contrapunctus/MapComplete
1
0
Fork 0
forked from MapComplete/MapComplete
Code Pull requests Activity

Chore: update name suggestion index

Browse source
This commit is contained in:
Pieter Vander Vennet 2025-01-08 14:21:07 +01:00
parent 9f80e60ed6
commit c7e34f1a86
356 changed files with 46221 additions and 38409 deletions
Download patch file Download diff file Expand all files Collapse all files

7
scripts/downloadNsiLogos.ts
View file

@ -26,8 +26,8 @@ class DownloadNsiLogos extends Script {
let path = basePath + nsiItem.id
const logos = nsiWD["wikidata"][nsiItem?.tags?.[type + ":wikidata"]]?.logos
if (NameSuggestionIndex.isSvg(nsiItem, type)) {
const nsi = await NameSuggestionIndex.getNsiIndex()
if (nsi.isSvg(nsiItem, type)) {
path = path + ".svg"
}
@ -92,7 +92,8 @@ class DownloadNsiLogos extends Script {
}
async downloadFor(type: "brand" | "operator"): Promise<void> {
const items = NameSuggestionIndex.allPossible(type)
const nsi = await NameSuggestionIndex.getNsiIndex()
const items = nsi.allPossible(type)
const basePath = "./public/assets/data/nsi/logos/"
let downloadCount = 0
const stepcount = 5

99
scripts/generateStats.ts
View file

@ -2,7 +2,7 @@ import known_layers from "../src/assets/generated/known_layers.json"
import { LayerConfigJson } from "../src/Models/ThemeConfig/Json/LayerConfigJson"
import { TagUtils } from "../src/Logic/Tags/TagUtils"
import { Utils } from "../src/Utils"
import { existsSync, readFileSync, writeFileSync } from "fs"
import { copyFileSync, existsSync, readFileSync, writeFileSync } from "fs"
import ScriptUtils from "./ScriptUtils"
import TagRenderingConfig from "../src/Models/ThemeConfig/TagRenderingConfig"
import { And } from "../src/Logic/Tags/And"
@ -13,7 +13,7 @@ import TagInfo from "../src/Logic/Web/TagInfo"
class Utilities {
static mapValues<X extends string | number, T, TOut>(
record: Record<X, T>,
f: (t: T) => TOut
f: (t: T) => TOut,
): Record<X, TOut> {
const newR = <Record<X, TOut>>{}
for (const x in record) {
@ -26,7 +26,7 @@ class Utilities {
class GenerateStats extends Script {
async createOptimizationFile(includeTags = true) {
ScriptUtils.fixUtils()
const layers = <LayerConfigJson[]>known_layers.layers
const layers = <LayerConfigJson[]>known_layers["layers"]
const keysAndTags = new Map<string, Set<string>>()
@ -77,10 +77,10 @@ class GenerateStats extends Script {
const count = tagData.data.find((item) => item.type === "all").count
tagTotal.get(key).set(value, count)
console.log(key + "=" + value, "-->", count)
})
}),
)
}
})
}),
)
writeFileSync(
"./src/assets/key_totals.json",
@ -92,8 +92,8 @@ class GenerateStats extends Script {
tags: Utils.MapToObj(tagTotal, (v) => Utils.MapToObj(v, (t) => t)),
},
null,
" "
)
" ",
),
)
}
@ -139,7 +139,7 @@ class GenerateStats extends Script {
async createNameSuggestionIndexFile(basepath: string, type: "brand" | "operator" | string) {
const path = basepath + type + ".json"
let allBrands = <Record<string, Record<string, number>>>{}
let allBrands: Record<string, Record<string, number>> = {}
if (existsSync(path)) {
allBrands = JSON.parse(readFileSync(path, "utf8"))
console.log(
@ -147,82 +147,65 @@ class GenerateStats extends Script {
Object.keys(allBrands).length,
" previously loaded " + type,
"from",
path
path,
)
}
let skipped = 0
const nsi = await NameSuggestionIndex.getNsiIndex()
const allBrandNames: string[] = Utils.Dedup(
NameSuggestionIndex.allPossible(type).map((item) => item.tags[type])
nsi.allPossible(<any>type).map((item) => item.tags[type]),
)
const missingBrandNames: string[] = []
for (let i = 0; i < allBrandNames.length; i++) {
const brand = allBrandNames[i]
if (!!allBrands[brand] && Object.keys(allBrands[brand]).length == 0) {
delete allBrands[brand]
console.log("Deleted", brand, "as no entries at all")
}
if (allBrands[brand] !== undefined) {
const max = Math.max(...Object.values(allBrands[brand]))
skipped++
if (skipped % 100 == 0) {
console.warn("Busy; ", i + "/" + allBrandNames.length, "; skipped", skipped)
}
if (max < 0) {
console.log("HMMMM:", allBrands[brand])
delete allBrands[brand]
} else {
continue
}
}
missingBrandNames.push(brand)
}
const batchSize = 101
for (let i = 0; i < missingBrandNames.length; i += batchSize) {
const batchSize = 50
for (let i = 0; i < allBrandNames.length; i += batchSize) {
console.warn(
"Downloading",
batchSize,
"items: ",
i + "/" + missingBrandNames.length,
"; skipped",
skipped,
"total:",
allBrandNames.length
i + "/" + allBrandNames.length,
)
const distributions = await Promise.all(
let downloaded = 0
await Promise.all(
Utils.TimesT(batchSize, async (j) => {
await ScriptUtils.sleep(j * 250)
return TagInfo.getGlobalDistributionsFor(type, missingBrandNames[i + j])
})
)
for (let j = 0; j < distributions.length; j++) {
const brand = missingBrandNames[i + j]
const distribution: Record<string, number> = Utilities.mapValues(
distributions[j],
(s) => s.data.find((t) => t.type === "all").count
)
allBrands[brand] = distribution
}
const brand = allBrandNames[i + j]
if (!allBrands[brand]) {
allBrands[brand] = {}
}
const writeInto = allBrands[brand]
const dloaded = await TagInfo.getGlobalDistributionsFor(
writeInto, (stats) => stats.data.find((t) => t.type === "all").count,
type, brand)
downloaded += dloaded
}))
console.log("Downloaded ", downloaded, " values this batch")
writeFileSync(path, JSON.stringify(allBrands), "utf8")
console.log("Checkpointed", path)
}
console.log("Written:", path)
writeFileSync(path, JSON.stringify(allBrands), "utf8")
}
constructor() {
super(
"Downloads stats on osmSource-tags and keys from tagInfo. There are two usecases with separate outputs:\n 1. To optimize the query before sending it to overpass (generates ./src/assets/key_totals.json) \n 2. To amend the Name Suggestion Index "
"Downloads stats on osmSource-tags and keys from tagInfo. There are two usecases with separate outputs:\n 1. To optimize the query before sending it to overpass (generates ./src/assets/key_totals.json) \n 2. To amend the Name Suggestion Index ",
)
}
async main(_: string[]) {
const basepath = "./src/assets/generated/stats/"
await this.createOptimizationFile()
const target = "./public/assets/data/nsi/"
const basepath = target + "stats/"
{
const src = "./node_modules/name-suggestion-index/dist/"
const files = ["featureCollection.min.json", "nsi.min.json", "wikidata.min.json"]
console.log(process.cwd())
for (const file of files) {
console.log("Copying ", src + file, target + "/" + file)
copyFileSync(src + file, target + file)
}
}
for (const type of ["operator", "brand"]) {
await this.createNameSuggestionIndexFile(basepath, type)
this.summarizeNSI(basepath + type + ".json", "./public/assets/data/nsi/stats/" + type)
}
await this.createOptimizationFile()
}
}