Custom Transform Component, Change Type Or Add Output Column
Jun 26, 2006
Would anyone happen to have any pointers or know of any good code examples to either programmatically change the type of an input column when it is passed through the component, or add a new column to the output? I am extracting data from an Oracle database which is in Julian date format (represented within SSIS as a DT_NUMERIC column) and I need to to either transform the input column holding it into a date column, or to dynamically add a new output column holding the transformed data.
I'm writing a custom source component that reads data from a SharePoint list with dynamic mapping to output columns. It's my first custom component and it's based on several samples and tutorials from Internet
Output columns are not created by the component itself, they must be added by user at design time. The component makes dynamically an association between SharePoint fields and available output columns at run-time (based on an mapping table).
I made a very basic skeleton and I encounter a problem when I add a column to output: it has no datatype and when I try to set one I have an the error Property value is not valid, The component xxxxxx does not allow setting output column datatype properties.
Imports System Imports Microsoft.SqlServer.Dts.Pipeline Imports Microsoft.SqlServer.Dts.Pipeline.Wrapper Imports Microsoft.SqlServer.Dts.Runtime.Wrapper <DtsPipelineComponent(ComponentType:=ComponentType.SourceAdapter, DisplayName:="SharePoint Dynamic Assoc List Source",
Hi all, I saw a couple of other posts here on this topic, but none quite got to my issue. I'm trying to add error output to a custom source component (not a script, a custom component). The samples all seem to deal with transform components, and my issues seem to be unique to source components.
I have the following code related to error handling ...
Public Overloads Overrides Sub ProvideComponentProperties()
...
Dim output As IDTSOutput90 = ComponentMetaData.OutputCollection.New
Dim errorOutput As IDTSOutput90 = ComponentMetaData.OutputCollection(ERRORCOLUMNNAME)
Dim errorColumn As IDTSOutputColumn90 = errorOutput.OutputColumnCollection.New
errorColumn.Name = outColumn.Name
errorColumn.SetDataTypeProperties(DataType.DT_I4, 0, 0, 0, 0) ... End Sub
The confusions I have are: a) the stock advanced properties editor (I haven't provided a custom one) doesn't seem to "realize" that I have an error output, so provides no method to configure. I am believing it would need to know which Output columns can have their error/truncation redirected. I'd have thought setting ErrorRowDisposition on my output column would have been enough to trigger this ?? b) since I don't have any means of configuring, not surprisingly, when I try to connect my error output, designer complains that, "Ths error output cannot receive any error rows. This occurs for several reasons: Input columns or output columns are not yet defined. Error handling is not supported by the component. Error handling is not configured for the component." c) UsesDispoistions would seem to be appropriate only for a transform component
Thanks for reading, and appreciate any help or pointers. Bill
I created a SSIS custom component, transformation (Asynchronous) with one Input collection and 2 output collections.
The SSIS Package which includes the Component I created works well in the Business Intelligence Studio, but when the same Package is run in the 'Execute Package utility' It fails to run. ( when you Double click on the dtsx file)
The cause of the failiure is
public override void PrimeOutput(int outputs, int[] outputIDs, PipelineBuffer[] buffers)
method receives only one output buffer when executed using the 'Execute Package Utility' { outputs = 1 , buffer.Length = 1 } ( when executed in the BI studio, the method receives parameters of both the output buffers that I expect { outputs = 2 , buffer.Length = 2 } )
The property ComponentMetaData.OutputCollection.Count = 2 as well. Yet the PrimeOutput method provides only 1 buffer.
The Validation Succeeds on both instances, which I assume means that Meta Data is Provided Properly.
What would be the reason for the same pakage to run in 2 different ways like this,
What might I have missed out to do, to make the package run in different ways on 'Business Intelligence Studio' and 'Execute Package Utility'
Thanks a lot
Below are some of the lines from the ProvideComonentProperties Method which deals with the output Collection, Isn't this sufficient for the PrimeOutput to provide 2 output buffers?
ProvideComponentProperties()
public override void ProvideComponentProperties() {
I'm having a tad bit of trouble getting output from an asynchronous component that I've written and am looking for some insight.
This component takes in a name string passed from upstream and parses the name components into standardized output fields. I'm using an asynchronous component because if the name string contains two names ("Fred & Wilma Flintstone") I'm outputting one row for Fred and one for Wilma. I've gotten it to run and with debugging have observed what appeared to me to be proper execution, but zero rows are flowing out of it.
