Add alternative method to fetch user

2020-02-03 02:24:14 +13:00
parent 96fa4758c0
commit 272f4b6e3b
10 changed files with 191 additions and 65 deletions
--- a/src/lib/collectors.js
+++ b/src/lib/collectors.js
@@ -12,28 +12,87 @@ const requestCache = new RequestCache(constants.caching.resource_cache_time)
 const timelineEntryCache = new TtlCache(constants.caching.resource_cache_time)
 const history = new RequestHistory(["user", "timeline", "post"])

-function fetchUser(username) {
-	return requestCache.getOrFetch("user/"+username, () => {
-		return request(`https://www.instagram.com/${username}/`).then(res => {
-			if (res.status === 302) {
-				history.report("user", false)
-				throw constants.symbols.INSTAGRAM_DEMANDS_LOGIN
-			} else if (res.status === 404) {
-				throw constants.symbols.NOT_FOUND
-			} else return res.text().then(text => {
-				// require down here or have to deal with require loop. require cache will take care of it anyway.
-				// User -> Timeline -> TimelineImage -> collectors -/> User
-				const User = require("./structures/User")
-				const sharedData = extractSharedData(text)
-				const user = new User(sharedData.entry_data.ProfilePage[0].graphql.user)
-				history.report("user", true)
-				if (constants.caching.db_user_id) {
-					db.prepare("INSERT OR IGNORE INTO Users (username, user_id) VALUES (@username, @user_id)")
-						.run({username: user.data.username, user_id: user.data.id})
-				}
-				return user
-			})
+async function fetchUser(username) {
+	if (constants.allow_user_from_reel === "never") {
+		return fetchUserFromHTML(username)
+	} else if (constants.allow_user_from_reel === "prefer") {
+		const userID = db.prepare("SELECT user_id FROM Users WHERE username = ?").pluck().get(username)
+		if (userID) return fetchUserFromCombined(userID, username)
+		else return fetchUserFromHTML(username)
+	} else { // === "fallback"
+		return fetchUserFromHTML(username).catch(error => {
+			if (error === constants.symbols.INSTAGRAM_DEMANDS_LOGIN || error === constants.symbols.RATE_LIMITED) {
+				const userID = db.prepare("SELECT user_id FROM Users WHERE username = ?").pluck().get(username)
+				if (userID) return fetchUserFromCombined(userID, username)
+			}
+			throw error
 		})
+	}
+}
+
+function fetchUserFromHTML(username) {
+	return requestCache.getOrFetch("user/"+username, () => {
+		return switcher.request("user_html", `https://www.instagram.com/${username}/`, async res => {
+			if (res.status === 302) throw constants.symbols.INSTAGRAM_DEMANDS_LOGIN
+			if (res.status === 429) throw constants.symbols.RATE_LIMITED
+			return res
+		}).then(res => {
+			if (res.status === 404) {
+				throw constants.symbols.NOT_FOUND
+			} else {
+				return res.text().then(text => {
+					// require down here or have to deal with require loop. require cache will take care of it anyway.
+					// User -> Timeline -> TimelineEntry -> collectors -/> User
+					const User = require("./structures/User")
+					const sharedData = extractSharedData(text)
+					const user = new User(sharedData.entry_data.ProfilePage[0].graphql.user)
+					history.report("user", true)
+					if (constants.caching.db_user_id) {
+						db.prepare("INSERT OR IGNORE INTO Users (username, user_id) VALUES (@username, @user_id)")
+							.run({username: user.data.username, user_id: user.data.id})
+					}
+					return user
+				})
+			}
+		}).catch(error => {
+			if (error === constants.symbols.INSTAGRAM_DEMANDS_LOGIN || error === constants.symbols.RATE_LIMITED) {
+				history.report("user", false)
+			}
+			throw error
+		})
+	})
+}
+
+function fetchUserFromCombined(userID, username) {
+	// Fetch basic user information
+	const p = new URLSearchParams()
+	p.set("query_hash", constants.external.reel_query_hash)
+	p.set("variables", JSON.stringify({
+		user_id: userID,
+		include_reel: true
+	}))
+	return requestCache.getOrFetch("user/"+username, () => {
+		return switcher.request("reel_graphql", `https://www.instagram.com/graphql/query/?${p.toString()}`, async res => {
+			if (res.status === 429) throw constants.symbols.RATE_LIMITED
+			return res
+		}).then(res => res.json()).then(root => {
+			const result = root.data.user
+			if (!result) throw constants.symbols.NOT_FOUND
+			// require down here or have to deal with require loop. require cache will take care of it anyway.
+			// ReelUser -> Timeline -> TimelineEntry -> collectors -/> User
+			const ReelUser = require("./structures/ReelUser")
+			const user = new ReelUser(result.reel.user)
+			return user
+		}).catch(error => {
+			throw error
+		})
+	}).then(async user => {
+		// Add first timeline page
+		if (!user.timeline.pages[0]) {
+			const page = await fetchTimelinePage(userID, "")
+			user.timeline.addPage(page)
+		}
+		return user
 	})
 }

