Refactoring notebook managers and adding Azure backed storage.

I have created a base class for all notebook managers. Our
existing, file-based store, is now in filenbmanager.py. I have
also created a new Azure Blob based backed notebook manager.

IPython/frontend/html/notebook/azurenbmanager.py

@@ -0,0 +1,140 b''
+"""A notebook manager that uses Azure blob storage.
+
+Authors:
+
+* Brian Granger
+"""
+
+#-----------------------------------------------------------------------------
+#  Copyright (C) 2012  The IPython Development Team
+#
+#  Distributed under the terms of the BSD License.  The full license is in
+#  the file COPYING, distributed as part of this software.
+#-----------------------------------------------------------------------------
+
+#-----------------------------------------------------------------------------
+# Imports
+#-----------------------------------------------------------------------------
+
+import datetime
+
+import azure
+from azure.storage import BlobService
+
+from tornado import web
+
+from .basenbmanager import BaseNotebookManager
+from IPython.nbformat import current
+from IPython.utils.traitlets import Unicode, Instance
+
+
+#-----------------------------------------------------------------------------
+# Classes
+#-----------------------------------------------------------------------------
+
+class AzureNotebookManager(BaseNotebookManager):
+
+    account_name = Unicode('', config=True, help='Azure storage account name.')
+    account_key = Unicode('', config=True, help='Azure storage account key.')
+    container = Unicode('', config=True, help='Container name for notebooks.')
+
+    blob_service_host_base = Unicode('.blob.core.windows.net', config=True,
+        help='The basename for the blob service URL. If running on the preview site this '
+             'will be .blob.core.azure-preview.com.')
+    def _blob_service_host_base_changed(self, new):
+        self._update_service_host_base(new)
+
+    blob_service = Instance('azure.storage.BlobService')
+    def _blob_service_default(self):
+        return BlobService(account_name=self.account_name, account_key=self.account_key)
+
+    def __init__(self, **kwargs):
+        super(BaseNotebookManager,self).__init__(**kwargs)
+        self._update_service_host_base(self.blob_service_host_base)
+        self._create_container()
+
+    def _update_service_host_base(self, shb):
+        azure.BLOB_SERVICE_HOST_BASE = shb
+
+    def _create_container(self):
+        self.blob_service.create_container(self.container)
+
+    def load_notebook_names(self):
+        """On startup load the notebook ids and names from Azure.
+
+        The blob names are the notebook ids and the notebook names are stored
+        as blob metadata.
+        """
+        self.mapping = {}
+        blobs = self.blob_service.list_blobs(self.container)
+        ids = [blob.name for blob in blobs]
+        
+        for id in ids:
+            md = self.blob_service.get_blob_metadata(self.container, id)
+            name = md['x-ms-meta-nbname']
+            self.mapping[id] = name
+
+    def list_notebooks(self):
+        """List all notebooks in the container.
+
+        This version uses `self.mapping` as the authoritative notebook list.
+        """
+        data = [dict(notebook_id=id,name=name) for id, name in self.mapping.items()]
+        data = sorted(data, key=lambda item: item['name'])
+        return data
+
+    def read_notebook_object(self, notebook_id):
+        """Get the object representation of a notebook by notebook_id."""
+        if not self.notebook_exists(notebook_id):
+            raise web.HTTPError(404, u'Notebook does not exist: %s' % notebook_id)
+        try:
+            s = self.blob_service.get_blob(self.container, notebook_id)
+        except:
+            raise web.HTTPError(500, u'Notebook cannot be read.')
+        try:
+            # v1 and v2 and json in the .ipynb files.
+            nb = current.reads(s, u'json')
+        except:
+            raise web.HTTPError(500, u'Unreadable JSON notebook.')
+        # Todo: The last modified should actually be saved in the notebook document.
+        # We are just using the current datetime until that is implemented.
+        last_modified = datetime.datetime.utcnow()
+        return last_modified, nb
+
+    def write_notebook_object(self, nb, notebook_id=None):
+        """Save an existing notebook object by notebook_id."""
+        try:
+            new_name = nb.metadata.name
+        except AttributeError:
+            raise web.HTTPError(400, u'Missing notebook name')
+
+        if notebook_id is None:
+            notebook_id = self.new_notebook_id(new_name)
+
+        if notebook_id not in self.mapping:
+            raise web.HTTPError(404, u'Notebook does not exist: %s' % notebook_id)
+
+        try:
+            data = current.writes(nb, u'json')
+        except Exception as e:
+            raise web.HTTPError(400, u'Unexpected error while saving notebook: %s' % e)
+
+        metadata = {'nbname': new_name}
+        try:
+            self.blob_service.put_blob(self.container, notebook_id, data, 'BlockBlob', x_ms_meta_name_values=metadata)
+        except Exception as e:
+            raise web.HTTPError(400, u'Unexpected error while saving notebook: %s' % e)
+
+        self.mapping[notebook_id] = new_name
+        return notebook_id
+
+    def delete_notebook(self, notebook_id):
+        """Delete notebook by notebook_id."""
+        if not self.notebook_exists(notebook_id):
+            raise web.HTTPError(404, u'Notebook does not exist: %s' % notebook_id)
+        try:
+            self.blob_service.delete_blob(self.container, notebook_id)
+        except Exception as e:
+            raise web.HTTPError(400, u'Unexpected error while deleting notebook: %s' % e)
+        else:
+            self.delete_notebook_id(notebook_id)