In my ProvideComponentProperties method, I add the three fields and there associated metadata to the OutputColumnCollection. Is this method where this should occur? It's before the PrimeOutput method, so I didn't know if I should be creating the output columns in ProcessInput (i.e., after the output buffer is provided by PrimeOutput.)
In ProcessInput, I'm using AddRow for each input row and another if it contains a second name, setting the value for each index using the buffer's SetString method, to no avail. I can observe it to this point, but then don't know what's in that output buffer (if I'm using the wrong buffer index value, etc)
I'm trying to create a fairly simple custom transform component (because I've read that's the easiest one to create) which will take one column from a flat file source and based on the first row create the output columns. I'm actually trying to write a component that will solve the now well known problem with parsing CSV files in SSIS. I have a lot of source files and all have many columns so a component that can read in the first line from the CSV file and create the output columns automatically will save me lots of time when migrating the old DTS packages.
I have the basic component set up but I'm stuck when trying to override the OnInputPathAttached method because I don't know how to use the inputID to get the first line from the input (the buffer). Are there any good examples for creating output columns dynamically based on the input buffer? Should I just give up on on the transform and create a custom source component instead?
When data is imported from our legacy system, the same functions need to be applied to several columns on different tables. I want to build a kind of "Function Library", so that the functions I define can be re-used for columns in several packages.
The "Derived Column" transform seems ideal, if only I could add my list of user-defined functions to it. Basically I want to inherit from it, and add my own list of functions for the users to select.
Is this possible ?
What other approaches could I take to building about 30 re-usable functions?
I am creating a customer data flow component for SSIS for use in a package. I've got some custom properties that I am exposing using the supplied advanced editor (no custom property editor here).
Some of my properties are enumerated types, and I have deciphered how to get those properties to show as dropdown lists of their respective enumerations. (For those of you who may be looking as hard as I did as to how to accomplish this, see the end of this post.)
I also have a few properties which request SSIS package variable names - such as an file name variable. However, I can't figure out how to tell the advanced editor that the property is looking for an SSIS variable, so that it can show a dropdown list of package variables, much like virtually any other Microsoft supplied Data Flow component can.
Is there a Type Converter I could specify for those custom properties? Is there another way to instruct SSIS that my custom property is expecting a variable? Or do I need to code a custom UI for editing my Data Flow Task?
To create a dropdown list of values for a custom property that represents an enum, do the following:
1. Create your enum definition, such as "public enum ThisIsMyEnum { one, two }"
2. Create a new class that inherits from TypeConverter, such as "public class MyEnumConverter : TypeConverter"
3. Override "CanConvertFrom", and return true if "sourceType == typeof(string)"
4. Override "CanConvertTo", and return true if "destinationType == typeof(string)"
5. Override "ConvertFrom", and return the enum value (such as "one" or "two" in my example) that corresponds to the string passed in the parameter "value"
6. Override "ConvertTo", and return a string that corresponds to the enum value passed in the parameter "value"
7. Override "GetStandardValuesSupported" and return true
8. Override "GetStandarValuesExclusive" and return true to indicate that ONLY the enum values should be accepted
9. Override "GetStandardValues", and return a new StandardValuesCollection constructed with Enum.GetValues() of your enum, such as "return new StandardValuesCollection(Enum.GetValues(typeof(ThisIsMyEnum)));"
10. Just above your "public enum" declaration, add a "TypeConverter" attribute to link your type converter to your enum, such as "[TypeConverter(typeof(MyEnumConverter))]"
11. In "ProvideComponentProperties", after you've created your custom property like this: "IDTSCustomProperty90 propEnum = ComponentMetaData.CustomPropertyCollection.New()", add another line to specify the TypeConverter property of the property to the full assembly name of your type converter, like so: "propEnum.TypeConverter = typeof(MyEnumConverter).AssemblyQualifiedName;"
Wanted to enquire how this is done. Tried doing it in the setUsageType method I have overwritten but only allows description to be changed. Basically need to change "Name".
Best option would be to change it instantly when a user selects a column from the inputs in the custom component, ie. it changes the Output Alias to a desired value. (Input tab in advanced editor)
All this is being done in a custom component which I would like to be synchronous, can achieve a similar result asynchronously.
Does anyone know how to get destination coulmns to show up in the advanced editor for a custom component? I have a custom flat file destination component that builds the output based on a specific layout. It works as long as the upstream column names match my output names. What I want is to allow non-matching columns to be mapped by the user as they can in a stock flat file destination. The closest that I have been able to come is to get the "column mappings" tab to show up and populate the "Available Input Columns" by setting ExternalmetadataColumnCollection.IsUsed to true on the input. The problem is that the "Available destination columns" box is always empty. I have tried the IsUsed property on the output and pretty much every other property that I could find. On the Input and Output properties all of my columns show up under the output as both External and Output columns. Is there a separate collection for "destination" columns that I can't find? It's getting a little frustrating, is this something that can be done or do I have to write a custom UI to make it happen?
This is trivial I'm sure but I'll be dogged if I can find someone who mentions how to do it. I am attempting to develop a Data Flow Transformation that appends a new column (a string value) into the current stream.
I have found plenty of references on how to replace an existing column but I'd really like to just add my new column in there. It doesn't need to be configurable, it can be a static column name. I'll take a solution that allows the column name to be set at design time, don't get me wrong but the magic I'm looking for is how to implement a new column in a stream.
Yes, I am well aware of the derived column task but I will be replacing a few hundred instances and I'd much rather just drag an item onto the designer than to drag a derived column, double click it, type in the column name, set the expression and then set the datatype, etc.
Anyone spare a moment to enlighten me?
Pardon the lack of formatting, this BB doesn't play with Opera (I know, I'm a heretic)
using System; using System.Collections; using System.Runtime.InteropServices; using Microsoft.SqlServer.Dts.Pipeline; using Microsoft.SqlServer.Dts.Pipeline.Wrapper; using Microsoft.SqlServer.Dts.Runtime.Wrapper; using Microsoft.SqlServer.Dts.Runtime;
namespace Microsoft.Samples.SqlServer.Dts { [ DtsPipelineComponent ( DisplayName = "Nii", Description = "This is the component that says Nii.", ComponentType = ComponentType.Transform ) ] public class Nii : PipelineComponent {
public override void ProcessInput(int inputID, PipelineBuffer buffer) { if (!buffer.EndOfRowset) { while (buffer.NextRow()) { try { // do something here to } catch (Exception e) { ComponentMetaData.FireInformation(0, ComponentMetaData.Name, "There was an error on row " + buffer.CurrentRow.ToString() + ". The error is: " + e.Message + " : " + e.Source + " : " + e.StackTrace, "", 0, ref fireEventAgain); } } } } }
I want to recalculate some columns in a custom component, due to a formula using a single column as factor. What will be the best way to let the user choose the single column (factor) beneath the columns to work with.
Is there a way to use customproperties other than strings. DropDowns for example. If so, do you know a tutorial oder code snippet?
I'm building a custom transform component. I want to mark some input columns as keys for deduplicating. In a similar way to the provided Sort component, I want to check those columns and allow pass-throughs (or not) for the others - so next to each input column name I need two checkboxes (1:use for dedupe; 2:include in output if 1 not checked). If a column is checked for use in the dedupe, I want some other attributes to be shown indicating how it will be used. How do I display the checkboxes to let users select which columns to include for deduplication, and then how do I add further attributes underneath (copying the Sort component's look) for selection?
Thanks in advance for guidance and pointers on this.
A month or so ago I instigated this thread- http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=243117&SiteID=1 which talked about how to conditionally add a column to my component depending on the value of a custom property. If the custom property is TRUE then the column should appear in the output (and vice versa).
Bob Bojanic said I should use the SetComponentProperty() method to do this and that is working pretty well. However, it bothers me that SetComponentProperty() could be called, the column will then be added, and then the package developer could press 'Cancel'. In this instance the value of my custom property would be inconsistent with the presence of the extra column.
I'm having my first go at developing a destination adapter which will send data to an update Web Service.
I've got some rather big gaps in my understanding. I've been following the various samples I've found on the net and have validated my mapping and picked up all the available column names and datatypes which are appearing in the Input and Output Properties tab of the Advanced Editor but I only have a tab for "Input Columns" and not "Column Mappings".
Which method defines the availble columns for the user to map?
Let me know if I haven't given enough information.
My script component defines a (DT_WSTR, 450) output column named keyword. How can I access its length within the script, instead of hard-coding 450 as a constant?
My feeble attempts to tease this out through Visual Studio got me this far:
I have created a program that imports a csv into the sql server. but during that import I need to track all the errors that occured for some malformed rows. I think I need to use the error output collection of the dataflow components to track the errors. I figured out that every dataflow component has a error output collection along with the data output collection. I want to write those error outputs into a separete database. So, I have created a SQL server data destination component and created a path between derived columns error output and it input collection. But it is not working as expected. can any body help on this?
or can anyone give me any example how to use/handle error output collection in SSIS?
