cinemaanalytics

JFuture 2019 Cinema Analytics System Challenge

Challenge

Imagine that you're a developer working on Cinema Analytics System, who needs to collect data and make calculations on the following items:

1. Display the dynamics of film releases in China and the United States over the past 3 years. Please take into account 5 different film genres.

   Example: Fiction, 2017: China — 23 films; USA — 42. Fiction, 2018: China — 10 films; USA — 22.

2. Make the list of top 5 Directors of the highest-rated films.

Assumptions

1. The data is collected for 2019, 2018 and 2017 years, as 2019 is almost over. Due to automatic data parsing and loading it is easy to add support for 2016 year instead of 2019 by using the corresponding link to the Wikipedia page.

2. When setting the gross (in USD) value for the film the mapping is done by the film's title, so there is the potential chance of duplicate films with the same title per year per country.

3. The actual output format for the "top 5 Directors of the highest-rated films" has not been explicitly specified, so it is set to "Director (Country) / Title (month day, year) / Gross (in USD) / [Genres]".

4. "Top 5 Directors of the highest-rated films" query is run on top of the whole data set, not per country.

5. Criteria for choosing the highest-rated films is based on the gross (in USD) value of the film.

Information gathering and calculation

• Data for the films is taken from Wikipedia pages using Wikipedia REST API.

• For each of the page the corresponding GET /page/title/{title} has been called manually to locate revision value (rev in the response), so the data is "frozen" and parsing can rely on known page HTML format.

• Each page is then programmatically fetched using Wikipedia REST API and parsed for the corresponding data, both films and rankings (gross value in USD). HTML parsing is done using jsoup Java library.

• Data is "flushed" (see configuration parameter below) using analytics.action=flush option when running into the local JSON data files. Which are then used for the analytics queries, unless analytics.mode=online when local data is ignored, and the data is fetched from Wikipedia.

• There are subtle differences in both China and USA Wikipedia films pages, so there are different parsers for each of the countries.

• In addition, there are subtle differences in parsing current 2019 year data, and passed years, like 2018 and 2017, this is also reflected in the parser implementations.

Here is the list of the Wikipedia pages used to fetch and parse the data (these are also controlled by china.sources and usa.sources configuration parameters described below):

• List of Chinese films of 2017
• List of Chinese films of 2018
• List of Chinese films of 2019
• List of American films of 2017
• List of American films of 2018
• List of American films of 2019

Configuration

The application supports a few configuration parameters:

parameter	values	default	description
analytics.action	flush - (it then ignores the analytics.mode property, i.e. it is always run in online mode)	run	action to run
	run - (run actual analytics)
analytics.mode	online - (fetches the data from the Wikipedia REST API)	offline	either read local data or remote
	offline - (reads the data from the locally stored JSON data files)
analytics.genres	list of genres (from Genreenum separated by comma)	COMEDY,DRAMA,SCI_FI,HISTORY,FANTASY	genres to use in the dynamics of film releases analytics
analytics.topdirectors	number	5	number of top directors
useragent.email	email of user running the analytics in the online mode	[email protected]	Wikipedia REST API requires to set the contact information in the User-Agent header
china.sources	list of Wikipedia pages to retrieve data from	see application.properties	list of china films pages in Wikipedia
usa.sources	list of Wikipedia pages to retrieve data from	see application.properties	list of usa films pages in Wikipedia

Note: if you are using analytics.mode=online be sure you change the value of the useragent.email unless you want to use the default value!

To set the parameter on run either (in the priority order):

• set JVM system property by adding -D for the running command, ex.: gradle -Danalytics.action=flush run or gradle -Danalytics.mode=online -Danalytics.genres=DRAMA,COMEDY,NATURE run, etc.
• OS specific environment variable (in that case upper case the property name and replace . with _, i.e. analytics.action becomes ANALYTICS_ACTION
• change the value directly in the application.properties

Running

gradle -q run

It assumes you have Gradle installed. Otherwise it is easy to install it with SDKMAN: sdk install gradle.

Output

Comedy, 2017: China - 99 films; USA - 81. Comedy, 2018: China - 100 films; USA - 68. Comedy, 2019: China - 29 films; USA - 75.
Drama, 2017: China - 146 films; USA - 109. Drama, 2018: China - 169 films; USA - 117. Drama, 2019: China - 51 films; USA - 102.
Sci-Fi, 2017: China - 5 films; USA - 30. Sci-Fi, 2018: China - 5 films; USA - 25. Sci-Fi, 2019: China - 4 films; USA - 21.
History, 2017: China - 12 films; USA - 3. History, 2018: China - 4 films; USA - 4. History, 2019: China - 3 films; USA - 1.
Fantasy, 2017: China - 34 films; USA - 19. Fantasy, 2018: China - 32 films; USA - 20. Fantasy, 2019: China - 10 films; USA - 20.
------------------------------
Anthony and Joe Russo (USA) / Avengers: Endgame (APRIL 26, 2019) / $858,373,000 / [Action, Adventure, Drama, Epic, Fantasy, Sci-Fi, Superhero]
Wu Jing (CHINA) / Wolf Warriors 2 (JULY 27, 2017) / $854,248,869 / [Action]
Yu Yang (CHINA) / Nezha (JULY 26, 2019) / $710,400,000 / [Animation, Comedy, Drama, Fantasy]
Ryan Coogler (USA) / Black Panther (FEBRUARY 16, 2018) / $700,059,566 / [Action, Adventure, Superhero]
Anthony Russo and Joe Russo (USA) / Avengers: Infinity War (APRIL 27, 2018) / $678,815,482 / [Action, Adventure, Drama, Epic, Fantasy, Sci-Fi, Superhero]

License

MIT License.