Show More
@@ -1,222 +1,224 | |||||
1 | """Utilities to manipulate JSON objects. |
|
1 | """Utilities to manipulate JSON objects. | |
2 | """ |
|
2 | """ | |
3 | #----------------------------------------------------------------------------- |
|
3 | #----------------------------------------------------------------------------- | |
4 | # Copyright (C) 2010-2011 The IPython Development Team |
|
4 | # Copyright (C) 2010-2011 The IPython Development Team | |
5 | # |
|
5 | # | |
6 | # Distributed under the terms of the BSD License. The full license is in |
|
6 | # Distributed under the terms of the BSD License. The full license is in | |
7 | # the file COPYING.txt, distributed as part of this software. |
|
7 | # the file COPYING.txt, distributed as part of this software. | |
8 | #----------------------------------------------------------------------------- |
|
8 | #----------------------------------------------------------------------------- | |
9 |
|
9 | |||
10 | #----------------------------------------------------------------------------- |
|
10 | #----------------------------------------------------------------------------- | |
11 | # Imports |
|
11 | # Imports | |
12 | #----------------------------------------------------------------------------- |
|
12 | #----------------------------------------------------------------------------- | |
13 | # stdlib |
|
13 | # stdlib | |
14 | import math |
|
14 | import math | |
15 | import re |
|
15 | import re | |
16 | import types |
|
16 | import types | |
17 | from datetime import datetime |
|
17 | from datetime import datetime | |
18 |
|
18 | |||
19 | try: |
|
19 | try: | |
20 | # base64.encodestring is deprecated in Python 3.x |
|
20 | # base64.encodestring is deprecated in Python 3.x | |
21 | from base64 import encodebytes |
|
21 | from base64 import encodebytes | |
22 | except ImportError: |
|
22 | except ImportError: | |
23 | # Python 2.x |
|
23 | # Python 2.x | |
24 | from base64 import encodestring as encodebytes |
|
24 | from base64 import encodestring as encodebytes | |
25 |
|
25 | |||
26 | from IPython.utils import py3compat |
|
26 | from IPython.utils import py3compat | |
27 | from IPython.utils.encoding import DEFAULT_ENCODING |
|
27 | from IPython.utils.encoding import DEFAULT_ENCODING | |
28 | next_attr_name = '__next__' if py3compat.PY3 else 'next' |
|
28 | next_attr_name = '__next__' if py3compat.PY3 else 'next' | |
29 |
|
29 | |||
30 | #----------------------------------------------------------------------------- |
|
30 | #----------------------------------------------------------------------------- | |
31 | # Globals and constants |
|
31 | # Globals and constants | |
32 | #----------------------------------------------------------------------------- |
|
32 | #----------------------------------------------------------------------------- | |
33 |
|
33 | |||
34 | # timestamp formats |
|
34 | # timestamp formats | |
35 | ISO8601="%Y-%m-%dT%H:%M:%S.%f" |
|
35 | ISO8601="%Y-%m-%dT%H:%M:%S.%f" | |
36 | ISO8601_PAT=re.compile(r"^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d+$") |
|
36 | ISO8601_PAT=re.compile(r"^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d+$") | |
37 |
|
37 | |||
38 | #----------------------------------------------------------------------------- |
|
38 | #----------------------------------------------------------------------------- | |
