Add docs to download overture places data & ollama model (#13)

* Add docs to download overture places data & ollama model
* Hit local overture parquet files
* Add osx gitignore
* Add .env.example
* Make overture data source selectable using .env
* Add pytest marker to set right ENV vars during CI

---------

Co-authored-by: Daniel Wiesmann <yellowcap@users.noreply.github.com>

src/geo_assistant/agent/llms.py

@@ -3,11 +3,11 @@ from dotenv import load_dotenv
 
 from langchain_ollama import ChatOllama
 
-# Load environment variables from .env file
+# Load environment variables from env file
 load_dotenv()
 
 # Get model name from environment variable, default to llama3.2
-MODEL_NAME = os.environ.get("OLLAMA_MODEL", "llama3.2")
+MODEL_NAME = os.environ.get("OLLAMA_AGENT_MODEL", "llama3.2")
 OLLAMA_BASE_URL = os.environ.get("OLLAMA_BASE_URL", "http://localhost:11434")
 
 llm = ChatOllama(

src/geo_assistant/tools/overture.py

@@ -1,12 +1,17 @@
 import json
+import os
 from typing import Annotated
 
 import duckdb
+from dotenv import load_dotenv
 from langchain_core.messages import ToolMessage
 from langchain_core.tools import tool
 from langchain_core.tools.base import InjectedToolCallId
 from langgraph.types import Command
 
+# Load environment variables
+load_dotenv()
+
 
 def create_database_connection():
     """Create and configure a DuckDB connection with necessary extensions.
@@ -33,13 +38,17 @@ async def get_place(
     """Get place location from Overture Maps based on user input place name."""
 
     db_connection = create_database_connection()
+    source = os.getenv("OVERTURE_SOURCE", "local")
+    if source == "s3":
+        data_path = os.getenv("OVERTURE_S3_PATH")
+        db_connection.execute("SET s3_region='us-west-2';")
+    else:
+        data_path = os.getenv("OVERTURE_LOCAL_PATH")
 
     location_results = db_connection.execute(
         f"""
       LOAD spatial;
 
-      SET s3_region='us-west-2';
-
       SELECT
           id,
           jaro_winkler_similarity(LOWER(names.primary), LOWER('{place_name}')) AS similarity_score,
@@ -48,7 +57,7 @@ async def get_place(
           CAST(socials AS JSON) AS socials,
           ST_AsGeoJSON(geometry) AS geometry,
       FROM read_parquet(
-          's3://overturemaps-us-west-2/release/2025-11-19.0/theme=places/type=place/*',
+          '{data_path}',
           filename=true,
           hive_partitioning=1
       )

src/geo_assistant/tools/summarize.py

@@ -1,5 +1,6 @@
 """Tools for summarizing satellite images using LLM-based analysis."""
 
+import os
 from typing import Annotated, Optional
 import dspy
 from langchain_core.tools import tool
@@ -7,6 +8,10 @@ from langgraph.types import Command
 from langchain_core.messages import ToolMessage
 from langchain_core.tools.base import InjectedToolCallId
 
+import dotenv
+
+dotenv.load_dotenv()
+
 
 class SatImgSummary(dspy.Signature):
     "Describe things you see in the satellite image."
@@ -20,8 +25,8 @@ class SatImgSummaryAgent(dspy.Module):
 
     def __init__(
         self,
-        model: str = "ministral-3:14b-cloud",
-        api_base: str = "http://localhost:11434",
+        model: str = os.environ.get("OLLAMA_IMAGE_MODEL", "ministral-3:14b-cloud"),
+        api_base: str = os.environ.get("OLLAMA_BASE_URL", "http://localhost:11434"),
         temperature: float = 0.5,
         max_tokens: int = 4_096,
     ) -> None: