Something went wrong. Try again.
a tool for shared writing and social publishing
Something went wrong. Try again.
TypeScript
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156// Reading chapters out of post titles.//// A publication that posts one page per document — a comic, a serialised story// — names its pages in a run: `chapter 1 - 001`, `chapter 1 - 002`, `FITV #2 |// Pg. 7`. Its archive then reads as dozens of near-identical rows when what a// reader wants to browse is chapters. Chapter view collapses those pages back// into the chapters they came from and shows the art instead.//// Two things make that readable without a vocabulary of chapter words, which// there is no writing — `#`, `chapter`, `Ch.`, `Episode`, `Capítulo` are all// the same idea. Titles in a run name the chapter first and the page second,// divided by one of a handful of characters people reach for. And posts go up// in reading order, so a chapter is always a *run* of consecutive posts and// only neighbours need comparing.
import { AtUri } from "@atproto/api";import { getDocumentURL } from "./getPublicationURL";import { blobRefToSrc, COVER_THUMBNAIL_WIDTH } from "./blobRefToSrc";import type { NormalizedDocument } from "./normalizeRecords";
// The characters a title uses to divide its chapter from its page. Dash// variants count as the dash people meant to type.const CHAPTER_SEPARATOR = /[:,\-–—/|]/;
/** * The chapter a title names, or null when it names none. */function chapterOf(title: string): string | null { const match = title.match(CHAPTER_SEPARATOR); if (!match || match.index === undefined) return null; return title.slice(0, match.index).trim() || null;}
// ignore casingfunction sameChapter(a: string | null, b: string | null) { return a !== null && b !== null && a.toLowerCase() === b.toLowerCase();}
export type ChapterListItem<T> = { key: string; label: string; posts: T[];};
type ChapterablePost = { uri: string; record: { title?: string; publishedAt?: string };};
/** * Collapse the posts whose titles follow the archive's own chapter convention * into one item per chapter, leaving every other post an item of its own. * * Items come back oldest-first */export function groupPostsIntoChapters<T extends ChapterablePost>( posts: T[],): ChapterListItem<T>[] { // The convention only shows up as a run of neighbouring titles, so the // archive has to be walked the way it was written. const inOrder = [...posts].sort((a, b) => { const ad = a.record.publishedAt ? new Date(a.record.publishedAt).getTime() : 0; const bd = b.record.publishedAt ? new Date(b.record.publishedAt).getTime() : 0; if (ad !== bd) return ad - bd; return a.uri < b.uri ? -1 : 1; }); const titleOf = (post: T) => (post.record.title || "").trim(); const chapters = inOrder.map((post) => chapterOf(titleOf(post)));
const items: ChapterListItem<T>[] = []; let i = 0; while (i < inOrder.length) { // A chapter needs a second page to be one: a lone title that happens to // carry a separator is still just a post. if (!sameChapter(chapters[i], chapters[i + 1] ?? null)) { items.push({ key: inOrder[i].uri, label: titleOf(inOrder[i]), posts: [inOrder[i]], }); i++; continue; }
let last = i + 1; while ( last + 1 < inOrder.length && sameChapter(chapters[i], chapters[last + 1]) ) last++;
const label = chapters[i] as string; items.push({ // Two runs could carry the same name (a chapter posted in two sittings), // so where the run started is part of its identity. key: `${label.toLowerCase()}|${inOrder[i].uri}`, label, posts: inOrder.slice(i, last + 1), }); i = last + 1; }
return items;}
// Everything a chapter's card on the shelf renders — plain strings, so a// server render can group the whole archive and ship one small card per// chapter instead of the posts behind it.export type ChapterCard = { key: string; label: string; href: string; pageCount: number; coverImageSrc?: string; membersOnly: boolean;};
type CardablePost = { uri: string; membersOnly?: boolean; record: NormalizedDocument;};
/** * Group posts into chapters and reduce each to its card: label, the URL of its * first page in reading order, that page's cover, and whether any page is * members-only. Runs wherever the full archive already lives (the SSR page * query, the editor's in-memory documents) so the shelf never needs the posts * themselves shipped to it. */export function buildChapterCards( posts: CardablePost[], publication: { uri: string; record: unknown },): ChapterCard[] { return groupPostsIntoChapters(posts).map((item) => { const first = item.posts[0]; const coverImage = first.record.coverImage; return { key: item.key, label: item.label, href: getDocumentURL(first.record, first.uri, publication), pageCount: item.posts.length, coverImageSrc: coverImage ? blobRefToSrc(coverImage.ref, new AtUri(first.uri).host, undefined, { width: COVER_THUMBNAIL_WIDTH.medium, }) : undefined, membersOnly: item.posts.some((post) => !!post.membersOnly), }; });}