Spark SQL in PySpark throws ParseError Exception - pyspark

Question

I have a pyspark code running in an AWS glue job. I have some temporary tables created from dataframes in the same code using the function createOrReplaceTempView().
I am running a multiline SQL query within the code to fetch data from these temporary tables and store in a dataframe. However, I am getting ParseError Exception. The same query runs absolutely fine in a SQL client. Below is the query. Can someone please help? Unfortunately I don't get any error message other than the text "ParseError Exception"
def getSCMCaseHistLanding(self):
landingDF = self.spark.sql("""SELECT aud.case_id,
r.name AS role_name,
cas.account_id,
cas.created_time AS case_created_time,
cas.last_updated_time,
cas.screening_decision,
aud.created_time AS record_created_time,
aud.username,
row_number() OVER (PARTITION BY aud.case_id ORDER BY date_trunc('minute', aud.created_time) + date_part('second', aud.created_time)::INT / 10 * INTERVAL '10 sec') rank,
max(CASE WHEN field = 'status_id' THEN aud.value ELSE NULL END) AS d_status,
max(CASE WHEN field = 'status_id' THEN SPLIT_PART(SPLIT_PART(aud.description, 'from ', 2), ' to', 1) ELSE NULL END) AS src_status,
max(CASE WHEN lower(field) = 'annotation' THEN [value] ELSE NULL END) AS annotation,
max(CASE WHEN lower(field) = 'reason_id' THEN [value] ELSE NULL END) AS reason,
max(CASE WHEN lower(field) = 'approver_id' THEN [value] ELSE NULL END) AS approver,
r.team_name AS approver_source,
max(CASE WHEN lower(field) = 'decision_id' THEN [value] ELSE NULL END) AS decision,
max(CASE WHEN field = 'status_id' THEN aud.created_time ELSE NULL END) AS lsd,
last_value(lsd IGNORE NULLS) OVER (PARTITION BY aud.case_id ORDER BY record_created_time ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS last_status_changed_date,
max(CASE WHEN aud.field = 'assigned_to' THEN REVERSE(SPLIT_PART(REVERSE(aud.value), ' ', 1)) ELSE NULL END) AS assigned_to_t,
last_value(assigned_to_t IGNORE NULLS)
OVER (PARTITION BY aud.case_id ORDER BY record_created_time ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS assigned_to,
max(CASE WHEN aud.field = 'assigned_to' THEN aud.username ELSE NULL END) AS assigned_by,
max(CASE
WHEN (aud.field = 'assigned_to' AND aud.description LIKE 'Workbasket%') OR
(aud.field = 'assigned_to' AND aud.description LIKE '%by%') THEN 'Auto Assign'
WHEN aud.field = 'assigned_to' AND aud.description LIKE '%themself%' THEN 'Get Next'
END) AS assignment_method,
max(
CASE WHEN aud.field = 'assigned_to' THEN aud.description ELSE NULL END) AS assignment_detail,
max(CASE WHEN lower(field) = 'state_id' THEN [value] ELSE NULL END) AS ll_state,
last_value(d_status IGNORE NULLS)
OVER (PARTITION BY aud.case_id ORDER BY record_created_time ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS dest_status,
last_value(ll_state IGNORE NULLS)
OVER (PARTITION BY aud.case_id ORDER BY record_created_time ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS l_state,
CASE WHEN l_state IS NULL THEN 'Open' ELSE l_state END AS state,
max(CASE WHEN lower(field) = 'responsive_action_id' THEN [value] ELSE NULL END) AS responsive_action,
max(CASE WHEN type_name = 'Language' THEN skill.skill_name ELSE NULL END) AS language_skill,
max(
CASE WHEN type_name = 'Business' THEN skill.skill_name ELSE NULL END) AS business_skill,
max(CASE
WHEN lower(field) = 'annotation' AND lower(value) LIKE 'f+ applied by jarvis%' THEN 'jarvis'
WHEN lower(field) = 'annotation' AND (lower(value) LIKE '%bulk action%'
OR lower(value) LIKE '%moving cases to invalid data%') THEN 'bulk action'
WHEN lower(field) = 'annotation' AND lower(value) LIKE '%t+ decision recommended by scr model%'
THEN 'SCR T+ Model'
WHEN lower(field) = 'annotation' AND lower(value) LIKE '%f+ applied by scr model%'
THEN 'SCR F+ Model'
ELSE NULL END) AS source_of_action,
max(CASE
WHEN lower(field) = 'annotation' AND lower(value) LIKE '%duplicate case%' THEN 'Y'
ELSE 'N' END) AS duplicate_case,
max(CASE
WHEN field = 'state_id' AND aud.description = 'Case reopened' THEN
date_trunc('minute', aud.created_time) +
date_part('second', aud.created_time)::INT / 10 * INTERVAL '10 sec'
ELSE NULL END) AS case_reopened_time,
max(0) AS active_record_flag
FROM cm_spectre_case_audit aud
JOIN cm_spectre_case cas
ON cas.case_id = aud.case_id
LEFT JOIN v_cm_user_snapshot u ON
CASE
WHEN aud.field = 'assigned_to' THEN REVERSE(SPLIT_PART(REVERSE(aud.value), ' ', 1))
ELSE aud.username
END = u.alias AND aud.created_time BETWEEN u.start_time AND u.end_time
LEFT JOIN cm_role r ON u.current_role_id = r.role_id
LEFT JOIN cm_lookup_skills lookup_skill ON lookup_skill.case_id = cas.case_id
LEFT JOIN cm_skill skill ON skill.skill_id = lookup_skill.skill_id
LEFT JOIN cm_skill_type skill_type ON skill_type.type_id = skill.type_id
WHERE 1 = 1
AND NOT (field = 'created_time'
AND username = 'System')
AND NOT (field = 'annotation'
AND username = 'System')
AND NOT (field = 'skill_name_')
AND NOT (field = 'assigned_to'
AND lower(aud.description) LIKE '%unassigned%')
AND NOT (field = 'attachment')
AND NOT (field = 'accept_list_id')
AND NOT (field LIKE 'screening_match_id%')
GROUP BY aud.case_id, cas.account_id, role_name, r.team_name, cas.created_time, cas.last_updated_time,
cas.screening_decision,
aud.created_time,
aud.username
""")
return landingDF