mirror of
https://github.com/aljazceru/dev-gpt.git
synced 2025-12-24 17:14:18 +01:00
Florian Hönicke
34
.github/workflows/ci.yml
vendored
34
.github/workflows/ci.yml
vendored
@@ -5,12 +5,12 @@ on:
|
||||
pull_request:
|
||||
|
||||
jobs:
|
||||
test_level:
|
||||
test_cognitive_level:
|
||||
runs-on: ubuntu-latest
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
group: [0, 1, 2]
|
||||
group: [0, 1, 2, 3, 4]
|
||||
steps:
|
||||
- uses: actions/checkout@v2
|
||||
- name: Set up Python 3.8
|
||||
@@ -27,9 +27,35 @@ jobs:
|
||||
- name: Test
|
||||
id: test
|
||||
run: |
|
||||
pytest -vs test/test_generator.py::test_generation_level_${{ matrix.group }}
|
||||
timeout-minutes: 10
|
||||
pytest -vs test/integration/test_generator.py::test_generation_level_${{ matrix.group }}
|
||||
timeout-minutes: 15
|
||||
env:
|
||||
OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
|
||||
SCENEX_API_KEY: ${{ secrets.SCENEX_API_KEY }}
|
||||
WHISPER_API_KEY: ${{ secrets.WHISPER_API_KEY }}
|
||||
|
||||
test_unit:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: actions/checkout@v2
|
||||
- name: Set up Python 3.8
|
||||
uses: actions/setup-python@v2
|
||||
with:
|
||||
python-version: 3.8
|
||||
- name: Prepare environment
|
||||
run: |
|
||||
python -m pip install --upgrade pip
|
||||
python -m pip install wheel
|
||||
pip install --no-cache-dir ".[full,test]"
|
||||
pip install pytest
|
||||
pip install pytest-split
|
||||
- name: Test
|
||||
id: test
|
||||
run: |
|
||||
pytest -vs test/unit
|
||||
timeout-minutes: 15
|
||||
env:
|
||||
OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
|
||||
SCENEX_API_KEY: ${{ secrets.SCENEX_API_KEY }}
|
||||
WHISPER_API_KEY: ${{ secrets.WHISPER_API_KEY }}
|
||||
|
||||
|
||||
@@ -11,7 +11,7 @@ from langchain.chat_models import ChatOpenAI
|
||||
from openai.error import RateLimitError
|
||||
from langchain.schema import HumanMessage, SystemMessage, BaseMessage, AIMessage
|
||||
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
|
||||
from requests.exceptions import ConnectionError
|
||||
from requests.exceptions import ConnectionError, ChunkedEncodingError
|
||||
from urllib3.exceptions import InvalidChunkLength
|
||||
|
||||
from src.constants import PRICING_GPT4_PROMPT, PRICING_GPT4_GENERATION, PRICING_GPT3_5_TURBO_PROMPT, \
|
||||
@@ -132,7 +132,7 @@ class _GPTConversation:
|
||||
try:
|
||||
response = self._chat(self.messages)
|
||||
break
|
||||
except (ConnectionError, InvalidChunkLength) as e:
|
||||
except (ConnectionError, InvalidChunkLength, ChunkedEncodingError) as e:
|
||||
print('There was a connection error. Retrying...')
|
||||
if i == 9:
|
||||
raise e
|
||||
|
||||
12
src/apis/pypi.py
Normal file
12
src/apis/pypi.py
Normal file
@@ -0,0 +1,12 @@
|
||||
import requests
|
||||
|
||||
def is_package_on_pypi(package_name, version=None):
|
||||
optional_version = f"/{version}" if version else ""
|
||||
url = f"https://pypi.org/pypi/{package_name}{optional_version}/json"
|
||||
response = requests.get(url)
|
||||
if response.status_code == 200:
|
||||
return True
|
||||
elif response.status_code == 404:
|
||||
return False
|
||||
else:
|
||||
return None
|
||||
@@ -14,7 +14,6 @@ DOCKER_FILE_TAG = 'dockerfile'
|
||||
CLIENT_FILE_TAG = 'python'
|
||||
STREAMLIT_FILE_TAG = 'python'
|
||||
|
||||
|
||||
FILE_AND_TAG_PAIRS = [
|
||||
(EXECUTOR_FILE_NAME, EXECUTOR_FILE_TAG),
|
||||
(IMPLEMENTATION_FILE_NAME, IMPLEMENTATION_FILE_TAG),
|
||||
@@ -39,9 +38,16 @@ MAX_DEBUGGING_ITERATIONS = 10
|
||||
|
||||
DEMO_TOKEN = '45372338e04f5a41af949024db929d46'
|
||||
|
||||
PROBLEMATIC_PACKAGES = [
|
||||
# 'Pyrender', 'Trimesh',
|
||||
'moderngl', 'pyopengl', 'pyglet', 'pythreejs', 'panda3d', # because they need a screen,
|
||||
BLACKLISTED_PACKAGES = [
|
||||
'moderngl', 'pyopengl', 'pyglet', 'pythreejs', 'panda3d', # because they need a screen,
|
||||
'tika', # because it needs java
|
||||
]
|
||||
UNNECESSARY_PACKAGES = [
|
||||
'fastapi', 'uvicorn', 'starlette' # because the wrappers are used instead
|
||||
]
|
||||
|
||||
UNNECESSARY_PACKAGES = ['fastapi']
|
||||
LANGUAGE_PACKAGES = [
|
||||
'allennlp', 'bertopic', 'fasttext', 'flair', 'gensim', 'nltk',
|
||||
'pattern', 'polyglot', 'pytorch-transformers', 'rasa', 'sentence-transformers',
|
||||
'spacy', 'stanza', 'summarizer', 'sumy', 'textblob', 'textstat', 'transformers'
|
||||
]
|
||||
|
||||
@@ -4,7 +4,6 @@ import random
|
||||
import re
|
||||
import shutil
|
||||
from typing import Callable
|
||||
|
||||
from typing import List, Text, Optional
|
||||
|
||||
from langchain import PromptTemplate
|
||||
@@ -14,20 +13,20 @@ from pydantic.dataclasses import dataclass
|
||||
from src.apis import gpt
|
||||
from src.apis.gpt import _GPTConversation
|
||||
from src.apis.jina_cloud import process_error_message, push_executor, is_executor_in_hub
|
||||
from src.apis.pypi import is_package_on_pypi
|
||||
from src.constants import FILE_AND_TAG_PAIRS, NUM_IMPLEMENTATION_STRATEGIES, MAX_DEBUGGING_ITERATIONS, \
|
||||
PROBLEMATIC_PACKAGES, EXECUTOR_FILE_NAME, TEST_EXECUTOR_FILE_NAME, TEST_EXECUTOR_FILE_TAG, \
|
||||
REQUIREMENTS_FILE_NAME, REQUIREMENTS_FILE_TAG, DOCKER_FILE_NAME, UNNECESSARY_PACKAGES, IMPLEMENTATION_FILE_NAME, \
|
||||
IMPLEMENTATION_FILE_TAG
|
||||
from src.options.generate.templates_system import system_task_iteration, system_task_introduction, system_test_iteration
|
||||
BLACKLISTED_PACKAGES, EXECUTOR_FILE_NAME, TEST_EXECUTOR_FILE_NAME, TEST_EXECUTOR_FILE_TAG, \
|
||||
REQUIREMENTS_FILE_NAME, REQUIREMENTS_FILE_TAG, DOCKER_FILE_NAME, IMPLEMENTATION_FILE_NAME, \
|
||||
IMPLEMENTATION_FILE_TAG, LANGUAGE_PACKAGES, UNNECESSARY_PACKAGES
|
||||
from src.options.generate.templates_system import system_task_iteration, system_task_introduction, system_test_iteration
|
||||
from src.options.generate.templates_user import template_generate_microservice_name, \
|
||||
template_generate_possible_packages, \
|
||||
template_solve_code_issue, \
|
||||
template_solve_pip_dependency_issue, template_is_dependency_issue, template_generate_playground, \
|
||||
template_generate_function, template_generate_test, template_generate_requirements, \
|
||||
template_chain_of_thought, template_summarize_error, \
|
||||
template_generate_apt_get_install, template_solve_apt_get_dependency_issue, template_pm_task_iteration, \
|
||||
template_solve_apt_get_dependency_issue, template_pm_task_iteration, \
|
||||
template_pm_test_iteration
|
||||
|
||||
from src.options.generate.ui import get_random_employee
|
||||
from src.utils.io import persist_file, get_all_microservice_files_with_content, get_microservice_path
|
||||
from src.utils.string_tools import print_colored
|
||||
@@ -38,6 +37,7 @@ class TaskSpecification:
|
||||
task: Optional[Text]
|
||||
test: Optional[Text]
|
||||
|
||||
|
||||
class Generator:
|
||||
def __init__(self, task_description, path, model='gpt-4'):
|
||||
self.gpt_session = gpt.GPTSession(task_description, model=model)
|
||||
@@ -45,14 +45,14 @@ class Generator:
|
||||
self.microservice_root_path = path
|
||||
|
||||
def extract_content_from_result(self, plain_text, file_name, match_single_block=False, can_contain_code_block=True):
|
||||
optional_line_break = '\n' if can_contain_code_block else '' # the \n at the end makes sure that ``` within the generated code is not matched because it is not right before a line break
|
||||
optional_line_break = '\n' if can_contain_code_block else '' # the \n at the end makes sure that ``` within the generated code is not matched because it is not right before a line break
|
||||
pattern = fr"\*?\*?{file_name}\*?\*?\n```(?:\w+\n)?([\s\S]*?){optional_line_break}```"
|
||||
match = re.search(pattern, plain_text, re.MULTILINE)
|
||||
if match:
|
||||
return match.group(1).strip()
|
||||
elif match_single_block:
|
||||
# Check for a single code block
|
||||
single_code_block_pattern = r"^```(?:\w+\n)?([\s\S]*?)```"
|
||||
single_code_block_pattern = r"```(?:\w+\n)?([\s\S]*?)```"
|
||||
single_code_block_match = re.findall(single_code_block_pattern, plain_text, re.MULTILINE)
|
||||
if len(single_code_block_match) == 1:
|
||||
return single_code_block_match[0].strip()
|
||||
@@ -79,10 +79,11 @@ metas:
|
||||
def _default_parse_result_fn(x):
|
||||
_parsed_results = {}
|
||||
for _file_name in files_names:
|
||||
_content = self.extract_content_from_result(x, _file_name, match_single_block=len(files_names)==1)
|
||||
_content = self.extract_content_from_result(x, _file_name, match_single_block=len(files_names) == 1)
|
||||
if _content != '':
|
||||
_parsed_results[_file_name] = _content
|
||||
return _parsed_results
|
||||
|
||||
return _default_parse_result_fn
|
||||
|
||||
def generate_and_persist_file(
|
||||
@@ -113,7 +114,9 @@ metas:
|
||||
parse_result_fn = self.get_default_parse_result_fn(file_name_s)
|
||||
|
||||
print_colored('', f'\n\n############# {section_title} #############', 'blue')
|
||||
system_introduction_message = _GPTConversation._create_system_message(self.microservice_specification.task, self.microservice_specification.test, system_definition_examples)
|
||||
system_introduction_message = _GPTConversation._create_system_message(self.microservice_specification.task,
|
||||
self.microservice_specification.test,
|
||||
system_definition_examples)
|
||||
conversation = self.gpt_session.get_conversation(messages=[system_introduction_message])
|
||||
template_kwargs = {k: v for k, v in template_kwargs.items() if k in template.input_variables}
|
||||
if 'file_name' in template.input_variables and len(file_name_s) == 1:
|
||||
@@ -125,7 +128,8 @@ metas:
|
||||
)
|
||||
content = parse_result_fn(content_raw)
|
||||
if content == {}:
|
||||
content_raw = conversation.chat('You must add the content' + (f' for {file_name_s[0]}' if len(file_name_s) == 1 else ''))
|
||||
content_raw = conversation.chat(
|
||||
'You must add the content' + (f' for {file_name_s[0]}' if len(file_name_s) == 1 else ''))
|
||||
content = parse_result_fn(content_raw)
|
||||
for _file_name, _file_content in content.items():
|
||||
persist_file(_file_content, os.path.join(destination_folder, _file_name))
|
||||
@@ -137,12 +141,14 @@ metas:
|
||||
packages,
|
||||
num_approach,
|
||||
):
|
||||
MICROSERVICE_FOLDER_v1 = get_microservice_path(self.microservice_root_path, microservice_name, packages, num_approach, 1)
|
||||
MICROSERVICE_FOLDER_v1 = get_microservice_path(self.microservice_root_path, microservice_name, packages,
|
||||
num_approach, 1)
|
||||
os.makedirs(MICROSERVICE_FOLDER_v1)
|
||||
|
||||
with open(os.path.join(os.path.dirname(__file__), 'static_files', 'microservice', 'microservice.py'), 'r') as f:
|
||||
microservice_executor_boilerplate = f.read()
|
||||
microservice_executor_code = microservice_executor_boilerplate.replace('class GPTDeployExecutor(Executor):', f'class {microservice_name}(Executor):')
|
||||
microservice_executor_code = microservice_executor_boilerplate.replace('class GPTDeployExecutor(Executor):',
|
||||
f'class {microservice_name}(Executor):')
|
||||
persist_file(microservice_executor_code, os.path.join(MICROSERVICE_FOLDER_v1, EXECUTOR_FILE_NAME))
|
||||
|
||||
with open(os.path.join(os.path.dirname(__file__), 'static_files', 'microservice', 'apis.py'), 'r') as f:
|
||||
@@ -201,14 +207,14 @@ metas:
|
||||
# })
|
||||
# )
|
||||
|
||||
|
||||
with open(os.path.join(os.path.dirname(__file__), 'static_files', 'microservice', 'Dockerfile'), 'r', encoding='utf-8') as f:
|
||||
with open(os.path.join(os.path.dirname(__file__), 'static_files', 'microservice', 'Dockerfile'), 'r',
|
||||
encoding='utf-8') as f:
|
||||
docker_file_template_lines = f.readlines()
|
||||
docker_file_template_lines = [line for line in docker_file_template_lines if not line.startswith('RUN apt-get update')]
|
||||
docker_file_template_lines = [line for line in docker_file_template_lines if
|
||||
not line.startswith('RUN apt-get update')]
|
||||
docker_file_content = '\n'.join(docker_file_template_lines)
|
||||
persist_file(docker_file_content, os.path.join(MICROSERVICE_FOLDER_v1, 'Dockerfile'))
|
||||
|
||||
|
||||
self.write_config_yml(microservice_name, MICROSERVICE_FOLDER_v1)
|
||||
|
||||
print('\nFirst version of the microservice generated. Start iterating on it to make the tests pass...')
