Merge pull request #3969 from seleniumbase/pdf-improvements

PDF improvements

Author: mdmintz

examples/cdp_mode/raw_print_to_pdf.py

@@ -1,12 +1,10 @@
-import base64
 from seleniumbase import SB
-from selenium.webdriver.common.print_page_options import PrintOptions
 
-with SB(uc=True, test=True, ad_block=True) as sb:
+with SB(uc=True, test=True, pls="none") as sb:
     url = "https://seleniumbase.io"
     sb.activate_cdp_mode(url)
-    sb.reconnect()  # To access WebDriver methods
-    print_options = PrintOptions()
-    pdf_base64 = sb.driver.print_page(print_options)
-    with open("downloaded_files/sb.pdf", "wb") as f:
-        f.write(base64.b64decode(pdf_base64))
+    sb.assert_title("SeleniumBase Docs")
+    file_path = "downloaded_files/sb.pdf"
+    sb.print_to_pdf(file_path)
+    sb.assert_downloaded_file("sb.pdf")
+    sb.assert_pdf_text(file_path, "SeleniumBase")

help_docs/method_summary.md

@@ -341,6 +341,7 @@ self.save_file_as(file_url, new_file_name, destination_folder=None)
 self.save_data_as(data, file_name, destination_folder=None)
 self.append_data_to_file(data, file_name, destination_folder=None)
 self.get_file_data(file_name, folder=None)
+self.print_to_pdf(name, folder=None)
 self.get_downloads_folder()
 self.get_browser_downloads_folder()
 self.get_downloaded_files(regex=None, browser=False)

requirements.txt

@@ -13,7 +13,7 @@ filelock~=3.16.1;python_version<"3.9"
 filelock>=3.19.1;python_version>="3.9"
 fasteners>=0.20
 mycdp>=1.2.0
-pynose>=1.5.4
+pynose>=1.5.5
 platformdirs>=4.3.6;python_version<"3.9"
 platformdirs>=4.4.0;python_version>="3.9"
 typing-extensions>=4.13.2
@@ -36,15 +36,15 @@ charset-normalizer>=3.4.3,<4
 urllib3>=1.26.20,<2;python_version<"3.10"
 urllib3>=1.26.20,<2.6.0;python_version>="3.10"
 requests==2.32.4;python_version<"3.9"
-requests>=2.32.5,<2.33;python_version>="3.9"
+requests~=2.32.5;python_version>="3.9"
 sniffio==1.3.1
 h11==0.16.0
 outcome==1.3.0.post0
 trio==0.27.0;python_version<"3.9"
-trio==0.30.0;python_version>="3.9"
-trio-websocket==0.12.2
+trio~=0.30.0;python_version>="3.9"
+trio-websocket~=0.12.2
 wsproto==1.2.0
-websocket-client==1.8.0
+websocket-client~=1.8.0
 selenium==4.27.1;python_version<"3.9"
 selenium==4.32.0;python_version>="3.9" and python_version<"3.10"
 selenium==4.35.0;python_version>="3.10"
@@ -82,7 +82,7 @@ rich>=14.1.0,<15
 coverage>=7.6.1;python_version<"3.9"
 coverage>=7.10.6;python_version>="3.9"
 pytest-cov>=5.0.0;python_version<"3.9"
-pytest-cov>=6.3.0;python_version>="3.9"
+pytest-cov>=7.0.0;python_version>="3.9"
 flake8==5.0.4;python_version<"3.9"
 flake8==7.3.0;python_version>="3.9"
 mccabe==0.7.0

seleniumbase/__version__.py

@@ -1,2 +1,2 @@
 # seleniumbase package
-__version__ = "4.41.5"
+__version__ = "4.41.6"

seleniumbase/fixtures/base_case.py

@@ -7245,6 +7245,7 @@ def get_pdf_text(
             page_search = [page]
         else:
             page_search = None
+        logging.getLogger("pdfminer").setLevel(logging.ERROR)
         pdf_text = extract_text(
             file_path,
             password="",
@@ -7552,6 +7553,47 @@ def get_file_data(self, file_name, folder=None):
             folder = constants.Files.DOWNLOADS_FOLDER
         return page_utils._get_file_data(folder, file_name)
 
