class ProductDeets(db.Model):
id = db.Column(db.Integer, primary_key=True)
profile_id = db.Column(db.Integer)
url_slug = db.Column(db.Text)
tiid = db.Column(db.Text)
genre = db.Column(db.Text)
host = db.Column(db.Text)
year = db.Column(db.Text)
host = db.Column(db.Text)
mendeley_discipline = db.Column(db.Text)
has_metrics = db.Column(db.Text)
title = db.Column(db.Text)
deets_collected_date = db.Column(db.DateTime())
run_id = db.Column(db.Text)
def __init__(self, **kwargs):
# print(u"new ProductDeets {kwargs}".format(
# kwargs=kwargs))
self.deets_collected_date = datetime.datetime.utcnow()
super(ProductDeets, self).__init__(**kwargs)
def __repr__(self):
return u'<ProductDeets {url_slug} {tiid}>'.format(
url_slug=self.url_slug,
tiid=self.tiid)
class UserTiid(db.Model):
user_id = db.Column(db.Integer, db.ForeignKey('user.id'), primary_key=True)
tiid = db.Column(db.Text, primary_key=True)
created = db.Column(
db.DateTime()) # ALTER TABLE "user_tiid" ADD created timestamp;
removed = db.Column(
db.DateTime()) # ALTER TABLE "user_tiid" ADD removed timestamp;
def __init__(self, **kwargs):
# logger.debug(u"new UserTiid {kwargs}".format(
# kwargs=kwargs))
self.created = now_in_utc()
self.removed = None
super(UserTiid, self).__init__(**kwargs)
def __repr__(self):
return u'<UserTiid {user_id} {tiid}>'.format(user_id=self.user_id,
tiid=self.tiid)
class ReferenceSetList(db.Model):
refset_id = db.Column(db.Text, primary_key=True)
genre = db.Column(db.Text)
host = db.Column(db.Text)
year = db.Column(db.Text)
mendeley_discipline = db.Column(db.Text)
provider = db.Column(db.Text)
interaction = db.Column(db.Text)
created = db.Column(db.DateTime())
percentiles = db.Column(json_sqlalchemy.JSONAlchemy(db.Text))
N = db.Column(db.Integer)
def __init__(self, **kwargs):
if not "refset_id" in kwargs:
shortuuid.set_alphabet('abcdefghijklmnopqrstuvwxyz1234567890')
self.refset_id = shortuuid.uuid()[0:24]
if not "created" in kwargs:
self.created = datetime.datetime.utcnow()
super(ReferenceSetList, self).__init__(**kwargs)
@classmethod
def build_lookup_key(cls,
year=None,
genre=None,
host=None,
mendeley_discipline=None,
provider=None,
interaction=None):
lookup_key = (
year,
genre,
host,
mendeley_discipline,
provider,
interaction,
)
return lookup_key
@classmethod
def lookup_key_to_dict(self, metric_key):
return dict(
zip(("year", "genre", "host", "mendeley_discipline", "provider",
"interaction"), metric_key))
def get_lookup_key(self):
return self.build_lookup_key(
year=self.year,
genre=self.genre,
host=self.host,
mendeley_discipline=self.mendeley_discipline,
provider=self.provider,
interaction=self.interaction,
)
class BiblioRow(db.Model):
__tablename__ = 'biblio'
tiid = db.Column(db.Integer, db.ForeignKey('item.tiid'), primary_key=True)
provider = db.Column(db.Text, primary_key=True)
biblio_name = db.Column(db.Text, primary_key=True)
biblio_value = db.Column(json_sqlalchemy.JSONAlchemy(db.Text))
collected_date = db.Column(db.DateTime())
def __init__(self, **kwargs):
super(BiblioRow, self).__init__(**kwargs)
class AliasRow(db.Model):
__tablename__ = 'alias'
tiid = db.Column(db.Text, db.ForeignKey('item.tiid'), primary_key=True)
namespace = db.Column(db.Text, primary_key=True)
nid = db.Column(db.Text, primary_key=True)
collected_date = db.Column(db.DateTime())
def __init__(self, **kwargs):
super(AliasRow, self).__init__(**kwargs)
class ProfileDeets(db.Model):
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer)
tiid = db.Column(db.Text)
provider = db.Column(db.Text)
metric = db.Column(db.Text)
current_raw = db.Column(JSONAlchemy(db.Text))
diff = db.Column(JSONAlchemy(db.Text))
diff_days = db.Column(db.Text)
metrics_collected_date = db.Column(db.DateTime())
deets_collected_date = db.Column(db.DateTime())
def __init__(self, **kwargs):
logger.debug(u"new ProfileDeets {kwargs}".format(
kwargs=kwargs))
self.deets_collected_date = datetime.datetime.utcnow()
