded_item_size);
        void *p = data;

        // Copy first chunk of `arr` if needed:
        if (index > 1) {
            if (arr->stride == padded_item_size) {
                p = mempcpy(p, arr->data, (index-1)*padded_item_size);
            } else {
                for (int64_t i = 0; i < index-1; i++)
                    p = mempcpy(p, arr->data + arr->stride*i, padded_item_size);
            }
        }

        // Copy `to_insert`
        if (to_insert.stride == padded_item_size) {
            p = mempcpy(p, to_insert.data, to_insert.length*padded_item_size);
        } else {
            for (int64_t i = 0; i < index-1; i++)
                p = mempcpy(p, to_insert.data + to_insert.stride*i, padded_item_size);
        }

        // Copy last chunk of `arr` if needed:
        if (index < arr->length + 1) {
            if (arr->stride == padded_item_size) {
                p = mempcpy(p, arr->data + padded_item_size*(index-1), (arr->length - index + 1)*padded_item_size);
            } else {
                for (int64_t i = index-1; i < arr->length-1; i++)
                    p = mempcpy(p, arr->data + arr->stride*i, padded_item_size);
            }
        }
        arr->length = new_len;
        arr->stride = padded_item_size;
        arr->data = data;
        arr->data_refcount = 0;
    }
}

public void Array$remove_at(array_t *arr, Int_t int_index, Int_t int_count, int64_t padded_item_size)
{
    int64_t index = Int_to_Int64(int_index, false);
    if (index < 1) index = arr->length + index + 1;

    int64_t count = Int_to_Int64(int_count, false);
    if (index < 1 || index > (int64_t)arr->length || count < 1) return;

    if (count > arr->length - index + 1)
        count = (arr->length - index) + 1;

    if (index == 1) {
        arr->data += arr->stride * count;
    } else if (index + count > arr->length) {
        if (arr->free >= 0)
            arr->free += count;
    } else if (arr->data_refcount != 0 || (int64_t)arr->stride != padded_item_size) {
        void *copy = arr->atomic ? GC_MALLOC_ATOMIC((arr->length-1) * padded_item_size) : GC_MALLOC((arr->length-1) * padded_item_size);
        for (int64_t src = 1, dest = 1; src <= (int64_t)arr->length; src++) {
            if (src < index || src >= index + count) {
                memcpy(copy + (dest - 1)*padded_item_size, arr->data + arr->stride*(src - 1), padded_item_size);
                ++dest;
            }
        }
        arr->data = copy;
        arr->free = 0;
        arr->data_refcount = 0;
    } else {
        memmove((void*)arr->data + (index-1)*padded_item_size, arr->data + (index-1 + count)*padded_item_size, (arr->length - index + count - 1)*padded_item_size);
        arr->free += count;
    }
    arr->length -= count;
    if (arr->length == 0) arr->data = NULL;
}

public void Array$remove_item(array_t *arr, void *item, Int_t max_removals, const TypeInfo *type)
{
    int64_t padded_item_size = get_padded_item_size(type);
    const Int_t ZERO = (Int_t){.small=(0<<2)|1};
    const Int_t ONE = (Int_t){.small=(1<<2)|1};
    const TypeInfo *item_type = type->ArrayInfo.item;
    for (int64_t i = 0; i < arr->length; ) {
        if (max_removals.small == ZERO.small) // zero
            break;

        if (generic_equal(item, arr->data + i*arr->stride, item_type)) {
            Array$remove_at(arr, I(i+1), ONE, padded_item_size);
            max_removals = Int$minus(max_removals, ONE);
        } else {
            i++;
        }
    }
}

public Int_t Array$find(array_t arr, void *item, const TypeInfo *type)
{
    const TypeInfo *item_type = type->ArrayInfo.item;
    for (int64_t i = 0; i < arr.length; i++) {
        if (generic_equal(item, arr.data + i*arr.stride, item_type))
            return I(i+1);
    }
    return I(0);
}

public void *Array$first(array_t arr, closure_t predicate)
{
    bool (*is_good)(void*, void*) = (void*)predicate.fn;
    for (int64_t i = 0; i < arr.length; i++) {
        if (is_good(arr.data + i*arr.stride, predicate.userdata))
            return arr.data + i*arr.stride;
    }
    return NULL;
}


public void Array$sort(array_t *arr, closure_t comparison, int64_t padded_item_size)
{
    if (arr->data_refcount != 0 || (int64_t)arr->stride != padded_item_size)
        Array$compact(arr, padded_item_size);

    qsort_r(arr->data, arr->length, padded_item_size, comparison.fn, comparison.userdata);
}

public array_t Array$sorted(array_t arr, closure_t comparison, int64_t padded_item_size)
{
    Array$compact(&arr, padded_item_size);
    qsort_r(arr.data, arr.length, padded_item_size, comparison.fn, comparison.userdata);
    return arr;
}

public void Array$shuffle(array_t *arr, int64_t padded_item_size)
{
    if (arr->data_refcount != 0 || (int64_t)arr->stride != padded_item_size)
        Array$compact(arr, padded_item_size);

    char tmp[padded_item_size];
    for (int64_t i = arr->length-1; i > 1; i--) {
        int64_t j = arc4random_uniform(i+1);
        memcpy(tmp, arr->data + i*padded_item_size, padded_item_size);
        memcpy((void*)arr->data + i*padded_item_size, arr->data + j*padded_item_size, padded_item_size);
        memcpy((void*)arr->data + j*padded_item_size, tmp, padded_item_size);
    }
}

public array_t Array$shuffled(array_t arr, int64_t padded_item_size)
{
    Array$compact(&arr, padded_item_size);
    Array$shuffle(&arr, padded_item_size);
    return arr;
}

public void *Array$random(array_t arr)
{
    if (arr.length == 0)
        return NULL; // fail("Cannot get a random item from an empty array!");
    int64_t index = arc4random_uniform(arr.length);
    return arr.data + arr.stride*index;
}

public table_t Array$counts(array_t arr, const TypeInfo *type)
{
    table_t counts = {};
    const TypeInfo count_type = {.size=sizeof(table_t), .align=__alignof__(table_t),
        .tag=TableInfo, .TableInfo.key=type->ArrayInfo.item, .TableInfo.value=&$Int};
    for (int64_t i = 0; i < arr.length; i++) {
        void *key = arr.data + i*arr.stride;
        int64_t *count = Table$get(counts, key, &count_type);
        int64_t val = count ? *count + 1 : 1;
        Table$set(&counts, key, &val, &count_type);
    }
    return counts;
}

public array_t Array$sample(array_t arr, Int_t int_n, array_t weights, int64_t padded_item_size)
{
    int64_t n = Int_to_Int64(int_n, false);
    if (arr.length == 0 || n <= 0)
        return (array_t){};

    array_t selected = {
        .data=arr.atomic ? GC_MALLOC_ATOMIC(n * padded_item_size) : GC_MALLOC(n * padded_item_size),
        .length=n,
        .stride=padded_item_size, .atomic=arr.atomic};

    double total = 0.0;
    for (int64_t i = 0; i < weights.length && i < arr.length; i++) {
        double weight = *(double*)(weights.data + weights.stride*i);
        if (isinf(weight))
            fail("Infinite weight!");
        else if (isnan(weight))
            fail("NaN weight!");
        else if (weight < 0.0)
            fail("Negative weight!");
        else
            total += weight;
    }

    if (isinf(total))
        fail("Sample weights have overflowed to infinity");

    if (total == 0.0) {
        for (int64_t i = 0; i < n; i++) {
            int64_t index = arc4random_uniform(arr.length);
            memcpy(selected.data + i*padded_item_size, arr.data + arr.stride*index, padded_item_size);
        }
    } else {
        double inverse_average = (double)arr.length / total;

        struct {
            int64_t alias;
            double odds;
        } aliases[arr.length] = {};

        for (int64_t i = 0; i < arr.length; i++) {
            double weight = i >= weights.length ? 0.0 : *(double*)(weights.data + weights.stride*i);
            aliases[i].odds = weight * inverse_average;
            aliases[i].alias = -1;
        }

        int64_t small = 0;
        for (int64_t big = 0; big < arr.length; big++) {
            while (aliases[big].odds >= 1.0) {
                while (small < arr.length && (aliases[small].odds >= 1.0 || aliases[small].alias != -1))
                    ++small;

                if (small >= arr.length) {
                    aliases[big].odds = 1.0;
                    aliases[big].alias = big;
                    break;
                }

                aliases[small].alias = big;
                aliases[big].odds = (aliases[small].odds + aliases[big].odds) - 1.0;
            }
            if (big < small) small = big;
        }

        for (int64_t i = small; i < arr.length; i++)
            if (aliases[i].alias == -1)
                aliases[i].alias = i;

        for (int64_t i = 0; i < n; i++) {
            double r = drand48() * arr.length;
            int64_t index = (int64_t)r;
            if ((r - (double)index) > aliases[index].odds)
                index = aliases[index].alias;
            memcpy(selected.data + i*selected.stride, arr.data + index*arr.stride, padded_item_size);
        }
    }
    return selected;
}

