add submissions

provsql/src/provenance.py

@@ -13,8 +13,114 @@ import traceback
 
 tablesWithProvenance = {}
 
+twojoins = ["INNERJOIN", "LEFTJOIN", "RIGHTJOIN", "FULLJOIN", "NATURALJOIN"]
+threejoins = ["LEFTOUTERJOIN", "RIGHTOUTERJOIN", "FULLOUTERJOIN"]
+
 def tablesInQuery(query):
-    parsed = sp.parse(query)[0]
+    output = []
+    tokens = query.split(" ")
+    tokens = list(filter(lambda a: a != "", tokens))
+    state = "not in"
+    idx = 0
+    while idx < len(tokens):
+        if tokens[idx] == "":
+            idx += 1
+            continue
+        if state == "not in":
+            if tokens[idx].upper() == "FROM":
+                state = "from in"
+            if tokens[idx].upper() == "JOIN":
+                state = "join in"
+        elif state == "from in":
+            if tokens[idx].upper() in ["WHERE", "ON"]:
+                state = "not in"
+                idx += 1
+                continue
+            if tokens[idx].upper() == "JOIN":
+                state = "join in"
+                idx += 1
+                continue
+            if (tokens[idx]+tokens[idx+1]).upper() in twojoins:
+                state = "join in"
+                idx += 2
+                continue
+            if (tokens[idx]+tokens[idx+1]+tokens[idx+2]).upper() in threejoins:
+                state = "join in"
+                idx += 3
+                continue
+            if "SELECT" in tokens[idx].upper():
+                idx += 1
+                state = "not in"
+                continue
+            if "," in tokens[idx]:
+                output.append(tokens[idx].split(",")[0])
+            else:
+                if tokens[idx + 1].upper() == "AS":
+                    output.append(tokens[idx])
+                    idx += 2
+                elif "," in tokens[idx + 1]:
+                    output.append(tokens[idx])
+                    idx += 1
+                elif tokens[idx+1].upper() == "WHERE" :
+                    state = "not in"
+                    output.append(tokens[idx])
+                    idx += 1
+                elif tokens[idx+2].upper() == "WHERE":
+                    state = "not in"
+                    output.append(tokens[idx])
+                    idx += 2
+                elif (tokens[idx+2]+tokens[idx+3]).upper() in twojoins:
+                    output.append(tokens[idx])
+                    idx += 1
+                elif (tokens[idx+2]+tokens[idx+3]+tokens[idx+4]).upper() in threejoins:
+                    output.append(tokens[idx])
+                    idx += 1
+                else:
+                    output.append(tokens[idx])
+        elif state == "join in":
+            if tokens[idx].upper() in ["WHERE", "ON"]:
+                state = "not in"
+                idx += 1
+                continue
+            if tokens[idx].upper() == "JOIN":
+                state = "join in"
+                idx += 1
+                continue
+            if (tokens[idx]+tokens[idx+1]).upper() in twojoins:
+                state = "join in"
+                idx += 2
+                continue
+            if (tokens[idx]+tokens[idx+1]+tokens[idx+2]).upper() in threejoins:
+                state = "join in"
+                idx += 3
+                continue
+            if "SELECT" in tokens[idx].upper():
+                idx += 1
+                state = "not in"
+                continue
+            if tokens[idx+2].upper() == "USING":
+                output.append(tokens[idx])
+                idx += 2
+                state = "not in"
+                continue
+            if tokens[idx + 1].upper() == "AS":
+                output.append(tokens[idx])
+                idx += 2
+            elif tokens[idx+1].upper() in ["WHERE", "ON", "JOIN"] or (tokens[idx+1]+tokens[idx+2]).upper() in twojoins or (tokens[idx+1]+tokens[idx+2]+tokens[idx+3]).upper() in threejoins:
+                output.append(tokens[idx])
+            else:
+                output.append(tokens[idx])
+                idx+=1
+        idx += 1
+                
+
+    
+    return list(set(output))
+
+            
+
+
+    '''parsed = sp.parse(query)[0]
     output = []
     state = ""
     for token in parsed.tokens:
@@ -43,7 +149,8 @@ def tablesInQuery(query):
                 output.append(cur.split("as")[0].strip())
             break
 