39 | # Classes and functions |
|
39 | # Classes and functions | |
40 | #----------------------------------------------------------------------------- |
|
40 | #----------------------------------------------------------------------------- | |
41 |
|
41 | |||
42 | def rekey(dikt): |
|
42 | def rekey(dikt): | |
43 | """Rekey a dict that has been forced to use str keys where there should be |
|
43 | """Rekey a dict that has been forced to use str keys where there should be | |
44 | ints by json.""" |
|
44 | ints by json.""" | |
45 | for k in dikt.iterkeys(): |
|
45 | for k in dikt.iterkeys(): | |
46 | if isinstance(k, basestring): |
|
46 | if isinstance(k, basestring): | |
47 | ik=fk=None |
|
47 | ik=fk=None | |
48 | try: |
|
48 | try: | |
49 | ik = int(k) |
|
49 | ik = int(k) | |
50 | except ValueError: |
|
50 | except ValueError: | |
51 | try: |
|
51 | try: | |
52 | fk = float(k) |
|
52 | fk = float(k) | |
53 | except ValueError: |
|
53 | except ValueError: | |
54 | continue |
|
54 | continue | |
55 | if ik is not None: |
|
55 | if ik is not None: | |
56 | nk = ik |
|
56 | nk = ik | |
57 | else: |
|
57 | else: | |
58 | nk = fk |
|
58 | nk = fk | |
59 | if nk in dikt: |
|
59 | if nk in dikt: | |
60 | raise KeyError("already have key %r"%nk) |
|
60 | raise KeyError("already have key %r"%nk) | |
61 | dikt[nk] = dikt.pop(k) |
|
61 | dikt[nk] = dikt.pop(k) | |
62 | return dikt |
|
62 | return dikt | |
63 |
|
63 | |||
64 |
|
64 | |||
65 | def extract_dates(obj): |
|
65 | def extract_dates(obj): | |
66 | """extract ISO8601 dates from unpacked JSON""" |
|
66 | """extract ISO8601 dates from unpacked JSON""" | |
67 | if isinstance(obj, dict): |
|
67 | if isinstance(obj, dict): | |
68 | obj = dict(obj) # don't clobber |
|
68 | obj = dict(obj) # don't clobber | |
69 | for k,v in obj.iteritems(): |
|
69 | for k,v in obj.iteritems(): | |
70 | obj[k] = extract_dates(v) |
|
70 | obj[k] = extract_dates(v) | |
71 | elif isinstance(obj, (list, tuple)): |
|
71 | elif isinstance(obj, (list, tuple)): | |
72 | obj = [ extract_dates(o) for o in obj ] |
|
72 | obj = [ extract_dates(o) for o in obj ] | |
73 | elif isinstance(obj, basestring): |
|
73 | elif isinstance(obj, basestring): | |
74 | if ISO8601_PAT.match(obj): |
|
74 | if ISO8601_PAT.match(obj): | |
75 | obj = datetime.strptime(obj, ISO8601) |
|
75 | obj = datetime.strptime(obj, ISO8601) | |
76 | return obj |
|
76 | return obj | |
77 |
|
77 | |||
78 | def squash_dates(obj): |
|
78 | def squash_dates(obj): | |
79 | """squash datetime objects into ISO8601 strings""" |
|
79 | """squash datetime objects into ISO8601 strings""" | |
80 | if isinstance(obj, dict): |
|
80 | if isinstance(obj, dict): | |
81 | obj = dict(obj) # don't clobber |
|
81 | obj = dict(obj) # don't clobber | |
82 | for k,v in obj.iteritems(): |
|
82 | for k,v in obj.iteritems(): | |
83 | obj[k] = squash_dates(v) |
|
83 | obj[k] = squash_dates(v) | |
84 | elif isinstance(obj, (list, tuple)): |
|
84 | elif isinstance(obj, (list, tuple)): | |
85 | obj = [ squash_dates(o) for o in obj ] |
|
85 | obj = [ squash_dates(o) for o in obj ] | |
86 | elif isinstance(obj, datetime): |
|
86 | elif isinstance(obj, datetime): | |
87 | obj = obj.strftime(ISO8601) |
|
87 | obj = obj.strftime(ISO8601) | |
88 | return obj |
|
88 | return obj | |
89 |
|
89 | |||
