组合/合并日期范围的算法

Question

我正在尝试找到如何将日期范围合并到一个数据库记录（数组元素）中的最佳方法。

这是我拥有的数据：

  Array
(
    [0] => Array
        (
            [id] => 18298
            [start_date] => 2011-07-09
            [end_date] => 2011-10-01
        )

    [1] => Array
        (
            [id] => 18297
            [start_date] => 2011-06-01
            [end_date] => 2011-06-30
        )

    [2] => Array
        (
            [id] => 17113
            [start_date] => 2011-03-31
            [end_date] => 2011-05-31
        )

    [3] => Array
        (
            [id] => 20555
            [start_date] => 2011-01-03
            [end_date] => 2011-03-31
        )
)

在我们组合它们之后，数组（或数据库）应该是这样的：

Array
(
    [0] => Array
        (
            [merged_ids] => 18298
            [start_date] => 2011-07-09
            [end_date] => 2011-10-01
        )

    [1] => Array
        (
            [merged_ids] => 18297, 17113, 20555
            [start_date] => 2011-01-03
            [end_date] => 2011-06-30
        )
)

是否有任何算法可以遍历所有元素/范围并将它们组合起来？哪种方式更好/更容易 - 通过数据库 (MYSQL) 或编码 (PHP)？

非常感谢任何建议。

谢谢！

更新：抱歉，我没有提供足够的信息：我们应该合并任何连续和重叠的日期范围。

Answer 1

按开始日期排序。

然后遍历并检查下一项的开始日期是在当前结束日期之前还是之后。如果是，则将下一个合并到当前的。然后继续。

Answer 2

我编写了组合/合并范围列表的函数。它是用 Python 编写的，但用 PHP 重写应该很容易。这是完整的代码：https://gist.github.com/barszczmm/8447665，这是简化的算法（仍在 Python 中）：

list_of_ranges.sort() # sort input list
new_list_of_ranges = [] # output list

new_range_item_start = None
new_range_item_end = None

length = len(list_of_ranges)
for i, range_item in enumerate(list_of_ranges):
    if new_range_item_start is None:
        new_range_item_start = range_item[0]
        new_range_item_end = range_item[1]
    elif new_range_item_end >= range_item[0]:
        new_range_item_end = max(range_item[1], new_range_item_end)
    else:
        new_list_of_ranges.append((new_range_item_start, new_range_item_end))
        new_range_item_start = range_item[0]
        new_range_item_end = range_item[1]
    # save if this is last item
    if i + 1 == length:
        new_list_of_ranges.append((new_range_item_start, new_range_item_end))

Answer 3

实现是这样的：

function mergeDateTimeRanges($ranges)
{
    $retVal = [];
    //sort date ranges by begin time
    usort($ranges, function ($a, $b) {
        return strcmp($a['begin_at'], $b['begin_at']);
    });

    $currentRange = [];
    foreach ($ranges as $range) {
        // bypass invalid value
        if ($range['begin_at'] >= $range['end_at']) {
            continue;
        }
        //fill in the first element
        if (empty($currentRange)) {
            $currentRange = $range;
            continue;
        }

        if ($currentRange['end_at'] < $range['begin_at']) {
            $retVal[] = $currentRange;
            $currentRange = $range;
        } elseif ($currentRange['end_at'] < $range['end_at']) {
            $currentRange ['end_at'] = $range['end_at'];
        }
    }

    if ($currentRange) {
        $retVal[] = $currentRange;
    }

    return $retVal;
}

Answer 4

我的方法将生成一个合并数组，其中重叠或连续的日期被组合在一起，组的 id 被存储为逗号分隔值的字符串。

我正在使用现代“宇宙飞船操作员”（<=>）进行usort()s 比较。如果你的代码在php7以下的版本上运行，你可以使用：

usort($array,function($a,$b){ return strcmp($a['start_date'],$b['start_date']); });

请参阅内联 cmets 了解分步说明。

代码：(Demo)

function mergeRanges($array){
    usort($array,function($a,$b){ return $a['start_date']<=>$b['start_date']; });  // order by start_date ASC

    foreach($array as $i=>$row){
         if($i && $row['start_date']<=date('Y-m-d',strtotime("{$result[$x]['end_date']} +1 day"))){  // not the first iteration and dates are within current group's range
            if($row['end_date']>$result[$x]['end_date']){  // only if current end_date is greater than existing end_date
                $result[$x]['end_date']=$row['end_date'];  // overwrite end_date with new end_date in group
            }
            $result[$x]['merged_ids'][]=$row['id'];  // append id to merged_ids subarray
        }else{  // first iteration or out of range; start new group
            if($i){  // if not first iteration
                $result[$x]['merged_ids']=implode(', ',$result[$x]['merged_ids']);  // convert previous group's id elements to csv string
            }else{  // first iteration
                $x=-1;  // declare $x as -1 so that it becomes 0 when incremented with ++$x
            }
            $result[++$x]=['merged_ids'=>[$row['id']],'start_date'=>$row['start_date'],'end_date'=>$row['end_date']]; // declare new group
        }
    }
    $result[$x]['merged_ids']=implode(', ',$result[$x]['merged_ids']);  // convert final merged_ids subarray to csv string
    return $result;
}

$array=[
    ['id'=>18298,'start_date'=>'2011-07-09','end_date'=>'2011-10-01'],
    ['id'=>18297,'start_date'=>'2011-06-01','end_date'=>'2011-06-30'],
    ['id'=>17113,'start_date'=>'2011-03-31','end_date'=>'2011-05-31'],  // tests that 17113 and 18297 belong in same group
    ['id'=>20556,'start_date'=>'2011-02-03','end_date'=>'2011-02-13'],  // tests that "fully overlapped" date range is included
    ['id'=>20555,'start_date'=>'2011-01-03','end_date'=>'2011-03-31']
];

print_r(mergeRanges($array));

输出：

Array
(
    [0] => Array
        (
            [merged_ids] => 20555, 20556, 17113, 18297
            [start_date] => 2011-01-03
            [end_date] => 2011-06-30
        )

    [1] => Array
        (
            [merged_ids] => 18298
            [start_date] => 2011-07-09
            [end_date] => 2011-10-01
        )

)