添加浏览器能力封装
This commit is contained in:
zk
@@ -2,3 +2,4 @@ cryptography>=41
|
||||
rapidocr>=3.9
|
||||
onnxruntime>=1.17
|
||||
opencv-python>=4.5
|
||||
playwright>=1.49
|
||||
|
||||
+254
@@ -0,0 +1,254 @@
|
||||
"""Playwright 页面抓取工具。
|
||||
|
||||
对外仅暴露两个动作：
|
||||
- open_page(url, wait_ms): 打开页面，返回 page_id
|
||||
- query(page_id, selector): 选中节点并提取文字、图片 URL 与属性
|
||||
|
||||
实现上内部常驻一个 Browser；每次 open_page 创建独立 BrowserContext 和 Page，
|
||||
query 完成后自动释放对应资源。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import threading
|
||||
import uuid
|
||||
from dataclasses import dataclass
|
||||
|
||||
from playwright.async_api import Browser, BrowserContext, Page, async_playwright
|
||||
|
||||
|
||||
@dataclass
|
||||
class NodeSnapshot:
|
||||
"""页面节点快照。"""
|
||||
|
||||
tag: str
|
||||
text: str
|
||||
image_urls: list[str]
|
||||
attrs: dict[str, str]
|
||||
|
||||
|
||||
@dataclass
|
||||
class _PageSession:
|
||||
context: BrowserContext
|
||||
page: Page
|
||||
|
||||
|
||||
class _BrowserRuntime:
|
||||
"""在后台事件循环中管理 Playwright 生命周期。"""
|
||||
|
||||
def __init__(self) -> None:
|
||||
self._thread: threading.Thread | None = None
|
||||
self._loop: asyncio.AbstractEventLoop | None = None
|
||||
self._ready = threading.Event()
|
||||
self._lock = threading.RLock()
|
||||
self._playwright = None
|
||||
self._browser: Browser | None = None
|
||||
self._pages: dict[str, _PageSession] = {}
|
||||
|
||||
def _thread_main(self) -> None:
|
||||
loop = asyncio.new_event_loop()
|
||||
asyncio.set_event_loop(loop)
|
||||
self._loop = loop
|
||||
self._ready.set()
|
||||
loop.run_forever()
|
||||
|
||||
pending = asyncio.all_tasks(loop)
|
||||
for task in pending:
|
||||
task.cancel()
|
||||
|
||||
if pending:
|
||||
loop.run_until_complete(asyncio.gather(*pending, return_exceptions=True))
|
||||
|
||||
if self._browser is not None:
|
||||
loop.run_until_complete(self._browser.close())
|
||||
self._browser = None
|
||||
|
||||
if self._playwright is not None:
|
||||
loop.run_until_complete(self._playwright.stop())
|
||||
self._playwright = None
|
||||
|
||||
loop.run_until_complete(loop.shutdown_asyncgens())
|
||||
loop.close()
|
||||
|
||||
def _ensure_thread(self) -> None:
|
||||
with self._lock:
|
||||
if self._thread and self._thread.is_alive():
|
||||
return
|
||||
|
||||
self._thread = threading.Thread(
|
||||
target=self._thread_main,
|
||||
name="offerpie-playwright",
|
||||
daemon=True,
|
||||
)
|
||||
self._thread.start()
|
||||
self._ready.wait()
|
||||
|
||||
def _run(self, coro):
|
||||
self._ensure_thread()
|
||||
assert self._loop is not None
|
||||
future = asyncio.run_coroutine_threadsafe(coro, self._loop)
|
||||
return future.result()
|
||||
|
||||
async def _ensure_browser(self) -> Browser:
|
||||
if self._browser is not None:
|
||||
return self._browser
|
||||
|
||||
if self._playwright is None:
|
||||
self._playwright = await async_playwright().start()
|
||||
|
||||
self._browser = await self._playwright.chromium.launch(headless=True)
|
||||
return self._browser
|
||||
|
||||
async def _open_page(self, url: str, wait_ms: int) -> str:
|
||||
browser = await self._ensure_browser()
|
||||
context = await browser.new_context()
|
||||
page = await context.new_page()
|
||||
await page.goto(url, wait_until="domcontentloaded", timeout=30000)
|
||||
if wait_ms > 0:
|
||||
await page.wait_for_timeout(wait_ms)
|
||||
|
||||
page_id = uuid.uuid4().hex
|
||||
self._pages[page_id] = _PageSession(context=context, page=page)
|
||||
return page_id
|
||||
|
||||
async def _query(self, page_id: str, selector: str) -> list[NodeSnapshot]:
|
||||
session = self._pages.get(page_id)