@@ -50,8 +109,8 @@ function fetchTimelinePage(userID, after) {
 		first: constants.external.timeline_fetch_first,
 		after: after
 	}))
-	return requestCache.getOrFetchPromise("page/"+after, () => {
-		return switcher.request(`https://www.instagram.com/graphql/query/?${p.toString()}`, async res => {
+	return requestCache.getOrFetchPromise(`page/${userID}/${after}`, () => {
+		return switcher.request("timeline_graphql", `https://www.instagram.com/graphql/query/?${p.toString()}`, async res => {
 			if (res.status === 429) throw constants.symbols.RATE_LIMITED
 			return res
 		}).then(res => res.json()).then(root => {
@@ -77,7 +136,7 @@ function getOrCreateShortcode(shortcode) {
 		return timelineEntryCache.get(shortcode)
 	} else {
 		// require down here or have to deal with require loop. require cache will take care of it anyway.
-		// TimelineImage -> collectors -/> TimelineImage
+		// TimelineEntry -> collectors -/> TimelineEntry
 		const TimelineEntry = require("./structures/TimelineEntry")
 		const result = new TimelineEntry()
 		timelineEntryCache.set(shortcode, result)
@@ -108,7 +167,7 @@ function fetchShortcodeData(shortcode) {
 	p.set("query_hash", constants.external.shortcode_query_hash)
 	p.set("variables", JSON.stringify({shortcode}))
 	return requestCache.getOrFetchPromise("shortcode/"+shortcode, () => {
-		return switcher.request(`https://www.instagram.com/graphql/query/?${p.toString()}`, async res => {
+		return switcher.request("post_graphql", `https://www.instagram.com/graphql/query/?${p.toString()}`, async res => {
 			if (res.status === 429) throw constants.symbols.RATE_LIMITED
 			return res
 		}).then(res => res.json()).then(root => {
@@ -123,6 +182,14 @@ function fetchShortcodeData(shortcode) {
 					db.prepare("REPLACE INTO Posts (shortcode, id, id_as_numeric, username, json) VALUES (@shortcode, @id, @id_as_numeric, @username, @json)")
 						.run({shortcode: data.shortcode, id: data.id, id_as_numeric: data.id, username: data.owner.username, json: JSON.stringify(data)})
 				}
+				// if we have the owner but only a reelUser, update it. this code is gross.
+				if (requestCache.hasNotPromise("user/"+data.owner.username)) {
+					const user = requestCache.getWithoutClean("user/"+data.owner.username)
+					if (user.fromReel) {
+						user.data.full_name = data.owner.full_name
+						user.data.is_verified = data.owner.is_verified
+					}
+				}
 				return data
 			}
 		}).catch(error => {