Release 1.0.2

LuckyYam · Feb 25, 2024 · c4ae9db · c4ae9db
2 parents 0eb3958 + d3febf5
commit c4ae9db
Show file tree

Hide file tree

Showing 10 changed files with 262 additions and 88 deletions.
diff --git a/README.md b/README.md
@@ -2,7 +2,7 @@
 
 [![NPM](https://img.shields.io/badge/Available%20On-NPM-lightgrey.svg?logo=npm&logoColor=339933&labelColor=white&style=flat-square)](https://www.npmjs.com/package/@shineiichijo/nhentai-ts)
 
-Scrap and build a PDF of a doujin from NHentai (only mirror sites, [check the available sites here](https://github.com/LuckyYam/nhentai-ts/blob/master/src/lib/constants.ts#L1)).
+Scrap and build a PDF of a doujin from NHentai. [Check the available sites here](https://github.com/LuckyYam/nhentai-ts/blob/master/src/lib/constants.ts#L1).
 
 [Documentation](https://luckyyam.github.io/nhentai-ts/)
 
@@ -13,10 +13,34 @@ Scrap and build a PDF of a doujin from NHentai (only mirror sites, [check the av
 yarn add @shineiichijo/nhentai-ts
 ```
 
+## Note
+
+If you're choosing `nhentai.net` for the site, make sure to follow the following steps (as the site has enabled cloudflare protection):
+
+- Open https://nhentai.net/ in your browser.
+- Open Dev Tools and set the User Agent to what you want (you'll need the user agent).
+- Reload the site and wait for the clearance of cloudflare (without closing the Dev Tools).
+- Save the cookie value from the Network Tab (`cf_clearance` value).
+
+After following all these steps, you are all set. You can also check the [example](#usage-examples) of it (at the first one). Remember, the cookie value expires after 30 minutes of inactivity. So, you might have to do the above steps again (in case you're gonna use it again).
+
 ## Usage Examples
 ```ts
 import { NHentai } from '@shineiichijo/nhentai-ts'
 
+const user_agent = 'User Agent'
+const cookie_value = 'cookie'
+const nhentai = new NHentai({ site: 'nhentai.net', user_agent, cookie_value }) //check above
+;(async () => {
+    //Explores the home page
+    const { data } = await nhentai.explore()
+    console.log(data)
+})()
+```
+
+```ts
+import { NHentai } from '@shineiichijo/nhentai-ts'
+
 const nhentai = new NHentai()
 ;(async () => {
     //searches for manga
@@ -33,16 +57,16 @@ const nhentai = new NHentai()
 ```ts
 import { NHentai } from '@shineiichijo/nhentai-ts'
 
-const nhentai = new NHentai({ site: 'nhentai.website' }) //configuring a mirror site of the class (you can check the available sites here: https://github.com/LuckyYam/nhentai-ts/blob/master/src/lib/constants.ts#L1)
+const nhentai = new NHentai({ site: 'nhentai.to' }) //configuring a mirror site of the class (you can check the available sites here: https://github.com/LuckyYam/nhentai-ts/blob/master/src/lib/constants.ts#L1)
 //validates the ID of a doujin
 nhentai.validate('172').then(console.log)
 ```
 
 ```ts
 import { NHentai } from '@shineiichijo/nhentai-ts'
 
-const nhentai = new NHentai({ site: 'https://nhentai.xxx' })
-//explores all of the available doujin
+const nhentai = new NHentai()
+//explores all of the available doujin (homepage)
 nhentai.explore(2 /* Page number of exploring the doujin */).then(console.log)
 ```
 

diff --git a/package.json b/package.json
@@ -1,7 +1,7 @@
 {
   "name": "@shineiichijo/nhentai-ts",
   "description": "A scraper for NHentai with types",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
   "license": "MIT",
@@ -38,14 +38,17 @@
     "docs": "yarn typedoc"
   },
   "dependencies": {
-    "axios": "^0.27.2",
+    "axios": "^1.6.7",
     "cheerio": "^1.0.0-rc.12",
+    "http-cookie-agent": "^6.0.1",
     "jszip": "^3.10.0",
-    "pdfkit": "^0.13.0"
+    "pdfkit": "^0.13.0",
+    "tough-cookie": "^4.1.3"
   },
   "devDependencies": {
     "@types/node": "^18.6.1",
     "@types/pdfkit": "^0.12.6",
+    "@types/tough-cookie": "^4.0.5",
     "@typescript-eslint/eslint-plugin": "^5.31.0",
     "@typescript-eslint/parser": "^5.31.0",
     "eslint": "^8.20.0",

diff --git a/src/Parser/doujin.ts b/src/Parser/doujin.ts
@@ -1,21 +1,35 @@
 import { CheerioAPI } from 'cheerio'
-import { baseURLS, imageSites, Pages } from '../lib'
+import { baseURLS, clean, getExtension, Pages, imageSites } from '../lib'
 import { TURL, IDoujinInfo } from '../Types'
 
 export const parseDoujinInfo = (
     $: CheerioAPI,
-    site: keyof typeof baseURLS
+    site: keyof typeof baseURLS,
+    api_pages?: { t: string }[]
 ): IDoujinInfo => {
     const pages: string[] = []
-    $('.thumb-container').each((i, el) => {
-        const url = $(el).find('a > img').attr('data-src')
-        if (url)
+    const gallery_id = (
+        $('.thumb-container').first().find('a > img').attr('data-src') ||
+        '/galleries/'
+    ).split('/galleries/')[0]
+    if (site === 'net' && api_pages)
+        api_pages.forEach((page, i) =>
             pages.push(
-                url
-                    .replace(`${i + 1}t`, `${i + 1}`)
-                    .replace(imageSites[site], 'i.nhentai.net')
+                `https://i.nhentai.net/galleries/${gallery_id}/${
+                    i + 1
+                }.${getExtension(page.t)}`
             )
-    })
+        )
+    else
+        $('.thumb-container').each((i, el) => {
+            const url = $(el).find('a > img').attr('data-src')
+            if (url)
+                pages.push(
+                    url
+                        .replace(`${i + 1}t`, `${i + 1}`)
+                        .replace(imageSites[site], 'i.nhentai.net')
+                )
+        })
     const cover =
         $('#cover').find('a > img').attr('data-src') ||
         $('#cover').find('a > img').attr('src')
@@ -67,14 +81,16 @@ export const parseDoujinInfo = (
         id,
         title: titles.english,
         originalTitle: titles.original,
-        parodies,
-        characters,
-        tags,
-        artists,
-        groups,
-        languages,
-        categories,
-        cover: cover ? cover.replace('cdn.dogehls.xyz', 't3.nhentai.net') : null,
+        parodies: clean(parodies),
+        characters: clean(characters),
+        tags: clean(tags),
+        artists: clean(artists),
+        groups: clean(groups),
+        languages: clean(languages),
+        categories: clean(categories),
+        cover: cover
+            ? cover.replace('cdn.dogehls.xyz', 't3.nhentai.net')
+            : null,
         images,
         url
     }