-    return list(set(output))
+    return list(set(output))'''
+
 
 
 #Preprocess query

submissions/Solutions/Q1-solution.sql

+SELECT Students.FirstName, Students.LastName, Courses.Instructor FROM Students INNER JOIN Enrollments ON Students.NetId=Enrollments.NetId INNER JOIN Courses ON Enrollments.CRN=Courses.CRN WHERE (Students.FirstName LIKE 'A%' OR Students.LastName LIKE 'A%') AND (Courses.Instructor LIKE 'A%');
\ No newline at end of file

submissions/Solutions/Q2-solution.sql

+SELECT DISTINCT S.NetId, S.FirstName, S.LastName FROM   Students S, Enrollments E, Courses C WHERE  S.NetId = E.NetId  AND E.CRN = C.CRN  AND S.Department = 'CS' AND C.Department = 'ECE' AND E.Score <=  All (SELECT E2.Score FROM   Enrollments E2  WHERE  E2.CRN = E.CRN) ORDER BY S.LastName, S.NetId DESC;
\ No newline at end of file

submissions/Solutions/Q3-solution.sql

+SELECT E.CRN,C.Title,C.Department, AVG(Score) AS avgscore FROM Enrollments E, Courses C WHERE E.CRN = C.CRN AND C.Department='ECE' GROUP BY E.CRN HAVING avgscore > 70 ORDER BY avgscore DESC;
\ No newline at end of file

submissions/Solutions/Q4-solution.sql

+SELECT p1.BrandName, COUNT(DISTINCT p2.CustomerId) as NumOfCustomers FROM Purchases p2 RIGHT JOIN Products p1 ON p2.ProductId = p1.ProductId GROUP BY p1.BrandName ORDER BY NumOfCustomers DESC, p1.BrandName ASC;
\ No newline at end of file

submissions/Solutions/Q5-solution.sql

+SELECT Students.FirstName, Students.LastName, ROUND(Enrollments.Score,2), Courses.Title, ROUND(E.avgscore,2) FROM Students NATURAL JOIN Enrollments NATURAL JOIN (SELECT CRN, AVG(Score) AS avgscore FROM Enrollments GROUP BY CRN) AS E JOIN Courses USING (CRN) WHERE Score > avgscore ORDER BY Students.FirstName, Courses.Title DESC;
\ No newline at end of file

submissions/Solutions/Q6-solution.sql

+(SELECT S.NetID, S.FirstName, S.LastName FROM Students S NATURAL JOIN Enrollments E JOIN Courses C USING (CRN) WHERE S.Department = 'CS' AND C.Department = 'CS' AND E.Score  > 80) UNION (SELECT S1.NetID, S1.FirstName, S1.LastName FROM Students S1 NATURAL JOIN Enrollments E1 JOIN Courses C1 USING (CRN) WHERE S1.Department = 'ECE' AND C1.Department = 'CS' AND E1.Score  > 70) order by NetId
\ No newline at end of file

submissions/Solutions/Q7-solution.sql

+SELECT FirstName, LastName, COUNT(PurchaseId) AS cnt
+FROM Customers LEFT JOIN Purchases USING (CustomerId)
+GROUP BY FirstName, LastName
+ORDER BY FirstName, LastName DESC;
\ No newline at end of file

submissions/Solutions/Q8-solution.sql

+SELECT C.CustomerId, C.FirstName, C.LastName, C.PhoneNumber
+FROM Customers C NATURAL JOIN Purchases P
+WHERE C.CustomerId IN (SELECT C1.CustomerId
+					     FROM Customers C1 NATURAL JOIN Purchases P1
+					     WHERE P1.Price >=  ALL(SELECT Price
+											     FROM Purchases
+											     WHERE ProductId = P1.ProductId))
+GROUP BY P.CustomerId
+ORDER BY P.CustomerId DESC
\ No newline at end of file

submissions/extract.py

@@ -56,7 +56,7 @@ def main():
         os.makedirs("hw1-provenance")
     except OSError as e:
         print(e)
-    conn = psycopg2.connect("")
+    conn = psycopg2.connect("dbname=provdb user=joker password=m")
     cur = conn.cursor()
     extractData(conn, cur)
     conn.close()
@@ -68,7 +68,7 @@ def main():
             submissions = question_submissions[student_id]
             for sub_id in list(submissions.keys()):
                 try:
-                    conn = psycopg2.connect("")
+                    conn = psycopg2.connect("dbname=provdb user=joker password=m")
                     cur = conn.cursor()
                     commitAndExecute(conn, cur, 'SET search_path TO public, provsql;')
                     provenance.exe_query(submissions[sub_id], f, cur, conn)