diegoripley/tippecanoe
1
0
Fork 0
mirror of https://github.com/felt/tippecanoe.git synced 2026-10-02 16:35:40 +02:00
Code Issues Packages Projects Releases Wiki Activity

Choose the megatile features from those that will be in the next N zooms

Browse Source
This commit is contained in:
Erica Fischer
2024-10-07 12:42:24 -07:00
parent 78661f1cb2
commit 467a7c59e2
7 changed files with 11683 additions and 12272 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Makefile
+13 -13
View File
@@ -527,22 +527,22 @@ accumulate-test:
# and 99 without it
test `grep '"POP1950": null' tests/ne_110m_populated_places_nulls/in.json | wc -l` == 99
./tippecanoe -yNAME -yPOP1950 -yclustered:cluster_size -yclustered:unrelated -q -z3 -r1.75 -b0 -f -e tests/pbf/accum.dir --accumulate-numeric-attributes=clustered --set-attribute '{"clustered:cluster_size":1}' --accumulate-attribute '{"clustered:cluster_size":"sum"}' --retain-points-multiplier 3 tests/ne_110m_populated_places_nulls/in.json
# at this drop rate, there are 6 points at z0 that have no POP1950s clustered onto them....
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:count:POP1950' | wc -l` == 78
# at this drop rate, there are 78 points at z0 that have no POP1950s clustered onto them....
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:count:POP1950' | wc -l` == 63
# 35 of which have no POP1950 at all
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'POP1950' | wc -l` == 35
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'POP1950' | wc -l` == 27
# 43 of which do have POP1950
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:count:POP1950' | grep 'POP1950' | wc -l` == 43
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:count:POP1950' | grep 'POP1950' | wc -l` == 36
# plus 59 that are clustered
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:count:POP1950' | wc -l` == 59
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:count:POP1950' | wc -l` == 60
# the 59 clustered POP1950s have a total count of 101
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:count:POP1950' | sed 's/.*"clustered:count:POP1950": //' | awk '{sum += $$1} END {print sum}'` == 101
# we have already established that there are 43 bare POP1950s
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:count:POP1950' | sed 's/.*"clustered:count:POP1950": //' | awk '{sum += $$1} END {print sum}'` == 108
# we have already established that there are 36 bare POP1950s
# which makes a total of 144, which is the total count expected
#
# meanwhile, regular attribute accumulation.
# there are 137 features in the z0 tile, and they all have clustered:cluster_size
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:cluster_size' | wc -l` == 137
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:cluster_size' | wc -l` == 123
# there are no features that lack it.
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:cluster_size' | wc -l` == 0
# they add up to the 243 original features
@@ -550,15 +550,15 @@ accumulate-test:
# Make sure we do *not* accumulate a numeric attribute that already has the magic prefix:
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep sum:clustered:unrelated | wc -l` == 0
# But that we *do* preserve those attributes into the output features:
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep clustered:unrelated | wc -l` == 66
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep clustered:unrelated | wc -l` == 62
#
# on to the sums:
# in the original data set, the POP1950s that are present add up to 161590
test `grep '"POP1950": [0-9]' tests/ne_110m_populated_places_nulls/in.json | sed 's/.*"POP1950": //' | awk '{sum += $$1} END {print sum}' ` == 161590
# in the z0 tile, the clustered POP1950s add up to 113357
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:sum:POP1950' | sed 's/.*"clustered:sum:POP1950": //' | awk '{sum += $$1} END {print sum}'` == 113357
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep 'clustered:sum:POP1950' | sed 's/.*"clustered:sum:POP1950": //' | awk '{sum += $$1} END {print sum}'` == 115552
# and the non-clustered ones add up to 48233
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:sum:POP1950' | grep POP1950 | sed 's/.*"POP1950": //' | awk '{sum += $$1} END {print sum}'` == 48233
test `./tippecanoe-decode -c tests/pbf/accum.dir/0/0/0.pbf 0 0 0 | grep -v 'clustered:sum:POP1950' | grep POP1950 | sed 's/.*"POP1950": //' | awk '{sum += $$1} END {print sum}'` == 46038
# which is the correct 161590
#
# OK, so do these still hold after megatile filtering?
@@ -599,11 +599,11 @@ accumulate-test:
