From b2937d3c9462ac1adacee2ba6aa80e80acf9650e Mon Sep 17 00:00:00 2001
From: neobooru <50623835+neobooru@users.noreply.github.com>
Date: Wed, 5 Jul 2023 13:57:59 +0200
Subject: [PATCH] Find PNG version of posts on konachan + test fixes

---
 package.json                 |  2 +-
 src/engines/Moebooru.test.ts | 18 ++++++++++++++++++
 src/engines/Moebooru.ts      | 12 ++++++++++++
 src/engines/Shimmie2.test.ts |  2 +-
 src/test/jest.d.ts           |  1 -
 5 files changed, 32 insertions(+), 3 deletions(-)

diff --git a/package.json b/package.json
index f0e9ad6..f0d627a 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "neo-scraper",
-  "version": "0.7.0",
+  "version": "0.7.1",
   "description": "Scrape a booru-like post object from the DOM.",
   "type": "module",
   "main": "dist/index.js",
diff --git a/src/engines/Moebooru.test.ts b/src/engines/Moebooru.test.ts
index 1f2bf65..33d9ad3 100644
--- a/src/engines/Moebooru.test.ts
+++ b/src/engines/Moebooru.test.ts
@@ -76,3 +76,21 @@ test("get konachan.com post 285782 with notes", async () => {
   );
   expect(res.posts[0]).toHaveSource("https://www.pixiv.net/member_illust.php?mode=medium&illust_id=75747354");
 });
+
+test("get konachan.com post 358247 jpeg and png", async () => {
+  const pageUrl = "https://konachan.com/post/show/358247/3d-animal_ears-arknights-black_hair-blonde_hair-br";
+  const res = await scrapeUrl(pageUrl);
+  expect(res.posts.length).toBe(2);
+  expect(res.posts[0]).toHaveTag("texas_(arknights)", "character");
+  expect(res.posts[0]).toHaveResolution([6144, 3456]);
+  expect(res.posts[0].pageUrl).toBe(pageUrl);
+  expect(res.posts[0].contentUrl).toBe(
+    "https://konachan.com/image/56983a2166a13f4af963774ccd424f0e/Konachan.com%20-%20358247%203d%20arknights%20computer%20couch%20dress%20drink%20gloves%20gun%20halo%20hoodie%20necklace%20pantyhose%20red_eyes%20red_hair%20shirt%20shorts%20skirt%20tie%20twintails%20weapon%20wolfgirl.png"
+  );
+  expect(res.posts[0]).toHaveSource("https://www.pixiv.net/en/artworks/108925945");
+
+  // JPEG variant
+  expect(res.posts[1].contentUrl).toBe(
+    "https://konachan.com/jpeg/56983a2166a13f4af963774ccd424f0e/Konachan.com%20-%20358247%203d%20arknights%20computer%20couch%20dress%20drink%20gloves%20gun%20halo%20hoodie%20necklace%20pantyhose%20red_eyes%20red_hair%20shirt%20shorts%20skirt%20tie%20twintails%20weapon%20wolfgirl.jpg"
+  );
+});
diff --git a/src/engines/Moebooru.ts b/src/engines/Moebooru.ts
index e52c2a7..d337b34 100644
--- a/src/engines/Moebooru.ts
+++ b/src/engines/Moebooru.ts
@@ -110,6 +110,18 @@ export default class Moebooru extends ScrapeEngineBase {
       post.notes = createNotesFromMoebooruBoxes(document, boxSize);
     }
 
+    // Sometimes there is both a JPEG and a PNG version of a post.
+    // If this is the case, copy all the information from the JPEG
+    // version and only replace the contentUrl with the link to the PNG version.
+    // We want to add this PNG version before the other version, so that it appears on top.
+    const unchangedImgEl = document.querySelector<HTMLAnchorElement>("a.original-file-unchanged");
+    if (unchangedImgEl && unchangedImgEl.href != post.contentUrl) {
+      const copy = Object.assign(new ScrapedPost(), post);
+      copy.contentUrl = unchangedImgEl.href;
+      result.tryAddPost(copy);
+    }
+
+    // Add normal/non-png post.
     result.tryAddPost(post);
 
     return result;
diff --git a/src/engines/Shimmie2.test.ts b/src/engines/Shimmie2.test.ts
index 29b514e..5450f33 100644
--- a/src/engines/Shimmie2.test.ts
+++ b/src/engines/Shimmie2.test.ts
@@ -10,6 +10,6 @@ test("get rule34.paheal.net post 2505982", async () => {
     "http://law-zilla.tumblr.com/post/171249285285/finished-subdraw-37-kaisa-from-league-of-legends"
   );
   expect(res.posts[0].contentUrl).toContain(
-    ".paheal.net/_images/15427ab5ed2cddbeef947e30f5c9cab6/2505982%20-%20Kai%27Sa%20League_of_Legends%20lawzilla.png"
+    ".paheal.net/_images/15427ab5ed2cddbeef947e30f5c9cab6/2505982%20-%20Kai%27Sa%20lawzilla%20League_of_Legends.png"
   );
 });
diff --git a/src/test/jest.d.ts b/src/test/jest.d.ts
index 22ff885..96be272 100644
--- a/src/test/jest.d.ts
+++ b/src/test/jest.d.ts
@@ -4,7 +4,6 @@ import { TagCategory } from "../BooruTypes.js";
 declare global {
   namespace jest {
     interface Matchers<R> {
-      // The `= undefined` part is incorrect according to VSCode, but it is actually required.
       toHaveTag(tagName: string, tagCategory?: TagCategory): CustomMatcherResult;
       toHaveNote(noteText: string): CustomMatcherResult;
       toHaveResolution(resolution: [number, number]): CustomMatcherResult;