atomicshop 2.19.12__py3-none-any.whl → 2.19.14__py3-none-any.whl

atomicshop/__init__.py

@@ -1,4 +1,4 @@
 """Atomic Basic functions and classes to make developer life easier"""
 
 __author__ = "Den Kras"
-__version__ = '2.19.12'
+__version__ = '2.19.14'

atomicshop/basics/classes.py

@@ -10,6 +10,8 @@ from ..file_io.file_io import read_file
 
 
 """
+Using logger in the class only once during the import of the module.
+ 
 class ParserParent:
     # Initializing the logger in the "class variable" section will leave the instance of the logger initiated
     # and the rest of the instances of the class will use the same logger.
@@ -28,6 +30,58 @@ class ParserParent:
 """
 
 
+"""
+Using Base class for easier interfacing on subclasses.
+
+recognition/recognition_base.py:
+    from abc import abstractmethod
+    
+    
+    class Recognizer:
+        @abstractmethod
+        def recognize_vendor(self, file_path: str) -> str:
+            pass
+    
+        @abstractmethod
+        def recognize_family(self, bytes_list: list[str]]) -> str:
+            pass
+
+
+recognition/super_vendor.py:
+    from .recognition_base import Recognizer
+    
+    class SupervendorRecognizer(Recognizer):
+        def recognize_vendor(self, file_path: str) -> str:
+            classification_string: str = <Some logic to classify the SuperVendor>
+            return classification_string
+    
+        def recognize_family(self, bytes_list: list[str]]) -> str:
+            family_classification_string: str = <Some logic to classify the family of the SuperVendor>
+            return family_classification_string
+            
+
+main_script.py:
+    from . import recognition
+    from .recognition.recognition_base import Recognizer
+    
+    # Get the list of all the recognizers in the recognition package.
+    recognizers_list: list = classes.get_list_of_classes_in_module(
+        imported_package=recognition, imported_base_class=Recognizer)
+    
+    # Get the list of all the vendors from the file.
+    vendors_list: list = list()
+    for recognizer in recognizers_list:
+        recognizer_instance = recognizer()
+        vendor_name: str = recognizer_instance.recognize_vendor(file_object=file_path)
+        if vendor_name:
+            vendors_list.append((vendor_name, recognizer_instance))
+    
+    # Get the families of the vendors.
+    for vendor_name, recognizer_instance in vendors_list:
+        family_name: str = recognizer_instance.recognize_family(bytes_list=file_bytes_list)
+        print(f"Vendor: {vendor_name}, Family: {family_name}")
+"""
+
 def get_list_of_classes_in_module(
         imported_package,
         imported_base_class
@@ -66,6 +120,22 @@ def get_list_of_classes_in_module(
         # Get the list of classes
         unpacker_classes = get_list_of_classes_in_module(imported_package=unpackers, imported_base_class=Unpacker)
 
+        # Initialize the classes
+        for unpacker_class in unpacker_classes:
+            unpacker_instance = unpacker_class()
+            unpacker_instance.unpack("file_path")
+        ----------------------------
+        # You can also initialize the list of classes dynamically and after that execute methods.
+        # Example:
+        unpacker_classes = get_list_of_classes_in_module(imported_package=unpackers, imported_base_class=Unpacker)
+
+        instance_list: list = []
+        for unpacker_class in unpacker_classes:
+            instance_list.append(unpacker_class())
+
+        for instance in instance_list:
+            instance.unpack("file_path")
+
     :param imported_package:
     :param imported_base_class:
     :return:

atomicshop/process_poller/simple_process_pool.py

@@ -2,9 +2,11 @@ import threading
 from pathlib import Path
 import time
 import multiprocessing.managers
+import queue
+
 
 from ..wrappers.pywin32w.win_event_log.subscribes import process_create, process_terminate
-from .. import get_process_list, get_process_name_cmd_dll
+from .. import get_process_list
 from ..print_api import print_api
 
 
@@ -15,6 +17,10 @@ WAIT_FOR_PROCESS_POLLER_PID_SECONDS: int = 3
 WAIT_FOR_PROCESS_POLLER_PID_COUNTS: int = WAIT_FOR_PROCESS_POLLER_PID_SECONDS * 10
 
