abstract-webtools 0.1.6.154__py3-none-any.whl → 0.1.6.156__py3-none-any.whl

abstract_webtools/__init__.py

@@ -1,5 +1,5 @@
 from .abstract_webtools import *
-from .url_grabber import url_grabber_component
+
 from .abstract_webtools import get_url_mgr
 from .managers import *
 from .abstract_usurpit import usurpManager,usurpit

{abstract_webtools-0.1.6.154.dist-info → abstract_webtools-0.1.6.156.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: abstract_webtools
-Version: 0.1.6.154
+Version: 0.1.6.156
 Summary: Utilities for fetching/parsing web content with requests/urllib3/BS4 and helpers.
 Home-page: https://github.com/AbstractEndeavors/abstract_webtools
 Author: putkoff

{abstract_webtools-0.1.6.154.dist-info → abstract_webtools-0.1.6.156.dist-info}/RECORD

@@ -1,4 +1,4 @@
-abstract_webtools/__init__.py,sha256=yzVSyqPwhUPBijfAj3AMl9GgewR_jvVNgFEKwkXV0TY,198
+abstract_webtools/__init__.py,sha256=P2eRTWMkZPqdrX3K5CjVoYZE0kA9xxYc74K8VVJi5Fo,152
 abstract_webtools/abstract_usurpit.py,sha256=2idbYXLFhXh8VPfdYgWICNH8dehnZRCdt4U5sTsVxo4,9663
 abstract_webtools/abstract_webtools.py,sha256=rBSuzveYUtXugRcsEMrXCRbuMps7ntFeJG86ZHvw_-k,91587
 abstract_webtools/big_user_agent_list.py,sha256=5ZkrUWmfzYL5yaULREslh9ZiRQeITbSjqZlp2KQON3w,131923
@@ -8,7 +8,6 @@ abstract_webtools/find_dirs.py,sha256=BlE4ruzMABqmv03NcutZ1j5N3pCc-Q4uNEAMpNolZC
 abstract_webtools/k2s_downloader.py,sha256=t0tCKAfDNQGn9tKh3eg0XVU0bY-MmYITwJa3ANf7090,6988
 abstract_webtools/main.py,sha256=_I7pPXPkoLZOoYGLQDrSLGhGuQt6-PVyXEHZSmglk2g,1329
 abstract_webtools/soup_gui.py,sha256=n95YAps1R6DpMwR4UbthSqQby0C5WHUa9tsW-f2qpLg,5184
-abstract_webtools/url_grabber.py,sha256=FAry0gmZH4Iy646mYq9yKN4hh5ME6MiTr_AmP5WAR90,10440
 abstract_webtools/url_grabber_new.py,sha256=xb23qo4anOY0Ax3CAfaHJ8s5VEz61Sinh-XpEDFW7Is,3621
 abstract_webtools/managers/__init__.py,sha256=RXQAK5z9nYlocM91P2OC4jR352-MiqT5bAi4xZl7_FU,470
 abstract_webtools/managers/allss.py,sha256=IBhlyRQHfK-BtwUnSEbIPqlI1MtZ8-XsdaHv0b91HQ0,269
@@ -47,7 +46,7 @@ abstract_webtools/managers/soupManager/soupManager.py,sha256=6vWYnZGuimStbNiuH_V
 abstract_webtools/managers/urlManager/__init__.py,sha256=gaJCHeK91Z-eYsBnxgdhbIUten1-gbx-zqx70R6ag-Y,26
 abstract_webtools/managers/urlManager/urlManager (Copy).py,sha256=vCFuLADmv3h7icaaoAsImGqb_49VizPY_ZvMl-C7PYk,7756
 abstract_webtools/managers/urlManager/urlManager.py,sha256=vY4KQXtcrlC2YtlultxQpVe581l5kAuT5VGA0WrI16g,8945
-abstract_webtools-0.1.6.154.dist-info/METADATA,sha256=LePpjPoDmdzjKrrzctyP_nsCoa04MMVrSFzZZgoa8Vw,16573
-abstract_webtools-0.1.6.154.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-abstract_webtools-0.1.6.154.dist-info/top_level.txt,sha256=2DMJ7RmjTcjCsa-uwAV0K6eXXlIIkFDEjBLg_uyCmCI,18
-abstract_webtools-0.1.6.154.dist-info/RECORD,,
+abstract_webtools-0.1.6.156.dist-info/METADATA,sha256=qtxMiWowCso_y87NvZFOaObZZwu9hrf1OfWxVhOwOl8,16573
+abstract_webtools-0.1.6.156.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+abstract_webtools-0.1.6.156.dist-info/top_level.txt,sha256=2DMJ7RmjTcjCsa-uwAV0K6eXXlIIkFDEjBLg_uyCmCI,18
+abstract_webtools-0.1.6.156.dist-info/RECORD,,

