This is part 2 of the article for creating custom SSMA report using SSIS and SSRS. In the previous post, we discussed how to use SSIS package to extract XML files containing SSMA assessment report information into SQL Server table. This week, we will discuss how to parse the XML data.
First, let's create tables to hold the parsed data:
CREATETABLE tblSSMAReport_Object (
Project VARCHAR(255),SessionVARCHAR(26),
ObjectID CHAR(36),PathVARCHAR(255),
Category Varchar(32), Name Varchar(32),ConvertedWithError TINYINT,
SourceStatement Varchar(max),
PRIMARYKEY (Project, Session, ObjectID)
)
CREATETABLE tblSSMAReport_MessageDetails (
Project VARCHAR(255),SessionVARCHAR(26),
ObjectID CHAR(36), Code CHAR(8), SourceCodeLink VARCHAR(800),Description VARCHAR(MAX),
PRIMARYKEY (Project, Session, ObjectID, Code)
)
tblSSMAReport_Object table contains the information about the database object and its overall status. tblSSMAReport_MessageDetails table stores conversion message .
We will need to parse text in order to extract information based on relative location. I created a function which we will use several times in this example:
CREATEFUNCTION [dbo].[UDF_ParsePath]
(
@string VARCHAR(800), @delimiter VARCHAR(16),@mode VARCHAR(6) = 'last',
@keyword VARCHAR(16) = null
)
RETURNSVARCHAR(100)
ASBEGINIF CHARINDEX(@delimiter, @string) > 0BEGINDECLARE @first_position INT, @last_position INTSET @first_position = CASEWHENLOWER(@mode) = 'last'THEN
LEN(REPLACE(@string,' ','|')) -
(CHARINDEX(REVERSE(ISNULL(@keyword,@delimiter)), REVERSE(@string)) +
LEN(REPLACE(ISNULL(@keyword,@delimiter),' ','|')) - 1) +
LEN(REPLACE(@delimiter,' ','|'))
WHENLOWER(@mode) = 'after'AND CHARINDEX(@keyword, @string) > 0 THEN
CASEWHEN CHARINDEX(@delimiter, @string, CHARINDEX(@keyword, @string) + LEN(REPLACE(@keyword,' ','|'))) = 0
THEN 0 ELSECHARINDEX(@delimiter, @string, CHARINDEX(@keyword, @string) +
LEN(REPLACE(@keyword,' ','|')))+ LEN(REPLACE(@delimiter,' ','|'))
ENDWHENLOWER(@mode) = 'before'AND CHARINDEX(@keyword, @string) > 0 THEN
LEN(REPLACE(@string,' ','|')) -
(CHARINDEX(REVERSE(@delimiter), REVERSE(@string),
CHARINDEX(REVERSE(@keyword), REVERSE(@string)) +
LEN(REPLACE(@keyword,' ','|')) +
LEN(REPLACE(@delimiter,' ','|')))
) + 2
ELSE 0ENDSET @last_position = CASE
WHEN CHARINDEX(@delimiter, @string, @first_position) = 0 THEN LEN(REPLACE(@string,' ','|')) + 1
ELSE CHARINDEX(@delimiter, @string, @first_position) ENDIF @first_position > 0 AND @last_position - @first_position > 0
RETURNSUBSTRING(@string, @first_position, @last_position - @first_position)
ENDRETURNNULL
ENDThe UDF returns segment of the string based on the following parameter value:
- Last: return an entire value of last segment (as identified by delimiter character) containing the keyword
- After: return the segment after the segment containing the keyword
- Before: return the segment before the segment containing the keyword
For example:
DECLARE @string VARCHAR(255) = 'C:\SSMA\Projects\ProjectName\report\report_2010_09_22T09_34_43\GUID\path.xml'
SELECTdbo.UDF_ParsePath(@string, '\','before','report\') as Project,
dbo.UDF_ParsePath(@string, '\','last','report') asSession,
dbo.UDF_ParsePath(@string, '\','before','path.xml') as ObjectID
results in the following:
| Project | Session | ObjectID |
| ProjectName | report_2010_09_22T09_34_43 | 0027b853-988c-4be4-b937-9859503d014d |
We can then use the user defined function to parse the XML and populate tblSSMAReport_Object:
INSERT INTO tblSSMAReport_Object (ObjectID, Project, Session, Path)
SELECT ObjectID, Project, SESSION,
MAX(CASEWHEN rn=6 THENpath + ' > 'ELSE''END)
+ MAX(CASEWHEN rn=5 THENpath + ' > 'ELSE''END)
+ MAX(CASEWHEN rn=4 THENpath + ' > 'ELSE''END)
+ MAX(CASEWHEN rn=3 THENpath + ' > 'ELSE''END)
+ MAX(CASEWHEN rn=2 THENpath + ' > 'ELSE''END)
+ MAX(CASEWHEN rn=1 THENpathELSE''END)
AsPATH
FROM(SELECT ROW_NUMBER() OVER (PARTITION BY FileName ORDERBY FileName ) as rn,
dbo.UDF_ParsePath(rpt.FileName, '\','before','report\') as Project,
