Show More
@@ -1,310 +1,312 b'' | |||
|
1 | 1 | """Functions for signing notebooks""" |
|
2 | 2 | #----------------------------------------------------------------------------- |
|
3 | 3 | # Copyright (C) 2014, The IPython Development Team |
|
4 | 4 | # |
|
5 | 5 | # Distributed under the terms of the BSD License. The full license is in |
|
6 | 6 | # the file COPYING, distributed as part of this software. |
|
7 | 7 | #----------------------------------------------------------------------------- |
|
8 | 8 | |
|
9 | 9 | #----------------------------------------------------------------------------- |
|
10 | 10 | # Imports |
|
11 | 11 | #----------------------------------------------------------------------------- |
|
12 | 12 | |
|
13 | 13 | import base64 |
|
14 | 14 | from contextlib import contextmanager |
|
15 | 15 | import hashlib |
|
16 | 16 | from hmac import HMAC |
|
17 | 17 | import io |
|
18 | 18 | import os |
|
19 | 19 | |
|
20 | 20 | from IPython.utils.py3compat import string_types, unicode_type, cast_bytes |
|
21 | 21 | from IPython.utils.traitlets import Instance, Bytes, Enum, Any, Unicode, Bool |
|
22 | 22 | from IPython.config import LoggingConfigurable, MultipleInstanceError |
|
23 | 23 | from IPython.core.application import BaseIPythonApplication, base_flags |
|
24 | 24 | |
|
25 | 25 | from .current import read, write |
|
26 | 26 | |
|
27 | 27 | #----------------------------------------------------------------------------- |
|
28 | 28 | # Code |
|
29 | 29 | #----------------------------------------------------------------------------- |
|
30 | 30 | try: |
|
31 | 31 | # Python 3 |
|
32 | 32 | algorithms = hashlib.algorithms_guaranteed |
|
33 | 33 | except AttributeError: |
|
34 | 34 | algorithms = hashlib.algorithms |
|
35 | 35 | |
|
36 | 36 | def yield_everything(obj): |
|
37 | 37 | """Yield every item in a container as bytes |
|
38 | 38 | |
|
39 | 39 | Allows any JSONable object to be passed to an HMAC digester |
|
40 | 40 | without having to serialize the whole thing. |
|
41 | 41 | """ |
|
42 | 42 | if isinstance(obj, dict): |
|
43 | 43 | for key in sorted(obj): |
|
44 | 44 | value = obj[key] |
|
45 | 45 | yield cast_bytes(key) |
|
46 | 46 | for b in yield_everything(value): |
|
47 | 47 | yield b |
|
48 | 48 | elif isinstance(obj, (list, tuple)): |
|
49 | 49 | for element in obj: |
|
50 | 50 | for b in yield_everything(element): |
|
51 | 51 | yield b |
|
52 | 52 | elif isinstance(obj, unicode_type): |
|
53 | 53 | yield obj.encode('utf8') |
|
54 | 54 | else: |
|
55 | 55 | yield unicode_type(obj).encode('utf8') |
|
56 | 56 | |
|
57 | 57 | |
|
58 | 58 | @contextmanager |
|
59 | 59 | def signature_removed(nb): |
|
60 | 60 | """Context manager for operating on a notebook with its signature removed |
|
61 | 61 | |
|
62 | 62 | Used for excluding the previous signature when computing a notebook's signature. |
|
63 | 63 | """ |
|
64 | 64 | save_signature = nb['metadata'].pop('signature', None) |
|
65 | 65 | try: |
|
66 | 66 | yield |
|
67 | 67 | finally: |
|
68 | 68 | if save_signature is not None: |
|
69 | 69 | nb['metadata']['signature'] = save_signature |
|
70 | 70 | |
|
71 | 71 | |
|
72 | 72 | class NotebookNotary(LoggingConfigurable): |
|
73 | 73 | """A class for computing and verifying notebook signatures.""" |
|
74 | 74 | |
|
75 | 75 | profile_dir = Instance("IPython.core.profiledir.ProfileDir") |
|
76 | 76 | def _profile_dir_default(self): |
|
77 | 77 | from IPython.core.application import BaseIPythonApplication |
|
78 | 78 | app = None |
|
79 | 79 | try: |
|
80 | 80 | if BaseIPythonApplication.initialized(): |
|
81 | 81 | app = BaseIPythonApplication.instance() |
|
82 | 82 | except MultipleInstanceError: |
|
83 | 83 | pass |
|
84 | 84 | if app is None: |
|
85 | 85 | # create an app, without the global instance |
|
86 | 86 | app = BaseIPythonApplication() |
|
87 | 87 | app.initialize(argv=[]) |
|
88 | 88 | return app.profile_dir |
|
89 | 89 | |
|
90 | 90 | algorithm = Enum(algorithms, default_value='sha256', config=True, |
|
91 | 91 | help="""The hashing algorithm used to sign notebooks.""" |
|
92 | 92 | ) |
|
93 | 93 | def _algorithm_changed(self, name, old, new): |
|
94 | 94 | self.digestmod = getattr(hashlib, self.algorithm) |
|
95 | 95 | |
|
96 | 96 | digestmod = Any() |
|
97 | 97 | def _digestmod_default(self): |
|
98 | 98 | return getattr(hashlib, self.algorithm) |
|
99 | 99 | |
|
100 |
secret_file = Unicode( |
|
|
100 | secret_file = Unicode(config=True, | |
|
101 | help="""The file where the secret key is stored.""" | |
|
102 | ) | |
|
101 | 103 | def _secret_file_default(self): |
|
102 | 104 | if self.profile_dir is None: |
|
103 | 105 | return '' |
|
104 | 106 | return os.path.join(self.profile_dir.security_dir, 'notebook_secret') |
|
105 | 107 | |
|
106 | 108 | secret = Bytes(config=True, |
|
107 | 109 | help="""The secret key with which notebooks are signed.""" |
|
108 | 110 | ) |
|
109 | 111 | def _secret_default(self): |
|
110 | 112 | # note : this assumes an Application is running |
|
111 | 113 | if os.path.exists(self.secret_file): |
|
112 | 114 | with io.open(self.secret_file, 'rb') as f: |
|
113 | 115 | return f.read() |
|
114 | 116 | else: |
|
115 | 117 | secret = base64.encodestring(os.urandom(1024)) |
|
116 | 118 | self._write_secret_file(secret) |
|
117 | 119 | return secret |
|
118 | 120 | |
|
119 | 121 | def _write_secret_file(self, secret): |
|
120 | 122 | """write my secret to my secret_file""" |
|
121 | 123 | self.log.info("Writing notebook-signing key to %s", self.secret_file) |
|
122 | 124 | with io.open(self.secret_file, 'wb') as f: |
|
123 | 125 | f.write(secret) |
|
124 | 126 | try: |
|
125 | 127 | os.chmod(self.secret_file, 0o600) |
|
126 | 128 | except OSError: |
|
127 | 129 | self.log.warn( |
|
128 | 130 | "Could not set permissions on %s", |
|
129 | 131 | self.secret_file |
|
130 | 132 | ) |
|
131 | 133 | return secret |
|
132 | 134 | |
|
133 | 135 | def compute_signature(self, nb): |
|
134 | 136 | """Compute a notebook's signature |
|
135 | 137 | |
|
136 | 138 | by hashing the entire contents of the notebook via HMAC digest. |
|
137 | 139 | """ |
|
138 | 140 | hmac = HMAC(self.secret, digestmod=self.digestmod) |
|
139 | 141 | # don't include the previous hash in the content to hash |
|
140 | 142 | with signature_removed(nb): |
|
141 | 143 | # sign the whole thing |
|
142 | 144 | for b in yield_everything(nb): |
|
143 | 145 | hmac.update(b) |
|
144 | 146 | |
|
145 | 147 | return hmac.hexdigest() |
|
146 | 148 | |
|
147 | 149 | def check_signature(self, nb): |
|
148 | 150 | """Check a notebook's stored signature |
|
149 | 151 | |
|
150 | 152 | If a signature is stored in the notebook's metadata, |
|
151 | 153 | a new signature is computed and compared with the stored value. |
|
152 | 154 | |
|
153 | 155 | Returns True if the signature is found and matches, False otherwise. |
