daily.py

from totalimpactwebapp.snap import Snap
from totalimpactwebapp.product import Product
from totalimpactwebapp.profile import Profile
from totalimpactwebapp.profile import get_profile_summary_dict
from totalimpactwebapp.product import refresh_products_from_tiids
from totalimpactwebapp.pinboard import Pinboard
# from totalimpactwebapp.pinboard import auto_populate_pinboard
from totalimpactwebapp.reference_set import save_all_reference_set_lists
from totalimpactwebapp.reference_set import RefsetBuilder
from totalimpactwebapp.product_deets import populate_product_deets
from totalimpactwebapp.drip_email import log_drip_email
from totalimpactwebapp.tweeter import Tweeter
from totalimpactwebapp.tweeter import get_and_save_tweeter_followers
from totalimpact.providers.provider import ProviderFactory
from totalimpact import default_settings

from util import commit
from util import dict_from_dir
from core_tasks import provider_method_wrapper
from totalimpactwebapp import db
from totalimpactwebapp import ti_queues
from db_backup_to_s3 import upload_to_s3
import tasks

from sqlalchemy import and_, or_, func, between
import datetime
import os
import requests
import argparse
import logging
import time
import pickle
from collections import defaultdict, Counter, OrderedDict
import StringIO
import csv
import urllib
import hashlib
import json

# logger is set below, in main


"""
requires these env vars be set in this environment:
DATABASE_URL
"""

try:
    # set jason's env variables for local running.
    import config
    config.set_env_vars_from_dot_env()
except ImportError:
    pass


def page_query(q):
    offset = 0
    while True:
        r = False
        for elem in q.limit(50).offset(offset):
           r = True
           yield elem
        offset += 50
        if not r:
            break

# from https://bitbucket.org/zzzeek/sqlalchemy/wiki/UsageRecipes/WindowedRangeQuery
# cited from many webpages as good way to do paging
def column_windows(session, column, windowsize):
    """Return a series of WHERE clauses against 
    a given column that break it into windows.

    Result is an iterable of tuples, consisting of
    ((start, end), whereclause), where (start, end) are the ids.

    Requires a database that supports window functions, 
    i.e. Postgresql, SQL Server, Oracle.

    Enhance this yourself !  Add a "where" argument
    so that windows of just a subset of rows can
    be computed.

    """
    def int_for_range(start_id, end_id):
        if end_id:
            return and_(
                column>=start_id,
                column<end_id
            )
        else:
            return column>=start_id

    q = session.query(
                column, 
                func.row_number().\
                        over(order_by=column).\
                        label('rownum')
                ).\
                from_self(column)
    if windowsize > 1:
        q = q.filter("rownum %% %d=1" % windowsize)

    intervals = [id for id, in q]

    while intervals:
        start = intervals.pop(0)
        if intervals:
            end = intervals[0]
        else:
            end = None
        yield int_for_range(start, end)


def windowed_query(q, column, windowsize, desc=False):
    """"Break a Query into windows on a given column."""

    for whereclause in column_windows(
                                        q.session, 
                                        column, windowsize):
        if desc:
            for row in q.filter(whereclause).order_by(column.desc()):
                yield row
        else:
            for row in q.filter(whereclause).order_by(column):
                yield row


def csv_of_dict(mydicts):
    (header, rows) = build_csv_rows_from_dict(mydicts)

    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_from_dict(mydicts):

    columns = []
    for dictrow in mydicts:
        columns += dictrow.keys()
    columns = sorted(list(set(columns)))

    # make header row
    ordered_column_names = OrderedDict([(col, None) for col in columns])

    # body rows
    rows = []
    for dictrow in mydicts:
        ordered_contents = OrderedDict()
        for col in ordered_column_names:
            try:
                ordered_contents[col] = unicode(dictrow[col]).encode("utf8")
            except (AttributeError, KeyError):
                ordered_contents[col] = u""
        rows += [ordered_contents]
    return(ordered_column_names, rows)


def profile_deets(url_slug=None, 
        min_url_slug=None, 
        start_days_ago=44, 
        end_days_ago=30):

    start_date = datetime.datetime.utcnow() - datetime.timedelta(days=start_days_ago)
    end_date = datetime.datetime.utcnow() - datetime.timedelta(days=end_days_ago)
    q = db.session.query(Profile) \
            .filter(Profile.created.between(start_date, end_date))

    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    elif min_url_slug:
        q = q.filter(Profile.url_slug>=min_url_slug)

    profile_deets = []
    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"profile_deets: {url_slug}".format(
            url_slug=profile.url_slug))
        profile_deets += [get_profile_summary_dict(profile)]
        db.session.expunge(profile)
        # print csv_of_dict(profile_deets)
        # with open("profile_deets.pickle", "wb") as handle:
        #   pickle.dump(profile_deets, handle)

    # print json.dumps(profile_deets, sort_keys=True, indent=4)

    print "****"
    csv_contents = csv_of_dict(profile_deets)
    print csv_contents
    import tempfile
    temp_csv_file = tempfile.NamedTemporaryFile(delete=False)
    temp_csv_file.write(csv_contents)
    temp_csv_file.close()

    upload_to_s3(temp_csv_file.name, "exploring/profile_deets.csv")
    time.sleep(30)


def profile_deets_live(args):

    url_slug = args.get("url_slug", None)
    min_url_slug = args.get("min_url_slug", None)

    q = profile_query(url_slug, min_url_slug)

    number_considered = 0.0
    start_time = datetime.datetime.utcnow()
    profile_deets = []

