Don't do work for guessing maxzoom unless needed. Add a better test.

This commit is contained in:
Eric Fischer
2017-05-30 16:58:56 -07:00
parent 8776f17980
commit c7d146ea48
6 changed files with 2381 additions and 34 deletions
+28 -26
View File
@@ -185,7 +185,7 @@ long long parse_geometry(int t, json_object *j, long long *bbox, drawvec &out, i
return g;
}
int serialize_geometry(json_object *geometry, json_object *properties, json_object *id, const char *reading, int line, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, const char *fname, int basezoom, int layer, double droprate, long long *file_bbox, json_object *tippecanoe, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, json_object *feature, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count) {
int serialize_geometry(json_object *geometry, json_object *properties, json_object *id, const char *reading, int line, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, const char *fname, int basezoom, int layer, double droprate, long long *file_bbox, json_object *tippecanoe, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, json_object *feature, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count, bool want_dist) {
json_object *geometry_type = json_hash_get(geometry, "type");
if (geometry_type == NULL) {
static int warned = 0;
@@ -426,30 +426,32 @@ int serialize_geometry(json_object *geometry, json_object *properties, json_obje
dv = fix_polygon(dv);
}
std::vector<unsigned long long> locs;
for (size_t i = 0; i < dv.size(); i++) {
if (dv[i].op == VT_MOVETO || dv[i].op == VT_LINETO) {
locs.push_back(encode(dv[i].x << geometry_scale, dv[i].y << geometry_scale));
if (want_dist) {
std::vector<unsigned long long> locs;
for (size_t i = 0; i < dv.size(); i++) {
if (dv[i].op == VT_MOVETO || dv[i].op == VT_LINETO) {
locs.push_back(encode(dv[i].x << geometry_scale, dv[i].y << geometry_scale));
}
}
}
std::sort(locs.begin(), locs.end());
size_t n = 0;
double sum = 0;
for (size_t i = 1; i < locs.size(); i++) {
if (locs[i - 1] != locs[i]) {
sum += log(locs[i] - locs[i - 1]);
n++;
std::sort(locs.begin(), locs.end());
size_t n = 0;
double sum = 0;
for (size_t i = 1; i < locs.size(); i++) {
if (locs[i - 1] != locs[i]) {
sum += log(locs[i] - locs[i - 1]);
n++;
}
}
}
if (n > 0) {
double avg = exp(sum / n);
// Convert approximately from tile units to feet
double dist_ft = sqrt(avg) / 33;
if (n > 0) {
double avg = exp(sum / n);
// Convert approximately from tile units to feet
double dist_ft = sqrt(avg) / 33;
*dist_sum += log(dist_ft) * n;
*dist_count += n;
*dist_sum += log(dist_ft) * n;
*dist_count += n;
}
locs.clear();
}
locs.clear();
bool inline_meta = true;
// Don't inline metadata for features that will span several tiles at maxzoom
@@ -601,7 +603,7 @@ void check_crs(json_object *j, const char *reading) {
}
}
void parse_json(json_pull *jp, const char *reading, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, char *fname, int basezoom, int layer, double droprate, long long *file_bbox, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count) {
void parse_json(json_pull *jp, const char *reading, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, char *fname, int basezoom, int layer, double droprate, long long *file_bbox, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count, bool want_dist) {
long long found_hashes = 0;
long long found_features = 0;
long long found_geometries = 0;
@@ -669,7 +671,7 @@ void parse_json(json_pull *jp, const char *reading, volatile long long *layer_se
}
found_geometries++;
serialize_geometry(j, NULL, NULL, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, NULL, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count);
serialize_geometry(j, NULL, NULL, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, NULL, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count, want_dist);
json_free(j);
continue;
}
@@ -712,10 +714,10 @@ void parse_json(json_pull *jp, const char *reading, volatile long long *layer_se
if (geometries != NULL) {
size_t g;
for (g = 0; g < geometries->length; g++) {
serialize_geometry(geometries->array[g], properties, id, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, tippecanoe, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count);
serialize_geometry(geometries->array[g], properties, id, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, tippecanoe, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count, want_dist);
}
} else {
serialize_geometry(geometry, properties, id, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, tippecanoe, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count);
serialize_geometry(geometry, properties, id, reading, jp->line, layer_seq, progress_seq, metapos, geompos, indexpos, exclude, include, exclude_all, metafile, geomfile, indexfile, poolfile, treefile, fname, basezoom, layer, droprate, file_bbox, tippecanoe, segment, initialized, initial_x, initial_y, readers, maxzoom, j, layermap, layername, uses_gamma, attribute_types, dist_sum, dist_count, want_dist);
}
json_free(j);
@@ -727,7 +729,7 @@ void parse_json(json_pull *jp, const char *reading, volatile long long *layer_se
void *run_parse_json(void *v) {
struct parse_json_args *pja = (struct parse_json_args *) v;
parse_json(pja->jp, pja->reading, pja->layer_seq, pja->progress_seq, pja->metapos, pja->geompos, pja->indexpos, pja->exclude, pja->include, pja->exclude_all, pja->metafile, pja->geomfile, pja->indexfile, pja->poolfile, pja->treefile, pja->fname, pja->basezoom, pja->layer, pja->droprate, pja->file_bbox, pja->segment, pja->initialized, pja->initial_x, pja->initial_y, pja->readers, pja->maxzoom, pja->layermap, *pja->layername, pja->uses_gamma, pja->attribute_types, pja->dist_sum, pja->dist_count);
parse_json(pja->jp, pja->reading, pja->layer_seq, pja->progress_seq, pja->metapos, pja->geompos, pja->indexpos, pja->exclude, pja->include, pja->exclude_all, pja->metafile, pja->geomfile, pja->indexfile, pja->poolfile, pja->treefile, pja->fname, pja->basezoom, pja->layer, pja->droprate, pja->file_bbox, pja->segment, pja->initialized, pja->initial_x, pja->initial_y, pja->readers, pja->maxzoom, pja->layermap, *pja->layername, pja->uses_gamma, pja->attribute_types, pja->dist_sum, pja->dist_count, pja->want_dist);
return NULL;
}
+2 -1
View File
@@ -31,10 +31,11 @@ struct parse_json_args {
std::map<std::string, int> const *attribute_types;
double *dist_sum;
size_t *dist_count;
bool want_dist;
};
struct json_pull *json_begin_map(char *map, long long len);
void json_end_map(struct json_pull *jp);
void parse_json(json_pull *jp, const char *reading, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, char *fname, int basezoom, int layer, double droprate, long long *file_bbox, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count);
void parse_json(json_pull *jp, const char *reading, volatile long long *layer_seq, volatile long long *progress_seq, long long *metapos, long long *geompos, long long *indexpos, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, FILE *metafile, FILE *geomfile, FILE *indexfile, struct memfile *poolfile, struct memfile *treefile, char *fname, int basezoom, int layer, double droprate, long long *file_bbox, int segment, int *initialized, unsigned *initial_x, unsigned *initial_y, struct reader *readers, int maxzoom, std::map<std::string, layermap_entry> *layermap, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, double *dist_sum, size_t *dist_count, bool want_dist);
void *run_parse_json(void *v);
+10 -7
View File
@@ -372,7 +372,7 @@ void *run_sort(void *v) {
return NULL;
}
void do_read_parallel(char *map, long long len, long long initial_offset, const char *reading, struct reader *reader, volatile long long *progress_seq, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, char *fname, int basezoom, int source, int nlayers, std::vector<std::map<std::string, layermap_entry> > *layermaps, double droprate, int *initialized, unsigned *initial_x, unsigned *initial_y, int maxzoom, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, int separator, double *dist_sum, size_t *dist_count) {
void do_read_parallel(char *map, long long len, long long initial_offset, const char *reading, struct reader *reader, volatile long long *progress_seq, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, char *fname, int basezoom, int source, int nlayers, std::vector<std::map<std::string, layermap_entry> > *layermaps, double droprate, int *initialized, unsigned *initial_x, unsigned *initial_y, int maxzoom, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, int separator, double *dist_sum, size_t *dist_count, bool want_dist) {
long long segs[CPUS + 1];
segs[0] = 0;
segs[CPUS] = len;
@@ -437,6 +437,7 @@ void do_read_parallel(char *map, long long len, long long initial_offset, const
pja[i].attribute_types = attribute_types;
pja[i].dist_sum = &(dist_sums[i]);
pja[i].dist_count = &(dist_counts[i]);
pja[i].want_dist = want_dist;
if (pthread_create(&pthreads[i], NULL, run_parse_json, &pja[i]) != 0) {
perror("pthread_create");
@@ -487,6 +488,7 @@ struct read_parallel_arg {
std::map<std::string, int> const *attribute_types;
double *dist_sum;
size_t *dist_count;
bool want_dist;
};
void *run_read_parallel(void *v) {
@@ -508,7 +510,7 @@ void *run_read_parallel(void *v) {
}
madvise(map, rpa->len, MADV_RANDOM); // sequential, but from several pointers at once
do_read_parallel(map, rpa->len, rpa->offset, rpa->reading, rpa->reader, rpa->progress_seq, rpa->exclude, rpa->include, rpa->exclude_all, rpa->fname, rpa->basezoom, rpa->source, rpa->nlayers, rpa->layermaps, rpa->droprate, rpa->initialized, rpa->initial_x, rpa->initial_y, rpa->maxzoom, rpa->layername, rpa->uses_gamma, rpa->attribute_types, rpa->separator, rpa->dist_sum, rpa->dist_count);
do_read_parallel(map, rpa->len, rpa->offset, rpa->reading, rpa->reader, rpa->progress_seq, rpa->exclude, rpa->include, rpa->exclude_all, rpa->fname, rpa->basezoom, rpa->source, rpa->nlayers, rpa->layermaps, rpa->droprate, rpa->initialized, rpa->initial_x, rpa->initial_y, rpa->maxzoom, rpa->layername, rpa->uses_gamma, rpa->attribute_types, rpa->separator, rpa->dist_sum, rpa->dist_count, rpa->want_dist);
madvise(map, rpa->len, MADV_DONTNEED);
if (munmap(map, rpa->len) != 0) {
@@ -525,7 +527,7 @@ void *run_read_parallel(void *v) {
return NULL;
}
void start_parsing(int fd, FILE *fp, long long offset, long long len, volatile int *is_parsing, pthread_t *parallel_parser, bool &parser_created, const char *reading, struct reader *reader, volatile long long *progress_seq, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, char *fname, int basezoom, int source, int nlayers, std::vector<std::map<std::string, layermap_entry> > &layermaps, double droprate, int *initialized, unsigned *initial_x, unsigned *initial_y, int maxzoom, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, int separator, double *dist_sum, size_t *dist_count) {
void start_parsing(int fd, FILE *fp, long long offset, long long len, volatile int *is_parsing, pthread_t *parallel_parser, bool &parser_created, const char *reading, struct reader *reader, volatile long long *progress_seq, std::set<std::string> *exclude, std::set<std::string> *include, int exclude_all, char *fname, int basezoom, int source, int nlayers, std::vector<std::map<std::string, layermap_entry> > &layermaps, double droprate, int *initialized, unsigned *initial_x, unsigned *initial_y, int maxzoom, std::string layername, bool uses_gamma, std::map<std::string, int> const *attribute_types, int separator, double *dist_sum, size_t *dist_count, bool want_dist) {
// This has to kick off an intermediate thread to start the parser threads,
// so the main thread can get back to reading the next input stage while
// the intermediate thread waits for the completion of the parser threads.
