Normalizer

Short Description

Normalizer creates one or more output records from each single input record. Input records do not have to be sorted.

COMPONENTSAME INPUT METADATASORTED INPUTSINPUTSOUTPUTSJAVACTLAUTO-PROPAGATED METADATA
Normalizer-x11x

Ports

PORT TYPENUMBERREQUIREDDESCRIPTIONMETADATA
Input0For input data recordsAny1
Output0For normalized data recordsAny2

Metadata

Normalizer does not propagate metadata.
Normalizer does not have metadata templates.
Normalizer does not require any specific metadata fields.

Normalizer Attributes

ATTRIBUTEREQDESCRIPTIONPOSSIBLE VALUES
BASIC
Normalize[1]The definition of the way how records should be normalized, written in the graph in CTL or Java.
Normalize URL[1]The name of an external file, including the path, containing the definition of the way how records should be normalized written in CTL or Java.
Normalize class[1]The name of an external class defining the way how records should be normalized.
Normalize source charsetEncoding of an external file defining the transformation.

The default encoding depends on DEFAULT_SOURCE_CODE_CHARSET in defaultProperties.
E.g. UTF-8
DEPRECATED
Error actionsDefinition of the action that should be performed when the specified transformation returns some Error code. See Return Values of Transformations.
Error logThe URL of the file to which error messages for specified Error actions should be written. If not set, they are written to Console.

[1] One of these must specified.

Details

Normalizer requires transformation. The transformation can be defined in CTL (see CTL Interface below) or in Java (see Java Interface below).
The transformation is defined using several functions. Each of them has its own purpose. The order of function calls is depicted in diagram below.

The number of calls of transform() function depends on the result of the count() (or countOnError()) function.

CTL Interface

The transformation written in CTL uses a CTL template for Normalizer. Only the functions count() and transform() are mandatory. Other functions are optional.

Once you have written your transformation, you can also convert it to Java language code using the corresponding button in the upper right corner of the tab.

CTL Templates for Normalizer

CTL TEMPLATE FUNCTIONSboolean init()
RequiredNo
DescriptionInitializes the component, sets up the environment and global variables
InvocationCalled before processing the first record
Returnstrue | false (in case of false graph fails)
CTL TEMPLATE FUNCTIONSinteger count()
RequiredYes
Input ParametersNone
ReturnsThe returned number defines the number of new output records that will be created by the transform() function.

If the count() function returns 0, the subsequent call of transform() is skipped (transform() is called zero times).
InvocationCalled repeatedly, once for each input record
DescriptionFor each input record it generates the number of output records that will be created from this input.

If count() fails and user has not defined any countOnError(), the whole graph will fail.

If any of the input records causes the count() function to fail, and if user has defined another function (countOnError()), processing continues in this countOnError() at the place where count() failed. The countOnError() function gets the information gathered by count() that was received from previously successfully processed input records. Also the error message and stack trace are passed to countOnError().
ExampleSee here below
function integer count() {
   customers = split($in.0.customers,"-");
   return length(customers);
}
CTL TEMPLATE FUNCTIONSinteger transform()
RequiredYes
Input Parametersinteger idx integer numbers from 0 to count-1 (Here count is the number returned by the count() function.)
ReturnsInteger numbers. The number corresponds to the return value of transformation. For detailed information, see Return Values of Transformations.
InvocationCalled repeatedly, once for each output record. The number of calls is defined by return value of function count().
DescriptionCreates output records.

If transform() fails and the user has not defined any transformOnError(), the whole graph will fail.

If any part of the transform() function for some output record causes fail of the transform() function, and if user has defined another function (transformOnError()), processing continues in this transformOnError() at the place where transform() failed. The transformOnError() function gets the information gathered by transform() that was received from previously successfully processed code. Also the error message and stack trace are passed to transformOnError().
ExampleSee here below
function integer transform(integer idx) {
   myString = customers[idx];
   $out.0.OneCustomer = str2integer(myString);
   $out.0.RecordNo = $in.0.recordNo;
   $out.0.OrderWithinRecord = idx;
   return OK;
}
CTL TEMPLATE FUNCTIONSvoid clean()
RequiredNo
Input ParametersNone
Returnsvoid
InvocationCalled repeatedly, once for each input record.

The function is called after the corresponding call(s) of transform() function.
DescriptionReturns the component to the initial settings.
ExampleSee here below
function void clean() {
   clear(customers);
}
CTL TEMPLATE FUNCTIONSinteger countOnError(string errorMessage, string stackTrace)
RequiredNo
Input Parametersstring errorMessage
string stackTrace
ReturnsThe returned number defines the number of new output records that will be created by the transform() function.