    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"profile_deets: {url_slug}".format(
            url_slug=profile.url_slug))
        profile_deets += [get_profile_summary_dict(profile)]
        db.session.expunge(profile)
        # print csv_of_dict(profile_deets)
        # with open("profile_deets.pickle", "wb") as handle:
        #   pickle.dump(profile_deets, handle)

    # print json.dumps(profile_deets, sort_keys=True, indent=4)

    print "****"
    csv_contents = csv_of_dict(profile_deets)
    print csv_contents
    import tempfile
    temp_csv_file = tempfile.NamedTemporaryFile(delete=False)
    temp_csv_file.write(csv_contents)
    temp_csv_file.close()

    upload_to_s3(temp_csv_file.name, "exploring/profile_deets.csv")
    time.sleep(30)


def add_product_deets_for_everyone(url_slug=None, skip_until_url_slug=None):
    if url_slug:
        profile_iterator = [Profile.query.filter_by(url_slug=url_slug).first()]
    else:
        q = db.session.query(Profile)
        profile_iterator = windowed_query(q, Profile.url_slug, 25)
        # profile_iterator = page_query(Profile.query.order_by(Profile.url_slug.asc()))

    run_id = datetime.datetime.utcnow().isoformat()
    for profile in profile_iterator:
        if skip_until_url_slug and skip_until_url_slug.lower() > profile.url_slug.lower():
            logger.info(u"in add_product_deets_for_everyone and skipping {url_slug}".format(
                url_slug=profile.url_slug))
            continue

        logger.info(u"add_product_deets_for_everyone: {url_slug}".format(
            url_slug=profile.url_slug))

        for product in profile.products_not_removed:
            # logger.info(u"add_product_deets_for_everyone: {url_slug}, tiid={tiid}".format(
            #     url_slug=profile.url_slug, tiid=product.tiid))
            product_deets = populate_product_deets(profile, product)  # not delayed
            product_deets.run_id = run_id
            db.session.add(product_deets)
        db.session.commit()


def dedup_everyone(url_slug=None, min_url_slug=None):
    q = db.session.query(Profile)
    if url_slug:
        q = q.filter(Profile.url_slug==url_slug)
    elif min_url_slug:
        q = q.filter(Profile.url_slug>=min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):
        if profile.is_live:
            logger.info(u"dedup: {url_slug}".format(url_slug=profile.url_slug))
            response = profile.remove_duplicates()


def mint_stripe_customers_for_all_profiles():

    stripe.api_key = os.getenv("STRIPE_API_KEY")

    for profile in page_query(Profile.query.order_by(Profile.email.asc())):

        if profile.stripe_id:
            print u"Already a Stripe customer for {email}; skipping".format(
                email=profile.email
            )
            continue


        full_name = u"{first} {last}".format(
            first=profile.given_name,
            last=profile.surname
        )
        if profile.is_advisor:
            print u"making an Advisor Stripe customer for {email} ".format(email=profile.email)
            stripe_customer = stripe.Customer.create(
                description=full_name,
                email=profile.email,
                plan="base",
                coupon="ADVISOR_96309"
            )
        else:
            print u"making a regular Stripe customer for {email} ".format(email=profile.email)            
            stripe_customer = stripe.Customer.create(
                description=full_name,
                email=profile.email,
                plan="base" 
            )

        print u"Successfully made stripe id " + stripe_customer.id

        profile.stripe_id = stripe_customer.id
        db.session.merge(profile)
        db.session.commit()


def write_500_random_profile_urls():
    urls = []
    sample_size = 500
    for profile in page_query(Profile.query):
        products_count = len(profile.tiids)
        if products_count > 0:
            url = "https://staging-impactstory.org/" + profile.url_slug
            urls.append([products_count, url])
            logger.info(u"getting a new profile url out: {url}".format(
                url=url
            ))

    sampled_urls = random.sample(urls, sample_size)

    logger.info(u"writing our {sample_size} sampled profile URLs".format(
        sample_size=sample_size
    ))

    for row in sampled_urls:
        try:
            print "{products_count},{url}".format(
                products_count=row[0],
                url=row[1]
            )
        except UnicodeEncodeError:
            pass  # whatever, we don't need exactly 500


def email_report_to_live_profiles(url_slug=None, min_url_slug=None, max_emails=None):
    number_emails_sent = 0

    q = profile_query(url_slug, min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):

        # logger.debug(u"in email_report_to_live_profiles for {url_slug}".format(
        #     url_slug=profile.url_slug))

        try:
            if not profile.is_live:
                pass
                logger.info(u"not sending, profile is not live {url_slug}".format(url_slug=profile.url_slug))                
            elif not profile.email or (u"@" not in profile.email):
                pass
                logger.info(u"not sending, no email address for {url_slug}".format(url_slug=profile.url_slug))
            elif profile.notification_email_frequency == "none":
                pass
                logger.info(u"not sending, {url_slug} is unsubscribed".format(url_slug=profile.url_slug))
            elif profile.last_email_sent and ((datetime.datetime.utcnow() - profile.last_email_sent).days < 7):
                pass
                logger.info(u"not sending, {url_slug} already got email this week".format(url_slug=profile.url_slug))
            else:
                # logger.info(u"checking email for {url_slug}".format(url_slug=profile.url_slug))
                # status = tasks.send_email_if_new_diffs.delay(profile.id)
                status = tasks.send_email_if_new_diffs(profile)
                if status=="email sent":
                    number_emails_sent += 1    
                    if max_emails:
                        logger.info(u"sent an email, have {num} left before hitting max".format(
                        num = max_emails-number_emails_sent))
                logger.info(u"checked email for {url_slug}, status={status}".format(
                    url_slug=profile.url_slug, status=status))

        except Exception as e:
            logger.warning(u"EXCEPTION in email_report_to_everyone_who_needs_one for {url_slug}, skipping to next profile.  Error {e}".format(
                url_slug=profile.url_slug, e=e))
            pass

        if max_emails and number_emails_sent >= max_emails:
            logger.info(u"Reached max_number_profiles_to_consider, so no done queueing email")
            break

    return


def email_all_profiles_about_tng(url_slug=None, min_url_slug=None, max_emails=None):
    number_emails_sent = 0

    q = profile_query(url_slug, min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):

        logger.debug(u"in email_all_profiles_about_tng for {url_slug}".format(
            url_slug=profile.url_slug))

        try:
            if not profile.email or (u"@" not in profile.email):
                pass
                logger.info(u"not sending, no email address for {url_slug}".format(url_slug=profile.url_slug))
            else:
                logger.info(u"emailing {url_slug} about tng".format(url_slug=profile.url_slug))
                tasks.send_tng_email(profile)

        except Exception as e:
            logger.warning(u"EXCEPTION in email_all_profiles_about_tng for {url_slug}, skipping to next profile.  Error {e}".format(
                url_slug=profile.url_slug, e=e))
            pass


    return

def build_refsets(save_after_every_profile=False):
    refset_builder = RefsetBuilder()

    q = db.session.query(Profile)
    for profile in windowed_query(q, Profile.url_slug, 25):
        refset_builder.process_profile(profile)
        if save_after_every_profile:
            save_all_reference_set_lists(refset_builder)

    save_all_reference_set_lists(refset_builder)


def collect_embed(url_slug=None, min_url_slug=None):
    q = profile_query(url_slug, min_url_slug)

    start_time = datetime.datetime.utcnow()
    number_considered = 0.0
    number_markups = 0.0
    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.debug("-->collecting embed for {url_slug}".format(
            url_slug=profile.url_slug))

        for product in profile.display_products:
            if not product.embed_markup:

                number_considered += 1

                try:
                    embed_markup = product.get_embed_markup()
                except Exception:
                    print "got an exception, skipping", product.aliases.best_url
                    continue

                if embed_markup:

                    print "GOT MARKUP for", product.tiid, product.host, product.aliases.best_url, embed_markup
                    # print "  got an embed for", product.genre, "!"
                    product.embed_markup = embed_markup
                    db.session.add(product)
                    commit(db)
                    number_markups += 1
                    elapsed_seconds = (datetime.datetime.utcnow() - start_time).seconds
                    print "elapsed seconds=", elapsed_seconds, ";  number per second=", number_considered/(0.1+elapsed_seconds)


def live_profile_query():
    from totalimpactwebapp.profile import default_free_trial_days
    min_created_date = datetime.datetime.utcnow() - datetime.timedelta(days=default_free_trial_days)
    q = db.session.query(Profile).filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None, Profile.created>=min_created_date))
    return q

def profile_query(url_slug=None, min_url_slug=None):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        q = live_profile_query()
        if min_url_slug:
            q = q.filter(Profile.url_slug>=min_url_slug)
    return q

def borked_pinboards_for_life_profiles(url_slug=None, min_url_slug=None):
    problem_urls = []
    blank_urls = []
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        q = live_profile_query()
        if min_url_slug:
            q = q.filter(Profile.url_slug>=min_url_slug)
    for profile in windowed_query(q, Profile.url_slug, 25):
        all_profile_tiids = [product.tiid for product in profile.products] #includes removed
        print profile.url_slug, profile.id        
        board = Pinboard.query.filter_by(profile_id=profile.id).first()
        if board:
            tiids = [tiid for (dummy, tiid) in board.contents["one"]]
            if not tiids:
                logger.debug("{url} has blank pinboard".format(
                    url=profile.url_slug))
                blank_urls.append(profile.url_slug)
            for tiid in tiids:
                if tiid not in all_profile_tiids:
                    logger.debug("{url} does not own this pinboard {board}".format(
                        url=profile.url_slug, board=board))
                    problem_urls.append(profile.url_slug)
    logger.debug("problem urls: {urls}".format(
        urls=problem_urls))
    logger.debug("blank urls: {urls}".format(
        urls=blank_urls))


def new_metrics_for_live_profiles(url_slug=None, min_url_slug=None, start_days_ago=7):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        min_created_date = datetime.datetime.utcnow() - datetime.timedelta(days=30)
        q = db.session.query(Profile).filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None, Profile.created>=min_created_date))
        if min_url_slug:
            q = q.filter(Profile.url_slug>=min_url_slug)

