File Readers

Learn how to configure Connect FilePulse for a specific file format.

The FilePulseSourceTask uses the FileInputReader. configured in the connector’s configuration for reading object files (i.e., tasks.reader.class).

Currently, Connect FilePulse provides the following FileInputReader implementations :

Amazon S3

package: io.streamthoughts.kafka.connect.filepulse.fs.reader

  • AmazonS3AvroFileInputReader
  • AmazonS3BytesArrayInputReader
  • AmazonS3RowFileInputReader
  • AmazonS3XMLFileInputReader
  • AmazonS3MetadataFileInputReader

Azure Blob Storage

package: io.streamthoughts.kafka.connect.filepulse.fs.reader

  • AzureBlobStorageAvroFileInputReader
  • AzureBlobStorageBytesArrayInputReader
  • AzureBlobStorageRowFileInputReader
  • AzureBlobStorageXMLFileInputReader
  • AzureBlobStorageMetadataFileInputReader

Google Cloud Storage

package: io.streamthoughts.kafka.connect.filepulse.fs.reader

  • GcsAvroFileInputReader
  • GcsBytesArrayInputReader
  • GcsRowFileInputReader
  • GcsXMLFileInputReader
  • GcsMetadataFileInputReader

Local Filesystem

package: io.streamthoughts.kafka.connect.filepulse.fs.reader

  • LocalAvroFileInputReader
  • LocalBytesArrayInputReader
  • LocalRowFileInputReader
  • LocalXMLFileInputReader
  • LocalMetadataFileInputReader

RowFileInputReader (default)

The <PREFIX>RowFileInputReaders can be used to read files line by line. This reader creates one record per row. It should be used for reading delimited text files, application log files, etc.


file.encodingThe text file encoding to useStringUTF_8High
buffer.initial.bytes.sizeThe initial buffer size used to read input files.String4096Medium minimum number of records to read from file before returning to task.Integer1Medium
skip.headersThe number of rows to be skipped in the beginning of file.Integer0Medium
skip.footersThe number of rows to be skipped at the end of file.Integer0Medium
read.max.wait.msThe maximum time to wait in milliseconds for more bytes after hitting end of file.Long0Medium


The <PREFIX>BytesArrayInputReaders create a single byte array record from a source file.


The <PREFIX>AvroFileInputReaders can be used to read Avro files.


The <PREFIX>XMLFileInputReaders can be used to read XML files.


xpath.expressionThe XPath expression used extract data from XML input filesString/High
xpath.result.typeThe expected result type for the XPath expression in [NODESET, STRING]StringNODESETHigh
force.array.on.fieldsThe comma-separated list of fields for which an array-type must be forcedList-High


The FileInputMetadataReaders can be used to send a single record per file containing metadata, i.e.: name, path, hash, lastModified, size, etc.