hyper-sdk 2.12.4__tar.gz → 2.13.0__tar.gz

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hyper_sdk
-Version: 2.12.4
+Version: 2.13.0
 Summary: Hyper Solutions Python SDK
 License: MIT License
         
@@ -386,6 +386,34 @@ device_link = parse_slider_device_check_link(
 )
 ```
 
+### Challenge Script
+
+DataDome sometimes serves a challenge page with the **challenge script in its own file** instead of inlining it in the page:
+
+```html
+<script defer src="https://ct.captcha-delivery.com/interstitial.1.33.0.202609141.js"></script>
+```
+
+It switches between the two forms per request, so check every challenge page. When the page uses a separate file, fetch it with your own client so the request keeps your proxy, TLS fingerprint and headers, then pass the body as `script`:
+
+```python
+from hyper_sdk import DataDomeInterstitialInput
+from hyper_sdk.datadome import parse_challenge_script_url
+
+script = ""
+script_url = parse_challenge_script_url(html_content)
+if script_url is not None:
+    script = your_client.get(script_url).text
+
+result = session.generate_interstitial_payload(DataDomeInterstitialInput(
+    # other interstitial input fields
+    html=html_content,
+    script=script,  # empty when the page inlines the script
+))
+```
+
+The same applies to `DataDomeSliderInput`. One helper covers both challenge types, since the captcha page uses the same tag. `script` is optional and leaving it empty keeps working.
+
 ## 📖 Documentation
 
 For detailed documentation on how to use the SDK, including examples and API reference, please visit our documentation website:

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/README.md

@@ -343,6 +343,34 @@ device_link = parse_slider_device_check_link(
 )
 ```
 
+### Challenge Script
+
+DataDome sometimes serves a challenge page with the **challenge script in its own file** instead of inlining it in the page:
+
+```html
+<script defer src="https://ct.captcha-delivery.com/interstitial.1.33.0.202609141.js"></script>
+```
+
+It switches between the two forms per request, so check every challenge page. When the page uses a separate file, fetch it with your own client so the request keeps your proxy, TLS fingerprint and headers, then pass the body as `script`:
+
+```python
+from hyper_sdk import DataDomeInterstitialInput
+from hyper_sdk.datadome import parse_challenge_script_url
+
+script = ""
+script_url = parse_challenge_script_url(html_content)
+if script_url is not None:
+    script = your_client.get(script_url).text
+
+result = session.generate_interstitial_payload(DataDomeInterstitialInput(
+    # other interstitial input fields
+    html=html_content,
+    script=script,  # empty when the page inlines the script
+))
+```
+
+The same applies to `DataDomeSliderInput`. One helper covers both challenge types, since the captcha page uses the same tag. `script` is optional and leaving it empty keeps working.
+
 ## 📖 Documentation
 
 For detailed documentation on how to use the SDK, including examples and API reference, please visit our documentation website:

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/hyper_sdk/datadome/parse.py

@@ -1,6 +1,16 @@
 import json
+import re
+from typing import Optional
 from urllib.parse import urlencode
 
+# Matches the tag a challenge page uses to load the bundle from
+# ct.captcha-delivery.com, capturing the URL. The name in front of the version
+# is left open so it keeps matching whatever DataDome calls the bundle, e.g.
+# interstitial.1.33.0.202609141.js or captcha.1.34.0.202609141.js.
+_CHALLENGE_SCRIPT_REGEX = re.compile(
+    r"""<script[^>]+src\s*=\s*["']([^"']*/[a-z_-]+\.\d+\.\d+\.\d+\.[^"']*\.js)["']"""
+)
+
 
 def parse_slider_device_check_link(src: str, datadome_cookie: str, referer: str) -> str:
     """
@@ -82,3 +92,49 @@ def parse_interstitial_device_check_link(src: str, datadome_cookie: str, referer
     }
 
