2023-01-09 20:30:13 +01:00
|
|
|
import Script from "./Script"
|
2023-07-27 01:48:54 +02:00
|
|
|
import { Overpass } from "../src/Logic/Osm/Overpass"
|
|
|
|
import { RegexTag } from "../src/Logic/Tags/RegexTag"
|
|
|
|
import { ImmutableStore } from "../src/Logic/UIEventSource"
|
|
|
|
import { BBox } from "../src/Logic/BBox"
|
2023-01-09 20:30:13 +01:00
|
|
|
import * as fs from "fs"
|
2023-10-17 13:31:06 +02:00
|
|
|
import { writeFileSync } from "fs"
|
2023-06-01 14:32:45 +02:00
|
|
|
import { Feature } from "geojson"
|
2023-01-09 20:30:13 +01:00
|
|
|
import ScriptUtils from "./ScriptUtils"
|
2023-07-27 01:48:54 +02:00
|
|
|
import { Imgur } from "../src/Logic/ImageProviders/Imgur"
|
|
|
|
import { LicenseInfo } from "../src/Logic/ImageProviders/LicenseInfo"
|
|
|
|
import { Utils } from "../src/Utils"
|
|
|
|
import Constants from "../src/Models/Constants"
|
2023-01-09 20:30:13 +01:00
|
|
|
|
|
|
|
export default class GenerateImageAnalysis extends Script {
|
|
|
|
constructor() {
|
|
|
|
super(
|
|
|
|
"Downloads (from overpass) all tags which have an imgur-image; then analyses the licenses"
|
|
|
|
)
|
|
|
|
}
|
|
|
|
|
2023-05-18 13:07:14 +02:00
|
|
|
async fetchImages(key: string, datapath: string, refresh: boolean): Promise<void> {
|
2023-01-09 20:30:13 +01:00
|
|
|
const targetPath = `${datapath}/features_with_${key.replace(/[:\/]/, "_")}.geojson`
|
2023-05-18 13:07:14 +02:00
|
|
|
if (fs.existsSync(targetPath) && !refresh) {
|
2023-01-09 20:30:13 +01:00
|
|
|
console.log("Skipping", key)
|
|
|
|
return
|
|
|
|
}
|
2023-01-10 02:52:09 +01:00
|
|
|
const tag = new RegexTag(key, /^https:\/\/i.imgur.com\/.*$/i)
|
2023-01-09 20:30:13 +01:00
|
|
|
const overpass = new Overpass(
|
|
|
|
tag,
|
|
|
|
[],
|
2023-05-18 13:07:14 +02:00
|
|
|
Constants.defaultOverpassUrls[0], //"https://overpass.kumi.systems/api/interpreter",
|
2023-04-20 23:47:51 +02:00
|
|
|
new ImmutableStore(500),
|
2023-01-09 20:30:13 +01:00
|
|
|
false
|
|
|
|
)
|
|
|
|
console.log("Starting query...")
|
|
|
|
const data = await overpass.queryGeoJson(BBox.global)
|
2023-06-01 14:32:45 +02:00
|
|
|
console.log(
|
|
|
|
"Got data:",
|
|
|
|
data[0].features.length,
|
|
|
|
"items; timestamp:",
|
|
|
|
data[1].toISOString()
|
|
|
|
)
|
2023-01-09 20:30:13 +01:00
|
|
|
fs.writeFileSync(targetPath, JSON.stringify(data[0]), "utf8")
|
|
|
|
console.log("Written", targetPath)
|
|
|
|
}
|
|
|
|
|
2023-05-18 13:07:14 +02:00
|
|
|
async downloadData(datapath: string, refresh: boolean): Promise<void> {
|
2023-01-09 20:30:13 +01:00
|
|
|
if (!fs.existsSync(datapath)) {
|
|
|
|
fs.mkdirSync(datapath)
|
|
|
|
}
|
2023-05-18 13:07:14 +02:00
|
|
|
await this.fetchImages("image", datapath, refresh)
|
|
|
|
await this.fetchImages("image:streetsign", datapath, refresh)
|
2023-01-09 20:30:13 +01:00
|
|
|
for (let i = 0; i < 5; i++) {
|
2023-05-18 13:07:14 +02:00
|
|
|
await this.fetchImages("image:" + i, datapath, refresh)
|
2023-01-09 20:30:13 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
loadData(datapath: string): Feature[] {
|
|
|
|
const allFeatures: Feature[] = []
|
|
|
|
|
|
|
|
const files = ScriptUtils.readDirRecSync(datapath)
|
|
|
|
for (const file of files) {
|
|
|
|
if (!file.endsWith(".geojson")) {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
const contents = JSON.parse(fs.readFileSync(file, "utf8"))
|
|
|
|
allFeatures.push(...contents.features)
|
|
|
|
}
|
|
|
|
|
|
|
|
return allFeatures
|
|
|
|
}
|
|
|
|
|
|
|
|
async fetchImageMetadata(datapath: string, image: string): Promise<boolean> {
|
|
|
|
if (image === undefined) {
|
|
|
|
return false
|
|
|
|
}
|
2023-01-10 02:52:09 +01:00
|
|
|
if (!image.match(/https:\/\/i\.imgur\.com\/[a-zA-Z0-9]+\.jpg/)) {
|
2023-01-09 20:30:13 +01:00
|
|
|
return false
|
|
|
|
}
|
2023-05-18 13:07:14 +02:00
|
|
|
const filename = image.replace(/[\/:.\-%]/g, "_") + ".json"
|
|
|
|
const targetPath = datapath + "/" + filename
|
2023-01-09 20:30:13 +01:00
|
|
|
if (fs.existsSync(targetPath)) {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
const attribution = await Imgur.singleton.DownloadAttribution(image)
|
2023-05-18 13:07:14 +02:00
|
|
|
|
|
|
|
if ((attribution.artist ?? "") === "") {
|
|
|
|
// This is an invalid attribution. We save the raw response as well
|
|
|
|
const hash = image.substr("https://i.imgur.com/".length).split(".jpg")[0]
|
|
|
|
|
|
|
|
const apiUrl = "https://api.imgur.com/3/image/" + hash
|
|
|
|
const response = await Utils.downloadJsonCached(apiUrl, 365 * 24 * 60 * 60, {
|
|
|
|
Authorization: "Client-ID " + Constants.ImgurApiKey,
|
|
|
|
})
|
|
|
|
const rawTarget = datapath + "/raw/" + filename
|
|
|
|
console.log("Also storing the raw response to", rawTarget)
|
|
|
|
await fs.writeFileSync(rawTarget, JSON.stringify(response, null, " "))
|
|
|
|
}
|
|
|
|
|
2023-01-09 20:30:13 +01:00
|
|
|
await fs.writeFileSync(targetPath, JSON.stringify(attribution, null, " "))
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
|
2023-06-01 14:32:45 +02:00
|
|
|
loadImageUrls(datapath: string): { allImages: Set<string>; imageSource: Map<string, string> } {
|
2023-01-09 20:30:13 +01:00
|
|
|
let allImages = new Set<string>()
|
2023-05-18 13:07:14 +02:00
|
|
|
const features = this.loadData(datapath)
|
|
|
|
let imageSource: Map<string, string> = new Map<string, string>()
|
2023-01-09 20:30:13 +01:00
|
|
|
|
|
|
|
for (const feature of features) {
|
|
|
|
