added _experiment of transforming a ChatGPT thread into an CBR thread

Author: DinisCruz

tests/_experiments/test_chat_gpt_threads.py

@@ -0,0 +1,79 @@
+
+from unittest import TestCase
+import pytest
+import requests
+#from bs4 import BeautifulSoup
+from osbot_utils.helpers.Random_Guid import Random_Guid
+from osbot_utils.utils.Dev import pprint
+from osbot_utils.utils.Files import file_create, file_contains, file_contents, temp_file
+from osbot_utils.utils.Json import json_load, json_parse, json_file_create, json_file_load
+from osbot_utils.utils.Objects import dict_to_obj
+from osbot_llms.models.GPT_History import GPT_History
+from osbot_llms.models.LLMs__Chat_Completion import LLMs__Chat_Completion
+
+
+# notes:
+#       - this worked really well :)
+#       - the use of bs4 might be a bit overthe top since all we need is the contents of the last script tag
+#       - there was a bug with Html_To_Tag that prevented its use ( todo: see why and fix it)
+# @pytest.mark.skip("Needs to be integrated into a service")
+# class test_chat_gpt_threads(TestCase):
+#     def test_load_chat_gpt_saved_chat(self):
+#
+#         # url      = "https://chatgpt.com/share/6713bfef-3778-8006-9bbd-01f842395d6c"
+#         # response = requests.get(url)
+#         # assert response.status_code
+#         # html_code = response.text
+#         #html_file         = file_create(contents=html_code, extension=".html")
+#
+#         # html_file         = '/var/folders/z4/3k99j_cn39g_0jnqt6w57f6m0000gn/T/tmpqnpeu3bj.html'
+#         # html_code         = file_contents(path=html_file)
+#         # html_soup         = BeautifulSoup(html_code, 'html.parser')
+#         # js_code           = html_soup.find_all('script')[1].text
+#         # json_code         = js_code[24:-1]
+#         # json_data         = json.loads(json_code)
+#
+#         #json_file         = json_file_create(json_data, path=temp_file(extension=".json"))
+#         json_file          = '/var/folders/z4/3k99j_cn39g_0jnqt6w57f6m0000gn/T/tmpw28a0cpx.json'
+#         json_data          = json_file_load(json_file)
+#         json_obj          = dict_to_obj(json_data)
+#
+#         action_json = json_data.get('state').get('loaderData').get('routes/share.$shareId.($action)')
+#         action_obj = dict_to_obj(action_json)
+#
+#         data_obj   = action_obj.serverResponse.data
+#
+#         title = data_obj.title
+#         print()
+#
+#         question = None
+#         answer = None
+#         for item in data_obj.linear_conversation:
+#             if hasattr(item, 'message'):
+#                 if hasattr(item.message.metadata, 'is_visually_hidden_from_conversation'):
+#                     continue
+#                 role =  item.message.author.role
+#                 if item.message.content.content_type == 'text':
+#                     parts = item.message.content.parts
+#                     text = ''.join(parts)
+#                     if role == 'user':
+#                         question = text
+#                     if role == 'assistant':
+#                         answer = text
+#         gpt_history = GPT_History(question=question, answer=answer)
+#         llms_chat_completion = LLMs__Chat_Completion(histories=[gpt_history])
+#         llms_chat_completion.chat_thread_id = Random_Guid()
+#         llms_chat_completion.user_prompt  = "can you summarize this conversation?"
+#         llms_chat_completion.stream       = True
+#         llms_chat_completion.llm_platform = "Groq (Free)"
+#         llms_chat_completion.llm_provider = "1. Meta"
+#         llms_chat_completion.llm_model    = "llama-3.1-70b-versatile"
+#
+#         request_json = llms_chat_completion.json()
+#         #request_url  = "https://community.prod.aws.cyber-boardroom.com/api/llms/chat/completion_proxy"
+#         request_url   = "https://community.dev.aws.cyber-boardroom.com/api/open_ai/prompt_with_system__stream"
+#         response     = requests.post(request_url, json=request_json)
+#
+#         pprint(llms_chat_completion.chat_thread_id)
+#
+#         pprint(dict(response.headers))