|
154 | 156 | |
|
155 | 157 | The following conditions must all be met for a notebook to be trusted: |
|
156 | 158 | - a signature is stored in the form 'scheme:hexdigest' |
|
157 | 159 | - the stored scheme matches the requested scheme |
|
158 | 160 | - the requested scheme is available from hashlib |
|
159 | 161 | - the computed hash from notebook_signature matches the stored hash |
|
160 | 162 | """ |
|
161 | 163 | stored_signature = nb['metadata'].get('signature', None) |
|
162 | 164 | if not stored_signature \ |
|
163 | 165 | or not isinstance(stored_signature, string_types) \ |
|
164 | 166 | or ':' not in stored_signature: |
|
165 | 167 | return False |
|
166 | 168 | stored_algo, sig = stored_signature.split(':', 1) |
|
167 | 169 | if self.algorithm != stored_algo: |
|
168 | 170 | return False |
|
169 | 171 | my_signature = self.compute_signature(nb) |
|
170 | 172 | return my_signature == sig |
|
171 | 173 | |
|
172 | 174 | def sign(self, nb): |
|
173 | 175 | """Sign a notebook, indicating that its output is trusted |
|
174 | 176 | |
|
175 | 177 | stores 'algo:hmac-hexdigest' in notebook.metadata.signature |
|
176 | 178 | |
|
177 | 179 | e.g. 'sha256:deadbeef123...' |
|
178 | 180 | """ |
|
179 | 181 | signature = self.compute_signature(nb) |
|
180 | 182 | nb['metadata']['signature'] = "%s:%s" % (self.algorithm, signature) |
|
181 | 183 | |
|
182 | 184 | def mark_cells(self, nb, trusted): |
|
183 | 185 | """Mark cells as trusted if the notebook's signature can be verified |
|
184 | 186 | |
|
185 | 187 | Sets ``cell.trusted = True | False`` on all code cells, |
|
186 | 188 | depending on whether the stored signature can be verified. |
|
187 | 189 | |
|
188 | 190 | This function is the inverse of check_cells |
|
189 | 191 | """ |
|
190 | 192 | if not nb['worksheets']: |
|
191 | 193 | # nothing to mark if there are no cells |
|
192 | 194 | return |
|
193 | 195 | for cell in nb['worksheets'][0]['cells']: |
|
194 | 196 | if cell['cell_type'] == 'code': |
|
195 | 197 | cell['trusted'] = trusted |
|
196 | 198 | |
|
197 | 199 | def _check_cell(self, cell): |
|
198 | 200 | """Do we trust an individual cell? |
|
199 | 201 | |
|
200 | 202 | Return True if: |
|
201 | 203 | |
|
202 | 204 | - cell is explicitly trusted |
|
203 | 205 | - cell has no potentially unsafe rich output |
|
204 | 206 | |
|
205 | 207 | If a cell has no output, or only simple print statements, |
|
206 | 208 | it will always be trusted. |
|
207 | 209 | """ |
|
208 | 210 | # explicitly trusted |
|
209 | 211 | if cell.pop("trusted", False): |
|
210 | 212 | return True |
|
211 | 213 | |
|
212 | 214 | # explicitly safe output |
|
213 | 215 | safe = { |
|
214 | 216 | 'text/plain', 'image/png', 'image/jpeg', |
|
215 | 217 | 'text', 'png', 'jpg', # v3-style short keys |
|
216 | 218 | } |
|
217 | 219 | |
|
218 | 220 | for output in cell['outputs']: |
|
219 | 221 | output_type = output['output_type'] |
|
220 | 222 | if output_type in ('pyout', 'display_data'): |
|
221 | 223 | # if there are any data keys not in the safe whitelist |
|
222 | 224 | output_keys = set(output).difference({"output_type", "prompt_number", "metadata"}) |
|
223 | 225 | if output_keys.difference(safe): |
|
224 | 226 | return False |
|
225 | 227 | |
|
226 | 228 | return True |
|
227 | 229 | |
|
228 | 230 | def check_cells(self, nb): |
|
229 | 231 | """Return whether all code cells are trusted |
|
230 | 232 | |
|
