notebook/IPython/html/services/contents/filemanager.py

528 lines
18 KiB
Python
Raw Normal View History

"""A contents manager that uses the local file system for storage."""
# Copyright (c) IPython Development Team.
# Distributed under the terms of the Modified BSD License.
import base64
import io
import os
2011-08-19 10:31:16 +08:00
import glob
2013-04-10 04:49:10 +08:00
import shutil
2013-07-04 01:35:30 +08:00
from tornado import web
from .manager import ContentsManager
from IPython import nbformat
from IPython.utils.io import atomic_writing
from IPython.utils.path import ensure_dir_exists
from IPython.utils.traitlets import Unicode, Bool, TraitError
from IPython.utils.py3compat import getcwd
from IPython.utils import tz
from IPython.html.utils import is_hidden, to_os_path, url_path_join
class FileContentsManager(ContentsManager):
root_dir = Unicode(getcwd(), config=True)
save_script = Bool(False, config=True, help='DEPRECATED, IGNORED')
def _save_script_changed(self):
self.log.warn("""
Automatically saving notebooks as scripts has been removed.
Use `ipython nbconvert --to python [notebook]` instead.
""")
def _root_dir_changed(self, name, old, new):
"""Do a bit of validation of the root_dir."""
if not os.path.isabs(new):
# If we receive a non-absolute path, make it absolute.
self.root_dir = os.path.abspath(new)
return
if not os.path.isdir(new):
raise TraitError("%r is not a directory" % new)
checkpoint_dir = Unicode('.ipynb_checkpoints', config=True,
help="""The directory name in which to keep file checkpoints
This is a path relative to the file's own directory.
By default, it is .ipynb_checkpoints
2013-04-10 04:49:10 +08:00
"""
)
def _copy(self, src, dest):
"""copy src to dest
2014-03-20 03:21:55 +08:00
like shutil.copy2, but log errors in copystat
"""
2014-03-20 03:21:55 +08:00
shutil.copyfile(src, dest)
try:
shutil.copystat(src, dest)
except OSError as e:
self.log.debug("copystat on %s failed", dest, exc_info=True)
def _get_os_path(self, path=''):
"""Given a filename and API path, return its file system
path.
Parameters
----------
path : string
The relative API path to the named file.
Returns
-------
path : string
Native, absolute OS path to for a file.
"""
return to_os_path(path, self.root_dir)
def dir_exists(self, path):
"""Does the API-style path refer to an extant directory?
API-style wrapper for os.path.isdir
Parameters
----------
path : string
The path to check. This is an API path (`/` separated,
relative to root_dir).
Returns
-------
exists : bool
Whether the path is indeed a directory.
"""
path = path.strip('/')
os_path = self._get_os_path(path=path)
return os.path.isdir(os_path)
def is_hidden(self, path):
"""Does the API style path correspond to a hidden directory or file?
Parameters
----------
path : string
The path to check. This is an API path (`/` separated,
relative to root_dir).
Returns
-------
hidden : bool
Whether the path exists and is hidden.
"""
path = path.strip('/')
os_path = self._get_os_path(path=path)
return is_hidden(os_path, self.root_dir)
def file_exists(self, path):
"""Returns True if the file exists, else returns False.
API-style wrapper for os.path.isfile
Parameters
----------
path : string
The relative path to the file (with '/' as separator)
Returns
-------
exists : bool
Whether the file exists.
"""
path = path.strip('/')
nbpath = self._get_os_path(path)
return os.path.isfile(nbpath)
def exists(self, path):
"""Returns True if the path exists, else returns False.
API-style wrapper for os.path.exists
Parameters
----------
path : string
The API path to the file (with '/' as separator)
Returns
-------
exists : bool
Whether the target exists.