allImages.add(feature.properties["image"])
|
2023-05-18 13:07:14 +02:00
|
|
|
imageSource[feature.properties["image"]] = feature.properties.id
|
|
|
|
allImages.add(feature.properties["image:streetsign"])
|
2023-06-01 14:32:45 +02:00
|
|
|
imageSource[feature.properties["image:streetsign"]] =
|
|
|
|
feature.properties.id + " (streetsign)"
|
2023-05-18 13:07:14 +02:00
|
|
|
|
2023-01-09 20:30:13 +01:00
|
|
|
for (let i = 0; i < 10; i++) {
|
|
|
|
allImages.add(feature.properties["image:" + i])
|
2023-06-01 14:32:45 +02:00
|
|
|
imageSource[
|
|
|
|
feature.properties["image:" + i]
|
|
|
|
] = `${feature.properties.id} (image:${i})`
|
2023-01-09 20:30:13 +01:00
|
|
|
}
|
|
|
|
}
|
2023-05-18 13:07:14 +02:00
|
|
|
allImages.delete(undefined)
|
|
|
|
allImages.delete(null)
|
|
|
|
imageSource.delete(undefined)
|
|
|
|
imageSource.delete(null)
|
2023-06-01 14:32:45 +02:00
|
|
|
return { allImages, imageSource }
|
2023-05-18 13:07:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
async downloadMetadata(datapath: string): Promise<void> {
|
2023-06-01 14:32:45 +02:00
|
|
|
const { allImages, imageSource } = this.loadImageUrls(datapath)
|
2023-01-09 20:30:13 +01:00
|
|
|
console.log("Detected", allImages.size, "images")
|
|
|
|
let i = 0
|
|
|
|
let d = 0
|
|
|
|
let s = 0
|
|
|
|
let f = 0
|
|
|
|
let start = Date.now()
|
|
|
|
for (const image of Array.from(allImages)) {
|
|
|
|
i++
|
|
|
|
try {
|
|
|
|
const downloaded = await this.fetchImageMetadata(datapath, image)
|
|
|
|
const runningSecs = (Date.now() - start) / 1000
|
|
|
|
const left = allImages.size - i
|
|
|
|
|
|
|
|
const estimatedActualSeconds = Math.floor((left * runningSecs) / (f + d))
|
|
|
|
const estimatedActualMinutes = Math.floor(estimatedActualSeconds / 60)
|
|
|
|
|
|
|
|
const msg = `${i}/${
|
|
|
|
allImages.size
|
2023-05-07 23:50:39 +02:00
|
|
|
} downloaded: ${d},skipped: ${s}, failed: ${f}, running: ${Math.floor(
|
|
|
|
runningSecs
|
|
|
|
)}sec, ETA: ${estimatedActualMinutes}:${estimatedActualSeconds % 60}`
|
2023-06-01 14:32:45 +02:00
|
|
|
if (d + (f % 1000) === 1 || downloaded) {
|
2023-05-18 13:07:14 +02:00
|
|
|
ScriptUtils.erasableLog(msg)
|
|
|
|
}
|
2023-01-09 20:30:13 +01:00
|
|
|
if (downloaded) {
|
|
|
|
d++
|
|
|
|
} else {
|
|
|
|
s++
|
|
|
|
}
|
|
|
|
if (d + f == 75000) {
|
|
|
|
console.log("Used 75000 API calls, leaving 5000 for the rest of the day...")
|
2023-05-18 13:07:14 +02:00
|
|
|
break
|
2023-01-09 20:30:13 +01:00
|
|
|
}
|
|
|
|
} catch (e) {
|
2023-05-18 13:07:14 +02:00
|
|
|
// console.log(e)
|
2023-06-01 14:32:45 +02:00
|
|
|
console.log(
|
|
|
|
"Offending image hash is",
|
|
|
|
image,
|
|
|
|
"from https://openstreetmap.org/" + imageSource[image]
|
|
|
|
)
|
2023-01-09 20:30:13 +01:00
|
|
|
f++
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-05-18 13:07:14 +02:00
|
|
|
async downloadImage(url: string, imagePath: string): Promise<boolean> {
|
|
|
|
const filenameLong = url.replace(/[\/:.\-%]/g, "_") + ".jpg"
|
|
|
|
const targetPathLong = imagePath + "/" + filenameLong
|
|
|
|
|
|
|
|
const filename = url.substring("https://i.imgur.com/".length)
|
|
|
|
const targetPath = imagePath + "/" + filename
|
|
|
|
if (fs.existsSync(targetPathLong)) {
|
|
|
|
if (fs.existsSync(targetPath)) {
|
|
|
|
fs.unlinkSync(targetPathLong)
|
|
|
|
console.log("Unlinking duplicate")
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
console.log("Renaming...")
