Add infrastructure for doing relation type stats

5 files changed, 50 insertions, 23 deletions

diff --git a/tagstats/tagstats.cpp b/tagstats/tagstats.cpp
index 9c026fa..aeca840 100644
--- a/tagstats/tagstats.cpp
+++ b/tagstats/tagstats.cpp
@@ -57,39 +57,42 @@ void print_help() {
               << "This program is part of Taginfo. It calculates statistics\n" \
               << "on OSM tags from OSMFILE and puts them into DATABASE (an SQLite database).\n" \
               << "\nOptions:\n" \
-              << "  -H, --help           This help message\n" \
-              << "  -d, --debug          Enable debugging output\n";
+              << "  -H, --help                    This help message\n" \
+              << "  -d, --debug                   Enable debugging output\n";
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-    std::cout << "  -T, --tags-list      File with tags we are especially interested in\n";
+    std::cout << "  -T, --tags=FILENAME           File with tags we are interested in\n";
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
-    std::cout << "  -t, --top=NUMBER     Top of bounding box for distribution images\n" \
-              << "  -r, --right=NUMBER   Right of bounding box for distribution images\n" \
-              << "  -b, --bottom=NUMBER  Bottom of bounding box for distribution images\n" \
-              << "  -l, --left=NUMBER    Left of bounding box for distribution images\n" \
-              << "  -w, --width=NUMBER   Width of distribution images (default: 360)\n" \
-              << "  -h, --height=NUMBER  Height of distribution images (default: 180)\n" \
+    std::cout << "  -R, --relation-types=FILENAME File with relation types we are interested in\n" \
+              << "  -t, --top=NUMBER              Top of bounding box for distribution images\n" \
+              << "  -r, --right=NUMBER            Right of bounding box for distribution images\n" \
+              << "  -b, --bottom=NUMBER           Bottom of bounding box for distribution images\n" \
+              << "  -l, --left=NUMBER             Left of bounding box for distribution images\n" \
+              << "  -w, --width=NUMBER            Width of distribution images (default: 360)\n" \
+              << "  -h, --height=NUMBER           Height of distribution images (default: 180)\n" \
               << "\nDefault for bounding box is: (-180, -90, 180, 90).\n";
 }
 
 int main(int argc, char *argv[]) {
     static struct option long_options[] = {
-        {"debug",     no_argument, 0, 'd'},
-        {"help",      no_argument, 0, 'H'},
+        {"debug",          no_argument, 0, 'd'},
+        {"help",           no_argument, 0, 'H'},
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-        {"tags-list", required_argument, 0, 'T'},
+        {"tags",           required_argument, 0, 'T'},
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
-        {"top",       required_argument, 0, 't'},
-        {"right",     required_argument, 0, 'r'},
-        {"bottom",    required_argument, 0, 'b'},
-        {"left",      required_argument, 0, 'l'},
-        {"width",     required_argument, 0, 'w'},
-        {"height",    required_argument, 0, 'h'},
+        {"relation-types", required_argument, 0, 'R'},
+        {"top",            required_argument, 0, 't'},
+        {"right",          required_argument, 0, 'r'},
+        {"bottom",         required_argument, 0, 'b'},
+        {"left",           required_argument, 0, 'l'},
+        {"width",          required_argument, 0, 'w'},
+        {"height",         required_argument, 0, 'h'},
         {0, 0, 0, 0}
     };
 
     bool debug = false;
 
     std::string tags_list;
+    std::string relation_type_list;
 