# Now on to binning!
./tippecanoe-overzoom --assign-to-bins tests/pbf/h3-0-0-0.geojson --accumulate-numeric-attributes=clustered --accumulate-attribute '{"clustered:cluster_size":"sum"}' -o tests/pbf/bins-0-0-0.pbf tests/pbf/accum.dir/0/0/0.pbf 0/0/0 0/0/0
# Now there are 30 bins with POP1950 clusters
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep 'clustered:count:POP1950' | wc -l` == 44
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep 'clustered:count:POP1950' | wc -l` == 41
# There are none with bare POP1950 (which is expected; we should only have summary statistics)
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep -v 'clustered:count:POP1950' | grep 'POP1950' | wc -l` == 0
# And 4 with no POP1950 at all
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep -v 'POP1950' | wc -l` == 4
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep -v 'POP1950' | wc -l` == 3
#
# the clustered and megatile-filtered and binned POP1950s add up to 161590
test `./tippecanoe-decode -c tests/pbf/bins-0-0-0.pbf 0 0 0 | grep 'clustered:sum:POP1950' | sed 's/.*"clustered:sum:POP1950": //' | awk '{sum += $$1} END {print sum}'` == 161590
tests/muni/out/-z10_--retain-points-multiplier_10_-M10000_--drop-smallest-as-needed.json
+1983 -2183
View File
File diff suppressed because it is too large Load Diff
tests/muni/out/-z11_--retain-points-multiplier_2_--extend-zooms-if-still-dropping.json
+2813 -2875
View File
File diff suppressed because it is too large Load Diff
tests/muni/out/-z11_-ycount_--set-attribute_count%3a1_--accumulate-attribute_count%3asum_--retain-points-multiplier_5.json
+3911 -4179
View File
File diff suppressed because it is too large Load Diff
tests/muni/out/-zg_--retain-points-multiplier_2.json
+2813 -2875
View File
File diff suppressed because it is too large Load Diff
tests/ne_110m_populated_places/out/-z3_-b0_--accumulate-numeric-attributes_numnum_--set-attribute_accum%3a1_--set-attribute_accum2%3a1_--accumulate-attribute_accum%3asum_--retain-points-multiplier_3.json
+142 -142
View File
File diff suppressed because one or more lines are too long
tile.cpp
+8 -5
View File
@@ -1056,7 +1056,8 @@ static bool skip_next_feature(decompressor *geoms, std::atomic<long long> *geomp
// This function is called repeatedly from write_tile() to retrieve the next feature
// from the input stream. If the stream is at an end, it returns a feature with the
// geometry type set to -2.
static serial_feature next_feature(decompressor *geoms, std::atomic<long long> *geompos_in, int z, unsigned tx, unsigned ty, unsigned *initial_x, unsigned *initial_y, long long *original_features, long long *unclipped_features, int nextzoom, int maxzoom, int minzoom, int max_zoom_increment, size_t pass, std::atomic<long long> *along, long long alongminus, int buffer, std::atomic<bool> *within, compressor **geomfile, std::atomic<long long> *geompos, long long start_geompos[], std::atomic<double> *oprogress, double todo, const char *fname, int child_shards, json_object *filter, const char *global_stringpool, long long *pool_off, std::vector<std::vector<std::string>> *layer_unmaps, bool first_time, bool compressed, multiplier_state *multiplier_state, std::shared_ptr<std::string> &tile_stringpool, std::vector<std::string> const &unidecode_data, unsigned long long &previndex) {
static serial_feature next_feature(decompressor *geoms, std::atomic<long long> *geompos_in, int z, unsigned tx, unsigned ty, unsigned *initial_x, unsigned *initial_y, long long *original_features, long long *unclipped_features, int nextzoom, int maxzoom, int minzoom, int max_zoom_increment, size_t pass, std::atomic<long long> *along, long long alongminus, int buffer, std::atomic<bool> *within, compressor **geomfile, std::atomic<long long> *geompos, long long start_geompos[], std::atomic<double> *oprogress, double todo, const char *fname, int child_shards, json_object *filter, const char *global_stringpool, long long *pool_off, std::vector<std::vector<std::string>> *layer_unmaps, bool first_time, bool compressed, multiplier_state *multiplier_state, std::shared_ptr<std::string> &tile_stringpool, std::vector<std::string> const &unidecode_data, unsigned long long &previndex, double droprate) {