"""
path = path.strip('/')
os_path = self._get_os_path(path=path)
return os.path.exists(os_path)
2013-07-04 01:35:30 +08:00
def _base_model(self, path):
"""Build the common base of a contents model"""
os_path = self._get_os_path(path)
2014-02-03 07:50:57 +08:00
info = os.stat(os_path)
last_modified = tz.utcfromtimestamp(info.st_mtime)
created = tz.utcfromtimestamp(info.st_ctime)
# Create the base model.
model = {}
model['name'] = path.rsplit('/', 1)[-1]
2014-02-03 07:50:57 +08:00
model['path'] = path
model['last_modified'] = last_modified
model['created'] = created
model['content'] = None
model['format'] = None
model['message'] = None
return model
def _dir_model(self, path, content=True):
"""Build a model for a directory
if content is requested, will include a listing of the directory
"""
os_path = self._get_os_path(path)
four_o_four = u'directory does not exist: %r' % os_path
if not os.path.isdir(os_path):
raise web.HTTPError(404, four_o_four)
elif is_hidden(os_path, self.root_dir):
self.log.info("Refusing to serve hidden directory %r, via 404 Error",
os_path
)
raise web.HTTPError(404, four_o_four)
model = self._base_model(path)
2014-02-03 07:50:57 +08:00
model['type'] = 'directory'
if content:
model['content'] = contents = []
for os_path in glob.glob(self._get_os_path('%s/*' % path)):
name = os.path.basename(os_path)
# skip over broken symlinks in listing
if not os.path.exists(os_path):
self.log.warn("%s doesn't exist", os_path)
continue
if self.should_list(name) and not is_hidden(os_path, self.root_dir):
contents.append(self.get_model(
path='%s/%s' % (path, name),
content=False)
)
model['format'] = 'json'
2014-02-03 07:50:57 +08:00
return model
def _file_model(self, path, content=True):
"""Build a model for a file
if content is requested, include the file contents.
UTF-8 text files will be unicode, binary files will be base64-encoded.
"""
model = self._base_model(path)
model['type'] = 'file'
if content:
os_path = self._get_os_path(path)
with io.open(os_path, 'rb') as f:
bcontent = f.read()
try:
model['content'] = bcontent.decode('utf8')
except UnicodeError as e:
model['content'] = base64.encodestring(bcontent).decode('ascii')
model['format'] = 'base64'
else:
model['format'] = 'text'
return model
def _notebook_model(self, path, content=True):
"""Build a notebook model
if content is requested, the notebook content will be populated
as a JSON structure (not double-serialized)
"""
model = self._base_model(path)
model['type'] = 'notebook'
if content:
os_path = self._get_os_path(path)
with io.open(os_path, 'r', encoding='utf-8') as f:
try:
nb = nbformat.read(f, as_version=4)
except Exception as e:
2014-09-11 05:22:01 +08:00
raise web.HTTPError(400, u"Unreadable Notebook: %s %r" % (os_path, e))
self.mark_trusted_cells(nb, path)
model['content'] = nb
model['format'] = 'json'
self.validate_notebook_model(model)
return model
def get_model(self, path, content=True):
""" Takes a path for an entity and returns its model
Parameters
----------
path : str
the API path that describes the relative path for the target
Returns
-------
model : dict
the contents model. If content=True, returns the contents
of the file or directory as well.
"""
path = path.strip('/')
if not self.exists(path):
raise web.HTTPError(404, u'No such file or directory: %s' % path)
os_path = self._get_os_path(path)
if os.path.isdir(os_path):
model = self._dir_model(path, content=content)
elif path.endswith('.ipynb'):
model = self._notebook_model(path, content=content)
else:
model = self._file_model(path, content=content)
return model
def _save_notebook(self, os_path, model, path=''):
"""save a notebook file"""
# Save the notebook file
nb = nbformat.from_dict(model['content'])
self.check_and_sign(nb, path)
with atomic_writing(os_path, encoding='utf-8') as f:
nbformat.write(nb, f, version=nbformat.NO_CONVERT)
def _save_file(self, os_path, model, path=''):
"""save a non-notebook file"""
fmt = model.get('format', None)
if fmt not in {'text', 'base64'}:
raise web.HTTPError(400, "Must specify format of file contents as 'text' or 'base64'")
try:
content = model['content']
if fmt == 'text':
bcontent = content.encode('utf8')
else:
b64_bytes = content.encode('ascii')
bcontent = base64.decodestring(b64_bytes)
except Exception as e:
raise web.HTTPError(400, u'Encoding error saving %s: %s' % (os_path, e))
with atomic_writing(os_path, text=False) as f:
f.write(bcontent)
def _save_directory(self, os_path, model, path=''):
"""create a directory"""
if is_hidden(os_path, self.root_dir):
raise web.HTTPError(400, u'Cannot create hidden directory %r' % os_path)
if not os.path.exists(os_path):
os.mkdir(os_path)
elif not os.path.isdir(os_path):
raise web.HTTPError(400, u'Not a directory: %s' % (os_path))
else:
self.log.debug("Directory %r already exists", os_path)
def save(self, model, path=''):
"""Save the file model and return the model with no content."""
