hurry.file fields

The file widget is built on top of the HurryFile object:

> >>> from hurry.file import HurryFile >>> file = > HurryFile('foo.txt', 'mydata') >>> file.filename 'foo.txt' > >>> file.data 'mydata' >>> file.size 6 >>> > f = file.file >>> f.read() 'mydata'

HurryFile objects are equal if data and filename of two of them match:

> >>> file1 = HurryFile('foo.txt', 'mydata') >>> file2 > = HurryFile('foo.txt', 'mydata') >>> file3 = > HurryFile('bar.txt', 'otherdata') >>> file1 == file2 True > > >>> file1 != file2 False > > >>> file1 == file3 False > > >>> file1 != file3 True

We can also create HurryFile objects from file-like objects:

> >>> from StringIO import StringIO >>> from zope > import component >>> from hurry.file.interfaces import > IFileRetrieval >>> fileretrieval = > component.getUtility(IFileRetrieval) >>> file = > fileretrieval.createFile('bar.txt', StringIO('test data')) > >>> file.filename 'bar.txt' >>> file.size 9 > >>> file.data 'test data' >>> f = file.file > >>> f.read() 'test data'

This does exactly the same, but may be easier to use:

> >>> from hurry.file import createHurryFile >>> file > = createHurryFile('test2.txt', StringIO('another test file')) > >>> file.filename 'test2.txt' >>> file.size 17

The HurryFile object normally stores the file data using ZODB persistence. Files can however also be stored by tramline. If tramline is installed in Apache, the Tramline takes care of generating ids for files and storing the file on the filesystem directly. The ids are then passed as file data to be stored in the ZODB.

Let's first enable tramline.

The tramline directory structure is a directory with two subdirectories, one called 'repository' and the other called 'upload':

> >>> import tempfile, os >>> dirpath = > tempfile.mkdtemp() >>> repositorypath = os.path.join(dirpath, > 'repository') >>> uploadpath = os.path.join(dirpath, > 'upload') >>> os.mkdir(repositorypath) >>> > os.mkdir(uploadpath)

We create a TramlineFileRetrieval object knowing about this directory, and register it as a utility:

> >>> from hurry.file.file import TramlineFileRetrievalBase > >>> class TramlineFileRetrieval(TramlineFileRetrievalBase): > ... def getTramlinePath(self): ... return dirpath >>> > retrieval = TramlineFileRetrieval() >>> > component.provideUtility(retrieval, IFileRetrieval)

Now let's store a file the way tramline would during upload:

> >>> f = open(os.path.join(repositorypath, '1'), 'wb') > >>> f.write('test data') >>> f.close()

The file with underlying name '1' (the data stored in the ZODB will be just '1') will now be created:

> >>> file = HurryFile('foo.txt', '1')

The data is now '1', referring to the real file:

> >>> file.data '1'

Retrieving the file results in the real file:

> >>> f = file.file >>> f.read() 'test data'

We can also retrieve its size:

> >>> file.size 9L

It should be possible to create Hurry File objects that are stored in the directory structure directly:

> >>> file = retrieval.createFile('test.txt', StringIO('my test > data')) >>> file.filename 'test.txt'

We get an id for the data now:

> >>> file.data != 'my test data' True

And we can retrieve the file itself:

> >>> f = file.file >>> f.read() 'my test data' > >>> file.size 12L

Now let's disable tramline in our utility:

> >>> class TramlineFileRetrieval(TramlineFileRetrievalBase): > ... def getTramlinePath(self): ... return dirpath ... def > isTramlineEnabled(self): ... return False >>> > component.provideUtility(TramlineFileRetrieval(), IFileRetrieval)

We expect the same behavior as when tramline is not installed:

> >>> file = HurryFile('foo.txt', 'data') >>> f = > file.file >>> f.read() 'data' >>> file.size 4

Clean up:

> >>> import shutil >>> shutil.rmtree(dirpath)

hurry.file widgets

This is an infrastructure to create a file widget that behaves as much as possible like a normal text widget in formlib. Normally a file widget loses its file data when a form is re-presented for reasons of failing form validation. A hurry.file widget retains the file, for example by storing it in a session.