     return f"https://geo.captcha-delivery.com/interstitial/?{urlencode(params)}"
+
+
+def parse_challenge_script_url(html: str) -> Optional[str]:
+    """
+        Return the challenge bundle URL a device check or captcha page loads with a
+        <script defer src="..."> tag, or None when the page has no such tag.
+
+        DataDome serves some challenge pages with the bundle inlined in the HTML and
+        others with it in its own file, switching between the two per request, so this
+        has to be checked on every challenge rather than configured once.
+
+        When it returns a URL, GET that URL with the same client, proxy and headers you
+        used for the challenge page, and pass the response body as the `script` field of
+        DataDomeInterstitialInput or DataDomeSliderInput. When it returns None the
+        bundle is already in the HTML and `script` stays empty.
+
+        Args:
+            html (str): The response body of the GET request to the device check link.
+
+        Returns:
+            Optional[str]: The challenge bundle URL, or None when the page inlines it.
+
+        Example:
+            device_link = parse_interstitial_device_check_link(body, cookie, referer)
+            # ... GET device_link with your own client, read the body into html ...
+
+            script = ""
+            script_url = parse_challenge_script_url(html)
+            if script_url is not None:
+                # fetch script_url with your own client, read the body into script
+                script = your_client.get(script_url).text
+
+            result = session.generate_interstitial_payload(DataDomeInterstitialInput(
+                user_agent=user_agent,
+                device_link=device_link,
+                html=html,
+                accept_language=accept_language,
+                ip=ip,
+                script=script,
+            ))
+    """
+    match = _CHALLENGE_SCRIPT_REGEX.search(html)
+    if match is None:
+        return None
+
+    return match.group(1)

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/hyper_sdk/datadome_input.py

@@ -1,6 +1,6 @@
 
 class DataDomeSliderInput:
-    def __init__(self, user_agent: str, device_link: str, html: str, puzzle: str, piece: str, parent_url: str, accept_language: str, ip: str):
+    def __init__(self, user_agent: str, device_link: str, html: str, puzzle: str, piece: str, parent_url: str, accept_language: str, ip: str, script: str = ""):
         # UserAgent must be a Chrome Windows User-Agent.
         self.user_agent = user_agent
 
@@ -22,11 +22,21 @@ class DataDomeSliderInput:
         self.piece = piece
 
         self.parent_url = parent_url
+
+        # Script is the DataDome challenge bundle. It is only needed when the
+        # challenge page loads the bundle with a <script defer src="..."> tag
+        # instead of inlining it, which DataDome does per request. Use
+        # hyper_sdk.datadome.parse_challenge_script_url on the html to find out,
+        # then GET that URL with the same client and pass the response body here.
+        #
+        # Leave it empty when the page inlines the bundle.
+        self.script = script
+
         self.accept_language = accept_language
         self.ip = ip
 
     def to_dict(self):
-        return {
+        data = {
             "userAgent": self.user_agent,
             "deviceLink": self.device_link,
             "html": self.html,
@@ -36,10 +46,13 @@ class DataDomeSliderInput:
             "acceptLanguage": self.accept_language,
             "ip": self.ip,
         }
+        if self.script:  # Only include script if it's not empty
+            data["script"] = self.script
+        return data
 
 
 class DataDomeInterstitialInput:
-    def __init__(self, user_agent: str, device_link: str, html: str, accept_language: str, ip: str):
+    def __init__(self, user_agent: str, device_link: str, html: str, accept_language: str, ip: str, script: str = ""):
         # UserAgent must be a Chrome Windows User-Agent.
         self.user_agent = user_agent
 
@@ -50,17 +63,29 @@ class DataDomeInterstitialInput:
         # Html is the response body of the GET request to the DeviceLink
         self.html = html
 
+        # Script is the DataDome challenge bundle. It is only needed when the
+        # challenge page loads the bundle with a <script defer src="..."> tag
+        # instead of inlining it, which DataDome does per request. Use
+        # hyper_sdk.datadome.parse_challenge_script_url on the html to find out,
+        # then GET that URL with the same client and pass the response body here.
+        #
+        # Leave it empty when the page inlines the bundle.
+        self.script = script
+
         self.accept_language = accept_language
         self.ip = ip
 
     def to_dict(self):
-        return {
+        data = {
             "userAgent": self.user_agent,
             "deviceLink": self.device_link,
             "html": self.html,
             "acceptLanguage": self.accept_language,
             "ip": self.ip,
         }
+        if self.script:  # Only include script if it's not empty
+            data["script"] = self.script
+        return data
 
 
 class DataDomeTagsInput:

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/hyper_sdk/incapsula/dynamic.py

@@ -2,7 +2,7 @@ import re
 from urllib.parse import urlparse
 
 # Precompiled regular expressions
-reese_script_regex = re.compile(r'src\s*=\s*"((/[^/]+/\d+)(?:\?.*)?)"')
+reese_script_regex = re.compile(r'src\s*=\s*"((/[^/]+/[^/?"]+)\?s=[^"]*)"')
 
 
 def parse_dynamic_reese_script(html_content: str, url_str: str) -> tuple[str, str]:

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/hyper_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hyper_sdk
-Version: 2.12.4
+Version: 2.13.0
 Summary: Hyper Solutions Python SDK
 License: MIT License
         
@@ -386,6 +386,34 @@ device_link = parse_slider_device_check_link(
 )
 ```
 
+### Challenge Script
+
+DataDome sometimes serves a challenge page with the **challenge script in its own file** instead of inlining it in the page:
+
+```html
+<script defer src="https://ct.captcha-delivery.com/interstitial.1.33.0.202609141.js"></script>
+```
+
+It switches between the two forms per request, so check every challenge page. When the page uses a separate file, fetch it with your own client so the request keeps your proxy, TLS fingerprint and headers, then pass the body as `script`:
+
+```python
+from hyper_sdk import DataDomeInterstitialInput
+from hyper_sdk.datadome import parse_challenge_script_url
+
+script = ""
+script_url = parse_challenge_script_url(html_content)
+if script_url is not None:
+    script = your_client.get(script_url).text
+
+result = session.generate_interstitial_payload(DataDomeInterstitialInput(
+    # other interstitial input fields
+    html=html_content,
+    script=script,  # empty when the page inlines the script
+))
+```
+
+The same applies to `DataDomeSliderInput`. One helper covers both challenge types, since the captcha page uses the same tag. `script` is optional and leaving it empty keeps working.
+
 ## 📖 Documentation
 
 For detailed documentation on how to use the SDK, including examples and API reference, please visit our documentation website:

{hyper_sdk-2.12.4 → hyper_sdk-2.13.0}/pyproject.toml

@@ -1,38 +1,38 @@
-[build-system]
-requires = ["setuptools>=42", "wheel"]
-build-backend = "setuptools.build_meta"
-
-[project]
-name = "hyper_sdk"
-version = "2.12.4"
-description = "Hyper Solutions Python SDK"
-readme = "README.md"
-requires-python = ">=3.7"
-classifiers = [
-    "Intended Audience :: Developers",
-    "License :: OSI Approved :: MIT License",
-    "Programming Language :: Python :: 3",
-    "Framework :: AsyncIO",
-]
-dependencies = [
-    "httpx[http2]>=0.24.0",
-    "certifi>=2024.2.2",
-    "charset-normalizer>=3.3.2",
-    "idna>=3.6",
-    "jsonpickle>=3.0.3",
-    "PyJWT>=2.8.0",
-    "urllib3>=2.2.1"
-]
-license = {file = "LICENSE"}
-
-[project.urls]
-homepage = "https://github.com/Hyper-Solutions/hyper-sdk-py"
-
-[tool.setuptools]
-packages = [
-    "hyper_sdk",
-    "hyper_sdk.akamai",
-    "hyper_sdk.incapsula",
-    "hyper_sdk.kasada",
-    "hyper_sdk.datadome",
+[build-system]
+requires = ["setuptools>=42", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "hyper_sdk"
+version = "2.13.0"
+description = "Hyper Solutions Python SDK"
+readme = "README.md"
+requires-python = ">=3.7"
+classifiers = [
+    "Intended Audience :: Developers",
+    "License :: OSI Approved :: MIT License",
+    "Programming Language :: Python :: 3",
+    "Framework :: AsyncIO",
+]
+dependencies = [
+    "httpx[http2]>=0.24.0",
+    "certifi>=2024.2.2",
+    "charset-normalizer>=3.3.2",
+    "idna>=3.6",
+    "jsonpickle>=3.0.3",
+    "PyJWT>=2.8.0",
+    "urllib3>=2.2.1"
+]
+license = {file = "LICENSE"}
+
+[project.urls]
+homepage = "https://github.com/Hyper-Solutions/hyper-sdk-py"
+
+[tool.setuptools]
+packages = [
+    "hyper_sdk",
+    "hyper_sdk.akamai",
+    "hyper_sdk.incapsula",
+    "hyper_sdk.kasada",
+    "hyper_sdk.datadome",
 ]