diff --git a/src/Parser/list.ts b/src/Parser/list.ts
@@ -9,34 +9,26 @@ export const parseDoujinList = (
 ): IList => {
     const data: IList['data'] = []
     const baseURL = baseURLS[site]
-    let currentPage = 1
-    let totalPages = 1
-    const pageElements =
-        site === 'to'
-            ? $('body > section')
-            : site === 'xxx'
-            ? $('#content > section')
-            : $('#fnh > section')
-    if (pageElements.find('.page.current').html()) {
-        currentPage = Number(pageElements.find('.page.current').text())
-        const split = $(pageElements)
-            .last()
-            .text()
-            .split('\n')
-            .filter((el) => Number(el) >= 1)
-        totalPages = Number(split[split.length - 1])
-    }
-    const pagination = {
-        currentPage,
-        hasNextPage: totalPages > currentPage,
-        totalPages
-    }
+    const currentPage = Number($('.pagination').find('.page.current').text())
+    const totalPages = Number(
+        ($('.pagination').find('a.last').attr('href') || '').split('page=')[1]
+    )
+    const pagination =
+        currentPage === 0
+            ? null
+            : {
+                  currentPage,
+                  hasNextPage: totalPages > currentPage,
+                  totalPages
+              }
     $('.gallery').each((i, el) => {
         const contentElements = $(el).find('a')
         const slug = contentElements.attr('href')
         const id = slug ? slug.split('g/')[1].replace('/', '') : ''
         const url = `${baseURL}/g/${id}` as TURL
-        const coverSlug = contentElements.find('img').attr('data-src')
+        const coverSlug =
+            contentElements.find('a > img').attr('data-src') ||
+            contentElements.find('a > img').attr('src')
         const cover = coverSlug
             ? `${
                   coverSlug.startsWith('/galleries/')

diff --git a/src/Types/list.ts b/src/Types/list.ts
@@ -3,7 +3,7 @@ import { List } from '../lib/Classes/List'
 
 export interface IList {
     /** Pagination of the doujin list */
-    pagination: IPagination
+    pagination: IPagination | null
     /** Data of the list */
     data: List[]
 }

diff --git a/src/lib/Classes/NHentai.ts b/src/lib/Classes/NHentai.ts
@@ -1,43 +1,75 @@
-import axios from 'axios'
+import axios, { AxiosInstance } from 'axios'
 import { load } from 'cheerio'
+import { CookieJar } from 'tough-cookie'
+import { HttpsCookieAgent } from 'http-cookie-agent/http'
 import { parseDoujinList, parseDoujinInfo } from '../../Parser'
 import { sites } from '../constants'
 import { IDoujinInfo, TSite, IList } from '../../Types'
+import { getAPIGalleryPages } from '../util'
 
 export class NHentai {
+    #axios: AxiosInstance
     /**
      * Constructs an instance of the NHentai class
-     * @param options Options of the NHentai class
+     * @param _options Options of the NHentai class
      */
     constructor(
-        private options: { site: TSite | `https://${TSite}` } = {
+        private readonly _options: {
+            site: TSite | `https://${TSite}`
+            user_agent?: string
+            cookie_value?: string
+        } = {
             site: 'https://nhentai.to'
         }
     ) {
+        this.#axios = axios
         if (
             !sites.includes(
-                this.options.site
+                this._options.site
                     .replace('https:', '')
                     .replace(/\//g, '') as TSite
             )
         )
-            this.options.site = 'https://nhentai.website'
-        if (!this.options.site.startsWith('https://'))
-            this.options.site =
-                `https://${this.options.site}` as `https://${TSite}`
+            this._options.site = 'https://nhentai.to'
+        if (!this._options.site.startsWith('https://'))
+            this._options.site =
+                `https://${this._options.site}` as `https://${TSite}`
+        if (
+            this._options.site.includes('nhentai.net') &&
+            (!this._options.cookie_value || !this._options.user_agent)
+        )
+            throw new Error(
+                `Assign the ${
+                    !this._options.cookie_value
+                        ? "'cookie_value'"
+                        : "'user_agent'"
+                } in the instance of the class to use this site.`
+            )
+        if (this._options.cookie_value) {
+            const jar = new CookieJar()
+            jar.setCookie(this._options.cookie_value, this._options.site)
+            const httpsAgent = new HttpsCookieAgent({ cookies: { jar } })
+            this.#axios = axios.create({ httpsAgent })
+        }
+        if (this._options.user_agent)
+            this.#axios.defaults.headers.common['User-Agent'] =
+                this._options.user_agent
     }
 
