birdclaw 0.5.1 → 0.7.0

package/src/lib/timeline-live.ts

@@ -1,13 +1,44 @@
+import { Effect } from "effect";
 import type { Database } from "./sqlite";
-import { listHomeTimelineViaBird } from "./bird";
+import { listHomeTimelineViaBirdEffect } from "./bird";
 import { getNativeDb } from "./db";
+import { runEffectPromise } from "./effect-runtime";
 import { buildMediaJsonFromIncludes, countTweetMedia } from "./media-includes";
 import { readSyncCache, writeSyncCache } from "./sync-cache";
-import type { XurlMentionsResponse } from "./types";
+import type {
+	XurlMediaItem,
+	XurlMentionUser,
+	XurlMentionsResponse,
+} from "./types";
 import { upsertTweetAccountEdge } from "./tweet-account-edges";
 import { ensureStubProfileForXUser, upsertProfileFromXUser } from "./x-profile";
+import { listHomeTimelineViaXurlEffect } from "./xurl";
 
 const DEFAULT_TIMELINE_CACHE_TTL_MS = 2 * 60_000;
+const MAX_XURL_TIMELINE_PAGE_SIZE = 100;
+
+export type HomeTimelineMode = "bird" | "xurl" | "auto";
+export interface HomeTimelineProgress {
+	source: "bird" | "xurl" | "cache";
+	fetched: number;
+	total?: number;
+	page?: number;
+	maxPages?: number;
+	pageSize?: number;
+	done: boolean;
+}
+export interface SyncHomeTimelineOptions {
+	account?: string;
+	mode?: HomeTimelineMode;
+	limit?: number;
+	maxPages?: number;
+	startTime?: string;
+	following?: boolean;
+	refresh?: boolean;
+	cacheTtlMs?: number;
+	timeoutMs?: number;
+	onProgress?: (progress: HomeTimelineProgress) => void;
+}
 
 function parseCacheTtlMs(value?: number) {
 	if (typeof value !== "number" || !Number.isFinite(value) || value < 0) {
@@ -17,32 +48,130 @@ function parseCacheTtlMs(value?: number) {
 }
 
 function assertLimit(limit: number) {
-	if (!Number.isFinite(limit) || limit < 1) {
+	if ((!Number.isFinite(limit) && limit !== Infinity) || limit < 1) {
 		throw new Error("--limit must be at least 1");
 	}
 }
 
