def run(self):
access_token = os.getenv('FB_ACCESS_TOKEN')
if not access_token:
raise EnvironmentError("FB Access token is not set")
with self.input().open('r') as facts_file:
facts = json.load(facts_file)
page_id = facts['ids']['instagram']['pageId']
all_media = []
fields = ','.join(self.columns.keys())
# use limit=100 to keep amount of requests small
# 100 is the maximum value the Graph API will accept
limit = 100
media_url = (f'{API_BASE}/{page_id}/media'
f'?fields={fields}&limit={limit}')
response = try_request_multiple_times(media_url)
response_json = response.json()
current_count = len(response_json['data'])
all_media.extend(response_json['data'])
logger.info("Fetching Instagram posts ...")
while 'next' in response_json['paging']:
next_url = response_json['paging']['next']
response = try_request_multiple_times(next_url)
response_json = response.json()
current_count += len(response_json['data'])
if sys.stdout.isatty():
print(
f"\rFetched {current_count} Instagram posts",
end='',
flush=True)
for media in response_json['data']:
all_media.append(media)
if self.minimal_mode:
logger.info("Running in minimal mode, stopping now")
response_json['paging'].pop('next')
if sys.stdout.isatty():
print() # have to manually print newline
logger.info("Fetching of Instagram posts complete")
df = pd.DataFrame([
{
column: adapter(media[column])
for (column, adapter)
in self.columns.items()
}
for media
in all_media
])
with self.output().open('w') as output_file:
df.to_csv(output_file, index=False, header=True)
def run(self):
with self.input().open('r') as facts_file:
facts = json.load(facts_file)
page_id = facts['ids']['instagram']['pageId']
df = pd.DataFrame(columns=self.columns)
metrics = ','.join([
'impressions', 'reach', 'profile_views', 'follower_count',
'website_clicks'
])
period = 'day'
url = f'{API_BASE}/{page_id}/insights?metric={metrics}&period={period}'
response = try_request_multiple_times(url)
response_data = response.json()['data']
timestamp = response_data[0]['values'][0]['end_time']
impressions = response_data[0]['values'][0]['value']
reach = response_data[1]['values'][0]['value']
profile_views = response_data[2]['values'][0]['value']
follower_count = response_data[3]['values'][0]['value']
website_clicks = response_data[4]['values'][0]['value']
df.loc[0] = [
timestamp, impressions, reach, profile_views, follower_count,
website_clicks
]
with self.output().open('w') as output_file:
df.to_csv(output_file, index=False, header=True)
def run(self):
with self.input().open('r') as facts_file:
facts = json.load(facts_file)
page_id = facts['ids']['instagram']['pageId']
df = pd.DataFrame(columns=self.columns)
fields = ','.join([
'followers_count',
'media_count'
])
url = f'{API_BASE}/{page_id}?fields={fields}'
response = try_request_multiple_times(url)
response_data = response.json()
timestamp = dt.datetime.now()
follower_count = response_data.get('followers_count')
media_count = response_data.get('media_count')
df.loc[0] = [
timestamp,
follower_count,
media_count
]
with self.output().open('w') as output_file:
df.to_csv(output_file, index=False, header=True)
def run(self):
with self.input().open('r') as facts_file:
facts = json.load(facts_file)
page_id = facts['ids']['instagram']['pageId']
df = pd.DataFrame(columns=self.columns)
metrics = ','.join([
'impressions',
'reach',
'profile_views',
'follower_count',
'website_clicks'
])
period = 'day'
url = f'{API_BASE}/{page_id}/insights?metric={metrics}&period={period}'
response = try_request_multiple_times(url)
response_data = response.json()['data']
timestamp = response_data[0]['values'][0]['end_time']
metrics = self.extract_metrics(response_data)
df = df.append({'timestamp': timestamp, **metrics}, ignore_index=True)
with self.output().open('w') as output_file:
df.to_csv(output_file, index=False, header=True)
def run(self):
with self.input().open('r') as input_file:
post_df = pd.read_csv(input_file)
if self.minimal_mode:
post_df = post_df.head(5)
generic_metrics = ['impressions', 'reach', 'engagement', 'saved']
performance_df = pd.DataFrame(columns=[
column for column in self.columns
if not column.startswith('delta_')
])
fetch_time = dt.datetime.now()
for i, row in self.tqdm(post_df.iterrows(),
desc="Fetching insights for instagram posts",
total=len(post_df)):
# Fetch only insights for less than 2 months old posts
post_time = dtparser.parse(row['timestamp'])
if post_time.date() < \
fetch_time.date() - self.timespan:
continue
metrics = ','.join(generic_metrics)
if row['media_type'] == 'VIDEO':
metrics += ',video_views' # causes error if used on non-video
url = f'{API_BASE}/{row["id"]}/insights?metric={metrics}'
response = try_request_multiple_times(url)
response_data = response.json()['data']
impressions = response_data[0]['values'][0]['value']
reach = response_data[1]['values'][0]['value']
engagement = response_data[2]['values'][0]['value']
saved = response_data[3]['values'][0]['value']
video_views = response_data[4]['values'][0]['value']\
if row['media_type'] == 'VIDEO'\
else 0 # for non-video posts
performance_df.loc[i] = [
str(row['id']), # The type was lost during CSV conversion
fetch_time,
impressions,
reach,
engagement,
saved,
video_views
]
performance_df = self.filter_fkey_violations(performance_df)
performance_df = self.condense_performance_values(
performance_df,
delta_function=PerformanceValueCondenser.linear_delta)
with self.output().open('w') as output_file:
performance_df.to_csv(output_file, index=False, header=True)
def _get_single_metric(page_id, metric, period='lifetime'):
url = f'{API_BASE}/{page_id}/insights?metric={metric}&period={period}'
res = try_request_multiple_times(url)
return res.json()['data'][0]['values'][0]['value']