def top_year(year, connection):
resp = connection.read("""
SELECT SUM(t.pdf) as downloads, t.article, a.url,
a.title, a.abstract, a.collection, a.posted, a.doi
FROM article_traffic t
INNER JOIN articles a ON t.article=a.id
WHERE t.year = %s
AND a.posted >= '%s-01-01'
AND a.posted <= '%s-12-31'
GROUP BY 2,3,4,5,6,7,8
ORDER BY 1 DESC
LIMIT 25
""", (year,year,year))
if len(resp) == 0:
return []
results = [models.SearchResultArticle(a, connection) for a in resp]
return results
def paper_query(q, categories, timeframe, metric, page, page_size, connection):
"""Returns a list of the most downloaded papers that meet a given set of constraints.
Arguments:
- connection: a database Connection object.
- q: A search string to compare against article abstracts,
titles and author names. (Title matches are weighted more heavily.)
- categories: A list of bioRxiv categories the results can be in.
- timeframe: A description of the range of dates on which to
base the rankings (i.e. "alltime" or "lastmonth")
- metric: Which article-level statistic to use when sorting results
- page: Which page of the results to display (0-indexed)
- page_size: How many entries should be returned
Returns:
- An list of Article objects that meet the search criteria, sorted by the
specified metric in descending order.
"""
# We build two queries, 'select' and 'countselect': one to get the
# current page of results, and one to figure out the total number
# of results
select = "SELECT "
if metric == "downloads":
select += "r.downloads"
elif metric == "twitter":
select += "SUM(r.count)"
select += ", a.id, a.url, a.title, a.abstract, a.collection, a.posted, a.doi"
countselect = "SELECT COUNT(DISTINCT a.id)"
params = ()
query = ""
if q != "": # if there's a text search specified
params = (q, )
query += f' FROM {config.db["schema"]}.articles AS a INNER JOIN {config.db["schema"]}.'
if metric == "twitter":
query += "crossref_daily"
elif metric == "downloads":
query_times = {
"alltime": "alltime_ranks",
"ytd": "ytd_ranks",
"lastmonth": "month_ranks",
}
query += query_times[timeframe]
if metric == "twitter":
query += " AS r ON r.doi=a.doi"
elif metric == "downloads":
query += " AS r ON r.article=a.id"
if q != "":
query += """, plainto_tsquery(%s) query,
coalesce(setweight(a.title_vector, 'A') || setweight(a.abstract_vector, 'C') || setweight(a.author_vector, 'D')) totalvector
"""
# add a WHERE clause if we need one:
# (all-time twitter stats don't require it)
if metric == "downloads" or (metric == "twitter" and timeframe != "alltime"
) or len(categories) > 0:
query += " WHERE "
if metric == "downloads":
query += "r.downloads > 0"
if q != "" or len(categories) > 0:
query += " AND "
if q != "":
query += "query @@ totalvector "
if len(categories) > 0 or (metric == "twitter"
and timeframe != "alltime"):
query += " AND "
if len(categories) > 0:
query += "collection=ANY(%s)"
if q != "":
params = (q, categories)
else:
params = (categories, )
if metric == "twitter" and timeframe != "alltime":
query += " AND "
if metric == "twitter" and timeframe != "alltime":
query += "r.source_date > now() - interval "
query_times = {"day": 2, "week": 7, "month": 30, "year": 365}
query += f"'{query_times[timeframe]} days' "
# this is the last piece of the query we need for the one
# that counts the total number of results
countselect += query
resp = connection.read(countselect, params)
total = resp[0][0]
# continue building the query to get the full list of results:
if metric == "twitter":
query += " GROUP BY a.id"
query += " ORDER BY "
if metric == "downloads":
query += "r.rank ASC"
elif metric == "twitter":
query += "SUM(r.count) DESC"
query += f" LIMIT {page_size}"
if page > 0:
query += f" OFFSET {page * page_size}"
query += ";"
select += query
result = connection.read(select, params)
results = [models.SearchResultArticle(a, connection) for a in result]
return results, total