        # also, only if not refreshed recently
        min_last_refreshed = datetime.datetime.utcnow() - datetime.timedelta(days=start_days_ago)
        q = q.filter(Profile.last_refreshed <= min_last_refreshed)

    start_time = datetime.datetime.utcnow()
    number_profiles = 0.0
    total_refreshes = 0
    for profile in windowed_query(q, Profile.url_slug, 25):
        number_profiles += 1
        print profile.url_slug, profile.id, profile.last_refreshed, len(profile.display_products)
        number_refreshes = len(profile.display_products)
        if number_refreshes:
            profile.refresh_products(source="scheduled")
            total_refreshes += number_refreshes
            pause_length = min(number_refreshes * 3, 120)
            print "pausing", pause_length, "seconds after refreshing", number_refreshes, "products"
            time.sleep(pause_length)
            print total_refreshes, "total refreshes across", number_profiles, "profiles"
            elapsed_seconds = (datetime.datetime.utcnow() - start_time).seconds
            print "elapsed seconds=", elapsed_seconds, ";  number profiles per second=", number_profiles/(0.1+elapsed_seconds)


from totalimpactwebapp.countries_info import country_iso_by_name


def update_mendeley_countries_for_live_profiles(url_slug=None, min_url_slug=None):
    q = profile_query(url_slug, min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"{url_slug} processing mendeley countries".format(
            url_slug=profile.url_slug))
        for product in profile.display_products:
            metric = product.get_metric_by_name("mendeley", "countries")
            if metric:
                snap = metric.most_recent_snap
                if not snap.raw_value:
                    # logger.error(u"{url_slug} has NO SNAP for tiid {tiid}".format(
                    #     url_slug=profile.url_slug, tiid=product.tiid))
                    # don't save this one to the db
                    continue
                new_snap_value = {}
                for country_name, country_count in snap.raw_value.iteritems():
                    if country_name in country_iso_by_name:
                        iso = country_iso_by_name[country_name]
                        new_snap_value[iso] = country_count
                        # logger.error(u"{country_name} -> {iso}".format(
                        #     country_name=country_name, iso=iso))
                    else:
                        if len(country_name) != 2:
                            logger.error(u"Can't find country {country} in lookup".format(
                                country=country_name))
                        new_snap_value[country_name] = country_count
                if new_snap_value:
                    logger.info(u"New snap value {snap}".format(
                        snap=new_snap_value))
                    snap.raw_value = new_snap_value
                    db.session.add(snap)
                    commit(db)


def collect_new_mendeley(url_slug=None, min_url_slug=None):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        q = db.session.query(Profile).filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None))
        if min_url_slug:
            q = q.filter(Profile.url_slug>=min_url_slug)

    start_time = datetime.datetime.utcnow()
    number_profiles = 0.0
    total_refreshes = 0
    for profile in windowed_query(q, Profile.url_slug, 25):
        number_profiles += 1
        number_refreshes = 0.0
        print profile.url_slug, profile.id
        for product in profile.display_products:
            if product.get_metric_by_name("mendeley", "readers"):
                number_refreshes += 1
                refresh_products_from_tiids(product.profile_id, [product.tiid], source="scheduled")
        if number_refreshes:
            total_refreshes += number_refreshes
            pause_length = min(number_refreshes * 2, 60)
            print "pausing", pause_length, "seconds after refreshing", number_refreshes, "products"
            time.sleep(pause_length)
            print total_refreshes, "total refreshes across", number_profiles, "profiles"
            elapsed_seconds = (datetime.datetime.utcnow() - start_time).seconds
            print "elapsed seconds=", elapsed_seconds, ";  number profiles per second=", number_profiles/(0.1+elapsed_seconds)


def linked_accounts(account_type, url_slug=None, min_url_slug=None):
    column_name = account_type+"_id"
    if url_slug:
        q = db.session.query(Profile).filter(getattr(Profile, column_name) != None).filter(Profile.url_slug==url_slug)
    else:
        if min_url_slug:
            q = db.session.query(Profile).filter(getattr(Profile, column_name) != None).filter(Profile.url_slug>=min_url_slug)

        else:
            q = db.session.query(Profile).filter(getattr(Profile, column_name) != None)

    number_considered = 0.0
    for profile in windowed_query(q, Profile.url_slug, 25):
        number_considered += 1
        logger.info(u"{url_slug} previous number of account products: {num}".format(
            url_slug=profile.url_slug, num=len(profile.account_products)))
        existing_account_product_list = [p for p in profile.account_products if p.index_name==account_type]
        if existing_account_product_list:
            existing_account_product = existing_account_product_list[0]
            if existing_account_product.followers:
                logger.info(u"{url_slug} already has an account_product for {account_type}, so skipping".format(
                    url_slug=profile.url_slug, account_type=account_type))
            else:
                logger.info(u"{url_slug} already has an account_product for {account_type}, but no followers, so refreshing".format(
                    url_slug=profile.url_slug, account_type=account_type))
                refresh_products_from_tiids(existing_account_product.profile_id, [existing_account_product.tiid], source="scheduled")
        else:
            logger.info(u"{url_slug} had no account_product for {account_type}, so adding".format(
                url_slug=profile.url_slug, account_type=account_type))
            tiids = profile.update_products_from_linked_account(account_type, update_even_removed_products=False)
            if tiids:
                logger.info(u"{url_slug} added {num} products for {account_type}".format(
                    url_slug=profile.url_slug, num=len(tiids), account_type=account_type))