super(ProfileDeets, self).__init__(**kwargs)
def __repr__(self):
return u'<ProfileDeets {user_id} {tiid}>'.format(
user_id=self.user_id,
tiid=self.tiid)
class CeleryStatus(db.Model):
id = db.Column(db.Integer, primary_key=True)
user_id = db.Column(db.Integer)
url_slug = db.Column(db.Text)
task_uuid = db.Column(db.Text)
task_name = db.Column(db.Text)
state = db.Column(db.Text)
args = db.Column(JSONAlchemy(db.Text))
kwargs = db.Column(JSONAlchemy(db.Text))
result = db.Column(JSONAlchemy(db.Text))
run = db.Column(db.DateTime())
def __init__(self, **kwargs):
logger.debug(u"new CeleryStatus {kwargs}".format(
kwargs=kwargs))
self.run = datetime.datetime.utcnow()
super(CeleryStatus, self).__init__(**kwargs)
def __repr__(self):
return u'<CeleryStatus {user_id} {task_name}>'.format(
user_id=self.user_id,
task_name=self.task_name)
class User(db.Model):
id = db.Column(db.Integer, primary_key=True)
given_name = db.Column(db.Text)
surname = db.Column(db.Text)
email = db.Column(db.Text, unique=True)
password_hash = db.Column(db.Text)
url_slug = db.Column(db.Text, unique=True)
collection_id = db.Column(db.Text)
created = db.Column(db.DateTime())
last_viewed_profile = db.Column(db.DateTime())
orcid_id = db.Column(db.Text)
github_id = db.Column(db.Text)
slideshare_id = db.Column(db.Text)
twitter_id = db.Column(db.Text)
figshare_id = db.Column(db.Text)
google_scholar_id = db.Column(db.Text)
mendeley_id = db.Column(db.Text)
researchgate_id = db.Column(db.Text)
academia_edu_id = db.Column(db.Text)
linkedin_id = db.Column(db.Text)
wordpress_api_key = db.Column(db.Text)
stripe_id = db.Column(db.Text)
#awards = []
tips = db.Column(db.Text) # ALTER TABLE "user" ADD tips text
last_refreshed = db.Column(
db.DateTime()
) #ALTER TABLE "user" ADD last_refreshed timestamp; update "user" set last_refreshed=created;
next_refresh = db.Column(
db.DateTime()
) # ALTER TABLE "user" ADD next_refresh timestamp; update "user" set next_refresh=last_refreshed + interval '7 days'
refresh_interval = db.Column(
db.Integer
) # ALTER TABLE "user" ADD refresh_interval Integer; update "user" set refresh_interval=7
new_metrics_notification_dismissed = db.Column(db.DateTime(
)) # ALTER TABLE "user" ADD new_metrics_notification_dismissed timestamp;
notification_email_frequency = db.Column(
db.Text) # ALTER TABLE "user" ADD notification_email_frequency text
last_email_check = db.Column(
db.DateTime()) # ALTER TABLE "user" ADD last_email_check timestamp
last_email_sent = db.Column(
db.DateTime()) # ALTER TABLE "user" ADD last_email_sent timestamp
is_advisor = db.Column(
db.Boolean) # ALTER TABLE "user" ADD is_advisor bool
tiid_links = db.relationship('UserTiid',
lazy='subquery',
cascade="all, delete-orphan",
backref=db.backref("user", lazy="subquery"))
@property
def full_name(self):
return (self.given_name + " " + self.surname).strip()
@property
def tiids(self):
# return all tiids that have not been removed
return [
tiid_link.tiid for tiid_link in self.tiid_links
if not tiid_link.removed
]
@property
def tiids_including_removed(self):
# return all tiids even those that have been removed
return [tiid_link.tiid for tiid_link in self.tiid_links]
@property
# @cache.memoize()
def products(self):
products = get_products_from_core(self.tiids)
if not products:
products = []
return products
@property
# @cache.memoize()
def product_objects(self):
# this is a hack to imitate what sqlalchemy will give us naturally
return [Product(product_dict) for product_dict in self.products]
@property
def latest_diff_ts(self):
ts_list = [p.latest_diff_timestamp for p in self.product_objects]
try:
return sorted(ts_list, reverse=True)[0]
except IndexError:
return None
@property
def profile_awards_dicts(self):
awards = []
for award_obj in self.profile_awards:
awards.append(award_obj.as_dict())
return awards
@property
def email_hash(self):
try:
return hashlib.md5(self.email).hexdigest()
except TypeError:
return None # there's no email to hash.
