add playwright basic script

Author: ks6088ts

apps/13_langchain_toolkits/README.md

@@ -33,7 +33,18 @@ $ pip install -r requirements.txt
 
 ### Examples
 
-#### [PlayWright Browser Toolkit](https://python.langchain.com/docs/integrations/tools/playwright/)
+#### Playwright
+
+##### [Playwright for Python](https://playwright.dev/python/docs/intro)
+
+- [Test generator](https://playwright.dev/docs/codegen)
+
+```shell
+# Generate Python code from PlayWright
+$ npx playwright codegen --target=python
+```
+
+##### [PlayWright Browser Toolkit](https://python.langchain.com/docs/integrations/tools/playwright/)
 
 ```shell
 $ poetry run playwright install

apps/13_langchain_toolkits/playwright_basic.py

@@ -0,0 +1,100 @@
+import asyncio
+import logging
+import time
+
+import typer
+from dotenv import load_dotenv
+from playwright.async_api import Page, async_playwright
+from playwright.sync_api import sync_playwright
+
+load_dotenv()
+logger = logging.getLogger(__name__)
+app = typer.Typer()
+
+
+async def scrape_task(page: Page, page_number: int):
+    url = f"https://expert.visasq.com/issue/?keyword=&is_started_only=true&page={page_number}"
+    print(f"Downloading {url}...")
+    try:
+        await page.goto(url)
+        await page.wait_for_load_state()
+        await page.screenshot(
+            path=f"screenshot_{page_number}.png",
+            full_page=True,
+        )
+    except Exception as e:
+        logger.error(f"error: {e}, url: {url}")
+    print(f"Downloaded {url}")
+
+
+async def _get_visasq_cases():
+    tasks = []
+    async with async_playwright() as p:
+        logger.debug("Launch browser")
+
+        browser = await p.chromium.launch()
+        context = await browser.new_context()
+
+        for i in range(1, 20):
+            page = await context.new_page()
+            tasks.append(
+                asyncio.create_task(
+                    scrape_task(
+                        page=page,
+                        page_number=i,
+                    )
+                )
+            )
+        await asyncio.gather(*tasks)
+
+        # Finalize
+        await context.close()
+        await browser.close()
+
+
+@app.command()
+def get_visasq_cases(
+    verbose: bool = False,
+):
+    if verbose:
+        logging.basicConfig(level=logging.DEBUG)
+
+    start = time.time()
+    asyncio.run(_get_visasq_cases())
+    end = time.time()
+    elapsed_time = end - start
+    print(f"経過時間:{elapsed_time:.2f} 秒")
+
+
+@app.command()
+def get_yahoo_realtime_trends(
+    verbose: bool = False,
+) -> None:
+    if verbose:
+        logging.basicConfig(level=logging.DEBUG)
+
+    with sync_playwright() as p:
+        logger.debug("Launch browser")
+        browser = p.chromium.launch()
+        context = browser.new_context()
+        page = context.new_page()
+        logger.debug("Go to Yahoo Realtime Trends")
+        page.goto(url="https://search.yahoo.co.jp/realtime")
+        trends = []
+        for i in range(20):
+            keyword = page.get_by_role("link", name=f"{i+1}", exact=True).text_content()
+            trends.append(
+                {
+                    "rank": i + 1,
+                    "keyword": keyword,
+                },
+            )
+        # Finalize
+        context.close()
+        browser.close()
+
+    print(trends)
+
+
+if __name__ == "__main__":
+    app()

poetry.lock

@@ -35,6 +35,7 @@ langgraph-checkpoint-sqlite = "^1.0.4"
 playwright = "^1.47.0"
 lxml = "^5.3.0"
 nest-asyncio = "^1.6.0"
+typer = "^0.12.5"
 
 [tool.poetry.group.dev.dependencies]
 pre-commit = "^3.8.0"

pyproject.toml

@@ -18,6 +18,10 @@ langgraph==0.2.23
 langchain-chroma==0.1.4
 beautifulsoup4==4.12.3
 langgraph-checkpoint-sqlite==1.0.4
+playwright==1.47.0
+lxml==5.3.0
+nest-asyncio==1.6.0
+typer==0.12.5
 
 # To run 99_streamlit_examples/pages/10_Object_Detection.py
 # ultralytics==8.2.89