webapp.forms
18 January 2008
18:21
as NumPy
Simple HTML forms library thing
| 1 | """Tools for form processing, data conversion and validation. |
|---|---|
| 2 | |
| 3 | Assume the following raw input data, which may for example come from the body |
| 4 | of a POST request, which basically represents a flat dictionary: |
| 5 | |
| 6 | >>> data = { |
| 7 | ... 'people[0][name]': 'johnny', 'people[0][age]': '42', |
| 8 | ... 'people[1][name]': 'anna', 'people[1][age]': '23', |
| 9 | ... } |
| 10 | |
| 11 | Decoding this data will produce a properly nested structure of dictionaries and |
| 12 | lists: |
| 13 | |
| 14 | >>> data = decode(data) |
| 15 | >>> data |
| 16 | {'people': [{'age': '42', 'name': 'johnny'}, {'age': '23', 'name': 'anna'}]} |
| 17 | |
| 18 | Note though that neither validation nor data type conversions have been |
| 19 | performed. That is the responsibility of the validators: |
| 20 | |
| 21 | >>> validate = DictValidator(people=ListValidator(DictValidator( |
| 22 | ... name=TextValidator(), |
| 23 | ... age=IntValidator() |
| 24 | ... ))) |
| 25 | >>> validate(data) |
| 26 | {'people': [{'age': 42, 'name': u'johnny'}, {'age': 23, 'name': u'anna'}]} |
| 27 | |
| 28 | You can see that type conversions have been applied as requested. |
| 29 | |
| 30 | If invalid data is encountered, a `ValidationError` will be raised: |
| 31 | |
| 32 | >>> validate(decode({'people[0][age]': 'forty-two'})) |
| 33 | Traceback (most recent call last): |
| 34 | ... |
| 35 | ValidationErrors: 1 error |
| 36 | |
| 37 | To get useful information out of such an exception, it needs to be "unpacked": |
| 38 | |
| 39 | >>> try: |
| 40 | ... validate(decode({'people[0][age]': 'forty-two'})) |
| 41 | ... except ValidationError, e: |
| 42 | ... errors = e.unpack() #doctest: +ELLIPSIS |
| 43 | >>> print errors['people'][0]['age'][0] |
| 44 | Please enter a whole number. |
| 45 | |
| 46 | The structure of the unpacked errors mirrors that of the decoded data and the |
| 47 | validators, using nested dictionary and lists. |
| 48 | |
| 49 | On top of validators, there's a third abstraction layer: the `Form` class. The |
| 50 | example given above could also be defined and used as follows: |
| 51 | |
| 52 | >>> class PeopleForm(Form): |
| 53 | ... people = [{ |
| 54 | ... 'name': TextValidator(), |
| 55 | ... 'age': IntValidator() |
| 56 | ... }] |
| 57 | |
| 58 | >>> form = PeopleForm() |
| 59 | >>> form.validate(data) |
| 60 | True |
| 61 | >>> form['people'] |
| 62 | [{'age': 42, 'name': u'johnny'}, {'age': 23, 'name': u'anna'}] |
| 63 | """ |
| 64 | |
| 65 | from cgi import parse_header |
| 66 | import logging |
| 67 | from md5 import md5 |
| 68 | from random import random |
| 69 | import re |
| 70 | from types import FunctionType |
| 71 | |
| 72 | from genshi.builder import tag |
| 73 | from genshi.core import Stream |
| 74 | from genshi.filters import HTMLFormFiller, Transformer |
| 75 | |
| 76 | from webapp import app |
| 77 | from webapp.http import abort |
| 78 | from webapp.i18n import gettext, ngettext |
| 79 | |
| 80 | __all__ = ['Form', 'BoolValidator', 'DictValidator', 'IntValidator', |
| 81 | 'ListValidator', 'TextValidator', 'decode', 'encode'] |
| 82 | __docformat__ = 'restructuredtext en' |
| 83 | |
| 84 | log = logging.getLogger('webapp.forms') |
| 85 | |
| 86 | |
| 87 | _decode = re.compile(r'\[(\w+)\]').findall |
| 88 | |
| 89 | def decode(data): |
| 90 | """Decodes the flat dictionary d into a nested structure. |
| 91 | |
| 92 | >>> decode({'foo': 'bar'}) |
| 93 | {'foo': 'bar'} |
| 94 | |
| 95 | >>> decode({'foo[0]': 'bar', 'foo[1]': 'baz'}) |
| 96 | {'foo': ['bar', 'baz']} |
| 97 | |
| 98 | >>> data = decode({'foo[bar]': '1', 'foo[baz]': '2'}) |
| 99 | >>> assert data == {'foo': {'bar': '1', 'baz': '2'}} |
| 100 | |
| 101 | >>> decode({'foo[bar][0]': 'baz', 'foo[bar][1]': 'buzz'}) |
| 102 | {'foo': {'bar': ['baz', 'buzz']}} |
| 103 | |
| 104 | >>> decode({'foo[0][bar]': '23', 'foo[1][baz]': '42'}) |
| 105 | {'foo': [{'bar': '23'}, {'baz': '42'}]} |
| 106 | |
| 107 | >>> decode({'foo[0][0]': '23', 'foo[0][1]': '42'}) |
| 108 | {'foo': [['23', '42']]} |
| 109 | |
| 110 | >>> decode({'foo': ['23', '42']}) |
| 111 | {'foo': ['23', '42']} |
| 112 | """ |
| 113 | result = {} |
| 114 | lists = [] |
| 115 | |
| 116 | for key, value in data.items(): |
| 117 | pos = key.find('[') |
| 118 | if pos == -1: |
| 119 | result[key] = value |
| 120 | else: |
| 121 | names = [key[:pos]] + _decode(key[pos:]) |
| 122 | container = result |
| 123 | end = len(names) - 1 |
| 124 | for idx in range(end): |
| 125 | curname = names[idx] |
| 126 | if curname not in container: |
| 127 | nextname = names[idx + 1] |
| 128 | container[curname] = {} |
| 129 | if nextname.isdigit(): |
| 130 | container[curname]['__list__'] = True |
| 131 | container = container[curname] |
| 132 | container[names[-1]] = value |
| 133 | |
| 134 | def _convert(data): |
| 135 | if type(data) is dict: |
| 136 | if '__list__' in data: |
| 137 | del data['__list__'] |
| 138 | data = [_convert(v) for k, v in sorted(data.items())] |
| 139 | else: |
| 140 | data = dict((k, _convert(v)) for k, v in data.items()) |
| 141 | return data |
| 142 | |
| 143 | return _convert(result) |
| 144 | |
| 145 | def encode(data): |
| 146 | """Encodes a nested structure into a flat dictionary. |
| 147 | |
| 148 | >>> encode({'foo': 'bar'}) |
| 149 | {'foo': 'bar'} |
| 150 | |
| 151 | >>> encode({'foo': ['bar', 'baz']}) |
| 152 | {'foo[0]': 'bar', 'foo[1]': 'baz'} |
| 153 | |
| 154 | >>> encode({'foo': [{'bar': '42', 'baz': '43'}]}) |
| 155 | {'foo[0][baz]': '43', 'foo[0][bar]': '42'} |
| 156 | """ |
| 157 | def _encode(data=data, prefix='', result={}): |
| 158 | if isinstance(data, dict): |
| 159 | for key, value in data.items(): |
| 160 | if key is None: |
| 161 | name = prefix |
| 162 | elif not prefix: |
| 163 | name = key |
| 164 | else: |
| 165 | name = "%s[%s]" % (prefix, key) |
| 166 | _encode(value, name, result) |
| 167 | elif isinstance(data, list): |
| 168 | for i in range(len(data)): |
| 169 | _encode(data[i], "%s[%d]" % (prefix, i), result=result) |
| 170 | else: |
| 171 | result[prefix] = data |
| 172 | return result |
| 173 | return _encode() |
| 174 | |
| 175 | |
| 176 | class ValidationError(Exception): |
| 177 | """Exception raised when invalid data is encountered.""" |
| 178 | |
| 179 | def unpack(self, key=None): |
| 180 | return {key: [self.args[0]]} |
| 181 | |
| 182 | |
| 183 | class Validator(object): |
| 184 | """Abstract validator base class.""" |
| 185 | |
| 186 | def __call__(self, string): |
| 187 | return unicode(string) |
| 188 | |
| 189 | |
| 190 | class DictValidator(Validator): |
| 191 | """Apply a set of validators to a dictionary of values. |
| 192 | |
| 193 | >>> validator = DictValidator(name=TextValidator(), age=IntValidator()) |
| 194 | >>> validator({'name': u'John Doe', 'age': u'42'}) |