|
||||
@@ -223,13 +229,15 @@ metas:
|
||||
packages = ' '.join(json.loads(json_string)['packages'])
|
||||
|
||||
docker_file_template = self.read_docker_template()
|
||||
return {DOCKER_FILE_NAME: docker_file_template.replace('{{apt_get_packages}}', '{apt_get_packages}').format(apt_get_packages=packages)}
|
||||
return {DOCKER_FILE_NAME: docker_file_template.replace('{{apt_get_packages}}', '{apt_get_packages}').format(
|
||||
apt_get_packages=packages)}
|
||||
|
||||
def parse_result_fn_requirements(self, content_raw: str):
|
||||
content_parsed = self.extract_content_from_result(content_raw, 'requirements.txt', match_single_block=True)
|
||||
|
||||
lines = content_parsed.split('\n')
|
||||
lines = [line for line in lines if not any([pkg in line for pkg in ['jina', 'docarray', 'openai', 'pytest', 'gpt_3_5_turbo_api']])]
|
||||
lines = [line for line in lines if
|
||||
not any([pkg in line for pkg in ['jina', 'docarray', 'openai', 'pytest', 'gpt_3_5_turbo']])]
|
||||
content_modified = f'''jina==3.15.1.dev14
|
||||
docarray==0.21.0
|
||||
openai==0.27.5
|
||||
@@ -291,12 +299,14 @@ pytest
|
||||
for i in range(1, MAX_DEBUGGING_ITERATIONS):
|
||||
print('Debugging iteration', i)
|
||||
print('Trying to debug the microservice. Might take a while...')
|
||||
previous_microservice_path = get_microservice_path(self.microservice_root_path, microservice_name, packages, num_approach, i)
|
||||
next_microservice_path = get_microservice_path(self.microservice_root_path, microservice_name, packages, num_approach, i + 1)
|
||||
previous_microservice_path = get_microservice_path(self.microservice_root_path, microservice_name, packages,
|
||||
num_approach, i)
|
||||
next_microservice_path = get_microservice_path(self.microservice_root_path, microservice_name, packages,
|
||||
num_approach, i + 1)
|
||||
log_hubble = push_executor(previous_microservice_path)
|
||||
error = process_error_message(log_hubble)
|
||||
if error:
|
||||
print('An error occurred during the build process. Feeding the error back to the assistent...')
|
||||
print('An error occurred during the build process. Feeding the error back to the assistant...')
|
||||
self.do_debug_iteration(error, next_microservice_path, previous_microservice_path)
|
||||
if i == MAX_DEBUGGING_ITERATIONS - 1:
|
||||
raise self.MaxDebugTimeReachedException('Could not debug the microservice.')
|
||||
@@ -356,7 +366,8 @@ pytest
|
||||
summarized_error=summarized_error,
|
||||
task_description=self.microservice_specification.task,
|
||||
test_description=self.microservice_specification.test,
|
||||
all_files_string=self.files_to_string({key: val for key, val in file_name_to_content.items() if key != EXECUTOR_FILE_NAME}),
|
||||
all_files_string=self.files_to_string(
|
||||
{key: val for key, val in file_name_to_content.items() if key != EXECUTOR_FILE_NAME}),
|
||||
)
|
||||
|
||||
class MaxDebugTimeReachedException(BaseException):
|
||||
@@ -367,16 +378,22 @@ pytest
|
||||
|
||||
def is_dependency_issue(self, summarized_error, dock_req_string: str, package_manager: str):
|
||||
# a few heuristics to quickly jump ahead
|
||||
if any([error_message in summarized_error for error_message in ['AttributeError', 'NameError', 'AssertionError']]):
|
||||
if any([error_message in summarized_error for error_message in
|
||||
['AttributeError', 'NameError', 'AssertionError']]):
|
||||
return False
|
||||
if package_manager.lower() == 'pip' and any([em in summarized_error for em in ['ModuleNotFoundError', 'ImportError']]):
|
||||
if package_manager.lower() == 'pip' and any(
|
||||
[em in summarized_error for em in ['ModuleNotFoundError', 'ImportError']]):
|
||||
return True
|
||||
|
||||
print_colored('', f'Is it a {package_manager} dependency issue?', 'blue')
|
||||
conversation = self.gpt_session.get_conversation()
|
||||
answer = conversation.chat(
|
||||
template_is_dependency_issue.format(summarized_error=summarized_error, all_files_string=dock_req_string).replace('PACKAGE_MANAGER', package_manager)
|
||||
answer_raw = conversation.chat(
|
||||
template_is_dependency_issue.format(summarized_error=summarized_error,
|
||||
all_files_string=dock_req_string).replace('PACKAGE_MANAGER',
|
||||
package_manager)
|
||||
)
|
||||
answer_json_string = self.extract_content_from_result(answer_raw, 'response.json', match_single_block=True, )
|
||||
answer = json.loads(answer_json_string)['dependency_installation_failure']
|
||||
return 'yes' in answer.lower()
|
||||
|
||||
def generate_microservice_name(self, description):
|
||||
@@ -400,14 +417,10 @@ pytest
|
||||
description=self.microservice_specification.task
|
||||
)['strategies.json']
|
||||
packages_list = [[pkg.strip().lower() for pkg in packages] for packages in json.loads(packages_json_string)]
|
||||
packages_list = [[self.replace_with_gpt_3_5_turbo_if_possible(pkg) for pkg in packages] for packages in packages_list]
|
||||
packages_list = [[self.replace_with_gpt_3_5_turbo_if_possible(pkg) for pkg in packages] for packages in
|
||||
packages_list]
|
||||
|
||||
packages_list = [
|
||||
packages for packages in packages_list if len(set(packages).intersection(set(PROBLEMATIC_PACKAGES))) == 0
|
||||
]
|
||||
packages_list = [
|
||||
[package for package in packages if package not in UNNECESSARY_PACKAGES] for packages in packages_list
|
||||
]
|
||||
packages_list = self.filter_packages_list(packages_list)
|
||||
packages_list = packages_list[:NUM_IMPLEMENTATION_STRATEGIES]
|
||||
return packages_list
|
||||
|
||||
@@ -461,8 +474,10 @@ gptdeploy deploy --path {self.microservice_root_path}
|
||||
'task',
|
||||
'',
|
||||
template_pm_task_iteration,
|
||||
micro_service_initial_description=f'''Microservice description:
|
||||
micro_service_initial_description=f'''Microservice description:
|
||||
```
|
||||
{self.microservice_specification.task}
|
||||
```
|
||||
''',
|
||||
)
|
||||
self.refine_requirements(
|
||||
@@ -497,11 +512,14 @@ Test scenario:
|
||||
{self.microservice_specification.test}
|
||||
''')
|
||||
|
||||
def refine_requirements(self, pm, messages, refinement_type, custom_suffix, template_pm_iteration, micro_service_initial_description=None):
|
||||
def refine_requirements(self, pm, messages, refinement_type, custom_suffix, template_pm_iteration,
|
||||
micro_service_initial_description=None):
|
||||
user_input = self.microservice_specification.task
|
||||
num_parsing_tries = 0
|
||||
while True:
|
||||
conversation = self.gpt_session.get_conversation(messages, print_stream=os.environ['VERBOSE'].lower() == 'true', print_costs=False)
|
||||
conversation = self.gpt_session.get_conversation(messages,
|
||||
print_stream=os.environ['VERBOSE'].lower() == 'true',
|
||||
print_costs=False)
|
||||
agent_response_raw = conversation.chat(
|
||||
template_pm_iteration.format(
|
||||
custom_suffix=custom_suffix,
|
||||
@@ -510,22 +528,24 @@ Test scenario:
|
||||
role='user'
|
||||
)
|
||||
messages.append(HumanMessage(content=user_input))
|
||||
agent_question = self.extract_content_from_result(agent_response_raw, 'prompt.txt', can_contain_code_block=False)
|
||||
final = self.extract_content_from_result(agent_response_raw, 'final.txt', can_contain_code_block=False)
|
||||
agent_question = self.extract_content_from_result(agent_response_raw, 'prompt.json',
|
||||
can_contain_code_block=False)
|
||||
final = self.extract_content_from_result(agent_response_raw, 'final.json', can_contain_code_block=False)
|
||||
if final:
|
||||
messages.append(AIMessage(content=final))
|
||||
setattr(self.microservice_specification, refinement_type, final)
|
||||
break
|
||||
elif agent_question:
|
||||
messages.append(AIMessage(content=agent_question))
|
||||
user_input = self.get_user_input(pm, agent_question)
|
||||
question_parsed = json.loads(agent_question)['question']
|
||||
messages.append(AIMessage(content=question_parsed))
|
||||
user_input = self.get_user_input(pm, question_parsed)
|
||||
else:
|
||||
if num_parsing_tries > 2:
|
||||
raise self.TaskRefinementException()
|
||||
num_parsing_tries += 1
|
||||
messages.append(AIMessage(content=agent_response_raw))
|
||||
messages.append(SystemMessage(content='You did not put your answer into the right format using *** and ```.'))
|
||||
|
||||
messages.append(
|
||||
SystemMessage(content='You did not put your answer into the right format using *** and ```.'))
|
||||
|
||||
@staticmethod
|
||||
def get_user_input(employee, prompt_to_user):
|
||||
@@ -537,9 +557,28 @@ Test scenario:
|
||||
|
||||
@staticmethod
|
||||
def replace_with_gpt_3_5_turbo_if_possible(pkg):
|
||||
if pkg in ['allennlp', 'bertopic', 'fasttext', 'flair', 'gensim', 'nltk',
|
||||
'pattern', 'polyglot', 'pytorch-transformers', 'rasa', 'sentence-transformers',
|
||||
'spacy', 'stanza', 'textblob', 'textstat', 'transformers']:
|
||||
|
||||
return 'gpt_3_5_turbo_api'
|
||||
if pkg in LANGUAGE_PACKAGES:
|
||||
return 'gpt_3_5_turbo'
|
||||
return pkg
|
||||
|
||||
@staticmethod
|
||||
def filter_packages_list(packages_list):
|
||||
# filter out complete package lists
|
||||
packages_list = [
|
||||
packages for packages in packages_list if all([
|
||||
pkg not in BLACKLISTED_PACKAGES # no package is allowed to be blacklisted
|
||||
for pkg in packages
|
||||
])
|
||||
]
|
||||
# filter out single packages
|
||||
packages_list = [
|
||||
[
|
||||
package for package in packages
|
||||
if (package not in UNNECESSARY_PACKAGES)
|
||||
and ( # all packages must be on pypi or it is gpt_3_5_turbo
|
||||
is_package_on_pypi(package)
|
||||
or package == 'gpt_3_5_turbo'
|
||||
)
|
||||
] for packages in packages_list
|
||||
]
|
||||
return packages_list
|
||||
|
||||
@@ -6,7 +6,7 @@ RUN apt-get update && apt-get install --no-install-recommends -y {{apt_get_packa
|
||||
|
||||
## install requirements for the executor
|
||||
COPY requirements.txt .
|
||||
RUN pip install --compile -r requirements.txt
|
||||
RUN pip -v install --compile -r requirements.txt
|
||||
|
||||
# setup the workspace
|
||||
COPY . /workdir/
|
||||
|
||||
@@ -5,7 +5,7 @@ import openai
|
||||
openai.api_key = os.getenv("OPENAI_API_KEY")
|
||||
|
||||
|
||||
class GPT_3_5_Turbo_API:
|
||||
class GPT_3_5_Turbo:
|
||||
def __init__(self, system: str = ''):
|
||||
self.system = system
|
||||
|
||||
|
||||
@@ -40,19 +40,26 @@ a)
|
||||
If the description is not sufficiently specified, then ask for the missing information.
|
||||
Your response must exactly match the following block code format (double asterisks for the file name and triple backticks for the file block):
|
||||
|
||||
**prompt.txt**
|
||||
```text
|
||||
<prompt to the client here>
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "<prompt to the client here>"
|
||||
}}
|
||||
```
|
||||
|
||||
b)
|
||||
Otherwise you respond with the summarized description.
|
||||
The summarized description must contain all the information mentioned by the client.
|
||||
Otherwise you respond with the detailed description.
|
||||
The detailed description must contain all the information mentioned by the client.
|
||||
Your response must exactly match the following block code format (double asterisks for the file name and triple backticks for the file block):
|
||||
|
||||
**final.txt**
|
||||
```text
|
||||
<task here>
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"description": "<microservice description here>",
|
||||
"code_samples": "<code samples from the client here>",
|
||||
"documentation_info": "<documentation info here>",
|
||||
"credentials: "<credentials here>"
|
||||
}}
|
||||
```
|
||||
|
||||
The character sequence ``` must always be at the beginning of the line.
|
||||
@@ -64,9 +71,11 @@ output: defined
|
||||
api access: not defined
|
||||
database access: n/a
|
||||
|
||||
**prompt.txt**
|
||||
```text
|
||||
Please provide the url of the weather api and a valid api key or some other way accessing the api. Or let our engineers try to find a free api.