If the count() function returns 0, the subsequent call of transform() is skipped.
InvocationCalled if count() throws an exception.
DescriptionFor each input record it generates the number of output records that will be created from this input.

If any of the input records causes fail of the count() function, and if user has defined another function (countOnError()), processing continues in this countOnError() at the place where count() failed.
ExampleSee here below
function integer countOnError(
                  string errorMessage,
                  string stackTrace) {
   printErr(errorMessage);
   return 1;
}
CTL TEMPLATE FUNCTIONSinteger transformOnError(string errorMessage, string stackTrace, integer idx)
RequiredNo
Input Parametersstring errorMessage
string stackTrace
integer idx
ReturnsInteger numbers. For more information, see Return Values of Transformations.
InvocationCalled if transform() throws an exception.
DescriptionCreates output records.

If any part of the transform() function for some output record causes fail of the transform() function, and if the user has defined another function (transformOnError()), processing continues in this transformOnError() at the place where transform() failed.

The transformOnError() function gets the information gathered by transform() that was received from previously successfully processed code. Also the error message and stack trace are passed to transformOnError().
ExampleSee here below
function integer transformOnError(
                  string errorMessage,
                  string stackTrace,
                  integer idx) {
   printErr(errorMessage);
   printErr(stackTrace);
   $out.0.OneCustomerOnError = customers[idx];
   $out.0.RecordNo = $recordNo;
   $out.0.OrderWithinRecord = idx;
   return OK;
}
CTL TEMPLATE FUNCTIONSstring getMessage()
RequiredNo
DescriptionPrints the error message specified and invoked by the user.
InvocationCalled in any time specified by the user (called only when either count(), transform(), countOnError(), or transformOnError() returns value less than or equal to -2).
Returnsstring
CTL TEMPLATE FUNCTIONSvoid preExecute()
RequiredNo
Input ParametersNone
Returnsvoid
DescriptionMay be used to allocate and initialize resources required by the transform. All resources allocated within this function should be released by the postExecute() function.
InvocationCalled during each graph run before the transform is executed.
CTL TEMPLATE FUNCTIONSvoid postExecute()
RequiredNo
Input ParametersNone
Returnsvoid
DescriptionShould be used to free any resources allocated within the preExecute() function.
InvocationCalled during each graph run after the entire transform was executed.

Access to input and output fields

Input records or fields
Input records or fields are accessible within the count(), countOnError(),transform() and transformOnError() functions only.

Output records or fields
Output records or fields are accessible within the transform() and transformOnError() functions only.

Capt. Eddie to ground control:

All of the other CTL template functions allow to access neither inputs nor outputs.

Remember that if you do not hold these rules, NPE will be thrown!

Java Interface

The transformation implements methods of the RecordNormalize interface and inherits other common methods from the Transform interface. See Common Java Interfaces.
Following are the methods of the RecordNormalize interface:

  • boolean init(Properties parameters, DataRecordMetadata sourceMetadata, DataRecordMetadata targetMetadata)
    Initializes normalize class/function. This method is called only once at the beginning of normalization process. Any object allocation/initialization should happen here.
  • int count(DataRecord source)
    Returns the number of output records which will be created from specified input record.
  • int countOnError(Exception exception, DataRecord source)
    Called only if count(DataRecord) throws an exception.
  • int transform(DataRecord source, DataRecord target, int idx)
    idx is a sequential number of output record (starting from 0). For detailed information about return values and their meaning, see Return Values of Transformations. In Normalizer, only ALL, 0, SKIP, and Error codes have some meaning.
  • int transformOnError(Exception exception, DataRecord source, DataRecord target, int idx)
    Called only if transform(DataRecord, DataRecord, int) throws an exception.
  • void clean()
    Finalizes current round/clean after current round - called after the transform method was called for the input record.

Examples

Converting multivalue fields to multiple records

Input records contain group name and list of users of the group. Convert records into tuples having group name and one username.

accounting  | [johnsmith, elisabethtaylor]
development | [georgegreen, janegreen, peterbrown]

Solution
Define the transformation using Normalize attribute.

//#CTL2

function integer count() {
    return length($in.0.users);
}

function integer transform(integer idx) {
    $out.0.group = $in.0.group;
    $out.0.user = $in.0.users[idx];
    return OK;
}

Normalizer returns following records:

accounting |johnsmith
accounting |elisabethtaylor
development|georgegreen
development|janegreen
development|peterbrown

Best Practices

If the transformation is specified in an external file (with Normalize URL), we recommend users to explicitly specify Normalize source charset.

See also

Denormalizer
Rollup
Common Properties of Components
Specific attribute types
Common Properties of Transformers
Defining Transformations