In order to do this, we have a special way to store file data along with its filename:

>>> from hurry.file import HurryFile
>>> some_file = HurryFile('foo.txt', 'the contents')
>>> some_file.filename
'foo.txt'
>>> some_file.data
'the contents'

We can provide a download widget. In this case, there's nothing to download:

>>> from hurry.file.browser import DownloadWidget
>>> from hurry.file.schema import File
>>> from zope.publisher.browser import TestRequest
>>> field = File(__name__='foo', title=u'Foo')
>>> field = field.bind(None)
>>> request = TestRequest()
>>> widget = DownloadWidget(field, request)
>>> widget()
u'<div>Download not available</div>'

Even if there were data in the request, there'd be nothing to download:

>>> from zope.publisher.browser import FileUpload
>>> request = TestRequest(form={'field.foo': FileUpload(some_file)})
>>> widget = DownloadWidget(field, request)
>>> widget()
u'<div>Download not available</div>'

Now set a value:

>>> widget.setRenderedValue(some_file)
>>> widget()
u'<a href="foo.txt">foo.txt</a>'

Now on to an edit widget. First the case in an add form with no data already available, and no data in request:

>>> from hurry.file.browser import EncodingFileWidget
>>> field = File(__name__='foo', title=u'Foo', required=False)
>>> field = field.bind(None)
>>> request = TestRequest()
>>> widget = EncodingFileWidget(field, request)

>>> def normalize(s):
...   return '\n  '.join(filter(None, s.split(' ')))

>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">

Now let's try a situation where data is available in the request, but it's an empty string for the file:

>>> request = TestRequest(form={'field.foo': u''})
>>> widget = EncodingFileWidget(field, request)

>>> def normalize(s):
...   return '\n  '.join(filter(None, s.split(' ')))

>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">

Now let's render again when there's already available data. What should show up is an extra, hidden field which contains the file_id:

>>> widget.setRenderedValue(some_file)
>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">
  (foo.txt)<input value="Zm9vLnR4dAp0aGUgY29udGVudHM=" type="hidden" name="field.foo.file_id" id="field.foo.file_id" class="hiddenType">

Now let's render again, this time with file data available in the request instead. The same should happen:

>>> request = TestRequest(form={'field.foo': FileUpload(some_file)})
>>> widget = EncodingFileWidget(field, request)
>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">
  (foo.txt)<input value="Zm9vLnR4dAp0aGUgY29udGVudHM=" type="hidden" name="field.foo.file_id" id="field.foo.file_id" class="hiddenType">

Now let's render again, this time not with file data available in the request, but an id. Again, we should see the same:

>>> request = TestRequest(form={'field.foo.file_id': 
...                             'Zm9vLnR4dAp0aGUgY29udGVudHM='})
>>> widget = EncodingFileWidget(field, request)
>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">
  (foo.txt)<input value="Zm9vLnR4dAp0aGUgY29udGVudHM=" type="hidden" name="field.foo.file_id" id="field.foo.file_id" class="hiddenType">

If there is both file data and an id, something else happens. First, let's prepare some new file:

>>> another_file = HurryFile('bar.txt', 'bar contents')

We happen to know, due to the implementation of EncodingFileWidget, that the file_id is going to be "YmFyLnR4dApiYXIgY29udGVudHM=". Let's make a request with the original id, but a new file upload:

>>> request = TestRequest(form={'field.foo': FileUpload(another_file),
...                             'field.foo.file_id': 
...                             'Zm9vLnR4dAp0aGUgY29udGVudHM='})

We expect the new file to be the one that's uploaded:

>>> widget = EncodingFileWidget(field, request)
>>> print normalize(widget())
<input type="file" size="20" name="field.foo" id="field.foo" class="fileType">
  (bar.txt)<input value="YmFyLnR4dApiYXIgY29udGVudHM=" type="hidden" name="field.foo.file_id" id="field.foo.file_id" class="hiddenType">

hurry.file changes

1.2.1 (2011-08-09)

• Fix buggy inequality comparison of HurryFiles.

1.2 (2009-03-11)

• Add a 'size' attribute that knows the size of the file in bytes (no matter whether it's stored using tramline or in the ZODB).

1.1 (2008-08-07)

• Add in a buildout.cfg that installs the test runner.
• List dependencies in setup.py.
• Rely on zope.session instead of zope.app.session to stop deprecation warnings.
• Add long description in setup.py based on README.txt and file.txt doctests, and CHANGES.txt.

1.0 (2006-10-25)

• Support for Tramline (fast file uploads/downloads) through IFileRetrieval. By default, nothing changes.

  If a subclass of TramlineFileRetrievalBase is registered as a IFileRetrieval utility, hurry.file becomes Tramline aware. If files are created manually, they can be created through the createHurryFile function, or the 'createFile' method of the IFileRetrieval service. This will take care of storing the file in the right place.

  Tramline can be found here:

0.9.3 (2006-10-23)

• Send tramline_ok header back when redisplaying widget, in case we're working with tramline.

0.9.2 (2006-09-28)

• Zope 3.3 has a change in the way it deals with file name encoding which broke hurry.file. This includes a workaround.

0.9.1 (2006-09-22)

• first cheeseshop release.

0.9 (2006-06-15)

• separation from general hurry package into hurry.file
• eggification

0.8 (2006-05-01)

Initial public release.