feat: Implement passive XHR/fetch capture using Playwright
- Added `capture_passive_playwright.py` for synchronous passive capture of XHR/fetch responses. - Added `capture_passive_playwright_async.py` for asynchronous passive capture using Playwright's async API. - Introduced `save_page_snapshots.py` to save full page HTML snapshots with deduplication. - Removed obsolete CSRF token capture JSON file.
This commit is contained in:
Ross
@@ -0,0 +1,307 @@
|
||||
#!/usr/bin/env python3
|
||||
"""Async passive XHR/fetch capture using Playwright's async API.
|
||||
|
||||
This script mirrors the sync version but uses async/await to avoid blocking
|
||||
the event loop. It attaches listeners before navigating and prints a small
|
||||
heartbeat that shows how many responses we've captured so you can see
|
||||
activity live.
|
||||
|
||||
Usage:
|
||||
python scrapers/capture_passive_playwright_async.py --output-dir xhr_captured_async --continuous
|
||||
"""
|
||||
import argparse
|
||||
import asyncio
|
||||
import json
|
||||
import mimetypes
|
||||
import hashlib
|
||||
import os
|
||||
from datetime import datetime, timezone
|
||||
from pathlib import Path
|
||||
|
||||
from dotenv import load_dotenv
|
||||
load_dotenv()
|
||||
|
||||
try:
|
||||
from playwright.async_api import async_playwright
|
||||
except Exception:
|
||||
raise
|
||||
|
||||
|
||||
def sanitize(s: str) -> str:
|
||||
out = []
|
||||
for ch in s:
|
||||
if ch.isalnum() or ch in ('-', '_', '.'):
|
||||
out.append(ch)
|
||||
else:
|
||||
out.append('_')
|
||||
name = ''.join(out)
|
||||
if len(name) > 180:
|
||||
name = name[:120] + '_' + hashlib.sha1(s.encode('utf-8')).hexdigest()[:8]
|
||||
return name
|
||||
|
||||
|
||||
def guess_ext(ctype: str) -> str:
|
||||
if not ctype:
|
||||
return 'bin'
|
||||
mt = ctype.split(';')[0].strip()
|
||||
ext = mimetypes.guess_extension(mt)
|
||||
if ext:
|
||||
return ext.lstrip('.')
|
||||
if mt.startswith('image/'):
|
||||
return mt.split('/')[1]
|
||||
if mt == 'application/json':
|
||||
return 'json'
|
||||
if mt == 'text/html':
|
||||
return 'html'
|
||||
if mt.startswith('text/'):
|
||||
return 'txt'
|
||||
return 'bin'
|
||||
|
||||
|
||||
def now_ts():
|
||||
return datetime.now(timezone.utc).strftime('%Y%m%dT%H%M%SZ')
|
||||
|
||||
|
||||
def save_text(path: Path, text: str):
|
||||
path.parent.mkdir(parents=True, exist_ok=True)
|
||||
with open(path, 'w', encoding='utf-8') as f:
|
||||
f.write(text)
|
||||
|
||||
|
||||
def save_binary(path: Path, data: bytes):
|
||||
path.parent.mkdir(parents=True, exist_ok=True)
|
||||
with open(path, 'wb') as f:
|
||||
f.write(data)
|
||||
|
||||
|
||||
async def run(args):
|
||||
output = Path(args.output_dir)
|
||||
output.mkdir(parents=True, exist_ok=True)
|
||||
index_path = output / 'capture_index.jsonl'
|
||||
|
||||
capture_count = 0
|
||||
capture_count_lock = asyncio.Lock()
|
||||
|
||||
async def heartbeat():
|
||||
while True:
|
||||
await asyncio.sleep(5)
|
||||
async with capture_count_lock:
|
||||
print(f"[heartbeat] captured={capture_count}")
|
||||
|
||||
async with async_playwright() as p:
|
||||
browser_type = p.chromium
|
||||
# Launch persistent context
|
||||
if args.channel:
|
||||
try:
|
||||
context = await browser_type.launch_persistent_context(user_data_dir=args.profile, headless=args.headless, channel=args.channel)
|
||||
except Exception:
|
||||
print(f'Launch with channel {args.channel} failed; falling back to bundled Chromium')
|
||||
context = await browser_type.launch_persistent_context(user_data_dir=args.profile, headless=args.headless)
|
||||
else:
|
||||
context = await browser_type.launch_persistent_context(user_data_dir=args.profile, headless=args.headless)
|
||||
