mirror of
https://github.com/aljazceru/dev-gpt.git
synced 2025-12-21 15:44:19 +01:00
fix: stable version
This commit is contained in:
31
main.py
31
main.py
@@ -6,18 +6,11 @@ from jina import Client
|
|||||||
from src import gpt, jina_cloud
|
from src import gpt, jina_cloud
|
||||||
from src.constants import TAG_TO_FILE_NAME, EXECUTOR_FOLDER
|
from src.constants import TAG_TO_FILE_NAME, EXECUTOR_FOLDER
|
||||||
from src.prompt_examples import executor_example, docarray_example
|
from src.prompt_examples import executor_example, docarray_example
|
||||||
from src.prompt_tasks import general_guidelines, executor_name_task, executor_file_task, requirements_file_task, \
|
from src.prompt_tasks import general_guidelines, executor_file_task, requirements_file_task, \
|
||||||
test_executor_file_task, docker_file_task
|
test_executor_file_task, docker_file_task
|
||||||
from src.utils.io import recreate_folder
|
from src.utils.io import recreate_folder
|
||||||
from src.utils.string import find_between, clean_content
|
from src.utils.string import find_between, clean_content
|
||||||
|
|
||||||
input_executor_description = "Write an executor that takes image bytes as input (document.blob within a DocumentArray) and use BytesIO to convert it to PIL and detects ocr " \
|
|
||||||
"and returns the texts as output (as DocumentArray). "
|
|
||||||
|
|
||||||
input_test_description = 'The test downloads the image ' \
|
|
||||||
'https://upload.wikimedia.org/wikipedia/commons/thumb/a/aa/Onlineocr.png/640px-Onlineocr.png ' \
|
|
||||||
' loads it as bytes, takes it as input to the executor and asserts that the output is "Hello World".'
|
|
||||||
|
|
||||||
|
|
||||||
def extract_content_from_result(plain_text, tag):
|
def extract_content_from_result(plain_text, tag):
|
||||||
content = find_between(plain_text, f'$$$start_{tag}$$$', f'$$$end_{tag}$$$')
|
content = find_between(plain_text, f'$$$start_{tag}$$$', f'$$$end_{tag}$$$')
|
||||||
@@ -45,7 +38,7 @@ def write_config_yml(executor_name):
|
|||||||
f.write(config_content)
|
f.write(config_content)
|
||||||
|
|
||||||
|
|
||||||
def main():
|
def main(executor_name, input_executor_description, input_test_description):
|
||||||
recreate_folder(EXECUTOR_FOLDER)
|
recreate_folder(EXECUTOR_FOLDER)
|
||||||
system_definition = (
|
system_definition = (
|
||||||
"You are a principal engineer working at Jina - an open source company."
|
"You are a principal engineer working at Jina - an open source company."
