Make min_tag_combination_count configurable at runtime

3 files changed, 17 insertions, 8 deletions

diff --git a/tagstats/tagstats.cpp b/tagstats/tagstats.cpp
index 2dcbfb9..91b63ed 100644
--- a/tagstats/tagstats.cpp
+++ b/tagstats/tagstats.cpp
@@ -59,7 +59,9 @@ void print_help() {
               << "\nOptions:\n" \
               << "  -H, --help                    This help message\n";
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-    std::cout << "  -T, --tags=FILENAME           File with tags we are interested in\n";
+    std::cout << "  -T, --tags=FILENAME           File with tags we are interested in\n" \
+              << "  -m, --min-tag-combination-count=N  Tag combinations not appearing this often\n" \
+              << "                                     are not written to database\n";
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
     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" \
@@ -75,7 +77,8 @@ int main(int argc, char *argv[]) {
     static struct option long_options[] = {
         {"help",           no_argument, 0, 'H'},
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-        {"tags",           required_argument, 0, 'T'},
+        {"tags",                      required_argument, 0, 'T'},
+        {"min-tag-combination-count", required_argument, 0, 'm'},
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
         {"relation-types", required_argument, 0, 'R'},
         {"top",            required_argument, 0, 't'},
@@ -98,10 +101,12 @@ int main(int argc, char *argv[]) {
     unsigned int width  = 360;
     unsigned int height = 180;
 
+    unsigned int min_tag_combination_count = 1000;
+
     while (true) {
         int c = getopt_long(argc, argv,
 #ifdef TAGSTATS_COUNT_TAG_COMBINATIONS
-                            "dHT:R:t:r:b:l:w:h:",
+                            "dHR:t:r:b:l:w:h:T:m:",
 #else
                             "dHR:t:r:b:l:w:h:",
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
@@ -118,6 +123,9 @@ int main(int argc, char *argv[]) {
             case 'T':
                 tags_list = optarg;
                 break;
+            case 'm':
+                min_tag_combination_count = atoi(optarg);
+                break;
 #endif // TAGSTATS_COUNT_TAG_COMBINATIONS
             case 'R':
                 relation_type_list = optarg;
@@ -154,7 +162,7 @@ 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, relation_type_list, map_to_int);
+    TagStatsHandler handler(db, tags_list, relation_type_list, map_to_int, min_tag_combination_count);
     Osmium::Input::read(infile, handler);
 }
 
diff --git a/tagstats/tagstats_handler.hpp b/tagstats/tagstats_handler.hpp
index d3ea136..4069dd1 100644
--- a/tagstats/tagstats_handler.hpp
+++ b/tagstats/tagstats_handler.hpp
@@ -243,7 +243,7 @@ class TagStatsHandler : public Osmium::Handler::Base {
      * Tag combination not appearing at least this often are not written
      * to database.
      */
-    static const unsigned int min_tag_combination_count = 1000;
+    unsigned int m_min_tag_combination_count;
 
     time_t timer;
 
@@ -446,8 +446,9 @@ class TagStatsHandler : public Osmium::Handler::Base {
 
 public:
 
-    TagStatsHandler(Sqlite::Database& database, const std::string& tags_list, const std::string& relation_type_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, unsigned int min_tag_combination_count) :
         Base(),
+        m_min_tag_combination_count(min_tag_combination_count),
         m_max_timestamp(0),
         m_string_store(string_store_size),
         m_database(database),
@@ -691,7 +692,7 @@ public:
             for (combination_hash_map_t::const_iterator it = stat->m_key_value_combination_hash.begin(); it != stat->m_key_value_combination_hash.end(); ++it) {
                 const Counter* s = &(it->second);
 
-                if (s->all() >= min_tag_combination_count) {
+                if (s->all() >= m_min_tag_combination_count) {
                     std::vector<std::string> kv2;
                     boost::split(kv2, it->first, boost::is_any_of("="));
                     kv2.push_back(""); // if there is no = in key, make sure there is an empty value
diff --git a/tagstats/test_tagstats.sh b/tagstats/test_tagstats.sh
index d07b5b7..c699d55 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 --tags test_tags.txt --relation-types test_relation_types.txt $OSMFILE $DATABASE
+./tagstats --tags test_tags.txt --min-tag-combination-count=100 --relation-types test_relation_types.txt $OSMFILE $DATABASE