path = path.strip('/')
if 'type' not in model:
raise web.HTTPError(400, u'No file type provided')
if 'content' not in model and model['type'] != 'directory':
raise web.HTTPError(400, u'No file content provided')
# One checkpoint should always exist
if self.file_exists(path) and not self.list_checkpoints(path):
self.create_checkpoint(path)
os_path = self._get_os_path(path)
self.log.debug("Saving %s", os_path)
try:
if model['type'] == 'notebook':
self._save_notebook(os_path, model, path)
elif model['type'] == 'file':
self._save_file(os_path, model, path)
elif model['type'] == 'directory':
self._save_directory(os_path, model, path)
else:
raise web.HTTPError(400, "Unhandled contents type: %s" % model['type'])
except web.HTTPError:
raise
except Exception as e:
raise web.HTTPError(400, u'Unexpected error while saving file: %s %s' % (os_path, e))
validation_message = None
if model['type'] == 'notebook':
self.validate_notebook_model(model)
validation_message = model.get('message', None)
model = self.get_model(path, content=False)
if validation_message:
model['message'] = validation_message
return model
def update(self, model, path):
"""Update the file's path
For use in PATCH requests, to enable renaming a file without
re-uploading its contents. Only used for renaming at the moment.
"""
path = path.strip('/')
new_path = model.get('path', path).strip('/')
if path != new_path:
self.rename(path, new_path)
model = self.get_model(new_path, content=False)
return model
def delete(self, path):
"""Delete file at path."""
path = path.strip('/')
os_path = self._get_os_path(path)
rm = os.unlink
if os.path.isdir(os_path):
listing = os.listdir(os_path)
# don't delete non-empty directories (checkpoints dir doesn't count)
if listing and listing != [self.checkpoint_dir]:
raise web.HTTPError(400, u'Directory %s not empty' % os_path)
elif not os.path.isfile(os_path):
raise web.HTTPError(404, u'File does not exist: %s' % os_path)
2013-04-10 04:49:10 +08:00
# clear checkpoints
for checkpoint in self.list_checkpoints(path):
checkpoint_id = checkpoint['id']
cp_path = self.get_checkpoint_path(checkpoint_id, path)
if os.path.isfile(cp_path):
self.log.debug("Unlinking checkpoint %s", cp_path)
os.unlink(cp_path)
if os.path.isdir(os_path):
self.log.debug("Removing directory %s", os_path)
shutil.rmtree(os_path)
else:
self.log.debug("Unlinking file %s", os_path)
rm(os_path)
def rename(self, old_path, new_path):
"""Rename a file."""
old_path = old_path.strip('/')
new_path = new_path.strip('/')
if new_path == old_path:
return
new_os_path = self._get_os_path(new_path)
old_os_path = self._get_os_path(old_path)