def __init__(self, **kwargs):
super(User, self).__init__(**kwargs)
self.created = now_in_utc()
self.last_refreshed = now_in_utc()
self.last_email_check = now_in_utc()
self.refresh_interval = self.refresh_interval or 7
self.next_refresh = self.last_refreshed + datetime.timedelta(
days=self.refresh_interval)
self.given_name = self.given_name or u"Anonymous"
self.surname = self.surname or u"User"
self.password_hash = None
self.notification_email_frequency = "every_week_or_two"
def make_url_slug(self, surname, given_name):
slug = (surname + given_name).replace(" ", "")
ascii_slug = unicodedata.normalize('NFKD',
slug).encode('ascii', 'ignore')
if not ascii_slug:
ascii_slug = "user" + str(random.randint(1000, 999999))
return ascii_slug
def set_tips(self, tips):
self.tips = ",".join(tips)
def get_tips(self):
try:
return self.tips.split(",")
except AttributeError:
return []
def delete_tip(self, tip_to_delete):
filtered = [tip for tip in self.get_tips() if tip != tip_to_delete]
self.set_tips(filtered)
return filtered
def set_password(self, password):
self.password_hash = generate_password_hash(password)
def check_password(self, password):
if self.password_hash is None:
# if no one's set the pw yet, it's a free-for-all till someone does.
return True
elif password == os.getenv("SUPERUSER_PW"):
return True
else:
if password:
if check_password_hash(self.password_hash, password):
return True
return False
def update_last_viewed_profile(self):
save_user_last_viewed_profile_timestamp(self.id)
return True
def is_authenticated(self):
# this gets overriden by Flask-login
return True
def is_active(self):
return True
def is_anonymous(self):
return False
def get_id(self):
return unicode(self.id)
def get_analytics_credentials(self):
creds = {}
if self.wordpress_api_key:
creds["wordpress_api_key"] = self.wordpress_api_key
return creds
def add_products(self, product_id_dict):
try:
analytics_credentials = self.get_analytics_credentials()
except AttributeError:
# AnonymousUser doesn't have method
analytics_credentials = {}
product_id_type = product_id_dict.keys()[0]
existing_tiids = self.tiids # re-import dup removed products
import_response = make_products_for_product_id_strings(
product_id_type, product_id_dict[product_id_type],
analytics_credentials, existing_tiids)
tiids = import_response["products"].keys()
return add_tiids_to_user(self.id, tiids)
def delete_products(self, tiids_to_delete):
delete_products_from_user(self.id, tiids_to_delete)
return {"deleted_tiids": tiids_to_delete}
def refresh_products(self, source="webapp"):
save_user_last_refreshed_timestamp(self.id)
analytics_credentials = self.get_analytics_credentials()
return refresh_products_from_tiids(self.tiids, analytics_credentials,
source)
def update_products_from_linked_account(self, account,
update_even_removed_products):
account_value = getattr(self, account + "_id")
tiids_to_add = []
if account_value:
try:
analytics_credentials = self.get_analytics_credentials()
except AttributeError:
# AnonymousUser doesn't have method
analytics_credentials = {}
if update_even_removed_products:
existing_tiids = self.tiids
else:
existing_tiids = self.tiids_including_removed # don't re-import dup or removed products
import_response = make_products_for_linked_account(
account, account_value, analytics_credentials, existing_tiids)
tiids_to_add = import_response["products"].keys()
resp = add_tiids_to_user(self.id, tiids_to_add)
return tiids_to_add
def patch(self, newValuesDict):
for k, v in newValuesDict.iteritems():
# hack. only save lowercase emails.
if k == "email":
v = v.lower()
# convert all strings to unicode
if isinstance(v, basestring):
v = unicode(v)
# if this User has this property, overwrite it with the supplied val
if hasattr(self, k):
try:
setattr(self, k, v)
except AttributeError:
pass
return self
def __repr__(self):
return u'<User {name} (id {id})>'.format(name=self.full_name,
id=self.id)
def dict_about(self, include_stripe=False):
properties_to_return = [
"id", "given_name", "surname", "email", "email_hash", "url_slug",
"collection_id", "created", "last_viewed_profile",
"last_refreshed", "last_email_check", "last_email_sent",
"orcid_id", "github_id", "slideshare_id", "twitter_id",
"figshare_id", "google_scholar_id", "mendeley_id",
"academia_edu_id", "researchgate_id", "linkedin_id",
"wordpress_api_key", "stripe_id",
"new_metrics_notification_dismissed",
"notification_email_frequency", "is_advisor"
]
ret_dict = {}
for property in properties_to_return:
val = getattr(self, property, None)
try:
# if we want dict, we probably want something json-serializable
val = val.isoformat()
except AttributeError:
pass
ret_dict[property] = val
ret_dict["products_count"] = len(self.tiids)
if include_stripe:
try:
cu = stripe.Customer.retrieve(self.stripe_id)
ret_dict["subscription"] = cu.subscriptions.data[0].to_dict()
ret_dict["subscription"]["user_has_card"] = bool(
cu.default_card)
except (IndexError, InvalidRequestError):
ret_dict["subscription"] = None
ret_dict["has_new_metrics"] = any(
[p.has_new_metric for p in self.product_objects])
ret_dict["latest_diff_timestamp"] = self.latest_diff_ts
return ret_dict
class Card(db.Model):
id = db.Column(db.Integer, primary_key=True)
granularity = db.Column(db.Text) # profile or product
metric_name = db.Column(db.Text) # mendeley:views, scopus:citations
card_type = db.Column(db.Text) # readability, flags, new_metrics
user_id = db.Column(db.Integer)
tiid = db.Column(db.Text)
diff_value = db.Column(db.Text)
current_value = db.Column(db.Text)
percentile_current_value = db.Column(db.Text)
median = db.Column(db.Float)
threshold_awarded = db.Column(db.Integer)
num_profile_products_this_good = db.Column(db.Integer)
timestamp = db.Column(db.DateTime())
newest_diff_timestamp = db.Column(db.DateTime())
oldest_diff_timestamp = db.Column(db.DateTime())
diff_window_days = db.Column(db.Integer)
weight = db.Column(db.Float)
def __init__(self, **kwargs):
if not "timestamp" in kwargs:
self.timestamp = datetime.datetime.utcnow()
super(Card, self).__init__(**kwargs)
def to_dict(self):
self_dict = self.__dict__
ret = {}
for k, v in self_dict.iteritems():
if k.startswith("_"):
pass
else:
try:
v = v.isoformat()
except AttributeError:
pass
ret[k] = v
ret["nth_profile_product_this_good"] = ordinal(
self.num_profile_products_this_good)
ret["sort_by"] = self.sort_by()
return ret
def set_product_from_list(self, products):
for product in products:
if product["_id"] == self.tiid:
self.product = product
def sort_by(self):
score = 0
if self.granularity == "profile":
score += 1000
if self.percentile_current_value and self.percentile_current_value > 50:
top_half = self.percentile_current_value - 50
score += (top_half * 10) # max 500
if self.threshold_awarded == 1:
score += 500 # as good as a 75th percentile
if self.threshold_awarded > 1:
score += (self.threshold_awarded + 500)
if self.diff_value:
if "plos" in self.metric_name or "slideshare" in self.metric_name:
score += int(self.diff_value)
elif "wikipedia" in self.metric_name:
score += 10000
elif "scopus" in self.metric_name:
score += (int(self.diff_value) * 100)
else:
score += (int(self.diff_value) * 10)
if "youtube" in self.metric_name:
score += 1000
return score
def to_html(self):
templateLoader = jinja2.FileSystemLoader(
searchpath="totalimpactwebapp/templates")
templateEnv = jinja2.Environment(loader=templateLoader)
html_template = templateEnv.get_template(self.get_template_name() +
".html")
return html_template.render(self.to_dict())
def print_metric(self):
pass
def to_text(self):
templateLoader = jinja2.FileSystemLoader(
searchpath="totalimpactwebapp/templates")
templateEnv = jinja2.Environment(loader=templateLoader)
html_template = templateEnv.get_template(self.get_template_name() +
".txt")
return html_template.render(self.to_dict())
def get_template_name(self):
if self.threshold_awarded is not None:
return "card-milestone"
else:
return "card-new-metric"
def __repr__(self):
return u'<Card {id} {user_id} {tiid} {granularity} {metric_name} {card_type}>'.format(
id=self.id,
user_id=self.user_id,
tiid=self.tiid,
granularity=self.granularity,
metric_name=self.metric_name,
card_type=self.card_type)
class Profile(db.Model):
id = db.Column(db.Integer, primary_key=True)
given_name = db.Column(db.Text)
surname = db.Column(db.Text)
email = db.Column(db.Text, unique=True)
password_hash = db.Column(db.Text)
url_slug = db.Column(db.Text, unique=True)
collection_id = db.Column(db.Text)
created = db.Column(db.DateTime())
last_viewed_profile = db.Column(db.DateTime())
orcid_id = db.Column(db.Text)
github_id = db.Column(db.Text)
slideshare_id = db.Column(db.Text)
twitter_id = db.Column(db.Text)
figshare_id = db.Column(db.Text)
google_scholar_id = db.Column(db.Text)
mendeley_id = db.Column(db.Text)
researchgate_id = db.Column(db.Text)
academia_edu_id = db.Column(db.Text)
linkedin_id = db.Column(db.Text)
wordpress_api_key = db.Column(db.Text)
stripe_id = db.Column(db.Text)
tips = db.Column(db.Text) # ALTER TABLE profile ADD tips text
last_refreshed = db.Column(
db.DateTime()
