"Filling in the gaps" with a single-line query

Hi,
I've got the following scenario:
Files are being stored in a database, with a number of name-value
pairs associated with each file. This happens by storing the files in
one table (File), the list of property names in another table
(FileMetaDataSchema) and the values of properties in a third table
(FileMetaData), which references both File and FileMetaDataSchema.
The frontend of my application assumes there is a record for each
property of each file in the FileMetaData table, even if the value is
an empty string. In other words, if i have a list of 3 properties and
2 files, FileMetaData will contain 6 records.
Due to a bug in the system, this does not always happen. Suppose one
adds a new property and neglects to insert the "blank" records for the
new properties for all the files, or a new file is added, but the
associated meta-data records are not... (why and how this happens is
not the topic of discussion, so don't worry about that).
I have written a sql script to insert all the missing "blank" records,
but I feel it is very clumsy and intuitively, I just know there must
be a simpler way, my knowledge is just too limited. What it does is,
it iterates (using cursors) through all the files and all the
properties, checks if there is a record for each combination FileID
and FileMetaDataSchemaID and if not, it inserts one. I am looking for
a better way out of curiosity, for my own benefit.
Here's the script and thanks for any input:
DECLARE MetadataSchemaCursor CURSOR FOR
SELECT FileMetaDataSchemaID FROM FileMetaDataSchema
DECLARE @.FileMetaDataSchemaID INT,
@.FileID INT
OPEN MetadataSchemaCursor
FETCH NEXT FROM MetaDataSchemaCursor INTO @.FileMetaDataSchemaID
WHILE (@.@.FETCH_STATUS = 0)
BEGIN
DECLARE FileCursor CURSOR FOR
SELECT FileID FROM [File]
OPEN FileCursor
FETCH NEXT FROM FileCursor INTO @.FileID
WHILE (@.@.FETCH_STATUS = 0)
BEGIN
IF NOT EXISTS (SELECT 1 FROM FileMetaData WHERE FileID = @.FileID AND
FileMetaDataSchemaID = @.FileMetaDataSchemaID)
INSERT INTO FileMetaData (FileID, FileMetaDataSchemaID,
PropertyValue) SELECT @.FileID, @.FileMetaDataSchemaID, ''
FETCH NEXT FROM FileCursor INTO @.FileID
END
CLOSE FileCursor
DEALLOCATE FileCursor
FETCH NEXT FROM MetaDataSchemaCursor INTO @.FileMetaDataSchemaID
END
CLOSE MetadataSchemaCursor
DEALLOCATE MetadataSchemaCursor>I just know there must
>be a simpler way, my knowledge is just too limited.
You are correct, there is a simpler way.
INSERT INTO FileMetaData (FileID, FileMetaDataSchemaID, PropertyValue)
SELECT A.FileID, B.FileMetaDataSchemaID, ''
FROM [File] as A
CROSS
JOIN FileMetaDataSchema as B
WHERE NOT EXISTS
(select * from FileMetaData as X
where A.FileID = X.FileID
and B.FileMetaDataSchemaID = X.FileMetaDataSchemaID)
Roy Harvey
Beacon Falls, CT
On 22 Feb 2007 06:48:53 -0800, "Velislav" <vgebrev@.gmail.com> wrote:

>Hi,
>I've got the following scenario:
>Files are being stored in a database, with a number of name-value
>pairs associated with each file. This happens by storing the files in
>one table (File), the list of property names in another table
>(FileMetaDataSchema) and the values of properties in a third table
>(FileMetaData), which references both File and FileMetaDataSchema.
>The frontend of my application assumes there is a record for each
>property of each file in the FileMetaData table, even if the value is
>an empty string. In other words, if i have a list of 3 properties and
>2 files, FileMetaData will contain 6 records.
>Due to a bug in the system, this does not always happen. Suppose one
>adds a new property and neglects to insert the "blank" records for the
>new properties for all the files, or a new file is added, but the
>associated meta-data records are not... (why and how this happens is
>not the topic of discussion, so don't worry about that).
>I have written a sql script to insert all the missing "blank" records,
>but I feel it is very clumsy and intuitively, I just know there must
>be a simpler way, my knowledge is just too limited. What it does is,
>it iterates (using cursors) through all the files and all the
>properties, checks if there is a record for each combination FileID
>and FileMetaDataSchemaID and if not, it inserts one. I am looking for
>a better way out of curiosity, for my own benefit.
>Here's the script and thanks for any input:
>DECLARE MetadataSchemaCursor CURSOR FOR
> SELECT FileMetaDataSchemaID FROM FileMetaDataSchema
>DECLARE @.FileMetaDataSchemaID INT,
> @.FileID INT
>OPEN MetadataSchemaCursor
>FETCH NEXT FROM MetaDataSchemaCursor INTO @.FileMetaDataSchemaID
>WHILE (@.@.FETCH_STATUS = 0)
>BEGIN
> DECLARE FileCursor CURSOR FOR
> SELECT FileID FROM [File]
> OPEN FileCursor
> FETCH NEXT FROM FileCursor INTO @.FileID
> WHILE (@.@.FETCH_STATUS = 0)
> BEGIN
> IF NOT EXISTS (SELECT 1 FROM FileMetaData WHERE FileID = @.FileID AND
>FileMetaDataSchemaID = @.FileMetaDataSchemaID)
> INSERT INTO FileMetaData (FileID, FileMetaDataSchemaID,
>PropertyValue) SELECT @.FileID, @.FileMetaDataSchemaID, ''
> FETCH NEXT FROM FileCursor INTO @.FileID
> END
> CLOSE FileCursor
> DEALLOCATE FileCursor
>FETCH NEXT FROM MetaDataSchemaCursor INTO @.FileMetaDataSchemaID
>END
>CLOSE MetadataSchemaCursor
>DEALLOCATE MetadataSchemaCursor|||On Feb 22, 5:42 pm, Roy Harvey <roy_har...@.snet.net> wrote:
> You are correct, there is a simpler way.
> INSERT INTO FileMetaData (FileID, FileMetaDataSchemaID, PropertyValue)
> SELECT A.FileID, B.FileMetaDataSchemaID, ''
> FROM [File] as A
> CROSS
> JOIN FileMetaDataSchema as B
> WHERE NOT EXISTS
> (select * from FileMetaData as X
> where A.FileID = X.FileID
> and B.FileMetaDataSchemaID = X.FileMetaDataSchemaID)
> Roy Harvey
> Beacon Falls, CT
>
Thank you
Note to self - look up cross joins.