| 195 | {'age': 42, 'name': u'John Doe'} |
| 196 | """ |
| 197 | |
| 198 | class ValidationErrors(ValidationError): |
| 199 | |
| 200 | def __init__(self, errors): |
| 201 | ValidationError.__init__(self, '%d error%s' % ( |
| 202 | len(errors), len(errors) != 1 and 's' or '' |
| 203 | )) |
| 204 | self.errors = errors |
| 205 | |
| 206 | def __unicode__(self): |
| 207 | return ', '.join([str(e) for e in self.errors.values()]) |
| 208 | |
| 209 | def unpack(self, key=None): |
| 210 | retval = {} |
| 211 | for name, error in self.errors.items(): |
| 212 | retval.update(error.unpack(key=name)) |
| 213 | return retval |
| 214 | |
| 215 | |
| 216 | def __init__(self, **validators): |
| 217 | self.validators = validators |
| 218 | |
| 219 | def __call__(self, string): |
| 220 | errors = {} |
| 221 | result = {} |
| 222 | for name, validator in self.validators.items(): |
| 223 | try: |
| 224 | result[name] = validator(string.get(name, '')) |
| 225 | except ValidationError, e: |
| 226 | errors[name] = e |
| 227 | if errors: |
| 228 | raise self.ValidationErrors(errors) |
| 229 | return result |
| 230 | |
| 231 | def __repr__(self): |
| 232 | return '<%s %r>' % (self.__class__.__name__, self.validators) |
| 233 | |
| 234 | |
| 235 | class ListValidator(Validator): |
| 236 | """Apply a single validator to a sequence of values. |
| 237 | |
| 238 | >>> validator = ListValidator(IntValidator()) |
| 239 | >>> validator([u'1', u'2', u'3']) |
| 240 | [1, 2, 3] |
| 241 | """ |
| 242 | |
| 243 | class ValidationErrors(ValidationError): |
| 244 | |
| 245 | def __init__(self, errors): |
| 246 | ValidationError.__init__(self, '%d error%s' % ( |
| 247 | len(errors), len(errors) != 1 and 's' or '' |
| 248 | )) |
| 249 | self.errors = errors |
| 250 | |
| 251 | def __unicode__(self): |
| 252 | return ', '.join([str(e) for e in self.errors]) |
| 253 | |
| 254 | def unpack(self, key=None): |
| 255 | return {key: [e.unpack() for e in self.errors]} |
| 256 | |
| 257 | |
| 258 | def __init__(self, validator, min_size=None, max_size=None): |
| 259 | self.validator = validator |
| 260 | self.min_size = min_size |
| 261 | self.max_size = max_size |
| 262 | |
| 263 | def __call__(self, string): |
| 264 | errors = [] |
| 265 | if self.min_size is not None and len(string) < self.min_size: |
| 266 | errors.append(ValidationError( |
| 267 | ngettext('Please provide at least %(num)s item.', |
| 268 | 'Please provide at least %(num)s items.', |
| 269 | self.min_size) |
| 270 | )) |
| 271 | if self.max_size is not None and len(string) > self.max_size: |
| 272 | errors.append(ValidationError( |
| 273 | ngettext('Please provide no more than %(num)s item.', |
| 274 | 'Please provide no more than %(num)s items.', |
| 275 | self.max_size) |
| 276 | )) |
| 277 | result = [] |
| 278 | for item in string: |
| 279 | try: |
| 280 | result.append(self.validator(item)) |
| 281 | except ValidationError, e: |
| 282 | errors.append(e) |
| 283 | if errors: |
| 284 | raise self.ValidationErrors(errors) |
| 285 | return result |
| 286 | |
| 287 | def __repr__(self): |
| 288 | return '<%s %r>' % (self.__class__.__name__, self.validator) |
| 289 | |
| 290 | |
| 291 | class TextValidator(Validator): |
| 292 | """Validator for strings. |
| 293 | |
| 294 | >>> validator = TextValidator(required=True, min_length=6) |
| 295 | >>> validator('foo bar') |
| 296 | u'foo bar' |
| 297 | >>> validator('') |
| 298 | Traceback (most recent call last): |
| 299 | ... |
