simplex 1.0.0__tar.gz → 1.2.0__tar.gz

{simplex-1.0.0 → simplex-1.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: simplex
-Version: 1.0.0
+Version: 1.2.0
 Summary: Official Python SDK for Simplex API
 Home-page: https://github.com/shreyka/simplex-python
 Author: Simplex Labs, Inc.
@@ -21,6 +21,15 @@ Requires-Dist: rich>=13.0.0
 Requires-Dist: prompt_toolkit>=3.0.0
 Requires-Dist: playwright>=1.0.0
 Requires-Dist: Pillow>=9.0.0
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # Simplex AI Python SDK
 

{simplex-1.0.0 → simplex-1.2.0}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="simplex",
-    version="1.0.0",
+    version="1.2.0",
     packages=find_packages(),
     install_requires=[
         "openai>=1.0.0",

{simplex-1.0.0 → simplex-1.2.0}/simplex/simplex.py

@@ -1,42 +1,35 @@
-from playwright.sync_api import Page, sync_playwright
+from playwright.sync_api import Page, sync_playwright, Browser
 from PIL import Image
 import requests
 from typing import List
 import io
+import time
 
 from .utils import center_bbox, screenshot_to_image
 
 BASE_URL = "https://u3mvtbirxf.us-east-1.awsapprunner.com"
 
 class Simplex:
-    def __init__(self, api_key: str, driver: Page = None):
+    def __init__(self, api_key: str, browser: Browser = None):
         """
         Initialize Simplex instance
         
         Args:
             api_key (str): API key for authentication
-            driver (playwright.sync_api.Page, optional): Playwright page object. If not provided, 
-                                                       a new headless browser instance will be created.
+            browser (optional): Browser instance (for Hyperbrowser integration)
         """
         self.api_key = api_key
+        self.browser = browser
         
-        if driver is None:
+        if browser is None:
             self.playwright = sync_playwright().start()
             self.browser = self.playwright.chromium.launch(headless=True)
-            self.driver = self.browser.new_page()
         else:
-            self.driver = driver
-            self.browser = None
             self.playwright = None
+            
+        self.driver = self.browser.new_page(viewport={"width": 1280, "height": 720})
 
-    def __del__(self):
-        """Cleanup Playwright resources"""
-        if self.browser:
-            self.browser.close()
-        if self.playwright:
-            self.playwright.stop()
-
-    def find_element(self, element_description: str, state: Image.Image | None = None) -> List[int]:
+    def find_element(self, element_description: str, state: Image.Image | None = None, annotate: bool = True) -> List[int]:
         """
         Find an element in the screenshot using the element description
 
@@ -69,11 +62,41 @@ class Simplex:
             files=files
         )
 
+
         # Print the results
         print(f"Status Code: {response.status_code}")
         if response.status_code == 200:
             res = response.json()
             bbox = [int(res['x1']), int(res['y1']), int(res['x2']), int(res['y2'])]
+
+            # Add overlay directly to the page if driver exists
+            if hasattr(self, 'driver') and annotate:
+                # Create and inject overlay element
+                self.driver.evaluate("""
+                    (bbox) => {
+                        // Remove any existing overlay
+                        const existingOverlay = document.getElementById('simplex-bbox-overlay');
+                        if (existingOverlay) {
+                            existingOverlay.remove();
+                        }
+                        
+                        // Create new overlay
+                        const overlay = document.createElement('div');
+                        overlay.id = 'simplex-bbox-overlay';
+                        overlay.style.position = 'absolute';
+                        overlay.style.border = '2px solid red';
+                        overlay.style.left = bbox[0] + 'px';
+                        overlay.style.top = bbox[1] + 'px';
+                        overlay.style.width = (bbox[2] - bbox[0]) + 'px';
+                        overlay.style.height = (bbox[3] - bbox[1]) + 'px';
+                        overlay.style.pointerEvents = 'none';
+                        overlay.style.zIndex = '10000';
+                        
+                        document.body.appendChild(overlay);
+                    }
+                """, bbox)
+                self.driver.wait_for_selector('#simplex-bbox-overlay')
+                time.sleep(5)
             return bbox 
         else:
             print("Error:", response.text)
@@ -130,7 +153,7 @@ class Simplex:
         """
         self.driver.goto(url)
 
