From e7cd5510d54914238bce6d32a2cffc9c36de7d67 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Mike=20F=C3=A4hrmann?= Date: Thu, 11 Apr 2019 19:27:02 +0200 Subject: [PATCH] [pixnet] add extractors (closes #177) for: - users/blogs: http://albertayu773.pixnet.net/ - folders: https://albertayu773.pixnet.net/album/folder/1405768 - sets : https://albertayu773.pixnet.net/album/set/15078995 - photos : https://albertayu773.pixnet.net/album/photo/159443828 --- docs/supportedsites.rst | 2 + gallery_dl/extractor/__init__.py | 1 + gallery_dl/extractor/pixnet.py | 179 +++++++++++++++++++++++++++++++ 3 files changed, 182 insertions(+) create mode 100644 gallery_dl/extractor/pixnet.py diff --git a/docs/supportedsites.rst b/docs/supportedsites.rst index f6ed52db..08142555 100644 --- a/docs/supportedsites.rst +++ b/docs/supportedsites.rst @@ -73,6 +73,7 @@ Photobucket https://photobucket.com/ Albums, individual Imag Piczel https://piczel.tv/ Images from Users, Folders, individual Images Pinterest https://www.pinterest.com/ Boards, Pins, pin.it Links, related Pins Pixiv https://www.pixiv.net/ |pixiv-C| Required +Pixnet https://www.pixnet.net/ |pixnet-C| Pornreactor http://pornreactor.cc/ |pornreactor-C| PowerManga https://read.powermanga.org/ Chapters, Manga Pururin https://pururin.io/ Galleries @@ -118,6 +119,7 @@ Turboimagehost https://www.turboimagehost.com/ individual Images .. |joyreactor-C| replace:: Images from Users, Posts, Search Results, Tag-Searches .. |nijie-C| replace:: Images from Users, Doujin, Favorites, individual Images .. |pixiv-C| replace:: Images from Users, Favorites, Follows, pixiv.me Links, Rankings, Search Results, Individual Images +.. |pixnet-C| replace:: Images from Users, Folders, individual Images, Sets .. |pornreactor-C| replace:: Images from Users, Posts, Search Results, Tag-Searches .. |smugmug-C| replace:: Albums, individual Images, Images from Users and Folders .. |yuki-S| replace:: yuki.la 4chan archive diff --git a/gallery_dl/extractor/__init__.py b/gallery_dl/extractor/__init__.py index 21a0ef18..f1673df4 100644 --- a/gallery_dl/extractor/__init__.py +++ b/gallery_dl/extractor/__init__.py @@ -67,6 +67,7 @@ modules = [ "piczel", "pinterest", "pixiv", + "pixnet", "pururin", "reactor", "readcomiconline", diff --git a/gallery_dl/extractor/pixnet.py b/gallery_dl/extractor/pixnet.py new file mode 100644 index 00000000..9cada6bf --- /dev/null +++ b/gallery_dl/extractor/pixnet.py @@ -0,0 +1,179 @@ +# -*- coding: utf-8 -*- + +# Copyright 2016-2019 Mike Fährmann +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License version 2 as +# published by the Free Software Foundation. + +"""Extractors for https://www.pixnet.net/""" + +from .common import Extractor, Message +from .. import text + + +BASE_PATTERN = r"(?:https?://)?(?!www\.)([^.]+)\.pixnet.net" + + +class PixnetExtractor(Extractor): + """Base class for pixnet extractors""" + category = "pixnet" + filename_fmt = "{num:>03}_{id}.{extension}" + archive_fmt = "{id}" + url_fmt = "" + + def __init__(self, match): + Extractor.__init__(self, match) + self.blog, self.item_id = match.groups() + self.root = "https://{}.pixnet.net".format(self.blog) + + def items(self): + url = self.url_fmt.format(self.root, self.item_id) + page = self.request(url, encoding="utf-8").text + user = text.extract(page, '')[0] + if "href" not in pnext: + return + url = self.root + text.extract(pnext, 'href="', '"')[0] + page = self.request(url, encoding="utf-8").text + + +class PixnetImageExtractor(PixnetExtractor): + """Extractor for a single photo from pixnet.net""" + subcategory = "image" + filename_fmt = "{id}.