Mercurial > repos > shellac > guppy_basecaller

comparison env/lib/python3.7/site-packages/schema_salad/validate.py @ 2:6af9afd405e9 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 0a63dd5f4d38a1f6944587f52a8cd79874177fc1"
author shellac
date Thu, 14 May 2020 14:56:58 -0400
parents 26e78fe6e8c4
children
comparison
equal deleted inserted replaced
1:75ca89e9b81c 2:6af9afd405e9
1 from __future__ import absolute_import
2
3 import logging
4 import pprint
5 from typing import ( # noqa: F401
6 Any,
7 List,
8 MutableMapping,
9 MutableSequence,
10 Optional,
11 Set,
12 Union,
13 )
14
15 import six
16 from six.moves import urllib
17 from typing_extensions import Text # pylint: disable=unused-import
18
19 from . import avro
20 from .exceptions import (
21 ClassValidationException,
22 ValidationException,
23 SchemaSaladException,
24 )
25 from .avro import schema # noqa: F401
26 from .avro.schema import ( # pylint: disable=unused-import, no-name-in-module, import-error
27 Schema,
28 )
29 from .sourceline import SourceLine
30
31 # move to a regular typing import when Python 3.3-3.6 is no longer supported
32
33
34 _logger = logging.getLogger("salad")
35
36
37 def validate(
38 expected_schema, # type: Schema
39 datum, # type: Any
40 identifiers=None, # type: Optional[List[Text]]
41 strict=False, # type: bool
42 foreign_properties=None, # type: Optional[Set[Text]]
43 ):
44 # type: (...) -> bool
45 if not identifiers:
46 identifiers = []
47 if not foreign_properties:
48 foreign_properties = set()
49 return validate_ex(
50 expected_schema,
51 datum,
52 identifiers,
53 strict=strict,
54 foreign_properties=foreign_properties,
55 raise_ex=False,
56 )
57
58
59 INT_MIN_VALUE = -(1 << 31)
60 INT_MAX_VALUE = (1 << 31) - 1
61 LONG_MIN_VALUE = -(1 << 63)
62 LONG_MAX_VALUE = (1 << 63) - 1
63
64
65 def friendly(v): # type: (Any) -> Any
66 if isinstance(v, avro.schema.NamedSchema):
67 return v.name
68 if isinstance(v, avro.schema.ArraySchema):
69 return "array of <{}>".format(friendly(v.items))
70 elif isinstance(v, avro.schema.PrimitiveSchema):
71 return v.type
72 elif isinstance(v, avro.schema.UnionSchema):
73 return " or ".join([friendly(s) for s in v.schemas])
74 else:
75 return v
76
77
78 def vpformat(datum): # type: (Any) -> str
79 a = pprint.pformat(datum)
80 if len(a) > 160:
81 a = a[0:160] + "[...]"
82 return a
83
84
85 def validate_ex(
86 expected_schema, # type: Schema
87 datum, # type: Any
88 identifiers=None, # type: Optional[List[Text]]
89 strict=False, # type: bool
90 foreign_properties=None, # type: Optional[Set[Text]]
91 raise_ex=True, # type: bool
92 strict_foreign_properties=False, # type: bool
93 logger=_logger, # type: logging.Logger
94 skip_foreign_properties=False, # type: bool
95 ):
96 # type: (...) -> bool
97 """Determine if a python datum is an instance of a schema."""