tiids_that_need_twitter = """sl1uu922rwpl1htii64upwjs
zh61suvqesowwqi1qcn0v413
5jugcbn9anjkrb84y3zyqwix""".split()

def refresh_twitter(min_tiid=None):
    if min_tiid:
        q = db.session.query(Product).filter(Product.profile_id != None).filter(Product.tiid>min_tiid)
    else:
        q = db.session.query(Product).filter(Product.profile_id != None).filter(Product.tiid.in_(tiids_that_need_twitter))

    start_time = datetime.datetime.utcnow()
    number_considered = 0.0
    number_refreshed = 0
    for product in windowed_query(q, Product.tiid, 25):
        number_considered += 1
        try:
            if product.biblio.repository=="Twitter" and len(product.metrics)==0:
                print "refreshing", product.tiid, number_refreshed
                refresh_products_from_tiids(product.profile_id, [product.tiid], source="scheduled")
                number_refreshed += 1
                if number_refreshed >= 15:
                    #api limit
                    print "refreshed 15, so breaking now"
                    break
        except AttributeError:
            pass


def refresh_tweeted_products(min_tiid=None):

    if min_tiid:
        q = db.session.query(Product).filter(Product.profile_id != None).filter(Product.tiid>min_tiid)
    else:
        q = db.session.query(Product).filter(Product.profile_id != None)

    start_time = datetime.datetime.utcnow()
    number_considered = 0.0
    number_refreshed = 0
    for product in windowed_query(q, Product.tiid, 25):
        number_considered += 1
        try:
            if product.get_metric_by_name("altmetric_com", "tweets"):
                print number_refreshed, ". refreshing: ", product.tiid
                refresh_products_from_tiids(product.profile_id, [product.tiid], source="scheduled")
                number_refreshed += 1
                time.sleep(0.5)
                elapsed_seconds = (datetime.datetime.utcnow() - start_time).seconds
                print "elapsed seconds=", elapsed_seconds, ";  number per second=", number_considered/(0.1+elapsed_seconds)
                
        except AttributeError:
            pass


def run_through_altmetric_tweets(url_slug=None, min_url_slug=None):
    q = profile_query(url_slug, min_url_slug)

    total_objects_saved = 0
    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"{url_slug}".format(
            url_slug=profile.url_slug))

        new_objects = save_product_tweets_for_profile(profile)
        total_objects_saved += len(new_objects)
        print "total_objects_saved", total_objects_saved


def run_through_twitter_pages(url_slug=None, min_url_slug=None):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.twitter_id != None).filter(Profile.url_slug==url_slug)
    else:
        q = db.session.query(Profile).filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None)).filter(Profile.twitter_id != None)
        if min_url_slug:
            q = q.filter(Profile.url_slug>=min_url_slug)

    from totalimpactwebapp.tweet import save_recent_tweets

    for profile in windowed_query(q, Profile.url_slug, 25):

        logger.info(u"{url_slug} has twitter handle {twitter_handle}, now saving tweets".format(
            url_slug=profile.url_slug, twitter_handle=profile.twitter_id))
        save_recent_tweets(profile.id, profile.twitter_id)


# def star_best_products(args):
#     url_slug = args.get("url_slug", None)
#     min_url_slug = args.get("min_url_slug", None)

#     q = profile_query(url_slug, min_url_slug)

#     number_considered = 0.0
#     start_time = datetime.datetime.utcnow()
#     for profile in windowed_query(q, Profile.url_slug, 25):
#         number_considered += 1

#         board = Pinboard.query.filter_by(profile_id=profile.id).first()
#         if board:
#             # already has one!  skip and keep going
#             continue

#         if not profile.products:
#             # print "no products"
#             continue

#         logger.info(u"*******saved pinboard for {url_slug}".format(
#             url_slug=profile.url_slug))

#         contents = auto_populate_pinboard(profile)
#         board = Pinboard(profile_id=profile.id, contents=contents)
#         db.session.add(board)
#         commit(db)

#         elapsed_seconds = (datetime.datetime.utcnow() - start_time).seconds
#         print "elapsed seconds=", elapsed_seconds, ";  number per second=", number_considered/(0.1+elapsed_seconds)
  

def count_news_for_subscribers(url_slug=None, min_url_slug=None):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        if min_url_slug:
            q = db.session.query(Profile).filter(Profile.stripe_id!=None).filter(Profile.url_slug>=min_url_slug)

        else:
            q = db.session.query(Profile).filter(Profile.stripe_id!=None)


    number_considered = 0.0
    total_with_news = 0
    total_number_of_products_with_news = defaultdict(int)
    start_time = datetime.datetime.utcnow()
    for profile in windowed_query(q, Profile.url_slug, 25):
        if profile.is_paid_subscriber:

            number_considered += 1

            logger.info(u"count_news_for_subscribers: {url_slug}".format(
                url_slug=profile.url_slug))
            number_of_products_with_news = 0
            for product in profile.products_not_removed:
                metric = product.get_metric_by_name("altmetric_com", "news")
                if metric:
                    number_of_products_with_news += 1
            if number_of_products_with_news:
                total_number_of_products_with_news[number_of_products_with_news] += 1
                total_with_news += 1

            logger.info(u"of {total} profiles, total_with_news:{total_with_news} ({percent}%)\ntotal_number_of_products_with_news:{total_number_of_products_with_news}".format(
                total=number_considered,
                total_with_news=total_with_news, 
                percent=100*total_with_news/number_considered,
                total_number_of_products_with_news=total_number_of_products_with_news))
        else:
            logger.info(u"count_news_for_subscribers: not counting {url_slug} because not a subscriber".format(
                url_slug=profile.url_slug))


def send_drip_emails(url_slug=None, min_url_slug=None):
    MIN_AGE_DAYS_FOR_DRIP_EMAIL = 28
    DRIP_MILESTONE = "last-chance"

