add methods to fetch chapters

Author: Essential Randomness

src/page-loaders.ts

@@ -77,3 +77,16 @@ export const loadUserProfilePage = async ({
     (await axios.get<string>(getUserProfileUrl({ username }))).data
   ) as UserProfile;
 };
+
+export interface ChapterIndexPage extends CheerioAPI {
+  kind: "ChapterIndexPage";
+}
+export const loadChaptersIndexPage = async ({ workId }: { workId: string }) => {
+  return load(
+    (
+      await axios.get<string>(
+        `https://archiveofourown.org/works/${workId}/navigate`
+      )
+    ).data
+  ) as ChapterIndexPage;
+};

src/works/chapter-getters.ts

@@ -0,0 +1,56 @@
+import { Author, Chapter } from "types/entities";
+import { getWorkDetailsFromUrl, getWorkUrl } from "src/urls";
+
+import { ChapterIndexPage } from "src/page-loaders";
+
+const TITLE_SEPARATOR = ". ";
+export const getChaptersList = ($chapterIndexPage: ChapterIndexPage) => {
+  const chapters: Chapter[] = [];
+  $chapterIndexPage("ol.index > li").each((index, li) => {
+    const link = $chapterIndexPage(li).find("a")[0];
+    const chapterText = $chapterIndexPage(link).text();
+    const { workId, chapterId } = getWorkDetailsFromUrl({
+      url: link.attribs.href,
+    });
+    const title = chapterText.substring(
+      chapterText.indexOf(TITLE_SEPARATOR) + TITLE_SEPARATOR.length
+    );
+    const dateNode = $chapterIndexPage(
+      $chapterIndexPage(li).find(".datetime")[0]
+    );
+
+    chapters.push({
+      id: chapterId!,
+      workId,
+      index: index + 1,
+      title,
+      // Remove parenthesis from the date
+      publishedAt: dateNode.text().replace(/[\(\)]/g, ""),
+      // We rebuild the url so it gets the full path
+      url: getWorkUrl({ workId, chapterId }),
+    });
+  });
+  return chapters;
+};
+
+export const getWorkTitle = ($chapterIndexPage: ChapterIndexPage) => {
+  return $chapterIndexPage(".works-navigate h2 a[href^='/works/']").text();
+};
+
+export const getWorkAuthors = ($chapterIndexPage: ChapterIndexPage) => {
+  const authors: Author[] = [];
+  const authorNode = $chapterIndexPage(".works-navigate h2 a[rel='author']");
+  if (authorNode.text().trim() === "Anonymous") {
+    return "Anonymous";
+  }
+
+  if (authorNode.length !== 0) {
+    authorNode.each((i, element) => {
+      const url = element.attribs.href;
+      const [, username, pseud] = url.match(/users\/(.+)\/pseuds\/(.+)/)!;
+
+      authors.push({ username: username, pseud: decodeURI(pseud) });
+    });
+  }
+  return authors;
+};

src/works/index.ts

@@ -1,4 +1,14 @@
-import { LockedWorkSummary, WorkSummary } from "types/entities";
+import {
+  Author,
+  Chapter,
+  LockedWorkSummary,
+  WorkSummary,
+} from "types/entities";
+import {
+  getChaptersList,
+  getWorkAuthors as getWorkAuthorsFromChaptersIndex,
+  getWorkTitle as getWorkTitleFromChaptersIndex,
+} from "./chapter-getters";
 import {
   getWorkAdditionalTags,
   getWorkAuthors,
@@ -21,9 +31,8 @@ import {
   getWorkUpdateDate,
   getWorkWarnings,
   getWorkWordCount,
-} from "./getters";
-
-import { loadWorkPage } from "../page-loaders";
+} from "./work-getters";
+import { loadChaptersIndexPage, loadWorkPage } from "../page-loaders";
 
 export const getWork = async ({
   workId,
@@ -64,7 +73,6 @@ export const getWork = async ({
       total: totalChapters,
     },
     complete: totalChapters !== null && totalChapters === publishedChapters,
-
     summary: getWorkSummary(workPage),
     stats: {
       bookmarks: getWorkBookmarkCount(workPage),
@@ -75,3 +83,23 @@ export const getWork = async ({
     locked: false,
   };
 };
+
+export const getWorkWithChapters = async ({
+  workId,
+}: {
+  workId: string;
+}): Promise<{
+  title: string;
+  authors: "Anonymous" | Author[];
+  workId: string;
+  chapters: Chapter[];
+}> => {
+  const page = await loadChaptersIndexPage({ workId });
+
+  return {
+    title: getWorkTitleFromChaptersIndex(page),
+    authors: getWorkAuthorsFromChaptersIndex(page),
+    workId,
+    chapters: getChaptersList(page),
+  };
+};