Consistency matters.

Redesign how information is stored and accessed. This way code can be written once. A developer using this API should not have to build contingencies just because Wolfram Alpha changes its output. The API should handle that.
These changes move the parsing of the tree out of the Wolfram API's returned XML and into Python dictionaries. This allows the full suite of tools that work with Python dictionaries to be used as well as simplifying and unifying how the responses should be handled.
In addition, I have added consistencies in certain areas that allow information to be accessed in one, unified way; regardless of how it would otherwise have been formatted by the library.
With this I want to encourage simplicity. You shouldn't have to look back through the code to figure out exactly what is being iterated over when someone decided to write iter(self) instead of simply iter(self.pods).
Also moved the Client class to the top of the file where it can be immediately seen.
master
Zenohm 8 years ago
parent bcd372031d
commit f653d1e3a9
  1. 201
      wolframalpha/__init__.py

@ -1,84 +1,179 @@
from xml.etree import ElementTree as etree
from six.moves import urllib
import xmltodict
from . import compat
compat.fix_HTTPMessage()
class Client(object):
"""
Wolfram|Alpha v2.0 client
"""
def __init__(self, app_id='Q59EW4-7K8AHE858R'):
self.app_id = app_id
def query(self, query, assumption=None):
"""
Query Wolfram|Alpha with query using the v2.0 API
"""
data = {
'input': query,
'appid': self.app_id
}
if assumption:
data.update({'assumption': assumption})
query = urllib.parse.urlencode(data)
url = 'https://api.wolframalpha.com/v2/query?' + query
resp = urllib.request.urlopen(url)
assert resp.headers.get_content_type() == 'text/xml'
assert resp.headers.get_param('charset') == 'utf-8'
return Result(resp)
class Result(object):
def __init__(self, stream):
self.tree = etree.parse(stream)
self.tree = xmltodict.parse(stream, dict_constructor=dict)['queryresult']
self._handle_error()
self.info = []
try:
self.pods = list(map(Pod, self.tree['pod']))
self.info.append(self.pods)
except KeyError:
self.pods = None
try:
self.assumptions = list(map(Assumption, self.tree['assumptions']))
self.info.append(self.assumptions)
except KeyError:
self.assumptions = None
try:
self.warnings = list(map(Warning, self.tree['warnings']))
self.info.append(self.warnings)
except KeyError:
self.warnings = None
def _handle_error(self):
error = self.tree.find('error')
if not error:
error_state = self.tree['@error']
if error_state == 'false':
return
code = error.find('code').text
msg = error.find('msg').text
tmpl = 'Error {code}: {msg}'
raise Exception(tmpl.format(code=code, msg=msg))
error = self.tree['error']
code = error['code']
msg = error['msg']
template = 'Error {code}: {msg}'
raise Exception(template.format(code=code, msg=msg))
def _flatten(self, lists):
'''
src: http://stackoverflow.com/a/952952/4241708
usr: intuited
'''
from itertools import chain
return list(chain.from_iterable(lists))
def __iter__(self):
return (Pod(node) for node in self.tree.findall('pod'))
return iter(self.info)
def __len__(self):
return len(self.tree)
@property
def pods(self):
return list(iter(self))
@property
def results(self):
return (pod for pod in self if pod.title=='Result')
return self._flatten([pod.details for pod in self.pods if pod.primary or pod.title=='Result'])
@property
def details(self):
return {pod.title: pod.details for pod in self.pods}
class Pod(object):
def __init__(self, node):
self.node = node
self.__dict__.update(node.attrib)
self._handle_error()
self.title = node['@title']
self.scanner = node['@scanner']
self.id = node['@id']
self.position = float(node['@position'])
self.error = node['@error']
self.number_of_subpods = int(node['@numsubpods'])
self.subpods = node['subpod']
# Allow subpods to be accessed in a consistent way,
# as a list, regardless of how many there are.
if type(self.subpods) != list:
self.subpods = [self.subpods]
self.subpods = list(map(Subpod, self.subpods))
self.primary = '@primary' in node and node['@primary'] != 'false'
def _handle_error(self):
error_state = self.node['@error']
if error_state == 'false':
return
error = self.tree['error']
code = error['code']
msg = error['msg']
template = 'Error {code}: {msg}'
raise Exception(template.format(code=code, msg=msg))
def __iter__(self):
return (Content(node) for node in self.node.findall('subpod'))
return iter(self.subpods)
def __len__(self):
return self.number_of_subpods
@property
def main(self):
"The main content of this pod"
return next(iter(self))
def details(self):
return [subpod.text for subpod in self.subpods]
# Needs work. At the moment this should be considered a placeholder.
class Assumption(object):
def __init__(self, node):
self.assumption = node
#self.description = self.assumption[0]['desc'] # We only care about our given assumption.
def __iter__(self):
return iter(self.assumption)
def __len__(self):
return len(self.assumption)
@property
def text(self):
return self.main.text
@property
def img(self):
return self.main.img
class Content(object):
text = self.template.replace('${desc1}', self.description)
try:
text = text.replace('${word}', self.word)
except:
pass
return text[:text.index('. ') + 1]
# Needs work. At the moment this should be considered a placeholder.
class Warning(object):
def __init__(self, node):
self.node = node
self.__dict__.update(node.attrib)
self.text = node.find('plaintext').text
self.img = node.find('img').attrib['src']
class Client(object):
"""
Wolfram|Alpha v2.0 client
"""
def __init__(self, app_id):
self.app_id = app_id
def __iter__(self):
return iter(node)
def __len__(self):
return len(node)
def query(self, query):
"""
Query Wolfram|Alpha with query using the v2.0 API
"""
query = urllib.parse.urlencode(dict(
input=query,
appid=self.app_id,
))
url = 'https://api.wolframalpha.com/v2/query?' + query
resp = urllib.request.urlopen(url)
assert resp.headers.get_content_type() == 'text/xml'
assert resp.headers.get_param('charset') == 'utf-8'
return Result(resp)
class Subpod(object):
def __init__(self, node):
self.node = node
self.title = node['@title']
self.text = node['plaintext']
self.img = node['img']
# Allow images to be accessed in a consistent way,
# as a list, regardless of how many there are.
if type(self.img) != list:
self.img = [self.img]
self.img = list(map(Image, self.img))
class Image(object):
def __init__(self, node):
self.node = node
self.title = node['@title']
self.alt = node['@alt']
self.height = node['@height']
self.width = node['@width']
self.src = node['@src']

Loading…
Cancel
Save