将nvarchar转换为int错误

Question

I have this query that I can't quite work out the error I'm getting.

Conversion failed when converting the nvarchar value '1.5' to data type int.

This is due to pvlaue for one case pulling a '1.5' string and trying to convert it to an int implicitly. I tried to rectify this by casting to a float but this doesn't seem to work for me. Any help in this regard would be very helpful.

I'm running this on SQL Server 2012 Management Studio.

BEGIN DECLARE @Item NVARCHAR(100)
SET @Item = 'Water'

SELECT 
    d.DESCRIPTION, d.ITEM_CODE,  
    el.ENUM_LABEL as Building_Block, 
    CAST(itpC.pvalue AS DECIMAL(22, 3)), ffi.where_used,
    d.STATUS_IND
FROM 
    FSITEM d
LEFT JOIN 
    FSITEMTECHPARAM itp ON d.ITEM_CODE = itp.ITEM_CODE 
                        AND itp.PARAM_CODE = 'BUILDING_BLOCK'
LEFT JOIN 
    FSVALIDENUMVALCF ev ON CAST(coalesce(itp.PVALUE, 0) as float) = CAST(ev.ENUM_VALUE AS FLOAT)
                        AND ev.ENUM_CODE = 'C_BUILDING_BLOCK_TYP'
LEFT JOIN 
    FSVALIDENUMLABELCF el ON ev.ENUM_CODE = el.ENUM_CODE 
                          AND ev.ENUM_ORDER = el.ENUM_ORDER 
                          AND el.LANGUAGE_CODE = 'EN-US'
LEFT JOIN
    (SELECT 
         item_Code, max(CAST(pvalue AS FLOAT)) as pvalue
     FROM 
         FSITEMTECHPARAM
     JOIN
         KC_SITE_COST_MAP ON FSITEMTECHPARAM.PARAM_CODE = 'COST_'+KC_SITE_COST_MAP.COST_CODE
     GROUP BY 
         ITEM_CODE

     UNION

     SELECT
         ITEM_CODE, CAST(pvalue AS FLOAT) as pvalue
     FROM  
         FSFORMULA
     JOIN
         FSFORMULATECHPARAM ON FSFORMULA.FORMULA_ID = FSFORMULATECHPARAM.FORMULA_ID
                            AND PARAM_CODE = 'COST_TOTAL_BASE'
     WHERE 
         FSFORMULA.FORMULA_ID IN (SELECT FORMULA_ID FROM FSITEM)) itpC ON D.item_code = itpC.item_code
LEFT JOIN 
    (SELECT
         ffi.item_code, COUNT(distinct ffi.formula_id) where_used 
     FROM
         fsformulaingred ffi 
     JOIN
         fsformula ff ON ffi.formula_id = ff.formula_id 
     WHERE
         ff.status_ind = 500 AND ff.logical_delete = 0  
     GROUP BY
         ffi.item_code) ffi ON D.item_code = ffi.item_code 
WHERE 
    d.LOGICAL_DELETE = 0 
    AND d.COMPONENT_IND <> 2 
    AND d.Status_IND < 600
    AND (((CAST(ev.ENUM_VALUE as FLOAT) = 3 or CAST(ev.ENUM_VALUE as FLOAT) = 2)  
    AND d.Status_Ind < 600) OR
       CAST(ev.ENUM_VALUE as FLOAT) = 1 or CAST(ev.ENUM_VALUE as FLOAT) = '0') 
    AND (d.Description Like @Item + '%')
    --Or d.Description Like replace(@Item, '!', '%'))
ORDER BY 
    CAST(coalesce(ev.enum_value,0) as FLOAT) DESC, status_ind DESC, d.Item_Code
END

Answer 1

It's the coalesce that is doing the conversion to int , so casting the result from the coalesce is too late.

The result of an expression has to have the same type regardless of how the expression is evaluated. The expression coalesce(itp.PVALUE, 0) will be of the type int regardless of whether the value itp.PVALUE or the value 0 is used, because the type of 0 is int .

If you cast the 0 to float , the type of the expression will be float , so use coalesce(itp.PVALUE, cast(0 as float)) .