98
99 if not identifiers:
100 identifiers = []
101
102 if not foreign_properties:
103 foreign_properties = set()
104
105 schema_type = expected_schema.type
106
107 if schema_type == "null":
108 if datum is None:
109 return True
110 else:
111 if raise_ex:
112 raise ValidationException(u"the value is not null")
113 else:
114 return False
115 elif schema_type == "boolean":
116 if isinstance(datum, bool):
117 return True
118 else:
119 if raise_ex:
120 raise ValidationException(u"the value is not boolean")
121 else:
122 return False
123 elif schema_type == "string":
124 if isinstance(datum, six.string_types):
125 return True
126 elif isinstance(datum, bytes):
127 datum = datum.decode(u"utf-8")
128 return True
129 else:
130 if raise_ex:
131 raise ValidationException(u"the value is not string")
132 else:
133 return False
134 elif schema_type == "int":
135 if (
136 isinstance(datum, six.integer_types)
137 and INT_MIN_VALUE <= datum <= INT_MAX_VALUE
138 ):
139 return True
140 else:
141 if raise_ex:
142 raise ValidationException(u"`{}` is not int".format(vpformat(datum)))
143 else:
144 return False
145 elif schema_type == "long":
146 if (
147 isinstance(datum, six.integer_types)
148 ) and LONG_MIN_VALUE <= datum <= LONG_MAX_VALUE:
149 return True
150 else:
151 if raise_ex:
152 raise ValidationException(
153 u"the value `{}` is not long".format(vpformat(datum))
154 )
155 else:
156 return False
157 elif schema_type in ["float", "double"]:
158 if isinstance(datum, six.integer_types) or isinstance(datum, float):
159 return True
160 else:
161 if raise_ex:
162 raise ValidationException(
163 u"the value `{}` is not float or double".format(vpformat(datum))
164 )
165 else:
166 return False
167 elif isinstance(expected_schema, avro.schema.EnumSchema):
168 if expected_schema.name == "Any":
169 if datum is not None:
170 return True
171 else:
172 if raise_ex:
173 raise ValidationException(u"'Any' type must be non-null")
174 else:
175 return False
176 if not isinstance(datum, six.string_types):
177 if raise_ex:
178 raise ValidationException(
179 u"value is a {} but expected a string".format(
180 (type(datum).__name__)
181 )
182 )
183 else:
184 return False
185 if expected_schema.name == "Expression":
186 if "$(" in datum or "${" in datum:
187 return True
188 if raise_ex:
189 raise ValidationException(
190 u"value `%s` does not contain an expression in the form $() or ${}"
191 % datum
192 )
193 else:
194 return False
195 if datum in expected_schema.symbols:
196 return True
197 else:
198 if raise_ex:
199 raise ValidationException(
200 u"the value {} is not a valid {}, expected {}{}".format(
201 vpformat(datum),
202 expected_schema.name,
203 "one of " if len(expected_schema.symbols) > 1 else "",
204 "'" + "', '".join(expected_schema.symbols) + "'",
205 )
206 )
207 else:
208 return False
209 elif isinstance(expected_schema, avro.schema.ArraySchema):
210 if isinstance(datum, MutableSequence):
211 for i, d in enumerate(datum):
212 try:
213 sl = SourceLine(datum, i, ValidationException)
214 if not validate_ex(
215 expected_schema.items,
216 d,
217 identifiers,
218 strict=strict,
219 foreign_properties=foreign_properties,
220 raise_ex=raise_ex,
221 strict_foreign_properties=strict_foreign_properties,
222 logger=logger,
223 skip_foreign_properties=skip_foreign_properties,
224 ):
225 return False
226 except ValidationException as v:
227 if raise_ex:
228 raise ValidationException("item is invalid because", sl, [v])
229 else:
230 return False
231 return True
232 else:
233 if raise_ex:
234 raise ValidationException(
235 u"the value {} is not a list, expected list of {}".format(
236 vpformat(datum), friendly(expected_schema.items)
237 )
238 )
239 else:
240 return False
241 elif isinstance(expected_schema, avro.schema.UnionSchema):
242 for s in expected_schema.schemas:
243 if validate_ex(
244 s,
245 datum,
246 identifiers,
247 strict=strict,
248 raise_ex=False,
249 strict_foreign_properties=strict_foreign_properties,
250 logger=logger,
251 skip_foreign_properties=skip_foreign_properties,
252 ):
253 return True
254
255 if not raise_ex:
256 return False
257
258 errors = [] # type: List[SchemaSaladException]
259 checked = []
260 for s in expected_schema.schemas:
261 if isinstance(datum, MutableSequence) and not isinstance(
262 s, avro.schema.ArraySchema
263 ):
264 continue
265 elif isinstance(datum, MutableMapping) and not isinstance(
266 s, avro.schema.RecordSchema
267 ):
268 continue
269 elif isinstance(