90 | def date_default(obj): |
|
90 | def date_default(obj): | |
91 | """default function for packing datetime objects in JSON.""" |
|
91 | """default function for packing datetime objects in JSON.""" | |
92 | if isinstance(obj, datetime): |
|
92 | if isinstance(obj, datetime): | |
93 | return obj.strftime(ISO8601) |
|
93 | return obj.strftime(ISO8601) | |
94 | else: |
|
94 | else: | |
95 | raise TypeError("%r is not JSON serializable"%obj) |
|
95 | raise TypeError("%r is not JSON serializable"%obj) | |
96 |
|
96 | |||
97 |
|
97 | |||
98 | # constants for identifying png/jpeg data |
|
98 | # constants for identifying png/jpeg data | |
99 | PNG = b'\x89PNG\r\n\x1a\n' |
|
99 | PNG = b'\x89PNG\r\n\x1a\n' | |
100 | PNG64 = encodebytes(PNG) |
|
100 | # front of PNG base64-encoded | |
|
101 | PNG64 = b'iVBORw0KG' | |||
101 | JPEG = b'\xff\xd8' |
|
102 | JPEG = b'\xff\xd8' | |
102 | JPEG64 = encodebytes(JPEG) |
|
103 | # front of JPEG base64-encoded | |
|
104 | JPEG64 = b'/9' | |||
103 |
|
105 | |||
104 | def encode_images(format_dict): |
|
106 | def encode_images(format_dict): | |
105 | """b64-encodes images in a displaypub format dict |
|
107 | """b64-encodes images in a displaypub format dict | |
106 |
|
108 | |||
107 | Perhaps this should be handled in json_clean itself? |
|
109 | Perhaps this should be handled in json_clean itself? | |
108 |
|
110 | |||
109 | Parameters |
|
111 | Parameters | |
110 | ---------- |
|
112 | ---------- | |
111 |
|
113 | |||
112 | format_dict : dict |
|
114 | format_dict : dict | |
113 | A dictionary of display data keyed by mime-type |
|
115 | A dictionary of display data keyed by mime-type | |
114 |
|
116 | |||
115 | Returns |
|
117 | Returns | |
116 | ------- |
|
118 | ------- | |
117 |
|
119 | |||
118 | format_dict : dict |
|
120 | format_dict : dict | |
119 | A copy of the same dictionary, |
|
121 | A copy of the same dictionary, | |
120 | but binary image data ('image/png' or 'image/jpeg') |
|
122 | but binary image data ('image/png' or 'image/jpeg') | |
121 | is base64-encoded. |
|
123 | is base64-encoded. | |
122 |
|
124 | |||
123 | """ |
|
125 | """ | |
124 | encoded = format_dict.copy() |
|
126 | encoded = format_dict.copy() | |
125 |
|
127 | |||
126 | pngdata = format_dict.get('image/png') |
|
128 | pngdata = format_dict.get('image/png') | |
127 | if isinstance(pngdata, bytes): |
|
129 | if isinstance(pngdata, bytes): | |
128 | # make sure we don't double-encode |
|
130 | # make sure we don't double-encode | |
129 |
if pngdata |
|
131 | if not pngdata.startswith(PNG64): | |
130 | pngdata = encodebytes(pngdata) |
|
132 | pngdata = encodebytes(pngdata) | |
131 | encoded['image/png'] = pngdata.decode('ascii') |
|
133 | encoded['image/png'] = pngdata.decode('ascii') | |
132 |
|
134 | |||
133 | jpegdata = format_dict.get('image/jpeg') |
|
135 | jpegdata = format_dict.get('image/jpeg') | |
134 | if isinstance(jpegdata, bytes): |
|
136 | if isinstance(jpegdata, bytes): | |
135 | # make sure we don't double-encode |
|
137 | # make sure we don't double-encode | |
136 |
if jpegdata |
|
138 | if not jpegdata.startswith(JPEG64): | |
137 | jpegdata = encodebytes(jpegdata) |
|
139 | jpegdata = encodebytes(jpegdata) | |
138 | encoded['image/jpeg'] = jpegdata.decode('ascii') |
|
140 | encoded['image/jpeg'] = jpegdata.decode('ascii') | |
139 |
|
141 | |||
