nyuclassesfrom.xyz

Make NYU course search fast.

If we think Harvard's online catalog is slow... we don't even have a graphQL endpoint at NYU XD.

All jokes aside, our course search is huge, and it's very annoying to wait for a online catalog search. Furthermore, NYU's status as a global campus with over 40+ locations globally makes it harder to manage from a catagorical point of view, compared to other course searches.Different schools, different classes, and different timezones all blend together, which is why the current website is a slow process (Not to mention highly cryptic and bundled weirdly).

As such, this project is an attempt to gather fast results for course search, as inspired from classes.wtf.

Please keep in mind this is a hobby project between friends, and in no ways can we gaurantee that everything will be working 100% of the time. We do not hold any responsibility towards registering for courses, or any implications to using this service.

With all this being said, it is our aspiration to help fellow NYU students easily find classes to their upcoming semesters, and we hope this helps!

Data Fetching

• catch all the school codes and subject codes

  • ex: SHU = shanghai CSCI = computer science

• go through and fetch each course based off subject & school

  • https://schedge.a1liu.com/2022/fa/SHU/CSCI

• search for course based on full=true, query=courseName, school, subject https://schedge.a1liu.com/2022/fa/search?full=true&query={courseName}&school={schoolCode}&subject={subjectCode}

• stitch and format the json

• school codes (fetch) + subject codes (fetch) => list of combos (SHU/CSCI) => courses (fetch) => list of name, courseId, school, subject => search (fetches) => final list of everything per each course

Running the Frontend

cd frontend
npm install
npm run dev

TODOs

• [] Add Wrapping Formatting for frontend
  • Listen for the search query -> send to request on Upstash
  • Return results (via readData)

Future Work

To keep the "class status" field up to date with what's displayed on Albert, we need to synchronize the cache and the data on Albert. The consistency model will be eventual consistency, and we use the following mechanism to uphold this consistency model

• Each course record has a TTL
• When a user queries a certain course, we check the TTL value
  • If the record is not expired, we return all the data as stored
  • If the record is expired (given the TTL), we return all but the class_status field. Meanwhile, we perform another scrap on Albert. Once the data is acquired, we update the data on the cache, and push it to the client.

For now we assume that data fetched from the schedge api is up to date. To achieve the above consistency model we need a server that records the expiry dates of every course record. Everytime a user requests a course record, the server checks the expiry date of the record then return the user the course detail from the databse.