django-rest-framework/djangorestframework/parsers.py

171 lines
6.1 KiB
Python
Raw Normal View History

2011-03-04 13:28:20 +03:00
from StringIO import StringIO
from django.http.multipartparser import MultiPartParser as DjangoMPParser
2011-02-19 13:47:26 +03:00
from djangorestframework.response import ResponseException
from djangorestframework import status
try:
import json
except ImportError:
import simplejson as json
2011-03-04 13:28:20 +03:00
try:
from urlparse import parse_qs
except ImportError:
from cgi import parse_qs
class ParserMixin(object):
parsers = ()
def parse(self, content_type, content):
# See RFC 2616 sec 3 - http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.7
split = content_type.split(';', 1)
if len(split) > 1:
content_type = split[0]
content_type = content_type.strip()
media_type_to_parser = dict([(parser.media_type, parser) for parser in self.parsers])
try:
parser = media_type_to_parser[content_type]
except KeyError:
raise ResponseException(status.HTTP_415_UNSUPPORTED_MEDIA_TYPE,
{'error': 'Unsupported media type in request \'%s\'.' % content_type})
return parser(self).parse(content)
@property
def parsed_media_types(self):
"""Return an list of all the media types that this ParserMixin can parse."""
return [parser.media_type for parser in self.parsers]
@property
def default_parser(self):
"""Return the ParerMixin's most prefered emitter.
(This has no behavioural effect, but is may be used by documenting emitters)"""
return self.parsers[0]
class BaseParser(object):
2011-01-26 11:58:09 +03:00
"""All parsers should extend BaseParser, specifing a media_type attribute,
and overriding the parse() method."""
2011-01-26 23:31:47 +03:00
media_type = None
def __init__(self, resource):
2011-01-26 11:58:09 +03:00
"""Initialise the parser with the Resource instance as state,
in case the parser needs to access any metadata on the Resource object."""
self.resource = resource
def parse(self, input):
2011-01-26 11:58:09 +03:00
"""Given some serialized input, return the deserialized output.
The input will be the raw request content body. The return value may be of
any type, but for many parsers/inputs it might typically be a dict."""
return input
class JSONParser(BaseParser):
2011-01-26 23:31:47 +03:00
media_type = 'application/json'
def parse(self, input):
try:
return json.loads(input)
except ValueError, exc:
raise ResponseException(status.HTTP_400_BAD_REQUEST, {'detail': 'JSON parse error - %s' % str(exc)})
2011-01-26 11:58:09 +03:00
class XMLParser(BaseParser):
2011-01-26 23:31:47 +03:00
media_type = 'application/xml'
class DataFlatener(object):
2011-03-10 17:49:11 +03:00
"""Utility object for flatening dictionaries of lists. Useful for "urlencoded" decoded data."""
def flatten_data(self, data):
2011-03-10 17:49:11 +03:00
"""Given a data dictionary {<key>: <value_list>}, returns a flattened dictionary
with information provided by the method "is_a_list"."""
flatdata = dict()
2011-03-09 15:07:06 +03:00
for key, attr_value in data.items():
if self.is_a_list(key):
if isinstance(attr_value, list):
2011-03-09 15:07:06 +03:00
flatdata[key] = attr_value
else:
2011-03-09 15:07:06 +03:00
flatdata[key] = [attr_value]
else:
if isinstance(attr_value, list):
2011-03-09 15:07:06 +03:00
flatdata[key] = attr_value[0]
else:
2011-03-09 15:07:06 +03:00
flatdata[key] = attr_value
return flatdata
2011-03-10 17:49:11 +03:00
def is_a_list(self, key, val):
"""Returns True if the parameter with name *key* is expected to be a list, or False otherwise.
*val* which is the received value for parameter *key* can be used to guess the answer."""
return False
class FormParser(BaseParser, DataFlatener):
"""The default parser for form data.
Return a dict containing a single value for each non-reserved parameter.
2011-03-10 17:49:11 +03:00
In order to handle select multiple (and having possibly more than a single value for each parameter),
you can customize the output by subclassing the method 'is_a_list'.
"""
# TODO: writing tests for PUT files + normal data
2011-01-26 23:31:47 +03:00
media_type = 'application/x-www-form-urlencoded'
2011-03-10 17:49:11 +03:00
"""The value of the parameter when the select multiple is empty.
Browsers are usually stripping the select multiple that have no option selected from the parameters sent.
A common hack to avoid this is to send the parameter with a value specifying that the list is empty.
This value will always be stripped before the data is returned."""
EMPTY_VALUE = '_empty'
def parse(self, input):
data = parse_qs(input)
# Flatening data and removing EMPTY_VALUEs from the lists
data = self.flatten_data(data)
for key in filter(lambda k: self.is_a_list(k), data):
self.remove_empty_val(data[key])
# Strip any parameters that we are treating as reserved
2011-03-08 18:19:55 +03:00
for key in data.keys():
2011-03-04 13:28:20 +03:00
if key in self.resource.RESERVED_FORM_PARAMS:
data.pop(key)
return data
def remove_empty_val(self, val_list):
""" """
while(1): # Because there might be several times EMPTY_VALUE in the list
try:
ind = val_list.index(self.EMPTY_VALUE)
except ValueError:
break
else:
val_list.pop(ind)
# TODO: Allow parsers to specify multiple media_types
class MultipartParser(BaseParser, DataFlatener):
media_type = 'multipart/form-data'
2011-03-04 13:28:20 +03:00
def parse(self, input):
2011-03-10 17:49:11 +03:00
request = self.resource.request
#TODO : that's pretty dumb : files are loaded with
#upload_handlers, but as we read the request body completely (input),
#then it kind of misses the point. Why not input as a stream ?
upload_handlers = request._get_upload_handlers()
django_mpp = DjangoMPParser(request.META, StringIO(input), upload_handlers)
data, files = django_mpp.parse()
# Flatening data, files and combining them
data = self.flatten_data(data)
files = self.flatten_data(files)
2011-03-04 13:28:20 +03:00
data.update(files)
# Strip any parameters that we are treating as reserved
2011-03-08 18:19:55 +03:00
for key in data.keys():
2011-03-04 13:28:20 +03:00
if key in self.resource.RESERVED_FORM_PARAMS:
data.pop(key)
return data