Add OSM area preflight

scripts/lib/area-diagnostics.js

@@ -67,6 +67,7 @@ function parseOsm(xml) {
   const nodePattern = /<node\b([^>]*?)\/>|<node\b([^>]*?)>([\s\S]*?)<\/node>/g;
   for (const match of xml.matchAll(nodePattern)) {
     const attrs = xmlAttrs(match[1] || match[2] || "");
+    if (attrs.action === "delete") continue;
     if (attrs.id) nodeIds.add(attrs.id);
     nodeStats.total += 1;
     const tags = parseTags(match[3] || "");
@@ -80,6 +81,8 @@ function parseOsm(xml) {
     buildingsWithHeight: 0,
     buildingsWithLevels: 0,
     buildingsWithBadHeight: 0,
+    buildingsWithBadLevels: 0,
+    buildingGeometryIssues: [],
     missingNodeRefs: 0,
     grass: 0,
     scrub: 0,
@@ -87,6 +90,7 @@ function parseOsm(xml) {
   };
   for (const match of xml.matchAll(/<way\b([^>]*)>([\s\S]*?)<\/way>/g)) {
     const attrs = xmlAttrs(match[1]);
+    if (attrs.action === "delete") continue;
     const body = match[2];
     const tags = parseTags(body);
     const refs = [];
@@ -110,6 +114,14 @@ function parseOsm(xml) {
       if (isExplicitHeight(tags)) wayStats.buildingsWithHeight += 1;
       if (tags["building:levels"]) wayStats.buildingsWithLevels += 1;
       if (tags.height && !parseHeightMeters(tags.height)) wayStats.buildingsWithBadHeight += 1;
+      if (tags["building:levels"] && !parseBuildingLevels(tags["building:levels"])) {
+        wayStats.buildingsWithBadLevels += 1;
+      }
+      if (refs.length < 4) {
+        wayStats.buildingGeometryIssues.push({ id: way.id, issue: "has fewer than 4 node refs" });
+      } else if (!way.closed) {
+        wayStats.buildingGeometryIssues.push({ id: way.id, issue: "is not closed" });
+      }
     }
     if (tags.landuse === "grass") wayStats.grass += 1;
     if (tags.natural === "scrub") wayStats.scrub += 1;
@@ -122,11 +134,13 @@ function parseOsm(xml) {
     buildingsWithHeight: 0,
     buildingsWithLevels: 0,
     buildingsWithBadHeight: 0,
+    buildingsWithBadLevels: 0,
     healthyBuildingMultipolygons: 0,
     issues: [],
   };
   for (const match of xml.matchAll(/<relation\b([^>]*)>([\s\S]*?)<\/relation>/g)) {
     const attrs = xmlAttrs(match[1]);
+    if (attrs.action === "delete") continue;
     const body = match[2];
     const tags = parseTags(body);
     const members = [];
@@ -139,6 +153,9 @@ function parseOsm(xml) {
     if (isExplicitHeight(tags)) relationStats.buildingsWithHeight += 1;
     if (tags["building:levels"]) relationStats.buildingsWithLevels += 1;
     if (tags.height && !parseHeightMeters(tags.height)) relationStats.buildingsWithBadHeight += 1;
+    if (tags["building:levels"] && !parseBuildingLevels(tags["building:levels"])) {
+      relationStats.buildingsWithBadLevels += 1;
+    }
 
     const health = buildingRelationHealth(attrs.id || "", members, ways);
     if (health.ok) relationStats.healthyBuildingMultipolygons += 1;
@@ -163,7 +180,10 @@ function parseBounds(xml) {
     maxLon: Number(attrs.maxlon),
     maxLat: Number(attrs.maxlat),
   };
-  return Object.values(bounds).every(Number.isFinite) ? bounds : null;
+  return Object.values(bounds).every(Number.isFinite) &&
+    bounds.minLon < bounds.maxLon && bounds.minLat < bounds.maxLat
+    ? bounds
+    : null;
 }
 
 function isExplicitHeight(tags) {
@@ -177,6 +197,48 @@ function parseHeightMeters(value) {
   return Number.isFinite(height) && height > 0 ? height : null;
 }
 
+function parseBuildingLevels(value) {
+  const levels = Number(String(value).trim());
+  return Number.isFinite(levels) && levels > 0;
+}
+
+function analyzeOsmPreflight(osm) {
+  const errors = [];
+  const warnings = [];
+  if (!osm.bounds) errors.push("OSM has no valid <bounds>.");
+  if (osm.ways.missingNodeRefs) {
+    errors.push(`OSM ways reference ${osm.ways.missingNodeRefs} missing node(s).`);
+  }
+  for (const issue of osm.ways.buildingGeometryIssues) {
+    errors.push(`Building way ${issue.id}: ${issue.issue}.`);
+  }
+  if (osm.ways.buildingsWithBadHeight || osm.relations.buildingsWithBadHeight) {
+    errors.push("Some building height tags could not be parsed as positive meters.");
+  }
+  for (const issue of osm.relations.issues) {
+    errors.push(`Building relation ${issue.id}: ${issue.issues.join("; ")}.`);
+  }
+  const badLevels = osm.ways.buildingsWithBadLevels + osm.relations.buildingsWithBadLevels;
+  if (badLevels) {
+    warnings.push(`${badLevels} building:levels tag(s) could not be parsed as positive numbers.`);
+  }
+  return {
+    errors: uniqueLines(errors),
+    warnings: uniqueLines(warnings),
+    summary: {
+      bounds: osm.bounds,
+      nodes: osm.nodes.total,
+      ways: osm.ways.total,
+      relations: osm.relations.total,
+      buildingWays: osm.ways.buildings,
+      buildingMultipolygons: osm.relations.buildingMultipolygons,
+      missingNodeRefs: osm.ways.missingNodeRefs,
+      buildingWayIssues: osm.ways.buildingGeometryIssues.length,
+      buildingRelationIssues: osm.relations.issues.length,
+    },
+  };
+}
+
 function buildingRelationHealth(id, members, ways) {
   const issues = [];
   const outerMembers = members.filter((member) => member.type === "way" && member.role === "outer");
@@ -307,6 +369,15 @@ function stageManifestStatus(area, configPath = null) {
   const hasReimportManifest = fs.existsSync(reimportManifest);
   const derivedConfig = path.join(area.outputs.pipelineDir, "osm2streets-qgis.config.json");
   const stages = [
+    {
+      stage: "preflight",
+      expected: fs.existsSync(stageManifestPath(area, "preflight")),
+      inputs: {
+        ...(configPath ? { config: configPath } : {}),
+        osm: area.input,
+      },
+      outputs: {},
+    },
     {
       stage: "intermediates",
       expected: (
@@ -763,6 +834,7 @@ function formatBytes(bytes) {
 
 module.exports = {
   BUDGETS,
+  analyzeOsmPreflight,
   analyzeArea,
   artifactStatus,
   classifyAreaQuality,