 
+class PidProcessConverterPIDNotFoundError(Exception):
+    pass
+
+
 class SimpleProcessPool:
     """
     THE POOL OF PROCESSES IS NOT REAL TIME!!!
@@ -106,14 +112,16 @@ class SimpleProcessPool:
 
         self._processes: dict = dict()
         self._running: bool = False
+        self.shared_dict_update_queue: queue.Queue = queue.Queue()
+        self.empty_cmdline_queue: queue.Queue = queue.Queue()
 
     def start(self):
         self._running = True
 
         self._processes = get_process_list.GetProcessList(
-            get_method='pywin32', connect_on_init=True).get_processes(as_dict=True)
+            get_method='psutil', connect_on_init=True).get_processes(as_dict=True)
 
-        thread_get_queue = threading.Thread(target=self._start_main_thread, args=(self.process_pool_shared_dict_proxy,))
+        thread_get_queue = threading.Thread(target=self._start_main_thread, args=(self.shared_dict_update_queue,))
         thread_get_queue.daemon = True
         thread_get_queue.start()
 
@@ -121,13 +129,21 @@ class SimpleProcessPool:
         thread_process_termination.daemon = True
         thread_process_termination.start()
 
+        thread_get_psutil_commandline = threading.Thread(target=self._thread_get_psutil_commandline)
+        thread_get_psutil_commandline.daemon = True
+        thread_get_psutil_commandline.start()
+
+        thread_update_shared_dict = threading.Thread(target=self._update_shared_dict, args=(self.shared_dict_update_queue,))
+        thread_update_shared_dict.daemon = True
+        thread_update_shared_dict.start()
+
     def stop(self):
         self._running = False
 
     def get_processes(self):
         return self._processes
 
-    def _start_main_thread(self, process_pool_shared_dict_proxy):
+    def _start_main_thread(self, shared_dict_update_queue):
         get_instance = process_create.ProcessCreateSubscriber()
         get_instance.start()
 
@@ -137,17 +153,35 @@ class SimpleProcessPool:
             process_name = Path(event['NewProcessName']).name
             command_line = event['CommandLine']
 
+            # The event log tracing method doesn't always give the command line, unlike the psutil method.
+            # So, we'll get the command line from the current psutil snapshot separately.
+            if command_line == '':
+                self.empty_cmdline_queue.put(process_id)
+
             self._processes[process_id] = {
                 'name': process_name,
                 'cmdline': command_line
             }
 
             # Update the multiprocessing shared dict proxy.
-            if process_pool_shared_dict_proxy is not None:
-                process_pool_shared_dict_proxy.clear()
-                process_pool_shared_dict_proxy.update(self._processes)
+            shared_dict_update_queue.put(dict(self._processes))
 
-            # print_api(f'Process [{process_id}] added to the pool.', color='blue')
+    def _thread_get_psutil_commandline(self):
+        """
+        This function will get an entry from the queue where command line is missing and get the command line
+        from the psutil snapshot.
+        """
+
+        while self._running:
+            empty_cmd_pid = self.empty_cmdline_queue.get()
+            current_psutil_process_snapshot: dict = get_process_list.GetProcessList(
+                get_method='psutil', connect_on_init=True).get_processes(as_dict=True)
+            command_line = current_psutil_process_snapshot[empty_cmd_pid]['cmdline']
+
+            self._processes[empty_cmd_pid]['cmdline'] = command_line
+
+            # Update the multiprocessing shared dict proxy.
+            self.shared_dict_update_queue.put(dict(self._processes))
 
     def _thread_process_termination(self):
         process_terminate_instance = process_terminate.ProcessTerminateSubscriber()
@@ -188,6 +222,15 @@ class SimpleProcessPool:
         _ = self._processes.pop(process_id, None)
         # print_api(f'Process [{process_id}] removed from the pool.', color='yellow')
 
+        self.shared_dict_update_queue.put(dict(self._processes))
+
+    def _update_shared_dict(self, shared_dict_update_queue):
+        while self._running:
+            current_process_pool = shared_dict_update_queue.get()
+            if self.process_pool_shared_dict_proxy is not None:
+                self.process_pool_shared_dict_proxy.clear()
+                self.process_pool_shared_dict_proxy.update(current_process_pool)
+
 
 class PidProcessConverter:
     """
@@ -204,7 +247,7 @@ class PidProcessConverter:
 
         self.process_pool_shared_dict_proxy: multiprocessing.managers.DictProxy = process_pool_shared_dict_proxy
 
-        self.get_process_with_dll_instance = get_process_name_cmd_dll.ProcessNameCmdline(load_dll=True)
+        self.get_process_with_psutil = get_process_list.GetProcessList(get_method='psutil', connect_on_init=True)
 
     def get_process_by_pid(self, pid: int):
         """
@@ -219,18 +262,20 @@ class PidProcessConverter:
         counter = 0
         process_dict: dict = dict()
         while counter < WAIT_FOR_PROCESS_POLLER_PID_COUNTS:
-            if pid not in self.process_pool_shared_dict_proxy:
+            # We need it so that the pool will not change in the middle of the process.
+            current_pid_pool = convert_proxy_dict_to_dict(self.process_pool_shared_dict_proxy)
+            if pid not in current_pid_pool:
                 # print(dict(self.process_pool_shared_dict_proxy))
                 time.sleep(0.1)
                 counter += 1
             else:
-                process_dict = self.process_pool_shared_dict_proxy[pid]
+                process_dict = current_pid_pool[pid]
                 break
 
-        if counter == WAIT_FOR_PROCESS_POLLER_PID_COUNTS and not process_dict:
-            print_api(f"Error: The PID [{pid}] is not in the pool, trying DLL snapshot.", color='yellow')
+        if not process_dict:
+            print_api(f"Error: The PID [{pid}] is not in the pool, trying psutil snapshot.", color='yellow')
             # Last resort, try to get the process name by current process snapshot.
-            processes = self.get_process_with_dll_instance.get_process_details(as_dict=True)
+            processes = self.get_process_with_psutil.get_processes(as_dict=True)
             if pid not in processes:
                 print_api(f"Error: Couldn't get the process name for PID: {pid}.", color='red')
                 process_dict = {
@@ -242,3 +287,25 @@ class PidProcessConverter:
 
         return process_dict
 
+
+def convert_proxy_dict_to_dict(proxy_dict: multiprocessing.managers.DictProxy) -> dict:
+    """
+    Convert the multiprocessing shared dict proxy to a normal dict.
+
+    :param proxy_dict: multiprocessing.managers.DictProxy, the shared dict proxy.
+    :return: dict, the normal dict.
+    """
+
+    # Create a snapshot of the keys
+    keys = list(proxy_dict.keys())
+    current_pid_pool = {}
+
+    for key in keys:
+        try:
+            # Attempt to retrieve the value for each key
+            current_pid_pool[key] = proxy_dict[key]
+        except KeyError:
+            # The key was removed concurrently; skip it
+            continue
+
+    return current_pid_pool

atomicshop/wrappers/playwrightw/scenarios.py

@@ -207,7 +207,9 @@ def _fetch_content(
             'playwright_text',
             'js_text',
             'playwright_html',
+            'playwright_html_to_text',
             'js_html',
+            'js_html_to_text',
             'playwright_copypaste'
         ],
         headless: bool = True):
@@ -260,18 +262,12 @@ def _fetch_content(
         elif text_fetch_method == "js_text":
             # Use JavaScript to extract only the visible text from the page
             text_content: str = page.evaluate("document.body.innerText")
-        elif text_fetch_method == "playwright_html":
+        elif "playwright_html" in text_fetch_method:
             # Get the full HTML content of the page
-            html = page.content()
-            # Parse the HTML using BeautifulSoup and extract the text
-            soup = BeautifulSoup(html, 'html.parser')
-            text_content = soup.get_text()
-        elif text_fetch_method == "js_html":
+            text_content = page.content()
+        elif "js_html" in text_fetch_method:
             # Use JavaScript to extract the full text from the page
-            html = page.evaluate('document.documentElement.outerHTML')
-            # Parse the HTML using BeautifulSoup and extract the text
-            soup = BeautifulSoup(html, 'html.parser')
-            text_content = soup.get_text()
+            text_content = page.evaluate('document.documentElement.outerHTML')
         elif text_fetch_method == "playwright_copypaste":
             # Focus the page and simulate Ctrl+A and Ctrl+C
             page.keyboard.press("Control+a")  # Select all text
@@ -281,6 +277,11 @@ def _fetch_content(
         else:
             raise ValueError(f"Invalid text_fetch_method: {text_fetch_method}")
 
+        if "to_text" in text_fetch_method:
+            # Convert HTML to plain text using BeautifulSoup
+            soup = BeautifulSoup(text_content, "html.parser")
+            text_content = soup.get_text()
+
         # text = page.evaluate('document.body.textContent')
         # text = page.eval_on_selector('body', 'element => element.innerText')
         # text = page.eval_on_selector('body', 'element => element.textContent')

{atomicshop-2.19.12.dist-info → atomicshop-2.19.14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: atomicshop
-Version: 2.19.12
+Version: 2.19.14
 Summary: Atomic functions and classes to make developer life easier
 Author: Denis Kras
 License: MIT License

{atomicshop-2.19.12.dist-info → atomicshop-2.19.14.dist-info}/RECORD

@@ -1,4 +1,4 @@
-atomicshop/__init__.py,sha256=XtGdtCQBXBDdG8_6wWA54tuzPJ5IE_hCw_e_N2FoXLM,124
+atomicshop/__init__.py,sha256=jnzkSiVKr7rNaRPe51te8qpwDBNj2V37gzYMQVqPkoQ,124
 atomicshop/_basics_temp.py,sha256=6cu2dd6r2dLrd1BRNcVDKTHlsHs_26Gpw8QS6v32lQ0,3699
 atomicshop/_create_pdf_demo.py,sha256=Yi-PGZuMg0RKvQmLqVeLIZYadqEZwUm-4A9JxBl_vYA,3713
 atomicshop/_patch_import.py,sha256=ENp55sKVJ0e6-4lBvZnpz9PQCt3Otbur7F6aXDlyje4,6334
@@ -92,7 +92,7 @@ atomicshop/basics/ansi_escape_codes.py,sha256=uGVRW01v2O052701sOfAdCaM_GLF_cu_jr
 atomicshop/basics/argparse_template.py,sha256=horwgSf3MX1ZgRnYxtmmQuz9OU_vKrKggF65gmjlmfg,5836
 atomicshop/basics/booleans.py,sha256=V36NaMf8AffhCom_ovQeOZlYcdtGyIcQwWKki6h7O0M,1745
 atomicshop/basics/bytes_arrays.py,sha256=xfFW9CBQyzf0iXNWpx-EfrxtN3-tiKzuczPzOb6cOdU,7190
-atomicshop/basics/classes.py,sha256=UayCzPs3eynI3wOzSu-2IJSmTwOB4HwPwgVI2F-7_lQ,12648
+atomicshop/basics/classes.py,sha256=dBSQktjc1GmphpD_qOGYNDl-mQcmwg8OkhSqtXPfHEU,15301
 atomicshop/basics/dicts.py,sha256=DeYHIh940pMMBrFhpXt4dsigFVYzTrlqWymNo4Pq_Js,14049
 atomicshop/basics/dicts_nested.py,sha256=StYxYnYPa0SEJr1lmEwAv5zfERWWqoULeyG8e0zRAwE,4107
 atomicshop/basics/enumerations.py,sha256=41VVQYh_vnVapggxKg2IRU5e_EiMpZzX1n1mtxvoSzM,1364
@@ -170,7 +170,7 @@ atomicshop/permissions/ubuntu_permissions.py,sha256=n8z1vcIXDts4zLVue33dtJiTopjg
 atomicshop/permissions/win_permissions.py,sha256=eDQm1jfK9x_hkbLqIJjFTwfqinAWQ0iSr0kW3XrF1BE,1272
 atomicshop/process_poller/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 atomicshop/process_poller/process_pool.py,sha256=4Qs427qd7OcBxu5PMFU5PTmyuxRy0vgj2GLsRt0IoEw,9565
-atomicshop/process_poller/simple_process_pool.py,sha256=E0w66imLQQ0phHuTQfTaApOTurX0duE9vw1trjUY7rI,9894