     double top    =   90;
     double right  =  180;
@@ -102,9 +105,9 @@ int main(int argc, char *argv[]) {
     while (true) {
         int c = getopt_long(argc, argv,
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-        "dHT:t:r:b:l:w:h:",
+        "dHT:R:t:r:b:l:w:h:",
 #else
-        "dHt:r:b:l:w:h:",
+        "dHR:t:r:b:l:w:h:",
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
         long_options, 0);
         if (c == -1) {
@@ -123,6 +126,9 @@ int main(int argc, char *argv[]) {
                 tags_list = optarg;
                 break;
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
+            case 'R':
+                relation_type_list = optarg;
+                break;
             case 't':
                 top = atof(optarg);
                 break;
@@ -155,7 +161,7 @@ int main(int argc, char *argv[]) {
     Osmium::OSMFile infile(argv[optind]);
     Sqlite::Database db(argv[optind+1]);
     MapToInt<rough_position_t> map_to_int(left, bottom, right, top, width, height);
-    TagStatsHandler handler(db, tags_list, map_to_int);
+    TagStatsHandler handler(db, tags_list, relation_type_list, map_to_int);
     Osmium::Input::read(infile, handler);
 }
 
diff --git a/tagstats/tagstats_handler.hpp b/tagstats/tagstats_handler.hpp
index b2f4416..2955d53 100644
--- a/tagstats/tagstats_handler.hpp
+++ b/tagstats/tagstats_handler.hpp
@@ -25,6 +25,7 @@
 #include <string>
 #include <fstream>
 #include <iostream>
+#include <map>
 
 #include <google/sparse_hash_map>
 #include <boost/algorithm/string/split.hpp>
@@ -181,6 +182,15 @@ public:
 typedef google::sparse_hash_map<const char *, KeyValueStats *, djb2_hash, eqstr> key_value_hash_map_t;
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
 
+class RelationTypeStats {
+
+public:
+
+    RelationTypeStats() {
+    }
+
+}; // class RelationTypeStats
+
 /**
  * Osmium handler that creates statistics for Taginfo.
  */
@@ -200,6 +210,8 @@ class TagStatsHandler : public Osmium::Handler::Base {
     key_value_hash_map_t m_key_value_stats;
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
 
+    std::map<std::string, RelationTypeStats> m_relation_type_stats;
+
     time_t m_max_timestamp;
 
     // this must be much bigger than the largest string we want to store
@@ -392,7 +404,7 @@ class TagStatsHandler : public Osmium::Handler::Base {
 
 public:
 
-    TagStatsHandler(Sqlite::Database& database, const std::string& tags_list, MapToInt<rough_position_t>& map_to_int) :
+    TagStatsHandler(Sqlite::Database& database, const std::string& tags_list, const std::string& relation_type_list, MapToInt<rough_position_t>& map_to_int) :
         Base(),
         m_max_timestamp(0),
         m_string_store(string_store_size),
@@ -410,6 +422,11 @@ public:
             m_key_value_stats[m_string_store.add(key_value.c_str())] = new KeyValueStats();
         }
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
+        std::ifstream relation_type_list_file(relation_type_list.c_str(), std::ifstream::in);
+        std::string type;
+        while (relation_type_list_file >> type) {
+            m_relation_type_stats[type] = RelationTypeStats();
+        }
     }
 
     void node(const shared_ptr<Osmium::OSM::Node const>& node) {
diff --git a/tagstats/test_relation_types.txt b/tagstats/test_relation_types.txt
new file mode 100644
index 0000000..ffd5691
--- /dev/null
+++ b/tagstats/test_relation_types.txt
@@ -0,0 +1,4 @@
+multipolygon
+site
+route
+associatedStreet
diff --git a/tagstats/test_tagslist.txt b/tagstats/test_tags.txt
index e53cd59..e53cd59 100644
--- a/tagstats/test_tagslist.txt
+++ b/tagstats/test_tags.txt
diff --git a/tagstats/test_tagstats.sh b/tagstats/test_tagstats.sh
index ec08b91..d07b5b7 100755
--- a/tagstats/test_tagstats.sh
+++ b/tagstats/test_tagstats.sh
@@ -21,5 +21,5 @@ ulimit -c 1000000000
 rm -f core
 
 #./tagstats --left=5.5 --bottom=47 --right=15 --top=55 --width=200 --height=320 $OSMFILE $DATABASE
-./tagstats -T test_tagslist.txt $OSMFILE $DATABASE
+./tagstats --tags test_tags.txt --relation-types test_relation_types.txt $OSMFILE $DATABASE