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "Please provide the url of the weather api and a valid api key or some other way accessing the api. Or let our engineers try to find a free api."
|
||||
}}
|
||||
```
|
||||
|
||||
Example for the description "convert png to svg":
|
||||
@@ -75,9 +84,14 @@ output: defined
|
||||
api access: n/a
|
||||
database access: n/a
|
||||
|
||||
**final.txt**
|
||||
```text
|
||||
The user inserts a png and gets an svg as response.
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"description": "The user inserts a png and gets an svg as response.",
|
||||
"code_samples": "n/a",
|
||||
"documentation_info": "n/a",
|
||||
"credentials: "n/a"
|
||||
}}
|
||||
```
|
||||
|
||||
Example for the description "parser":
|
||||
@@ -86,9 +100,11 @@ output: not defined
|
||||
api access: n/a
|
||||
database access: n/a
|
||||
|
||||
**prompt.txt**
|
||||
```text
|
||||
Please provide the input and output format.
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "Please provide the input and output format."
|
||||
}}
|
||||
```
|
||||
'''
|
||||
|
||||
@@ -104,9 +120,11 @@ Your response must exactly match the following block code format (double asteris
|
||||
1.
|
||||
contains example: no
|
||||
2.
|
||||
**prompt.txt**
|
||||
```text
|
||||
<prompt to the client here>
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "<prompt to the client here>"
|
||||
}}
|
||||
```
|
||||
|
||||
If you did a, you must not do b.
|
||||
@@ -117,10 +135,12 @@ Your response must exactly match the following block code format (double asteris
|
||||
1.
|
||||
contains example: yes (<insert example here>)
|
||||
2.
|
||||
**final.txt**
|
||||
```text
|
||||
input: "<input here>"
|
||||
assertion: the output is of type <type here>
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"input": "<input here>",
|
||||
"assertion": "the output contains the result that is of type <type here>"
|
||||
}}
|
||||
```
|
||||
|
||||
If you did b, you must not do a.
|
||||
@@ -129,37 +149,46 @@ Example for: "given a city like "Berlin", get the weather report for the next 5
|
||||
1.
|
||||
contains example: yes (Berlin)
|
||||
2.
|
||||
**final.txt**
|
||||
```text
|
||||
input: "Berlin"
|
||||
assertion: the output is of type string
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"input": "Berlin",
|
||||
"assertion": "the output is of type string"
|
||||
}}
|
||||
```
|
||||
|
||||
Example for "The user inserts a png and gets an svg as response.":
|
||||
1.
|
||||
contains example: no
|
||||
2.
|
||||
**prompt.txt**
|
||||
```text
|
||||
Please provide a png example input file as url.
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "Please provide a png example input file as url."
|
||||
}}
|
||||
```
|
||||
|
||||
|
||||
Example for "The user inserts a png like https://aquasecurity.github.io/kube-bench/v0.6.5/images/kube-bench-logo-only.png and gets an svg as response.":
|
||||
1.
|
||||
contains example: yes (https://aquasecurity.github.io/kube-bench/v0.6.5/images/kube-bench-logo-only.png)
|
||||
2.
|
||||
**final.txt**
|
||||
```text
|
||||
input: "https://aquasecurity.github.io/kube-bench/v0.6.5/images/kube-bench-logo-only.png"
|
||||
assertion: the output is of type svg
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"input": "https://aquasecurity.github.io/kube-bench/v0.6.5/images/kube-bench-logo-only.png",
|
||||
"assertion": "the output is of type svg"
|
||||
}}
|
||||
```
|
||||
|
||||
Example for "The microservice takes nothing as input and returns the current time.":
|
||||
1.
|
||||
contains example: n/a
|
||||
**final.txt**
|
||||
```text
|
||||
input: "nothing"
|
||||
assertion: the output is of type string
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"input": "nothing",
|
||||
"assertion": "the output is of type string"
|
||||
}}
|
||||
```
|
||||
'''
|
||||
|
||||
@@ -24,7 +24,8 @@ The implemented function and the test must not contain prototype or placeholder
|
||||
The implemented function and the test must run in a docker container based on debian.
|
||||
The implemented function and the test must not use libraries like Flask.
|
||||
The implemented function and the test must not have a __main__ function.
|
||||
The implemented function and the test must use gpt_3_5_turbo_api if the task requires understanding, generation, summarization, transformation, translation or classification of natural language or using any language model. Other language models are not allowed.'''
|
||||
The implemented function and the test must use gpt_3_5_turbo if the task requires the processing of language (e.g., understanding, generation, summarization, transformation, translation, modification or classification). Other language models are not allowed.
|
||||
The implemented function must follow a clean and elegant coding style.'''
|
||||
|
||||
|
||||
template_generate_microservice_name = PromptTemplate.from_template(
|
||||
@@ -49,7 +50,7 @@ list of lists wrapped into ``` and name it **strategies.json**. \
|
||||
Do not use quotation marks around packages names in the output. \
|
||||
Separate packages in a combination by comma. \
|
||||
Note that you can also leave a line empty to indicate that one of the strategies does not require any package and can be done in plain python.
|
||||
The output looks like this:
|
||||
Write the output using double asterisks and triple backticks like this:
|
||||
**strategies.json**
|
||||
```
|
||||
[
|
||||
@@ -67,8 +68,8 @@ template_generate_possible_packages = PromptTemplate.from_template(
|
||||
"{description}"
|
||||
1. Write down ut to 3 different strategies to solve the task. For each strategy write down all the non-trivial subtasks you need to solve. If there is a natural language understanding or generation stragegy, write it down.
|
||||
2. Find out what is the core problem to solve.
|
||||
3. List up to 10 Python packages that are specifically designed or have functionalities to solve the complete core problem with one of the defined strategies. You must add gpt_3_5_turbo_api if the task involves generating or understanding natural language or using a (pre-trained) language model.
|
||||
4. Exclude any package that can generate or understand natural language or enables using any language model, but you must not exclude gpt_3_5_turbo_api. Print the cleaned list of packages and give a brief reason for keeping it after its name.