| 300 | ValidationError: This field is required. |
| 301 | |
| 302 | You can also specify a regular expression that the content must match using |
| 303 | the `pattern` parameter: |
| 304 | |
| 305 | >>> validator = TextValidator(pattern=r'\w+') |
| 306 | >>> validator('foo bar') |
| 307 | Traceback (most recent call last): |
| 308 | ... |
| 309 | ValidationError: The value does not match pattern "^\w+$". |
| 310 | |
| 311 | Because displaying regular expression patterns to the user is ugly almost |
| 312 | always unhelpful, you can specify a custom error message that should be |
| 313 | used when the pattern does not match: |
| 314 | |
| 315 | >>> validator = TextValidator(pattern=r'\w+', |
| 316 | ... message='Only letters allowed here.') |
| 317 | >>> validator('foo bar') |
| 318 | Traceback (most recent call last): |
| 319 | ... |
| 320 | ValidationError: Only letters allowed here. |
| 321 | """ |
| 322 | |
| 323 | def __init__(self, required=False, min_length=None, max_length=None, |
| 324 | pattern=None, message=None): |
| 325 | self.required = required |
| 326 | self.min_length = min_length |
| 327 | self.max_length = max_length |
| 328 | if isinstance(pattern, basestring): |
| 329 | pattern = re.compile('^%s$' % pattern) |
| 330 | self.pattern = pattern |
| 331 | if pattern and not message: |
| 332 | message = gettext('The value does not match pattern "%(pattern)s".', |
| 333 | pattern=pattern.pattern) |
| 334 | self.message = message |
| 335 | |
| 336 | def __call__(self, string): |
| 337 | string = unicode(string) |
| 338 | if self.required and not string: |
| 339 | raise ValidationError(gettext('This field is required.')) |
| 340 | if self.min_length is not None and len(string) < self.min_length: |
| 341 | raise ValidationError( |
| 342 | ngettext('Please enter at least %(num)d character.', |
| 343 | 'Please enter at least %(num)d characters.', |
| 344 | self.min_length) |
| 345 | ) |
| 346 | if self.max_length is not None and len(string) > self.max_length: |
| 347 | raise ValidationError( |
| 348 | ngettext('Please enter no more than %(num)d character.', |
| 349 | 'Please enter no more than %(num)d characters.', |
| 350 | self.max_length) |
| 351 | ) |
| 352 | if self.pattern is not None and not self.pattern.match(string): |
| 353 | raise ValidationError(self.message) |
| 354 | return string |
| 355 | |
| 356 | |
| 357 | class IntValidator(Validator): |
| 358 | """Validator for integers. |
| 359 | |
| 360 | >>> validator = IntValidator(min_value=0, max_value=99) |
| 361 | >>> validator('13') |
| 362 | 13 |
| 363 | |
| 364 | >>> validator('thirteen') |
| 365 | Traceback (most recent call last): |
| 366 | ... |
| 367 | ValidationError: Please enter a whole number. |
| 368 | |
| 369 | >>> validator('193') |
| 370 | Traceback (most recent call last): |
| 371 | ... |
| 372 | ValidationError: Ensure this value is less than or equal to 99. |
| 373 | """ |
| 374 | |
| 375 | def __init__(self, required=False, min_value=None, max_value=None): |
| 376 | self.required = required |
| 377 | self.min_value = min_value |
| 378 | self.max_value = max_value |
| 379 | |
| 380 | def __call__(self, string): |
| 381 | if not string: |
| 382 | if self.required: |
| 383 | raise ValidationError(gettext('This field is required.')) |
| 384 | return None |
| 385 | try: |
| 386 | value = int(string) |
| 387 | except ValueError: |
| 388 | raise ValidationError(gettext('Please enter a whole number.')) |
| 389 | |
