mid summer night dream project

Author: ptyadana

Database Clinics - MySQL/03.Mid Summer Night Dream/01.Mid_Summer_Night_Dream_Schema_Table_Creation.sql

@@ -0,0 +1,17 @@
+CREATE SCHEMA shakespeare;
+
+USE shakespeare;
+
+/* play and characters table */
+CREATE TABLE amnd(
+	line_number INT NOT NULL AUTO_INCREMENT,
+    char_name TEXT,
+    play_text TEXT,
+    PRIMARY KEY (line_number)
+);
+
+/* performance table */
+CREATE TABLE performance(
+	query_type TEXT,
+    query_time FLOAT
+);

Database Clinics - MySQL/03.Mid Summer Night Dream/02.Mid_Summer_Night_Dream_Process_and_Insert_into_DB.py

@@ -0,0 +1,58 @@
+# read data from two text files and load the data into databases
+
+import pymysql
+import time
+
+myConnection = pymysql.connect(
+    host="localhost", user="root", password="root", db="shakespeare")
+
+cur = myConnection.cursor()
+
+# read character file and get all characters
+with open("datasets/characters.txt", "r") as char:
+    characterList = char.read().splitlines()
+
+# as first character is unknown in Play file, initialize as Unknown
+currentCharacter = "Unknown"
+
+start_time = time.time()
+
+createSQL = "INSERT INTO amnd(char_name, play_text) VALUES(%s, %s);"
+
+# Part 1: process Play file text
+# to Create a Record for each line in the play
+# The character who is speaking
+# The line number
+# The phrase itself, trimmed of spaces
+with open("datasets/A_Midsummer_Nights_Dream.txt", "r") as playlines:
+    for line in playlines:
+        if line.upper().strip() in characterList:
+            currentCharacter = line.upper().strip()
+            print("changing character to : ", currentCharacter)
+        else:
+            sql_values = currentCharacter, line.strip()
+            print("writing line : ", sql_values)
+            cur.execute(createSQL, sql_values)
+
+myConnection.commit()
+end_time = time.time()
+
+
+# Part 2: process for Query Performance Calculation
+cur.execute('SELECT COUNT(line_number) FROM amnd;')
+numPlayLines = cur.fetchall()[0][0]
+print(numPlayLines, 'rows')
+
+# calculate query execution time
+queryExecTime = end_time - start_time
+print("Total query time: ", queryExecTime)
+
+queryTimePerLine = queryExecTime / numPlayLines
+print("Query time per line: ", queryTimePerLine)
+
+# record query execution time into performance table
+insertPerformanceSQL = "INSERT INTO performance VALUES('CREATE',%s);"
+cur.execute(insertPerformanceSQL, queryTimePerLine)
+
+myConnection.commit()
+myConnection.close()

Database Clinics - MySQL/03.Mid Summer Night Dream/03.Mid_Summer_Night_Dream_Update_Character_Names.py

@@ -0,0 +1,44 @@
+# Update Character Name in uppercase for every line in plays
+# example: "Now, fair Hippolyta, our nuptial hour" => Now, fair HIPPOLYTA, our nuptial hour
+
+
+import pymysql
+import time
+
+myConnection = pymysql.connect(
+    host="localhost", user="root", password="root", db="shakespeare")
+
+cur = myConnection.cursor()
+start_time = time.time()
+
+# Part 1: Update Character Name in uppercase for every line in plays
+updateSQL = "UPDATE amnd SET play_text = REPLACE(play_text, %s, %s);"
+
+with open("datasets/characters.txt", "r") as char:
+    for character in char.read().splitlines():
+        print("capitalizing occurences of ", character)
+        updateStrings = character.capitalize(), character.upper()
+        cur.execute(updateSQL, updateStrings)
+
+myConnection.commit()
+
+end_time = time.time()
+
+# Part 2: process for Query Performance Calculation
+cur.execute('SELECT COUNT(line_number) FROM amnd;')
+numPlayLines = cur.fetchall()[0][0]
+print(numPlayLines, 'rows')
+
+# calculate query execution time
+queryExecTime = end_time - start_time
+print("Total query time: ", queryExecTime)
+
+queryTimePerLine = queryExecTime / numPlayLines
+print("Query time per line: ", queryTimePerLine)
+
+# record query execution time into performance table
+insertPerformanceSQL = "INSERT INTO performance VALUES('UPDATE',%s);"
+cur.execute(insertPerformanceSQL, queryTimePerLine)
+
+myConnection.commit()
+myConnection.close()

Database Clinics - MySQL/03.Mid Summer Night Dream/04.Mid_Summer_Night_Dream_Delete_Stage_Directions.py

@@ -0,0 +1,45 @@
+# Delete Stage Directions in Plays text such as "Exit", "Enter", etc
+
+import pymysql
+import time
+
+myConnection = pymysql.connect(
+    host="localhost", user="root", password="root", db="shakespeare")
+
+cur = myConnection.cursor()
+start_time = time.time()
+
+cur.execute('SELECT COUNT(line_number) FROM amnd;')
+numPlayLines_Before_Delete = cur.fetchall()[0][0]
+
+
+# delete stage directions
+# RLIKE : regular expression like, ^ means starts with , | means OR
+cur.execute(
+    "DELETE FROM amnd WHERE play_text RLIKE '^enter|^exit|^act|^scence|^exeunt';")
+
+print("Deleting lines..")
+
+end_time = time.time()
+
+myConnection.commit()
+
+
+cur.execute('SELECT COUNT(line_number) FROM amnd;')
+numPlayLines_After_Delete = cur.fetchall()[0][0]
+numPlayLInes_Deleted = numPlayLines_Before_Delete - numPlayLines_After_Delete
+print(numPlayLInes_Deleted, 'rows')
+
+# calculate query execution time
+queryExecTime = end_time - start_time
+print("Total query time: ", queryExecTime)
+
+queryTimePerLine = queryExecTime / numPlayLInes_Deleted
+print("Query time per line: ", queryTimePerLine)
+
+# record query execution time into performance table
+insertPerformanceSQL = "INSERT INTO performance VALUES('DELETE',%s);"
+cur.execute(insertPerformanceSQL, queryTimePerLine)
+
+myConnection.commit()
+myConnection.close()

Database Clinics - MySQL/03.Mid Summer Night Dream/05.Mid_Summer_Night_Dream_Get_Plays.py

@@ -0,0 +1,37 @@
+# get the plays and measure the performance of query
+
+import pymysql
+import time
+
+myConnection = pymysql.connect(
+    host="localhost", user="root", password="root", db="shakespeare")
+
+cur = myConnection.cursor()
+start_time = time.time()
+
+# Part 1: Get the plays from database
+cur.execute("SELECT play_text FROM amnd;")
+
+for line in cur.fetchall():
+    print(line[0])
+
+end_time = time.time()
+
+# Part 2: process for Query Performance Calculation
+cur.execute('SELECT COUNT(line_number) FROM amnd;')
+numPlayLines = cur.fetchall()[0][0]
+print(numPlayLines, 'rows')
+
+# calculate query execution time
+queryExecTime = end_time - start_time
+print("Total query time: ", queryExecTime)
+
+queryTimePerLine = queryExecTime / numPlayLines
+print("Query time per line: ", queryTimePerLine)
+
+# record query execution time into performance table
+insertPerformanceSQL = "INSERT INTO performance VALUES('READ',%s);"
+cur.execute(insertPerformanceSQL, queryTimePerLine)
+
+myConnection.commit()
+myConnection.close()