Spaces:

zhou12189108
/

h-point

Running

App Files Files Community

zhou12189108 commited on Jan 15, 2024

Commit

1488b3f

verified ·

1 Parent(s): b048c19

Upload hcaptcha_solver.py

Browse files

Files changed (1) hide show

hcaptcha_solver.py +137 -89

hcaptcha_solver.py CHANGED Viewed

@@ -1,8 +1,13 @@
-from __future__ import annotations
-from loguru import logger
-from playwright.async_api import BrowserContext as ASyncContext, async_playwright, Page
-from hcaptcha_challenger import ModelHub, install
-from hcaptcha_challenger.agents import AgentT, Malenia
 async def route_continuation(route, request, host, sitekey):
@@ -17,6 +22,7 @@ async def route_continuation(route, request, host, sitekey):
 <html lang="en">
 <head>
 <title>hCAPTCHA 演示</title>
 <meta http-equiv="X-UA-Compatible" content="IE=edge">
 <meta name="viewport" content="width=device-width, user-scalable=yes">
 <script src="https://js.hcaptcha.com/1/api.js" type="text/javascript" async defer></script>
@@ -72,96 +78,121 @@ async def route_continuation(route, request, host, sitekey):
         await route.continue_()
-def patch_modelhub(modelhub: ModelHub):
-    """
-    1. Patching clip_candidates allows you to handle all image classification tasks in self-supervised mode.
-    2. You need to inject hints for all categories that appear in a batch of images
-    3. The ObjectsYaml in the GitHub repository are updated regularly,
-    but if you find something new, you can imitate the following and patch some hints.
-    4. Note that this should be a regularly changing table.
-    If after a while certain labels no longer appear, you should not fill them in clip_candidates
-    5. Please note that you only need a moderate number of candidates prompts,
-    too many prompts will increase the computational complexity
-    :param modelhub:
-    :return:
-    """
-    modelhub.clip_candidates.update(
-        {
-            "the largest animal in real life": [
-                "parrot",
-                "bee",
-                "ladybug",
-                "frog",
-                "crab",
-                "bat",
-                "butterfly",
-                "dragonfly",
-                "giraffe",
-                "tiger",
-                "owl",
-                "duck"
-            ]
-        }
-    )
-def prelude(page: Page) -> AgentT:
-    # 1. You need to deploy sub-thread tasks and actively run `install(upgrade=True)` every 20 minutes
-    # 2. You need to make sure to run `install(upgrade=True, clip=True)` before each instantiation
-    install(upgrade=True, clip=True)
-    modelhub = ModelHub.from_github_repo()
-    modelhub.parse_objects()
-    # Make arbitrary pre-modifications to modelhub, which is very useful for CLIP models
-    patch_modelhub(modelhub)
-    agent = AgentT.from_page(
-        # page, the control handle of the Playwright Page
-        page=page,
-        # modelhub, Register modelhub externally, and the agent can patch custom configurations
-        modelhub=modelhub,
-        tmp_dir="tmp_dir",
-        # clip, Enable CLIP zero-shot image classification method
-        clip=True,
     )
-    return agent
 async def hit_challenge(context: ASyncContext, host, sitekey, times: int = 8):
     await context.route('**/*', lambda route, request: route_continuation(route, request, host, sitekey))
     page = await context.new_page()
-    agent = prelude(page)
     await page.goto(f"https://{host}")
-    logger.info("startup sitelink", url=f"https://{host}")
-    await agent.handle_checkbox()
-    for pth in range(1, times):
-        # Handle challenge
-        result = await agent.execute()
-        if not agent.qr:
-            return
-        # Post-processing
-        match result:
-            case agent.status.CHALLENGE_BACKCALL | agent.status.CHALLENGE_RETRY:
-                logger.warning(f"retry", pth=pth, ash=agent.ash)
-                await page.wait_for_timeout(500)
-                fl = page.frame_locator(agent.HOOK_CHALLENGE)
-                await fl.locator("//div[@class='refresh button']").click()
-            case agent.status.CHALLENGE_SUCCESS:
-                logger.success(f"task done", pth=pth, ash=agent.ash)
-                rqdata = agent.cr.__dict__
-                await context.close()
-                return rqdata["generated_pass_UUID"]
 async def bytedance(host, sitekey):
@@ -171,6 +202,23 @@ async def bytedance(host, sitekey):
             locale="en-US"
         )
         await Malenia.apply_stealth(context)
-        token = await hit_challenge(context, host, sitekey)
         return token

+import asyncio
+import base64
+import random
+import nopecha
+import requests
+from hcaptcha_challenger.agents import Malenia
+from playwright.async_api import BrowserContext as ASyncContext, async_playwright
+nopecha.api_key = '43dkmk7lmv9qhqzx'
 async def route_continuation(route, request, host, sitekey):
 <html lang="en">
 <head>
 <title>hCAPTCHA 演示</title>
+<meta charset="UTF-8">
 <meta http-equiv="X-UA-Compatible" content="IE=edge">
 <meta name="viewport" content="width=device-width, user-scalable=yes">
 <script src="https://js.hcaptcha.com/1/api.js" type="text/javascript" async defer></script>
         await route.continue_()
+def url_to_base64(url):
+    try:
+        # 获取 URL 的内容
+        response = requests.get(url)
+        # 将内容转换为 base64
+        content_base64 = base64.b64encode(response.content).decode('utf-8')
+        return content_base64
+    except requests.exceptions.RequestException as e:
+        print(f"Error fetching URL: {e}")
+        return None
+async def on_response(response, page):
+    if response.url.startswith("https://api.hcaptcha.com/getcaptcha"):
+        # 获取响应内容
+        data = await response.json()
+        # 打印响应内容
+        print(data)
+        examples_urls = data.get("requester_question_example")
+        examples = [url_to_base64(i) for i in examples_urls] if examples_urls else None
+        if data.get("request_type") == "image_label_binary":
+            await classify_click(page, data, 0, examples)
+        elif data.get("request_type") == "image_label_area_select":
+            # await asyncio.sleep(random.uniform(0.1, 0.3))
+            # await page.wait_for_selector("//iframe[contains(@title, 'hCaptcha challenge')]")
+            # frame_challenge = page.frame_locator("//iframe[contains(@title, 'hCaptcha challenge')]")
+            # await frame_challenge.locator("//div[@class='refresh button']").click()
+            await area_click(page, data, 0, examples)
+        else:
+            await asyncio.sleep(random.uniform(0.1, 0.3))
+            await page.wait_for_selector("//iframe[contains(@title, 'hCaptcha challenge')]")
+            frame_challenge = page.frame_locator("//iframe[contains(@title, 'hCaptcha challenge')]")
+            await frame_challenge.locator("//div[@class='refresh button']").click()
+    elif response.url.startswith("https://api.hcaptcha.com/checkcaptcha"):
+        data0 = await response.json()
+        if data0.get("pass"):
+            global tasks, token
+            await page.close()
+            token = data0.get("generated_pass_UUID")
+            tasks.cancel()
+        print(data0)
+async def classify_click(page, data, round0, examples):
+    await page.wait_for_selector("//iframe[contains(@title, 'hCaptcha challenge')]")
+    frame_challenge = page.frame_locator("//iframe[contains(@title, 'hCaptcha challenge')]")
+    samples = frame_challenge.locator("//div[@class='task-image']")
+    await frame_challenge.locator("//div[@tabindex='0']").nth(0).wait_for()
+    count = await samples.count()
+    print(count)
+    clicks = nopecha.Recognition.solve(
+        type='hcaptcha',
+        task=data.get("requester_question").get("en"),
+        image_urls=[data.get("tasklist")[i]["datapoint_uri"] for i in (range(9) if round0 == 0 else range(9, 18))],
+        image_examples=examples
     )
+    for i in range(count):
+        sample = samples.nth(i)
+        await sample.wait_for()
+        if clicks[i]:
+            print("try to click")
+            await sample.click(delay=200)
+    print(clicks)
+    await asyncio.sleep(random.uniform(0.1, 0.3))
+    fl = frame_challenge.locator("//div[@class='button-submit button']")
+    await fl.click()
+    if round0 == 0:
+        await classify_click(page, data, 1, examples)
+async def area_click(page, data, round0, examples):
+    await page.wait_for_selector("//iframe[contains(@title, 'hCaptcha challenge')]")
+    frame_challenge = page.frame_locator("//iframe[contains(@title, 'hCaptcha challenge')]")
+    locator = frame_challenge.locator("//div[@class='challenge-view']//canvas")
+    await locator.wait_for(state="visible")
+    image = await locator.screenshot()
+    if examples:
+        clicks = nopecha.Recognition.solve(
+            type='hcaptcha_area_select',
+            task=data.get("requester_question").get("en"),
+            image_data=[url_to_base64(data.get("tasklist")[0]["datapoint_uri"])],
+            image_examples=examples
+        )
+    else:
+        clicks = nopecha.Recognition.solve(
+            type='hcaptcha_area_select',
+            task=data.get("requester_question").get("en"),
+            image_data=[url_to_base64(data.get("tasklist")[0]["datapoint_uri"])]
+        )
+    print(clicks)
+    print(clicks["x"], clicks["y"])
+    print("try to click")
+    bounds = await locator.bounding_box()
+    print(bounds)
+    await locator.click(delay=200, position={"x": int(bounds["width"] * clicks["x"] / 100),
+                                             "y": int(bounds["height"] * clicks["y"] / 100)})
+    print("done")
+    fl = frame_challenge.locator("//div[@class='button-submit button']")
+    await fl.click()
+    await asyncio.sleep(random.uniform(0.1, 0.3))
+    if round0 == 0:
+        await area_click(page, data, 1, examples)
 async def hit_challenge(context: ASyncContext, host, sitekey, times: int = 8):
     await context.route('**/*', lambda route, request: route_continuation(route, request, host, sitekey))
     page = await context.new_page()
+    page.on('response', lambda response: on_response(response, page))
     await page.goto(f"https://{host}")
+    checkbox = page.frame_locator("//iframe[contains(@title,'checkbox')]")
+    await checkbox.locator("#checkbox").click()
+    await asyncio.sleep(3000)
 async def bytedance(host, sitekey):
             locale="en-US"
         )
         await Malenia.apply_stealth(context)
+        await hit_challenge(context, host, sitekey)
+question = {}
+tasks = None
+token = None
+async def main(host,key):
+    global tasks, token
+    try:
+        tasks = asyncio.gather(bytedance(host,key),
+                               return_exceptions=True)
+        await tasks
+    except asyncio.CancelledError:
+        print("task done")
         return token