Partitions in Proc SQL - sas

Question

Can we use partitions inside Proc Sql, if not could you please help me the equivalent logic to use?
Proc sql;
select user_id,
content_title,
calendar_date,
sum(watch_second)/3600 as hours_watched,
sum(hours_watched) over (partition by user_id, content_title order by calendar_date) AS cumulative_hours,
available_hours,
cumulative_hours / available_hours as pct_completed
from watch_history as a
inner join dim_user s
ON s.USER_ID = a.USER_ID
left join dim_content_meta pb
ON pb.metrics_video_id = a.metrics_video_id
inner join sascidm.coop_top50 v
on pb.hummus_show_id = v.hummus_show_id
where pb.hummus_playback_type in ('VOD')
AND pb.hummus_show_type = 'series'
order by 1,2,3;
quit;
Error:
sum(watch_second)/3600 as hours_watched,
29 sum(hours_watched) over (partition by user_id, content_title order by calendar_date) AS cumulative_hours,
____
22
76
ERROR 22-322: Syntax error, expecting one of the following: !, !!, &, *, **, +, ',', -, /, <, <=, <>, =, >, >=, ?, AND, BETWEEN,
CONTAINS, EQ, EQT, GE, GET, GT, GTT, LE, LET, LIKE, LT, LTT, NE, NET, OR, ^=, |, ||, ~=.
ERROR 76-322: Syntax error, statement will be ignored.

Answer 1

No, proc sql supports only a relatively limited set of SQL that's close to ANSI SQL from a few decades ago, and does not support partition by as that's much newer.
For the most part, a SAS programmer would instead use SAS procs to compute things like this rather than SQL - PROC TABULATE could probably do this I suspect.

Answer 2

It is simple in a DATA step to generate a cummulative sum. Just use a SUM statement. To reset by groups use a FIRST. flags to detect when a new group is starting.
Proc sql;
create table raw as
select
user_id
,content_title
,calendar_date
,sum(watch_second)/3600 as hours_watched
/*
,sum(hours_watched) over (partition by user_id, content_title order by calendar_date) AS cumulative_hours
*/
,available_hours
/*
,cumulative_hours / available_hours as pct_completed
*/
from watch_history as a
inner join dim_user s
ON s.USER_ID = a.USER_ID
left join dim_content_meta pb
ON pb.metrics_video_id = a.metrics_video_id
inner join sascidm.coop_top50 v
on pb.hummus_show_id = v.hummus_show_id
where pb.hummus_playback_type in ('VOD')
AND pb.hummus_show_type = 'series'
order by 1,2,3
;
quit;
data want;
set raw ;
by user_id content_title calendar_date;
if first.content_title then cumulative_hours=0;
cumulative_hours + hours_watched;
pct_completed = cumulative_hours / available_hours;
run;