Drop any large CSV into your browser. Run complex SQL with JOINs, Window Functions, and CTEs.
Instant pivot charts. 100% Client-Side. Zero servers. Zero cloud fees. Zero data leaks.
🚀 Explore Live Workbench » · ⭐️ Star on GitHub · View Benchmarks · Architecture · Quickstart
Every day, data analysts, engineers, and researchers face a dilemma when analyzing confidential CSVs:
- Excel/Sheets: Freezes or truncates after a few hundred thousand rows; no relational SQL.
- Cloud Data Warehouses (Snowflake, BigQuery): Requires provisioning clusters, creating IAM roles, managing storage buckets, and incurring billing costs.
- Data Privacy Risks: Uploading internal corporate telemetry, financial sheets, or PII to third-party SaaS converters violates compliance (GDPR/SOC-2/HIPAA).
QueryDrop solves this entirely in your browser. By cross-compiling the real SQLite C engine to WebAssembly and offloading data parsing to a background Web Worker thread, QueryDrop turns your browser into a local, isolated SQL warehouse capable of querying hundreds of thousands of rows with zero latency.
| Feature | Excel / Google Sheets | Cloud SaaS / Convert Tools | ⚡ QueryDrop |
|---|---|---|---|
| Row Handling | Sluggish > 100k rows | Depends on tier / paywall | 250,000+ rows in seconds |
| Data Privacy | Local or Cloud Synced | ❌ Uploaded to remote servers | 🔒 100% In-Memory (Zero Network Egress) |
| Full SQL Support | ❌ Formulas only | ✅ Full SQLite 3 (JOINs, CTEs, Window Fns) | |
| Off-Thread Processing | ❌ UI locks up | N/A | ✅ Dedicated Web Worker (60 FPS UI) |
| Visualizations | Manual chart wizard | Paid dashboards | ✅ 1-Click Apache ECharts + PNG Export |
| Database Export | CSV only | Proprietary | ✅ Native .sqlite + sanitized CSV |
| Infrastructure Cost | Subscription ($) | Per-query / Monthly ($$) | ⚡ 100% Free ($0 / Static SPA) |
- 🛡️ Air-Gapped Privacy: Files never touch an API. Everything lives inside isolated browser memory buffers using the HTML5 File & Stream APIs.
- ⚙️ Native SQLite C-Engine via WebAssembly: Powered by
sql.js(Emscripten-compiled SQLite 3.x). You get the exact same relational consistency, aggregate functions, and deterministic planner as native SQLite. - 🧵 Off-Thread Web Worker Orchestration: CSV chunking, type inference, transaction batching, and query execution run on a separate OS thread via
sql.worker.ts. Your scroll and UI never stutter. - 📊 Interactive Apache ECharts: Seamlessly switch between Bar, Line, Pie, and Scatter plots. Auto-detects categorical axes vs numerical measures, with instant high-res canvas exports.
- 🔍 Smart Schema & Type Coercion: Automatically analyzes headers and values to infer
INTEGER,REAL, andTEXT, with fallback type sanitization for complex nested JSON and ISO dates. - 🗃️ Sample Playgrounds: Instant preloaded datasets (Global E-Commerce Orders, Tech Hub Quality of Life) so anyone can evaluate capabilities immediately with zero setup.
flowchart TD
subgraph Browser_Main_Thread ["Browser UI (Main Thread)"]
UI["React 19 Dashboard"]
Dropzone["Drag & Drop File Zone"]
SQLEditor["Monaco/Code Query Console"]
Visualizer["Apache ECharts Engine"]
ResultsGrid["Virtual Paginated Data Grid"]
end
subgraph Background_Worker ["Web Worker (Background Thread)"]
Parser["Streaming PapaParse (RFC-4180)"]
TypeInfer["Schema & Type Inference"]
WasmEngine["SQLite 3 WebAssembly Engine"]
MemoryStore[("In-Memory SQLite DB")]
end
Dropzone -->|"File Blob / Stream"| Parser
Parser -->|"Batch Tuples"| TypeInfer
TypeInfer -->|"BEGIN TRANSACTION\nINSERT VALUES...\nCOMMIT"| MemoryStore
MemoryStore --- WasmEngine
SQLEditor -->|"postMessage('RUN_QUERY', sql)"| WasmEngine
WasmEngine -->|"Structured Row Tuples"| Background_Worker
Background_Worker -->|"postMessage('QUERY_RESULT')"| UI
UI --> ResultsGrid
UI --> Visualizer
Tested locally on Google Chrome (V8) with the enterprise telemetry stress dataset:
| Metric | Result |
|---|---|
| Dataset Size | 146.82 MB (ultra_complex_enterprise_telemetry_120mb.csv) |
| Total Rows & Columns | 250,000 records × 22 columns |
| Parse & Type Inference | ~5.0 seconds |
| SQLite Wasm Table Load | ~9.9 seconds (batched in single transaction) |
| Aggregate Query Latency | 12 ms (SELECT COUNT(*), AVG(duration_ms), MAX(bandwidth_bytes)) |
| Group-By Aggregation | 28 ms (SELECT geo_city, count(*), sum(bandwidth_bytes) GROUP BY 1) |
| Memory Footprint | Stays within browser WebAssembly 32-bit linear memory limit (~512MB heap) |
Once your CSV is dropped, execute standard SQLite queries directly:
-- 1. Identify high-risk error spikes by city and status
SELECT
geo_city,
http_status,
COUNT(*) AS total_errors,
ROUND(AVG(duration_ms), 2) AS avg_latency_ms
FROM telemetry
WHERE http_status >= 500
GROUP BY geo_city, http_status
ORDER BY total_errors DESC
LIMIT 10;-- 2. Window function: Top 3 longest API routes per city
WITH RankedRoutes AS (
SELECT
geo_city,
endpoint_route,
duration_ms,
DENSE_RANK() OVER (PARTITION BY geo_city ORDER BY duration_ms DESC) as rank
FROM telemetry
)
SELECT geo_city, endpoint_route, duration_ms
FROM RankedRoutes
WHERE rank <= 3;# 1. Clone repository
git clone https://github.com/Neeraj3102/querydrop.git
cd querydrop
# 2. Install dependencies
npm install
# 3. Start local development server
npm run dev
# 4. Run lint checks
npm run lint
# 5. Build optimized production bundle
npm run buildIf QueryDrop saved you time or cloud fees, please consider giving it a ⭐️ Star on GitHub — it helps more developers discover client-side SQL analytics!
Issues, PRs, and feature discussions are welcomed on GitHub Issues.
Distributed under the MIT License. Created with precision by Neeraj.