+function parseMode(mode: HomeTimelineMode | undefined) {
+	const parsed = mode ?? "bird";
+	if (parsed !== "bird" && parsed !== "xurl" && parsed !== "auto") {
+		throw new Error("--mode must be bird, xurl, or auto");
+	}
+	return parsed;
+}
+
+function parseMaxPages(maxPages: number | undefined) {
+	if (maxPages === undefined) return 1;
+	if (!Number.isFinite(maxPages) || maxPages < 1) {
+		throw new Error("--max-pages must be at least 1");
+	}
+	return Math.floor(maxPages);
+}
+
+function parseStartTime(value: string | undefined) {
+	if (!value?.trim()) return undefined;
+	const time = new Date(value).getTime();
+	if (!Number.isFinite(time)) {
+		throw new Error("--start-time must be a valid date");
+	}
+	return { iso: new Date(time).toISOString(), time };
+}
+
+function reachedStartTimeBoundary(
+	payload: XurlMentionsResponse,
+	startTimeMs: number | undefined,
+) {
+	if (startTimeMs === undefined) return false;
+	return payload.data.some((tweet) => {
+		const createdAt = new Date(tweet.created_at).getTime();
+		return Number.isFinite(createdAt) && createdAt <= startTimeMs;
+	});
+}
+
+function getReferencedTweetId(
+	tweet: XurlMentionsResponse["data"][number],
+	type: "replied_to" | "quoted",
+) {
+	return tweet.referenced_tweets?.find((reference) => reference.type === type)
+		?.id;
+}
+
+function mergeTimelinePayloads(
+	payloads: XurlMentionsResponse[],
+	limit: number,
+) {
+	const data: XurlMentionsResponse["data"] = [];
+	const usersById = new Map<string, XurlMentionUser>();
+	const mediaByKey = new Map<string, XurlMediaItem>();
+	let meta: XurlMentionsResponse["meta"] | undefined;
+
+	for (const payload of payloads) {
+		meta = payload.meta;
+		for (const tweet of payload.data) {
+			if (data.some((existing) => existing.id === tweet.id)) continue;
+			data.push(tweet);
+			if (data.length >= limit) break;
+		}
+		for (const user of payload.includes?.users ?? []) {
+			usersById.set(user.id, user);
+		}
+		for (const media of payload.includes?.media ?? []) {
+			mediaByKey.set(media.media_key, media);
+		}
+		if (data.length >= limit) break;
+	}
+
+	return {
+		data,
+		includes: {
+			users: [...usersById.values()],
+			media: [...mediaByKey.values()],
+		},
+		meta,
+	} satisfies XurlMentionsResponse;
+}
+
 function resolveAccount(db: Database, accountId?: string) {
 	const row = accountId
-		? (db.prepare("select id from accounts where id = ?").get(accountId) as
-				| { id: string }
+		? (db
+				.prepare(
+					"select id, handle, external_user_id, is_default as isDefault from accounts where id = ?",
+				)
+				.get(accountId) as
+				| ({ id: string; handle: string; external_user_id: string | null } & {
+						isDefault: number;
+				  })
 				| undefined)
 		: (db
 				.prepare(
 					`
-          select id
+          select id, handle, external_user_id, is_default as isDefault
           from accounts
           order by is_default desc, created_at asc
           limit 1
           `,
 				)
-				.get() as { id: string } | undefined);
+				.get() as
+				| ({ id: string; handle: string; external_user_id: string | null } & {
+						isDefault: number;
+				  })
+				| undefined);
 
 	if (!row) {
 		throw new Error(`Unknown account: ${accountId ?? "default"}`);
 	}
 
-	return row.id;
+	return {
+		accountId: row.id,
+		isDefault: row.isDefault === 1,
+		username: row.handle.replace(/^@/, ""),
+		externalUserId:
+			typeof row.external_user_id === "string" &&
+			row.external_user_id.trim().length > 0
+				? row.external_user_id.trim()
+				: undefined,
+	};
 }
 
 function replaceTweetFts(db: Database, tweetId: string, text: string) {
@@ -57,6 +186,7 @@ function mergeHomeTimelineIntoLocalStore(
 	db: Database,
 	accountId: string,
 	payload: XurlMentionsResponse,
+	source: "bird" | "xurl",
 ) {
 	const usersById = new Map(
 		(payload.includes?.users ?? []).map((user) => [user.id, user]),
@@ -67,13 +197,15 @@ function mergeHomeTimelineIntoLocalStore(
       id, account_id, author_profile_id, kind, text, created_at,
       is_replied, reply_to_id, like_count, media_count, bookmarked, liked,
       entities_json, media_json, quoted_tweet_id
-    ) values (?, ?, ?, 'home', ?, ?, 0, null, ?, ?, 0, 0, ?, ?, null)
+    ) values (?, ?, ?, 'home', ?, ?, ?, ?, ?, ?, 0, 0, ?, ?, ?)
     on conflict(id) do update set
       account_id = tweets.account_id,
       author_profile_id = excluded.author_profile_id,
       kind = tweets.kind,
       text = excluded.text,
       created_at = excluded.created_at,
+      is_replied = max(tweets.is_replied, excluded.is_replied),
+      reply_to_id = coalesce(tweets.reply_to_id, excluded.reply_to_id),
       like_count = excluded.like_count,
       media_count = max(tweets.media_count, excluded.media_count),
       entities_json = excluded.entities_json,
@@ -82,7 +214,8 @@ function mergeHomeTimelineIntoLocalStore(
         else tweets.media_json
       end,
       bookmarked = tweets.bookmarked,
-      liked = tweets.liked
+      liked = tweets.liked,
+      quoted_tweet_id = coalesce(tweets.quoted_tweet_id, excluded.quoted_tweet_id)
     `,
 	);
 