| 390 | if self.min_value is not None and value < self.min_value: |
| 391 | raise ValidationError( |
| 392 | gettext('Ensure this value is greater than or equal to ' |
| 393 | '%(value)s.', value=self.min_value) |
| 394 | ) |
| 395 | if self.max_value is not None and value > self.max_value: |
| 396 | raise ValidationError( |
| 397 | gettext('Ensure this value is less than or equal to ' |
| 398 | '%(value)s.', value=self.max_value) |
| 399 | ) |
| 400 | |
| 401 | return int(value) |
| 402 | |
| 403 | |
| 404 | class BoolValidator(Validator): |
| 405 | """Validator for boolean values. |
| 406 | |
| 407 | >>> validator = BoolValidator() |
| 408 | >>> validator('1') |
| 409 | True |
| 410 | |
| 411 | >>> validator = BoolValidator() |
| 412 | >>> validator('') |
| 413 | False |
| 414 | """ |
| 415 | |
| 416 | def __call__(self, string): |
| 417 | return bool(string) |
| 418 | |
| 419 | |
| 420 | class FormMeta(type): |
| 421 | """Meta class for forms.""" |
| 422 | |
| 423 | def __new__(cls, name, bases, d): |
| 424 | validators = {} |
| 425 | validate_functions = [] |
| 426 | |
| 427 | for base in bases: |
| 428 | if hasattr(base, '_validator'): |
| 429 | validators.update(base._validator.validators) |
| 430 | if hasattr(base, '_validators'): |
| 431 | validate_functions.extend(base._validators) |
| 432 | |
| 433 | def _walk(obj): |
| 434 | if isinstance(obj, Validator): |
| 435 | return obj |
| 436 | elif isinstance(obj, dict): |
| 437 | retval = {} |
| 438 | for key, value in obj.items(): |
| 439 | validator = _walk(value) |
| 440 | if validator: |
| 441 | retval[key] = validator |
| 442 | if retval: |
| 443 | return DictValidator(**retval) |
| 444 | elif isinstance(obj, list) and len(obj) == 1: |
| 445 | validator = _walk(obj[0]) |
| 446 | if validator: |
| 447 | return ListValidator(validator) |
| 448 | |
| 449 | for key, value in d.items(): |
| 450 | if type(value) is FunctionType and key.startswith('validate_'): |
| 451 | validate_functions.append(value) |
| 452 | else: |
| 453 | validator = _walk(value) |
| 454 | if validator: |
| 455 | validators[key] = validator |
| 456 | |
| 457 | d['_validator'] = DictValidator(**validators) |
| 458 | d['_validators'] = validate_functions |
| 459 | |
| 460 | return type.__new__(cls, name, bases, d) |
| 461 | |
| 462 | |
| 463 | class Form(object): |
| 464 | """Form base class. |
| 465 | |
| 466 | >>> class PersonForm(Form): |
| 467 | ... name = TextValidator(required=True) |
| 468 | ... age = IntValidator() |
| 469 | |
| 470 | >>> form = PersonForm() |
| 471 | >>> form.validate({'name': 'johnny', 'age': '42'}) |
| 472 | True |
| 473 | >>> form['name'] |
| 474 | u'johnny' |
| 475 | >>> form['age'] |
| 476 | 42 |
| 477 | |
| 478 | Let's cause a simple validation error: |
| 479 | |
| 480 | >>> form = PersonForm() |
| 481 | >>> form.validate({'name': '', 'age': 'fourty-two'}) |
| 482 | False |
| 483 | >>> print form.errors['age'][0] |
| 484 | Please enter a whole number. |
| 485 | >>> print form.errors['name'][0] |
| 486 | This field is required. |
| 487 | |
| 488 | You can also add custom validation routines independent of individual |
| 489 | fields, by adding methods that start with the prefix ``validate_``, and |
| 490 | take the data dictionary as argument. For example: |
| 491 | |
| 492 | >>> class PersonForm(Form): |
| 493 | ... name = TextValidator(required=True) |
| 494 | ... age = IntValidator() |
| 495 | ... |
| 496 | ... def validate_name_alpha(self, data): |
