#### Pollster API
## http://elections.huffingtonpost.com/pollster/api/v2
# pip install pollster
## adapted from example.py at https://github.com/huffpostdata/python-pollster
import datetime
import webbrowser
import pollster
api = pollster.Api()
## not sure what a tag is!
tags = api.tags_get() ## list of dictionary-looking objects
tags
## check out the slugs, anyway
for t in tags:
print(t.slug)
## 2016 president looks good
charts = api.charts_get(tags='2016-president')
len(charts.items)
## what's in a single chart (aka plot)
charts.items[0]
## navigate through it....
charts.items[0].question.name
## let's check out their plot
webbrowser.open(charts.items[0].url)
## next() does pagination
def run():
api = pollster.Api()
columns = [
"state", "pollster", "pop", "vtype", "method", "begmm", "begdd",
"begyy", "endmm", "enddd", "endyy", "trump", "clinton", "other",
"undecided", "Begdate", "Enddate", "Middate"
]
file = open("results.csv", "a")
file.write("{}\n".format(",".join(columns)))
results = pd.DataFrame(columns=columns)
polls_seen = set()
index = 0
try:
# Get all questions related to the 2016 presidential election
questions = api.questions_get(tags="2016-president")
for question in questions.items:
# Ignore poll questions that aren't Trump vs. Clinton
if "TrumpvClinton" not in question.slug:
continue
print("Question: {}".format(question.slug))
poll_cursor = "0"
# Grab all polls that contain this question
while True:
polls = api.polls_get(question=question.slug,
cursor=poll_cursor)
if len(polls.items) == 0:
break
poll_cursor = polls.next_cursor
for poll in polls.items:
# Ignore this poll if we've seen it before
if poll.slug in polls_seen:
continue
polls_seen.add(poll.slug)
print(" Poll: {}".format(poll.slug))
# Grab all questions in a given poll (yes, this is circular)
for poll_question in poll.poll_questions:
if poll_question.question is None or "TrumpvClinton" not in poll_question.question.slug:
continue
print(" PollQuestion: {}".format(
poll_question.question.slug))
# TODO: look at examples where there's more than one subpopulation
pop = poll_question.sample_subpopulations[0]
result = pd.DataFrame(None,
index=[index],
columns=columns)
result.state = poll_question.question.slug[3:5]
result.pollster = poll.survey_house
result.pop = pop.observations if pop.observations is not None else 1
result.vtype = pop.name
result.method = poll.mode
result.begmm = poll.start_date.month
result.begdd = poll.start_date.day
result.begyy = poll.start_date.year
result.endmm = poll.end_date.month
result.enddd = poll.end_date.day
result.endyy = poll.end_date.year
for response in pop.responses:
if response.pollster_label == "Clinton":
result.clinton = response.value
elif response.pollster_label == "Trump":
result.trump = response.value
elif response.pollster_label == "Undecided":
result.undecided = response.value
if result.clinton is None:
result.clinton = 0
if result.trump is None:
result.trump = 0
if result.undecided is None:
result.undecided = 0
result.other = 100 - result.clinton - result.trump - result.undecided
result.to_csv(file, header=False)
results.append(result)
index += 1
except pollster.rest.ApiException as e:
print(e)
file.close()
file.close()
return results