@@ -99,22 +232,27 @@ function mergeHomeTimelineIntoLocalStore(
 			const profile = usersById.has(tweet.author_id)
 				? upsertProfileFromXUser(db, author)
 				: ensureStubProfileForXUser(db, tweet.author_id);
+			const replyToId = getReferencedTweetId(tweet, "replied_to") ?? null;
+			const quotedTweetId = getReferencedTweetId(tweet, "quoted") ?? null;
 			upsertTweet.run(
 				tweet.id,
 				accountId,
 				profile.profile.id,
 				tweet.text,
 				tweet.created_at,
+				0,
+				replyToId,
 				Number(tweet.public_metrics?.like_count ?? 0),
 				countTweetMedia(tweet),
 				JSON.stringify(tweet.entities ?? {}),
 				buildMediaJsonFromIncludes(tweet, payload.includes?.media),
+				quotedTweetId,
 			);
 			upsertTweetAccountEdge(db, {
 				accountId,
 				tweetId: tweet.id,
 				kind: "home",
-				source: "bird",
+				source,
 				seenAt,
 				rawJson: JSON.stringify(tweet),
 			});
@@ -123,55 +261,191 @@ function mergeHomeTimelineIntoLocalStore(
 	})();
 }
 
-export async function syncHomeTimeline({
+export function syncHomeTimelineEffect({
 	account,
-	limit = 100,
+	mode,
+	limit,
+	maxPages,
+	startTime,
 	following = true,
 	refresh = false,
 	cacheTtlMs,
-}: {
-	account?: string;
-	limit?: number;
-	following?: boolean;
-	refresh?: boolean;
-	cacheTtlMs?: number;
-}) {
-	assertLimit(limit);
-	const db = getNativeDb();
-	const accountId = resolveAccount(db, account);
-	const cacheKey = `timeline:bird:${accountId}:${following ? "following" : "for-you"}:${String(limit)}`;
-	const ttlMs = parseCacheTtlMs(cacheTtlMs);
-	const cached = readSyncCache<XurlMentionsResponse>(cacheKey, db);
-	const cacheAgeMs = cached
-		? Date.now() - new Date(cached.updatedAt).getTime()
-		: Number.POSITIVE_INFINITY;
-
-	if (!refresh && cached && cacheAgeMs <= ttlMs) {
+	timeoutMs,
+	onProgress,
+}: SyncHomeTimelineOptions = {}): Effect.Effect<
+	{
+		ok: true;
+		source: "bird" | "xurl" | "cache";
+		kind: "timeline";
+		accountId: string;
+		feed: "following" | "for-you";
+		count: number;
+		payload: XurlMentionsResponse;
+	},
+	unknown
+> {
+	return Effect.gen(function* () {
+		const parsedStartTime = yield* Effect.try({
+			try: () => parseStartTime(startTime),
+			catch: (error) => error,
+		});
+		const parsedMode = parseMode(mode);
+		const finiteFallbackLimit = limit ?? (parsedStartTime ? 300 : 100);
+		const effectiveLimit =
+			limit ??
+			(parsedStartTime && (parsedMode === "xurl" || parsedMode === "auto")
+				? Infinity
+				: finiteFallbackLimit);
+		assertLimit(effectiveLimit);
+		const parsedMaxPages =
+			maxPages === undefined && parsedStartTime
+				? Infinity
+				: parseMaxPages(maxPages);
+		const db = getNativeDb();
+		const resolvedAccount = resolveAccount(db, account);
+		const accountId = resolvedAccount.accountId;
+		const effectiveMode =
+			parsedMode === "auto" &&
+			account !== undefined &&
+			!resolvedAccount.isDefault
+				? "xurl"
+				: parsedMode;
+		const cacheKey = `timeline:${effectiveMode}:${accountId}:${following ? "following" : "for-you"}:${Number.isFinite(effectiveLimit) ? String(effectiveLimit) : "all"}:${Number.isFinite(parsedMaxPages) ? String(parsedMaxPages) : "all-pages"}:${parsedStartTime?.iso ?? "no-start"}`;
+		const ttlMs = parseCacheTtlMs(cacheTtlMs);
+		const cached = readSyncCache<XurlMentionsResponse>(cacheKey, db);
+		const cacheAgeMs = cached
+			? Date.now() - new Date(cached.updatedAt).getTime()
+			: Number.POSITIVE_INFINITY;
+
+		if (!refresh && cached && cacheAgeMs <= ttlMs) {
+			yield* Effect.sync(() =>
+				onProgress?.({
+					source: "cache",
+					fetched: cached.value.data.length,
+					total: Number.isFinite(effectiveLimit) ? effectiveLimit : undefined,