{extension}" + directory_fmt = ("{category}", "{blog}") + pattern = BASE_PATTERN + r"/album/photo/(\d+)" + test = ("https://albertayu773.pixnet.net/album/photo/159443828", { + "url": "156564c422138914c9fa5b42191677b45c414af4", + "keyword": "19971bcd056dfef5593f4328a723a9602be0f087", + "content": "0e097bdf49e76dd9b9d57a016b08b16fa6a33280", + }) + + def items(self): + url = "https://api.pixnet.cc/oembed" + params = { + "url": "https://{}.pixnet.net/album/photo/{}".format( + self.blog, self.item_id), + "format": "json", + } + + data = self.request(url, params=params).json() + data["id"] = text.parse_int( + data["url"].rpartition("/")[2].partition("-")[0]) + data["filename"], _, data["extension"] = data["title"].rpartition(".") + data["blog"] = self.blog + data["user"] = data.pop("author_name") + + yield Message.Version, 1 + yield Message.Directory, data + yield Message.Url, data["url"], data + + +class PixnetSetExtractor(PixnetExtractor): + """Extractor for images from a pixnet set""" + subcategory = "set" + url_fmt = "{}/album/set/{}" + directory_fmt = ("{category}", "{blog}", + "{folder_id} {folder_title}", "{set_id} {set_title}") + pattern = BASE_PATTERN + r"/album/set/(\d+)" + test = ( + ("https://albertayu773.pixnet.net/album/set/15078995", { + "url": "6535712801af47af51110542f4938a7cef44557f", + "keyword": "bf25d59e5b0959cb1f53e7fd2e2a25f2f67e5925", + }), + ("https://anrine910070.pixnet.net/album/set/5917493", { + "url": "b3eb6431aea0bcf5003432a4a0f3a3232084fc13", + "keyword": "bf7004faa1cea18cf9bd856f0955a69be51b1ec6", + }), + ) + + def items(self): + url = self.url_fmt.format(self.root, self.item_id) + page = self.request(url, encoding="utf-8").text + data = self.metadata(page) + + yield Message.Version, 1 + yield Message.Directory, data + for num, info in enumerate(self._pagination(page), 1): + url, pos = text.extract(info, ' href="', '"') + src, pos = text.extract(info, ' src="', '"', pos) + alt, pos = text.extract(info, ' alt="', '"', pos) + + photo = { + "id": text.parse_int(url.rpartition("/")[2].partition("#")[0]), + "url": src.replace("_s.", "."), + "num": num, + "filename": alt, + "extension": src.rpartition(".")[2], + } + photo.update(data) + yield Message.Url, photo["url"], photo + + def metadata(self, page): + user , pos = text.extract(page, '', '<', pos) + sid , pos = text.extract(page, '/set/', '"', pos) + sname, pos = text.extract(page, '>', '<', pos) + return { + "blog": self.blog, + "user": user.rpartition(" (")[0], + "folder_id" : text.parse_int(fid, ""), + "folder_title": text.unescape(fname).strip(), + "set_id" : text.parse_int(sid), + "set_title" : text.unescape(sname), + } + + +class PixnetFolderExtractor(PixnetExtractor): + """Extractor for all sets in a pixnet folder""" + subcategory = "folder" + url_fmt = "{}/album/folder/{}" + pattern = BASE_PATTERN + r"/album/folder/(\d+)" + test = ("https://albertayu773.pixnet.net/album/folder/1405768", { + "pattern": PixnetSetExtractor.pattern, + "count": ">= 15", + }) + + +class PixnetUserExtractor(PixnetExtractor): + """Extractor for all sets and folders of a pixnet user""" + subcategory = "user" + url_fmt = "{}{}/album/list" + pattern = BASE_PATTERN + r"()(?:/blog|/album(?:/list)?)?/?(?:$|[?&#])" + test = ( + ("https://albertayu773.pixnet.net/"), + ("https://albertayu773.pixnet.net/blog"), + ("https://albertayu773.pixnet.net/album"), + ("https://albertayu773.pixnet.net/album/list", { + "pattern": PixnetFolderExtractor.pattern, + "count": ">= 30", + }), + ("https://anrine910070.pixnet.net/album/list", { + "pattern": PixnetSetExtractor.pattern, + "count": ">= 14", + }), + )