IPython/frontend/html/notebook/basenbmanager.py

@@ -0,0 +1,181 b''
+"""A base class notebook manager.
+
+Authors:
+
+* Brian Granger
+"""
+
+#-----------------------------------------------------------------------------
+#  Copyright (C) 2011  The IPython Development Team
+#
+#  Distributed under the terms of the BSD License.  The full license is in
+#  the file COPYING, distributed as part of this software.
+#-----------------------------------------------------------------------------
+
+#-----------------------------------------------------------------------------
+# Imports
+#-----------------------------------------------------------------------------
+
+import uuid
+
+from tornado import web
+
+from IPython.config.configurable import LoggingConfigurable
+from IPython.nbformat import current
+from IPython.utils.traitlets import List, Dict
+
+#-----------------------------------------------------------------------------
+# Classes
+#-----------------------------------------------------------------------------
+
+class BaseNotebookManager(LoggingConfigurable):
+
+    allowed_formats = List([u'json',u'py'])
+
+    # Map notebook_ids to notebook names
+    mapping = Dict()
+
+    def load_notebook_names(self):
+        """Load the notebook names into memory.
+
+        This should be called once immediately after the notebook manager
+        is created to load the existing notebooks into the mapping in
+        memory.
+        """
+        self.list_notebooks()
+
+    def list_notebooks(self):
+        """List all notebooks.
+
+        This returns a list of dicts, each of the form::
+
+            dict(notebook_id=notebook,name=name)
+
+        This list of dicts should be sorted by name::
+
+            data = sorted(data, key=lambda item: item['name'])
+        """
+        raise NotImplementedError('must be implemented in a subclass')
+
+
+    def new_notebook_id(self, name):
+        """Generate a new notebook_id for a name and store its mapping."""
+        # TODO: the following will give stable urls for notebooks, but unless
+        # the notebooks are immediately redirected to their new urls when their
+        # filemname changes, nasty inconsistencies result.  So for now it's
+        # disabled and instead we use a random uuid4() call.  But we leave the
+        # logic here so that we can later reactivate it, whhen the necessary
+        # url redirection code is written.
+        #notebook_id = unicode(uuid.uuid5(uuid.NAMESPACE_URL,
+        #                 'file://'+self.get_path_by_name(name).encode('utf-8')))
+        
+        notebook_id = unicode(uuid.uuid4())
+        self.mapping[notebook_id] = name
+        return notebook_id
+
+    def delete_notebook_id(self, notebook_id):
+        """Delete a notebook's id in the mapping.
+
+        This doesn't delete the actual notebook, only its entry in the mapping.
+        """
+        del self.mapping[notebook_id]
+
+    def notebook_exists(self, notebook_id):
+        """Does a notebook exist?"""
+        return notebook_id in self.mapping
+
+    def get_notebook(self, notebook_id, format=u'json'):
+        """Get the representation of a notebook in format by notebook_id."""
+        format = unicode(format)
+        if format not in self.allowed_formats:
+            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
+        last_modified, nb = self.read_notebook_object(notebook_id)
+        kwargs = {}
+        if format == 'json':
+            # don't split lines for sending over the wire, because it
+            # should match the Python in-memory format.
+            kwargs['split_lines'] = False
+        data = current.writes(nb, format, **kwargs)
+        name = nb.get('name','notebook')
+        return last_modified, name, data
+
+    def read_notebook_object(self, notebook_id):
+        """Get the object representation of a notebook by notebook_id."""
+        raise NotImplementedError('must be implemented in a subclass')
+
+    def save_new_notebook(self, data, name=None, format=u'json'):
+        """Save a new notebook and return its notebook_id.
+
+        If a name is passed in, it overrides any values in the notebook data
+        and the value in the data is updated to use that value.
+        """
+        if format not in self.allowed_formats:
+            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
+
+        try:
+            nb = current.reads(data.decode('utf-8'), format)
+        except:
+            raise web.HTTPError(400, u'Invalid JSON data')
+
+        if name is None:
+            try:
+                name = nb.metadata.name
+            except AttributeError:
+                raise web.HTTPError(400, u'Missing notebook name')
+        nb.metadata.name = name
+
+        notebook_id = self.write_notebook_object(nb)
+        return notebook_id
+
+    def save_notebook(self, notebook_id, data, name=None, format=u'json'):
+        """Save an existing notebook by notebook_id."""
+        if format not in self.allowed_formats:
+            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
+
+        try:
+            nb = current.reads(data.decode('utf-8'), format)
+        except:
+            raise web.HTTPError(400, u'Invalid JSON data')
+
+        if name is not None:
+            nb.metadata.name = name
+        self.write_notebook_object(nb, notebook_id)
+
+    def write_notebook_object(self, nb, notebook_id=None):
+        """Write a notebook object and return its notebook_id.
+
+        If notebook_id is None, this method should create a new notebook_id.
+        If notebook_id is not None, this method should check to make sure it
+        exists and is valid.
+        """
+        raise NotImplementedError('must be implemented in a subclass')
+
+    def delete_notebook(self, notebook_id):
+        """Delete notebook by notebook_id."""
+        raise NotImplementedError('must be implemented in a subclass')
+
+    def increment_filename(self, name):
+        """Increment a filename to make it unique.
+
+        This exists for notebook stores that must have unique names. When a notebook
+        is created or copied this method constructs a unique filename, typically
+        by appending an integer to the name.
+        """
+        return name
+
+    def new_notebook(self):
+        """Create a new notebook and return its notebook_id."""
+        name = self.increment_filename('Untitled')
+        metadata = current.new_metadata(name=name)
+        nb = current.new_notebook(metadata=metadata)
+        notebook_id = self.write_notebook_object(nb)
+        return notebook_id
+
+    def copy_notebook(self, notebook_id):
+        """Copy an existing notebook and return its notebook_id."""
+        last_mod, nb = self.read_notebook_object(notebook_id)
+        name = nb.metadata.name + '-Copy'
+        name = self.increment_filename(name)
+        nb.metadata.name = name
+        notebook_id = self.write_notebook_object(nb)
+        return notebook_id

IPython/frontend/html/notebook/filenbmanager.py

@@ -6,7 +6,7 b' Authors:'
 """
 
 #-----------------------------------------------------------------------------
-#  Copyright (C) 2008-2011  The IPython Development Team
+#  Copyright (C) 2011  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
@@ -19,20 +19,19 b' Authors:'
 import datetime
 import io
 import os
-import uuid
 import glob
 
 from tornado import web
 
-from IPython.config.configurable import LoggingConfigurable
+from .basenbmanager import BaseNotebookManager
 from IPython.nbformat import current
-from IPython.utils.traitlets import Unicode, List, Dict, Bool, TraitError
+from IPython.utils.traitlets import Unicode, Dict, Bool, TraitError
 
 #-----------------------------------------------------------------------------
 # Classes
 #-----------------------------------------------------------------------------
 
-class NotebookManager(LoggingConfigurable):
+class FileNotebookManager(BaseNotebookManager):
 
     notebook_dir = Unicode(os.getcwdu(), config=True, help="""
         The directory to use for notebooks.
@@ -59,24 +58,21 b' class NotebookManager(LoggingConfigurable):'
     )
     
     filename_ext = Unicode(u'.ipynb')
-    allowed_formats = List([u'json',u'py'])
 
-    # Map notebook_ids to notebook names
-    mapping = Dict()
     # Map notebook names to notebook_ids
     rev_mapping = Dict()
 
-    def list_notebooks(self):
-        """List all notebooks in the notebook dir.
-
-        This returns a list of dicts of the form::
-
-            dict(notebook_id=notebook,name=name)
-        """
+    def get_notebook_names(self):
+        """List all notebook names in the notebook dir."""
         names = glob.glob(os.path.join(self.notebook_dir,
                                        '*' + self.filename_ext))
         names = [os.path.splitext(os.path.basename(name))[0]
                  for name in names]
+        return names
+
+    def list_notebooks(self):
+        """List all notebooks in the notebook dir."""
+        names = self.get_notebook_names()
 
         data = []
         for name in names:
@@ -90,30 +86,20 b' class NotebookManager(LoggingConfigurable):'
 
     def new_notebook_id(self, name):
         """Generate a new notebook_id for a name and store its mappings."""
-        # TODO: the following will give stable urls for notebooks, but unless
-        # the notebooks are immediately redirected to their new urls when their
-        # filemname changes, nasty inconsistencies result.  So for now it's
-        # disabled and instead we use a random uuid4() call.  But we leave the
-        # logic here so that we can later reactivate it, whhen the necessary
-        # url redirection code is written.
-        #notebook_id = unicode(uuid.uuid5(uuid.NAMESPACE_URL,
-        #                 'file://'+self.get_path_by_name(name).encode('utf-8')))
-        
-        notebook_id = unicode(uuid.uuid4())
-        
-        self.mapping[notebook_id] = name
+        notebook_id = super(BaseNotebookManager, self).new_notebook_id(name)
         self.rev_mapping[name] = notebook_id
         return notebook_id
 
     def delete_notebook_id(self, notebook_id):
-        """Delete a notebook's id only. This doesn't delete the actual notebook."""
+        """Delete a notebook's id in the mapping."""
+        super(BaseNotebookManager, self).delete_notebook_id(notebook_id)
         name = self.mapping[notebook_id]
-        del self.mapping[notebook_id]
         del self.rev_mapping[name]
 
     def notebook_exists(self, notebook_id):
         """Does a notebook exist?"""
-        if notebook_id not in self.mapping:
+        exists = super(BaseNotebookManager, self).notebook_exists(notebook_id)
+        if not exists:
             return False
         path = self.get_path_by_name(self.mapping[notebook_id])
         return os.path.isfile(path)
@@ -132,22 +118,7 b' class NotebookManager(LoggingConfigurable):'
         path = os.path.join(self.notebook_dir, filename)
         return path       
 
-    def get_notebook(self, notebook_id, format=u'json'):
-        """Get the representation of a notebook in format by notebook_id."""
-        format = unicode(format)
-        if format not in self.allowed_formats:
-            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
-        last_modified, nb = self.get_notebook_object(notebook_id)
-        kwargs = {}
-        if format == 'json':
-            # don't split lines for sending over the wire, because it
-            # should match the Python in-memory format.
-            kwargs['split_lines'] = False
-        data = current.writes(nb, format, **kwargs)
-        name = nb.get('name','notebook')
-        return last_modified, name, data
-
-    def get_notebook_object(self, notebook_id):
+    def read_notebook_object(self, notebook_id):
         """Get the NotebookNode representation of a notebook by notebook_id."""
         path = self.find_path(notebook_id)
         if not os.path.isfile(path):
@@ -165,60 +136,27 b' class NotebookManager(LoggingConfigurable):'
         nb.metadata.name = os.path.splitext(os.path.basename(path))[0]
         return last_modified, nb
 
