Can I have SQL incrementally count XML elements while parsing? - sql

Question

So this is my first foray into parsing XML, and I'm trying to figure out how to get this to work how I want.
Given the following XML format:
<Tiles>
<TileRow>
<TileValue>2</TileValue>
<TileValue>3</TileValue>
<TileValue>4</TileValue>
</TileRow>
<TileRow>
<TileValue>2</TileValue>
<TileValue>7</TileValue>
</TileRow>
</Tiles>
I want it to put it in a SQL table as the following:
| X | Y | Val |
|---|---|-----|
| 1 | 1 | 2 |
| 1 | 2 | 3 |
| 1 | 3 | 4 |
| 2 | 1 | 2 |
| 2 | 2 | 7 |
Basically, imagine a grid, and each "TileRow" starts a new Row in that grid. Each "TileValue" assigns the Column position in that grid, with the actual TileValue being what's in the 'cell' in that grid.
Is there a way to make SQL 'count' each time it passes over an element, or something to that effect?

Answer 1

Please try the following solution.
It is based on the Node Order Comparison operator in XQuery.
Node Order Comparison Operators
SQL
DECLARE #xml XML =
N'<Tiles>
<TileRow>
<TileValue>2</TileValue>
<TileValue>3</TileValue>
<TileValue>4</TileValue>
</TileRow>
<TileRow>
<TileValue>2</TileValue>
<TileValue>7</TileValue>
</TileRow>
</Tiles>';
SELECT c.value('for $i in . return count(/Tiles/TileRow[. << $i])', 'INT') AS [X]
, c.value('for $i in . return count(../*[. << $i]) + 1', 'INT') AS [Y]
, c.value('(./text())[1]', 'INT') as Value
FROM #xml.nodes('/Tiles/TileRow/TileValue') AS t(c);
Output
+---+---+-------+
| X | Y | Value |
+---+---+-------+
| 1 | 1 | 2 |
| 1 | 2 | 3 |
| 1 | 3 | 4 |
| 2 | 1 | 2 |
| 2 | 2 | 7 |
+---+---+-------+

Answer 2

Unfortunately, SQL Server does not support returning position() directly, it only allows it inside a predicate. If it did then you could simply query:
select
v.TileValue.value('position(parent::*)[1]','int'),
v.TileValue.value('./position()[1]','int'),
v.TileValue.value('text()[1]','int')
from #x.nodes('/Tiles/TileRow/TileValue') v(TileValue)
Instead, you can simulate it with ROW_NUMBER()
select
v1.rn,
row_number() over (partition by v1.rn order by (select 1)),
v2.TileValue.value('text()[1]','int')
from (
select v1.TileRow.query('.') TileRow,
row_number() over (order by (select 1)) rn
from #x.nodes('/Tiles/TileRow') v1(TileRow)
) v1
cross apply v1.TileRow.nodes('TileRow/TileValue') v2(TileValue)