140 | return encoded |
|
142 | return encoded | |
141 |
|
143 | |||
142 |
|
144 | |||
143 | def json_clean(obj): |
|
145 | def json_clean(obj): | |
144 | """Clean an object to ensure it's safe to encode in JSON. |
|
146 | """Clean an object to ensure it's safe to encode in JSON. | |
145 |
|
147 | |||
146 | Atomic, immutable objects are returned unmodified. Sets and tuples are |
|
148 | Atomic, immutable objects are returned unmodified. Sets and tuples are | |
147 | converted to lists, lists are copied and dicts are also copied. |
|
149 | converted to lists, lists are copied and dicts are also copied. | |
148 |
|
150 | |||
149 | Note: dicts whose keys could cause collisions upon encoding (such as a dict |
|
151 | Note: dicts whose keys could cause collisions upon encoding (such as a dict | |
150 | with both the number 1 and the string '1' as keys) will cause a ValueError |
|
152 | with both the number 1 and the string '1' as keys) will cause a ValueError | |
151 | to be raised. |
|
153 | to be raised. | |
152 |
|
154 | |||
153 | Parameters |
|
155 | Parameters | |
154 | ---------- |
|
156 | ---------- | |
155 | obj : any python object |
|
157 | obj : any python object | |
156 |
|
158 | |||
157 | Returns |
|
159 | Returns | |
158 | ------- |
|
160 | ------- | |
159 | out : object |
|
161 | out : object | |
160 |
|
162 | |||
161 | A version of the input which will not cause an encoding error when |
|
163 | A version of the input which will not cause an encoding error when | |
162 | encoded as JSON. Note that this function does not *encode* its inputs, |
|
164 | encoded as JSON. Note that this function does not *encode* its inputs, | |
163 | it simply sanitizes it so that there will be no encoding errors later. |
|
165 | it simply sanitizes it so that there will be no encoding errors later. | |
164 |
|
166 | |||
165 | Examples |
|
167 | Examples | |
166 | -------- |
|
168 | -------- | |
167 | >>> json_clean(4) |
|
169 | >>> json_clean(4) | |
168 | 4 |
|
170 | 4 | |
169 | >>> json_clean(range(10)) |
|
171 | >>> json_clean(range(10)) | |
170 | [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] |
|
172 | [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] | |
171 | >>> sorted(json_clean(dict(x=1, y=2)).items()) |
|
173 | >>> sorted(json_clean(dict(x=1, y=2)).items()) | |
172 | [('x', 1), ('y', 2)] |
|
174 | [('x', 1), ('y', 2)] | |
173 | >>> sorted(json_clean(dict(x=1, y=2, z=[1,2,3])).items()) |
|
175 | >>> sorted(json_clean(dict(x=1, y=2, z=[1,2,3])).items()) | |
174 | [('x', 1), ('y', 2), ('z', [1, 2, 3])] |
|
176 | [('x', 1), ('y', 2), ('z', [1, 2, 3])] | |
175 | >>> json_clean(True) |
|
177 | >>> json_clean(True) | |
176 | True |
|
178 | True | |
177 | """ |
|
179 | """ | |
178 | # types that are 'atomic' and ok in json as-is. bool doesn't need to be |
|
180 | # types that are 'atomic' and ok in json as-is. bool doesn't need to be | |
179 | # listed explicitly because bools pass as int instances |
|
181 | # listed explicitly because bools pass as int instances | |
180 | atomic_ok = (unicode, int, types.NoneType) |
|
182 | atomic_ok = (unicode, int, types.NoneType) | |
181 |
|
183 | |||
182 | # containers that we need to convert into lists |
|
184 | # containers that we need to convert into lists | |
183 | container_to_list = (tuple, set, types.GeneratorType) |
|
