Update exam clash detection in the Module search page

[uyencfi]

Context

Fix #3663

Implementation

Given timetable = [m1, m2, m3] which is the list of modules currently added in the timetable.
To detect whether a module X clashes with timetable:

• current behaviour is to compare X.start == m.start (exact match) for each m in timetable
• updated behaviour:
  • assert that (X.start < m.end) && (X.end > m.start) for each m in timetable. This is the condition for 2 intervals to overlap, see https://stackoverflow.com/questions/325933/determine-whether-two-date-ranges-overlap

❗ However, the module struct only stores examStartTime (ISO timestamp, assuming UTC(?)) and examDuration (in minutes). Thus to get the endTime:

• m.end can be easily obtained since m is a known value from the timetable.
• X.end is more difficult since we need to compute it on-the-fly inside the Elasticsearch query.

The new Elasticsearch query I have looks like

query: {
  bool: { must_not: [  // this document must not match *any* condition in the list, i.e. not overlap any exam
    
    bool: { must: {               // condition 1: true if this exam overlaps m1
      range: { document.start < m1.start + m1.duration },
      script: { document.start + document.duration > m1.start }   // <-- can't use 'range' b/c ES doesn't allow accessing 2 fields at the same time
    }},

    bool: { must: { ... } },     // condition 2: true if this exam overlaps m2
    bool: { must: { ... } },     // condition 3: true if this exam overlaps m3
  ]},
},

TBD

• Is examStartTime really stored in Elasticsearch in UTC or UTC+08?
• (Unit?) Testing and how to verify correctness

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
nusmods-export	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Nov 18, 2024 10:49am
nusmods-website	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Nov 18, 2024 10:49am

[vercel]

@uyencfi is attempting to deploy a commit to a Personal Account owned by @nusmodifications on Vercel.

@nusmodifications first needs to authorize it.

[kokrui]

Thanks for the PR! We'll test out scraper changes over the next few days and get back to you ☺️

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 12 lines in your changes missing coverage. Please review.

Project coverage is 54.52%. Comparing base (988c6fd) to head (954ed9d).
Report is 13 commits behind head on master.

Files with missing lines	Patch %	Lines
website/src/views/modules/ModuleFinderSidebar.tsx	0.00%	12 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #3678   +/-   ##
=======================================
  Coverage   54.52%   54.52%           
=======================================
  Files         274      274           
  Lines        6076     6085    +9     
  Branches     1455     1453    -2     
=======================================
+ Hits         3313     3318    +5     
- Misses       2763     2767    +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[kokrui]

Quick live test case:

1. Add CM2112 to your Semester 2 timetable
2. In course search, enable "No Exam Clash (Sem 2)"
3. Search for HSI2001
4. You should not see HSI2001

[kokrui]

Hey, it looks like the exam for CM2112 got changed to be 2.30pm instead of 1pm, which makes it match HSI2001, so it's no longer that good of a candidate for a test case. A valid test case still exists: Sem 1, SH5110 vs FE5222

That said, let me take a look and test some things with your query and get back to you -- it looks pretty good right now, but for one, I think the query is inverted! (i.e. "No Exam Clash" gives us courses which do have an exam clash)