def add_story_screenshots(regenerate=False, article_id='storytext'):
"""
Utility. Used to generate a screenshot of every article.
pass regenerate=true to regenerate all screenshots (otherwise it'll skip
stories where that field already has a URL).
Pass an articleid to specify the CSS ID of the article. The image will be
cropped to that ID.
"""
if regenerate:
for story in Story.select():
logger.info("About to check {0}".format(story.name))
story.screenshot = screenshotter.get_story_image(story_url=story.url,
article_id=article_id)
story.save()
else:
for story in Story.select().where(Story.screenshot == None):
logger.info("About to check {0}".format(story.name))
story.screenshot = screenshotter.get_story_image(story_url=story.url,
article_id=article_id)
logger.info("Got screenshot {0}".format(story.screenshot))
story.save()
def add_story_screenshots(regenerate=False):
if regenerate:
for story in Story.select():
logger.info("About to check %s" % (story.name))
story.screenshot = screenshotter.get_story_image(story.url)
story.save()
else:
for story in Story.select().where(Story.screenshot == None):
logger.info("About to check %s" % (story.name))
story.screenshot = screenshotter.get_story_image(story.url)
story.save()
def add_story_screenshots(regenerate=False):
if regenerate:
for story in Story.select():
logger.info("About to check %s" % (story.name))
story.screenshot = screenshotter.get_story_image(story.url)
story.save()
else:
for story in Story.select().where(Story.screenshot == None):
logger.info("About to check %s" % (story.name))
story.screenshot = screenshotter.get_story_image(story.url)
story.save()
def test_change_tracking(self, mock_get_channel_name):
"""
Check if we can start tracking a URL, then update the slugs that are
tracked on it
"""
mock_get_channel_name.return_value = "default-channel"
clear_stories()
tracker = NPRStartTracking()
class FakeMessage(object):
body = {
"text": "@carebot track slug-a-b-c on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong",
"channel": "default-channel",
}
expected = "Ok, I've started tracking `slug-a-b-c` on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong"
message = tracker.respond(FakeMessage)
assert expected in message["text"]
# Now try to change the slug
FakeMessage.body[
"text"
] = "@carebot track slug-a-b-c,slug-x-y-z on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong"
message = tracker.respond(FakeMessage)
results = Story.select()
self.assertEqual(len(results), 1)
self.assertEqual(results[0].url, "http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong")
self.assertEqual(results[0].slug, "slug-a-b-c,slug-x-y-z")
def handle_overview_question(message):
message.reply(
"Let me check what's been happening. This may take a second.")
seven_days_ago = datetime.datetime.now() - datetime.timedelta(days=7)
stories = Story.select().where(Story.tracking_started > seven_days_ago)
slugs = Set()
for story in stories:
print story.name
story_slugs = story.slug.split(',')
for slug in story_slugs:
slugs.add(slug)
total_users = analytics.get_user_data(start_date='7daysAgo')
total_users = int(total_users['rows'][0][0])
total_users = "{:,}".format(total_users)
median = analytics.get_linger_rate(start_date='7daysAgo')
linger_rows = analytics.get_linger_rows(start_date='7daysAgo')
linger_histogram_url = ChartTools.linger_histogram_link(
linger_rows, median)
attachments = [{
"fallback": "linger update",
"color": "#eeeeee",
"title": "Time spent on graphics over the last week",
"image_url": linger_histogram_url
}]
slackTools.send_message(
message.body['channel'],
"In the past 7 days, I've tracked %s stories and %s graphics." %
(len(stories), len(slugs)))
slackTools.send_message(
message.body['channel'],
"%s people looked at graphics on those stories. Here's how much time they spent:"
% total_users,
attachments,
unfurl_links=False)
fields = []
for story in stories:
print "Adding %s" % story.name
fields.append({
"title": story.name.strip(),
"value": '<' + story.url + '|' + story.slug.strip() + '>',
"short": True
})
attachments = [{
"fallback": "linger update",
"color": "#eeeeee",
# "title": "What we have done",
"fields": fields
}]
slackTools.send_message(message.body['channel'],
"Here's everything:",
attachments,
unfurl_links=False)
def test_change_tracking(self, mock_get_channel_name):
"""
Check if we can start tracking a URL, then update the slugs that are
tracked on it
"""
mock_get_channel_name.return_value = 'default-channel'
clear_stories()
tracker = NPRStartTracking()
class FakeMessage(object):
body = {
'text': '@carebot track slug-a-b-c on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong',
'channel': 'default-channel'
}
expected = "Ok, I've started tracking `slug-a-b-c` on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong"
message = tracker.respond(FakeMessage)
assert expected in message['text']
# Now try to change the slug
FakeMessage.body['text'] = '@carebot track slug-a-b-c,slug-x-y-z on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong'
message = tracker.respond(FakeMessage)
results = Story.select()
self.assertEqual(len(results), 1)
self.assertEqual(results[0].url, 'http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong')
self.assertEqual(results[0].slug, 'slug-a-b-c,slug-x-y-z')
def get_story_stats():
analytics = GoogleAnalyticsScraper()
# TODO use a SQL query instead of app logic to exclude stories that are
# too old.
for story in Story.select():
logger.info("About to check %s" % (story.name))
story_time_bucket = time_bucket(story.date)
last_bucket = story.last_bucket
# Check when the story was last reported on
if last_bucket:
# Skip stories that have been checked recently
# And stories that are too old.
if (last_bucket == story_time_bucket):
logger.info("Checked recently. Bucket is still %s" % (story_time_bucket))
continue
if not story_time_bucket:
logger.info("Story is too new; skipping for now")
continue
# Some stories have multiple slugs
stats_per_slug = analytics.get_linger_data_for_story(story)
if len(stats_per_slug) is not 0:
slackTools.send_linger_time_update(story, stats_per_slug, story_time_bucket)
# Mark the story as checked
story.last_checked = datetime.datetime.now(pytz.timezone('US/Eastern'))
story.last_bucket = story_time_bucket
story.save()
def write(self, stories, team=None):
new_stories = []
for story in stories:
info_from_api = npr_api_scraper.get_story_details(story['story_url'])
if not info_from_api:
logger.info('Not adding %s to database: could not get story' % (story['story_headline']))
pass
exists = Story.select().where(Story.url == story['story_url'])
if exists:
logger.info('Not adding %s to database: already exists' % (story['story_headline']))
else:
try:
screenshot_url = screenshotter.get_story_image(story['story_url'])
story = Story.create(
name = story['story_headline'].strip(),
slug = story['graphic_slug'].strip(),
date = info_from_api['date'],
story_type = story['graphic_type'].strip(),
url = story['story_url'].strip(),
image = info_from_api['image'],
team = team,
screenshot = screenshot_url
)
new_stories.append(story)
except IntegrityError:
# Story probably already exists.
logger.info('Not adding %s to database: probably already exists' % (story['story_headline']))
pass
return new_stories
def respond(self, message):
"""
Respond to requests about the last seven days of data
TODO: Loop over all stories and report stats on each
"""
seven_days_ago = datetime.datetime.now() - datetime.timedelta(days=7)
stories = Story.select().where(Story.tracking_started > seven_days_ago)
slugs = Set()
for story in stories:
# print story.name
story_slugs = story.slug.split(',')
for slug in story_slugs:
slugs.add(slug)
try:
team = self.config.get_team_for_story(stories[0])
except:
team = self.config.get_default_team()
total_users = self.get_user_data(team=team, start_date='7daysAgo')
total_users = int(total_users['rows'][0][0])
total_users = "{:,}".format(total_users)
npr_linger = NPRLingerRate()
linger_rows = npr_linger.get_linger_data(team=team, start_date='7daysAgo')
median = NPRLingerRate.get_median(linger_rows)
linger_histogram_url = npr_linger.get_histogram_url(linger_rows, median)
attachments = [{
"fallback": "linger update",
"color": "#eeeeee",
"title": "Time spent on graphics over the last week",
"image_url": linger_histogram_url
}]
text = "In the past 7 days, I've tracked {0} stories and {1} graphics.".format(len(stories), len(slugs))
text += "\n\n"
text += "{0} people looked at graphics on the property. Here's how much time they spent:".format(total_users)
fields = []
for story in stories:
fields.append({
"title": story.name.strip(),
"value": "<{0}|{1}>".format(story.url, story.slug.strip()),
"short": True
})
attachments.append({
"fallback": "linger update",
"color": "#eeeeee",
# "title": "What we have done",
"fields": fields
})
return {
'text': text,
'attachments': attachments
}
def test_write_spreadsheet_duplicates(self, mock_upload):
mock_upload.return_value = 'http://image-url-here'
clear_stories()
scraper = SpreadsheetScraper()
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
# Insert the stories
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
# Now insert them again and make sure we don't have duplicates
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
def test_write_spreadsheet_duplicates(self, mock_upload):
mock_upload.return_value = 'http://image-url-here'
clear_stories()
scraper = SpreadsheetScraper()
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
