SQL update statement to populate numerical series for each distinct subset of table records - sql

Question

I need the SQL update statement to assign consecutive sequence numbers to subsets of records in a table. I'm using MS access.
Let's say the current table has records like:
notebook,blue
notebook.Yellow
pencil,yellow
chair,blue
desk,green
desk,blue
I would like to add another field to the table and populate it as follows:
notebook,blue,1
notebook.Yellow,1
pencil,yellow,2
chair,blue,2
desk,green,1
desk,blue,3
you see that I have given a consecutive number assignment based on a certain set of criteria. In this example, the criteria was a distinct value in the second field (in real life, the criteria will be a distinct combination of values from several fields, but all the relevant fields are within the same table... no join is needed to get the criteria). since there are three records with blue in field 2, these are numbered 1,2,3. And since there are two records with yellow, they are numbered 1,2.
So I can't derive the numbering from the row number, since I have several numbering series in the same table all starting with 1.
Also, I need it to be a query where I don't have to explicitly specify the value in the second field. I just want each unique value in the second field to get its own numbering series. that is, I don't want to have to explicitly write one query to generate the numbers for "blue", and write a separate query to generate the numbers for "yellow"
The maximum number of records in the series is under 1000. So I don't mind if I would need to create and auxiliary table with 1000 records, with a field containing the values 1 to 1000. Then the update statement to the primary table could pull in the next value from the auxiliary table.
But I don't know the SQL syntax to use for this update statement, or for the update statement for any other approach. So I need your advice.

Answer 1

I'm not sure how to do this with a single SQL statement, but here are 2 SQL statements that could be used to handle each case:
insert into table ('desk', 'blue', 1)
where not exists (select field3 from table where field1 = 'desk' and field2 = 'blue');
insert into table (field1, field2, field3)
select field1, field2, count(1) + 1
from table
where field1 = 'desk'
and field2 = 'blue'
group by field1, field2;

Answer 2

Create Table #TableAutoIncrement (ID int identity(1000 , 1) , item varchar(20), COLOR varchar(20) )
Insert INTO #TableAutoIncrement
(item, COLOR )
SELECT item, COLOR FROM YOURTABLE
--- GETTING all the values from the temporary table
SELECT * FROM #TableAutoIncrement

Answer 3

A colleague of mine worked out the necessary SQL. Here's the generalized solution (note that I really needed to number the multiple series in my data set based on a combination of two fields. In my simplified example in the original post, I was using only one field--color--but since I really need two fields, that's what I show in this solution.
SELECT *,
(SELECT COUNT(T1.ID)
FROM
[TableName] AS T1
WHERE T1.ID >= T2.ID and t1.[NameCriteriaField1] = t2.[NameCriteriaField1]
and t1.[NameCriteriaField2]= t2.[NameCriteriaField2])
AS Sequence into OutputResultsTableName
FROM
[TableName] AS T2
ORDER BY [NameCriteriaField2] , [NameCriteriaField1]
The source table is set up with "ID" as field with an integer value. Every record has a unique value of ID, but it does not matter if there are gaps in the ID or how the records are sequenced against the ID. (e.g., the typical MS access auto numbered primary key field serves this purpose)
This query is set up to assume that there are two fields in your data set that you want to use to group your records and assign a numerical series count to each record within each group. (Thus your table may contain multiple groups, and each group has its own numbering series starting with 1. But the way the query is formulated, there are exactly 2 criteria that define the group.) You cannot use any where clauses to further filter the records that get counted. Through experimentation, I found that adding where clauses gives unreliable results where records can get omitted. So if you need the results to be filtered so that some records are not to be included in the numerical series for a particular group, then do one of the following before running my query:
run a query to delete the undesired records from the source table
first copy all records from the source table into a new table and delete the records from the new table that should not be numbered, and run my query on the new table
deleting extraneous records before running this query is needed only if those records qualify as members of a group defined by criteria 1 and criteria 2. If there are extraneous records that don't match those two criteria, you can leave them in the table, because they will not impact the numbering of the records within the groups that you care about. They will just get their own independent numbering, which you can just ignore.**
The numbering of each group starts at 1, and the query dynamically defines the groups based on the distinct combinations of criteria1 and criteria2. However, if you have records that do not belong to any group, these records will all be numbered with 0. (Criteria1 and criteria2--at least to the extent of my testing--are non-null values. (In theory--at least on Microsoft Access, an empty string is different than Null, but I did not test this with empty strings either.) If you have records that have null in the criteria1 or criteria2 fields, MS Access consider these records as not belonging to any group and thus numbers them with 0. That is, these distinct groups need to define by non-null values for criteria1 and criteria2, and thus this is different than the way SQL DISTINCT statement works.
If you need to have NULL as a valid criteria for defining the group (and thus to have groups defined by NULL numbered), it's very simple. Prior to running my query, first run an update statement that changes all instances of null values in criteria1 or criteria2 to the phrase "placeholder for null field". Then run my query. On the result set (after the numbering has been assigned to the groups), run another update to change all occurrences of the placeholder phrase back to null.
Adjustment to syntax if your group is defined by only one field criteria
SELECT *,
(SELECT COUNT(T1.ID)
FROM
[TableName] AS T1
WHERE T1.ID >= T2.ID and t1.[NameCriteriaField1] = t2.[NameCriteriaField1] )
AS Sequence into OutputResultsTableName
FROM
[TableName] AS T2
ORDER BY [NameCriteriaField2] , [NameCriteriaField1]
Adjustment to syntax if your group is defined by combination of 3 field criteria
SELECT *,
(SELECT COUNT(T1.ID)
FROM
[TableName] AS T1
WHERE T1.ID >= T2.ID and t1.[NameCriteriaField1] = t2.[NameCriteriaField1]
and t1.[NameCriteriaField2]= t2.[NameCriteriaField2]
and t1.[NameCriteriaField3]= t2.[NameCriteriaField3])
AS Sequence into OutputResultsTableName
FROM
[TableName] AS T2
ORDER BY [NameCriteriaField2] , [NameCriteriaField1]