+atomicshop/process_poller/simple_process_pool.py,sha256=ap9nCywHZ-eta7EHe0mTYS45-zYvMjgvfESXAuG0Wi4,12578
 atomicshop/process_poller/tracer_base.py,sha256=IOiHcnmF-MccOSCErixN5mve9RifZ9cPnGVHCIRchrs,1091
 atomicshop/process_poller/pollers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 atomicshop/process_poller/pollers/psutil_pywin32wmi_dll.py,sha256=XRRfOIy62iOYU8IKRcyECWiL0rqQ35DeYbPsv_SHDVM,4510
@@ -284,7 +284,7 @@ atomicshop/wrappers/playwrightw/javascript.py,sha256=_bW7CAtm0Y8IHYrAalg5HpPFnk6
 atomicshop/wrappers/playwrightw/keyboard.py,sha256=zN3YddGO-qUkn6C0CRVFejP4cTuaUwXLDNFhFREjERY,422
 atomicshop/wrappers/playwrightw/locators.py,sha256=6wsLywZxDuii7mwv-zQsRbqQC8r7j96Bma5b5_7ZoVo,2411
 atomicshop/wrappers/playwrightw/mouse.py,sha256=-2FZbQtjgH7tdXWld6ZPGqlKFUdf5in0ujN0hewxa50,656
-atomicshop/wrappers/playwrightw/scenarios.py,sha256=Xvl1jUmQhd4l0MmOUgQKfgGleblIyE-qC3wuoyx16tU,11531
+atomicshop/wrappers/playwrightw/scenarios.py,sha256=oLrZiWnIc09I0Zvh3IxPVDPsQ-Kbr4QM0TeyXGvlsaY,11494
 atomicshop/wrappers/playwrightw/waits.py,sha256=PBFdz_PoM7Fo7O8hLqMrxNPzBEYgPoXwZceFFCGGeu8,7182
 atomicshop/wrappers/psutilw/cpus.py,sha256=w6LPBMINqS-T_X8vzdYkLS2Wzuve28Ydp_GafTCngrc,236
 atomicshop/wrappers/psutilw/disks.py,sha256=3ZSVoommKH1TWo37j_83frB-NqXF4Nf5q5mBCX8G4jE,9221
@@ -329,8 +329,8 @@ atomicshop/wrappers/socketw/statistics_csv.py,sha256=fgMzDXI0cybwUEqAxprRmY3lqbh
 atomicshop/wrappers/winregw/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 atomicshop/wrappers/winregw/winreg_installed_software.py,sha256=Qzmyktvob1qp6Tjk2DjLfAqr_yXV0sgWzdMW_9kwNjY,2345
 atomicshop/wrappers/winregw/winreg_network.py,sha256=AENV88H1qDidrcpyM9OwEZxX5svfi-Jb4N6FkS1xtqA,8851
-atomicshop-2.19.12.dist-info/LICENSE.txt,sha256=lLU7EYycfYcK2NR_1gfnhnRC8b8ccOTElACYplgZN88,1094
-atomicshop-2.19.12.dist-info/METADATA,sha256=IjBkWX6cc09byHYWcjX8hdW_eEmXddbKu1h0hNYUybM,10631
-atomicshop-2.19.12.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-atomicshop-2.19.12.dist-info/top_level.txt,sha256=EgKJB-7xcrAPeqTRF2laD_Np2gNGYkJkd4OyXqpJphA,11
-atomicshop-2.19.12.dist-info/RECORD,,
+atomicshop-2.19.14.dist-info/LICENSE.txt,sha256=lLU7EYycfYcK2NR_1gfnhnRC8b8ccOTElACYplgZN88,1094
+atomicshop-2.19.14.dist-info/METADATA,sha256=1ZQp-Wt5osHkFGnEavFkPEC41-jmfqJsSsktj0T7aBk,10631
+atomicshop-2.19.14.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+atomicshop-2.19.14.dist-info/top_level.txt,sha256=EgKJB-7xcrAPeqTRF2laD_Np2gNGYkJkd4OyXqpJphA,11
+atomicshop-2.19.14.dist-info/RECORD,,