Hash-analyze

Hash-analyze is a browser-based tool for reviewing fleet camera installation records. Upload job logs or spreadsheets, get an instant dashboard of hours, audit outcomes, and technician workload, then ask follow-up questions in a chat panel powered by Google Gemini.

Everything runs in the browser. Your files are not uploaded to a backend server.

What it does

1. Import data — paste text, or upload CSV, Excel (.xlsx / .xls), PDF, or plain-text logs.
2. Parse automatically — the app detects tabular data (CSV/Excel) or structured job-sheet blocks and fills the dashboard.
3. Review at a glance — see total jobs, hours logged, pass/fail audit counts, a technician workload chart, and a sortable table of records.
4. Ask questions — use the chat tab or one-click audit shortcuts to analyze the imported data with Gemini 2.5 Flash.

If no API key is provided, the app runs in demo mode with sample responses so you can explore the UI without calling Gemini.

Supported file formats

Format	Extensions	Notes
CSV / TSV	.csv, .txt	Best with a header row containing columns like Date, Technician, Vehicle ID
Excel	.xlsx, .xls	Converted client-side via SheetJS
PDF	.pdf	Text is extracted with PDF.js. Works for text-based PDFs only — scanned documents need OCR and are not supported
Job sheets	.txt, .log	Structured blocks using the --- JOB SHEET #001 --- template (see below)
JSON / logs	.json, .log	Treated as plain text

The parser skips title rows, sheet markers, and PDF page headers, and handles quoted CSV fields.

Quick start

Prerequisites

• Node.js 18 or later
• npm

Run locally

git clone https://github.com/hashnori/hashAnalyze.git
cd hashAnalyze
npm install
npm run dev

Open the URL shown in the terminal (usually http://localhost:5173).

Build for production

npm run build
npm run preview   # optional — serve the production build locally

The build output goes to the dist/ folder.

Gemini API key

To enable live AI analysis:

1. Go to Google AI Studio.
2. Create an API key.
3. Paste it into the Gemini API Key field in the app header.

The key is stored in your browser's localStorage only. It is sent directly from your browser to Google's Gemini API when you send a chat message — it never passes through a Hash-analyze server.

Data formats

Tabular (CSV / Excel)

Include a header row with recognizable column names. Examples that work well:

Date,Technician,Vehicle ID,Installation Type,Time Taken,Status,Audit Status
2026-06-15,Alex Mercer,VT-102,3-Camera Fleet,2.5,Completed,Passed
2026-06-16,Sarah Connor,VT-409,4-Camera System,4.0,Completed with Warnings,Needs Verification

Recognized header keywords:

• Date — date
• Technician — technician, tech, installer
• Vehicle — vehicle id, vehicle, car, truck
• Install type — installation, type
• Equipment — equipment, parts, tools
• Duration — time taken, hours, duration
• Status — status, state (excluding audit columns)
• Audit — audit, inspect, qa

A title row above the headers (e.g. Fleet Installation Report) is fine — the parser will find the header row automatically.

Job sheet template

For free-form logs, use repeated blocks like this:

--- JOB SHEET #001 ---
Date: 2026-06-15
Technician: Alex Mercer
Vehicle ID: VT-102 (Cargo Van)
Installation Type: 3-Camera Fleet Security
Equipment Used: Hash-Cam Pro HD, MDVR-Lite
Time Taken: 2.5 hours
Status: Fully Completed
Audit Status: Passed
Notes: Front camera aligned and secured. All feeds tested.

Load the built-in sample from the Load Sample Sheets button to see a working example.

Express audit shortcuts

The left panel includes three preset prompts:

• Quality Assurance Audit — find failed audits, wiring issues, and incomplete jobs
• Technician Performance Index — compare hours, job counts, and error rates
• Hardware & Cable Consumption — summarize parts used and inventory gaps

Each shortcut opens the chat tab and sends a structured prompt to Gemini using your imported file as context.

Project structure

hashAnalyze/
├── index.tsx          # Main React app (UI, parsing, Gemini integration)
├── main.tsx           # App entry point
├── index.html         # Vite HTML shell
├── index.css          # Tailwind styles
├── vite.config.ts     # Vite configuration
├── public/            # Static assets (favicon)
└── package.json

Tech stack

• React 19 + Vite 6 — frontend and bundler
• Tailwind CSS — styling
• SheetJS (xlsx) — Excel parsing (loaded from CDN)
• PDF.js — PDF text extraction
• Gemini 2.5 Flash — AI analysis (gemini-2.5-flash)

Deployment

Hash-analyze is a static Vite app and deploys to any static host (Vercel, Netlify, Cloudflare Pages, etc.).

Typical settings:

Setting	Value
Build command	npm run build
Output directory	dist
Framework preset	Vite

No server-side environment variables are required. Users enter their own Gemini API key in the browser.

Limitations

• PDFs must contain selectable text. Image-only or scanned PDFs will not parse.
• Parsing is heuristic — unusual column names or layouts may need manual adjustment or pasting into the job-sheet template.
• Demo mode returns canned summaries, not real analysis of your data.
• API usage counts against your Google AI Studio quota when a key is provided.

License

See the repository for license details.