Sentiment Analysis Web App

A multi-faceted sentiment analysis project with:

• Web app (Flask + HTML/CSS)
• JSON API (for programmatic use and the browser extension)
• Chrome extension (Manifest V3) to analyze selected or pasted text

Core analysis uses VADER as the primary analyzer (great for short/social text), with TextBlob as a fallback. The API returns the final label plus polarity/subjectivity.

Requirements

• Python 3.9+
• Chrome (for the extension)

Setup

# From the project directory
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Run the Flask app

python app.py
# App will run on http://127.0.0.1:5000/

Web App

Open http://127.0.0.1:5000/ and paste text to analyze.

API

POST /api/sentiment
Content-Type: application/json
{
  "text": "I love this!"
}

Response:

{
  "sentiment": "Positive",
  "polarity": 0.625,
  "subjectivity": 0.6
}

CORS headers are enabled for convenience so the Chrome extension can call the API.

Chrome Extension (MV3)

You can use the extension in two modes:

• Development (local API): run the Flask app locally, the extension will fall back to http://127.0.0.1:5000 if production is unreachable.
• Production (Vercel API): no local server required; the extension tries your deployed API first.

Setup:

1. In Chrome, go to chrome://extensions
2. Enable "Developer mode" (top right)
3. Click "Load unpacked" and select the extension/ folder
4. Pin the extension. Use it in any of these ways:
   • Select text on a page and click the floating blue button that appears near the selection. The popup opens (or a tooltip appears inline) with the analysis.
   • Or click the extension icon to open the popup, then:
     • Use "Analyze Selection" to fetch selected text from the current tab
     • Or paste/enter text and click "Analyze Input"

Notes:

• The extension is configured to call production first: https://sentanalysis.vercel.app/api/sentiment, then fall back to local: http://127.0.0.1:5000/api/sentiment.
• The in-page floating button is injected by content.js; if the popup cannot be opened (e.g., on Edge), it shows an inline tooltip result instead.
• Internal browser pages (chrome://, edge://, about:, extension pages) are restricted. Use the extension on normal sites.
• After changing any extension files, reload it from chrome://extensions (or edge://extensions).

Notes

• The project uses VADER primarily, with TextBlob fallback, implemented in sentiment/core.py.
• For production, consider stricter CORS and authentication.

Deploy to Vercel (Serverless)

This project is configured to run on Vercel's Python Runtime as a serverless WSGI app.

What was added:

• vercel.json routes all incoming requests to api/index.py.
• api/index.py exports the Flask app, which Vercel detects as a WSGI application.

Steps:

1. Ensure you have a Git repo (GitHub/GitLab/Bitbucket) for this folder.
2. Push your code and import the repo in Vercel (https://vercel.com/new).
3. Vercel detects Python, installs requirements.txt, and deploys.
4. After deploy, your app will be available at https://<your-project>.vercel.app/.

Verify after deploy:

• Open your site root, e.g. https://sentanalysis.vercel.app/.
• Test the API directly:

  Invoke-WebRequest -Uri https://sentanalysis.vercel.app/api/sentiment -Method POST `
    -Body (@{text="I love this!"} | ConvertTo-Json) `
    -ContentType "application/json"

Local test with Vercel CLI (optional):

npm i -g vercel
vercel dev
# Visit http://localhost:3000

Extension after deploy:

• Ensure extension/manifest.json host_permissions includes your production URL, e.g. "https://<your-project>.vercel.app/*" (already includes https://sentanalysis.vercel.app/*).
• The extension's popup.js already tries your production URL first. If you fork/rename, update the domain there.

Microsoft Edge:

• Load unpacked via edge://extensions and enable Developer mode.
• Allow the extension on all sites and (optionally) in InPrivate.
• If opening the popup fails, the floating button falls back to inline analysis tooltip.