Check order of input files and warn if they are out of order

Osm2pgsql relies on the order nodes, then ways, then relations for
correct functioning. Future versions will also need ids to be in order
for efficiency and to allow merging and simplification of input files.

This commits adds a warning output when the order is not correct.

See #1167.

Author: joto

src/CMakeLists.txt

@@ -1,5 +1,6 @@
 
 set(osm2pgsql_lib_SOURCES
+  check-order.cpp
   db-copy.cpp
   dependency-manager.cpp
   expire-tiles.cpp

src/check-order.cpp

@@ -0,0 +1,80 @@
+#include "check-order.hpp"
+#include "format.hpp"
+
+#include <osmium/osm/relation.hpp>
+#include <osmium/osm/types.hpp>
+#include <osmium/osm/way.hpp>
+
+void check_order_t::warning(char const *msg, osmid_t id)
+{
+    fmt::print(stderr,
+               "\nWARNING: {}: {}\n"
+               "         Unordered input files do not work correctly in all\n"
+               "         cases. Future versions of osm2pgsql will require\n"
+               "         ordered files. Use the 'sort' command of osmium tool\n"
+               "         to sort them first.\n\n",
+               msg, id);
+    m_issued_warning = true;
+}
+
+void check_order_t::node(const osmium::Node &node)
+{
+    if (m_issued_warning) {
+        return;
+    }
+
+    if (m_max_way_id > std::numeric_limits<osmid_t>::min()) {
+        warning("Found a node after a way", node.id());
+    }
+    if (m_max_relation_id > std::numeric_limits<osmid_t>::min()) {
+        warning("Found a node after a relation", node.id());
+    }
+
+    if (m_max_node_id == node.id()) {
+        warning("Node ID twice in input. Maybe you are using a history or "
+                "non-simplified change file?",
+                node.id());
+    }
+    if (node.id() < m_max_node_id) {
+        warning("Node IDs out of order", node.id());
+    }
+    m_max_node_id = node.id();
+}
+
+void check_order_t::way(const osmium::Way &way)
+{
+    if (m_issued_warning) {
+        return;
+    }
+
+    if (m_max_relation_id > std::numeric_limits<osmid_t>::min()) {
+        warning("Found a way after a relation", way.id());
+    }
+
+    if (m_max_way_id == way.id()) {
+        warning("Way ID twice in input. Maybe you are using a history or "
+                "non-simplified change file?",
+                way.id());
+    }
+    if (way.id() < m_max_way_id) {
+        warning("Way IDs out of order", way.id());
+    }
+    m_max_way_id = way.id();
+}
+
+void check_order_t::relation(const osmium::Relation &relation)
+{
+    if (m_issued_warning) {
+        return;
+    }
+
+    if (m_max_relation_id == relation.id()) {
+        warning("Relation ID twice in input. Maybe you are using a history or "
+                "non-simplified change file?",
+                relation.id());
+    }
+    if (relation.id() < m_max_relation_id) {
+        warning("Relation IDs out of order", relation.id());
+    }
+    m_max_relation_id = relation.id();
+}

src/check-order.hpp

@@ -0,0 +1,40 @@
+#ifndef OSM2PGSQL_CHECK_ORDER_HPP
+#define OSM2PGSQL_CHECK_ORDER_HPP
+
+#include "osmtypes.hpp"
+
+#include <osmium/fwd.hpp>
+#include <osmium/handler.hpp>
+
+#include <limits>
+
+/**
+ * Handler that can be used to check that an OSM file is ordered
+ * correctly. Ordered in this case refers to the usual order in OSM
+ * files: First nodes in the order of their IDs, then ways in the order
+ * of their IDs, then relations in the order or their IDs.
+ *
+ * IDs have to be unique for each type. This check will fail for
+ * history files.
+ */
+class check_order_t : public osmium::handler::Handler
+{
+
+public:
+    void node(const osmium::Node &node);
+
+    void way(const osmium::Way &way);
+
+    void relation(const osmium::Relation &relation);
+
+private:
+    void warning(char const *msg, osmid_t id);
+
+    osmid_t m_max_node_id = std::numeric_limits<osmid_t>::min();
+    osmid_t m_max_way_id = std::numeric_limits<osmid_t>::min();
+    osmid_t m_max_relation_id = std::numeric_limits<osmid_t>::min();
+
+    bool m_issued_warning = false;
+}; // class check_order_t
+
+#endif // OSM2PGSQL_CHECK_ORDER_HPP

src/osmdata.cpp

@@ -11,6 +11,7 @@
 #include <osmium/io/any_input.hpp>
 #include <osmium/visitor.hpp>
 
+#include "check-order.hpp"
 #include "db-copy.hpp"
 #include "format.hpp"
 #include "input-handler.hpp"
@@ -375,9 +376,10 @@ progress_display_t osmdata_t::process_file(osmium::io::File const &file,
     fmt::print(stderr, "Using {} parser.\n",
                osmium::io::as_string(file.format()));
 
+    check_order_t check_order{};
     input_handler_t handler{bbox, m_append, this};
     osmium::io::Reader reader{file};
-    osmium::apply(reader, handler);
+    osmium::apply(reader, check_order, handler);
     reader.close();
 
     return handler.progress();