dbo.UDF_ParsePath(rpt.FileName, '\','last','report') asSession,
dbo.UDF_ParsePath(rpt.FileName, '\','before','path.xml') as ObjectID,
col.value('./@name','VARCHAR(100)') AsPath
FROM tblRawData rptCROSS APPLY rpt.ObjectPath.nodes('//path/node') Tab(col) ) ASData
GROUPBY ObjectID, Project, Session
The inner statement parse object path and XML values while the outer statement perform string aggregation to form the object path across multiple rows. The tblSSMAReport_Object table needs to be updated with additional attribute such as object category, name and conversion status. The object category can be parsed from the object path. I use the following UDF to help with the parsing:
CREATEFUNCTION [dbo].[UDF_IdentifyObjectType]
(
@string VARCHAR(800) )
RETURNSVARCHAR(32)
ASBEGINDECLARE @ObjectType VARCHAR(32)
DECLARE @ObjectTypes TABLE(Seq INTIDENTITY(1,1), Name VARCHAR(32), MatchVARCHAR(32))
INSERT INTO @ObjectTypes (Name, Match)
VALUES ('packaged function', 'Packaged Functions'),('packaged type', 'Packaged Types'),('private packaged function', 'Private Packaged Functions'),('private packaged procedure', 'Private Packaged Procedures'),('private packaged type','Private Packaged Types'),('index','Indexes'),('trigger','Triggers'),('function','Functions'),('package','Packages'),('procedure','Procedures'),('sequence','Sequences'),('statement','Statements'),('synonym','Synonyms'),('table','Tables'),('user defined type','User Defined Types'),('view','Views')
SELECTTOP 1 @ObjectType= Name from @ObjectTypes
WHERE CHARINDEX(Match,@string) > 0
ORDERBY Seq
RETURN @ObjectTypeENDGOThe function locates the specified keyword from the path and returns the first value found based on the sequence specified in the function.
For example:
SELECT [dbo].[UDF_IdentifyObjectType]('Schemas > HR > Tables > LOCATIONS > Indexes > LOC_COUNTRY_IX')
returns 'Index' even though both 'Tables' and 'Indexes' keyword exists, but since index appears first in the sequence, the index keyword is returned. With UDF_IdentifyObjectType created, I can now update the object table and cross apply with the value from cat.xml values in the tblRawData to get the conversion status.
UPDATE obj SETobj.Category = dbo.UDF_IdentifyObjectType(Path),obj.Name = dbo.UDF_ParsePath(Path, ' > ','after',dbo.UDF_IdentifyObjectType(Path)),
obj.ConvertedWithError = CASEWHEN NotCvt.NotConvertedCount > 0 THEN 1 ELSE 0 END
FROM tblSSMAReport_Object obj LEFTJOIN tblRawData rawdata ON dbo.UDF_ParsePath(rawdata.FileName, '\','before','src.sql.txt') = obj.ObjectID
AND dbo.UDF_ParsePath(rawdata.FileName, '\','before','report\') = obj.Project
AND dbo.UDF_ParsePath(rawdata.FileName, '\','last','report') = obj.Session
LEFTJOIN (
SELECT dbo.UDF_ParsePath(rpt.FileName, '\','before','cat.xml') as ObjectID,
dbo.UDF_ParsePath(rpt.FileName, '\','before','report\') as Project,
dbo.UDF_ParsePath(rpt.FileName, '\','last','report') asSession,
col.value('./@category','VARCHAR(32)') As Category,
col.value('./@not-converted-count','INT') As NotConvertedCount
FROM tblRawData rptCROSS APPLY rpt.ObjectCategory.nodes('//statistics/statistic') Tab(col)) NotCvt ON NotCvt.ObjectID = obj.ObjectID
AND NotCvt.Session = obj.Sessionand NotCvt.Project = obj.Project
and dbo.UDF_IdentifyObjectType(Path) = NotCvt.Category
where obj.Category isnull
INSERT INTO tblSSMAReport_MessageDetails (Project, Session, ObjectID, Code, Description)
SELECTdbo.UDF_ParsePath(rpt.FileName, '\','before','report\') as Project,
dbo.UDF_ParsePath(rpt.FileName, '\','last','report') asSession,
dbo.UDF_ParsePath(rpt.FileName, '\','before','messages.xml') as ObjectID,
LEFT(col.value('./@name','VARCHAR(max)'),8) as Code,
col.value('./@name','VARCHAR(max)') As Description
FROM tblRawData rptCROSS APPLY rpt.ObjectMessage.nodes('//categories/category/node/record') Tab(col)
In the part 3 of this article, I will show an example of SSRS report which consume from the tables we just created.
Image may be NSFW.Clik here to view.