|
||||||
@@ -57,9 +50,8 @@ def main():
|
|||||||
user_query = (
|
user_query = (
|
||||||
input_executor_description
|
input_executor_description
|
||||||
+ general_guidelines
|
+ general_guidelines
|
||||||
+ executor_name_task()
|
|
||||||
+ executor_file_task()
|
+ executor_file_task()
|
||||||
+ test_executor_file_task()
|
+ test_executor_file_task(executor_name)
|
||||||
+ requirements_file_task()
|
+ requirements_file_task()
|
||||||
+ docker_file_task()
|
+ docker_file_task()
|
||||||
+ input_test_description
|
+ input_test_description
|
||||||
@@ -68,8 +60,6 @@ def main():
|
|||||||
plain_text = gpt.get_response(system_definition, user_query)
|
plain_text = gpt.get_response(system_definition, user_query)
|
||||||
extract_and_write(plain_text)
|
extract_and_write(plain_text)
|
||||||
|
|
||||||
executor_name = extract_content_from_result(plain_text, 'executor_name')
|
|
||||||
|
|
||||||
write_config_yml(executor_name)
|
write_config_yml(executor_name)
|
||||||
|
|
||||||
jina_cloud.push_executor()
|
jina_cloud.push_executor()
|
||||||
@@ -85,4 +75,17 @@ def main():
|
|||||||
|
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
main()
|
main(
|
||||||
|
executor_name='MyBelovedOcrExecutor',
|
||||||
|
input_executor_description=(
|
||||||
|
"Write an executor that takes image bytes as input (document.blob within a DocumentArray) "
|
||||||
|
# "and use BytesIO to convert it to PIL " \
|
||||||
|
"and detects ocr "
|
||||||
|
"and returns the texts as output (as DocumentArray). "
|
||||||
|
),
|
||||||
|
|
||||||
|
input_test_description='The test downloads the image ' \
|
||||||
|
'https://miro.medium.com/v2/resize:fit:1024/0*4ty0Adbdg4dsVBo3.png ' \
|
||||||
|
' loads it as bytes, takes it as input to the executor and asserts that the output is "> Hello, world!_".',
|
||||||
|
|
||||||
|
)
|
||||||
|
|||||||
@@ -1 +1,4 @@
|
|||||||
jina[perf]==3.14.2.dev18
|
jina[perf]==3.14.2.dev18
|
||||||
|
openai
|
||||||
|
ptest
|
||||||
|
jcloud
|
||||||
@@ -1,15 +1,21 @@
|
|||||||
import os
|
import os
|
||||||
|
|
||||||
|
import hubble
|
||||||
from jcloud.flow import CloudFlow
|
from jcloud.flow import CloudFlow
|
||||||
|
from jina import Flow
|
||||||
|
|
||||||
|
|
||||||
def push_executor():
|
def push_executor():
|
||||||
cmd = 'jina hub push executor/. --verbose'
|
cmd = 'jina hub push executor/. --verbose'
|
||||||
os.system(cmd)
|
os.system(cmd)
|
||||||
|
|
||||||
|
def get_user_name():
|
||||||
|
client = hubble.Client(max_retries=None, jsonify=True)
|
||||||
|
response = client.get_user_info()
|
||||||
|
return response['data']['name']
|
||||||
|
|
||||||
|
|
||||||
def deploy_flow(executor_name):
|
def deploy_flow(executor_name):
|
||||||
|
|
||||||
flow = f'''
|
flow = f'''
|
||||||
jtype: Flow
|
jtype: Flow
|
||||||
with:
|
with:
|
||||||
@@ -19,13 +25,13 @@ def deploy_flow(executor_name):
|
|||||||
jcloud:
|
jcloud:
|
||||||
version: '3.14.2.dev18'
|
version: '3.14.2.dev18'
|
||||||
labels:
|
labels:
|
||||||
team: now
|
team: microchain
|
||||||
gateway:
|
gateway:
|
||||||
jcloud:
|
jcloud:
|
||||||
expose: true
|
expose: true
|
||||||
executors:
|
executors:
|
||||||
- name: {executor_name.lower()}
|
- name: {executor_name.lower()}
|
||||||
uses: jinaai+docker://team-now-prod/{executor_name}
|
uses: jinaai+docker://{get_user_name()}/{executor_name}:latest
|
||||||
env:
|
env:
|
||||||
JINA_LOG_LEVEL: DEBUG
|
JINA_LOG_LEVEL: DEBUG
|
||||||
jcloud:
|
jcloud:
|
||||||
@@ -39,4 +45,9 @@ def deploy_flow(executor_name):
|
|||||||
with open(full_flow_path, 'w') as f:
|
with open(full_flow_path, 'w') as f:
|
||||||
f.write(flow)
|
f.write(flow)
|
||||||
|
|
||||||
|
# try local first
|
||||||
|
flow = Flow.load_config(full_flow_path)
|
||||||
|
with flow:
|
||||||
|
pass
|
||||||
|
|
||||||
return CloudFlow(path=full_flow_path).__enter__().endpoints['gateway']
|
return CloudFlow(path=full_flow_path).__enter__().endpoints['gateway']
|
||||||
@@ -1,8 +1,9 @@
|
|||||||
executor_example = "Here is an example of how an executor can be defined. It always starts with a comment:"
|
executor_example = "Here is an example of how an executor can be defined. It always starts with a comment:"
|
||||||
'''
|
'''
|
||||||
|
|
||||||
# this executor takes ... as input and returns ... as output
|
# this executor takes ... as input and returns ... as output
|
||||||
# it processes each document in the following way: ...
|
# it processes each document in the following way: ...
|
||||||
from jina import Executor, requests, DocumentArray, Document, Deployment
|
from jina import Executor, requests, DocumentArray, Document
|
||||||
class MyExecutor(Executor):
|
class MyExecutor(Executor):
|
||||||
def __init__(self, **kwargs):
|
def __init__(self, **kwargs):
|
||||||
super().__init__()
|
super().__init__()
|
||||||
@@ -15,8 +16,47 @@ class MyExecutor(Executor):
|
|||||||
'''
|
'''
|
||||||
"An executor gets a DocumentArray as input and returns a DocumentArray as output."
|
"An executor gets a DocumentArray as input and returns a DocumentArray as output."
|
||||||
|
|
||||||
docarray_example = "Here is an example of how a DocumentArray can be defined:"
|
docarray_example = (
|
||||||
|
"A DocumentArray is a python class that can be seen as a list of Documents. "
|
||||||
|
"A Document is a python class that represents a single document. "
|
||||||
|
"Here is the protobuf definition of a Document: "
|
||||||
'''
|
'''
|
||||||
|
|
||||||
|
message DocumentProto {
|
||||||
|
// A hexdigest that represents a unique document ID
|
||||||
|
string id = 1;
|
||||||
|
|
||||||
|
oneof content {
|
||||||
|
// the raw binary content of this document, which often represents the original document when comes into jina
|
||||||
|
bytes blob = 2;
|
||||||
|
|
||||||
|
// the ndarray of the image/audio/video document
|
||||||
|
NdArrayProto tensor = 3;
|
||||||
|
|
||||||
|
// a text document
|
||||||
|
string text = 4;
|
||||||
|
}
|
||||||
|
|
||||||
|
// a uri of the document could be: a local file path, a remote url starts with http or https or data URI scheme
|
||||||
|
string uri = 5;
|
||||||
|
|
||||||
|
// list of the sub-documents of this document (recursive structure)
|
||||||
|
repeated DocumentProto chunks = 6;
|
||||||
|
|
||||||
|
// the matched documents on the same level (recursive structure)
|
||||||
|
repeated DocumentProto matches = 7;
|
||||||
|
|
||||||
|
// the embedding of this document
|
||||||
|
NdArrayProto embedding = 8;
|
||||||
|
|
||||||
|
// a structured data value, consisting of field which map to dynamically typed values.
|
||||||
|
google.protobuf.Struct tags = 9;
|
||||||
|
|
||||||
|
}
|
||||||
|
'''
|
||||||
|
"Here is an example of how a DocumentArray can be defined: "
|
||||||
|
'''
|
||||||
|
|
||||||
from jina import DocumentArray, Document
|
from jina import DocumentArray, Document
|
||||||
|
|
||||||
d1 = Document(text='hello')
|
d1 = Document(text='hello')
|
||||||
@@ -26,8 +66,12 @@ d4 = Document(
|
|||||||
uri='https://docs.docarray.org',
|
uri='https://docs.docarray.org',
|
||||||
tags={'foo': 'bar'},
|
tags={'foo': 'bar'},
|
||||||
)
|
)
|
||||||
|
d5 = Document()
|
||||||
|
d5.tensor = np.ones((2,4))
|
||||||
|
d6 = Document()
|
||||||
|
d6.blob = b'RIFF\x00\x00\x00\x00WAVEfmt \x10\x00...'