# Should we proceed with the move?
if os.path.isfile(new_os_path):
raise web.HTTPError(409, u'File already exists: %s' % new_os_path)
# Move the file
try:
shutil.move(old_os_path, new_os_path)
except Exception as e:
raise web.HTTPError(500, u'Unknown error renaming file: %s %s' % (old_os_path, e))
# Move the checkpoints
old_checkpoints = self.list_checkpoints(old_path)
for cp in old_checkpoints:
checkpoint_id = cp['id']
old_cp_path = self.get_checkpoint_path(checkpoint_id, old_path)
new_cp_path = self.get_checkpoint_path(checkpoint_id, new_path)
if os.path.isfile(old_cp_path):
self.log.debug("Renaming checkpoint %s -> %s", old_cp_path, new_cp_path)
shutil.move(old_cp_path, new_cp_path)
2013-04-10 04:49:10 +08:00
# Checkpoint-related utilities
def get_checkpoint_path(self, checkpoint_id, path):
"""find the path to a checkpoint"""
path = path.strip('/')
parent, name = ('/' + path).rsplit('/', 1)
parent = parent.strip('/')
basename, ext = os.path.splitext(name)
filename = u"{name}-{checkpoint_id}{ext}".format(
2013-10-24 01:37:08 +08:00
name=basename,
2013-04-10 04:49:10 +08:00
checkpoint_id=checkpoint_id,
ext=ext,
2013-04-10 04:49:10 +08:00
)
os_path = self._get_os_path(path=parent)
cp_dir = os.path.join(os_path, self.checkpoint_dir)
ensure_dir_exists(cp_dir)
cp_path = os.path.join(cp_dir, filename)
return cp_path
def get_checkpoint_model(self, checkpoint_id, path):
"""construct the info dict for a given checkpoint"""
path = path.strip('/')
cp_path = self.get_checkpoint_path(checkpoint_id, path)
stats = os.stat(cp_path)
last_modified = tz.utcfromtimestamp(stats.st_mtime)
info = dict(
id = checkpoint_id,
last_modified = last_modified,
)
return info
2013-04-10 04:49:10 +08:00
# public checkpoint API
def create_checkpoint(self, path):
"""Create a checkpoint from the current state of a file"""
path = path.strip('/')
src_path = self._get_os_path(path)
# only the one checkpoint ID:
checkpoint_id = u"checkpoint"
cp_path = self.get_checkpoint_path(checkpoint_id, path)
self.log.debug("creating checkpoint for %s", path)
self._copy(src_path, cp_path)
# return the checkpoint info
return self.get_checkpoint_model(checkpoint_id, path)
def list_checkpoints(self, path):
"""list the checkpoints for a given file
This contents manager currently only supports one checkpoint per file.
2013-04-10 04:49:10 +08:00
"""
path = path.strip('/')
2013-07-04 01:35:30 +08:00
checkpoint_id = "checkpoint"
os_path = self.get_checkpoint_path(checkpoint_id, path)
if not os.path.exists(os_path):
2013-04-10 04:49:10 +08:00
return []
else:
return [self.get_checkpoint_model(checkpoint_id, path)]
def restore_checkpoint(self, checkpoint_id, path):
"""restore a file to a checkpointed state"""
path = path.strip('/')
self.log.info("restoring %s from checkpoint %s", path, checkpoint_id)
nb_path = self._get_os_path(path)
cp_path = self.get_checkpoint_path(checkpoint_id, path)
2013-04-10 04:49:10 +08:00
if not os.path.isfile(cp_path):
self.log.debug("checkpoint file does not exist: %s", cp_path)
2013-04-10 04:49:10 +08:00
raise web.HTTPError(404,
u'checkpoint does not exist: %s@%s' % (path, checkpoint_id)
2013-04-10 04:49:10 +08:00
)
# ensure notebook is readable (never restore from an unreadable notebook)
if cp_path.endswith('.ipynb'):
with io.open(cp_path, 'r', encoding='utf-8') as f:
nbformat.read(f, as_version=4)
self._copy(cp_path, nb_path)
2013-04-10 04:49:10 +08:00
self.log.debug("copying %s -> %s", cp_path, nb_path)
def delete_checkpoint(self, checkpoint_id, path):
"""delete a file's checkpoint"""
path = path.strip('/')
cp_path = self.get_checkpoint_path(checkpoint_id, path)
if not os.path.isfile(cp_path):
2013-04-10 04:49:10 +08:00
raise web.HTTPError(404,
u'Checkpoint does not exist: %s@%s' % (path, checkpoint_id)
2013-04-10 04:49:10 +08:00
)
self.log.debug("unlinking %s", cp_path)
os.unlink(cp_path)
def info_string(self):
return "Serving notebooks from local directory: %s" % self.root_dir
def get_kernel_path(self, path, model=None):
"""Return the initial working dir a kernel associated with a given notebook"""
if '/' in path:
os_dir = path.rsplit('/', 1)[0]
else:
os_dir = ''
return self._get_os_path(os_dir)