197 lines
7.5 KiB
Python
197 lines
7.5 KiB
Python
|
import os
|
||
|
|
import json
|
||
|
|
from openai import OpenAI
|
||
|
|
from pydantic import BaseModel, Field
|
||
|
|
from typing import List, Dict, Optional
|
||
|
|
from src.prompts.sops import *
|
||
|
|
from src.models.response_schemas import *
|
||
|
|
from dotenv import load_dotenv
|
||
|
|
load_dotenv()
|
||
|
|
|
||
|
|
|
||
|
|
#SopGeneratorDocument
|
||
|
|
class DocumentParser:
|
||
|
|
def __init__(self):
|
||
|
|
self.api_key = os.getenv("OPENAI_API_KEY")
|
||
|
|
self.client = OpenAI(api_key=self.api_key)
|
||
|
|
self.model = "gpt-4o-2024-08-06"
|
||
|
|
|
||
|
|
def _extract_text_from_docs(self, docs):
|
||
|
|
"""Extract text content from document objects."""
|
||
|
|
return [doc.page_content for doc in docs]
|
||
|
|
# Existing methods...
|
||
|
|
|
||
|
|
def extract_sops_from_doc(self, docs) -> VisionMissionResponse:
|
||
|
|
"""
|
||
|
|
Extracts Vision, Mission, and SOPs categorized into 'must,' 'shall,' and 'will' from the document.
|
||
|
|
|
||
|
|
:param docs: The document(s) from which to extract information.
|
||
|
|
:return: VisionMissionResponse containing the vision, mission, and role-specific SOPs.
|
||
|
|
"""
|
||
|
|
|
||
|
|
try:
|
||
|
|
docs_text = self._extract_text_from_docs(docs)
|
||
|
|
prompt = get_sop_extraction_from_doc()
|
||
|
|
response = self.client.beta.chat.completions.parse(
|
||
|
|
model=self.model,
|
||
|
|
messages=[
|
||
|
|
{
|
||
|
|
"role": "system",
|
||
|
|
"content": f'''{prompt}'''
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"role": "user",
|
||
|
|
"content": [{"type": "text", "text": text} for text in docs_text],
|
||
|
|
}
|
||
|
|
],
|
||
|
|
response_format=SOPsResponse,
|
||
|
|
max_tokens=4096,
|
||
|
|
temperature=0.1
|
||
|
|
)
|
||
|
|
|
||
|
|
# Parse the response from the LLM
|
||
|
|
extracted_text = json.loads(response.choices[0].message.content)
|
||
|
|
|
||
|
|
return extracted_text
|
||
|
|
|
||
|
|
except:
|
||
|
|
return False
|
||
|
|
|
||
|
|
def extract_vision_mission(self, docs) -> VisionMissionResponse:
|
||
|
|
"""
|
||
|
|
Extracts Vision, Mission, and SOPs categorized into 'must,' 'shall,' and 'will' from the document.
|
||
|
|
|
||
|
|
:param docs: The document(s) from which to extract information.
|
||
|
|
:return: VisionMissionResponse containing the vision, mission, and role-specific SOPs.
|
||
|
|
"""
|
||
|
|
|
||
|
|
try:
|
||
|
|
docs_text = self._extract_text_from_docs(docs)
|
||
|
|
prompt = get_vision_mission_extraction_from_doc()
|
||
|
|
response = self.client.beta.chat.completions.parse(
|
||
|
|
model=self.model,
|
||
|
|
messages=[
|
||
|
|
{
|
||
|
|
"role": "system",
|
||
|
|
"content": f'''{prompt}'''
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"role": "user",
|
||
|
|
"content": [{"type": "text", "text": text} for text in docs_text],
|
||
|
|
}
|
||
|
|
],
|
||
|
|
response_format=VisionMissionResponse,
|
||
|
|
max_tokens=4096,
|
||
|
|
temperature=0.1
|
||
|
|
)
|
||
|
|
|
||
|
|
# Parse the response from the LLM
|
||
|
|
extracted_text = json.loads(response.choices[0].message.content)
|
||
|
|
|
||
|
|
return extracted_text
|
||
|
|
|
||
|
|
except:
|
||
|
|
return False
|
||
|
|
|
||
|
|
'''def extract_departments_and_managers(self, docs):
|
||
|
|
"""
|
||
|
|
Extract departments and managerial roles from the document.
|
||
|
|
|
||
|
|
:param docs: List of document chunks
|
||
|
|
:return: Dictionary containing departments and their managerial roles
|
||
|
|
"""
|
||
|
|
try:
|
||
|
|
docs_text = self._extract_text_from_docs(docs)
|
||
|
|
prompt = get_departments_and_roles_extraction_prompt()
|
||
|
|
|
||
|
|
response = self.client.beta.chat.completions.parse(
|
||
|
|
model=self.model,
|
||
|
|
messages=[
|
||
|
|
{"role": "system", "content": prompt},
|
||
|
|
{"role": "user", "content": [{"type": "text", "text": text} for text in docs_text]}
|
||
|
|
],
|
||
|
|
response_format=DepartmentsAndRolesResponse,
|
||
|
|
max_tokens=4096,
|
||
|
|
temperature=0.1
|
||
|
|
)
|
||
|
|
|
||
|
|
|
||
|
|
return json.loads(response.choices[0].message.content)
|
||
|
|
except json.JSONDecodeError:
|
||
|
|
return False'''
|
||
|
|
|
||
|
|
|
||
|
|
def extract_departments_and_managers_workers(self, docs):
|
||
|
|
"""
|
||
|
|
Extract departments, managers, and workers from the document.
|
||
|
|
|
||
|
|
:param docs: List of document chunks
|
||
|
|
:return: Dictionary containing departments, their managers, and workers.
|
||
|
|
"""
|
||
|
|
try:
|
||
|
|
docs_text = self._extract_text_from_docs(docs)
|
||
|
|
prompt = get_departments_managers_workers_extraction_prompt() # Update your prompt to handle managers and workers
|
||
|
|
|
||
|
|
response = self.client.beta.chat.completions.parse(
|
||
|
|
model=self.model,
|
||
|
|
messages=[
|
||
|
|
{"role": "system", "content": prompt},
|
||
|
|
{"role": "user", "content": [{"type": "text", "text": text} for text in docs_text]}
|
||
|
|
],
|
||
|
|
response_format=DepartmentsAndWorkersResponse, # Use the updated response schema
|
||
|
|
max_tokens=4096,
|
||
|
|
temperature=0.1
|
||
|
|
)
|
||
|
|
|
||
|
|
return json.loads(response.choices[0].message.content)
|
||
|
|
|
||
|
|
except json.JSONDecodeError:
|
||
|
|
return False
|
||
|
|
|
||
|
|
|
||
|
|
def generate_sops_(self, docs, reference_roles):
|
||
|
|
try:
|
||
|
|
# First, extract departments and managers from the document
|
||
|
|
sop_doc = DocumentParser()
|
||
|
|
departments_and_roles = sop_doc.extract_departments_and_managers(docs)
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
# Prepare extracted roles (only managers)
|
||
|
|
extracted_managers = []
|
||
|
|
for department in departments_and_roles['departments']:
|
||
|
|
extracted_managers.extend([
|
||
|
|
{
|
||
|
|
'name': manager['name'],
|
||
|
|
'position': manager.get('title', 'Unknown Position'),
|
||
|
|
'role': manager.get('classification', 'Unknown Role') # PRP or SRP classification
|
||
|
|
}
|
||
|
|
for manager in department['managerial_roles']
|
||
|
|
])
|
||
|
|
|
||
|
|
# Generate prompt for the LLM to compare reference roles with extracted roles
|
||
|
|
prompt = generate_llm_comparison_prompt(reference_roles, extracted_managers)
|
||
|
|
|
||
|
|
# Send prompt to the LLM for comparison
|
||
|
|
response = self.client.beta.chat.completions.parse(
|
||
|
|
model=self.model,
|
||
|
|
messages=[
|
||
|
|
{"role": "system", "content": "You are a role comparison assistant."},
|
||
|
|
{"role": "user", "content": prompt}
|
||
|
|
],
|
||
|
|
max_tokens=1024,
|
||
|
|
temperature=0.1
|
||
|
|
)
|
||
|
|
|
||
|
|
# Parse LLM response (assuming it returns a structured JSON with assigned, unassigned, and unavailable roles)
|
||
|
|
comparison_result = json.loads(response.choices[0].message.content)
|
||
|
|
|
||
|
|
# Return the result as a JSON response
|
||
|
|
return jsonify(comparison_result), 200
|
||
|
|
|
||
|
|
except Exception as e:
|
||
|
|
return jsonify({"error": "Processing error", "message": f"An error occurred: {str(e)}"}), 500
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
|