|
||||
3. List up to 10 Python packages that are specifically designed or have functionalities to solve the complete core problem with one of the defined strategies. You must add gpt_3_5_turbo if the task involves generating or understanding natural language or using a (pre-trained) language model.
|
||||
4. Exclude any package that can generate or understand natural language or enables using any language model, but you must not exclude gpt_3_5_turbo. Print the cleaned list of packages and give a brief reason for keeping it after its name.
|
||||
5. For each cleaned package think if it fulfills the following requirements:
|
||||
a) specifically designed or have functionalities to solve the complete core problem.
|
||||
b) has a stable api among different versions
|
||||
@@ -81,7 +82,7 @@ When answering, just write "yes" or "no".
|
||||
|
||||
6. Determine the 5 most suitable python package combinations, ordered from the best to the least suitable. Combine the packages to achieve a comprehensive solution.
|
||||
If the package is mentioned in the description, then it is automatically the best one.
|
||||
If you listed gpt_3_5_turbo_api earlier, you must use it. gpt_3_5_turbo_api is the best package for handling text-based tasks. Also, gpt_3_5_turbo_api doesn't need any other packages processing text or using language models. It can handle any text-based task alone.
|
||||
If you listed gpt_3_5_turbo earlier, you must use it. gpt_3_5_turbo is the best package for handling text-based tasks. Also, gpt_3_5_turbo doesn't need any other packages processing text or using language models. It can handle any text-based task alone.
|
||||
|
||||
''' + template_generate_possible_packages_output_format_string)
|
||||
|
||||
@@ -94,11 +95,11 @@ template_code_wrapping_string = '''The code will go into {file_name_purpose}. Ma
|
||||
You must provide the complete file with the exact same syntax to wrap the code.'''
|
||||
|
||||
|
||||
gpt_35_turbo_usage_string = """If need to use gpt_3_5_turbo_api, then this is an example on how to use it:
|
||||
gpt_35_turbo_usage_string = """If need to use gpt_3_5_turbo, then this is an example on how to use it:
|
||||
```
|
||||
from .apis import GPT_3_5_Turbo_API
|
||||
from .apis import GPT_3_5_Turbo
|
||||
|
||||
gpt_3_5_turbo_api = GPT_3_5_Turbo_API(
|
||||
gpt_3_5_turbo = GPT_3_5_Turbo(
|
||||
system=\'\'\'
|
||||
You are a tv-reporter who is specialized in C-list celebrities.
|
||||
When you get asked something like 'Who was having a date with <X>?', then you answer with a json like '{{"dates": ["<Y>", "<Z>"]}}'.
|
||||
@@ -118,12 +119,16 @@ The function must full-fill: '{microservice_description}'.
|
||||
It will be tested with the following scenario: '{test_description}'.
|
||||
For the implementation use the following package(s): '{packages}'.
|
||||
|
||||
The code must start with the following import:
|
||||
```
|
||||
from .apis import GPT_3_5_Turbo
|
||||
```
|
||||
Obey the following rules:
|
||||
''' + not_allowed_function_string + '''
|
||||
|
||||
Your approach:
|
||||
1. Identify the core challenge when implementing the function.
|
||||
2. Think about solutions for these challenges. If gpt_3_5_turbo_api is mentioned in the above list of packages, then you must use it.
|
||||
2. Think about solutions for these challenges. If gpt_3_5_turbo is mentioned in the above list of packages, then you must use it.
|
||||
3. Decide for one of the solutions.
|
||||
4. Write the code for the function. Don't write code for the test.
|
||||
''' + gpt_35_turbo_usage_string + '\n' + template_code_wrapping_string
|
||||
@@ -136,13 +141,12 @@ template_generate_test = PromptTemplate.from_template(
|
||||
{code_files_wrapped}
|
||||
|
||||
Write a single pytest case that tests the following scenario: '{test_description}'. In case the test scenario is not precise enough, test a general case without any assumptions.
|
||||
Start the test with an extensive comment about the test case. If gpt_3_5_turbo_api is used in the executor, then the test must not check the exact output of the executor as it is not deterministic.
|
||||
Start the test with an extensive comment about the test case. If gpt_3_5_turbo is used in the executor, then the test must not check the exact output of the executor as it is not deterministic.
|
||||
|
||||
Use the following import to import the function:
|
||||
The test must start with the following import:
|
||||
```
|
||||
from .implementation import func
|
||||
```
|
||||
|
||||
''' + not_allowed_function_string + '''
|
||||
The test must not open local files.
|
||||
The test must not mock a function of the executor.
|
||||
@@ -163,7 +167,7 @@ Write the content of the requirements.txt file like this:
|
||||
...
|
||||
```
|
||||
Add any more packages that are needed to run the code.
|
||||
You must not add gpt_3_5_turbo_api to the requirements.txt file.
|
||||
You must not add gpt_3_5_turbo to the requirements.txt file.
|
||||
|
||||
All versions are fixed using ~=, ==, <, >, <=, >=. The package versions must not have conflicts. Output only the requirements.txt file.
|
||||
''' + '\n' + template_code_wrapping_string
|
||||
@@ -221,7 +225,17 @@ You are given the following files:
|
||||
|
||||
{all_files_string}
|
||||
|
||||
Is this a PACKAGE_MANAGER dependency installation failure? Answer with "yes" or "no".'''
|
||||
Is this error happening because a PACKAGE_MANAGER package is missing or failed to install?
|
||||
1. Write down one bullet point on why the error might happen because a PACKAGE_MANAGER package is missing or failed to install.
|
||||
2. Write down one bullet point on why it is unlikely that the error happens because a PACKAGE_MANAGER package is missing or failed to install.
|
||||
3. Write down your final answer.
|
||||
4. Write down your final answer as json in the following format:
|
||||
**response.json**
|
||||
```json
|
||||
{{"dependency_installation_failure": "<yes/no>"}}
|
||||
```
|
||||
Note that you must obey the double asterisk and tripple backtick syntax from above.
|
||||
'''
|
||||
)
|
||||
|
||||
|
||||
@@ -259,16 +273,33 @@ jina==2.0.0
|
||||
|
||||
|
||||
template_solve_apt_get_dependency_issue = PromptTemplate.from_template(
|
||||
'''Your task is to provide guidance on how to solve an error that occurred during the Docker build process.
|
||||
Here is the summary of the error that occurred:
|
||||
{summarized_error}
|
||||
|
||||
'''Your task is to provide guidance on how to solve an error that occurred during the Docker build process.
|
||||
You are given the following files:
|
||||
|
||||
{all_files_string}
|
||||
|
||||
Here is the summary of the error that occurred:
|
||||
{summarized_error}
|
||||
|
||||
To solve this error, you should determine the list of packages that need to be installed via `apt-get install` in the Dockerfile.
|
||||
Output them as a white space separated list:'''
|
||||
Output the apt-get packages that need to be placed at {{apt_get_packages}} as json in the following format:
|
||||
**apt-get-packages.json**
|
||||
```json
|
||||
{{"packages": ["<package1>", "<package2>"]}}
|
||||
```
|
||||
Example for the following requirements.txt file:
|
||||
**requirements.txt**
|
||||
```
|
||||
numpy==1.19.5
|
||||
fitz
|
||||
```
|
||||
The output would be:
|
||||
**apt-get-packages.json**
|
||||
```json
|
||||
{{"packages": []}}
|
||||
```
|
||||
Note that you must not output any other files. Only output the apt-get-packages.json file.
|
||||
'''
|
||||
)
|
||||
|
||||
|
||||
@@ -374,22 +405,30 @@ The playground (app.py) must not import the executor.
|
||||
template_pm_task_iteration = PromptTemplate.from_template(
|
||||
'''{micro_service_initial_description}
|
||||
1.Quickly go through the checklist (input/output well defined? api or db access needed?) and think about if you should ask something to the client or if you should write the final description.
|
||||
2.Either write the prompt.txt or the final.txt file.
|
||||
2.Either write the prompt.json or the final.json file.
|
||||
Either ask for clarification like this:
|
||||
**prompt.txt**
|
||||
```text
|
||||
<prompt to the client here (must be only one question)>
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "<prompt to the client here (must be only one question)>"
|
||||
}}
|
||||
```
|
||||
|
||||
Or write the summarized microservice description like this:
|
||||
**final.txt**
|
||||
```text
|
||||
<microservice description here>
|
||||
Or write the detailed microservice description all mentioned code samples, documentation info and credentials like this:
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"description": "<microservice description here>",
|
||||
"example_input": "<example input file or string here if mentioned before otherwise n/a>",
|
||||
"code_samples": "<code samples from the client here>",
|
||||
"documentation_info": "<documentation info here>",
|
||||
"credentials: "<credentials here>"
|
||||
}}
|
||||
```
|
||||
Note that your response must be either prompt.txt or final.txt. You must not write both.
|
||||
Note that your response must be either prompt.json or final.json. You must not write both.
|
||||
Note that you must obey the double asterisk and tripple backtick syntax from above.
|
||||
Note that the last sequence of characters in your response must be ``` (triple backtick).
|
||||
Note that prompt.txt must not only contain one question.
|
||||
Note that prompt.json must not only contain one question.
|
||||
Note that if urls, secrets, database names, etc. are mentioned, they must be part of the summary.
|
||||
{custom_suffix}
|
||||
'''
|
||||
@@ -397,36 +436,44 @@ Note that if urls, secrets, database names, etc. are mentioned, they must be par
|
||||
|
||||
template_pm_test_iteration = PromptTemplate.from_template(
|
||||
'''{micro_service_initial_description}
|
||||
1. write down if the original description and the refined description contain an example input for the microservice.
|
||||
2. write down either prompt.txt or final.txt.
|
||||
If the example input for the microservice is mentioned in the refined description or the original description, then output final.txt.
|
||||
Otherwise, output prompt.txt where you ask for the example input file as URL or the example string.
|
||||
1. write down if the microservice requires input.
|
||||
2. if it requires input, then write down if the original description or the refined description contain an example input for the microservice.
|
||||
3. write down either prompt.json or final.json.
|
||||
If the example input for the microservice is mentioned in the refined description or the original description, then output final.json.
|
||||
Otherwise, output prompt.json where you ask for the example input file as URL or the example string.
|
||||
Except for urls, you should come up with your own example input that makes sense for the microservice description.
|
||||
|
||||
Example for the case where an example input file is required and was not mentioned before:
|
||||
**prompt.txt**
|
||||
```text
|
||||
Can you please provide an example input file as URL?
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "Can you please provide an example input file as URL?"
|
||||
}}
|
||||
```
|
||||
|
||||
Example for the case where the example input string is required and was not mentioned before:
|
||||
**prompt.txt**
|
||||
```text
|
||||
Can you please provide an example input string?
|
||||
**prompt.json**
|
||||
```json
|
||||
{{
|
||||
"question": "Can you please provide an example input string?"
|
||||
}}
|
||||
```
|
||||
Note that you must not ask for an example input in case the example input is already mentioned in the refined description or the original description.
|
||||
Note that you must not ask for an example input in case the microservice does not require input.
|
||||
|
||||
Example for the case where the example is already mentioned in the refined description or the original description:
|
||||
**final.txt**
|
||||
```text
|
||||
input: <input here>
|
||||
assertion: the output is of type <type here>
|
||||
```
|
||||
Note that your response must be either prompt.txt or final.txt. You must not write both.
|
||||
**final.json**
|
||||
```json
|
||||
{{
|
||||
"input": "<input here>",
|
||||
"assertion": "the output contains the result that is of type <type here>"
|
||||
}}
|
||||
```
|
||||
Note that your response must be either prompt.json or final.json. You must not write both.
|
||||
Note that you must obey the double asterisk and tripple backtick syntax from above.
|
||||
Note that the last sequence of characters in your response must be ``` (triple backtick).
|
||||
Note that your response must start with the character sequence ** (double asterisk).
|
||||
Note that prompt.txt must only contain one question.
|
||||
Note that prompt.json must only contain one question.
|
||||
{custom_suffix}
|
||||
'''
|
||||
)
|
||||
|
||||
0
test/integration/__init__.py
Normal file
0
test/integration/__init__.py
Normal file
@@ -20,16 +20,13 @@ def test_generation_level_0(tmpdir):
|
||||
os.environ['VERBOSE'] = 'true'
|
||||
generator = Generator(
|
||||
"The microservice is very simple, it does not take anything as input and only outputs the word 'test'",
|
||||
str(tmpdir) + 'microservice',
|
||||
str(tmpdir),
|
||||
'gpt-3.5-turbo'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
|
||||
|
||||
# fixture
|
||||
@pytest.fixture
|
||||
def tmpdir():
|
||||
return 'microservice'
|
||||
|
||||
|
||||
def test_generation_level_1(tmpdir):
|
||||
"""
|
||||
@@ -47,7 +44,7 @@ def test_generation_level_1(tmpdir):
|
||||
Example tweet:
|
||||
\'When your coworker microwaves fish in the break room... AGAIN. 🐟🤢
|
||||
But hey, at least SOMEONE's enjoying their lunch. #officelife\'''',
|
||||
str(tmpdir) + 'microservice',
|
||||
str(tmpdir),
|
||||
'gpt-3.5-turbo'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
@@ -66,27 +63,50 @@ def test_generation_level_2(tmpdir):
|
||||
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).",
|
||||
str(tmpdir) + 'microservice',
|
||||
str(tmpdir),
|
||||
'gpt-3.5-turbo'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
|
||||
|
||||
@pytest.mark.skip(reason="not possible")
|
||||
def test_generation_level_3(tmpdir):
|
||||
"""
|
||||
Requirements:
|
||||
coding challenge: ✅ (calculate the average closing price)
|
||||
pip packages: ❌
|
||||
environment: ❌
|
||||
GPT-3.5-turbo: ✅ (for processing the text)
|
||||
APIs: ✅ (financial data API)
|
||||
Databases: ❌
|
||||
"""
|
||||
os.environ['VERBOSE'] = 'true'
|
||||
generator = Generator(
|
||||
f'''The input is a stock symbol (e.g., AAPL for Apple Inc.).
