]> git.cworth.org Git - apitrace/blob - scripts/jsondiff.py
jsondiff: Try to be more accurate w/ commas/newlines.
[apitrace] / scripts / jsondiff.py
1 #!/usr/bin/env python
2 ##########################################################################
3 #
4 # Copyright 2011 Jose Fonseca
5 # All Rights Reserved.
6 #
7 # Permission is hereby granted, free of charge, to any person obtaining a copy
8 # of this software and associated documentation files (the "Software"), to deal
9 # in the Software without restriction, including without limitation the rights
10 # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
11 # copies of the Software, and to permit persons to whom the Software is
12 # furnished to do so, subject to the following conditions:
13 #
14 # The above copyright notice and this permission notice shall be included in
15 # all copies or substantial portions of the Software.
16 #
17 # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
18 # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
20 # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
21 # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
22 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
23 # THE SOFTWARE.
24 #
25 ##########################################################################/
26
27
28 import json
29 import sys
30
31
32 def strip_object_hook(obj):
33     if '__class__' in obj:
34         return None
35     for name in obj.keys():
36         if name.startswith('__') and name.endswith('__'):
37             del obj[name]
38     return obj
39
40
41 class Visitor:
42
43     def visit(self, node, *args, **kwargs):
44         if isinstance(node, dict):
45             return self.visit_object(node, *args, **kwargs)
46         elif isinstance(node, list):
47             return self.visit_array(node, *args, **kwargs)
48         else:
49             return self.visit_value(node, *args, **kwargs)
50
51     def visit_object(self, node, *args, **kwargs):
52         pass
53
54     def visit_array(self, node, *args, **kwargs):
55         pass
56
57     def visit_value(self, node, *args, **kwargs):
58         pass
59
60
61 class Dumper(Visitor):
62
63     def __init__(self, stream = sys.stdout):
64         self.stream = stream
65         self.level = 0
66
67     def _write(self, s):
68         self.stream.write(s)
69
70     def _indent(self):
71         self._write('  '*self.level)
72
73     def _newline(self):
74         self._write('\n')
75
76     def visit_object(self, node):
77         self.enter_object()
78
79         members = node.keys()
80         members.sort()
81         for i in range(len(members)):
82             name = members[i]
83             value = node[name]
84             self.enter_member(name)
85             self.visit(value)
86             self.leave_member(i == len(members) - 1)
87         self.leave_object()
88
89     def enter_object(self):
90         self._write('{')
91         self._newline()
92         self.level += 1
93
94     def enter_member(self, name):
95         self._indent()
96         self._write('%s: ' % name)
97
98     def leave_member(self, last):
99         if not last:
100             self._write(',')
101         self._newline()
102
103     def leave_object(self):
104         self.level -= 1
105         self._indent()
106         self._write('}')
107         if self.level <= 0:
108             self._newline()
109
110     def visit_array(self, node):
111         self.enter_array()
112         for i in range(len(node)):
113             value = node[i]
114             self._indent()
115             self.visit(value)
116             if i != len(node) - 1:
117                 self._write(',')
118             self._newline()
119         self.leave_array()
120
121     def enter_array(self):
122         self._write('[')
123         self._newline()
124         self.level += 1
125
126     def leave_array(self):
127         self.level -= 1
128         self._indent()
129         self._write(']')
130
131     def visit_value(self, node):
132         self._write(json.dumps(node))
133
134
135
136 class Comparer(Visitor):
137
138     def __init__(self, ignore_added = False):
139         self.ignore_added = ignore_added
140
141     def visit_object(self, a, b):
142         if not isinstance(b, dict):
143             return False
144         if len(a) != len(b) and not self.ignore_added:
145             return False
146         ak = a.keys()
147         bk = b.keys()
148         ak.sort()
149         bk.sort()
150         if ak != bk and not self.ignore_added:
151             return False
152         for k in ak:
153             ae = a[k]
154             try:
155                 be = b[k]
156             except KeyError:
157                 return False
158             if not self.visit(ae, be):
159                 return False
160         return True
161
162     def visit_array(self, a, b):
163         if not isinstance(b, list):
164             return False
165         if len(a) != len(b):
166             return False
167         for ae, be in zip(a, b):
168             if not self.visit(ae, be):
169                 return False
170         return True
171
172     def visit_value(self, a, b):
173         return a == b
174
175
176
177 class Differ(Visitor):
178
179     def __init__(self, stream = sys.stdout, ignore_added = False):
180         self.dumper = Dumper(stream)
181         self.comparer = Comparer(ignore_added = ignore_added)
182
183     def visit(self, a, b):
184         if self.comparer.visit(a, b):
185             return
186         Visitor.visit(self, a, b)
187
188     def visit_object(self, a, b):
189         if not isinstance(b, dict):
190             self.replace(a, b)
191         else:
192             self.dumper.enter_object()
193             names = set(a.keys())
194             if not self.comparer.ignore_added:
195                 names.update(b.keys())
196             names = list(names)
197             names.sort()
198
199             for i in range(len(names)):
200                 name = names[i]
201                 ae = a.get(name, None)
202                 be = b.get(name, None)
203                 if not self.comparer.visit(ae, be):
204                     self.dumper.enter_member(name)
205                     self.visit(ae, be)
206                     self.dumper.leave_member(i == len(names) - 1)
207
208             self.dumper.leave_object()
209
210     def visit_array(self, a, b):
211         if not isinstance(b, list):
212             self.replace(a, b)
213         else:
214             self.dumper.enter_array()
215             max_len = max(len(a), len(b))
216             for i in range(max_len):
217                 try:
218                     ae = a[i]
219                 except IndexError:
220                     ae = None
221                 try:
222                     be = b[i]
223                 except IndexError:
224                     be = None
225                 self.dumper._indent()
226                 if self.comparer.visit(ae, be):
227                     self.dumper.visit(ae)
228                 else:
229                     self.visit(ae, be)
230                 if i != max_len - 1:
231                     self.dumper._write(',')
232                 self.dumper._newline()
233
234             self.dumper.leave_array()
235
236     def visit_value(self, a, b):
237         if a != b:
238             self.replace(a, b)
239
240     def replace(self, a, b):
241         self.dumper.visit(a)
242         self.dumper._write(' -> ')
243         self.dumper.visit(b)
244
245
246 def load(stream, strip = True):
247     if strip:
248         object_hook = strip_object_hook
249     else:
250         object_hook = None
251     return json.load(stream, strict=False, object_hook = object_hook)
252
253
254 def main():
255     a = load(open(sys.argv[1], 'rt'))
256     b = load(open(sys.argv[2], 'rt'))
257
258     if False:
259         dumper = Dumper()
260         dumper.visit(a)
261
262     differ = Differ()
263     differ.visit(a, b)
264
265
266 if __name__ == '__main__':
267     main()