-    def save_new_notebook(self, data, name=None, format=u'json'):
-        """Save a new notebook and return its notebook_id.
-
-        If a name is passed in, it overrides any values in the notebook data
-        and the value in the data is updated to use that value.
-        """
-        if format not in self.allowed_formats:
-            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
-
-        try:
-            nb = current.reads(data.decode('utf-8'), format)
-        except:
-            raise web.HTTPError(400, u'Invalid JSON data')
-
-        if name is None:
-            try:
-                name = nb.metadata.name
-            except AttributeError:
-                raise web.HTTPError(400, u'Missing notebook name')
-        nb.metadata.name = name
-
-        notebook_id = self.new_notebook_id(name)
-        self.save_notebook_object(notebook_id, nb)
-        return notebook_id
-
-    def save_notebook(self, notebook_id, data, name=None, format=u'json'):
-        """Save an existing notebook by notebook_id."""
-        if format not in self.allowed_formats:
-            raise web.HTTPError(415, u'Invalid notebook format: %s' % format)
-
+    def write_notebook_object(self, nb, notebook_id=None):
+        """Save an existing notebook object by notebook_id."""
         try:
-            nb = current.reads(data.decode('utf-8'), format)
-        except:
-            raise web.HTTPError(400, u'Invalid JSON data')
+            new_name = nb.metadata.name
+        except AttributeError:
+            raise web.HTTPError(400, u'Missing notebook name')
 
-        if name is not None:
-            nb.metadata.name = name
-        self.save_notebook_object(notebook_id, nb)
+        if notebook_id is None:
+            notebook_id = self.new_notebook_id(new_name)
 
-    def save_notebook_object(self, notebook_id, nb):
-        """Save an existing notebook object by notebook_id."""
         if notebook_id not in self.mapping:
             raise web.HTTPError(404, u'Notebook does not exist: %s' % notebook_id)
+
         old_name = self.mapping[notebook_id]
-        try:
-            new_name = nb.metadata.name
-        except AttributeError:
-            raise web.HTTPError(400, u'Missing notebook name')
         path = self.get_path_by_name(new_name)
         try:
             with open(path,'w') as f:
                 current.write(nb, f, u'json')
         except Exception as e:
             raise web.HTTPError(400, u'Unexpected error while saving notebook: %s' % e)
+
         # save .py script as well
         if self.save_script:
             pypath = os.path.splitext(path)[0] + '.py'
@@ -228,6 +166,7 b' class NotebookManager(LoggingConfigurable):'
             except Exception as e:
                 raise web.HTTPError(400, u'Unexpected error while saving notebook as script: %s' % e)
         
+        # remove old files if the name changed
         if old_name != new_name:
             old_path = self.get_path_by_name(old_name)
             if os.path.isfile(old_path):
@@ -239,6 +178,8 b' class NotebookManager(LoggingConfigurable):'
             self.mapping[notebook_id] = new_name
             self.rev_mapping[new_name] = notebook_id
             del self.rev_mapping[old_name]
+        
+        return notebook_id
 
     def delete_notebook(self, notebook_id):
         """Delete notebook by notebook_id."""
@@ -263,24 +204,4 b' class NotebookManager(LoggingConfigurable):'
                 break
             else:
                 i = i+1
-        return path, name
-
-    def new_notebook(self):
-        """Create a new notebook and return its notebook_id."""
-        path, name = self.increment_filename('Untitled')
-        notebook_id = self.new_notebook_id(name)
-        metadata = current.new_metadata(name=name)
-        nb = current.new_notebook(metadata=metadata)
-        with open(path,'w') as f:
-            current.write(nb, f, u'json')
-        return notebook_id
-
-    def copy_notebook(self, notebook_id):
-        """Copy an existing notebook and return its notebook_id."""
-        last_mod, nb = self.get_notebook_object(notebook_id)
-        name = nb.metadata.name + '-Copy'
-        path, name = self.increment_filename(name)
-        nb.metadata.name = name
-        notebook_id = self.new_notebook_id(name)
-        self.save_notebook_object(notebook_id, nb)
-        return notebook_id
+        return name