double extra_multiplier_zooms = log(retain_points_multiplier) / log(droprate);
while (1) {
serial_feature sf;
long long len;
@@ -1221,7 +1222,7 @@ static serial_feature next_feature(decompressor *geoms, std::atomic<long long> *
if (z >= sf.feature_minzoom || sf.dropped == FEATURE_KEPT) {
count->second = 0;
sf.dropped = FEATURE_KEPT; // feature is kept
} else if (count->second + 1 < retain_points_multiplier) {
} else if (ceil(z + extra_multiplier_zooms) >= sf.feature_minzoom && count->second + 1 < retain_points_multiplier) {
count->second++;
sf.dropped = count->second;
} else {
@@ -1287,6 +1288,7 @@ struct run_prefilter_args {
std::vector<std::string> const *unidecode_data;
bool first_time = false;
bool compressed = false;
double droprate = 1;
};
void *run_prefilter(void *v) {
@@ -1297,7 +1299,7 @@ void *run_prefilter(void *v) {
unsigned long long previndex = 0;
while (1) {
serial_feature sf = next_feature(rpa->geoms, rpa->geompos_in, rpa->z, rpa->tx, rpa->ty, rpa->initial_x, rpa->initial_y, rpa->original_features, rpa->unclipped_features, rpa->nextzoom, rpa->maxzoom, rpa->minzoom, rpa->max_zoom_increment, rpa->pass, rpa->along, rpa->alongminus, rpa->buffer, rpa->within, rpa->geomfile, rpa->geompos, rpa->start_geompos, rpa->oprogress, rpa->todo, rpa->fname, rpa->child_shards, rpa->filter, rpa->global_stringpool, rpa->pool_off, rpa->layer_unmaps, rpa->first_time, rpa->compressed, &multiplier_state, tile_stringpool, *(rpa->unidecode_data), previndex);
serial_feature sf = next_feature(rpa->geoms, rpa->geompos_in, rpa->z, rpa->tx, rpa->ty, rpa->initial_x, rpa->initial_y, rpa->original_features, rpa->unclipped_features, rpa->nextzoom, rpa->maxzoom, rpa->minzoom, rpa->max_zoom_increment, rpa->pass, rpa->along, rpa->alongminus, rpa->buffer, rpa->within, rpa->geomfile, rpa->geompos, rpa->start_geompos, rpa->oprogress, rpa->todo, rpa->fname, rpa->child_shards, rpa->filter, rpa->global_stringpool, rpa->pool_off, rpa->layer_unmaps, rpa->first_time, rpa->compressed, &multiplier_state, tile_stringpool, *(rpa->unidecode_data), previndex, rpa->droprate);
if (sf.t < 0) {
break;
}
@@ -1804,6 +1806,7 @@ long long write_tile(decompressor *geoms, std::atomic<long long> *geompos_in, ch
rpa.unidecode_data = &unidecode_data;
rpa.first_time = first_time;
rpa.compressed = compressed_input;
rpa.droprate = arg->droprate;
// this does need to be a real thread, so we can pipe both to and from it
if (pthread_create(&prefilter_writer, NULL, run_prefilter, &rpa) != 0) {
@@ -1835,7 +1838,7 @@ long long write_tile(decompressor *geoms, std::atomic<long long> *geompos_in, ch
ssize_t which_serial_feature = -1;
if (prefilter == NULL) {
sf = next_feature(geoms, geompos_in, z, tx, ty, initial_x, initial_y, &original_features, &unclipped_features, nextzoom, maxzoom, minzoom, max_zoom_increment, pass, along, alongminus, buffer, within, geomfile, geompos, start_geompos, &oprogress, todo, fname, child_shards, filter, global_stringpool, pool_off, layer_unmaps, first_time, compressed_input, &multiplier_state, tile_stringpool, unidecode_data, next_feature_previndex);
sf = next_feature(geoms, geompos_in, z, tx, ty, initial_x, initial_y, &original_features, &unclipped_features, nextzoom, maxzoom, minzoom, max_zoom_increment, pass, along, alongminus, buffer, within, geomfile, geompos, start_geompos, &oprogress, todo, fname, child_shards, filter, global_stringpool, pool_off, layer_unmaps, first_time, compressed_input, &multiplier_state, tile_stringpool, unidecode_data, next_feature_previndex, arg->droprate);
} else {
sf = parse_feature(prefilter_jp, z, tx, ty, layermaps, tiling_seg, layer_unmaps, postfilter != NULL);
}
@@ -1894,7 +1897,7 @@ long long write_tile(decompressor *geoms, std::atomic<long long> *geompos_in, ch
}
if (sf.dropped == FEATURE_DROPPED || drop_rest) {
multiplier_seq = (multiplier_seq + 1) % retain_points_multiplier;
multiplier_seq = 0;
if (find_feature_to_accumulate_onto(features, sf, which_serial_feature, layer_unmaps, LLONG_MAX, multiplier_seq)) {
preserve_attributes(arg->attribute_accum, sf, features[which_serial_feature]);