270 datum, (bool, six.integer_types, float, six.string_types)
271 ) and isinstance(s, (avro.schema.ArraySchema, avro.schema.RecordSchema)):
272 continue
273 elif datum is not None and s.type == "null":
274 continue
275
276 checked.append(s)
277 try:
278 validate_ex(
279 s,
280 datum,
281 identifiers,
282 strict=strict,
283 foreign_properties=foreign_properties,
284 raise_ex=True,
285 strict_foreign_properties=strict_foreign_properties,
286 logger=logger,
287 skip_foreign_properties=skip_foreign_properties,
288 )
289 except ClassValidationException:
290 raise
291 except ValidationException as e:
292 errors.append(e)
293 if bool(errors):
294 raise ValidationException(
295 "",
296 None,
297 [
298 ValidationException(
299 "tried {} but".format(friendly(check)), None, [err]
300 )
301 for (check, err) in zip(checked, errors)
302 ],
303 "-",
304 )
305 else:
306 raise ValidationException(
307 "value is a {}, expected {}".format(
308 type(datum).__name__, friendly(expected_schema)
309 )
310 )
311
312 elif isinstance(expected_schema, avro.schema.RecordSchema):
313 if not isinstance(datum, MutableMapping):
314 if raise_ex:
315 raise ValidationException(u"is not a dict")
316 else:
317 return False
318
319 classmatch = None
320 for f in expected_schema.fields:
321 if f.name in ("class",):
322 d = datum.get(f.name)
323 if not d:
324 if raise_ex:
325 raise ValidationException(u"Missing '{}' field".format(f.name))
326 else:
327 return False
328 if expected_schema.name != d:
329 if raise_ex:
330 raise ValidationException(
331 u"Expected class '{}' but this is '{}'".format(
332 expected_schema.name, d
333 )
334 )
335 else:
336 return False
337 classmatch = d
338 break
339
340 errors = []
341 for f in expected_schema.fields:
342 if f.name in ("class",):
343 continue
344
345 if f.name in datum:
346 fieldval = datum[f.name]
347 else:
348 try:
349 fieldval = f.default
350 except KeyError:
351 fieldval = None
352
353 try:
354 sl = SourceLine(datum, f.name, six.text_type)
355 if not validate_ex(
356 f.type,
357 fieldval,
358 identifiers,
359 strict=strict,
360 foreign_properties=foreign_properties,
361 raise_ex=raise_ex,
362 strict_foreign_properties=strict_foreign_properties,
363 logger=logger,
364 skip_foreign_properties=skip_foreign_properties,
365 ):
366 return False
367 except ValidationException as v:
368 if f.name not in datum:
369 errors.append(
370 ValidationException(
371 u"missing required field `{}`".format(f.name)
372 )
373 )
374 else:
375 errors.append(
376 ValidationException(
377 u"the `{}` field is not valid because".format(f.name),
378 sl,
379 [v],
380 )
381 )
382
383 for d in datum:
384 found = False
385 for f in expected_schema.fields:
386 if d == f.name:
387 found = True
388 if not found:
389 sl = SourceLine(datum, d, six.text_type)
390 if d is None:
391 err = ValidationException(u"mapping with implicit null key", sl)
392 if strict:
393 errors.append(err)
394 else:
395 logger.warning(err)
396 continue
397 if (
398 d not in identifiers
399 and d not in foreign_properties
400 and d[0] not in ("@", "$")
401 ):
402 if (
403 (d not in identifiers and strict)
404 and (
405 d not in foreign_properties
406 and strict_foreign_properties
407 and not skip_foreign_properties
408 )
409 and not raise_ex
410 ):
411 return False
412 split = urllib.parse.urlsplit(d)
413 if split.scheme:
414 if not skip_foreign_properties:
415 err = ValidationException(
416 u"unrecognized extension field `{}`{}.{}".format(
417 d,
418 " and strict_foreign_properties checking is enabled"
419 if strict_foreign_properties
420 else "",
421 "\nForeign properties from $schemas:\n {}".format(
422 "\n ".join(sorted(foreign_properties))
423 )
424 if len(foreign_properties) > 0
425 else "",
426 ),
427 sl,
428 )
429 if strict_foreign_properties:
430 errors.append(err)
431 elif len(foreign_properties) > 0:
432 logger.warning(err)
433 else:
434 err = ValidationException(
435 u"invalid field `{}`, expected one of: {}".format(
436 d,
437 ", ".join(
438 "'{}'".format(fn.name)
439 for fn in expected_schema.fields
440 ),
441 ),
442 sl,
443 )
444 if strict:
445 errors.append(err)
446 else:
447 logger.warning(err)
448
449 if bool(errors):
450 if raise_ex:
451 if classmatch:
452 raise ClassValidationException("", None, errors, "*")
453 else:
454 raise ValidationException("", None, errors, "*")
455 else:
456 return False
457 else:
458 return True
459 if raise_ex:
460 raise ValidationException(u"Unrecognized schema_type {}".format(schema_type))
461 else:
462 return False