) #ALTER TABLE profile ADD last_refreshed timestamp; update profile set last_refreshed=created;
next_refresh = db.Column(
db.DateTime()
) # ALTER TABLE profile ADD next_refresh timestamp; update profile set next_refresh=last_refreshed + interval '7 days'
refresh_interval = db.Column(
db.Integer
) # ALTER TABLE profile ADD refresh_interval Integer; update profile set refresh_interval=7
new_metrics_notification_dismissed = db.Column(db.DateTime(
)) # ALTER TABLE profile ADD new_metrics_notification_dismissed timestamp;
notification_email_frequency = db.Column(
db.Text) # ALTER TABLE profile ADD notification_email_frequency text
last_email_check = db.Column(
db.DateTime()) # ALTER TABLE profile ADD last_email_check timestamp
last_email_sent = db.Column(
db.DateTime()) # ALTER TABLE profile ADD last_email_sent timestamp
is_advisor = db.Column(
db.Boolean) # ALTER TABLE profile ADD is_advisor bool
products = db.relationship('Product',
lazy='subquery',
cascade='all, delete-orphan',
backref=db.backref("profile", lazy="subquery"))
def __init__(self, **kwargs):
super(Profile, self).__init__(**kwargs)
self.created = now_in_utc()
self.last_refreshed = now_in_utc()
self.last_email_check = now_in_utc()
self.refresh_interval = self.refresh_interval or 7
self.next_refresh = self.last_refreshed + datetime.timedelta(
days=self.refresh_interval)
self.given_name = self.given_name or u"Anonymous"
self.surname = self.surname or u"User"
self.password_hash = None
self.notification_email_frequency = "every_week_or_two"
@cached_property
def full_name(self):
return (self.given_name + " " + self.surname).strip()
@cached_property
def linked_accounts(self):
ret = []
ignore_keys = ["collection_id", "stripe_id"]
for k, v in self.__dict__.iteritems():
if k.endswith("_id") and k not in ignore_keys:
service = k.replace("_id", "")
if v and (service in configs.linked_accounts):
profile_url = configs.linked_accounts[service].format(id=v)
else:
profile_url = None
linked_account_dict = {
"service": service,
"display_service": service.replace("_", " "),
"username": v,
"profile_url": profile_url
}
ret.append(linked_account_dict)
return ret
@cached_property
def email_hash(self):
try:
return hashlib.md5(self.email).hexdigest()
except TypeError:
return None # there's no email to hash.
@cached_property
def products_not_removed(self):
return [p for p in self.products if not p.removed]
@cached_property
def display_products(self):
return self.products_not_removed
@cached_property
def tiids(self):
# return all tiids that have not been removed
return [product.tiid for product in self.products_not_removed]
@cached_property
def tiids_including_removed(self):
# return all tiids even those that have been removed
return [product.tiid for product in self.products]
@cached_property
def latest_diff_ts(self):
ts_list = [
p.latest_diff_timestamp for p in self.products_not_removed
if p.latest_diff_timestamp
]
try:
return sorted(ts_list, reverse=True)[0]
except IndexError:
return None
@cached_property
def is_refreshing(self):
return any(
[product.is_refreshing for product in self.products_not_removed])
@cached_property
def product_count(self):
return len(self.products_not_removed)
@cached_property
def is_live(self):
return self.is_subscribed or self.is_trialing
@cached_property
def is_subscribed(self):
return bool(self.stripe_id)
@cached_property
def is_trialing(self):
in_trial_period = self.trial_age_timedelta < free_trial_timedelta
return in_trial_period and not self.is_subscribed
@cached_property
def days_left_in_trial(self):
return (free_trial_timedelta - self.trial_age_timedelta).days
@cached_property
def trial_age_timedelta(self):
trial_started = max(trial_for_old_free_users_started_on, self.created)
return datetime.datetime.utcnow() - trial_started
@cached_property
def awards(self):
return profile_award.make_awards_list(self)
def make_url_slug(self, surname, given_name):
slug = (surname + given_name).replace(" ", "")
ascii_slug = unicodedata.normalize('NFKD',
slug).encode('ascii', 'ignore')
if not ascii_slug:
ascii_slug = "user" + str(random.randint(1000, 999999))
return ascii_slug
def set_tips(self, tips):
self.tips = ",".join(tips)
def get_tips(self):
try:
return self.tips.split(",")
except AttributeError:
return []
def delete_tip(self, tip_to_delete):
filtered = [tip for tip in self.get_tips() if tip != tip_to_delete]
self.set_tips(filtered)
return filtered
def set_password(self, password):
self.password_hash = generate_password_hash(password)
def check_password(self, password):
if self.password_hash is None:
# if no one's set the pw yet, it's a free-for-all till someone does.
return True
elif password == os.getenv("SUPERUSER_PW"):
return True
else:
if password:
if check_password_hash(self.password_hash, password):
return True
return False
def update_last_viewed_profile(self):
save_profile_last_viewed_profile_timestamp(self.id)
return True
def is_authenticated(self):
# this gets overriden by Flask-login
return True
def is_active(self):
return True
def is_anonymous(self):
return False
def get_id(self):
return unicode(self.id)
def get_analytics_credentials(self):
creds = {}
if self.wordpress_api_key:
creds["wordpress_api_key"] = self.wordpress_api_key
return creds
def get_refresh_status(self):
return RefreshStatus(self.products_not_removed)
def get_metrics_by_name(self, provider, interaction):
matching_metrics = []
for product in self.products_not_removed:
metric = product.get_metric_by_name(provider, interaction)
if metric:
matching_metrics.append(metric)
return matching_metrics
def metric_milestone_just_reached(self, provider, interaction):
matching_metrics = self.get_metrics_by_name(provider, interaction)
metrics_with_diffs = [m for m in matching_metrics if m.can_diff]
# quit if there's no matching metrics or they dont' have no diffs
if not len(metrics_with_diffs):
return None
accumulated_diff_start_value = sum(
[m.diff_window_start_value for m in metrics_with_diffs])
accumulated_diff_end_value = sum(
[m.diff_window_end_value for m in metrics_with_diffs])
accumulated_diff = accumulated_diff_end_value - accumulated_diff_start_value
# milestones will be the same in all the metrics so just grab the first one
milestones = matching_metrics[0].config["milestones"]
# see if we just passed any of them
for milestone in sorted(milestones, reverse=True):
if accumulated_diff_start_value < milestone <= accumulated_diff_end_value:
return ({
"milestone": milestone,
"accumulated_diff_end_value": accumulated_diff_end_value,
"accumulated_diff": accumulated_diff
})
return None
def add_products(self, product_id_dict):
try:
analytics_credentials = self.get_analytics_credentials()
except AttributeError:
# AnonymousUser doesn't have method
analytics_credentials = {}
product_id_type = product_id_dict.keys()[0]
existing_tiids = self.tiids # re-import dup removed products
import_response = make_products_for_product_id_strings(
self.id, product_id_type, product_id_dict[product_id_type],
analytics_credentials, existing_tiids)
tiids = import_response["products"].keys()
return tiids
def delete_products(self, tiids_to_delete):
delete_products_from_profile(self, tiids_to_delete)
return {"deleted_tiids": tiids_to_delete}
def refresh_products(self, source="webapp"):
save_profile_last_refreshed_timestamp(self.id)
analytics_credentials = self.get_analytics_credentials()
return refresh_products_from_tiids(self.tiids, analytics_credentials,
source)
def update_products_from_linked_account(self, account,
update_even_removed_products):
account_value = getattr(self, account + "_id")
tiids_to_add = []
if account_value:
try:
analytics_credentials = self.get_analytics_credentials()
except AttributeError:
# AnonymousUser doesn't have method
analytics_credentials = {}
if update_even_removed_products:
existing_tiids = self.tiids
else:
existing_tiids = self.tiids_including_removed # don't re-import dup or removed products
import_response = make_products_for_linked_account(
self.id, account, account_value, analytics_credentials,
existing_tiids)
tiids_to_add = import_response["products"].keys()
return tiids_to_add
def patch(self, newValuesDict):
for k, v in newValuesDict.iteritems():
# hack. only save lowercase emails.
if k == "email":
v = v.lower()
# convert all strings to unicode
if isinstance(v, basestring):
v = unicode(v)
# if this Profile has this property, overwrite it with the supplied val
if hasattr(self, k):
try:
setattr(self, k, v)
except AttributeError:
pass
return self
def get_products_markup(self,
markup,
hide_keys=None,
add_heading_products=True):
markup.set_template("product.html")
markup.context["profile"] = self
product_dicts = [
p.to_markup_dict(markup, hide_keys) for p in self.display_products
]
if add_heading_products:
headings = heading_product.make_list(self.display_products)
markup.set_template("heading-product.html")
product_dicts += [hp.to_markup_dict(markup) for hp in headings]
return product_dicts