# Insert the stories
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
# Now insert them again and make sure we don't have duplicates
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
def handle_slug_inquiry(self, message):
"""
Respond to an inquiry about the slug with stats and charts
"""
match = re.search(self.SLUG_SEARCH_REGEX, message.body["text"])
slug = match.group(1)
if slug:
# Try to match the story to a slug to accurately get a team
# The Google Analytics property ID comes from the team config
# We use the default team if none is found
stories = Story.select().where(Story.slug.contains(slug))
team = self.config.get_team_for_stories(stories)
linger_rows = self.get_linger_data(team=team, slug=slug)
if not linger_rows:
return {"text": "Sorry, I wasn't able to find linger rate stats for %s" % slug}
median = NPRLingerRate.get_median(linger_rows)
print "Got median"
print median
people = "{:,}".format(median["total_people"])
time_text = TimeTools.humanist_time_bucket(median)
reply = u"*%s* people spent a median *%s* on `%s`." % (people, time_text, slug)
reply += "\n\nThis graphic appears in %s %s I am tracking:" % (
inflector.number_to_words(len(stories)),
inflector.plural("story", len(stories)),
)
for story in stories:
reply += "\n" + "*<%s|%s>*" % (story.url, story.name.strip())
# Get linger rate data for charting.
all_graphics_rows = self.get_linger_data(team=team)
all_graphics_median = NPRLingerRate.get_median(all_graphics_rows)
attachments = [
{
"fallback": slug + " update",
"color": "#eeeeee",
"title": slug,
"image_url": self.get_histogram_url(linger_rows, median),
},
{
"fallback": slug + " update",
"color": "#eeeeee",
"title": "How all graphics performed",
"image_url": self.get_histogram_url(all_graphics_rows, all_graphics_median),
},
]
return {"text": reply, "attachments": attachments}
def respond(self, message):
m = re.search(self.START_TRACKING_REGEX, message.body['text'])
url = re.search(self.GRUBER_URLINTEXT_PAT, message.body['text'])
if not m:
return False
slug = m.group(1)
url = url.group(1)
if slug:
# Check if the slug is in the database.
try:
story = Story.select().where(Story.url.contains(url)).get()
story.slug = slug
story.save()
text = "Ok! I'm already tracking `%s`, and I've updated the slug." % url
except Story.DoesNotExist:
# If it's not in the database, start tracking it.
if not url:
logger.error("Couldn't find story URL in message %s", message.body['text'])
text = "Sorry, I need a story URL to start tracking."
return
details = npr_api_scraper.get_story_details(url)
if not details:
logger.error("Couldn't find story in API for URL %s", url)
text = "Sorry, I wasn't able to find that story in the API, so I couldn't start tracking it."
return
# Find out what team we need to save this story to
channel = slack_tools.get_channel_name(message.body['channel'])
team = self.config.get_team_for_channel(channel)
# Create the story
story = Story.create(name=details['title'],
slug=slug,
date=details['date'],
url=url,
image=details['image'],
team=team
)
story.save()
text = "Ok, I've started tracking `%s` on %s. The first stats should arrive in 4 hours or less." % (slug, url)
else:
text = "Sorry, I wasn't able to start tracking `%s` right now." % slug
if text:
return {
'text': text
}
def start_tracking(message):
m = re.search(START_TRACKING_REGEX, message.body['text'])
if not m:
return False
slug = m.group(1)
if slug:
# Check if the slug is in the database.
try:
story = Story.select().where(Story.slug.contains(slug)).get()
message.reply(
"Thanks! I'm already tracking `%s`, and you should start seeing results within a couple hours."
% slug)
except Story.DoesNotExist:
# If it's not in the database, start tracking it.
url = re.search(GRUBER_URLINTEXT_PAT, message.body['text'])
if not url:
logger.error("Couldn't find story URL in message %s",
message.body['text'])
message.reply("Sorry, I need a story URL to start tracking.")
return
details = npr_api_scraper.get_story_details(url.group(1))
if not details:
logger.error("Couldn't find story in API for URL %s",
url.group(1))
message.reply(
"Sorry, I wasn't able to find that story in the API, so I couldn't start tracking it."