|
||||
page = await context.new_page()
|
||||
|
||||
async def on_response(resp):
|
||||
nonlocal capture_count
|
||||
try:
|
||||
req = resp.request
|
||||
rtype = (req.resource_type or '').lower()
|
||||
|
||||
# write quick debug line
|
||||
try:
|
||||
with open(output / '_debug_events.log', 'a', encoding='utf-8') as dbg:
|
||||
dbg.write(f"{now_ts()}\tRESP\t{resp.status}\t{rtype}\t{req.url}\n")
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
allow = False
|
||||
if not args.capture_types:
|
||||
allow = True
|
||||
elif rtype in args.capture_types:
|
||||
allow = True
|
||||
if '/document/content/' in req.url or '/document/summary/' in req.url:
|
||||
allow = True
|
||||
if not allow:
|
||||
return
|
||||
|
||||
url = req.url
|
||||
ts = now_ts()
|
||||
resp_headers = dict(resp.headers)
|
||||
ctype = (resp_headers.get('content-type') or '').lower()
|
||||
|
||||
safe = sanitize(url.replace('https://', '').replace('http://', ''))
|
||||
h = hashlib.sha1((url + ts).encode('utf-8')).hexdigest()[:8]
|
||||
|
||||
body_path = None
|
||||
excerpt = None
|
||||
|
||||
try:
|
||||
if 'application/json' in ctype or ctype.startswith('text/') or 'javascript' in ctype:
|
||||
try:
|
||||
txt = await resp.text()
|
||||
except asyncio.CancelledError:
|
||||
return
|
||||
except Exception as e:
|
||||
if 'Target page, context or browser has been closed' in str(e):
|
||||
return
|
||||
raise
|
||||
# try pretty JSON
|
||||
try:
|
||||
parsed = json.loads(txt)
|
||||
pretty = json.dumps(parsed, ensure_ascii=False, indent=2)
|
||||
fname = f"{safe}_{h}_{ts}.json"
|
||||
body_path = str(output / fname)
|
||||
save_text(output / fname, pretty)
|
||||
excerpt = pretty[:800]
|
||||
except Exception:
|
||||
fname = f"{safe}_{h}_{ts}.txt"
|
||||
body_path = str(output / fname)
|
||||
save_text(output / fname, txt)
|
||||
excerpt = txt[:800]
|
||||
else:
|
||||
try:
|
||||
data = await resp.body()
|
||||
except asyncio.CancelledError:
|
||||
return
|
||||
except Exception as e:
|
||||
if 'Target page, context or browser has been closed' in str(e):
|
||||
return
|
||||
raise
|
||||
ext = guess_ext(ctype)
|
||||
# Put images into an images/ subdirectory, other binaries into assets/
|
||||
if ctype.startswith('image/'):
|
||||
subdir = output / 'images'
|
||||
subdir.mkdir(parents=True, exist_ok=True)
|
||||
fname = f"{safe}_{h}_{ts}.{ext}"
|
||||
body_path = str(subdir / fname)
|
||||
save_binary(subdir / fname, data)
|
||||
else:
|
||||
subdir = output / 'assets'
|
||||
subdir.mkdir(parents=True, exist_ok=True)
|
||||
fname = f"{safe}_{h}_{ts}.{ext}"
|
||||
body_path = str(subdir / fname)
|
||||
save_binary(subdir / fname, data)
|
||||
except Exception as e:
|
||||
# final fallback: try binary
|
||||
try:
|
||||
data = await resp.body()
|
||||
ext = guess_ext(ctype)
|
||||
if ctype.startswith('image/'):
|
||||
subdir = output / 'images'
|
||||
subdir.mkdir(parents=True, exist_ok=True)
|
||||
fname = f"{safe}_{h}_{ts}.{ext}"
|
||||
body_path = str(subdir / fname)
|
||||
save_binary(subdir / fname, data)
|
||||
else:
|
||||
subdir = output / 'assets'
|
||||
subdir.mkdir(parents=True, exist_ok=True)
|
||||
fname = f"{safe}_{h}_{ts}.{ext}"
|
||||
body_path = str(subdir / fname)
|
||||
save_binary(subdir / fname, data)
|
||||
except Exception:
|
||||
# skip if still failing
|
||||
return
|
||||
|
||||
meta = {
|
||||
'url': url,
|
||||
'resource_type': rtype,
|
||||
'status': resp.status,
|
||||
'timestamp': ts,
|
||||
'response_headers': resp_headers,
|
||||
'response_body_file': body_path,
|
||||
'response_excerpt': excerpt,
|
||||
}
|
||||
meta_name = f"{safe}_{h}_{ts}.meta.json"
|
||||
save_text(output / meta_name, json.dumps(meta, ensure_ascii=False, indent=2))
|
