View Source

This section describes how to create datasets and replace segments by parameterizing the input paths to your data in .

Structuring Your Data

Each file that is included as part of the dataset with parameters should have identical structures:

Matching file formats
Matching column order, naming, and data type

Matching column headers. Each column in any row that is part of a column header in a dataset with parameters should have a valid value that is consistent with corresponding values across all files in the dataset.

NOTE: If your files have missing or empty values in rows that are used as headers in your recipe, these rows may be treated as data rows during the import process, which may result in unexpected or missing column values.

Within each column, the data format should be consistent.
- For example, if the date formats change between files in the source system, you and your recipe may not be able to manage the differences, and it is possible that data in the output may be missing.

NOTE: Avoid creating datasets with parameters where individual files or tables have differing schemas. Either import these sources separately and then correct in the application before performing a union on the datasets, or make corrections in the source application to standardize the schemas.

When working with datasets with parameters, it may be useful to do the following if you expect the underlying datasets to be less than 100% consistent with each other.

Recreate the dataset with parameters, except deselect the Detect Structure option during the import step.
In the Transformer page, collect a Random Sample using a full scan. This step attempts to gather data from multiple individual files, which may illuminate problems across the data.

Tip: If you suspect that there is a problem with a specific file or rows of data (e.g. from a specific date), you can create a static dataset from the file in question.

NOTE: For parameterized datasets sourced from , only the first 100,000 files are read.

Steps

NOTE: Matching file path patterns in a large directory can be slow. Where possible, avoid using multiple patterns to match a file pattern or scanning directories with a large number of files. To increase matching speed, avoid wildcards in top-level directories and be as specific as possible with your wildcards and patterns.

In the Import Data page, navigate your environment to locate one of the files or tables that you wish to parameterize.
Click Create Dataset with Parameters.
Create Dataset with Parameters
Within the Define Parameterized Path, select a segment of text. Then select one of the following options:
1. Add Datetime Parameter
2. Add Variable
3. Add Pattern Parameter - wildcards and patterns
4. For more information on limitations, see Overview of Parameterization.
5. If you need to navigate elsewhere, select Browse.
Specify the parameter. Click Save.
Click Update matches. Verify that all of your preferred datasets are matching.
NOTE: If you are matching with more datasets than you wish, you should review your parameters.
Click Create.
The parameterized dataset is loaded. See Import Data Page.

A flow containing a dataset with parameters has additional options for managing them. See Flow View Page.

Add Datetime Parameter

Datetime parameters require the following elements:

Format: You must specify the format of the matching date and/or time values using alphanumeric patterns. To review a list of example formats, click Browse Date/Timestamp Patterns.

You can also create custom formats using patterns. For example, the following regex pattern matches patterns like MM.DD.YYYY:

/[0-9][0-9]\.[0-9][0-9]\.[0-9][0-9][0-9][0-9]/

Date range: Use these controls to specify the range that matching dates must fall within.

NOTE: Date range parameters are case-insensitive.

Tip: Datetime parameters that you configure here are evaluated at the time of job execution. So, now refers to the time when the job is executed.

Time zone: The default time zone is the location of the host of the application. To change the current time zone, click Change.

For a list of supported time zone values, see Supported Time Zone Values.

Extend Datetime parameter

A parameterized dataset can support only one Datetime parameter. If you have multiple parts of the path that contain date information, you can create a Datetime element for each part.

Steps:

Within the Define Parameterized Path, select a segment of text for which to create the first part.
Create the Datetime parameter for this element. Remember to use the appropriate format for the part. For example, if you have highlighted a four-digit year for the part, the date format value should be: YYYY.

Then, select the second element and click the Extend Datetime Parameter icon.
Dataprep by Trifacta > Parameterize Files for Import > CreateDatasetWithParameters-ExtendDatetimeParameter.png

Click the Extend Datetime Parameter icon to create additional parts to your Datetime parameter.

In the dialog, you can specify the date format of the second element of your Datetime parameter. Matches are made on the two elements, as well as any static text in between them.

Add Variable

A variable parameter is a key-value pair that can be inserted into the path.

At execution time, the default value is applied, or you can choose to override the value.
A variable can have an empty default value.

Name: The name of the variable is used to identify its purpose.

NOTE: If multiple datasets within the same flow share the same variable name, they are treated as the same variable.

Default Value: If the variable value is not overridden at execution time, this value is inserted in the variable location in the path.