-    def execute_action(self, action: List[List[str]], state: Image.Image | None = None) -> None:
+    def execute_action(self, action: List[List[str]], state: Image.Image | None = None, annotate: bool = True) -> None:
         """
         Execute an action with playwright driver
 
@@ -143,12 +166,12 @@ class Simplex:
 
         try:
             if action_type == "CLICK":
-                bbox = self.find_element(description, state)
+                bbox = self.find_element(description, state, annotate=annotate)
                 center_x, center_y = center_bbox(bbox)
                 self.driver.mouse.click(center_x, center_y)
 
             elif action_type == "HOVER":
-                bbox = self.find_element(description, state)
+                bbox = self.find_element(description, state, annotate=annotate)
                 center_x, center_y = center_bbox(bbox)
                 self.driver.mouse.move(center_x, center_y)
 
@@ -165,14 +188,14 @@ class Simplex:
             print(f"Error executing action: {e}")
             return None
         
-    def do(self, step_description: str) -> None:
+    def do(self, step_description: str, annotate: bool = True) -> None:
         """
         Execute a step description
         """
         state = self.take_stable_screenshot()
         actions = self.step_to_action(step_description, state)
         for action in actions:
-            self.execute_action(action)
+            self.execute_action(action, annotate=annotate)
             
     def take_stable_screenshot(self) -> Image.Image:
         """

{simplex-1.0.0 → simplex-1.2.0}/simplex.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: simplex
-Version: 1.0.0
+Version: 1.2.0
 Summary: Official Python SDK for Simplex API
 Home-page: https://github.com/shreyka/simplex-python
 Author: Simplex Labs, Inc.
@@ -21,6 +21,15 @@ Requires-Dist: rich>=13.0.0
 Requires-Dist: prompt_toolkit>=3.0.0
 Requires-Dist: playwright>=1.0.0
 Requires-Dist: Pillow>=9.0.0
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: home-page
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # Simplex AI Python SDK
 