I would like to change the output of a column into a hyperlink however I am not sure that is possible without some further post-processing. I am aware I would have to do:
Code: select '<a href="' + URL + '" target="_blank" nav="web">' + tbl_ID.ID + '</a>' from tbl_ID however this obviously just exports the text <a href="[URL]" target="_blank" nav="web">[ID]</a>
This would work fine if I was exporting it into some sort of HTML table however this doesn't work with excel for example. Is there a way to make this possible or would I have to create a macro in excel to add the url with the ID separately?
Hi all I'm into a project which uses a lot of views for joining 2 or more tables. Using the MERGE component in SSIS will be a huge effort coz it only has 2 inputs and I gotta SORT the input too. Isnt it possible to have a VIEW like component that joins more than 2 tables and DOESNT need sorting?? (I've thought about creating views in database engine but it breaks my data floe in SSIS and is'nt a practical solution)
OLE DB source which calls a stored proc that returns a result set
data conversion
Excel destination I am in design mode in Business Intelligence studio. My excel destination (with an Excel Connection) shows no sheet name though I have an execute SQL task before the data flow to create the excel table called SHEET1. Needless to say, there are no output columns visible to do any mappings. I did go to the ExcelConnection to set the OpenRowset Property to SHEET1 but it seems to have no effect.
I can do the export in SQL Server Management studio and that works fine, but it is basic and does not meet my requirements. I have to customize the package to allow dynamic Excel filenames based on account names and have to split my result set into multiple excel sheets because excel 2003 has a max of 65536 rows per sheet. Also when I use the export wizard, I have the source as a table and eventually the source has to be a stored proc with input parms.
What am I missing or doing wrong? Thanks in advance
I am writing a custom dataflow transformation component and I need to get the name of the preceeding component.
I have been trying to find a way to get a reference to the Package object, MainPipe object or IDTSPath90 object (connecting to the IDTSInput90 of my component) from my component because I think from there I can get to the information I want.
How do I programatically change a column (say username) in a table (say tblusers) from varchar(25) to varchar(100)I am looking for something likealter tblusers set username as varchar(100) I know the above statement in nonsensical but it conveys the idea.
I have a table say Friend which has a column "IsSingle VARCHAR(10)" and this column has values like yes or No Now I want to change type of column IsSingle from VARCHAR to BIT, if I try to do it manaually SQL throws error that cannot convert yes to bit etc. I know that we can write a script to do this task but i dont know how ? Any pointers,links, suggestions will help me to start with.
I am using SQL Server 2000 Enterprise Edition fully patched. Database is in Simple Recovery mode.
I need to change a column's data type from "int(4)" to "smallint(2)". I know for sure that there will be no data (precision) lost, because I know the possible values that this column could have.
My problem is that the table I am dealing with has 600,000,000 records in it. I dropped all indexes before I tried to alter the table. But still it is taking forever and filling up my 280GB disk with transaction log file.
I know that in Oracle, if I want I can turn off logging and do these kind of modifications relatively faster.
I was wondering if there is a way of disabling logging before running this alter command.
What is the best practice to handle a situation of this sort?
What I want to accomplish is that at design time the designer can enter a value for some custom property on my custom task and that this value is accessed at executing time.
I am developing several charts with column type and sub type as stacked. There is a requirement from the users that they want an option to choose the type of chart.
Is it possible to change chart type dynamically from say Column type to Line type based on user option in front-end?
In a Data Flow, I have the necessity to use a SSIS variable of type €œObject€? inside Script Component and assign to it the content of 'n' variables of string type. On exiting from the script the variable of type object should contain something like in the following lines: AAAAAAAAAAAAAAAAAAAAAAAAAAAAA BBBBBBBBBBBBBBBBBBBBBBBBBBBBB CCCCCCCCCCCCCCCCCCCCCCCCCCCCC DDDDDDDDDDDDDDDDDDDDDDDDDDDDD €¦€¦€¦€¦€¦€¦€¦. €¦€¦€¦€¦€¦€¦€¦. On exiting from the data flow I will use the variable of type Object in a Script Task, by reading each element in a cyclic fashion. Is there anyone who have experienced something like this? Could anyone provide any example of that? Thanks in advance!
