v0.2.0
Home GitHub

Architecture

Last updated

This page shows the layers and module dependencies of go-rss-reader in one overview diagram; per-module structure, data flow, and the state machine are in doc/architecture.md.

Overview

graph TB
    Main[cmd/cli main] --> App[internal/app TUI Layer]
    App --> Collector[internal/util Collector]
    App --> Extractor[internal/util Extractor]
    App --> API[internal/api LLM Client]
    App --> DB[internal/database SQLite]
    Collector --> DB
    Collector --> Feeds[(RSS Feeds)]
    Extractor --> Web[(News Pages)]
    API --> OpenAI[(OpenAI Chat Completions)]
    Collector -.-> Model[internal/model]
    Extractor -.-> Model
    DB -.-> Model

Layers

Layer Package Responsibility
Entry point cmd/cli Calls app.New() and Run(), prints the error if launch fails
Application internal/app tview layout, hotkeys, command parsing, refresh scheduling, and coordination of the modules below
Collection internal/util (Collector) Feed add/remove, RSS fetch and parse, deduplication, time-window filtering
Extraction internal/util (Extractor) Fetches the original page and extracts title, author, body, and word count
LLM internal/api Streams OpenAI Chat Completions
Storage internal/database SQLite path resolution, table creation, article/feed/key-value access
Data model internal/model News, NewsContent, RSS/Channel/Item

Every package lives under internal/, so none of them can be imported from outside the module.

Cross-Cutting Principles

Principle Implementation
UI updates only on the tview main thread Background goroutines always update the screen through app.QueueUpdateDraw()
Cache first While the list is empty, database.Get(72) fills it before the network fetch
Failures do not stop the run A failing feed or a failing extraction is logged and skipped without affecting other items
Single data file Articles, feeds, the API key, and the digest all live in one SQLite file

Further Reading

中文