    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        drip_email_create_date = datetime.datetime.utcnow() - datetime.timedelta(days=MIN_AGE_DAYS_FOR_DRIP_EMAIL)
        logger.info(u"in send_drip_emails with drip_email_create_date:{drip_email_create_date}".format(
            drip_email_create_date=drip_email_create_date))

        # only profiles that have null stripe ids and are at least drip email days old
        q = db.session.query(Profile) \
                .filter(Profile.stripe_id==None) \
                .filter(Profile.created <= drip_email_create_date)
        if min_url_slug:
            q = q.filter(Profile.url_slug >= min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):
        # logger.info(u"in send_drip_emails with {url_slug}".format(
        #     url_slug=profile.url_slug))

        if profile.is_trialing and not profile.received_drip_email(DRIP_MILESTONE):
            logger.info(u"in send_drip_emails, sending email to: {url_slug}".format(
                url_slug=profile.url_slug))
            tasks.send_drip_email(profile, DRIP_MILESTONE)
            drip_log = log_drip_email(profile, DRIP_MILESTONE)
            logger.info(u"in send_drip_emails, SENT EMAIL to: {url_slug}".format(
                url_slug=profile.url_slug))
        else:
            pass
            # logger.info(u"in send_drip_emails, but NOT sending email to: {url_slug}".format(
            #     url_slug=profile.url_slug))


def ip_deets():
    from totalimpactwebapp.interaction import Interaction
    from totalimpactwebapp.interaction import get_ip_insights 
    q = db.session.query(Interaction)
    cache = {}
    for interaction in windowed_query(q, Interaction.ip, 25):
        if interaction.country:
            continue
            
        if interaction.ip in cache:
            interaction.country, interaction.user_type = cache[interaction.ip]
        else:
            insights = get_ip_insights(interaction.ip)
            interaction.country = insights.country.iso_code
            interaction.user_type = insights.traits.user_type
            cache[interaction.ip] = interaction.country, interaction.user_type
            print interaction.country, interaction.user_type
        db.session.add(interaction)
        commit(db)


def countries_for_all_profiles(url_slug=None, min_created_date=None):
    if url_slug:
        q = db.session.query(Profile).filter(Profile.url_slug==url_slug)
    else:
        q = db.session.query(Profile).filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None))
        if min_created_date:
            q = q.filter(Profile.created >= min_created_date)

    countries = {}
    start_time = datetime.datetime.utcnow()
    number_profiles = 0.0
    total_refreshes = 0
    for profile in windowed_query(q, Profile.created, 25):  # sort by created

        if number_profiles > 500:
            print "ok, got country data for 500 profiles. quitting."
            return True

        number_profiles += 1
        number_refreshes = 0.0
        print "getting countries for", profile.url_slug, profile.id, profile.created

        for my_country in profile.countries:
            pass


def refresh_tiid(tiid):
    tiids = refresh_products_from_tiids(None, [tiid])
    print tiids
    return tiids


def update_twitter_followers(max_pages):

    last_updated_days_ago = 1
    min_last_updated_date = datetime.datetime.utcnow() - datetime.timedelta(days=last_updated_days_ago)
    if not max_pages:
        max_pages = 10

    for i in range(max_pages):
        q = Tweeter.query.filter(Tweeter.is_deleted==None) \
            .filter(Tweeter.last_collected_date < min_last_updated_date) \
            .order_by(Tweeter.last_collected_date.asc()) \
            .limit(100)
        tweeters = q.all()
        if tweeters:
            get_and_save_tweeter_followers(tweeters)


def update_tweet_text_for_live_profiles(url_slug=None, min_url_slug=None):
    q = profile_query(url_slug, min_url_slug)

    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"in update_tweet_text_for_live_profiles for {url_slug}".format(
            url_slug=profile.url_slug))

        profile.parse_and_save_tweets()


def update_this_profile(profile):
    logger.info(u"**updating {url_slug: <16} is_live: {is_live}, next_refresh: {next_refresh}".format(
        url_slug=profile.url_slug, is_live=profile.is_live, next_refresh=profile.next_refresh.isoformat()[0:10]))

    try:
        if profile.is_live:
            number_products_before = len(profile.tiids)
            added_tiids = profile.update_all_linked_accounts(add_even_if_removed=False)
            number_products_after = number_products_before + len(added_tiids)
            if len(added_tiids)==0:
                logger.info(u"  NO CHANGE on update for {url_slug}, {number_products_before} products".format(
                    number_products_before=number_products_before,
                    url_slug=profile.url_slug))
            else:
                logger.info(u"  BEFORE={number_products_before}, AFTER={number_products_after}; {percent} for {url_slug}".format(
                    number_products_before=number_products_before,
                    number_products_after=number_products_after,
                    percent=100.0*(number_products_after-number_products_before)/number_products_before,
                    url_slug=profile.url_slug))

        # refresh all profiles, live and not, after the update from linked accounts is done
        profile.refresh_products("scheduled")  # puts them on celery

    except Exception as e:
        logger.exception(e)
        logger.debug(u"Exception in main loop on {url_slug}, so skipping".format(
            url_slug=profile.url_slug))


def update_profiles(limit=5, url_slug=None):
    # shouldn't need this, but hack to get around the fact that we limit profiles
    # before knowing how many have too many products
    # don't want to get too few profiles, then updates stall out
    hack_fudge_factor_for_too_big_profiles = 25

    if url_slug:
        q = db.session.query(Profile.id).filter(Profile.url_slug==url_slug)
    else:
        q = db.session.query(Profile.id).filter(Profile.next_refresh <= datetime.datetime.utcnow())
        q = q.limit(limit + hack_fudge_factor_for_too_big_profiles)

    number_profiles = 0.0
    for profile_id in q.all():
        product_count = db.session.query(Product.tiid).filter(Product.profile_id==profile_id, Product.removed==None).count()
        logger.info(u"profile {profile_id} has {product_count} products".format(
            profile_id=profile_id, product_count=product_count))

        if product_count > 500:
            logger.warning(u"Too many products (n={product_count}) for profile {profile_id}, skipping update".format(
                product_count=product_count, profile_id=profile_id))
        else:

            profile = Profile.query.get(profile_id)

            if limit and number_profiles >= limit:
                logger.info(u"updated all {limit} profiles, done for now.".format(
                    limit=limit))
                return
            
            update_this_profile(profile)

            number_profiles += 1


def live_profile_emails(args):
    url_slug = args.get("url_slug", None)
    min_url_slug = args.get("min_url_slug", None)

    q = profile_query(url_slug, min_url_slug)

    number_profiles_updated = 0.0
    for profile in windowed_query(q, Profile.url_slug, 25):
        print profile.email


def update_all_live_profiles(args):
    url_slug = args.get("url_slug", None)
    min_url_slug = args.get("min_url_slug", None)
    force_all = args.get("force_all", None)

    q = profile_query(url_slug, min_url_slug)
    if not force_all:
        q = q.filter(Profile.next_refresh <= datetime.datetime.utcnow())

    limit = args.get("limit", 5)
    if url_slug:
        limit = 1

    number_profiles_updated = 0.0
    for profile in windowed_query(q, Profile.next_refresh, 25, desc=False):
        product_count = len(profile.products_not_removed)
        logger.info(u"profile {url_slug} has {product_count} products".format(
            url_slug=profile.url_slug, product_count=product_count))

        if product_count > 500:
            logger.warning(u"Too many products (n={product_count}) for profile {url_slug}, skipping update".format(
                product_count=product_count, url_slug=profile.url_slug))
        else:            
            update_this_profile(profile)


            number_profiles_updated += 1
            if limit and number_profiles_updated >= limit:
                logger.info(u"updated all {limit} profiles, done for now.".format(
                    limit=limit))
                return

            pause_length = min(product_count * 3, 120)
            print "pausing", pause_length, "seconds after refreshing", product_count, "products"
            time.sleep(pause_length)


def say_hi(one, two, three):
    print "hi!"

def rq_metrics_for_all_live_profiles(args):
    url_slug = args.get("url_slug", None)
    tiid = args.get("tiid", None)
    no_rq = args.get("no_rq", False)
    limit = args.get("limit", 5)
    if url_slug:
        limit = 1

    queue_number = 0

    q = db.session.query(Product.tiid).select_from(Profile)
    q = q.filter(Product.removed == None)
    q = q.join(Profile.products)
    if url_slug:
        q = q.filter(Profile.url_slug==url_slug)
    elif tiid:
        q = q.filter(Product.tiid==tiid)
    else:
        from totalimpactwebapp.profile import default_free_trial_days
        min_created_date = datetime.datetime.utcnow() - datetime.timedelta(days=default_free_trial_days)
        q = q.filter(or_(Profile.is_advisor!=None, Profile.stripe_id!=None, Profile.created>=min_created_date))
        # q = q.filter(Profile.next_refresh <= datetime.datetime.utcnow())
        q = q.order_by(Product.last_refresh_finished)  # oldest first
        q = q.limit(limit)
    print "q=", q

    all_metrics_provider_names = [p.provider_name for p in ProviderFactory.get_providers(default_settings.PROVIDERS, "metrics")]

    for tiid in q.all():
        print "tiid", tiid

        for provider_name in all_metrics_provider_names:
            print "putting {} on rq queue to run metrics through {}".format(
                tiid, provider_name)
            if no_rq:
                print "asked for no-rq, so calling right now"
                provider_method_wrapper(tiid, provider_name, "metrics")
            else:
                job = ti_queues[queue_number].enqueue_call(
                    func=provider_method_wrapper,
                    args=(tiid, provider_name, "metrics"),
                    timeout=60 * 10,
                    result_ttl=0  # number of seconds
                    )
                job.save()


def debug_biblio_for_live_profiles(args):
    url_slug = args.get("url_slug", None)
    min_url_slug = args.get("min_url_slug", None)

    q = profile_query(url_slug, min_url_slug)

    from totalimpact.providers.bibtex import Bibtex
    bibtex_provider = Bibtex()

    from totalimpactwebapp.product import put_biblio_in_product

    for profile in windowed_query(q, Profile.url_slug, 25):
        logger.info(u"in debug_biblio_for_live_profiles for {url_slug}".format(
            url_slug=profile.url_slug))

        for product in profile.products_not_removed:
            if product.biblio \
                and hasattr(product.biblio, "journal") \
                and "journal =" in product.biblio.journal \
                and hasattr(product.biblio, "full_citation") \
                and "journal" in product.biblio.full_citation:
                    print "got one:", product.tiid, product.biblio.full_citation
                    aliases = bibtex_provider.member_items(product.biblio.full_citation)
                    print aliases
                    for alias in aliases:
                        (ns, nid) = alias
                        if ns=="biblio":
                            product = put_biblio_in_product(product, nid, provider_name="bibtex")
                            print product.biblio
                            db.session.merge(product)
                            commit(db)
            else:
                pass
                # print ".",


def main(function, args):
    if function=="emailreports":
        email_report_to_live_profiles(args["url_slug"], args["min_url_slug"], args["max_emails"])
    elif function=="email_tng":
        email_all_profiles_about_tng(args["url_slug"], args["min_url_slug"], args["max_emails"])
    elif function=="dedup":
        dedup_everyone(args["url_slug"], args["min_url_slug"])
    elif function=="productdeets":
        add_product_deets_for_everyone(args["url_slug"], args["skip_until_url_slug"])
    elif function=="refsets":
        build_refsets(args["save_after_every_profile"])
    elif function=="embed":
        collect_embed(args["url_slug"], args["min_url_slug"])
    elif function=="linked_accounts":
        linked_accounts(args["account_type"], args["url_slug"], args["min_url_slug"])
    elif function=="refresh_tweeted_products":
        refresh_tweeted_products(args["min_tiid"])
    elif function=="run_through_twitter_pages":
        run_through_twitter_pages(args["url_slug"], args["min_url_slug"])
    elif function=="count_news":
        count_news_for_subscribers(args["url_slug"], args["min_url_slug"])
    elif function=="drip_email":
        send_drip_emails(args["url_slug"], args["min_url_slug"])
    elif function=="profile_deets":
        profile_deets(args["url_slug"], args["min_url_slug"], args["start_days_ago"], args["end_days_ago"])
    elif function=="new_mendeley":
        collect_new_mendeley(args["url_slug"], args["min_url_slug"])
    elif function=="ip_deets":
        ip_deets()
    elif function=="run_through_altmetric_tweets":
        run_through_altmetric_tweets(args["url_slug"], args["min_url_slug"])
    elif function=="new_metrics_for_live_profiles":
        new_metrics_for_live_profiles(args["url_slug"], args["min_url_slug"], args["start_days_ago"])
    elif function=="borked_pinboards_for_life_profiles":
        borked_pinboards_for_life_profiles(args["url_slug"], args["min_url_slug"])
    elif function=="update_mendeley_countries_for_live_profiles":
        update_mendeley_countries_for_live_profiles(args["url_slug"], args["min_url_slug"])
    elif function=="update_profiles":
        update_profiles(args["limit"], args["url_slug"])
    elif function=="refresh_tiid":
        refresh_tiid(args["tiid"])
    elif function=="update_twitter_followers":
        update_twitter_followers(args["max_pages"])
    elif function=="update_tweet_text_for_live_profiles":
        update_tweet_text_for_live_profiles(args["url_slug"], args["min_url_slug"])
    else:
        # call function by its name in this module, with all args :)
        # http://stackoverflow.com/a/4605/596939

        globals()[function](args)


if __name__ == "__main__":

    db.create_all()
    
    # get args from the command line:
    parser = argparse.ArgumentParser(description="Run stuff.")
    parser.add_argument('function', type=str, help="one of emailreports, refsets, dedup, productdeets")
    parser.add_argument('--url_slug', default=None, type=str, help="url slug")
    parser.add_argument('--min_url_slug', default=None, type=str, help="min_url_slug")
    parser.add_argument('--tiid', default=None, type=str, help="tiid")
    parser.add_argument('--min_tiid', default=None, type=str, help="min_tiid")
    parser.add_argument('--save_after_every_profile', action='store_true', help="use to debug refsets, saves refsets to db after every profile.  slow.")
    parser.add_argument('--max_emails', default=None, type=int, help="max number of emails to send")
    parser.add_argument('--account_type', default=None, type=str, help="account_type")
    parser.add_argument('--start_days_ago', type=int)
    parser.add_argument('--end_days_ago', type=int)
    parser.add_argument('--limit', type=int, default=5)
    parser.add_argument('--max_pages', type=int)
    parser.add_argument('--force_all', type=int)
    parser.add_argument('--no-rq', action="store_true", help="do jobs in this thread")

    args = vars(parser.parse_args())
    function = args["function"]

    arg_string = dict((k, v) for (k, v) in args.iteritems() if v and k!="function")
    print u"daily.py {function} with {arg_string}".format(
        function=function.upper(), arg_string=arg_string)

    global logger
    logger = logging.getLogger("ti.daily.{function}".format(
        function=function))
    
    main(function, args)

    db.session.remove()
    