Refactor middle: Replace analyze and flush methods

Instead of the analyze() and flush() functions on the middle, the more
generic functions after_nodes(), after_ways(), and after_relations() are
introduced. Each middle can do in them whatever it needs at that place
of reading the inputs. In the case of the middle-pgsql we'll now do the
analyze of each of the nodes, ways, and relations table right after they
have been filled.

This also fixes a bug: The analyze() functions in the middle were never
called.

Author: joto

src/middle-pgsql.cpp

@@ -26,6 +26,7 @@
 #include "node-ram-cache.hpp"
 #include "options.hpp"
 #include "osmtypes.hpp"
+#include "pgsql-helper.hpp"
 #include "util.hpp"
 
 /**
@@ -607,13 +608,29 @@ void middle_pgsql_t::relation_delete(osmid_t osm_id)
     m_db_copy.delete_object(osm_id);
 }
 
-void middle_pgsql_t::analyze()
+void middle_pgsql_t::after_nodes()
 {
-    for (auto const &table : m_tables) {
-        m_db_connection.exec("ANALYZE {}"_format(table.name()));
+    m_db_copy.sync();
+    if (m_options->flat_node_file.empty()) {
+        auto const &table = m_tables[NODE_TABLE];
+        analyze_table(m_db_connection, table.schema(), table.name());
     }
 }
 
+void middle_pgsql_t::after_ways()
+{
+    m_db_copy.sync();
+    auto const &table = m_tables[WAY_TABLE];
+    analyze_table(m_db_connection, table.schema(), table.name());
+}
+
+void middle_pgsql_t::after_relations()
+{
+    m_db_copy.sync();
+    auto const &table = m_tables[REL_TABLE];
+    analyze_table(m_db_connection, table.schema(), table.name());
+}
+
 middle_query_pgsql_t::middle_query_pgsql_t(
     std::string const &conninfo, std::shared_ptr<node_ram_cache> const &cache,
     std::shared_ptr<node_persistent_cache> const &persistent_cache)
@@ -658,8 +675,6 @@ void middle_pgsql_t::commit()
     m_copy_thread->finish();
 }
 
-void middle_pgsql_t::flush() { m_db_copy.sync(); }
-
 void middle_pgsql_t::stop(thread_pool_t &pool)
 {
     m_cache.reset();

src/middle-pgsql.hpp

@@ -60,14 +60,15 @@ struct middle_pgsql_t : public middle_t
 
     void start() override;
     void stop(thread_pool_t &pool) override;
-    void analyze() override;
     void commit() override;
 
     void node(osmium::Node const &node) override;
     void way(osmium::Way const &way) override;
     void relation(osmium::Relation const &rel) override;
 
-    void flush() override;
+    void after_nodes() override;
+    void after_ways() override;
+    void after_relations() override;
 
     idlist_t get_ways_by_node(osmid_t osm_id) override;
     idlist_t get_rels_by_node(osmid_t osm_id) override;

src/middle-ram.hpp

@@ -96,7 +96,6 @@ struct middle_ram_t : public middle_t, public middle_query_t
 
     void start() override {}
     void stop(thread_pool_t &pool) override;
-    void analyze() override {}
     void commit() override {}
 
     void node(osmium::Node const &node) override;
@@ -112,8 +111,6 @@ struct middle_ram_t : public middle_t, public middle_query_t
     bool relation_get(osmid_t id,
                       osmium::memory::Buffer &buffer) const override;
 
-    void flush() override {}
-
     std::shared_ptr<middle_query_t> get_query_instance() override;
 
 private:

src/middle.hpp

@@ -74,7 +74,6 @@ struct middle_t
 
     virtual void start() = 0;
     virtual void stop(thread_pool_t &pool) = 0;
-    virtual void analyze() = 0;
     virtual void commit() = 0;
 
     /// This is called for every added, changed or deleted node.
@@ -86,16 +85,14 @@ struct middle_t
     /// This is called for every added, changed or deleted relation.
     virtual void relation(osmium::Relation const &relation) = 0;
 
-    /**
-     * Ensure all pending data is written to the storage.
-     *
-     * You can only query objects from the storage after they have been
-     * flushed.
-     *
-     * The function is called after setting all the nodes, then after setting
-     * all the ways, and again after setting all the relations.
-     */
-    virtual void flush() = 0;
+    /// Called after all nodes from the input file(s) have been processed.
+    virtual void after_nodes() {}
+
+    /// Called after all ways from the input file(s) have been processed.
+    virtual void after_ways() {}
+
+    /// Called after all relations from the input file(s) have been processed.
+    virtual void after_relations() {}
 
     virtual idlist_t get_ways_by_node(osmid_t) { return {}; }
     virtual idlist_t get_rels_by_node(osmid_t) { return {}; }

src/osmdata.cpp

@@ -62,7 +62,7 @@ void osmdata_t::node(osmium::Node const &node)
     }
 }
 
-void osmdata_t::after_nodes() { flush(); }
+void osmdata_t::after_nodes() { m_mid->after_nodes(); }
 
 void osmdata_t::way(osmium::Way &way)
 {
@@ -79,7 +79,7 @@ void osmdata_t::way(osmium::Way &way)
     }
 }
 
-void osmdata_t::after_ways() { flush(); }
+void osmdata_t::after_ways() { m_mid->after_ways(); }
 
 void osmdata_t::relation(osmium::Relation const &rel)
 {
@@ -105,7 +105,7 @@ void osmdata_t::relation(osmium::Relation const &rel)
     }
 }
 
-void osmdata_t::after_relations() { flush(); }
+void osmdata_t::after_relations() { m_mid->after_relations(); }
 
 void osmdata_t::node_add(osmium::Node const &node) const
 {
@@ -187,11 +187,6 @@ void osmdata_t::start() const
     }
 }
 
-void osmdata_t::flush() const
-{
-    m_mid->flush();
-}
-
 namespace {
 
 /**

src/osmdata.hpp

@@ -65,8 +65,6 @@ class osmdata_t : public osmium::handler::Handler
     void way_delete(osmid_t id) const;
     void relation_delete(osmid_t id) const;
 
-    void flush() const;
-
     /**
      * Run stage 1b and stage 1c processing: Process dependent objects in
      * append mode.

src/pgsql-helper.cpp

@@ -28,3 +28,10 @@ void create_geom_check_trigger(pg_conn_t *db_connection,
         " {}();"_format(table + "_osm2pgsql_valid",
                         qualified_name(schema, table), func_name));
 }
+
+void analyze_table(pg_conn_t const &db_connection, std::string const &schema,
+                   std::string const &name)
+{
+    auto const qual_name = qualified_name(schema, name);
+    db_connection.exec("ANALYZE {}"_format(qual_name));
+}

src/pgsql-helper.hpp

@@ -1,11 +1,16 @@
 #ifndef OSM2PGSQL_PGSQL_HELPER_HPP
 #define OSM2PGSQL_PGSQL_HELPER_HPP
 
+#include <string>
+
 #include "pgsql.hpp"
 
 void create_geom_check_trigger(pg_conn_t *db_connection,
                                std::string const &schema,
                                std::string const &table,
                                std::string const &geom_column);
 
+void analyze_table(pg_conn_t const &db_connection, std::string const &schema,
+                   std::string const &name);
+
 #endif // OSM2PGSQL_PGSQL_HELPER_HPP