refactor: cleanup

dev_gpt/options/generate/chains/auto_refine_description.py

@@ -0,0 +1,57 @@
+import json
+
+from dev_gpt.apis.gpt import ask_gpt
+from dev_gpt.options.generate.chains.extract_information import extract_information
+from dev_gpt.options.generate.parser import identity_parser
+from dev_gpt.options.generate.prompt_factory import context_to_string
+
+
+
+def auto_refine_description(context):
+    context['microservice_description'] = ask_gpt(
+        better_description_prompt,
+        identity_parser,
+        context_string=context_to_string(context)
+    )
+    context['request_schema'] = ask_gpt(
+        generate_request_schema_prompt,
+        identity_parser,
+        context_string=context_to_string(context)
+    )
+    context['response_schema'] = ask_gpt(
+        generate_output_schema_prompt,
+        identity_parser,
+        context_string=context_to_string(context)
+    )
+    context['microservice_description'] = ask_gpt(
+        summarize_description_and_schemas_prompt,
+        identity_parser,
+        context_string=context_to_string(context)
+    )
+    # details = extract_information(context['microservice_description'], ['database connection details', 'URL', 'secret'])
+    # if details:
+    #     context['microservice_description'] += '\n\nAdditional information:' + json.dumps(details, indent=4)
+    # del context['details']
+
+
+better_description_prompt = f'''{{context_string}}
+Update the description of the Microservice to make it more precise without adding or removing information.
+Note: the output must be a list of tasks the Microservice has to perform.
+Example for the description: "return the average temperature of the 5 days weather forecast for a given location."
+1. get the 5 days weather forcast from the https://openweathermap.org/ API
+2. extract the temperature from the response
+3. calculate the average temperature'''
+
+generate_request_schema_prompt = '''{context_string}
+Generate the lean request json schema of the Microservice.
+Note: If you are not sure about the details, then come up with the minimal number of parameters possible.'''
+
+generate_output_schema_prompt = '''{context_string}
+Generate the lean response json schema for the Microservice.
+Note: If you are not sure about the details, then come up with the minimal number of parameters possible.'''
+
+summarize_description_and_schemas_prompt = '''{context_string}
+Write an updated microservice description by incorporating information about the request and response parameters in a concise way without losing any information.
+Note: You must not mention any details about algorithms or the technical implementation.
+Note: You must not mention that there is a request and response JSON schema
+Note: You must not use any formatting like triple backticks.'''

dev_gpt/options/generate/chains/extract_information.py

@@ -0,0 +1,29 @@
+from typing import Dict
+
+from dev_gpt.apis.gpt import ask_gpt
+from dev_gpt.options.generate.chains.question_answering import answer_yes_no_question
+from dev_gpt.options.generate.parser import identity_parser, boolean_parser
+
+
+def extract_information(text, info_keys) -> Dict[str, str]:
+    extracted_infos = {}
+    for info_key in info_keys:
+        is_information_in_text = answer_yes_no_question(text, f'Is a {info_key} mentioned above?')
+        if is_information_in_text:
+            extracted_info = ask_gpt(
+                extract_information_prompt,
+                identity_parser,
+                text=text,
+                info_key=info_key
+            )
+            extracted_infos[info_key] = extracted_info
+    return extracted_infos
+
+
+extract_information_prompt = '''\
+{text}
+
+Your task: 
+Return all {info_key}s from above.'
+Note: you must only output your answer.
+'''

dev_gpt/options/generate/chains/get_user_input_if_needed.py

@@ -4,13 +4,14 @@ from dev_gpt.options.generate.parser import identity_parser
 
 
 def get_user_input_if_needed(context, conditions, question_gen_prompt_part):
-    if all([c(context) for c in conditions]):
-        return ask_gpt(
+    if all([c(context_to_string(context)) for c in conditions]):
+        question_to_user = ask_gpt(
             generate_question_for_file_input_prompt,
             identity_parser,
             context_string=context_to_string(context),
             question_gen_prompt_part=question_gen_prompt_part
         )
+        return input(question_to_user)
     return None
 
