SQL Porosity - CSV Element Calculation

Background:
Recently "walked" into a scenario where a client was storing an variable sized CSV string in a field.  This field held data that would eventually be evaluated as a "pseudo" entity for reporting purposes.

Problem
Due to the need for performance and reliability, the normal practice of splitting the string into a result set and then performing a SQL Count on the result set, would be overly complex and may cause performance issues as the length is variable (i.e. varchar(max) )

Solution
After reviewing several search queries, which most ended up at my old faithful Stack Overflow.  I came across an approach that is nothing more than Porosity.  For my purpose, it fit the bill very cleanly without overly transitioning a single value into a data-set and then calculating the data-set.

Net Result
Data-Set:
Source:
SQL Porosity - Source Array
SQL Data - Source Value
SQL Porosity - Resulting Array
SQL Data - Comma's Voided
Math
n - Len([Field])
m - Len(Replace([Field], ',', ''))
Number of Elements: n - m + 1

Explanation:
First we need to know the source length (with comma's); next, we remove the comma's with empty space; next, we find the length of the string without the comma's; lastly, we subtract the "voided" length from the "source" length, to find the number of comma's that were in the string.  Then we add 1 to the difference to give visibility to the number of value elements within the CSV value.

Comments

Post a Comment

Popular posts from this blog

SysInternals - BgInfo for ALL Users

Image
I have seen various sites that have been pretty good at getting close to the point of getting Background Information created by the tool BgInfo but none of them have actually solved my specific problem. Solutions currently out there include: Placing into the "All Users" folder This did not work for me, as our implementation does not contain an "All Users" folder in the "Users" folder for profiles. Reg-Hacking and HKCU or HKLM entry This is the worst of all the bad choices available.  I am experienced with Registry Editing and I still would not want or find a need to do this. Placing into the "Default" folder This is good going forward but what about those 20+ user profiles that are already logged in.  FYI, if your server has 20+ user account profiles in the "Users" directory you have larger problems than I am willing to assist with in this posting. The solution that worked for me, will be described below in pain-staki...
Read more

Linked Server....

Well today, I started the fun filled adventure of managing data from/to a LinkedServer. During this learning curve, I found plenty of information about how to query from and instert into a Linked server but most of them utilized the OpenQuery  TSQL Statement to operate. One failure of the concepts is to use the inline statement to access the Linked Server. In my situation, i was linking to a MySQL database offsite.  Below you will find the MSSQL statement patterns to used when utilizing query statements against the Linked Server. Select: 1: Select {cols} 2: From {LinkedServer}...{table} 3: [Conditions] Insert: 1: Insert Into {LinkedServer}...{table} [cols,...] 2: Values ([cols,...]) There are more commands you can use but the point is that the developer/dba is utilizing only the 'server' and 'table' portions of the source information.  This being, as described by Microsoft, as ' {server}.{database}.{schema}.{table} ' is simply ' ...
Read more

TSQL - String Concatenation

On current project, I ran into a scenario where I needed to concatenate a series of rows into a single string value.  Nothing major, just concatenate them with a delimiter and return the result.  As all of us Microsoft SQL developer's know and deal with is the lack of a SQL supported function that is able to assist us in this "minor" feature. As this is my knowledge dump for techniques and other methods used for the various platforms, here we go: Construct: The first step is to identify what information you want, the basic's of query design.   SELECT Email FROM Contacts Group By Email Group by is conditional, depending on your situation and how you organize your referential integrity. Stage Next, lets build our environment, not the SProc just the supporting actors in the process. declare @email varchar(max) = '' SELECT @email += Email + ', ' FROM Contacts Group By Email This simply concatenates each of ...
Read more