@@ -566,6 +568,7 @@ void start_parsing(int fd, FILE *fp, long long offset, long long len, volatile i
rpa->attribute_types = attribute_types;
rpa->dist_sum = dist_sum;
rpa->dist_count = dist_count;
rpa->want_dist = want_dist;
if (pthread_create(parallel_parser, NULL, run_read_parallel, rpa) != 0) {
perror("pthread_create");
@@ -1250,7 +1253,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
}
if (map != NULL && map != MAP_FAILED && read_parallel_this) {
do_read_parallel(map, st.st_size - off, overall_offset, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, &layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, uses_gamma, attribute_types, read_parallel_this, &dist_sum, &dist_count);
do_read_parallel(map, st.st_size - off, overall_offset, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, &layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, uses_gamma, attribute_types, read_parallel_this, &dist_sum, &dist_count, guess_maxzoom);
overall_offset += st.st_size - off;
checkdisk(reader, CPUS);
@@ -1326,7 +1329,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
}
fflush(readfp);
start_parsing(readfd, readfp, initial_offset, ahead, &is_parsing, &parallel_parser, parser_created, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, gamma != 0, attribute_types, read_parallel_this, &dist_sum, &dist_count);
start_parsing(readfd, readfp, initial_offset, ahead, &is_parsing, &parallel_parser, parser_created, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, gamma != 0, attribute_types, read_parallel_this, &dist_sum, &dist_count, guess_maxzoom);
initial_offset += ahead;
overall_offset += ahead;
@@ -1363,7 +1366,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
fflush(readfp);
if (ahead > 0) {
start_parsing(readfd, readfp, initial_offset, ahead, &is_parsing, &parallel_parser, parser_created, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, gamma != 0, attribute_types, read_parallel_this, &dist_sum, &dist_count);
start_parsing(readfd, readfp, initial_offset, ahead, &is_parsing, &parallel_parser, parser_created, reading.c_str(), reader, &progress_seq, exclude, include, exclude_all, fname, basezoom, layer, nlayers, layermaps, droprate, initialized, initial_x, initial_y, maxzoom, sources[layer].layer, gamma != 0, attribute_types, read_parallel_this, &dist_sum, &dist_count, guess_maxzoom);
if (parser_created) {
if (pthread_join(parallel_parser, NULL) != 0) {
@@ -1380,7 +1383,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
long long layer_seq = overall_offset;
json_pull *jp = json_begin_file(fp);
parse_json(jp, reading.c_str(), &layer_seq, &progress_seq, &reader[0].metapos, &reader[0].geompos, &reader[0].indexpos, exclude, include, exclude_all, reader[0].metafile, reader[0].geomfile, reader[0].indexfile, reader[0].poolfile, reader[0].treefile, fname, basezoom, layer, droprate, reader[0].file_bbox, 0, &initialized[0], &initial_x[0], &initial_y[0], reader, maxzoom, &layermaps[0], sources[layer].layer, uses_gamma, attribute_types, &dist_sum, &dist_count);
parse_json(jp, reading.c_str(), &layer_seq, &progress_seq, &reader[0].metapos, &reader[0].geompos, &reader[0].indexpos, exclude, include, exclude_all, reader[0].metafile, reader[0].geomfile, reader[0].indexfile, reader[0].poolfile, reader[0].treefile, fname, basezoom, layer, droprate, reader[0].file_bbox, 0, &initialized[0], &initial_x[0], &initial_y[0], reader, maxzoom, &layermaps[0], sources[layer].layer, uses_gamma, attribute_types, &dist_sum, &dist_count, guess_maxzoom);
json_end(jp);
overall_offset = layer_seq;
checkdisk(reader, CPUS);
File diff suppressed because one or more lines are too long
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff