Tencent Realtime Quotes Design

Date: 2026-07-08

Goal

Add Tencent A-share quote integration so the GUI holdings table displays realtime prices from http://qt.gtimg.cn/q=<symbol>.

The first implementation is manual refresh only. It does not auto-refresh, generate buy/sell grid orders, or overwrite user-entered manual prices.

Confirmed Decision

Use Tencent's quote endpoint directly. Example:

http://qt.gtimg.cn/q=sz000001

The response is GBK/GB18030 encoded and shaped like:

v_sz000001="51~平安银行~000001~10.60~10.47~10.44~...";

The parser will use these fields:

1: name
2: code
3: current price
4: previous close
5: open
30: quote time
31: price change
32: change percent
33: high
34: low

Design

Add a market package with a Tencent quote client. The client converts local instrument codes into Tencent symbols:

Already prefixed sh or sz: keep as-is.
Codes starting with 6, 5, or 9: use sh.
Codes starting with 0, 1, 2, or 3: use sz.

TradingService keeps an in-memory quote cache. When the user clicks refresh in the GUI, the service fetches quotes for all active instruments and stores them in memory.

Position calculations will prefer prices in this order:

Tencent realtime quote
Manual price
Last trade price
Empty value

The holdings table and account summary will immediately reflect quote prices after refresh. Quote prices are not persisted to SQLite in this step, so stale prices do not silently survive app restarts.

Error Handling

If one symbol fails, the refresh reports the error and keeps existing data unchanged.
Empty or malformed Tencent responses raise a quote-specific error.
A zero or missing current price is ignored for that symbol.
Network timeouts use a short timeout so the GUI does not hang for long.

Testing

Unit-test Tencent response parsing with a fixture response.
Unit-test symbol inference.
Unit-test position calculations with a realtime quote overriding manual price.
Unit-test service quote refresh using a fake quote provider.
Keep existing GUI smoke tests passing.

2.1 KiB Raw Blame History