Quarterly Profit Chart（季度股价 vs 单季净利润对比图）

Overview

Given a stock (by name or code), produce a self-contained HTML report containing:
- A dual-axis chart: blue bars = single-quarter net profit attributable to parent (left axis,
in 亿元), red line = post-adjustment (后复权) closing price (right axis, in 元),
plus a sub-chart green line = single-quarter net profit YoY growth (%).
- A data-driven Chinese analysis section (key metrics + trend interpretation).

The chart is pure inline SVG (no CDN dependency), so it renders offline and is easy to save.

When to Use

• "画一下海康威视的季度股价和净利润图"
• "对比比亚迪的股价和利润趋势"
• "XX 季度业绩走势 / 利润和股价背离分析"
• Any request to chart a stock's quarterly price against its earnings.

Dependencies

• westock-data skill — provides the CLI (index.js) used to fetch K-line and financial data.
  Load it first via the westock-data skill to obtain the CLI path.
• node — to run the westock-data CLI.
• This skill's bundled script: scripts/gen_chart.py.

Workflow

Step 1 — Resolve the stock code

The westock-data CLI does not accept Chinese names (it returns empty). If the user gives a
name (e.g. "海康威视"), resolve it to a standard code first:

• Use WebSearch: search "名称 股票代码" and pick the standard code
  (formats: sz002415, sh600519, bj430047, or HK hk00700, US usAAPL).
• If the user already supplied a code, use it directly.

Step 2 — Load westock-data and locate its CLI

Invoke the westock-data skill to load its instructions, which expose the path to
.../westock-data/scripts/index.js. Pass that path to the script via --westock.
(The script also auto-discovers the path under ~/.workbuddy and /Applications/WorkBuddy.app,
so --westock is optional if the default layout matches.)

Step 3 — Run the generator

python3 SKILL_DIR/scripts/gen_chart.py CODE \
    --name "股票名称" \
    --westock "PATH/to/westock-data/scripts/index.js" \
    --out "OUTPUT.html"

The script prints a progress log and ends with a SUMMARY_JSON:... line containing the key
metrics (profit peak, price peak, latest quarter) — use it for your narration.

Step 4 — Present the result

Use present_files on the generated HTML so the user can view (and save) it. Then, based on the
SUMMARY_JSON and the script's auto-analysis, add a short contextual read-out (e.g. note whether
price and profit peaks are synchronized, or whether profit is at a high while valuation has
pulled back — a "profit up, price flat" divergence worth flagging in a Lynch-style lens).

Implementation Notes (non-obvious gotchas)

• Single-quarter profit: the finance API does not expose a single-quarter field. The
  script computes it as cumulative(this quarter) − cumulative(prior quarter in same year)
  (Q1 = cumulative directly). Finance returns cumulative net profit in NPParentCompanyOwners
  (unit: 元 → ÷1e8 for 亿元).
• Standard quarter-ends only: filter out non-standard report dates (e.g. re-statement dates
  like 2016-07-31) to avoid polluting the cumulative-difference calculation.
• Empty-result retry: the westock-data CLI intermittently returns empty [] / null on
  frequent/consecutive calls (likely internal rate-limiting). gen_chart.py already retries with
  exponential backoff; if a call still fails, simply re-run.
• History window: the finance endpoint typically returns ~42 quarters (from about 2016Q1).
  The X-axis is built dynamically from whatever quarters are available, so older/newer listings
  are handled automatically.
• Post-adjustment price: K-line uses --fq hfq (后复权) so splits/dividends don't distort
  the long-term trend.

Resources

scripts/

• gen_chart.py — fetches seasonal post-adjustment K-line + quarterly profit, computes
  single-quarter profit and YoY, and renders the self-contained HTML report.
  Args: CODE (required), --westock PATH, --name NAME, --out FILE.