PanSou: A High-Performance Cloud Drive Resource Search API 84

Finding study materials or movies is easy with cloud drives, but you often have to search back and forth between different cloud drives, and some resources are no longer available.

The open-source project PanSou solves these problems by integrating multiple search sources into a single API, allowing you to search across all major cloud drive platforms in one go.

PanSou is a high-performance cloud drive resource search API service that supports TG channel searches and custom plugin searches. The system is designed with performance and scalability at its core, enabling concurrent multi-channel and multi-plugin searches, with intelligent result ranking and categorization by cloud drive type.

PanSou Features

✅High-performance search: Execute multiple TG channel and asynchronous plugin searches simultaneously for faster search speeds.

✅Cloud drive type classification: Automatically identify various cloud drive links and display them categorized by type.

✅Intelligent ranking: Combines plugin level, recency, and priority keywords for sorting results.

✅Asynchronous plugin system: Expand search sources through plugins, supporting the "respond ASAP, keep processing" asynchronous search mode.

✅Two-level cache: Uses sharded memory plus sharded disk cache, significantly improving repeat query speed and concurrency performance.

Supported Cloud Drive Types

Includes Baidu Netdisk, Alibaba Cloud Drive, Quark Cloud Drive, Tianyi Cloud Drive, UC Netdisk, Mobile Cloud Drive, 115 Netdisk, PikPak, Xunlei Cloud Drive, 123 Cloud Drive, magnet links, eDonkey links, and more.

PanSou Technical Architecture

✅Tools layer (util/): Includes two-level cache system (cache/), worker pool (pool/), HTTP client, parsing utilities, etc.

✅Plugin system layer (plugin/): Defines plugin interfaces, asynchronous plugin base class, and various concrete plugins (such as jikepan, pan666, hunhepan, etc.).

✅Service layer (service/): Includes cache integration, search services, etc.

✅Model layer (model/): Defines data structures for plugin results, requests, responses, etc.

✅API layer (api/): Includes handlers, middleware, routes, etc.

✅Cache system: Uses two-level cache, including sharded memory cache and sharded disk cache, with GOB serialization. Some plugins also have their own caching mechanisms (such as search result cache, detail page cache, etc.).

✅Concurrency control: Implements worker pool management through util/pool to control concurrent tasks.

Deployment and Usage

✅Supports Docker deployment (integrated front-end and back-end version, and pure back-end API version), and can be quickly started with Docker commands or Docker Compose.

✅Can also be installed from source, requires Go 1.18+ environment, with optional SOCKS5 proxy (used to access Telegram sites).

Plugin System

PanSou can expand search sources through plugins, with a plugin development guide available. There are already multiple existing plugins (such as huban, pan666, hdr4k, etc.), and plugins can implement their own search logic and caching strategies.

In addition, the author has shared a detailed system development and design document for PanSou, which interested developers can read to understand the principles.