|
||||
1. Fetch stock data (open, high, low, close, volume) for the past 30 days using a financial data API Yahoo Finance.
|
||||
2. Calculate the average closing price over the 30 days.
|
||||
3. Generate a brief summary of the company's stock performance over the past 30 days, including the average closing price and the company name.
|
||||
4. Return the summary as a string.
|
||||
Example input: 'AAPL'
|
||||
''',
|
||||
str(tmpdir),
|
||||
'gpt-3.5-turbo'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
|
||||
def test_generation_level_4(tmpdir):
|
||||
"""
|
||||
Requirements:
|
||||
coding challenge: ❌
|
||||
pip packages: ✅ (text to speech)
|
||||
environment: ❌
|
||||
environment: ✅ (tts library)
|
||||
GPT-3.5-turbo: ✅ (summarizing the text)
|
||||
APIs: ✅ (whisper for speech to text)
|
||||
Databases: ❌
|
||||
"""
|
||||
os.environ['VERBOSE'] = 'true'
|
||||
generator = Generator(
|
||||
f'''Given an audio file of speech like https://www.signalogic.com/melp/EngSamples/Orig/ENG_M.wav,
|
||||
get convert it to text using the following api:
|
||||
f'''Given an audio file (1min wav) of speech,
|
||||
1. convert it to text using the Whisper API.
|
||||
Here is the documentation on how to use the API:
|
||||
import requests
|
||||
url = "https://transcribe.whisperapi.com"
|
||||
headers = {{
|
||||
@@ -95,30 +115,33 @@ headers = {{
|
||||
data = {{
|
||||
"url": "URL_OF_STORED_AUDIO_FILE"
|
||||
}}
|
||||
response = requests.post(url, headers=headers, files=file, data=data)
|
||||
print(response.text)
|
||||
Summarize the text.
|
||||
Create an audio file of the summarized text.
|
||||
response = requests.post(url, headers=headers, data=data)
|
||||
assert response.status_code == 200
|
||||
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(tmpdir) + 'microservice',
|
||||
'gpt-3.5-turbo'
|
||||
str(tmpdir),
|
||||
'gpt-4'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
|
||||
@pytest.mark.skip(reason="not possible")
|
||||
def test_generation_level_4(tmpdir):
|
||||
|
||||
def test_generation_level_5(tmpdir):
|
||||
"""
|
||||
Requirements:
|
||||
coding challenge: ✅ (putting text on the image)
|
||||
pip packages: ✅ (Pillow for image processing)
|
||||
environment: ❌
|
||||
environment: ✅ (image library)
|
||||
GPT-3.5-turbo: ✅ (for writing the joke)
|
||||
APIs: ✅ (scenex for image description)
|
||||
Databases: ❌
|
||||
"""
|
||||
os.environ['VERBOSE'] = 'true'
|
||||
generator = Generator(f'''
|
||||
The input is an image like this: https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/PNG_transparency_demonstration_1.png/560px-PNG_transparency_demonstration_1.png.
|
||||
The input is an image.
|
||||
Use the following api to get the description of the image:
|
||||
Request:
|
||||
curl "https://us-central1-causal-diffusion.cloudfunctions.net/describe" \\
|
||||
@@ -137,12 +160,18 @@ 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.''',
|
||||
str(tmpdir) + 'microservice',
|
||||
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(tmpdir),
|
||||
'gpt-3.5-turbo'
|
||||
)
|
||||
assert generator.generate() == 0
|
||||
|
||||
@pytest.fixture
|
||||
def tmpdir():
|
||||
return 'microservice'
|
||||
|
||||
|
||||
# further ideas:
|
||||
# Create a wrapper around google called Joogle. It modifies the page summary preview text of the search results to insert the word Jina as much as possible.
|
||||
# Create a wrapper around google called Joogle. It modifies the page summary preview text of the search results to insert the word Jina as much as possible.
|
||||
@@ -1,6 +0,0 @@
|
||||
from src.apis.jina_cloud import is_executor_in_hub
|
||||
|
||||
|
||||
def test_is_microservice_in_hub():
|
||||
assert is_executor_in_hub('reoihoflsnvoiawejeruhvflsfk') is False
|
||||
assert is_executor_in_hub('CLIPImageEncoder') is True
|
||||
0
test/unit/__init__.py
Normal file
0
test/unit/__init__.py
Normal file
34
test/unit/test_api.py
Normal file
34
test/unit/test_api.py
Normal file
@@ -0,0 +1,34 @@
|
||||
from src.apis.jina_cloud import is_executor_in_hub
|
||||
from src.apis.pypi import is_package_on_pypi
|
||||
from src.options.generate.generator import Generator
|
||||
|
||||
|
||||
def test_is_microservice_in_hub():
|
||||
assert is_executor_in_hub('reoihoflsnvoiawejeruhvflsfk') is False
|
||||
assert is_executor_in_hub('CLIPImageEncoder') is True
|
||||
|
||||
|
||||
def test_is_package_on_pypi():
|
||||
assert is_package_on_pypi('jina') is True
|
||||
assert is_package_on_pypi('jina', '0.9.25') is True
|
||||
assert is_package_on_pypi('jina', '10.10.10') is False
|
||||
assert is_package_on_pypi('jina-jina-jina') is False
|
||||
assert is_package_on_pypi('jina-jina-jina', '0.9.25') is False
|
||||
assert is_package_on_pypi('jina-jina-jina', '10.10.10') is False
|
||||
|
||||
|
||||
def test_filter_packages_list():
|
||||
filtered_list = Generator.filter_packages_list([
|
||||
["gpt_3_5_turbo", "requests", "base64", "gtts", "pydub"],
|
||||
["requests", "base64", "gtts", "pydub"],
|
||||
["gpt_3_5_turbo", "requests", "base64", "gtts"],
|
||||
["gpt_3_5_turbo", "requests", "base64", "pydub"],
|
||||
["requests", "base64", "gtts"]
|
||||
])
|
||||
assert filtered_list == [
|
||||
["gpt_3_5_turbo", "requests", "gtts", "pydub"],
|
||||
["requests", "gtts", "pydub"],
|
||||
["gpt_3_5_turbo", "requests", "gtts"],
|
||||
["gpt_3_5_turbo", "requests", "pydub"],
|
||||
["requests", "gtts"]
|
||||
]
|
||||
Reference in New Issue
Block a user