+    def print_to_pdf(self, name, folder=None):
+        """Saves the current page as a PDF.
+        If no folder is specified, uses the folder where pytest was called.
+        If the folder provided doesn't exist, it will get created.
+        @Params
+        name - The name to give the PDF file. Must end in ".pdf".
+        folder - The directory where you want to save the PDF."""
+        import base64
+        from selenium.webdriver.common.print_page_options import PrintOptions
+
+        if not name.lower().endswith(".pdf"):
+            raise Exception('PDF name {%s} must end in ".pdf"!)' % name)
+        download_file_lock = fasteners.InterProcessLock(
+            constants.MultiBrowser.DOWNLOAD_FILE_LOCK
+        )
+        if self.__is_cdp_swap_needed():
+            with download_file_lock:
+                with suppress(Exception):
+                    shared_utils.make_writable(
+                        constants.MultiBrowser.DOWNLOAD_FILE_LOCK
+                    )
+                if folder and not os.path.exists(folder):
+                    os.makedirs(folder)
+                self.cdp.print_to_pdf(name, folder)
+            return
+        self.wait_for_ready_state_complete()
+        print_options = PrintOptions()
+        pdf_base64 = self.driver.print_page(print_options)
+        with download_file_lock:
+            with suppress(Exception):
+                shared_utils.make_writable(
+                    constants.MultiBrowser.DOWNLOAD_FILE_LOCK
+                )
+            if folder and not os.path.exists(folder):
+                os.makedirs(folder)
+            filename = name
+            if folder:
+                filename = os.path.join(folder, name)
+            with open(filename, "wb") as f:
+                f.write(base64.b64decode(pdf_base64))
+
     def get_downloads_folder(self):
         """Returns the path of the SeleniumBase "downloaded_files/" folder.
         Calling self.download_file(file_url) will put that file in here.

setup.py

@@ -161,7 +161,7 @@
         'filelock>=3.19.1;python_version>="3.9"',
         'fasteners>=0.20',
         "mycdp>=1.2.0",
-        "pynose>=1.5.4",
+        "pynose>=1.5.5",
         'platformdirs>=4.3.6;python_version<"3.9"',
         'platformdirs>=4.4.0;python_version>="3.9"',
         'typing-extensions>=4.13.2',
@@ -184,15 +184,15 @@
         'urllib3>=1.26.20,<2;python_version<"3.10"',
         'urllib3>=1.26.20,<2.6.0;python_version>="3.10"',
         'requests==2.32.4;python_version<"3.9"',
-        'requests>=2.32.5,<2.33;python_version>="3.9"',
+        'requests~=2.32.5;python_version>="3.9"',
         'sniffio==1.3.1',
         'h11==0.16.0',
         'outcome==1.3.0.post0',
         'trio==0.27.0;python_version<"3.9"',
-        'trio==0.30.0;python_version>="3.9"',
-        'trio-websocket==0.12.2',
+        'trio~=0.30.0;python_version>="3.9"',
+        'trio-websocket~=0.12.2',
         'wsproto==1.2.0',
-        'websocket-client==1.8.0',
+        'websocket-client~=1.8.0',
         'selenium==4.27.1;python_version<"3.9"',
         'selenium==4.32.0;python_version>="3.9" and python_version<"3.10"',
         'selenium==4.35.0;python_version>="3.10"',
@@ -239,7 +239,7 @@
             'coverage>=7.6.1;python_version<"3.9"',
             'coverage>=7.10.6;python_version>="3.9"',
             'pytest-cov>=5.0.0;python_version<"3.9"',
-            'pytest-cov>=6.3.0;python_version>="3.9"',
+            'pytest-cov>=7.0.0;python_version>="3.9"',
         ],
         # pip install -e .[flake8]
         # Usage: flake8
@@ -271,8 +271,10 @@
             'pdfminer.six==20250506;python_version>="3.9"',
             'cryptography==39.0.2;python_version<"3.9"',
             'cryptography==45.0.7;python_version>="3.9"',
-            'cffi==1.17.1',
-            "pycparser==2.22",
+            'cffi==1.17.1;python_version<"3.9"',
+            'cffi==2.0.0;python_version>="3.9"',
+            'pycparser==2.22;python_version<"3.9"',
+            'pycparser==2.23;python_version>="3.9"',
         ],
         # pip install -e .[pillow]
         # (An optional library for image-processing.)