abstract_webtools/url_grabber.py

@@ -1,212 +0,0 @@
-from abstract_gui import make_component,sg
-import inspect
-import re
-from .managers import *
-window = None
-
-def get_attrs(values):
-  tags_js={'tag':[],'attribute':[],'input':[]}
-  for each in ['-SOUP_TAG-','-SOUP_ATTRIBUTE-','-SOUP_ATTRIBUTE_1-','-SOUP_ATTRIBUTE_2-']:
-    if values[each[:-1]+'_BOOL-'] == True:
-      for types in ['tag','attribute']:
-        if types in each.lower():
-          tags_js[types].append(values[each])
-  input_val = values['-SOUP_VALUES_INPUT-']
-  if input_val == '':
-    tags_js['input']=None
-  else:
-    tags_js['input']= input_val
-  if tags_js['tag']==[]:
-    tags_js['tag']=None if match.group(1) else None
-  else:
-    tags_js['tag']=tags_js['tag'][0]
-  if tags_js['attribute']==[]:
-    tags_js['attribute']=None
-  else:
-    tags_js['attribute']=tags_js['attribute'][0]
-  return tags_js
-def get_user_agent_mgr(user_agent=None):
-  return UserAgentManager(user_agent=user_agent)
-def get_cipher_list():
-  return CipherManager().get_default_ciphers()
-def get_parse_type_choices():
-    return ['html.parser', 'lxml', 'html5lib']
-def expandable(size:tuple=(None,None)):
-    return {"size": size,"resizable": True,"scrollable": True,"auto_size_text": True,"expand_x":True,"expand_y": True}
-def change_glob(var:any,val:any):
-    globals()[var]=val
-    return val
-def get_parse_type_choices():
-    bs4_module = inspect.getmodule(BeautifulSoup)
-    docstring = bs4_module.__builtins__
-    start_index = docstring.find("parse_types")
-    end_index = docstring.find(")", start_index)
-    choices_text = docstring[start_index:end_index]
-    choices = [choice.strip() for choice in choices_text.split(",")]
-    return choices
-def get_browsers():
-    return 'Chrome,Firefox,Safari,Microsoft Edge,Internet Explorer,Opera'.split(',')
-def get_user_agents():
-    from .big_user_agent_list import big_user_agent_list
-    return big_user_agent_list
-def create_user_agent(user_agent:str=get_user_agents()[0]):
-    return {"user-agent": user_agent}
-def get_operating_systems():
-  return ['Windows NT 10.0','Macintosh; Intel Mac OS X 10_15_7','Linux','Android','iOS']
-def create_columns(ls,i,k):
-    if float(i)%float(k)==float(0.00) and i != 0:
-        lsN = list(ls[:-k])
-        lsN.append(list(ls[-k:]))
-        ls = lsN
-    return ls
-def get_cypher_checks():
-    ciphers_list = get_cipher_list()
-    ls=[[[sg.Text('CIPHERS: ')],sg.Multiline('',key='-CIPHERS_OUTPUT-', size=(80, 5), disabled=False)]]
-    for k,cipher in enumerate(ciphers_list):
-        ls.append(sg.Checkbox(cipher,key=cipher,default=True,enable_events=True))
-        ls = create_columns(ls,k,5)
-    return ls
-def get_bs4_options():
-    bs4_options = [
-        'BeautifulSoup',
-        'Tag',
-        'NavigableString',
-        'Comment',
-        'ResultSet',
-        'SoupStrainer',
-        'CData'
-    ]
-    descriptions = [
-        'The main BeautifulSoup class used for parsing HTML.',
-        'Represents an HTML tag.',
-        'Represents a string within an HTML document.',
-        'Represents an HTML comment.',
-        'Represents a collection of tags found during a search.',
-        'Allows parsing only a specific subset of the HTML document.',
-        'Represents a CDATA section within an XML document.'
-    ]
-    return list(zip(bs4_options, descriptions))
-def get_multi_line(args):
-    return make_component("Multiline",**args,**expandable())
-def get_gpt_layout(url):
-    # Add a dropdown for selecting BeautifulSoup parsing capabilities