IPython/frontend/html/notebook/notebookapp.py

@@ -66,7 +66,11 b' from IPython.zmq.ipkernel import ('
     aliases as ipkernel_aliases,
     IPKernelApp
 )
-from IPython.utils.traitlets import Dict, Unicode, Integer, List, Enum, Bool
+from IPython.utils.importstring import import_item
+from IPython.utils.traitlets import (
+    Dict, Unicode, Integer, List, Enum, Bool,
+    DottedObjectName
+)
 from IPython.utils import py3compat
 from IPython.utils.path import filefind
 
@@ -404,6 +408,10 b' class NotebookApp(BaseIPythonApplication):'
         else:
             self.log.info("Using MathJax: %s", new)
 
+    notebook_manager_class = DottedObjectName('IPython.frontend.html.notebook.notebookmanager.NotebookManager',
+        config=True,
+        help='The notebook manager class to use.')
+
     def parse_command_line(self, argv=None):
         super(NotebookApp, self).parse_command_line(argv)
         if argv is None:
@@ -430,9 +438,10 b' class NotebookApp(BaseIPythonApplication):'
             config=self.config, log=self.log, kernel_argv=self.kernel_argv,
             connection_dir = self.profile_dir.security_dir,
         )
-        self.notebook_manager = NotebookManager(config=self.config, log=self.log)
+        kls = import_item(self.notebook_manager_class)
+        self.notebook_manager = kls(config=self.config, log=self.log)
         self.log.info("Serving notebooks from %s", self.notebook_manager.notebook_dir)
-        self.notebook_manager.list_notebooks()
+        self.notebook_manager.load_notebook_names()
         self.cluster_manager = ClusterManager(config=self.config, log=self.log)
         self.cluster_manager.update_profiles()
 

IPython/frontend/html/notebook/tests/test_nbmanager.py

@@ -7,28 +7,28 b' from tempfile import NamedTemporaryFile'
 from IPython.utils.tempdir import TemporaryDirectory
 from IPython.utils.traitlets import TraitError
 
-from IPython.frontend.html.notebook.notebookmanager import NotebookManager
+from IPython.frontend.html.notebook.filenbmanager import FileNotebookManager
 
 class TestNotebookManager(TestCase):
 
     def test_nb_dir(self):
         with TemporaryDirectory() as td:
-            km = NotebookManager(notebook_dir=td)
-            self.assertEqual(km.notebook_dir, td)
+            km = FileNotebookManager(notebook_dir=td)
+            self.assertEquals(km.notebook_dir, td)
 
     def test_create_nb_dir(self):
         with TemporaryDirectory() as td:
             nbdir = os.path.join(td, 'notebooks')
-            km = NotebookManager(notebook_dir=nbdir)
-            self.assertEqual(km.notebook_dir, nbdir)
+            km = FileNotebookManager(notebook_dir=nbdir)
+            self.assertEquals(km.notebook_dir, nbdir)
 
     def test_missing_nb_dir(self):
         with TemporaryDirectory() as td:
             nbdir = os.path.join(td, 'notebook', 'dir', 'is', 'missing')
-            self.assertRaises(TraitError, NotebookManager, notebook_dir=nbdir)
+            self.assertRaises(TraitError, FileNotebookManager, notebook_dir=nbdir)
 
     def test_invalid_nb_dir(self):
         with NamedTemporaryFile() as tf:
-            self.assertRaises(TraitError, NotebookManager, notebook_dir=tf.name)
+            self.assertRaises(TraitError, FileNotebookManager, notebook_dir=tf.name)