+					done: true,
+				}),
+			);
+			return {
+				ok: true,
+				source: "cache",
+				kind: "timeline",
+				accountId,
+				feed: following ? "following" : "for-you",
+				count: cached.value.data.length,
+				payload: cached.value,
+			} as const;
+		}
+
+		const fetchViaXurl = Effect.gen(function* () {
+			if (!following) {
+				return yield* Effect.fail(
+					new Error("xurl home timeline mode does not support --for-you"),
+				);
+			}
+			const pages: XurlMentionsResponse[] = [];
+			let nextToken: string | undefined;
+			for (let page = 0; page < parsedMaxPages; page += 1) {
+				const fetchedCount = pages.reduce(
+					(sum, item) => sum + item.data.length,
+					0,
+				);
+				const remaining = Number.isFinite(effectiveLimit)
+					? Math.max(1, effectiveLimit - fetchedCount)
+					: Infinity;
+				const pageSize = Math.min(
+					MAX_XURL_TIMELINE_PAGE_SIZE,
+					Math.max(5, remaining),
+				);
+				const pagePayload = yield* listHomeTimelineViaXurlEffect({
+					maxResults: pageSize,
+					userId: resolvedAccount.externalUserId,
+					username: resolvedAccount.username,
+					paginationToken: nextToken,
+					timeoutMs,
+				});
+				pages.push(pagePayload);
+				nextToken =
+					typeof pagePayload.meta?.next_token === "string"
+						? pagePayload.meta.next_token
+						: undefined;
+				const totalFetched = fetchedCount + pagePayload.data.length;
+				const done =
+					!nextToken ||
+					(Number.isFinite(parsedMaxPages) && page + 1 >= parsedMaxPages) ||
+					(Number.isFinite(effectiveLimit) && totalFetched >= effectiveLimit) ||
+					reachedStartTimeBoundary(pagePayload, parsedStartTime?.time);
+				yield* Effect.sync(() =>
+					onProgress?.({
+						source: "xurl",
+						fetched: totalFetched,
+						total: Number.isFinite(effectiveLimit) ? effectiveLimit : undefined,
+						page: page + 1,
+						maxPages: Number.isFinite(parsedMaxPages)
+							? parsedMaxPages
+							: undefined,
+						pageSize,
+						done,
+					}),
+				);
+				if (done) {
+					break;
+				}
+			}
+			return mergeTimelinePayloads(pages, effectiveLimit);
+		});
+		const fetchViaBird = listHomeTimelineViaBirdEffect({
+			maxResults: finiteFallbackLimit,
+			following,
+		});
+		let source: "bird" | "xurl";
+		let payload: XurlMentionsResponse;
+		if (effectiveMode === "xurl") {
+			payload = yield* fetchViaXurl;
+			source = "xurl";
+		} else if (effectiveMode === "auto") {
+			const fetched = yield* fetchViaXurl.pipe(
+				Effect.map((value) => ({ source: "xurl" as const, value })),
+				Effect.catchAll(() =>
+					fetchViaBird.pipe(
+						Effect.map((value) => ({ source: "bird" as const, value })),
+					),
+				),
+			);
+			payload = fetched.value;
+			source = fetched.source;
+		} else {
+			payload = yield* listHomeTimelineViaBirdEffect({
+				maxResults: finiteFallbackLimit,
+				following,
+			});
+			source = "bird";
+		}
+		if (source === "bird") {
+			yield* Effect.sync(() =>
+				onProgress?.({
+					source: "bird",
+					fetched: payload.data.length,
+					total: finiteFallbackLimit,
+					done: true,
+				}),
+			);
+		}
+		mergeHomeTimelineIntoLocalStore(db, accountId, payload, source);
+		writeSyncCache(cacheKey, payload, db);
+
 		return {
 			ok: true,
-			source: "cache",
+			source,
 			kind: "timeline",
 			accountId,
 			feed: following ? "following" : "for-you",
-			count: cached.value.data.length,
-			payload: cached.value,
-		};
-	}
-
-	const payload = await listHomeTimelineViaBird({
-		maxResults: limit,
-		following,
+			count: payload.data.length,
+			payload,
+		} as const;
 	});
-	mergeHomeTimelineIntoLocalStore(db, accountId, payload);
-	writeSyncCache(cacheKey, payload, db);
+}
 
