rt-scraper
v1.0.5
Published
This package, without an API key, will scrape the front page of RottenTomatoes.com for the movies opening this week, top box office, and coming soon.
Downloads
17
Maintainers
Readme
About
This package scrapes the front page of RottenTomatoes.com for data.
This is done without using an API key.
The information returned includes movies opening this week, top box office, and coming soon.
For each movie some information will be returned:
- Title
- Rotten Tomato meter rating (i.e. 93%, 45%, etc), or will show no score yet if not available
- Release date, for the opening this week and coming soon datasets
- US domestic gross, for the top box office dataset
Installation
npm install rt-scraper
Usage
There is only one method to this package: getRottenTomatoesScraperData(callback)
This method requires the use of a callback, which can be specified or anonymous. The callback has two parameters in the form of function(error, data)
.
If there was an error, like failed HTML request or bad parsing, error will be true and data will be null. Otherwise, error will be false and data will be a javascript object.
The data object has three keys: openingThisWeek, boxOffice, comingSoon. Each of these keys has for a value an array of movie objects.
Each movie object contains a title and meter key, where title is the name of the movie and meter is it's Rotten Tomato meter score. Note that if no score is available this value can be equal to "No Score Yet" rather than "93%" or "45%".
The openingThisWeek and comingSoon* keys will have movies with a date which references the movie's release date. This value will be in the form of "Mar 3" or "Sep 12".
The boxOffice key will have movies with a gross which references the US domestic gross of that movie. This value will be in the form of "$11.2M" or "$120.7M"
Examine the usage and data return examples for best results.
Usage Example
// Usage example of rt-scraper package
// See https://www.npmjs.com/package/rt-scraper
var rtscraper = require('rt-scraper');
console.log('Usage example of the rt-scraper package.')
rtscraper.getRottenTomatoesScraperData( function(error, data) {
if (!error) {
console.log(JSON.stringify(data, null, 2));
}
else {
console.log('Some error occured.');
}
});
Data Return Example
{
"openingThisWeek": [
{
"meter": "100%",
"title": "Get Out",
"date": "Feb 24"
},
{
"meter": "30%",
"title": "Collide",
"date": "Feb 24"
},
{
"meter": "44%",
"title": "Rock Dog",
"date": "Feb 24"
},
{
"meter": "85%",
"title": "The Girl With All the Gifts",
"date": "Feb 24"
},
{
"meter": "100%",
"title": "My Life as a Zucchini (Ma vie de courgette)",
"date": "Feb 24"
}
],
"boxOffice": [
{
"meter": "91%",
"title": "The Lego Batman Movie",
"gross": "$32.7M"
},
{
"meter": "8%",
"title": "Fifty Shades Darker",
"gross": "$20.3M"
},
{
"meter": "35%",
"title": "The Great Wall",
"gross": "$18.5M"
},
{
"meter": "90%",
"title": "John Wick: Chapter 2",
"gross": "$16.3M"
},
{
"meter": "30%",
"title": "Fist Fight",
"gross": "$12.3M"
},
{
"meter": "92%",
"title": "Hidden Figures",
"gross": "$7.3M"
},
{
"meter": "75%",
"title": "Split",
"gross": "$7.2M"
},
{
"meter": "34%",
"title": "A Dog's Purpose",
"gross": "$5.8M"
},
{
"meter": "93%",
"title": "La La Land",
"gross": "$4.6M"
},
{
"meter": "39%",
"title": "A Cure For Wellness",
"gross": "$4.4M"
}
],
"comingSoon": [
{
"meter": "96%",
"title": "Logan",
"date": "Mar 3"
},
{
"meter": "No Score Yet",
"title": "Table 19",
"date": "Mar 3"
},
{
"meter": "No Score Yet",
"title": "The Shack",
"date": "Mar 3"
},
{
"meter": "64%",
"title": "Before I Fall",
"date": "Mar 3"
},
{
"meter": "74%",
"title": "Lovesong",
"date": "Mar 3"
}
]
}