||||
try:
|
||||
with open(index_path, 'a', encoding='utf-8') as idx:
|
||||
idx.write(json.dumps({'url': url, 'resource_type': rtype, 'timestamp': ts, 'body_file': body_path, 'meta_file': str(output / meta_name), 'excerpt': excerpt}, ensure_ascii=False) + '\n')
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
async with capture_count_lock:
|
||||
capture_count += 1
|
||||
|
||||
except Exception:
|
||||
# swallow unexpected exceptions from handler so event loop stays healthy
|
||||
try:
|
||||
with open(output / '_debug_events.log', 'a', encoding='utf-8') as dbg:
|
||||
dbg.write(f"{now_ts()}\tRESP_HANDLER_ERROR\n")
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
# helpers to attach to page
|
||||
def attach_page_listeners(pg):
|
||||
pg.on('response', lambda r: asyncio.create_task(on_response(r)))
|
||||
pg.on('request', lambda req: open(output / '_debug_events.log', 'a', encoding='utf-8').write(f"{now_ts()}\tREQUEST\t{req.method}\t{req.url}\n"))
|
||||
pg.on('requestfinished', lambda req: open(output / '_debug_events.log', 'a', encoding='utf-8').write(f"{now_ts()}\tREQUEST_FINISHED\t{req.method}\t{req.url}\n"))
|
||||
|
||||
# attach listeners to existing pages
|
||||
for pg in context.pages:
|
||||
attach_page_listeners(pg)
|
||||
|
||||
# attach to future pages
|
||||
context.on('page', lambda new_page: attach_page_listeners(new_page))
|
||||
|
||||
# context-level response and request listeners
|
||||
context.on('response', lambda r: asyncio.create_task(on_response(r)))
|
||||
context.on('request', lambda req: open(output / '_debug_events.log', 'a', encoding='utf-8').write(f"{now_ts()}\tCTX_REQUEST\t{req.method}\t{req.url}\n"))
|
||||
|
||||
# Start heartbeat
|
||||
hb_task = asyncio.create_task(heartbeat())
|
||||
|
||||
# Navigate after listeners attached
|
||||
try:
|
||||
await page.goto(args.url)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
# If interactive, let user log in; otherwise start capture immediately
|
||||
if not args.no_prompt:
|
||||
print('When you have logged in in the opened browser, press Enter here to continue and capture...')
|
||||
try:
|
||||
await asyncio.get_event_loop().run_in_executor(None, input)
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
print(f'Starting async passive capture (types: {args.capture_types})')
|
||||
|
||||
try:
|
||||
if args.continuous:
|
||||
# keep running until KeyboardInterrupt
|
||||
while True:
|
||||
await asyncio.sleep(1)
|
||||
else:
|
||||
await asyncio.sleep(5)
|
||||
except KeyboardInterrupt:
|
||||
print('Interrupted; closing...')
|
||||
finally:
|
||||
hb_task.cancel()
|
||||
try:
|
||||
await context.close()
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
|
||||
def parse_args():
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--url', default='https://app.statdx.com/')
|
||||
parser.add_argument('--profile', default='playwright_profile')
|
||||
parser.add_argument('--output-dir', default='xhr_captured_async')
|
||||
parser.add_argument('--headless', action='store_true')
|
||||
parser.add_argument('--continuous', action='store_true')
|
||||
parser.add_argument('--no-prompt', action='store_true')
|
||||
parser.add_argument('--channel', default=os.getenv('PLAYWRIGHT_CHROME_CHANNEL', 'chrome'))
|
||||
parser.add_argument('--capture-types', default='xhr,fetch,document,other')
|
||||
return parser.parse_args()
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
args = parse_args()
|
||||
# normalize capture types to a set
|
||||
args.capture_types = {t.strip().lower() for t in args.capture_types.split(',') if t.strip()}
|
||||
try:
|
||||
asyncio.run(run(args))
|
||||
except KeyboardInterrupt:
|
||||
pass
|
||||
Reference in New Issue
Block a user