har-derived-api-client

1#!/usr/bin/env python32"""Record a HAR file while driving a website with Playwright.3 4Usage:5  python3 har_capture.py <url> <output.har> [--wait SECONDS] \6      [--action "fill:SELECTOR:TEXT"] [--action "press:SELECTOR:KEY"] \7      [--action "click:SELECTOR"] [--action "goto:URL"] [--action "sleep:SECONDS"]8 9Actions run in order after page load. The HAR embeds request/response bodies10(record_har_content='embed') so derived clients can see payload shapes.11 12NOTE: a failing action raises before the HAR is flushed -- you get no file.13Fix the selector (try --headed to watch) and rerun.14"""15import argparse16import sys17import time18 19from playwright.sync_api import sync_playwright20 21 22def run_action(page, spec: str) -> None:23    parts = spec.split(":", 2)24    kind = parts[0]25    if kind == "fill":26        page.fill(parts[1], parts[2])27    elif kind == "press":28        page.press(parts[1], parts[2])29    elif kind == "click":30        page.click(parts[1])31    elif kind == "goto":32        page.goto(parts[1] + (":" + parts[2] if len(parts) > 2 else ""))33    elif kind == "sleep":34        time.sleep(float(parts[1]))35    else:36        raise ValueError(f"unknown action: {spec}")37 38 39def main() -> int:40    ap = argparse.ArgumentParser()41    ap.add_argument("url")42    ap.add_argument("har_path")43    ap.add_argument("--wait", type=float, default=3.0,44                    help="seconds to idle at the end so late XHRs land in the HAR")45    ap.add_argument("--action", action="append", default=[],46                    help="fill:SEL:TEXT | press:SEL:KEY | click:SEL | goto:URL | sleep:SECS")47    ap.add_argument("--headed", action="store_true")48    args = ap.parse_args()49 50    with sync_playwright() as p:51        browser = p.chromium.launch(headless=not args.headed)52        context = browser.new_context(53            record_har_path=args.har_path,54            record_har_content="embed",  # keep response bodies in the HAR55        )56        page = context.new_page()57        page.goto(args.url, wait_until="domcontentloaded")58        for spec in args.action:59            run_action(page, spec)60            try:61                page.wait_for_load_state("networkidle", timeout=15000)62            except Exception:63                pass  # some pages never fully idle; the trailing --wait covers it64        time.sleep(args.wait)65        context.close()  # flushes the HAR66        browser.close()67    print(f"HAR written: {args.har_path}")68    return 069 70 71if __name__ == "__main__":72    sys.exit(main())73