)
return
# Find out what team we need to save this story to
channel = slackTools.get_channel_name(message.body['channel'])
team = config.get_team_for_channel(channel)
# Create the story
story = Story.create(name=details['title'],
slug=slug,
date=details['date'],
url=url.group(1),
image=details['image'],
team=team)
story.save()
message.reply(
"Ok, I've started tracking `%s`. The first stats should arrive in 4 hours or less."
% slug)
else:
message.reply(
"Sorry, I wasn't able to start tracking `%s` right now." % slug)
def respond(self, message):
m = re.search(self.START_TRACKING_REGEX, message.body['text'])
url = re.search(self.GRUBER_URLINTEXT_PAT, message.body['text'])
if not m:
return False
slug = m.group(1)
url = url.group(1)
if slug:
# Check if the slug is in the database.
try:
story = Story.select().where(Story.url.contains(url)).get()
story.slug = slug
story.save()
text = "Ok! I'm already tracking `%s`, and I've updated the slug." % url
except Story.DoesNotExist:
# If it's not in the database, start tracking it.
if not url:
logger.error("Couldn't find story URL in message %s",
message.body['text'])
text = "Sorry, I need a story URL to start tracking."
return
details = npr_api_scraper.get_story_details(url)
if not details:
logger.error("Couldn't find story in API for URL %s", url)
text = "Sorry, I wasn't able to find that story in the API, so I couldn't start tracking it."
return
# Find out what team we need to save this story to
channel = slack_tools.get_channel_name(message.body['channel'])
team = self.config.get_team_for_channel(channel)
# Create the story
story = Story.create(name=details['title'],
slug=slug,
date=details['date'],
url=url,
image=details['image'],
team=team)
story.save()
text = "Ok, I've started tracking `%s` on %s. The first stats should arrive in 4 hours or less." % (
slug, url)
else:
text = "Sorry, I wasn't able to start tracking `%s` right now." % slug
if text:
return {'text': text}
def add_story_screenshots(regenerate=False, article_id="storytext"):
"""
Utility. Used to generate a screenshot of every article.
pass regenerate=true to regenerate all screenshots (otherwise it'll skip
stories where that field already has a URL).
Pass an articleid to specify the CSS ID of the article. The image will be
cropped to that ID.
"""
if regenerate:
for story in Story.select():
logger.info("About to check {0}".format(story.name))
story.screenshot = screenshotter.get_story_image(story_url=story.url, article_id=article_id)
story.save()
else:
for story in Story.select().where(Story.screenshot == None):
logger.info("About to check {0}".format(story.name))
story.screenshot = screenshotter.get_story_image(story_url=story.url, article_id=article_id)
logger.info("Got screenshot {0}".format(story.screenshot))
story.save()
def handle_pageviews_inquiry(self, message):
match = re.search(self.UNSUCK_SLUG_SEARCH_REGEX, message.body["text"])
slug = match.group(1)
stories = Story.select().where(Story.slug.contains(slug))
team = config.get_team_for_stories(stories)
unique_pageviews = self.get_unique_visitor_data(team=team, slug=slug)
unique_pageviews = int(unique_pageviews)
unique_pageviews = "{:,}".format(unique_pageviews)
if not unique_pageviews:
return {"text": "Sorry, I wasn't able to find unique visitor stats for %s" % slug}
return {"text": "`{0}` has had *{1}* unique pageviews".format(slug, unique_pageviews)}
def get_story_stats():
"""
Loop through every story we know about.
If there hasn't been an update recently, fetch stats for that article.
