Alpha stage commit

app/config.py

@@ -0,0 +1,74 @@
+from __future__ import annotations
+
+from pathlib import Path
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    """Runtime settings.
+
+    SQLite is the default because this prototype should run immediately.
+    The schema is deliberately plain enough to migrate to PostGIS later.
+    """
+
+    database_url: str = "sqlite:///./data/workbench.sqlite"
+    data_dir: Path = Path("./data")
+    # 0 means import all stop_times. Use a positive value only for constrained
+    # demos where full timetable routing is not needed.
+    gtfs_stop_times_import_limit: int = 0
+    # "sidecar_stop_times" keeps the large timetable call table in a per-dataset
+    # SQLite file and stores compact GTFS tables in the main app database.
+    # Set to "main" for the old all-in-one SQLite layout.
+    gtfs_timetable_storage: str = "sidecar_stop_times"
+    gtfs_keep_activation_stage: bool = False
+    # "sidecar_features" keeps extracted OSM transport features in a per-dataset
+    # SQLite file. The main DB materializes only OSM rows that need stable
+    # foreign keys for matches or route-layer output.
+    osm_feature_storage: str = "sidecar_features"
+    osm_sidecar_create_visual_only_stops: bool = False
+    # Large OSM PBF extracts should be reduced to transport objects before the
+    # Python extractor scans them. XML fixtures stay unfiltered by default.
+    osm_pbf_prefilter_enabled: bool = True
+    osm_pbf_prefilter_formats: str = "osm_pbf"
+    osm_pbf_prefilter_script: Path = Path("scripts/osmium_transport_filter.sh")
+    osm_diff_max_sequence_gap: int = 14
+    osm_diff_apply_batch_size: int = 7
+    osm_diff_state_timeout_seconds: float = 30.0
+    sqlite_timeout_seconds: float = 120.0
+    sqlite_busy_timeout_ms: int = 120000
+    database_write_lock_timeout_seconds: float = 1.0
+    database_write_lock_cli_timeout_seconds: float = 3600.0
+    queue_worker_autostart: bool = True
+    queue_worker_count: int = 1
+    queue_worker_poll_interval_seconds: float = 2.0
+    queue_job_lease_seconds: int = 7200
+    route_matching_batch_size: int = 100
+    route_layer_osm_route_batch_size: int = 1000
+    route_layer_osm_stop_batch_size: int = 5000
+    # SQLite defaults to sidecar storage. PostgreSQL/PostGIS defaults to main
+    # table storage so indexes, joins, and spatial operators can work over the
+    # full imported datasets.
+    postgres_use_sidecars: bool = False
+    # Keep supervised workers alive across API server restarts. Stale workers are
+    # detected by PID files at the next startup; stale job leases are requeued.
+    queue_worker_stop_on_shutdown: bool = False
+
+    model_config = SettingsConfigDict(env_file=".env", env_file_encoding="utf-8")
+
+    @property
+    def normalized_database_url(self) -> str:
+        if self.database_url.startswith("postgresql://"):
+            return "postgresql+psycopg://" + self.database_url.removeprefix("postgresql://")
+        return self.database_url
+
+    @property
+    def is_sqlite_database(self) -> bool:
+        return self.normalized_database_url.startswith("sqlite")
+
+    @property
+    def is_postgresql_database(self) -> bool:
+        return self.normalized_database_url.startswith("postgresql")
+
+
+settings = Settings()
+settings.data_dir.mkdir(parents=True, exist_ok=True)