185 | container_to_list = (tuple, set, types.GeneratorType) | |
184 |
|
186 | |||
185 | if isinstance(obj, float): |
|
187 | if isinstance(obj, float): | |
186 | # cast out-of-range floats to their reprs |
|
188 | # cast out-of-range floats to their reprs | |
187 | if math.isnan(obj) or math.isinf(obj): |
|
189 | if math.isnan(obj) or math.isinf(obj): | |
188 | return repr(obj) |
|
190 | return repr(obj) | |
189 | return obj |
|
191 | return obj | |
190 |
|
192 | |||
191 | if isinstance(obj, atomic_ok): |
|
193 | if isinstance(obj, atomic_ok): | |
192 | return obj |
|
194 | return obj | |
193 |
|
195 | |||
194 | if isinstance(obj, bytes): |
|
196 | if isinstance(obj, bytes): | |
195 | return obj.decode(DEFAULT_ENCODING, 'replace') |
|
197 | return obj.decode(DEFAULT_ENCODING, 'replace') | |
196 |
|
198 | |||
197 | if isinstance(obj, container_to_list) or ( |
|
199 | if isinstance(obj, container_to_list) or ( | |
198 | hasattr(obj, '__iter__') and hasattr(obj, next_attr_name)): |
|
200 | hasattr(obj, '__iter__') and hasattr(obj, next_attr_name)): | |
199 | obj = list(obj) |
|
201 | obj = list(obj) | |
200 |
|
202 | |||
201 | if isinstance(obj, list): |
|
203 | if isinstance(obj, list): | |
202 | return [json_clean(x) for x in obj] |
|
204 | return [json_clean(x) for x in obj] | |
203 |
|
205 | |||
204 | if isinstance(obj, dict): |
|
206 | if isinstance(obj, dict): | |
205 | # First, validate that the dict won't lose data in conversion due to |
|
207 | # First, validate that the dict won't lose data in conversion due to | |
206 | # key collisions after stringification. This can happen with keys like |
|
208 | # key collisions after stringification. This can happen with keys like | |
207 | # True and 'true' or 1 and '1', which collide in JSON. |
|
209 | # True and 'true' or 1 and '1', which collide in JSON. | |
208 | nkeys = len(obj) |
|
210 | nkeys = len(obj) | |
209 | nkeys_collapsed = len(set(map(str, obj))) |
|
211 | nkeys_collapsed = len(set(map(str, obj))) | |
210 | if nkeys != nkeys_collapsed: |
|
212 | if nkeys != nkeys_collapsed: | |
211 | raise ValueError('dict can not be safely converted to JSON: ' |
|
213 | raise ValueError('dict can not be safely converted to JSON: ' | |
212 | 'key collision would lead to dropped values') |
|
214 | 'key collision would lead to dropped values') | |
213 | # If all OK, proceed by making the new dict that will be json-safe |
|
215 | # If all OK, proceed by making the new dict that will be json-safe | |
214 | out = {} |
|
216 | out = {} | |
215 | for k,v in obj.iteritems(): |
|
217 | for k,v in obj.iteritems(): | |
216 | out[str(k)] = json_clean(v) |
|
218 | out[str(k)] = json_clean(v) | |
217 | return out |
|
219 | return out | |
218 |
|
220 | |||
219 | # If we get here, we don't know how to handle the object, so we just get |
|
221 | # If we get here, we don't know how to handle the object, so we just get | |
220 | # its repr and return that. This will catch lambdas, open sockets, class |
|
222 | # its repr and return that. This will catch lambdas, open sockets, class | |
221 | # objects, and any other complicated contraption that json can't encode |
|
223 | # objects, and any other complicated contraption that json can't encode | |
222 | return repr(obj) |
|
224 | return repr(obj) |
General Comments 0
You need to be logged in to leave comments.
Login now