UniLab coursework
Assignment evidence and the retained application baseline for 1305493 Software Engineering Case Studies, Mae Fah Luang University.
Open UniLab · Product source and downloads · Coursework preview
Repository separation
This repository is the assignment and coursework record, with the v0.2.0 application baseline retained for traceability. The independently maintained product source, Android APK, desktop installers and all releases from v0.3.0 live in Thiha-Lynn/unilab-releases.
- Assignment / evidence: this repository and its
.docs/ directory.
- Product source / release builds:
unilab-releases/app, desktop and mobile; its build pipeline does not check out this assignment repository.
- Canonical product: unilab.ztvmm.live.
- Coursework preview: GitHub Pages, built from this repository's baseline.
The product packages use their documented dependency licenses (including AGPL background removal); original UniLab source retains its MIT notices. Product releases do not establish lecturer approval or a passed Week 5 gate. Those outcomes remain pending the presentation and must be recorded from actual feedback.
Coursework — start here if you are grading this
Everything for the course lives in .docs/. The one-screen map is
.docs/README.md; the traceability chain is
TRACEABILITY.md.
Evidence position, stated plainly: the gate also requires ≥ 5 real users interviewed.
15 were interviewed between 22 Aug and 8 Sep 2026 — MFU undergraduates from 15 programmes,
three per team member, recorded as role, pain and quote only (LR7) in
results.md. Twelve pains (P1–P12) came
out of them, and every requirement in the spec now traces to one. Two shipped features — OCR and
Workflows — traced to hypotheses no participant raised, so they are parked for the gate
(spec §2.1, Q5, Q6) rather than papered over: ids kept, code untouched, decision in W6. The interview instrument is in
all questions, EN/TH/MY.
Application baseline
The retained v0.2.0 source provides 59 tools across PDF, image, video, audio, text, study and utility categories. They share one workflow: select a file → process on this device → download.
Files are processed locally. Optional OCR languages and background-removal models need an initial download. Device memory, supported codecs and file-size limits affect what can run. Compression and lecture HTML conversion may rasterize output; inspect readability before submitting a document.
The current web app and native packages are maintained independently in the product repository. Its installation guide is the source of truth for current versions, offline capabilities and platform testing limitations.
Run it locally
git clone https://github.com/Thiha-Lynn/unilab.git
cd unilab
npm ci
npm run dev
Vite + vanilla JavaScript — no framework to learn. Processing is done by
pdf-lib, pdf.js,
mediabunny (WebCodecs video/audio),
tesseract.js (OCR),
mammoth (.docx),
browser-image-compression,
cropperjs,
heic2any,
gifenc,
jszip and
qrcode — all in-browser.
Two tools download a model on first use — OCR fetches its language data, and Remove
Background fetches an ONNX model. Both are gated behind an explicit button that states
the size, and in both cases it is the engine that is downloaded: your file stays here.
Development and contribution
Use this repository for course evidence and baseline corrections. Product changes belong in unilab-releases. Read CONTRIBUTING.md before opening a pull request. Main is protected and requires the test/build check.
Current collaborators with write access: 6631503097, 6631503088, zawwinhuttt, and myo-zin-thant18. Repository owner: Thiha-Lynn.
Repository access and historical contribution credit are different. The owner confirmed that nine early commits used the wrong Git identity. .mailmap corrects their mapped attribution while retaining commit IDs and history. GitHub's contributor display may still show the original identity.
Validation and publication
Run npm test and npm run build before proposing source changes. CI verifies changes before publishing the coursework Pages preview. Production is deployed from the independent product repository. Operations notes explain the separation.
Week 6 scope and evidence distinguishes shipped software from course approval and pending physical-device evidence. A successful build is not evidence of instructor approval or universal device compatibility.
AI use and licensing
Prototype scaffolding and development used AI assistance. Test results and documented limitations should be read alongside implementation claims; physical-device coverage remains incomplete.
Original UniLab source retains MIT notices. Dependencies have their own licenses. The combined product distribution is AGPL-3.0-only; see the product license and third-party notices.
