public interface BinaryFields
UI components
such as ListGrid
and FileItem
will offer UIs to download, view and upload
files stored in a binary field. DataSource.downloadFile
and DataSource.viewFile
can be used to programmatically trigger download/viewing as well. Note
that these automatic features, and components like TileGrid
that make use of them, only work correctly for
DataSources that declare a valid primaryKey
field.
As covered under the "binary"
field type description
, binary fields imply 3 other related metadata fields named after the
binary field, which store the file size, file name and date of creation.
When using one of the built-in server DataSource types (SQL, JPA, Hibernate), these metadata fields, UI controls and APIs will work with no special effort - just declare a binary field. For JPA or Hibernate the Java getter/setter for the binary field must be of type byte[] or Byte[] (this is imposed by the ORM system); if you are using the SQL DataSource, no additional declarations are required unless you are using beanClassName, in which case you must added getters/setters of type InputStream, or one of the types listed for JPA/Hibernate.
For metadata fields like
the file size, providing storage for the metadata fields (SQL columns or JPA/Hibernate
getter/setters) is optional but highly recommended. If you don't want a particular metadata
field, you can declare a <field> for it explicitly and set ignore="true". Note:
If you decide not to store the "_filename" metadata element, the built-in binary UI controls
in components like ListGrid, mentioned above, do not work properly. In the absence of a
filename metadata field, it is necessary to set mime-type
on the binary field -
otherwise, we can't make a reasonable guess at a MIME type, so the browser tends not to know
what to do with the downloaded content. Omiting the file size metadata will simply cause the
browser not to show a progress dialog during download, or not to show an accurate one.
Downloading from binary fields
If writing a custom DataSource or writing a DMI, implement the "downloadFile" and "viewFile" operationTypes, and return data for a single Record which has binary data as the field value for the binary field - accepted types are InputStream, byte[], java.sql.Blob or String.
For example, you could return a Java bean
with a getter method which returns a byte[], and which is named after the binaryField using
Java Beans conventions: getFieldName()
. Or, return a Java Map where a
byte[] is stored under a key named the same as the binary field's name.
As discussed under
FieldType
, additional metadata fields are implied when you
declare a binary field. Returning a value for the "_filesize" field will allow the browser to
show progress of the download. Returning a value for the "_filename" field will cause the
downloaded file to be named after the "_filename" value.
Uploading to binary fields
Again with the built-in server DataSource types, no special effort is required.
For a
custom DataSource, an upload from a FileItem
control (the default upload control) is treated just like an ordinary "add" or "update"
DataSource operation. The uploaded file(s), if any, are available from the server-side API
dsRequest.getUploadedFile()
. Metadata fields such as "_filename" can be populated
from the data in the returned ISCFileItem.
Binary fields and normal "fetch"
For an ordinary "fetch" operation, it's generally useless to return data for a binary field, because in most cases code running in the browser would not be able to do anything with a binary value (such as invoke a PDF viewing plugin). For this reason the Smart GWT Server will automatically omit values of type InputStream, byte[] or Blob during a normal "fetch".
However you can deliver the binary data to the browser by transforming it to a Base64-encoded
String by setting encodeInResponse
on your <field> declaration. This can be used with certain browser
features such as image data URIs, but note that some older browsers (notably IE7 and
earlier) do not support data URIs.
Downloads unrelated to binary fields
If you
want to download or view a file that is not stored in a DataSource record, for example, you
want to dynamically generate a report of some kind on the fly, any DMI can return a binary
response by calling the server-side API RPCManager.doCustomResponse()
and writing
binary data directly to the ServletResponse outputStream.
It's typical to represent an operation that returns a binary stream as a DataSource request with operationType "fetch", especially if it takes Criteria that identify DataSource Records that will be used to create the binary stream.
When initiating such a download, set downloadResult
via the
requestProperties
argument of DataSource.fetchData
so that the client-side framework knows that a download will occur
instead of a normal response. In this case, callbacks will not be fired because code in the
browser does not receive notification that the download has initiated or has completed.
Binary handling without the Smart GWT Server
To download or view files with the Smart GWT Server, you can write a servlet that streams back binary data. For a download, you can then redirect the main page to the servlet using by setting window.location - be sure the download can never fail if you this, because any error message returned by the server in lieu of a file will replace the application instead of triggering a "Save As.." dialog.
To view a file, you can open a new browser window to the URL of the servlet.