writer: Leaf directories: Find best base zoom

... to avoid extra indirection for as many tiles as we can
This commit is contained in:
EdwardO
2021-12-18 20:19:26 +01:00
parent 2ebc616cbd
commit dfe54a3c39

View File

@@ -2,6 +2,7 @@ import gzip
import itertools import itertools
import json import json
from contextlib import contextmanager from contextlib import contextmanager
from collections import defaultdict
@contextmanager @contextmanager
def write(fname): def write(fname):
@@ -19,6 +20,7 @@ class Writer:
self.tiles = [] self.tiles = []
self.hash_to_offset = {} self.hash_to_offset = {}
self.leaves = [] self.leaves = []
self.zoom_counts = defaultdict(int)
def write_tile(self,z,x,y,data): def write_tile(self,z,x,y,data):
# if the tile is GZIP-encoded, it won't work with range queries # if the tile is GZIP-encoded, it won't work with range queries
@@ -34,6 +36,7 @@ class Writer:
self.tiles.append((z,x,y,self.offset,len(data))) self.tiles.append((z,x,y,self.offset,len(data)))
self.hash_to_offset[hsh] = self.offset self.hash_to_offset[hsh] = self.offset
self.offset = self.offset + len(data) self.offset = self.offset + len(data)
self.zoom_counts[z] += 1
def write_entry(self,entry): def write_entry(self,entry):
self.f.write(entry[0].to_bytes(1,byteorder='little')) self.f.write(entry[0].to_bytes(1,byteorder='little'))
@@ -69,17 +72,24 @@ class Writer:
leafdir_tiles = [] leafdir_tiles = []
leafdir_len = 0 leafdir_len = 0
# Find best base zoom to avoid extra indirection for as many tiles as we can
base_zoom = 7
n_so_far = sum(self.zoom_counts[z] for z in range(0,8))
while n_so_far + self.zoom_counts[base_zoom+1] < 21845:
n_so_far += self.zoom_counts[base_zoom+1]
base_zoom += 1
def by_parent(t): def by_parent(t):
if t[0] >= 7: if t[0] >= base_zoom:
level_diff = t[0] - 7 level_diff = t[0] - base_zoom
return (7,t[1]//(1 << level_diff),t[2]//(1 << level_diff)) return (base_zoom,t[1]//(1 << level_diff),t[2]//(1 << level_diff))
else: else:
return (0,t[1]//(1 << t[0]),t[2]//(1 << t[0])) return (0,t[1]//(1 << t[0]),t[2]//(1 << t[0]))
# TODO optimize order # TODO optimize order
self.tiles.sort(key=by_parent) self.tiles.sort(key=by_parent)
for group in itertools.groupby(self.tiles,key=by_parent): for group in itertools.groupby(self.tiles,key=by_parent):
if group[0][0] != 7: if group[0][0] != base_zoom:
continue continue
entries = list(group[1]) entries = list(group[1])
if leafdir_len + len(entries) <= 21845: if leafdir_len + len(entries) <= 21845: