Alpha stage commit

app/osm_classification.py

@@ -0,0 +1,111 @@
+from __future__ import annotations
+
+import json
+import re
+from typing import Mapping
+
+
+LOCAL_SCOPE = "local"
+REGIONAL_SCOPE = "regional"
+LONG_DISTANCE_SCOPE = "long_distance"
+UNKNOWN_SCOPE = "unknown"
+OSM_ROUTE_SCOPE_CLASSIFIER_VERSION = "route_scope_v2"
+
+BUS_MODES = {"bus", "trolleybus"}
+LOCAL_MODES = {"tram", "light_rail", "subway", "ferry", "funicular", "aerialway", "monorail"}
+LONG_DISTANCE_MODES = {"coach"}
+
+LONG_DISTANCE_SERVICE_VALUES = {
+    "high_speed",
+    "long_distance",
+    "intercity",
+    "international",
+    "night",
+    "sleeper",
+}
+REGIONAL_SERVICE_VALUES = {"regional", "interurban", "commuter", "branch", "suburban"}
+LOCAL_SERVICE_VALUES = {"local", "urban", "city", "subway", "tram", "light_rail", "s-bahn", "sbahn"}
+
+LONG_DISTANCE_PREFIX_RE = re.compile(r"^(ICE|IC|EC|ECE|EN|NJ|RJ|RJX|TGV|THA|EST|FLX|WB)\b|^(ICE|IC|EC|ECE|EN|NJ|RJ|RJX|TGV|THA|EST|FLX|WB)\d")
+REGIONAL_PREFIX_RE = re.compile(r"^(IRE|RE|RB|RER|TER|REX|MEX|ALX|WFB|R)\b|^(IRE|RE|RB|RER|TER|REX|MEX|ALX|WFB|R)\d")
+LOCAL_TRAIN_PREFIX_RE = re.compile(r"^(S|S-BAHN)\b|^S\d")
+
+
+def infer_osm_route_scope(
+    *,
+    mode: str | None,
+    ref: str | None = None,
+    name: str | None = None,
+    network: str | None = None,
+    tags: Mapping[str, object] | str | None = None,
+) -> str | None:
+    """Classify a public-transport route into a display scope.
+
+    OSM tagging varies by country and operator, so this intentionally combines
+    explicit service tags with conservative reference-prefix heuristics.
+    """
+    normalized_mode = (mode or "").strip().lower()
+    tags_dict = _tags_dict(tags)
+    values = {
+        str(tags_dict.get(key) or "").strip().lower()
+        for key in ("service", "train", "bus", "passenger", "network:type", "route_scope")
+        if tags_dict.get(key)
+    }
+    if values & LONG_DISTANCE_SERVICE_VALUES:
+        return LONG_DISTANCE_SCOPE
+    if values & LOCAL_SERVICE_VALUES:
+        return LOCAL_SCOPE
+    if values & REGIONAL_SERVICE_VALUES:
+        return REGIONAL_SCOPE
+    if normalized_mode in LOCAL_MODES:
+        return LOCAL_SCOPE
+    if normalized_mode in LONG_DISTANCE_MODES:
+        return LONG_DISTANCE_SCOPE
+
+    text = _classification_text(ref, name, network, tags_dict)
+    if normalized_mode in BUS_MODES:
+        if any(marker in text for marker in ("FLIXBUS", "EUROLINES", "INTERCITYBUS", "IC BUS", "LONG DISTANCE", "FERNBUS")):
+            return LONG_DISTANCE_SCOPE
+        if any(marker in text for marker in ("REGIONALBUS", "REGIOBUS", "REGIONAL BUS", "REGIONALVERKEHR", "REGIONAL VERKEHR")):
+            return REGIONAL_SCOPE
+        return LOCAL_SCOPE
+
+    if normalized_mode == "train":
+        if LONG_DISTANCE_PREFIX_RE.search(text) or any(marker in text for marker in ("INTERCITY", "EUROCITY", "NIGHTJET", "FLIXTRAIN")):
+            return LONG_DISTANCE_SCOPE
+        if LOCAL_TRAIN_PREFIX_RE.search(text) or "S-BAHN" in text or "SBahn".upper() in text:
+            return LOCAL_SCOPE
+        if REGIONAL_PREFIX_RE.search(text) or any(marker in text for marker in ("REGIONAL", "REGIO", "REGIONALBAHN", "REGIONALEXPRESS")):
+            return REGIONAL_SCOPE
+        return UNKNOWN_SCOPE
+
+    return None
+
+
+def infer_osm_route_scope_from_tags(mode: str | None, ref: str | None, name: str | None, network: str | None, tags_json: str | None) -> str | None:
+    return infer_osm_route_scope(mode=mode, ref=ref, name=name, network=network, tags=tags_json)
+
+
+def _tags_dict(tags: Mapping[str, object] | str | None) -> dict[str, object]:
+    if isinstance(tags, str):
+        try:
+            data = json.loads(tags or "{}")
+        except json.JSONDecodeError:
+            return {}
+        return data if isinstance(data, dict) else {}
+    if isinstance(tags, Mapping):
+        return dict(tags)
+    return {}
+
+
+def _classification_text(ref: str | None, name: str | None, network: str | None, tags: Mapping[str, object]) -> str:
+    parts = [
+        ref or "",
+        name or "",
+        network or "",
+        str(tags.get("ref") or ""),
+        str(tags.get("name") or ""),
+        str(tags.get("network") or ""),
+        str(tags.get("network:short") or ""),
+    ]
+    return " ".join(parts).strip().upper().replace("_", " ")