simplex-1.2.0/tests/test_local.py

@@ -0,0 +1,206 @@
+import sys
+import os
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+from simplex import Simplex
+
+from PIL import ImageDraw
+
+from playwright.sync_api import sync_playwright
+from PIL import Image
+import time
+import os
+from hyperbrowser import Hyperbrowser
+
+from dotenv import load_dotenv
+
+
+load_dotenv()
+
+
+def screenshot_tests():
+    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"))
+    image = "/home/ubuntu/supreme-waffle/images/netflix.png"
+    screenshot = Image.open(image)
+
+    start_time = time.time()
+    bbox = simplex.find_element("dark mode icon", screenshot)
+    end_time = time.time()
+    print(f"Time taken: {end_time - start_time} seconds")
+    print(bbox)
+
+    start_time = time.time()
+    action = simplex.step_to_action("click and enter email address", screenshot)
+    end_time = time.time()
+    print(f"Time taken: {end_time - start_time} seconds")
+    print(action)
+
+
+def execute_action_test():
+    playwright = sync_playwright().start()
+    browser = playwright.chromium.launch(headless=False)
+    driver = browser.new_page()
+    
+    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), driver=driver)
+    simplex.goto("https://www.netflix.com/")
+    actions = [['CLICK', 'email field'], ['TYPE', 'email address']]
+    simplex.execute_action(actions[0])
+
+
+def cgtrader_test():
+    assets = ["apple watch"]
+    urls = []
+
+    with sync_playwright() as p:
+        driver = p.chromium.launch(headless=False).new_page()
+        simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), driver=driver)
+        simplex.goto("https://www.cgtrader.com/")
+
+        for asset in assets:
+            simplex.goto("https://www.cgtrader.com")
+            simplex.do(f"search for {asset}")  
+            simplex.do("click on search button")
+            simplex.do(f"click on the first product")
+            driver.wait_for_timeout(3000)
+
+            urls.append(simplex.driver.url)
+
+    print(urls)
+
+
+def test_find_element(): 
+    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"))
+    simplex.goto("https://www.cgtrader.com/")
+
+    state = simplex.take_stable_screenshot()
+    bbox = simplex.find_element("search bar")
+
+    copy_image = state.copy()
+    draw = ImageDraw.Draw(copy_image)
+    draw.rectangle(bbox, outline='red', width=2)
+    copy_image.save("annotated_state.png")
+    
+
+    # Get the page HTML and device scale factor
+    html = simplex.driver.content()
+    scale_factor = simplex.driver.evaluate("window.devicePixelRatio")
+    
+    # Get viewport dimensions and other relevant settings
+    viewport_size = simplex.driver.viewport_size
+    zoom_level = simplex.driver.evaluate("document.documentElement.style.zoom || 1")
+    
+    # Debug print
+    print(f"Original bbox: {bbox}")
+    print(f"Scale factor: {scale_factor}")
+    print(f"Viewport size: {viewport_size}")
+    print(f"Zoom level: {zoom_level}")
+    
+    # Transform coordinates from screenshot to HTML
+    html_bbox = [
+        bbox[0] / scale_factor,
+        bbox[1] / scale_factor,
+        bbox[2] / scale_factor,
+        bbox[3] / scale_factor
+    ]
+    print(f"Transformed bbox: {html_bbox}")
+    
+    # Create HTML wrapper with matching viewport settings and scaled overlay
+    html_with_settings = f"""
+    <!DOCTYPE html>
+    <html>
+    <head>
+        <meta name="viewport" content="width=device-width, initial-scale=1">
+        <style>
+            body {{
+                margin: 0;
+                width: {viewport_size['width']}px;
+                height: {viewport_size['height']}px;
+                zoom: {zoom_level};
+            }}
+            .viewport-container {{
+                width: 100%;
+                height: 100%;
+                overflow: hidden;
+                position: relative;
+                transform-origin: top left;
+                transform: scale({1/scale_factor});
+            }}
+            #page-content {{
+                width: {viewport_size['width'] * scale_factor}px;
+                height: {viewport_size['height'] * scale_factor}px;
+                transform-origin: top left;
+            }}
+            #bbox-overlay {{
+                position: absolute;
+                border: 2px solid red;
+                left: {bbox[0]}px;
+                top: {bbox[1]}px;
+                width: {bbox[2] - bbox[0]}px;
+                height: {bbox[3] - bbox[1]}px;
+                pointer-events: none;
+                z-index: 10000;
+            }}
+        </style>
+    </head>
+    <body>
+        <div class="viewport-container">
+            <div id="page-content">
+                {html}
+            </div>
+            <div id="bbox-overlay"></div>
+        </div>
+    </body>
+    </html>
+    """
+
+    # Save the HTML with viewport settings
+    with open('screenshot_with_viewport.html', 'w') as f:
+        f.write(html_with_settings)
+
+
+def test_find_element_2():
+    with sync_playwright() as p:
+        driver = p.chromium.launch(headless=False).new_page(viewport={"width": 1920, "height": 1080})
+        simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), driver=driver)
+        simplex.goto("https://www.cgtrader.com/")
+
+        state = simplex.take_stable_screenshot()
+        bbox = simplex.find_element("search bar")
+
+        print(bbox)
+
+def test_hyperbrowser_integration():
+    """Test Simplex integration with Hyperbrowser"""
+
+    # Initialize Hyperbrowser client
+    client = Hyperbrowser(api_key=os.getenv("HYPERBROWSER_API_KEY"))
+    
+    # Create a new session
+    session = client.sessions.create()
+    ws_endpoint = session.ws_endpoint
+
+    try:
+        with sync_playwright() as p:
+            # Connect browser to Hyperbrowser session
+            browser = p.chromium.connect_over_cdp(ws_endpoint)
+            context = browser.new_context()
+            page = context.new_page()
+            
+            # Initialize Simplex with the Hyperbrowser-connected page
+            simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), browser=browser)
+            
+            # Test basic functionality
+            simplex.goto("https://www.cgtrader.com/")
+            simplex.do("search for iphone")
+            
+            # Verify the search worked by finding the search results
+            bbox = simplex.find_element("search results")
+            assert bbox is not None, "Search results not found"
+            
+    finally:
+        # Always stop the Hyperbrowser session
+        client.sessions.stop(session.id)
+
+if __name__ == "__main__":
+    test_hyperbrowser_integration()
+    
+    

simplex-1.0.0/tests/test_local.py

@@ -1,111 +0,0 @@
-import sys
-import os
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-from simplex import Simplex
-
-from PIL import ImageDraw
-
-from playwright.sync_api import sync_playwright
-from PIL import Image
-import time
-import tempfile
-import webbrowser
-
-from dotenv import load_dotenv
-
-import io
-
-
-load_dotenv()
-
-
-def screenshot_tests():
-    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"))
-    image = "/home/ubuntu/supreme-waffle/images/netflix.png"
-    screenshot = Image.open(image)
-
-    start_time = time.time()
-    bbox = simplex.find_element("dark mode icon", screenshot)
-    end_time = time.time()
-    print(f"Time taken: {end_time - start_time} seconds")
-    print(bbox)
-
-    start_time = time.time()
-    action = simplex.step_to_action("click and enter email address", screenshot)
-    end_time = time.time()
-    print(f"Time taken: {end_time - start_time} seconds")
-    print(action)
-
-
-def execute_action_test():
-    playwright = sync_playwright().start()
-    browser = playwright.chromium.launch(headless=False)
-    driver = browser.new_page()
-    
-    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), driver=driver)
-    simplex.goto("https://www.netflix.com/")
-    actions = [['CLICK', 'email field'], ['TYPE', 'email address']]
-    simplex.execute_action(actions[0])
-
-
-def cgtrader_test():
-    assets = ["apple watch"]
-    urls = []
-
-    with sync_playwright() as p:
-        driver = p.chromium.launch(headless=False).new_page()
-        simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"), driver=driver)
-        simplex.goto("https://www.cgtrader.com/")
-
-        for asset in assets:
-            simplex.goto("https://www.cgtrader.com")
-            simplex.do(f"search for {asset}")  
-            simplex.do("click on search button")
-            simplex.do(f"click on the first product")
-            driver.wait_for_timeout(3000)
-
-            urls.append(simplex.driver.url)
-
-    print(urls)
-
-
-def test_find_element(): 
-    simplex = Simplex(api_key=os.getenv("SIMPLEX_API_KEY"))
-    simplex.goto("https://www.cgtrader.com/")
-
-    state = simplex.take_stable_screenshot()
-    bbox = simplex.find_element("cart")
-    
-    # Save the screenshot without annotation
-    state.save('screenshot_with_bbox.png')
-
-    # Create a minimal HTML file with just the visualization and bbox
-    with open('screenshot_with_bbox.html', 'w') as f:
-        f.write(f"""
-        <div style="position: relative; display: inline-block;">
-            <img src="screenshot_with_bbox.png">
-            <div style="
-                position: absolute;
-                border: 2px solid red;
-                left: {bbox[0]}px;
-                top: {bbox[1]}px;
-                width: {bbox[2] - bbox[0]}px;
-                height: {bbox[3] - bbox[1]}px;
-                pointer-events: none;
-            "></div>
-        </div>
-        """)
-
-    # Create a second HTML file with just the screenshot
-    with open('screenshot.html', 'w') as f:
-        f.write("""
-        <div style="position: relative; display: inline-block;">
-            <img src="screenshot_with_bbox.png">
-        </div>
-        """)
-
-
-if __name__ == "__main__":
-    cgtrader_test()
-    
-