"""
# TODO use a SQL query instead of app logic to exclude stories that are
# too old.
for story in Story.select():
logger.info("About to check %s" % (story.name))
team = config.get_team_for_story(story)
story_time_bucket = time_bucket(story.date)
last_bucket = story.last_bucket
# Check when the story was last reported on
if last_bucket:
# Skip stories that have been checked recently
# And stories that are too old.
if last_bucket == story_time_bucket:
logger.info("Checked recently. Bucket is still %s", story_time_bucket)
continue
if not story_time_bucket:
logger.info("Story is too new; skipping for now")
continue
plugins = [getattr(importlib.import_module(mod), cls) for (mod, cls) in (plugin.rsplit(".", 1) for plugin in team['plugins'])]
for plugin in plugins:
plugin = plugin()
try:
message = plugin.get_update_message(story)
if message:
slack_tools.send_message(
team['channel'],
message['text'],
message.get('attachments', None)
)
except NotImplementedError:
pass
# Mark the story as checked
story.last_checked = datetime.datetime.now(pytz.timezone(app_config.PROJECT_TIMEZONE))
story.last_bucket = story_time_bucket
story.save()
def test_write_spreadsheet(self, mock_upload):
mock_upload.return_value = 'http://image-url-here'
clear_stories()
scraper = SpreadsheetScraper()
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
for idx, story in enumerate(stories):
self.assertEqual(results[idx].name, story['story_headline'])
self.assertEqual(results[idx].url, story['story_url'])
def test_write_spreadsheet(self, mock_upload):
mock_upload.return_value = 'http://image-url-here'
clear_stories()
scraper = SpreadsheetScraper()
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
for idx, story in enumerate(stories):
self.assertEqual(results[idx].name, story['story_headline'])
self.assertEqual(results[idx].url, story['story_url'])
def get_story_stats():
"""
Loop through every story we know about.
If there hasn't been an update recently, fetch stats for that article.
"""
# TODO use a SQL query instead of app logic to exclude stories that are
# too old.
for story in Story.select():
logger.info("About to check %s" % (story.name))
team = config.get_team_for_story(story)
story_time_bucket = time_bucket(story.date)
last_bucket = story.last_bucket
# Check when the story was last reported on
if last_bucket:
# Skip stories that have been checked recently
# And stories that are too old.
if last_bucket == story_time_bucket:
logger.info("Checked recently. Bucket is still %s", story_time_bucket)
continue
if not story_time_bucket:
logger.info("Story is too new; skipping for now")
continue
plugins = [
getattr(importlib.import_module(mod), cls)
for (mod, cls) in (plugin.rsplit(".", 1) for plugin in team["plugins"])
]
for plugin in plugins:
plugin = plugin()
try:
message = plugin.get_update_message(story)
if message:
slack_tools.send_message(team["channel"], message["text"], message.get("attachments", None))
except NotImplementedError:
pass
# Mark the story as checked
story.last_checked = datetime.datetime.now(pytz.timezone(app_config.PROJECT_TIMEZONE))
story.last_bucket = story_time_bucket
story.save()
def handle_linger_update(message):
if 'doing' not in message.body['text']:
return
m = GRUBER_URLINTEXT_PAT.findall(message.body['text'])
if not m[0]:
return
url = str(m[0][0])
url = url.replace('&', '&')
logger.info("Looking for url %s" % url)
try:
story = Story.select().where(Story.url == url).get()
except:
message.reply("Sorry, I don't have stats for %s" % url)
return
story_time_bucket = story.time_bucket()
stats_per_slug = analytics.get_linger_data_for_story(story)
if len(stats_per_slug) is not 0:
reply = ("Here's what I know about the graphics on _%s_:") % (
story.name.strip())
fields = []
for stat in stats_per_slug:
time = TimeTools.humanist_time_bucket(stat['stats'])
fields.append({
"title": stat['slug'],
"value": time,
"short": True
})
attachments = [{
"fallback": story.name + " update",
"color": "#eeeeee",
"title": story.name,
"title_link": story.url,
"fields": fields
}]
# Use send_message instead of message.reply, otherwise we lose
# the bot icon.
slackTools.send_message(message.body['channel'], reply, attachments)
def test_write_spreadsheet(self, mock_upload):
"""
Make sure we save the stories to the database when scraping from a
spreadsheet
"""
clear_stories()
scraper = SpreadsheetScraper(self.source)
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
for idx, story in enumerate(stories):
self.assertEqual(results[idx].name, story['story_headline'])
self.assertEqual(results[idx].url, story['story_url'])
def test_write_spreadsheet(self, mock_upload):
"""
Make sure we save the stories to the database when scraping from a
spreadsheet
"""
clear_stories()
scraper = SpreadsheetScraper(self.source)
stories = scraper.scrape_spreadsheet('tests/data/stories.xlsx')
scraper.write(stories)
results = Story.select()
self.assertEqual(len(results), 4)
for idx, story in enumerate(stories):
self.assertEqual(results[idx].name, story['story_headline'])
self.assertEqual(results[idx].url, story['story_url'])
def handle_scroll_slug_question(message):
m = re.search(SCROLL_RATE_REGEX, message.body['text'])
if not m:
return
slug = m.group(1)
if slug:
stories = Story.select().where(Story.slug.contains(slug))
rows = analytics.get_depth_rate(slug)
if rows:
reply = u"Here's what I know about `%s`." % slug
reply += '\n\nThis graphic appears in %s %s:' % (
inflector.number_to_words(
len(stories)), inflector.plural('story', len(stories)))
for story in stories:
reply += '\n' + '*<%s|%s>*' % (story.url, story.name.strip())
histogram_url = ChartTools.scroll_histogram_link(rows)
if story.screenshot:
histogram_url = ChartTools.add_screenshot_to_chart(
story.screenshot, histogram_url)
attachments = [{
"fallback": slug + " update",
"color": "#eeeeee",
"title": slug,
"image_url": histogram_url
}]
slackTools.send_message(message.body['channel'],
reply,
attachments,
unfurl_links=False)
else:
message.reply("I wasn't able to find scroll data for %s" % slug)
def handle_url_inquiry(self, message):
"""
Respond to "How is http://example.com/foo doing?"
"""
if "doing" not in message.body["text"]:
return
match = self.GRUBER_URLINTEXT_PAT.findall(message.body["text"])
if not match[0]:
return
url = str(match[0][0])
url = url.replace("&", "&")
logger.info("Looking for url %s" % url)
try:
story = Story.select().where(Story.url == url).get()
except:
return {"text": "Sorry, I don't have stats for %s" % url}
return self.get_update_message(story)
def handle_url_inquiry(self, message):
"""
Respond to "How is http://example.com/foo doing?"
"""
if 'doing' not in message.body['text']:
return
match = self.GRUBER_URLINTEXT_PAT.findall(message.body['text'])
if not match[0]:
return
url = str(match[0][0])
url = url.replace('&', '&')
logger.info("Looking for url %s" % url)
try:
story = Story.select().where(Story.url == url).get()
except:
return {'text': "Sorry, I don't have stats for %s" % url}
return self.get_update_message(story)
def get_slug_message(self, slug, story=None):
# Try to match the story to a slug to accurately get a team
# The Google Analytics property ID comes from the team config
# We use the default team if none is found
stories = Story.select().where(Story.slug.contains(slug))
team = self.config.get_team_for_stories(stories)
params = self.get_slug_query_params(team=team, slug=slug)
data = GoogleAnalytics.query_ga(params)
if not data.get('rows'):
logger.info('No rows found for slug %s' % slug)
return
# Clean up the data
clean_data = self.clean_data(data.get('rows'))
total_people = self.get_total_people(clean_data)
friendly_people = "{:,}".format(total_people) # Comma-separated #s
median = self.get_median(clean_data)
# Set up the chart
scroll_histogram_url = self.get_chart(clean_data)
if story:
scroll_histogram_url = ChartTools.add_screenshot_to_chart(story,
scroll_histogram_url)
# TODO: Not confident in median calculations so far
# text = "*%s people* got a median of *%s percent* down the page." % (friendly_people, median)
text = ''
attachments = [{
"fallback": slug + " update",
"color": "#eeeeee",
"title": "How far down did people scroll?",
"image_url": scroll_histogram_url
}]
return {
'text': text,
'attachments': attachments
}
def test_start_tracking(self, mock_get_channel_name):
"""
Test if we can start tracking a new story given only a NPR URL and a
graphic slug
"""
mock_get_channel_name.return_value = "default-channel"
clear_stories()
tracker = NPRStartTracking()
class FakeMessage(object):
body = {
"text": "@carebot track slug-a-b-c on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong",
"channel": "default-channel",
}
expected = "Ok, I've started tracking `slug-a-b-c` on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong"
message = tracker.respond(FakeMessage)
print message
assert expected in message["text"]
results = Story.select()
self.assertEqual(len(results), 1)
self.assertEqual(results[0].url, "http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong")
def handle_pageviews_inquiry(self, message):
match = re.search(self.UNSUCK_SLUG_SEARCH_REGEX, message.body['text'])
slug = match.group(1)
stories = Story.select().where(Story.slug.contains(slug))
team = config.get_team_for_stories(stories)
unique_pageviews = self.get_unique_visitor_data(team=team, slug=slug)
unique_pageviews = int(unique_pageviews)
unique_pageviews = "{:,}".format(unique_pageviews)
if not unique_pageviews:
return {
'text':
"Sorry, I wasn't able to find unique visitor stats for %s" %
slug
}
return {
'text':
'`{0}` has had *{1}* unique pageviews'.format(
slug, unique_pageviews)
}
def test_start_tracking(self, mock_get_channel_name):
"""
Test if we can start tracking a new story given only a NPR URL and a
graphic slug
"""
mock_get_channel_name.return_value = 'default-channel'
clear_stories()
tracker = NPRStartTracking()
class FakeMessage(object):
body = {
'text': '@carebot track slug-a-b-c on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong',
'channel': 'default-channel'
}
expected = "Ok, I've started tracking `slug-a-b-c` on http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong"
message = tracker.respond(FakeMessage)
print message
assert expected in message['text']
results = Story.select()
self.assertEqual(len(results), 1)
self.assertEqual(results[0].url, 'http://www.npr.org/sections/13.7/2016/02/16/466109612/was-einstein-wrong')
def write(self, stories, team=None):
new_stories = []
for story in stories:
info_from_api = npr_api_scraper.get_story_details(
story['story_url'])
if not info_from_api:
logger.info('Not adding %s to database: could not get story' %
(story['story_headline']))
pass
exists = Story.select().where(Story.url == story['story_url'])
if exists:
logger.info('Not adding %s to database: already exists' %
(story['story_headline']))
else:
try:
screenshot_url = screenshotter.get_story_image(
story['story_url'])
story = Story.create(
name=story['story_headline'].strip(),
slug=story['graphic_slug'].strip(),
date=info_from_api['date'],
story_type=story['graphic_type'].strip(),
url=story['story_url'].strip(),
image=info_from_api['image'],
team=team,
screenshot=screenshot_url)
new_stories.append(story)
except IntegrityError:
# Story probably already exists.
logger.info(
'Not adding %s to database: probably already exists' %
(story['story_headline']))
pass
return new_stories
def get_story_stats():
analytics = GoogleAnalyticsScraper()
# TODO use a SQL query instead of app logic to exclude stories that are
# too old.
for story in Story.select():
logger.info("About to check %s" % (story.name))
story_time_bucket = time_bucket(story.date)
last_bucket = story.last_bucket
# Check when the story was last reported on
if last_bucket:
# Skip stories that have been checked recently
# And stories that are too old.
if (last_bucket == story_time_bucket):
logger.info("Checked recently. Bucket is still %s" %
(story_time_bucket))
continue
if not story_time_bucket:
logger.info("Story is too new; skipping for now")
continue
# Some stories have multiple slugs
stats_per_slug = analytics.get_linger_data_for_story(story)
if len(stats_per_slug) is not 0:
slackTools.send_linger_time_update(story, stats_per_slug,
story_time_bucket)
# Mark the story as checked
story.last_checked = datetime.datetime.now(pytz.timezone('US/Eastern'))
story.last_bucket = story_time_bucket
story.save()
def handle_slug_question(message):
m = re.search(LINGER_RATE_REGEX, message.body['text'])
if not m:
return
slug = m.group(1)
if slug:
median = analytics.get_linger_rate(slug)
stories = Story.select().where(Story.slug.contains(slug))
message.reply("Ok! I'm looking up %s. This may take a second." % slug)
if median:
people = "{:,}".format(median['total_people'])
time_text = TimeTools.humanist_time_bucket(median)
reply = u"*%s* people spent a median *%s* on `%s`." % (
people, time_text, slug)
# List the stories this slug appears on
reply += '\n\nThis graphic appears in %s %s:' % (
inflector.number_to_words(
len(stories)), inflector.plural('story', len(stories)))
for story in stories:
reply += '\n' + '*<%s|%s>*' % (story.url, story.name.strip())
# Get linger rate data
linger_rows = analytics.get_linger_rows(slug)
linger_histogram_url = ChartTools.linger_histogram_link(
linger_rows, median)
all_graphics_rows = analytics.get_linger_rows()
all_graphics_median = analytics.get_linger_rate()
all_histogram = ChartTools.linger_histogram_link(
all_graphics_rows, all_graphics_median)
attachments = [{
"fallback": slug + " update",
"color": "#eeeeee",
"title": slug,
"image_url": linger_histogram_url
}, {
"fallback": slug + " update",
"color": "#eeeeee",
"title": "How all graphics performed",
"image_url": all_histogram
}]
# Get scroll data, if any.
scroll_depth_rows = analytics.get_depth_rate(slug)
if scroll_depth_rows:
scroll_histogram_url = ChartTools.scroll_histogram_link(
scroll_depth_rows)
if stories[0].screenshot:
scroll_histogram_url = ChartTools.add_screenshot_to_chart(
stories[0].screenshot, scroll_histogram_url)
attachments.append({
"fallback": slug + " update",
"color": "#eeeeee",
"title": "How far down did people scroll?",
"image_url": scroll_histogram_url
})
slackTools.send_message(message.body['channel'],
reply,
attachments,
unfurl_links=False)
else:
message.reply("I wasn't able to figure out the linger rate of %s" %
slug)
def handle_slug_inquiry(self, message):
"""
Respond to an inquiry about the slug with stats and charts
"""
match = re.search(self.SLUG_SEARCH_REGEX, message.body['text'])
slug = match.group(1)
if slug:
# Try to match the story to a slug to accurately get a team
# The Google Analytics property ID comes from the team config
# We use the default team if none is found
stories = Story.select().where(Story.slug.contains(slug))
team = self.config.get_team_for_stories(stories)
linger_rows = self.get_linger_data(team=team, slug=slug)
if not linger_rows:
return {
'text':
"Sorry, I wasn't able to find linger rate stats for %s" %
slug
}
median = NPRLingerRate.get_median(linger_rows)
print "Got median"
print median
people = "{:,}".format(median['total_people'])
time_text = TimeTools.humanist_time_bucket(median)
reply = u"*%s* people spent a median *%s* on `%s`." % (
people, time_text, slug)
reply += '\n\nThis graphic appears in %s %s I am tracking:' % (
inflector.number_to_words(
len(stories)), inflector.plural('story', len(stories)))
for story in stories:
reply += '\n' + '*<%s|%s>*' % (story.url, story.name.strip())
# Get linger rate data for charting.
all_graphics_rows = self.get_linger_data(team=team)
all_graphics_median = NPRLingerRate.get_median(all_graphics_rows)
attachments = [{
"fallback":
slug + " update",
"color":
"#eeeeee",
"title":
slug,
"image_url":
self.get_histogram_url(linger_rows, median)
}, {
"fallback":
slug + " update",
"color":
"#eeeeee",
"title":
"How all graphics performed",
"image_url":
self.get_histogram_url(all_graphics_rows, all_graphics_median)
}]
return {'text': reply, 'attachments': attachments}
def respond(self, message):
"""
Respond to requests about the last seven days of data
TODO: Loop over all stories and report stats on each
"""
seven_days_ago = datetime.datetime.now() - datetime.timedelta(days=7)
stories = Story.select().where(Story.tracking_started > seven_days_ago)
slugs = Set()
for story in stories:
# print story.name
story_slugs = story.slug.split(',')
for slug in story_slugs:
slugs.add(slug)
try:
team = self.config.get_team_for_story(stories[0])
except:
team = self.config.get_default_team()
total_users = self.get_user_data(team=team, start_date='7daysAgo')
total_users = int(total_users['rows'][0][0])
total_users = "{:,}".format(total_users)
npr_linger = NPRLingerRate()
linger_rows = npr_linger.get_linger_data(team=team,
start_date='7daysAgo')
median = NPRLingerRate.get_median(linger_rows)
linger_histogram_url = npr_linger.get_histogram_url(
linger_rows, median)
attachments = [{
"fallback": "linger update",
"color": "#eeeeee",
"title": "Time spent on graphics over the last week",
"image_url": linger_histogram_url
}]
text = "In the past 7 days, I've tracked {0} stories and {1} graphics.".format(
len(stories), len(slugs))
text += "\n\n"
text += "{0} people looked at graphics on the property. Here's how much time they spent:".format(
total_users)
fields = []
for story in stories:
fields.append({
"title":
story.name.strip(),
"value":
"<{0}|{1}>".format(story.url, story.slug.strip()),
"short":
True
})
attachments.append({
"fallback": "linger update",
"color": "#eeeeee",
# "title": "What we have done",
"fields": fields
})
return {'text': text, 'attachments': attachments}