|
|
|
|
fs.renameSync(targetPathLong, targetPath)
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
if (fs.existsSync(targetPath)) {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
await ScriptUtils.DownloadFileTo(url, targetPath)
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
|
|
|
|
async downloadAllImages(datapath: string, imagePath: string): Promise<void> {
|
2023-06-01 14:32:45 +02:00
|
|
|
const { allImages } = this.loadImageUrls(datapath)
|
2023-05-18 13:07:14 +02:00
|
|
|
let skipped = 0
|
|
|
|
let failed = 0
|
|
|
|
let downloaded = 0
|
|
|
|
let invalid = 0
|
|
|
|
const startTime = Date.now()
|
2023-06-01 14:32:45 +02:00
|
|
|
const urls = Array.from(allImages).filter((url) => url.startsWith("https://i.imgur.com"))
|
2023-05-18 13:07:14 +02:00
|
|
|
for (const url of urls) {
|
2023-06-01 14:32:45 +02:00
|
|
|
const runningTime = (Date.now() - startTime) / 1000
|
2023-05-18 13:07:14 +02:00
|
|
|
const handled = skipped + downloaded + failed
|
|
|
|
const itemsLeft = allImages.size - handled
|
|
|
|
const speed = handled / runningTime
|
|
|
|
const timeLeft = Math.round(itemsLeft * speed)
|
|
|
|
try {
|
2023-06-04 22:14:23 +02:00
|
|
|
const urls = url.split(/[;,]/)
|
2023-06-01 14:32:45 +02:00
|
|
|
const downloadedStatus = await Promise.all(
|
2023-06-04 22:14:23 +02:00
|
|
|
urls.map((url) => this.downloadImage(url.trim(), imagePath))
|
2023-06-01 14:32:45 +02:00
|
|
|
)
|
2023-05-18 13:07:14 +02:00
|
|
|
|
|
|
|
for (const b of downloadedStatus) {
|
|
|
|
if (b) {
|
|
|
|
downloaded += 1
|
|
|
|
} else {
|
|
|
|
skipped += 1
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-01 14:32:45 +02:00
|
|
|
if (downloadedStatus.some((i) => i) || skipped % 10000 === 0) {
|
|
|
|
console.log(
|
|
|
|
"Handled",
|
|
|
|
url,
|
|
|
|
JSON.stringify({
|
|
|
|
skipped,
|
|
|
|
failed,
|
|
|
|
downloaded,
|
|
|
|
invalid,
|
|
|
|
total: allImages.size,
|
|
|
|
eta: timeLeft + "s",
|
|
|
|
})
|
|
|
|
)
|
2023-05-18 13:07:14 +02:00
|
|
|
}
|
|
|
|
} catch (e) {
|
|
|
|
console.log(e)
|
|
|
|
failed++
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-01-09 20:30:13 +01:00
|
|
|
analyze(datapath: string) {
|
|
|
|
const files = ScriptUtils.readDirRecSync(datapath)
|
|
|
|
const byAuthor = new Map<string, string[]>()
|
|
|
|
const byLicense = new Map<string, string[]>()
|
|
|
|
const licenseByAuthor = new Map<string, Set<string>>()
|
|
|
|
for (const file of files) {
|
|
|
|
if (!file.endsWith(".json")) {
|
|
|
|
continue
|
|
|
|
}
|
2023-06-01 14:32:45 +02:00
|
|
|
const attr = <LicenseInfo>JSON.parse(fs.readFileSync(file, { encoding: "utf8" }))
|
2023-01-10 02:52:09 +01:00
|
|
|
const license = attr.licenseShortName
|
2023-01-09 20:30:13 +01:00
|
|
|
|
2023-01-10 02:52:09 +01:00
|
|
|
if (license === undefined || attr.artist === undefined) {
|
|
|
|
continue
|
|
|
|
}
|
2023-01-09 20:30:13 +01:00
|
|
|
if (byAuthor.get(attr.artist) === undefined) {
|
|
|
|
byAuthor.set(attr.artist, [])
|
|
|
|
}
|
|
|
|
byAuthor.get(attr.artist).push(file)
|
|
|
|
|
|
|
|
if (byLicense.get(license) === undefined) {
|
|
|
|
byLicense.set(license, [])
|
|
|
|
}
|
|
|
|
byLicense.get(license).push(file)
|
|
|
|
|
|
|
|
if (licenseByAuthor.get(license) === undefined) {
|
|
|
|
licenseByAuthor.set(license, new Set<string>())
|
|
|
|
}
|
|
|
|
licenseByAuthor.get(license).add(attr.artist)
|
|
|
|
}
|
|
|
|
byAuthor.delete(undefined)
|
|
|
|
byLicense.delete(undefined)
|
|
|
|
licenseByAuthor.delete(undefined)
|
|
|
|
|
|
|
|
const byLicenseCount = Utils.MapToObj(byLicense, (a) => a.length)
|
|
|
|
const byAuthorCount = Utils.MapToObj(byAuthor, (a) => a.length)
|
|
|
|
const licenseByAuthorCount = Utils.MapToObj(licenseByAuthor, (a) => a.size)
|
2023-01-10 02:52:09 +01:00
|
|
|
|
|
|
|
const countsPerAuthor: number[] = Array.from(Object.keys(byAuthorCount)).map(
|
|
|
|
(k) => byAuthorCount[k]
|
|
|
|
)
|
|
|
|
console.log(countsPerAuthor)
|
|
|
|
countsPerAuthor.sort()
|
|
|
|
const median = countsPerAuthor[Math.floor(countsPerAuthor.length / 2)]
|
2023-10-30 13:44:27 +01:00
|
|
|
const json: {
|
|
|
|
leaderboard: { rank: number; account: string; name: string; nrOfImages: number }[]
|
|
|
|
} = {
|
|
|
|
leaderboard: [],
|
2023-10-17 13:31:06 +02:00
|
|
|
}
|
2023-01-10 02:52:09 +01:00
|
|
|
for (let i = 0; i < 100; i++) {
|
|
|
|
let maxAuthor: string = undefined
|
|
|
|
let maxCount = 0
|
|
|
|
for (const author in byAuthorCount) {
|
|
|
|
const count = byAuthorCount[author]
|
|
|
|
if (maxAuthor === undefined || count > maxCount) {
|
|
|
|
maxAuthor = author
|
|
|
|
maxCount = count
|
|
|
|
}
|
|
|
|
}
|
2023-10-17 13:31:06 +02:00
|
|
|
json.leaderboard.push({
|
2023-10-30 13:44:27 +01:00
|
|
|
rank: i + 1,
|
2023-10-17 13:31:06 +02:00
|
|
|
name: maxAuthor,
|
2023-10-30 13:44:27 +01:00
|
|
|
account: "https://openstreetmap.org/user/" + maxAuthor.replace(/ /g, "%20"),
|
|
|
|
nrOfImages: maxCount,
|
2023-10-17 13:31:06 +02:00
|
|
|
})
|
2023-01-10 02:52:09 +01:00
|
|
|
console.log(
|
|
|
|
"|",
|
|
|
|
i + 1,
|
|
|
|
"|",
|
|
|
|
`[${maxAuthor}](https://openstreetmap.org/user/${maxAuthor.replace(/ /g, "%20")})`,
|
|
|
|
"|",
|
|
|
|
maxCount,
|
|
|
|
"|"
|
|
|
|
)
|
|
|
|
delete byAuthorCount[maxAuthor]
|
|
|
|
}
|
2023-01-09 20:30:13 +01:00
|
|
|
|
|
|
|
const totalAuthors = byAuthor.size
|
|
|
|
let totalLicensedImages = 0
|
2023-10-17 13:31:06 +02:00
|
|
|
json["totalAuthors"] = totalAuthors
|
2023-01-09 20:30:13 +01:00
|
|
|
for (const license in byLicenseCount) {
|
|
|
|
totalLicensedImages += byLicenseCount[license]
|
|
|
|
}
|
2023-10-17 13:31:06 +02:00
|
|
|
json["byLicense"] = {}
|
2023-01-09 20:30:13 +01:00
|
|
|
for (const license in byLicenseCount) {
|
|
|
|
const total = byLicenseCount[license]
|
|
|
|
const authors = licenseByAuthorCount[license]
|
|
|
|
console.log(
|
|
|
|
`License ${license}: ${total} total pictures (${
|
|
|
|
Math.floor((1000 * total) / totalLicensedImages) / 10
|
|
|
|
}%), ${authors} authors (${
|
|
|
|
Math.floor((1000 * authors) / totalAuthors) / 10
|
|
|
|
}%), ${Math.floor(total / authors)} images/author`
|
|
|
|
)
|
2023-10-17 13:31:06 +02:00
|
|
|
json["byLicense"] = {
|
2023-10-30 13:44:27 +01:00
|
|
|
license,
|
|
|
|
total,
|
|
|
|
authors,
|
2023-10-17 13:31:06 +02:00
|
|
|
}
|
2023-01-09 20:30:13 +01:00
|
|
|
}
|
2023-01-10 02:52:09 +01:00
|
|
|
|
|
|
|
const nonDefaultAuthors = [
|
|
|
|
...Array.from(licenseByAuthor.get("CC-BY 4.0").values()),
|
|
|
|
...Array.from(licenseByAuthor.get("CC-BY-SA 4.0").values()),
|
|
|
|
]
|
|
|
|
|
2023-06-01 14:32:45 +02:00
|
|
|
console.log(
|
|
|
|
"Total number of correctly licenses pictures: ",
|
|
|
|
totalLicensedImages,
|
|
|
|
"(out of ",
|
|
|
|
files.length,
|
|
|
|
" images)"
|
|
|
|
)
|
2023-01-10 02:52:09 +01:00
|
|
|
console.log("Total number of authors:", byAuthor.size)
|
|
|
|
console.log(
|
|
|
|
"Total number of authors which used a valid, non CC0 license at one point in time",
|
|
|
|
nonDefaultAuthors.length
|
|
|
|
)
|
|
|
|
console.log("Median contributions per author:", median)
|
2023-10-17 13:31:06 +02:00
|
|
|
json["median"] = median
|
|
|
|
json["date"] = new Date().toISOString()
|
2023-10-30 13:44:27 +01:00
|
|
|
writeFileSync(
|
|
|
|
"../../git/MapComplete-data/picture-leaderboard.json",
|
|
|
|
JSON.stringify(json),
|
|
|
|
"utf8"
|
|
|
|
)
|
2023-01-09 20:30:13 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
async main(args: string[]): Promise<void> {
|
2023-05-18 13:07:14 +02:00
|
|
|
console.log("Usage: [--cached] to use the cached osm data")
|
|
|
|
console.log("Args are", args)
|
|
|
|
const cached = args.indexOf("--cached") < 0
|
2023-06-01 14:32:45 +02:00
|
|
|
args = args.filter((a) => a !== "--cached")
|
2023-04-20 23:47:51 +02:00
|
|
|
const datapath = args[0] ?? "../../git/MapComplete-data/ImageLicenseInfo"
|
2023-05-18 13:07:14 +02:00
|
|
|
await this.downloadData(datapath, cached)
|
2023-01-09 20:30:13 +01:00
|
|
|
|
2023-04-20 23:47:51 +02:00
|
|
|
await this.downloadMetadata(datapath)
|
2023-05-18 13:07:14 +02:00
|
|
|
await this.downloadAllImages(datapath, "/home/pietervdvn/data/imgur-image-backup")
|
2023-01-09 20:30:13 +01:00
|
|
|
this.analyze(datapath)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
new GenerateImageAnalysis().run()
|