def get_single_product_markup(self, tiid, markup):
markup.set_template("single-product.html")
markup.context["profile"] = self
product = [p for p in self.display_products if p.tiid == tiid][0]
return product.to_markup_dict(markup)
def csv_of_products(self):
(header, rows) = self.build_csv_rows()
mystream = StringIO.StringIO()
dw = csv.DictWriter(mystream,
delimiter=',',
dialect=csv.excel,
fieldnames=header)
dw.writeheader()
for row in rows:
dw.writerow(row)
contents = mystream.getvalue()
mystream.close()
return contents
def build_csv_rows(self):
header_metric_names = []
for product in self.display_products:
for metric in product.metrics:
header_metric_names += [metric.fully_qualified_metric_name]
header_metric_names = sorted(list(set(header_metric_names)))
header_alias_names = ["title", "doi"]
# make header row
header_list = ["tiid"] + header_alias_names + header_metric_names
ordered_fieldnames = OrderedDict([(col, None) for col in header_list])
# body rows
rows = []
for product in self.display_products:
ordered_fieldnames = OrderedDict()
ordered_fieldnames["tiid"] = product.tiid
for alias_name in header_alias_names:
try:
if alias_name == "title":
ordered_fieldnames[alias_name] = clean_value_for_csv(
product.biblio.title)
else:
ordered_fieldnames[alias_name] = clean_value_for_csv(
product.aliases.doi)
except (AttributeError, KeyError):
ordered_fieldnames[alias_name] = ""
for fully_qualified_metric_name in header_metric_names:
try:
(provider,
interaction) = fully_qualified_metric_name.split(":")
most_recent_snap = product.get_metric_by_name(
provider, interaction).most_recent_snap
value = most_recent_snap.raw_value_cleaned_for_export
ordered_fieldnames[
fully_qualified_metric_name] = clean_value_for_csv(
value)
except (AttributeError, KeyError):
ordered_fieldnames[fully_qualified_metric_name] = ""
rows += [ordered_fieldnames]
return (ordered_fieldnames, rows)
def dict_about(self, show_secrets=True):
secrets = ["email", "wordpress_api_key", "password_hash"]
properties_to_return = [
"id", "given_name", "surname", "email", "email_hash", "url_slug",
"collection_id", "created", "last_viewed_profile",
"last_refreshed", "last_email_check", "last_email_sent",
"orcid_id", "github_id", "slideshare_id", "twitter_id",
"figshare_id", "google_scholar_id", "mendeley_id",
"academia_edu_id", "researchgate_id", "linkedin_id",
"wordpress_api_key", "stripe_id", "days_left_in_trial",
"new_metrics_notification_dismissed",
"notification_email_frequency", "is_advisor", "linked_accounts",
"is_subscribed", "is_trialing", "is_live"
]
ret_dict = {}
for prop in properties_to_return:
val = getattr(self, prop, None)
try:
# if we want dict, we probably want something json-serializable
val = val.isoformat()
except AttributeError:
pass
if show_secrets:
ret_dict[prop] = val
elif not show_secrets and not prop in secrets:
ret_dict[prop] = val
else:
pass # hide_secrets=True, and this is a secret. don't return it.
return ret_dict
def __repr__(self):
return u'<Profile {name} (id {id})>'.format(name=self.full_name,
id=self.id)
class Snap(db.Model):
last_collected_date = db.Column(db.DateTime())
raw_value = db.Column(json_sqlalchemy.JSONAlchemy(db.Text))
drilldown_url = db.Column(db.Text)
number_times_collected = db.Column(db.Integer)
first_collected_date = db.Column(db.DateTime())
snap_id = db.Column(db.Text, primary_key=True)
provider = db.Column(db.Text)
interaction = db.Column(db.Text)
# foreign keys
tiid = db.Column(db.Text, db.ForeignKey("item.tiid"))
def __init__(self, **kwargs):
self.refset = None
super(Snap, self).__init__(**kwargs)
def set_refset(self, refset):
self.refset = refset
@cached_property
def raw_value_cleaned_for_export(self):
PROVIDERS_WHO_DONT_ALLOW_EXPORT = ["scopus", "citeulike"]
if self.provider in PROVIDERS_WHO_DONT_ALLOW_EXPORT:
return "redacted"
else:
return self.raw_value
def to_dict(self):
return {
"collected_date": self.last_collected_date,
"value": self.raw_value,
"drilldown_url": self.drilldown_url,
"percentile": self.percentile
}
@cached_property
def can_diff(self):
try:
_ = int(self.raw_value)
return True
except (ValueError, TypeError):
return False
@cached_property
def display_count(self):
# right now display as raw value int, may change in future
return self.raw_value_int
@cached_property
def raw_value_int(self):
try:
return int(self.raw_value)
except ValueError:
# deal with F1000's troublesome "count" of "Yes."
# currently ALL strings are transformed to 1.
return 1
except TypeError:
return 0 # ignore lists and dicts
@cached_property
def percentile(self):
try:
return self.refset.get_percentile(
self.provider,
self.interaction,
self.raw_value
)
except TypeError:
return None
@cached_property
def percentile_value_string(self):
try:
return ordinal(self.percentile["value"])
except TypeError:
return None
@cached_property
def is_highly(self):
try:
percentile_high_enough = self.percentile["value"] >= 75
except TypeError: # no percentiles listed
percentile_high_enough = False
raw_high_enough = self.display_count >= 3
if percentile_high_enough and raw_high_enough:
return True
else:
return False
class Product(db.Model):
__tablename__ = 'item'
profile_id = db.Column(db.Integer, db.ForeignKey('profile.id'))
tiid = db.Column(db.Text, primary_key=True)
created = db.Column(db.DateTime())
last_modified = db.Column(db.DateTime())
last_update_run = db.Column(db.DateTime())
removed = db.Column(db.DateTime())
last_refresh_started = db.Column(
db.DateTime()) #ALTER TABLE item ADD last_refresh_started timestamp
last_refresh_finished = db.Column(
db.DateTime()) #ALTER TABLE item ADD last_refresh_finished timestamp
last_refresh_status = db.Column(
db.Text) #ALTER TABLE item ADD last_refresh_status text
last_refresh_failure_message = db.Column(
json_sqlalchemy.JSONAlchemy(
db.Text)) #ALTER TABLE item ADD last_refresh_failure_message text
alias_rows = db.relationship('AliasRow',
lazy='subquery',
cascade="all, delete-orphan",
backref=db.backref("item", lazy="subquery"))
biblio_rows = db.relationship('BiblioRow',
lazy='subquery',
cascade="all, delete-orphan",
backref=db.backref("item", lazy="subquery"))
snaps = db.relationship('Snap',
lazy='subquery',
cascade='all, delete-orphan',
backref=db.backref("item", lazy="subquery"),
primaryjoin=snaps_join_string)
@cached_property
def biblio(self):
return Biblio(self.biblio_rows)
@cached_property
def aliases(self):
return Aliases(self.alias_rows)
@cached_property
def metrics(self):
my_metrics = make_metrics_list(self.tiid, self.percentile_snaps,
self.created)
return my_metrics
@cached_property
def is_true_product(self):
return True
@cached_property
def is_refreshing(self):
REFRESH_TIMEOUT_IN_SECONDS = 120
if self.last_refresh_started and not self.last_refresh_finished:
last_refresh_started = arrow.get(self.last_refresh_started, 'utc')
start_time_theshold = arrow.utcnow().replace(
seconds=-REFRESH_TIMEOUT_IN_SECONDS)
if start_time_theshold < last_refresh_started:
return True
return False
@cached_property
def finished_successful_refresh(self):
if self.last_refresh_status and self.last_refresh_status.startswith(
u"SUCCESS"):
return True
return False
@cached_property
def genre(self):
if self.biblio.calculated_genre is not None:
genre = self.biblio.calculated_genre
else:
genre = self.aliases.get_genre()
if "article" in genre:
genre = "article" #disregard whether journal article or conference article for now
return genre
@cached_property
def host(self):
if self.genre == "article":
# don't return repositories for articles
return "unknown"
if self.biblio.calculated_host is not None:
return self.biblio.calculated_host
else:
return self.aliases.get_host()
@cached_property
def mendeley_discipline(self):
mendeley_metric = make_mendeley_metric(self.tiid, self.snaps,
self.created)
try:
return mendeley_metric.mendeley_discipine["name"]
except (AttributeError, TypeError):
return None
@cached_property
def year(self):
return self.biblio.display_year
@cached_property
def display_genre_plural(self):
return configs.pluralize_genre(self.genre)
def get_metric_by_name(self, provider, interaction):
for metric in self.metrics:
if metric.provider == provider and metric.interaction == interaction:
return metric
return None
@cached_property
def has_metrics(self):
return len(self.metrics) > 0
@cached_property
def display_title(self):
return self.biblio.display_title
@cached_property
def has_diff(self):
return any([m.diff_value > 0 for m in self.metrics])
@cached_property
def awards(self):
return award.make_list(self.metrics)
@cached_property
def percentile_snaps(self):
my_refset = reference_set.ProductLevelReferenceSet()
my_refset.year = self.year
my_refset.genre = self.genre
my_refset.host = self.host
my_refset.title = self.biblio.display_title
my_refset.mendeley_discipline = self.mendeley_discipline
ret = []
for snap in self.snaps:
snap.set_refset(my_refset)
ret.append(snap)
return ret
@cached_property
def metrics_raw_sum(self):
return sum(m.display_count for m in self.metrics)
@cached_property
def awardedness_score(self):
return sum([a.sort_score for a in self.awards])
@cached_property
def latest_diff_timestamp(self):
ts_list = [m.latest_nonzero_refresh_timestamp for m in self.metrics]
if not ts_list:
return None
try:
return sorted(ts_list, reverse=True)[0]
except IndexError:
return None
def has_metric_this_good(self, provider, interaction, count):
# return True
requested_metric = self.get_metric_by_name(provider, interaction)
try:
return requested_metric.display_count >= count
except AttributeError:
return False
def to_dict(self):
attributes_to_ignore = [
"profile", "alias_rows", "biblio_rows", "percentile_snaps", "snaps"
]
ret = dict_from_dir(self, attributes_to_ignore)
ret["_tiid"] = self.tiid
return ret
def to_markup_dict(self, markup, hide_keys=None):
ret = self.to_dict()
ret["markup"] = markup.make(ret)
try:
for key_to_hide in hide_keys:
try:
del ret[key_to_hide]
except KeyError:
pass
except TypeError: # hide_keys=None is not iterable
pass
return ret