public array_t Array$from(array_t array, Int_t int_first)
{
    int64_t first = Int_to_Int64(int_first, false);
    if (first < 0)
        first = array.length + first + 1;

    if (first < 1 || first > array.length)
        return (array_t){.atomic=array.atomic};

    return (array_t){
        .atomic=array.atomic,
        .data=array.data + array.stride*(first-1),
        .length=array.length - first + 1,
        .stride=array.stride,
        .data_refcount=array.data_refcount,
    };
}

public array_t Array$to(array_t array, Int_t int_last)
{
    int64_t last = Int_to_Int64(int_last, false);
    if (last < 0)
        last = array.length + last + 1;

    if (last > array.length)
        last = array.length;

    if (last == 0)
        return (array_t){.atomic=array.atomic};

    return (array_t){
        .atomic=array.atomic,
        .data=array.data,
        .length=last,
        .stride=array.stride,
        .data_refcount=array.data_refcount,
    };
}

public array_t Array$by(array_t array, Int_t int_stride, int64_t padded_item_size)
{
    int64_t stride = Int_to_Int64(int_stride, false);
    // In the unlikely event that the stride value would be too large to fit in
    // a 15-bit integer, fall back to creating a copy of the array:
    if (__builtin_expect(array.stride*stride < ARRAY_MIN_STRIDE || array.stride*stride > ARRAY_MAX_STRIDE, 0)) {
        void *copy = NULL;
        int64_t len = (stride < 0 ? array.length / -stride : array.length / stride) + ((array.length % stride) != 0);
        if (len > 0) {
            copy = array.atomic ? GC_MALLOC_ATOMIC(len * padded_item_size) : GC_MALLOC(len * padded_item_size);
            void *start = (stride < 0 ? array.data + (array.stride * (array.length - 1)) : array.data);
            for (int64_t i = 0; i < len; i++)
                memcpy(copy + i*padded_item_size, start + array.stride*stride*i, padded_item_size);
        }
        return (array_t){
            .data=copy,
            .length=len,
            .stride=padded_item_size,
            .atomic=array.atomic,
        };
    }

    if (stride == 0)
        return (array_t){.atomic=array.atomic};

    return (array_t){
        .atomic=array.atomic,
        .data=(stride < 0 ? array.data + (array.stride * (array.length - 1)) : array.data),
        .length=(stride < 0 ? array.length / -stride : array.length / stride) + ((array.length % stride) != 0),
        .stride=array.stride * stride,
        .data_refcount=array.data_refcount,
    };
}

public array_t Array$reversed(array_t array, int64_t padded_item_size)
{
    // Just in case negating the stride gives a value that doesn't fit into a
    // 15-bit integer, fall back to Array$by()'s more general method of copying
    // the array. This should only happen if array.stride is MIN_STRIDE to
    // begin with (very unlikely).
    if (__builtin_expect(-array.stride < ARRAY_MIN_STRIDE || -array.stride > ARRAY_MAX_STRIDE, 0))
        return Array$by(array, I(-1), padded_item_size);

    array_t reversed = array;
    reversed.stride = -array.stride;
    reversed.data = array.data + (array.length-1)*array.stride;
    return reversed;
}

public array_t Array$concat(array_t x, array_t y, int64_t padded_item_size)
{
    void *data = x.atomic ? GC_MALLOC_ATOMIC(padded_item_size*(x.length + y.length)) : GC_MALLOC(padded_item_size*(x.length + y.length));
    if (x.stride == padded_item_size) {
        memcpy(data, x.data, padded_item_size*x.length);
    } else {
        for (int64_t i = 0; i < x.length; i++)
            memcpy(data + i*padded_item_size, x.data + i*padded_item_size, padded_item_size);
    }

    if (y.stride == padded_item_size) {
        memcpy(data + padded_item_size*x.length, y.data, padded_item_size*y.length);
    } else {
        for (int64_t i = 0; i < x.length; i++)
            memcpy(data + (x.length + i)*padded_item_size, y.data + i*padded_item_size, padded_item_size);
    }

    return (array_t){
        .data=data,
        .length=x.length + y.length,
        .stride=padded_item_size,
        .atomic=x.atomic,
    };
}

public bool Array$has(array_t array, void *item, const TypeInfo *type)
{
    const TypeInfo *item_type = type->ArrayInfo.item;
    for (int64_t i = 0; i < array.length; i++) {
        if (generic_equal(array.data + i*array.stride, item, item_type))
            return true;
    }
    return false;
}

public void Array$clear(array_t *array)
{
    *array = (array_t){.data=0, .length=0};
}

public int32_t Array$compare(const array_t *x, const array_t *y, const TypeInfo *type)
{
    // Early out for arrays with the same data, e.g. two copies of the same array:
    if (x->data == y->data && x->stride == y->stride)
        return (x->length > y->length) - (x->length < y->length);

    const TypeInfo *item = type->ArrayInfo.item;
    if (item->tag == PointerInfo || (item->tag == CustomInfo && item->CustomInfo.compare == NULL)) { // data comparison
        int64_t item_padded_size = type->ArrayInfo.item->size;
        if (type->ArrayInfo.item->align > 1 && item_padded_size % type->ArrayInfo.item->align)
            item_padded_size += type->ArrayInfo.item->align - (item_padded_size % type->ArrayInfo.item->align); // padding

        if ((int64_t)x->stride == item_padded_size && (int64_t)y->stride == item_padded_size && item->size == item_padded_size) {
            int32_t cmp = (int32_t)memcmp(x->data, y->data, MIN(x->length, y->length)*item_padded_size);
            if (cmp != 0) return cmp;
        } else {
            for (int32_t i = 0, len = MIN(x->length, y->length); i < len; i++) {
                int32_t cmp = (int32_t)memcmp(x->data+ x->stride*i, y->data + y->stride*i, item->size);
                if (cmp != 0) return cmp;
            }
        }
    } else {
        for (int32_t i = 0, len = MIN(x->length, y->length); i < len; i++) {
            int32_t cmp = generic_compare(x->data + x->stride*i, y->data + y->stride*i, item);
            if (cmp != 0) return cmp;
        }
    }
    return (x->length > y->length) - (x->length < y->length);
}

public bool Array$equal(const array_t *x, const array_t *y, const TypeInfo *type)
{
    return (Array$compare(x, y, type) == 0);
}

public CORD Array$as_text(const array_t *arr, bool colorize, const TypeInfo *type)
{
    if (!arr)
        return CORD_all("[", generic_as_text(NULL, false, type->ArrayInfo.item), "]");

    const TypeInfo *item_type = type->ArrayInfo.item;
    CORD c = "[";
    for (int64_t i = 0; i < arr->length; i++) {
        if (i > 0)
            c = CORD_cat(c, ", ");
        CORD item_cord = generic_as_text(arr->data + i*arr->stride, colorize, item_type);
        c = CORD_cat(c, item_cord);
    }
    c = CORD_cat(c, "]");
    return c;
}

public uint32_t Array$hash(const array_t *arr, const TypeInfo *type)
{
    // Array hash is calculated as a rolling, compacting hash of the length of the array, followed by
    // the hashes of its items (or the items themselves if they're small plain data)
    // In other words, it reads in a chunk of items or item hashes, then when it fills up the chunk,
    // hashes it down to a single item to start the next chunk. This repeats until the end, when it
    // hashes the last chunk down to a uint32_t.
    const TypeInfo *item = type->ArrayInfo.item;
    if (item->tag == PointerInfo || (item->tag == CustomInfo && item->CustomInfo.hash == NULL)) { // Raw data hash
        uint8_t hash_batch[4 + 8*item->size];
        memset(hash_batch, 0, sizeof(hash_batch));
        uint8_t *p = hash_batch, *end = hash_batch + sizeof(hash_batch);
        int64_t length = arr->length;
        *p = (uint32_t)length;
        p += sizeof(uint32_t);
        for (int64_t i = 0; i < arr->length; i++) {
            if (p >= end) {
                uint32_t chunk_hash;
                halfsiphash(&hash_batch, sizeof(hash_batch), TOMO_HASH_KEY, (uint8_t*)&chunk_hash, sizeof(chunk_hash));
                p = hash_batch;
                *(uint32_t*)p = chunk_hash;
                p += sizeof(uint32_t);
            }
            memcpy((p += item->size), arr->data + i*arr->stride, item->size);
        }
        uint32_t hash;
        halfsiphash(&hash_batch, ((int64_t)p) - ((int64_t)hash_batch), TOMO_HASH_KEY, (uint8_t*)&hash, sizeof(hash));
        return hash;
    } else {
        uint32_t hash_batch[16] = {(uint32_t)arr->length};
        uint32_t *p = &hash_batch[1], *end = hash_batch + sizeof(hash_batch)/sizeof(hash_batch[0]);
        for (int64_t i = 0; i < arr->length; i++) {
            if (p >= end) {
                uint64_t chunk_hash;
                halfsiphash(&hash_batch, sizeof(hash_batch), TOMO_HASH_KEY, (uint8_t*)&chunk_hash, sizeof(chunk_hash));
                p = hash_batch;
                *(p++) = chunk_hash;
            }
            *(p++) = generic_hash(arr->data + i*arr->stride, item);
        }
        uint32_t hash;
        halfsiphash(&hash_batch, ((int64_t)p) - ((int64_t)hash_batch), TOMO_HASH_KEY, (uint8_t*)&hash, sizeof(hash));
        return hash;
    }
}

static void siftdown(array_t *heap, int64_t startpos, int64_t pos, closure_t comparison, int64_t padded_item_size)
{
    assert(pos > 0 && pos < heap->length);
    char newitem[padded_item_size];
    memcpy(newitem, heap->data + heap->stride*pos, padded_item_size);
    while (pos > startpos) {
        int64_t parentpos = (pos - 1) >> 1;
        typedef int32_t (*cmp_fn_t)(void*, void*, void*);
        int32_t cmp = ((cmp_fn_t)comparison.fn)(newitem, heap->data + heap->stride*parentpos, comparison.userdata);
        if (cmp >= 0)
            break;

        memcpy(heap->data + heap->stride*pos, heap->data + heap->stride*parentpos, padded_item_size);
        pos = parentpos;
    }
    memcpy(heap->data + heap->stride*pos, newitem, padded_item_size);
}

static void siftup(array_t *heap, int64_t pos, closure_t comparison, int64_t padded_item_size)
{
    int64_t endpos = heap->length;
    int64_t startpos = pos;
    assert(pos < endpos);

    char old_top[padded_item_size];
    memcpy(old_top, heap->data + heap->stride*pos, padded_item_size);
    // Bubble up the smallest leaf node
    int64_t limit = endpos >> 1;
    while (pos < limit) {
        int64_t childpos = 2*pos + 1; // Smaller of the two child nodes
        if (childpos + 1 < endpos) {
            typedef int32_t (*cmp_fn_t)(void*, void*, void*);
            int32_t cmp = ((cmp_fn_t)comparison.fn)(
                heap->data + heap->stride*childpos,
                heap->data + heap->stride*(childpos + 1),
                comparison.userdata);
            childpos += (cmp >= 0);
        }

        // Move the child node up:
        memcpy(heap->data + heap->stride*pos, heap->data + heap->stride*childpos, padded_item_size);
        pos = childpos;
    }
    memcpy(heap->data + heap->stride*pos, old_top, padded_item_size);
    // Shift the node's parents down:
    siftdown(heap, startpos, pos, comparison, padded_item_size);
}

public void Array$heap_push(array_t *heap, const void *item, closure_t comparison, int64_t padded_item_size)
{
    Array$insert(heap, item, I(0), padded_item_size);

    if (heap->length > 1) {
        if (heap->data_refcount != 0)
            Array$compact(heap, padded_item_size);
        siftdown(heap, 0, heap->length-1, comparison, padded_item_size);
    }
}

public void Array$heap_pop(array_t *heap, closure_t comparison, int64_t padded_item_size)
{
    if (heap->length == 0)
        fail("Attempt to pop from an empty array");

    if (heap->length == 1) {
        *heap = (array_t){};
    } else if (heap->length == 2) {
        heap->data += heap->stride;
        --heap->length;
    } else {
        if (heap->data_refcount != 0)
            Array$compact(heap, padded_item_size);
        memcpy(heap->data, heap->data + heap->stride*(heap->length-1), padded_item_size);
        --heap->length;
        siftup(heap, 0, comparison, padded_item_size);
    }
}

public void Array$heapify(array_t *heap, closure_t comparison, int64_t padded_item_size)
{
    if (heap->data_refcount != 0)
        Array$compact(heap, padded_item_size);

    // It's necessary to bump the refcount because the user's comparison
    // function could do stuff that modifies the heap's data.
    ARRAY_INCREF(*heap);
    int64_t i, n = heap->length;
    for (i = (n >> 1) - 1 ; i >= 0 ; i--)
        siftup(heap, i, comparison, padded_item_size);
    ARRAY_DECREF(*heap);
}

public Int_t Array$binary_search(array_t array, void *target, closure_t comparison)
{
    typedef int32_t (*cmp_fn_t)(void*, void*, void*);
    int64_t lo = 0, hi = array.length-1;
    while (lo <= hi) {
        int64_t mid = (lo + hi) / 2;
        int32_t cmp = ((cmp_fn_t)comparison.fn)(
            array.data + array.stride*mid, target, comparison.userdata);
        if (cmp == 0)
            return I(mid+1);
        else if (cmp < 0)
            lo = mid + 1;
        else if (cmp > 0)
            hi = mid - 1;
    }
    return I(lo+1); // Return the index where the target would be inserted
}

// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0