     /**
      * Gets a random doujin
      * @returns Info of the random doujin
      */
     public getRandom = async (): Promise<IDoujinInfo> =>
-        await axios
-            .get<string>(`${this.options.site}/random`)
-            .then(({ data }) =>
+        await this.#axios
+            .get<string>(`${this._options.site}/random`)
+            .then(async ({ data }) =>
                 parseDoujinInfo(
                     load(data),
-                    this.options.site.split('nhentai.')[1] as 'to'
+                    this._options.site.split('nhentai.')[1] as 'to',
+                    this._options.site.includes('net')
+                        ? await getAPIGalleryPages(this.#axios, data)
+                        : undefined
                 )
             )
             .catch((err) => {
@@ -51,12 +83,12 @@ export class NHentai {
      */
     public explore = async (page: number = 1): Promise<IList> => {
         if (isNaN(page) || page < 1) page = 1
-        return await axios
-            .get<string>(`${this.options.site}?page=${page}`)
+        return await this.#axios
+            .get<string>(`${this._options.site}?page=${page}`)
             .then(({ data }) =>
                 parseDoujinList(
                     load(data),
-                    this.options.site.split('nhentai.')[1] as 'to'
+                    this._options.site.split('nhentai.')[1] as 'to'
                 )
             )
             .catch((err) => {
@@ -78,12 +110,12 @@ export class NHentai {
             throw new Error("The 'query' parameter shouldn't be undefined")
         let page = 1
         if (options?.page && options.page > 0) page = options.page
-        return await axios
-            .get<string>(`${this.options.site}/search?q=${query}&page=${page}`)
+        return await this.#axios
+            .get<string>(`${this._options.site}/search?q=${query}&page=${page}`)
             .then((res) => {
                 const results = parseDoujinList(
                     load(res.data),
-                    this.options.site.split('nhentai.')[1] as 'to'
+                    this._options.site.split('nhentai.')[1] as 'to'
                 )
                 if (!results.data.length)
                     throw new Error('No doujin results found')
@@ -100,12 +132,15 @@ export class NHentai {
         if (!id) throw new Error("The 'id' parameter shouldn't be undefined")
         const valid = await this.validate(id)
         if (!valid) throw new Error('Invalid doujin ID')
-        return await axios
-            .get(`${this.options.site}/g/${id}`)
-            .then((res) =>
+        return await this.#axios
+            .get(`${this._options.site}/g/${id}`)
+            .then(async (res) =>
                 parseDoujinInfo(
                     load(res.data),
-                    this.options.site.split('nhentai.')[1] as 'to'
+                    this._options.site.split('nhentai.')[1] as 'to',
+                    this._options.site.includes('net')
+                        ? await getAPIGalleryPages(this.#axios, res.data)
+                        : undefined
                 )
             )
             .catch((err) => {
@@ -118,8 +153,8 @@ export class NHentai {
      * @param id ID of the doujin to check
      */
     public validate = (id: string | number): Promise<boolean> =>
-        axios
-            .get(`${this.options.site}/g/${id}`)
+        this.#axios
+            .get(`${this._options.site}/g/${id}`)
             .then(() => true)
             .catch(() => false)
 }
diff --git a/src/lib/constants.ts b/src/lib/constants.ts
@@ -1,13 +1,11 @@
-export const sites = ['nhentai.to', 'nhentai.xxx', 'nhentai.website'] as const
+export const sites = ['nhentai.to', 'nhentai.net'] as const
 
 export const baseURLS = {
     to: 'https://nhentai.to',
-    xxx: 'https://nhentai.xxx',
-    website: 'https://nhentai.website'
-}
+    net: 'https://nhentai.net'
+} as const
 
 export const imageSites = {
     to: 'cdn.dogehls.xyz',
-    xxx: 'cdn.nhentai.xxx',
-    website: 'cdn.dogehls.xyz'
-}
+    net: /t[357].nhentai.net/
+} as const
diff --git a/src/lib/index.ts b/src/lib/index.ts
@@ -1,2 +1,3 @@
 export * from './constants'
 export * from './Classes'
+export * from './util'