| 497 | ... if not data['name'].isalpha(): |
| 498 | ... raise ValidationError('The value must only contain letters') |
| 499 | |
| 500 | >>> form = PersonForm() |
| 501 | >>> form.validate({'name': 'mr.t', 'age': '42'}) |
| 502 | False |
| 503 | >>> form.errors |
| 504 | {None: ['The value must only contain letters']} |
| 505 | |
| 506 | Note that validation errors that are not related to a specific field can be |
| 507 | found under the key `None` of the errors dictionary. |
| 508 | """ |
| 509 | __metaclass__ = FormMeta |
| 510 | |
| 511 | def __init__(self, id=None, name=None, defaults=None): |
| 512 | self.id = id |
| 513 | self.name = name |
| 514 | self.data = defaults or {} |
| 515 | self.errors = {} |
| 516 | if hasattr(app, 'ctxt') and hasattr(app.ctxt, 'forms'): |
| 517 | app.ctxt.forms.append(self) |
| 518 | |
| 519 | def __contains__(self, name): |
| 520 | return name in self.data |
| 521 | |
| 522 | def __iter__(self): |
| 523 | return iter(self.data) |
| 524 | |
| 525 | def __delitem__(self, name): |
| 526 | if name in self.data: |
| 527 | del self.data[name] |
| 528 | |
| 529 | def __getitem__(self, name): |
| 530 | return self.data.get(name) |
| 531 | |
| 532 | def __setitem__(self, name, value): |
| 533 | self.data[name] = value |
| 534 | |
| 535 | @classmethod |
| 536 | def add_validator(cls, validator, name=None): |
| 537 | if name: |
| 538 | cls._validator.validators[name] = validator |
| 539 | else: |
| 540 | cls._validators.append(validator) |
| 541 | |
| 542 | def validate(self, data): |
| 543 | d = self.data.copy() |
| 544 | d.update(decode(data)) |
| 545 | errors = {} |
| 546 | try: |
| 547 | data = self._validator(data) |
| 548 | for f in self._validators: |
| 549 | f(self, data) |
| 550 | except ValidationError, e: |
| 551 | log.debug('Validation failed: %r', errors) |
| 552 | errors = e.unpack() |
| 553 | self.data.update(data) |
| 554 | self.errors = errors |
| 555 | return not errors |
| 556 | |
| 557 | |
| 558 | def form_filter(request, response, chain): |
| 559 | if 'form_token' in request.cookies: |
| 560 | expected = request.cookies['form_token'] |
| 561 | else: |
| 562 | expected = md5(str(random())).hexdigest()[:24] |
| 563 | |
| 564 | if request.method == 'POST': |
| 565 | mimetype = request.headers.get('Content-Type') |
| 566 | if mimetype: |
| 567 | mimetype, _ = parse_header(mimetype) |
| 568 | if mimetype in ('application/x-www-form-urlencoded', |
| 569 | 'multipart/form-data'): |
| 570 | actual = request.POST.get('__token__') |
| 571 | if actual != expected: |
| 572 | abort(403, gettext('Missing or invalid form token. ' |
| 573 | 'Are cookies enabled?')) |
| 574 | |
| 575 | app.ctxt.forms = [] |
| 576 | |
| 577 | stream = chain.next()(request, response, chain) |
| 578 | if isinstance(stream, Stream): |
| 579 | mimetype = response.headers['Content-Type'] |
| 580 | if mimetype: |
| 581 | mimetype, _ = parse_header(mimetype) |
| 582 | if mimetype in ('application/xhtml+xml', 'text/html'): |
| 583 | response.set_cookie('form_token', expected) |
| 584 | for form in app.ctxt.forms: |
| 585 | stream |= HTMLFormFiller(name=form.name, id=form.id, |
| 586 | data=encode(form.data)) |
| 587 | stream |= Transformer('.//form[@method="post"]').prepend(tag.div( |
| 588 | tag.input(type='hidden', name='__token__', value=expected), |
| 589 | style='display: none' |
| 590 | )) |
| 591 | |
| 592 | return stream |
Comments
No comments so far.
