From 7b7727f234a751692dc798ac58c5c606a39cbcdb Mon Sep 17 00:00:00 2001
From: Emi Griffith <me@emigriffith.dev>
Date: Wed, 22 Jul 2026 14:45:26 -0700
Subject: [PATCH] paths.py: make DATA_DIR/LOGS_DIR env-overridable to avoid
 /app/data volume collision

After the repo split the code lives at the image root, so the Python package
data/ and the runtime data dir <root>/data are the same path (/app/data). The
deploy compose mounts the appdata volume there, shadowing data/*.py so
'import data.climate' fails at container boot. THERMOGRAPH_DATA_DIR (and
_LOGS_DIR) now let the container point runtime state at a path outside the code
tree; local dev is unchanged (defaults to <root>/data).
---
 paths.py | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/paths.py b/paths.py
index af629a5..aeb195d 100644
--- a/paths.py
+++ b/paths.py
@@ -12,8 +12,16 @@ keep doing so; these are only the defaults.
 import os
 
 REPO_DIR = os.path.dirname(os.path.abspath(__file__))
-DATA_DIR = os.path.join(REPO_DIR, "data")
-LOGS_DIR = os.path.join(REPO_DIR, "logs")
+# Runtime state dirs default under the repo root, but MUST be overridable: after
+# the repo split the code lives at the image root (/app), so the Python package
+# `data/` and the runtime data dir `<root>/data` collide at the same path. When
+# a persistent volume is mounted at that path (as the deploy compose does for
+# the parquet cache / notifier.lock / geonames) it shadows and erases the
+# `data/*.py` modules, so `import data.climate` fails at boot. Point
+# THERMOGRAPH_DATA_DIR (and _LOGS_DIR) at a path OUTSIDE the code tree in the
+# container to keep the volume clear of the package.
+DATA_DIR = os.environ.get("THERMOGRAPH_DATA_DIR") or os.path.join(REPO_DIR, "data")
+LOGS_DIR = os.environ.get("THERMOGRAPH_LOGS_DIR") or os.path.join(REPO_DIR, "logs")
 
 # Bundled reference data shipped alongside the code (generated by gen_cities /
 # gen_flavor), not runtime state — hence at the repo root, not data/.