-    parse_type_choices = ['html.parser', 'lxml', 'html5lib']
-    make_component("theme",'LightGrey1')
-    layout = [[sg.Text('URL:', size=(8, 1)), sg.Input(url, key='-URL-',enable_events=True),sg.Text('status:'),sg.Text('200',key="-STATUS_CODE-")
-               ,sg.Text(f'success: {url} is valid',key="-URL_WARNING-"),sg.Button('Grab URL',key='-GRAB_URL-',visible=True)],
-        [sg.Checkbox('Custom User-Agent', default=False, key='-CUSTOMUA-', enable_events=True)],
-        [sg.Text('User-Agent:', size=(8, 1)), sg.Combo(get_user_agents(), default_value='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.90 Safari/537.36', key='-USERAGENT-', disabled=False)],
-        [get_cypher_checks()],
-        [sg.Button('Grab URL'), sg.Button('Action'),sg.Button('Get All Text')],
-        [sg.Text('Parsing Capabilities:', size=(15, 1)), sg.DropDown(parse_type_choices, default_value='html.parser', key='-parse_type-',enable_events=True)],
-        [get_multi_line({"key":'-SOURCECODE-'})],
-        [sg.Text('find soup:'),[[sg.Checkbox('',default=True,key='-SOUP_TAG_BOOL-',enable_events=True),sg.Combo([], size=(15, 1),key='-SOUP_TAG-',enable_events=True)],
-                                [sg.Checkbox('',default=False,key='-SOUP_ATTRIBUTE_BOOL-',enable_events=True),sg.Combo([], size=(15, 1),key='-SOUP_ATTRIBUTE-',enable_events=True)],
-                                [sg.Checkbox('',default=False,key='-SOUP_ATTRIBUTE_1_BOOL-',enable_events=True),sg.Combo([], size=(15, 1),key='-SOUP_ATTRIBUTE_1-',enable_events=True)],
-                                [sg.Checkbox('',default=False,key='-SOUP_ATTRIBUTE_2_BOOL-',enable_events=True),sg.Combo([], size=(15, 1),key='-SOUP_ATTRIBUTE_2-',enable_events=True)],
-                                sg.Input(key='-SOUP_VALUES_INPUT-'), sg.Button('get soup'),sg.Button('all soup'),sg.Button('Send Soup')]],
-        [get_multi_line({"key":"-FIND_ALL_OUTPUT-"})]]
-    return layout
-def get_selected_cipher_list():
-  ls = []
-  ciphers_list = get_cipher_list()
-  event, values = window.read()
-  for cipher in ciphers_list:
-      if values[cipher] == True:
-          ls.append(cipher)
-  return ls
-def update_status(window,warn,warn_url,response_code,valid):
-    window['-URL-'].update(value=warn_url)
-    window['-STATUS_CODE-'].update(value=response_code)
-    window["-URL_WARNING-"].update(value=f"{warn} : {warn_url} is {valid}")
-def process_url(window,values):
-    response_code=False
-    temp_mgr=None
-    warn='warning'
-    valid='invalid'
-    warn_url = values['-URL-']
-    if warn_url=='' or warn_url == None:
-      update_status(window,warn,warn_url,response_code,valid)
-      return False
-    temp_url=urlManager(url=warn_url).url
-    if temp_url:
-      valid='valid'
-      response_code = requestManager(url=temp_mgr).response.status_code
-      warn = 'success'
-      warn_url = temp_mgr
-      update_status(window,warn,warn_url,response_code,valid)
-      return temp_mgr
-    update_status(window,warn,warn_url,response_code,valid)
-    return False
-def update_url(url_mgr,request_mgr,soup_mgr,link_mgr,values,cipher_list=get_cipher_list(),user_agent=get_user_agents()[0]):
-      ciphers = CipherManager(cipher_list=cipher_list).ciphers_string
-      request_mgr = requestManager(url_mgr=url_mgr,ciphers=ciphers,user_agent=get_user_agents()[0])
-      if request_mgr.source_code:
-        soup_mgr= SoupManager(url_mgr=url_mgr,request_mgr=request_mgr)
-        link_mgr= LinkManager(url_mgr=url_mgr,request_mgr=request_mgr,soup_mgr=soup_mgr)
-        window['-URL-'].update(value=url_mgr.url)
-        window['-CIPHERS_OUTPUT-'].update(value=request_mgr.ciphers)
-        return update_source_code(url_mgr,request_mgr,soup_mgr,link_mgr,values)
-      else:
-        return url_mgr,request_mgr,soup_mgr,link_mgr
-def update_source_code(url_mgr,request_mgr,soup_mgr,link_mgr,values):
-    parse_type = values['-parse_type-']
-    if parse_type != soup_mgr.parse_type:
-      soup_mgr.update_parse_type(parse_type=parse_type)
-    all_tags=soup_mgr.get_all_tags_and_attribute_names()
-    window['-SOURCECODE-'].update(value=soup_mgr.soup)
-    if values['-SOUP_TAG-'] != all_tags['tags']:
-      window['-SOUP_TAG-'].update(values=all_tags['tags'],value=all_tags['tags'][0])
-    if values['-SOUP_ATTRIBUTE-'] != all_tags['attributes']:
-      window['-SOUP_ATTRIBUTE-'].update(values=all_tags['attributes'],value=all_tags['attributes'][0])
-      window['-SOUP_ATTRIBUTE_1-'].update(values=all_tags['attributes'],value=all_tags['attributes'][0])
-      window['-SOUP_ATTRIBUTE_2-'].update(values=all_tags['attributes'],value=all_tags['attributes'][0])
-      return url_mgr,request_mgr,soup_mgr,link_mgr
-def url_grabber_while(window,initial_url="www.example.com"):
-    return_data=None
-    url_grab = False
-    url_mgr=urlManager(url=initial_url)
-    request_mgr = requestManager(url_mgr=url_mgr)
-    soup_mgr= SoupManager(url_mgr=url_mgr,request_mgr=request_mgr)
-    link_mgr= LinkManager(url_mgr=url_mgr,request_mgr=request_mgr,soup_mgr=soup_mgr)
-    while True:
-        event, values = window.read()
-        if event == sg.WINDOW_CLOSED:
-            break
-        if event=='-GRAB_URL-' or not url_grab:
-          url=values['-URL-']
-          if urlManager(url=url).url:
-            if url != url_mgr.url or url == initial_url:
-              url_mgr = urlManager(url=url)
-              
-              url_mgr,request_mgr,soup_mgr,link_mgr=update_url(url_mgr=url_mgr,request_mgr=request_mgr,soup_mgr=soup_mgr,link_mgr=link_mgr,values=values)
-              window['-URL-'].update(value=url_mgr.url)
-              url_grab=True
-        if event == 'get soup':
-            tags_js = get_attrs(values)
-            all_desired=soup_mgr.find_tags_by_attributes(tag=tags_js['tag'], attr=tags_js['attribute'],attr_values=tags_js['input'])
-            window['-FIND_ALL_OUTPUT-'].update(value=all_desired)
-        if event == '-CUSTOMUA-':
-            window['-SOURCECODE-'].update(disabled=values['-CUSTOMUA-'])
-            if not values['-CUSTOMUA-']:
-                window['-USERAGENT-'].update(value=user_agent_mgr.user_agent_header)
-                window['-USERAGENT-'].update(disabled=True)
-            else:
-                window['-USERAGENT-'].update(disabled=False)
-        if event=='Get All Text':
-            window['-FIND_ALL_OUTPUT-'].update(value=soup_mgr.extract_text_sections())
-        if event == 'Action':
-            parse_type = values['-parse_type-']
-            if parse_type != soup_mgr.parse_type:
-              soup_mgr.update_parse_type(parse_type=parse_type)
-            window['-SOURCECODE-'].update(value=soup_mgr.soup)
-        elif event == 'Send Soup':
-          return_data = values['-FIND_ALL_OUTPUT-']
-          break
-    window.close()
-    return return_data
-def url_grabber_component(url=None):
-    if url==None:
-      url = "www.example.com"
-    globals()['window'] = make_component('Window','URL Grabber', layout=get_gpt_layout(url),**expandable())
-    return url_grabber_while(window,initial_url=url)