Squash 4 commits into one

prediction_prophet/autonolas/research.py

@@ -2,6 +2,7 @@
 import os
 import math
 import tenacity
+from datetime import timedelta
 from sklearn.metrics.pairwise import cosine_similarity
 from typing import Any, Dict, Generator, List, Optional, Tuple, TypedDict
 from datetime import datetime, timezone
@@ -32,7 +33,7 @@
 from dateutil import parser
 from prediction_prophet.functions.utils import check_not_none
 from prediction_market_agent_tooling.gtypes import Probability
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from prediction_market_agent_tooling.tools.caches.db_cache import db_cache
 from prediction_prophet.functions.parallelism import par_map
 from pydantic.types import SecretStr
@@ -358,7 +359,7 @@ def fields_dict_to_bullet_list(fields_dict: Dict[str, str]) -> str:
     return bullet_list
 
 @tenacity.retry(stop=tenacity.stop_after_attempt(3), wait=tenacity.wait_fixed(1), reraise=True)
-@db_cache
+@db_cache(max_age=timedelta(days=1))
 def search_google(query: str, num: int = 3) -> List[str]:
     """Search Google using a custom search engine."""
     service = build("customsearch", "v1", developerKey=os.getenv("GOOGLE_SEARCH_API_KEY"))
@@ -1220,7 +1221,7 @@ def make_prediction(
     api_key: SecretStr | None = None,
 ) -> Prediction:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     current_time_utc = datetime.now(timezone.utc)
     formatted_time_utc = current_time_utc.strftime("%Y-%m-%dT%H:%M:%S.%f")[:-6] + "Z"

prediction_prophet/benchmark/agents.py

@@ -11,7 +11,6 @@
 from prediction_prophet.autonolas.research import EmbeddingModel
 from prediction_prophet.autonolas.research import make_prediction, get_urls_from_queries
 from prediction_prophet.autonolas.research import research as research_autonolas
-from prediction_market_agent_tooling.tools.is_predictable import is_predictable_binary
 from prediction_prophet.functions.rephrase_question import rephrase_question
 from prediction_prophet.functions.research import Research, research as prophet_research
 from prediction_prophet.functions.search import search
@@ -26,6 +25,7 @@
 from pydantic.types import SecretStr
 from prediction_prophet.autonolas.research import Prediction as LLMCompletionPredictionDict
 from prediction_market_agent_tooling.tools.langfuse_ import observe
+from prediction_market_agent_tooling.tools.is_predictable import is_predictable_binary
 
 if t.TYPE_CHECKING:
     from loguru import Logger

prediction_prophet/functions/create_embeddings_from_results.py

@@ -13,13 +13,13 @@
 from prediction_prophet.models.WebScrapeResult import WebScrapeResult
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from pydantic.types import SecretStr
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from prediction_market_agent_tooling.gtypes import secretstr_to_v1_secretstr
 
 
 def create_embeddings_from_results(results: list[WebScrapeResult], text_splitter: RecursiveCharacterTextSplitter, api_key: SecretStr | None = None) -> Chroma:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     collection = Chroma(embedding_function=OpenAIEmbeddings(api_key=secretstr_to_v1_secretstr(api_key)))
     texts = []

prediction_prophet/functions/debate_prediction.py

@@ -9,7 +9,7 @@
 from langchain.schema.output_parser import StrOutputParser
 from langchain_openai import ChatOpenAI
 from langchain.prompts import ChatPromptTemplate
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from prediction_market_agent_tooling.gtypes import secretstr_to_v1_secretstr
 
 
@@ -85,7 +85,7 @@
 
 def make_debated_prediction(prompt: str, additional_information: str, api_key: SecretStr | None = None) -> Prediction:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     formatted_time_utc = datetime.datetime.now(datetime.timezone.utc).isoformat(timespec='seconds') + "Z"
 

prediction_prophet/functions/generate_subqueries.py

@@ -3,7 +3,7 @@
 from langchain.output_parsers import CommaSeparatedListOutputParser
 from langchain.prompts import ChatPromptTemplate
 from pydantic.types import SecretStr
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from prediction_market_agent_tooling.gtypes import secretstr_to_v1_secretstr
 from prediction_market_agent_tooling.tools.langfuse_ import get_langfuse_langchain_config, observe
 
@@ -22,7 +22,7 @@ def generate_subqueries(query: str, limit: int, model: str, temperature: float,
         return [query]
 
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     subquery_generation_prompt = ChatPromptTemplate.from_template(template=subquery_generation_template)
 

prediction_prophet/functions/prepare_report.py

@@ -5,15 +5,15 @@
 from langchain.prompts import ChatPromptTemplate
 from langchain.schema.output_parser import StrOutputParser
 from prediction_prophet.functions.utils import trim_to_n_tokens
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from pydantic.types import SecretStr
 from prediction_market_agent_tooling.gtypes import secretstr_to_v1_secretstr
 from prediction_market_agent_tooling.tools.langfuse_ import get_langfuse_langchain_config, observe
 
 @observe()
 def prepare_summary(goal: str, content: str, model: str, api_key: SecretStr | None = None, trim_content_to_tokens: t.Optional[int] = None) -> str:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     prompt_template = """Write comprehensive summary of the following web content, that provides relevant information to answer the question: '{goal}'.
 But cut the fluff and keep it up to the point.
@@ -43,7 +43,7 @@ def prepare_summary(goal: str, content: str, model: str, api_key: SecretStr | No
 @observe()
 def prepare_report(goal: str, scraped: list[str], model: str, temperature: float, api_key: SecretStr | None = None) -> str:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     evaluation_prompt_template = """
     You are a professional researcher. Your goal is to provide a relevant information report

prediction_prophet/functions/rerank_subqueries.py

@@ -3,7 +3,7 @@
 from langchain.prompts import ChatPromptTemplate
 from langchain.schema.output_parser import StrOutputParser
 from pydantic.types import SecretStr
-from prediction_market_agent_tooling.tools.utils import secret_str_from_env
+from prediction_market_agent_tooling.config import APIKeys
 from prediction_market_agent_tooling.gtypes import secretstr_to_v1_secretstr
 from prediction_market_agent_tooling.tools.langfuse_ import get_langfuse_langchain_config, observe
 
@@ -20,7 +20,7 @@
 @observe()
 def rerank_subqueries(queries: list[str], goal: str, model: str, temperature: float, api_key: SecretStr | None = None) -> list[str]:
     if api_key == None:
-        api_key = secret_str_from_env("OPENAI_API_KEY")
+        api_key = APIKeys().openai_api_key
 
     rerank_results_prompt = ChatPromptTemplate.from_template(template=rerank_queries_template)
 

prediction_prophet/functions/web_scrape.py

@@ -4,18 +4,19 @@
 from bs4 import BeautifulSoup
 from requests import Response
 import tenacity
+from datetime import timedelta
 from prediction_market_agent_tooling.tools.caches.db_cache import db_cache
 
 
 @tenacity.retry(stop=tenacity.stop_after_attempt(3), wait=tenacity.wait_fixed(1), reraise=True)
-@db_cache
 def fetch_html(url: str, timeout: int) -> Response:
     headers = {
         "User-Agent": "Mozilla/5.0 (X11; Linux x86_64; rv:107.0) Gecko/20100101 Firefox/107.0"
     }
     response = requests.get(url, headers=headers, timeout=timeout)
     return response
 
+@db_cache(max_age=timedelta(days=1))
 def web_scrape_strict(url: str, timeout: int = 10) -> str:
     response = fetch_html(url=url, timeout=timeout)
 

pyproject.toml

@@ -30,7 +30,7 @@ scikit-learn = "^1.4.0"
 typer = ">=0.9.0,<1.0.0"
 types-requests = "^2.31.0.20240125"
 types-python-dateutil = "^2.9.0"
-prediction-market-agent-tooling = { version = "^0.56.0.dev112", extras = ["langchain", "google"] }
+prediction-market-agent-tooling = { version = "^0.56.0.dev113", extras = ["langchain", "google"] }
 langchain-community = "^0.2.6"
 memory-profiler = "^0.61.0"
 matplotlib = "^3.8.3"