Spaces:
Paused
Paused
File size: 10,125 Bytes
4c945e0 6a2a662 4c945e0 6183e7b 4c945e0 6183e7b 4ba3568 ba1c332 6183e7b ba1c332 4ba3568 ba1c332 4ba3568 ba1c332 6183e7b ba1c332 dbc1dab 4c945e0 ba1c332 4c945e0 4ba3568 4c945e0 4ba3568 4c945e0 6183e7b 4c945e0 6a2a662 4c945e0 6183e7b 6a2a662 4c945e0 6183e7b 4c945e0 6183e7b ddb005f 6183e7b 6a2a662 4c945e0 6183e7b 4c945e0 6a2a662 4c945e0 6183e7b 4c945e0 6a2a662 4c945e0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 | import concurrent.futures
import time
import os
import json
import re
import logging
from typing import List, Dict, Any
import tinytroupe
from tinytroupe.agent import TinyPerson
from tinytroupe.factory import TinyPersonFactory
from tinytroupe.extraction import ResultsExtractor
from backend.core.config import settings
from backend.services.persona_matcher import persona_matcher
logger = logging.getLogger(__name__)
class TinyTroupeSimulationManager:
def __init__(self):
self.max_concurrency = settings.MAX_CONCURRENCY
self._executor = concurrent.futures.ThreadPoolExecutor(max_workers=self.max_concurrency)
def generate_personas_async(self, business_description: str, customer_profile: str, num_personas: int, job_id: str, job_registry):
try:
# 1. First, check if there are any matching personas in the local /app/personas repo
matched_personas = persona_matcher.scan_tresor(business_description, customer_profile)
# Filter matches > 85% assureness score
valid_personas = [p for p in matched_personas if p.get("_assureness_score", 0) >= 85]
# Number of personas we still need to generate
missing_count = max(0, num_personas - len(valid_personas))
logger.info(f"Job {job_id}: Found {len(valid_personas)} matching personas. Generating {missing_count} new personas.")
job_registry.update_job(job_id, progress_percentage=20)
new_personas = []
# 2. Generate missing personas via TinyTroupe LLM call
if missing_count > 0:
try:
# Utilize the TinyPersonFactory dynamic population pattern
# Utilize the custom pipeline schema pattern for structured creation and validation
# Utilize the TinyPersonFactory dynamic population pattern natively with parallel generation
from tinytroupe.factory import TinyPersonFactory
factory = TinyPersonFactory(
sampling_space_description=customer_profile,
total_population_size=missing_count,
context=business_description
)
logger.info(f"Job {job_id}: Generating {missing_count} personas via TinyPersonFactory with parallelize=True...")
people = factory.generate_people(missing_count, parallelize=True)
for i, person in enumerate(people):
if person is not None and getattr(person, '_persona', None) is not None:
persona_data = person._persona
persona_data["_assureness_score"] = 100 # New ones are perfectly matched to the description
new_personas.append(persona_data)
# Safe filename parsing
safe_name = person.name.lower().strip()
safe_name = re.sub(r"[^\w\s-]", "", safe_name)
safe_name = re.sub(r"[\s-]+", "_", safe_name)[:60]
# Save to local file system for git sync
local_dir = "/app/personas"
os.makedirs(local_dir, exist_ok=True)
file_path = os.path.join(local_dir, f"{safe_name}.json")
with open(file_path, "w") as f:
json.dump(persona_data, f, indent=4)
job_registry.update_job(job_id, progress_percentage=20 + int((i+1)/missing_count * 60))
except Exception as e:
logger.error(f"Error during persona generation: {e}")
job_registry.update_job(job_id, status="FAILED", message=f"LLM Error: {str(e)}")
return
# Combine
all_personas = valid_personas[:num_personas] + new_personas
all_personas = all_personas[:num_personas] # Ensure we don't exceed the requested count
# Push new personas
if new_personas:
from backend.services.git_sync import git_sync
git_sync.background_push(commit_message=f"Added {len(new_personas)} new personas for job {job_id}")