231 | 233 | If there are no code cells, return True. |
|
232 | 234 | |
|
233 | 235 | This function is the inverse of mark_cells. |
|
234 | 236 | """ |
|
235 | 237 | if not nb['worksheets']: |
|
236 | 238 | return True |
|
237 | 239 | trusted = True |
|
238 | 240 | for cell in nb['worksheets'][0]['cells']: |
|
239 | 241 | if cell['cell_type'] != 'code': |
|
240 | 242 | continue |
|
241 | 243 | # only distrust a cell if it actually has some output to distrust |
|
242 | 244 | if not self._check_cell(cell): |
|
243 | 245 | trusted = False |
|
244 | 246 | return trusted |
|
245 | 247 | |
|
246 | 248 | |
|
247 | 249 | trust_flags = { |
|
248 | 250 | 'reset' : ( |
|
249 | 251 | {'TrustNotebookApp' : { 'reset' : True}}, |
|
250 | 252 | """Generate a new key for notebook signature. |
|
251 | 253 | All previously signed notebooks will become untrusted. |
|
252 | 254 | """ |
|
253 | 255 | ), |
|
254 | 256 | } |
|
255 | 257 | trust_flags.update(base_flags) |
|
256 | 258 | trust_flags.pop('init') |
|
257 | 259 | |
|
258 | 260 | |
|
259 | 261 | class TrustNotebookApp(BaseIPythonApplication): |
|
260 | 262 | |
|
261 | 263 | description="""Sign one or more IPython notebooks with your key, |
|
262 | 264 | to trust their dynamic (HTML, Javascript) output. |
|
263 | 265 | |
|
264 | 266 | Otherwise, you will have to re-execute the notebook to see output. |
|
265 | 267 | """ |
|
266 | 268 | |
|
267 | 269 | examples = """ipython trust mynotebook.ipynb and_this_one.ipynb""" |
|
268 | 270 | |
|
269 | 271 | flags = trust_flags |
|
270 | 272 | |
|
271 | 273 | reset = Bool(False, config=True, |
|
272 | 274 | help="""If True, generate a new key for notebook signature. |
|
273 | 275 | After reset, all previously signed notebooks will become untrusted. |
|
274 | 276 | """ |
|
275 | 277 | ) |
|
276 | 278 | |
|
277 | 279 | notary = Instance(NotebookNotary) |
|
278 | 280 | def _notary_default(self): |
|
279 | 281 | return NotebookNotary(parent=self, profile_dir=self.profile_dir) |
|
280 | 282 | |
|
281 | 283 | def sign_notebook(self, notebook_path): |
|
282 | 284 | if not os.path.exists(notebook_path): |
|
283 | 285 | self.log.error("Notebook missing: %s" % notebook_path) |
|
284 | 286 | self.exit(1) |
|
285 | 287 | with io.open(notebook_path, encoding='utf8') as f: |
|
286 | 288 | nb = read(f, 'json') |
|
287 | 289 | if self.notary.check_signature(nb): |
|
288 | 290 | print("Notebook already signed: %s" % notebook_path) |
|
289 | 291 | else: |
|
290 | 292 | print("Signing notebook: %s" % notebook_path) |
|
291 | 293 | self.notary.sign(nb) |
|
292 | 294 | with io.open(notebook_path, 'w', encoding='utf8') as f: |
|
293 | 295 | write(nb, f, 'json') |
|
294 | 296 | |
|
295 | 297 | def generate_new_key(self): |
|
296 | 298 | """Generate a new notebook signature key""" |
|
297 | 299 | print("Generating new notebook key: %s" % self.notary.secret_file) |
|
298 | 300 | self.notary._write_secret_file(os.urandom(1024)) |
|
299 | 301 | |
|
300 | 302 | def start(self): |
|
301 | 303 | if self.reset: |
|
302 | 304 | self.generate_new_key() |
|
303 | 305 | return |
|
304 | 306 | if not self.extra_args: |
|
305 | 307 | self.log.critical("Specify at least one notebook to sign.") |
|
306 | 308 | self.exit(1) |
|
307 | 309 | |
|
308 | 310 | for notebook_path in self.extra_args: |
|
309 | 311 | self.sign_notebook(notebook_path) |
|
310 | 312 |
General Comments 0
You need to be logged in to leave comments.
Login now