|
||||||
docs = DocumentArray([
|
docs = DocumentArray([
|
||||||
d1, d2, d3, d4
|
d1, d2, d3, d4
|
||||||
])
|
])
|
||||||
'''
|
'''
|
||||||
|
)
|
||||||
@@ -9,30 +9,31 @@ general_guidelines = (
|
|||||||
)
|
)
|
||||||
|
|
||||||
|
|
||||||
def _task(task, tag_name, file_name=None):
|
def _task(task, tag_name, file_name):
|
||||||
return task + f"{f'The code will go into {file_name}. ' if tag_name else ''}. Wrap the code in the string $$$start_{tag_name}$$$...$$$end_{tag_name}$$$ "
|
return task + f"The code will go into {file_name}. Wrap the code in the string $$$start_{tag_name}$$$...$$$end_{tag_name}$$$ "
|
||||||
|
|
||||||
|
|
||||||
def executor_name_task():
|
|
||||||
return _task("Write the executor name. "
|
|
||||||
"The executor name only consists of lower case and upper case letters.", 'executor_name'
|
|
||||||
)
|
|
||||||
|
|
||||||
|
|
||||||
def executor_file_task():
|
def executor_file_task():
|
||||||
return _task("Write the executor code. ", 'executor', EXECUTOR_FILE_NAME)
|
return _task("Write the executor code. ", 'executor', EXECUTOR_FILE_NAME)
|
||||||
|
|
||||||
|
|
||||||
def requirements_file_task():
|
def requirements_file_task():
|
||||||
return _task("Write the content of the requirements.txt file. Make sure to include pytest.", 'requirements',
|
return _task("Write the content of the requirements.txt file. "
|
||||||
|
"Make sure to include pytest. "
|
||||||
|
"All versions are fixed. " , 'requirements',
|
||||||
REQUIREMENTS_FILE_NAME)
|
REQUIREMENTS_FILE_NAME)
|
||||||
|
|
||||||
|
|
||||||
def test_executor_file_task():
|
def test_executor_file_task(executor_name):
|
||||||
return _task(
|
return _task(
|
||||||
"Write a small unit test for the executor. "
|
"Write a small unit test for the executor. "
|
||||||
"Start the test with an extensive comment about the test case. "
|
"Start the test with an extensive comment about the test case. "
|
||||||
"Never do relative imports.", 'test_executor', TEST_EXECUTOR_FILE_NAME)
|
"Use the following import to import the executor: "
|
||||||
|
f"from executor import {executor_name}",
|
||||||
|
'test_executor',
|
||||||
|
TEST_EXECUTOR_FILE_NAME
|
||||||
|
)
|
||||||
|
|
||||||
|
|
||||||
def docker_file_task():
|
def docker_file_task():
|
||||||
@@ -40,6 +41,8 @@ def docker_file_task():
|
|||||||
"Write the Dockerfile that defines the environment with all necessary dependencies that the executor uses. "
|
"Write the Dockerfile that defines the environment with all necessary dependencies that the executor uses. "
|
||||||
"The Dockerfile runs the test during the build process. "
|
"The Dockerfile runs the test during the build process. "
|
||||||
"It is important to make sure that all libs are installed that are required by the python packages. "
|
"It is important to make sure that all libs are installed that are required by the python packages. "
|
||||||
|
"Usually libraries are installed with apt-get. "
|
||||||
|
"Add the config.yml file to the Dockerfile. "
|
||||||
"The base image of the Dockerfile is FROM jinaai/jina:3.14.2-dev18-py310-standard. "
|
"The base image of the Dockerfile is FROM jinaai/jina:3.14.2-dev18-py310-standard. "
|
||||||
'The entrypoint is ENTRYPOINT ["jina", "executor", "--uses", "config.yml"] '
|
'The entrypoint is ENTRYPOINT ["jina", "executor", "--uses", "config.yml"] '
|
||||||
"The Dockerfile runs the test during the build process. "
|
"The Dockerfile runs the test during the build process. "
|
||||||
|
|||||||
Reference in New Issue
Block a user