job_registry.update_job(
job_id,
status="COMPLETED",
progress_percentage=100,
results={"personas": all_personas}
)
except Exception as e:
logger.error(f"Job {job_id} failed: {e}")
job_registry.update_job(job_id, status="FAILED", message=str(e))
def run_simulation_async(self, job_id: str, content_text: str, personas_data: List[Dict[str, Any]], format_type: str, parameters: Dict[str, Any], job_registry):
try:
job_registry.update_job(job_id, progress_percentage=10, status="RUNNING")
# Instantiate TinyPersons
persons = []
for p_data in personas_data:
try:
p = TinyPerson(name=p_data.get("name", "Unknown"))
# Make sure TinyPerson uses the loaded dictionary for the _persona structure
p._persona = p_data
persons.append(p)
except Exception as e:
logger.error(f"Failed to load person data: {e}")
if not persons:
job_registry.update_job(job_id, status="FAILED", message="No valid personas provided")
return
results = []
# Run in parallel using the ThreadPoolExecutor
def process_person(person: TinyPerson, index: int):
try:
# Send prompt context
prompt = f"Please read this {format_type}:\n\n'{content_text}'"
person.listen_and_act(prompt)
# Use TinyTroupe native ResultsExtractor
extractor = ResultsExtractor()
objective = "Rate the impact, attention, and relevance of the content based on the agent's background. Ratings must be integer scores between 0 and 100. Also provide a detailed analytical comment."
extracted_data = extractor.extract_results_from_agent(
person,
extraction_objective=objective,
situation=f"Testing a new {format_type}",
fields=["impact_score", "attention", "relevance", "comment"],
fields_hints={
"impact_score": "Integer between 0 and 100",
"attention": "Integer between 0 and 100",
"relevance": "Integer between 0 and 100",
"comment": "A string containing a descriptive paragraph"
},
verbose=False
)
# Safe parsing fallback
if extracted_data is None or type(extracted_data) is not dict:
extracted_data = {}
extracted_data = {}
parsed_response = {
"name": person.name,
"impact_score": int(extracted_data.get("impact_score", 50)),
"attention": int(extracted_data.get("attention", 50)),
"relevance": int(extracted_data.get("relevance", 50)),
"comment": extracted_data.get("comment", "No comment provided.")
}
# Update progress
current_prog = job_registry.get_job(job_id).get("progress_percentage", 10)
progress_increment = max(1, 80 // len(persons))
job_registry.update_job(job_id, progress_percentage=current_prog + progress_increment)
return parsed_response
except Exception as e:
logger.error(f"Person {person.name} failed to process: {e}")
return {"name": person.name, "error": str(e)}
futures = [self._executor.submit(process_person, p, i) for i, p in enumerate(persons)]
for future in concurrent.futures.as_completed(futures):
results.append(future.result())
# Aggregate results
total_impact = sum(r.get("impact_score", 0) for r in results if "error" not in r)
total_attention = sum(r.get("attention", 0) for r in results if "error" not in r)
total_relevance = sum(r.get("relevance", 0) for r in results if "error" not in r)
valid_count = len([r for r in results if "error" not in r])
agg_results = {
"impact_score": total_impact // valid_count if valid_count else 0,
"attention": total_attention // valid_count if valid_count else 0,
"relevance": total_relevance // valid_count if valid_count else 0,
"key_insights": [r.get("comment") for r in results if "error" not in r][:3],
"agent_dialogue": results
}
job_registry.update_job(
job_id,
status="COMPLETED",
progress_percentage=100,
results=agg_results
)
except Exception as e:
logger.error(f"Simulation Job {job_id} failed: {e}")
job_registry.update_job(job_id, status="FAILED", message=str(e))
tinytroupe_manager = TinyTroupeSimulationManager()
|