Skip to content

Latest commit

 

History

History
194 lines (139 loc) · 15.4 KB

README.md

File metadata and controls

194 lines (139 loc) · 15.4 KB
Raw

Fabflix Project

Members:

​ Qi He ([email protected]), Youan Lu ([email protected])

Part 1 Demo Video URL: https://youtu.be/LCVxxR6jmOk

Part 2 Demo Video URL: https://youtu.be/a5NNy3zP1j8

Part 3 Demo Video URL: https://youtu.be/YeDsWiWxFFc

Part 4 Demo Video URL: https://youtu.be/HlxYq-pPW9Y

Part 5 Demo Video URL: https://youtu.be/WvFB3SE39gQ

Deployment Instruction:

  • Require JDK 8 and Tomcat 8.
  • Open "context.xml" under META-INF folder.
  • Change the "username" and "password" to your own mysql username and password for testing, make sure you have moviedb and all data inserted correctly in your mysql database.
  • Build the project artifact, you will see a war file under the target folder.
  • Start Tomcat, and deploy the war file.
  • Open the fabflix app.

Contributions:

We also debug each other's code to implement the functionality.

Qi He:

  • Search
  • Multi search
  • Shopping Cart
  • Payment
  • Demo Video
  • PreparedStatements
  • XML Parsing
  • Android App

Youan Lu:

Single-instance Version Test Plan Graph Results Screenshot Average Query Time(ms) Average Search Servlet Time(ms) Average JDBC Time(ms) Analysis
Case 1: HTTP/1 thread 14 2.567011 2.493622 1 thread, time is relatively short.
Case 2: HTTP/10 threads 26 14.330229 14.273250 All three time are longer than 1 thread situation. The server has heavier workload.
Case 3: HTTPS/10 threads 34 21.995874 21.915524 Highest in TQ, TS, and TJ. Guess: the encryption process for Https took the extra time. TQ and TJ should be close to Case #2 but the result is higher than which of Case #2, the reason might be the network.
Case 4: HTTP/10 threads/No connection pooling 27 16.123943 16.054378 Without connection pooling, the performance is very close to Case #2, but it still adds a little more time because the connection is not pre-defined.
Scaled Version Test Plan Graph Results Screenshot Average Query Time(ms) Average Search Servlet Time(ms) Average JDBC Time(ms) Analysis
Case 1: HTTP/1 thread 76 3.845515 3.725063 Longer time compared to single-instance, because the server need to choose between master&slave first, and then route to it.
Case 2: HTTP/10 threads 75 3.504912 3.439293 very close to Case #1, but with shorter time, the scaled version works great when the server get heavier workload compared to the single-instance version, because the workload is distributed between to master and slave servers.
Case 3: HTTP/10 threads/No connection pooling 76 3.960356 3.890850 Still very close to previous two cases, but TJ and TS is higher because there is no pre-defined connections.

Substring matching design

Browse By Title:

  • if it's 0-9 and a-z, we use the substring matching by "title like 'x%'".
  • if it's *, we use the regex expression m.title regexp '^[a-z0-9]'.

Single search:

  • It searchs either title, year, director or stars by user input. We find any values that have search-item in any position where m.title like %search-item% or m.year= search-item or m.director like %search-item% or s.name like %search-item%
  • Notice that if you search by "Tom" for example. The movie list may display a movie that both title and star name do not include "Tom" sometimes. It's because we only need to display three star names. "Tom" is the actor in that movie but is not displayed in main movie page. If you click the movie name and go to single movie page, you can see "Tom" as the actor.

Advanced search/ Multiple search:

  • Title, year, director and stars can be searched together by user input. We allow user to type in these information by typing in text box. We find any values that include user input in any position
  • where m.title like '%title%' and m.year like '%year%' and m.director like '%director%' and s.name like '%star-name%'

Prepared Statement

We have changed all the statements of our servlets to prepared statements.

XML Parsers

Parsers:

MovieParser.java -- mains243.xml CastParser.java -- casts124.xml, actors63.xml

Insertion:

InsertParsedData.sql

The parsing process and insertion process are separated, the two .java files are for parsing, and they generate .txt files. We write the LOAD DATA query inside another file inside the root folder called InsertParsedData.sql.

XML Parsing Performance Tuning

  1. We put all existed useful data from Movies table, Genres table, Stars table, Genres_in_movies table, Stars_in_movies table into separated in-memory HashMaps before parsing, in order to filter out the duplicate data entry.
  2. We put all the parsed data into separated .txt files inside the self-generated ParsedFiles folder, each .txt file contains the data of a correlated table that will be inserted into the database later, each line in the .txt represents the values of an entry. The parsing process is approximately 1~2 seconds.
  3. We use the 'LOAD DATA' function of mysql to load the data from the .txt files, time is about 1~2 seconds, it is much faster compare to the naive approach.

XML Inconsistency Data

​ During the parsing process, if the program find the data is already existed in the database or some value are missing, the program will notify the user by printing a line of report in the terminal window, and the program also writes all the inconsistency data report and duplicate data report into a .txt file inside the ParsedFiles folder. After running MovieParser and CastParser, there will be two .txt files named moviesInconsistentDataReport.txt and castsInconsistentDataReport.txt shows all the inconsistency and duplicate report. The format is like this:

MovieParser at Movie No.1 ID=H1: Genre 'Drama' already exist
MovieParser at Movie No.2 ID=H2: Missing Movie Genre
ActorParser at Actor No.11 Name=Maury Abram: Missing Birth Year
ActorParser at Actor No.11 Name=Victoria Abril: Star 'Victoria Abril' already exist
CastParser at Index No.26649: Star 'Francisco Rabal' already exist
CastParser at Index No.26650: Star 's a' already exist

Note: The CastParser.java actually consist of both CastParser and ActorParser, and the inconsistency data report of these two parsers are in the same castsInconsistentDataReport.txt file.