 generate_question_for_file_input_prompt = '''\

dev_gpt/options/generate/chains/question_answering.py

@@ -0,0 +1,16 @@
+from dev_gpt.apis.gpt import ask_gpt
+from dev_gpt.options.generate.parser import boolean_parser
+
+
+def answer_yes_no_question(text, question):
+    prompt = question_prompt.format(
+        question=question,
+        text=text
+    )
+    return ask_gpt(prompt, boolean_parser)
+
+question_prompt = '''\
+{text}
+{question}
+Note: You must answer "yes" or "no".
+'''

dev_gpt/options/generate/condition.py

@@ -1,22 +1,10 @@
-from dev_gpt.apis.gpt import ask_gpt
-from dev_gpt.options.generate.prompt_factory import context_to_string
-from dev_gpt.options.generate.parser import boolean_parser
+from dev_gpt.options.generate.chains.question_answering import answer_yes_no_question
 
 
-def is_true(question):
-    def fn(context):
-        prompt = question_prompt.format(
-            question=question,
-            context_string=context_to_string(context)
-        )
-        return ask_gpt(prompt, boolean_parser)
+def is_question_true(question):
+    def fn(text):
+        return answer_yes_no_question(text, question)
     return fn
 
-def is_false(question):
-    return lambda context: not is_true(question)(context)
-
-question_prompt = '''\
-{context_string}
-{question}
-Note: You must answer "yes" or "no".
-'''
+def is_question_false(question):
+    return lambda context: not is_question_true(question)(context)

dev_gpt/options/generate/pm/pm.py

@@ -1,7 +1,8 @@
 from dev_gpt.apis import gpt
 from dev_gpt.apis.gpt import ask_gpt
+from dev_gpt.options.generate.chains.auto_refine_description import auto_refine_description
 from dev_gpt.options.generate.chains.user_confirmation_feedback_loop import user_feedback_loop
-from dev_gpt.options.generate.condition import is_false, is_true
+from dev_gpt.options.generate.condition import is_question_false, is_question_true
 from dev_gpt.options.generate.chains.get_user_input_if_needed import get_user_input_if_needed
 from dev_gpt.options.generate.parser import identity_parser
 # from dev_gpt.options.generate.pm.task_tree_schema import TaskTree
@@ -38,11 +39,9 @@ Description of the microservice:
         # return sub_task_tree
 
     def refine_description(self, microservice_description):
-        context = {
-            'microservice_description': microservice_description
-        }
-        self.auto_refine_description(context)
-        user_feedback_loop(context, microservice_description)
+        context = {'microservice_description': microservice_description}
+        auto_refine_description(context)
+        microservice_description = user_feedback_loop(context, context['microservice_description'])
 
         test_description = ask_gpt(
             generate_test_description_prompt,
@@ -56,8 +55,8 @@ Description of the microservice:
                 'Response schema': context['response_schema'],
             },
             conditions=[
-                is_true('Does request schema contain an example file url?'),
-                is_false('Is input url specified in the description?'),
+                is_question_true('Does request schema require a file?'),
+                # is_question_false('Is input url specified in the description?'),
             ],
             question_gen_prompt_part="Generate a question that asks for an example file url.",
         )
@@ -66,26 +65,7 @@ Description of the microservice:
 
         return microservice_description, test_description
 
-    def auto_refine_description(self, context):
-        context['microservice_description'] = ask_gpt(
-            better_description_prompt,
-            identity_parser,
-            **context
-        )
-        context['request_schema'] = ask_gpt(
-            generate_request_schema_prompt,
-            identity_parser,
-            **context
-        )
-        context['response_schema'] = ask_gpt(
-            generate_output_schema_prompt,
-            identity_parser,
-            **context
-        )
-        context['microservice_description'] = ask_gpt(
-            summarize_description_and_schemas_prompt, identity_parser,
-            **context
-        )
+
 
     # def get_nlp_fns(self, microservice_description):
     #     return ask_gpt(
@@ -229,28 +209,10 @@ Microservice description:
 {microservice_description}
 ```'''
 
-better_description_prompt = client_description + '''
-Update the description of the Microservice to make it more precise without adding or removing information.
-Note: the output must be a list of tasks the Microservice has to perform.
-Example for the description: "return the average temperature of the 5 days weather forecast for a given location."
-1. get the 5 days weather forcast from the https://openweathermap.org/ API
-2. extract the temperature from the response
-3. calculate the average temperature'''
 
 # better_description_prompt = client_description + '''
 # Update the description of the Microservice to make it more precise without adding or removing information.'''
 
-generate_request_schema_prompt = client_description + '''
-Generate the lean request json schema of the Microservice.
-Note: If you are not sure about the details, then come up with the minimal number of parameters possible.'''
-
-generate_output_schema_prompt = client_description + '''
-request json schema:
-```
-{request_schema}
-```
-Generate the lean response json schema for the Microservice.
-Note: If you are not sure about the details, then come up with the minimal number of parameters possible.'''
 
 # If we want to activate this back, then it first needs to work. Currently, it outputs "no" for too many cases.
 # is_feedback_valuable_prompt = client_description + '''
@@ -263,20 +225,6 @@ Note: If you are not sure about the details, then come up with the minimal numbe
 # Note: If the user does not want to provide feedback, then you must answer "no".'''
 
 
-summarize_description_and_schemas_prompt = client_description + '''
-Request json schema:
-```
-{request_schema}
-```
-Response json schema:
-```
-{response_schema}
-```
-Write an updated microservice description by incorporating information about the request and response parameters in a concise way without losing any information.
-Note: You must not mention any details about algorithms or the technical implementation.
-Note: You must not mention that there is a request and response JSON schema
-Note: You must not use any formatting like triple backticks.'''
-
 
 
 # summarize_description_prompt = client_description + '''

test/integration/test_generator.py

@@ -51,7 +51,7 @@ But hey, at least SOMEONE's enjoying their lunch. #officelife\'''',
     assert generator.generate() == 0
 
 
-@pytest.mark.parametrize('mock_input_sequence', [['y']], indirect=True)
+@pytest.mark.parametrize('mock_input_sequence', [['y', 'https://www.africau.edu/images/default/sample.pdf']], indirect=True)
 def test_generation_level_2(microservice_dir, mock_input_sequence):
     """
     Requirements:
@@ -64,7 +64,7 @@ def test_generation_level_2(microservice_dir, mock_input_sequence):
     """
     os.environ['VERBOSE'] = 'true'
     generator = Generator(
-        "The input is a PDF like https://www.africau.edu/images/default/sample.pdf and the output the summarized text (50 words).",
+        "The input is a PDF and the output the summarized text (50 words).",
         str(microservice_dir),
         'gpt-3.5-turbo'
     )
@@ -95,7 +95,7 @@ Example input: 'AAPL'
     )
     assert generator.generate() == 0
 
-@pytest.mark.parametrize('mock_input_sequence', [['y']], indirect=True)
+@pytest.mark.parametrize('mock_input_sequence', [['y', 'https://www.signalogic.com/melp/EngSamples/Orig/ENG_M.wav']], indirect=True)
 def test_generation_level_4(microservice_dir, mock_input_sequence):
     """
     Requirements:
@@ -125,14 +125,13 @@ print('This is the text from the audio file:', response.json()['text'])
 2. Summarize the text (~50 words) while still maintaining the key facts.
 3. Create an audio file of the summarized text using a tts library.
 4. Return the the audio file as base64 encoded binary.
-Example input file: https://www.signalogic.com/melp/EngSamples/Orig/ENG_M.wav
 ''',
         str(microservice_dir),
         'gpt-4'
     )
     assert generator.generate() == 0
 
-@pytest.mark.parametrize('mock_input_sequence', [['y']], indirect=True)
+@pytest.mark.parametrize('mock_input_sequence', [['y', 'https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/PNG_transparency_demonstration_1.png/560px-PNG_transparency_demonstration_1.png']], indirect=True)
 def test_generation_level_5(microservice_dir, mock_input_sequence):
     """
     Requirements:
@@ -165,7 +164,6 @@ Result format:
 The description is then used to generate a joke.
 The joke is the put on the image.
 The output is the image with the joke on it.
-Example input image: https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/PNG_transparency_demonstration_1.png/560px-PNG_transparency_demonstration_1.png
 ''',
                           str(microservice_dir),
                           'gpt-3.5-turbo'