-	return {
-		ok: true,
-		source: "bird",
-		kind: "timeline",
-		accountId,
-		feed: following ? "following" : "for-you",
-		count: payload.data.length,
-		payload,
-	};
+export function syncHomeTimeline(options: SyncHomeTimelineOptions = {}) {
+	return runEffectPromise(syncHomeTimelineEffect(options));
 }

package/src/lib/tweet-account-edges.ts

@@ -4,6 +4,8 @@ export type TweetAccountEdgeKind =
 	| "home"
 	| "mention"
 	| "authored"
+	| "search"
+	| "profile"
 	| "thread_context";
 
 export function upsertTweetAccountEdge(

package/src/lib/tweet-lookup.ts

@@ -1,6 +1,8 @@
-import { lookupTweetsByIdsViaBird } from "./bird";
+import { Effect } from "effect";
+import { lookupTweetsByIdsViaBirdEffect } from "./bird";
+import { runEffectPromise } from "./effect-runtime";
 import type { XurlTweetsResponse } from "./types";
-import { lookupTweetsByIds as lookupTweetsByIdsViaXurl } from "./xurl";
+import { lookupTweetsByIdsEffect as lookupTweetsByIdsViaXurlEffect } from "./xurl";
 
 export type TweetLookupMode = "auto" | "xurl" | "bird";
 
@@ -8,28 +10,37 @@ function errorMessage(error: unknown) {
 	return error instanceof Error ? error.message : String(error);
 }
 
-export async function lookupTweetsByIds(
+export function lookupTweetsByIdsEffect(
 	ids: string[],
 	mode: TweetLookupMode = "auto",
-): Promise<XurlTweetsResponse> {
+): Effect.Effect<XurlTweetsResponse, unknown> {
 	if (mode === "bird") {
-		return lookupTweetsByIdsViaBird(ids);
+		return lookupTweetsByIdsViaBirdEffect(ids);
 	}
 	if (mode === "xurl") {
-		return lookupTweetsByIdsViaXurl(ids);
+		return lookupTweetsByIdsViaXurlEffect(ids);
 	}
 
-	try {
-		return await lookupTweetsByIdsViaXurl(ids);
-	} catch (xurlError) {
-		try {
-			return await lookupTweetsByIdsViaBird(ids);
-		} catch (birdError) {
-			throw new Error(
-				`Tweet lookup failed via xurl and bird: xurl: ${errorMessage(
-					xurlError,
-				)}; bird: ${errorMessage(birdError)}`,
-			);
-		}
-	}
+	return lookupTweetsByIdsViaXurlEffect(ids).pipe(
+		Effect.catchAll((xurlError) =>
+			lookupTweetsByIdsViaBirdEffect(ids).pipe(
+				Effect.catchAll((birdError) =>
+					Effect.fail(
+						new Error(
+							`Tweet lookup failed via xurl and bird: xurl: ${errorMessage(
+								xurlError,
+							)}; bird: ${errorMessage(birdError)}`,
+						),
+					),
+				),
+			),
+		),
+	);
+}
+
+export function lookupTweetsByIds(
+	ids: string[],
+	mode: TweetLookupMode = "auto",
+): Promise<XurlTweetsResponse> {
+	return runEffectPromise(lookupTweetsByIdsEffect(ids, mode));
 }