|
||||
if session is None:
|
||||
raise KeyError(f"Unknown page_id: {page_id}")
|
||||
|
||||
try:
|
||||
locator = session.page.locator(selector)
|
||||
payload = await locator.evaluate_all(
|
||||
r"""(elements) => elements.map((el) => {
|
||||
const attrs = {};
|
||||
for (const attr of Array.from(el.attributes || [])) {
|
||||
attrs[attr.name] = attr.value;
|
||||
}
|
||||
|
||||
const imageUrls = [];
|
||||
const addImageUrl = (url) => {
|
||||
if (!url) return;
|
||||
try {
|
||||
const absUrl = new URL(url, location.href).href;
|
||||
if (
|
||||
absUrl.startsWith('data:') ||
|
||||
absUrl.startsWith('blob:') ||
|
||||
absUrl === 'about:blank'
|
||||
) {
|
||||
return;
|
||||
}
|
||||
imageUrls.push(absUrl);
|
||||
} catch {
|
||||
if (
|
||||
url.startsWith('data:') ||
|
||||
url.startsWith('blob:') ||
|
||||
url === 'about:blank'
|
||||
) {
|
||||
return;
|
||||
}
|
||||
imageUrls.push(url);
|
||||
}
|
||||
};
|
||||
|
||||
const resolveImageUrl = (img) => {
|
||||
const candidates = [
|
||||
img.getAttribute('data-src'),
|
||||
img.getAttribute('data-original'),
|
||||
img.getAttribute('data-croporisrc'),
|
||||
img.currentSrc,
|
||||
img.src,
|
||||
img.getAttribute('src'),
|
||||
];
|
||||
|
||||
const srcset = img.getAttribute('srcset');
|
||||
if (srcset) {
|
||||
for (const part of srcset.split(',')) {
|
||||
const candidate = part.trim().split(/\s+/)[0];
|
||||
candidates.push(candidate);
|
||||
}
|
||||
}
|
||||
|
||||
for (const candidate of candidates) {
|
||||
if (!candidate) continue;
|
||||
if (
|
||||
candidate.startsWith('data:') ||
|
||||
candidate.startsWith('blob:') ||
|
||||
candidate === 'about:blank'
|
||||
) {
|
||||
continue;
|
||||
}
|
||||
try {
|
||||
const absUrl = new URL(candidate, location.href).href;
|
||||
if (
|
||||
absUrl.startsWith('data:') ||
|
||||
absUrl.startsWith('blob:') ||
|
||||
absUrl === 'about:blank'
|
||||
) {
|
||||
continue;
|
||||
}
|
||||
return absUrl;
|
||||
} catch {
|
||||
return candidate;
|
||||
}
|
||||
}
|
||||
|
||||
return null;
|
||||
};
|
||||
|
||||
if (el.tagName === 'IMG') {
|
||||
addImageUrl(resolveImageUrl(el));
|
||||
}
|
||||
|
||||
for (const img of Array.from(el.querySelectorAll('img'))) {
|
||||
addImageUrl(resolveImageUrl(img));
|
||||
}
|
||||
|
||||
const text = (el.innerText || el.textContent || '').trim();
|
||||
|
||||
return {
|
||||
tag: (el.tagName || '').toLowerCase(),
|
||||
text,
|
||||
image_urls: Array.from(new Set(imageUrls)),
|
||||
attrs,
|
||||
};
|
||||
})"""
|
||||
)
|
||||
return [
|
||||
NodeSnapshot(
|
||||
tag=item.get("tag", ""),
|
||||
text=item.get("text", ""),
|
||||
image_urls=list(item.get("image_urls", [])),
|
||||
attrs=dict(item.get("attrs", {})),
|
||||
)
|
||||
for item in payload
|
||||
]
|
||||
finally:
|
||||
await session.page.close()
|
||||
await session.context.close()
|
||||
self._pages.pop(page_id, None)
|
||||
|
||||
|
||||
_runtime = _BrowserRuntime()
|
||||
|
||||
|
||||
def open_page(url: str, wait_ms: int = 3000) -> str:
|
||||
"""打开页面并返回 page_id。"""
|
||||
return _runtime._run(_runtime._open_page(url, wait_ms))
|
||||
|
||||
|
||||
def query(page_id: str, selector: str) -> list[NodeSnapshot]:
|
||||
"""按选择器抓取节点，提取文本、图片 URL 与属性。"""
|
||||
items = _runtime._run(_runtime._query(page_id, selector))
|
||||
seen_urls: set[str] = set()
|
||||
|
||||
for item in items:
|
||||
unique_urls: list[str] = []
|
||||
for url in item.image_urls:
|
||||
if url in seen_urls:
|
||||
continue
|
||||
seen_urls.add(url)
|
||||
unique_urls.append(url)
|
||||
item.image_urls = unique_urls
|
||||
|
||||
return items
|
||||
Reference in New Issue
Block a user