NOTE: When you edit an imported dataset, if a variable is renamed, a new variable is created using the new name. Any override values assigned under the old variable name for the dataset must be re-applied. Instances of the variable and override values used in other imported datasets remain unchanged.

Add Pattern Parameter

In the screen above, you can see an example of pattern-based parameterization. In this case, you are trying to parameterize the two digits after the value: POS-r.

Include nested folders

When you create a wildcard or pattern-based parameter, you have the option to scan any nested folders for matching sources.

If disabled, the scan stops when the next slash (/) in the path is encountered. Folders are not matched.
If enabled, the scan continues to any depth of folders.
NOTE: A high number of files and folders to scan can significantly increase the time required to load your dataset with parameters.

Example 1: all text files

Suppose your file and folder structure look like the following:

//source/user/me/datasets/thisfile.txt
//source/user/me/datasets/thatfile.txt
//source/user/me/datasets/anotherfile.csv
//source/user/me/datasets/detail/anestedfile.txt
//source/user/me/datasets/detail/anestedfile2.txt
//source/user/me/datasets/detail/anestedfile4.txt
//source/user/me/ahigherfile.txt

Since the filenames vary significantly, it may be easiest to create your pattern based on a wildcard. You create a wildcard parameter on the first file in the //source/user/me/datasets directory:

//source/user/me/datasets/*.txt

For the specified directory, the above pattern matches on any text file (.txt). In the example, it matches on the first two files but does not match on the CSV file.

When the Include nested folders checkbox is selected:

The first two files are matched.
The next three files inside a nested folder are matched.
The last file (ahgherfile.txt) is not matched, since it is not inside a nested folder.

Example 2: pattern-based files

Suppose your file and folder structure look like the following:

//source/user/me/datasets/file01.csv
//source/user/me/datasets/file02.csv
//source/user/me/datasets/file03.csv
//source/user/me/datasets/detail/file04.csv
//source/user/me/datasets/detail/file05.csv
//source/user/me/datasets/detail/file06.csv
//source/user/me/file07.csv

You create a pattern parameter on the first file in the //source/user/me/datasets directory with the following pattern-based parameter:

`file{digit}+`

The above pattern matches on the word file and a sequence of one or more digits. For example, suppose file100.csv lands in the directory at some point in the future. This pattern would capture it.

In the above example, this pattern matches on the first three files, which are all in the same directory.

When the Include nested folders checkbox is selected:

The first three files are matched.
The next three files inside a nested folder are matched.
The last file (file07.csv) is not matched, since it is not inside a nested folder.

Wildcard

The easiest way to is to add a wildcard: *

A wildcard can be any value of any length, including an empty string.

Tip: Wildcard matching is very broad. If you are using wildcards, you should constrain them to a very small part of the overall path. Some running environment may place limits on the number of files with which you can match.

Pattern - Regular expression

Instead of a wildcard match, you could specify a regular expression match. Regular expressions are a standardized means of expressing patterns.

Regular expressions are specified between forward slashes, as in the following:

/my_regular_expression/

NOTE: If regular expressions are poorly specified, they can create unexpected matches and results. Use them with care. For a list of limitations of regular expressions for parameterization, see Overview of Parameterization.

The following regular expression matches the same two sources in the previous screen:

/\_[0-9]*\_[0-9]*/

The above expression matches an underscore (_) followed by any number of digits, another underscore, and any number of digits.

Tip: In regular expressions, some characters have special meaning. To ensure that you are referencing the literal character, you can insert a backslash (\) before the character in question.

While the above matches the two sources, it also matches any of the following:

_2_1
__1
_1231231231231231235245234343_

These may not be proper matches. Instead, you can add some specificity to the expression to generate a better match:

/\_[0-9]{13}\_[0-9]{4}/

The above pattern matches an underscore, followed by exactly 13 digits, another underscore, and then another 4 digits. This pattern matches the above two sources exactly, without introducing the possibility of matching other numeric patterns.

Pattern - match

A is a platform-specific mechanism for specifying patterns, which is much simpler to use than regular expressions. These simple patterns can cover much of the same range of pattern expression as regular expressions without the same risks of expression and sometimes ugly syntax. For more information on , see Text Matching.

are specified between back-ticks, as in the following:

`my_pattern`

In the previous example, the following regular expression was used to match the proper set of files:

/\_[0-9]{13}\_[0-9]{4}/

In a , the above can be expressed in a simpler format:

`\_{digit}{13}\_{digit}{4}`

This simpler syntax is easier to parse and performs the same match as the regular expression version.