update extractor test results

This commit is contained in:
Mike Fährmann 2021-07-04 19:46:08 +02:00
parent 7f591c78cb
commit 21c2da454f
No known key found for this signature in database
GPG Key ID: 5680CA389D365A88
8 changed files with 43 additions and 38 deletions

View File

@ -140,7 +140,7 @@ class _500pxGalleryExtractor(_500pxExtractor):
}), }),
# unavailable photos (#1335) # unavailable photos (#1335)
("https://500px.com/p/Light_Expression_Photography/galleries/street", { ("https://500px.com/p/Light_Expression_Photography/galleries/street", {
"count": 0, "count": 4,
}), }),
("https://500px.com/fashvamp/galleries/lera"), ("https://500px.com/fashvamp/galleries/lera"),
) )

View File

@ -173,9 +173,8 @@ class BcyPostExtractor(BcyExtractor):
("https://bcy.net/item/detail/6950136331708144648", { ("https://bcy.net/item/detail/6950136331708144648", {
"pattern": r"https://p\d-bcy.byteimg.com/img/banciyuan/[0-9a-f]+" "pattern": r"https://p\d-bcy.byteimg.com/img/banciyuan/[0-9a-f]+"
r"~tplv-banciyuan-logo-v3:.+\.image", r"~tplv-banciyuan-logo-v3:.+\.image",
"count": 10, "count": 8,
"keyword": {"filter": "watermark"} "keyword": {"filter": "watermark"},
}), }),
# deleted # deleted
("https://bcy.net/item/detail/6780546160802143236", { ("https://bcy.net/item/detail/6780546160802143236", {

View File

@ -137,7 +137,7 @@ class FoolfuukaThreadExtractor(FoolfuukaExtractor):
"url": "61896d9d9a2edb556b619000a308a984307b6d30", "url": "61896d9d9a2edb556b619000a308a984307b6d30",
}), }),
("https://thebarchive.com/b/thread/739772332/", { ("https://thebarchive.com/b/thread/739772332/", {
"url": "07d39d2cb48f40fb337dc992993d965b0cd5f7cd", "url": "e8b18001307d130d67db31740ce57c8561b5d80c",
}), }),
("https://archive.wakarimasen.moe/a/thread/223157648/", { ("https://archive.wakarimasen.moe/a/thread/223157648/", {
"url": "fef0758d2eb81b1ba783051fd5ec491d70107a78", "url": "fef0758d2eb81b1ba783051fd5ec491d70107a78",

View File

@ -1,6 +1,6 @@
# -*- coding: utf-8 -*- # -*- coding: utf-8 -*-
# Copyright 2020 Mike Fährmann # Copyright 2020-2021 Mike Fährmann
# #
# This program is free software; you can redistribute it and/or modify # 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 # it under the terms of the GNU General Public License version 2 as
@ -19,21 +19,23 @@ class HentaihandGalleryExtractor(GalleryExtractor):
root = "https://hentaihand.com" root = "https://hentaihand.com"
pattern = r"(?:https?://)?(?:www\.)?hentaihand\.com/\w+/comic/([\w-]+)" pattern = r"(?:https?://)?(?:www\.)?hentaihand\.com/\w+/comic/([\w-]+)"
test = ( test = (
(("https://hentaihand.com/en/comic/kouda-tomohiro-chiyomi-" (("https://hentaihand.com/en/comic/c75-takumi-na-muchi-choudenji-hou-"
"blizzard-comic-aun-2016-12-english-nanda-sore-scans"), { "no-aishi-kata-how-to-love-a-super-electromagnetic-gun-toaru-kagaku-"
"pattern": r"https://cdn.hentaihand.com/.*/images/304546/\d+.jpg$", "no-railgun-english"), {
"count": 19, "pattern": r"https://cdn.hentaihand.com/.*/images/5669/\d+.jpg$",
"count": 50,
"keyword": { "keyword": {
"artists" : ["Kouda Tomohiro"], "artists" : ["Takumi Na Muchi"],
"date" : "dt:2020-02-06 00:00:00", "date" : "dt:2014-06-28 00:00:00",
"gallery_id": 304546, "gallery_id": 5669,
"lang" : "en", "lang" : "en",
"language" : "English", "language" : "English",
"relationships": ["Family", "Step family"], "parodies" : ["Toaru Kagaku No Railgun"],
"relationships": list,
"tags" : list, "tags" : list,
"title" : r"re:\[Kouda Tomohiro\] Chiyomi Blizzard", "title" : r"re:\(C75\) \[Takumi na Muchi\] Choudenji Hou ",
"title_alt" : r"re:\[幸田朋弘\] ちよみブリザード", "title_alt" : r"re:\(C75\) \[たくみなむち\] 超電磁砲のあいしかた",
"type" : "Manga", "type" : "Doujinshi",
}, },
}), }),
) )
@ -76,9 +78,9 @@ class HentaihandTagExtractor(Extractor):
r"/\w+/(parody|character|tag|artist|group|language" r"/\w+/(parody|character|tag|artist|group|language"
r"|category|relationship)/([^/?#]+)") r"|category|relationship)/([^/?#]+)")
test = ( test = (
("https://hentaihand.com/en/artist/himuro", { ("https://hentaihand.com/en/artist/takumi-na-muchi", {
"pattern": HentaihandGalleryExtractor.pattern, "pattern": HentaihandGalleryExtractor.pattern,
"count": ">= 18", "count": ">= 6",
}), }),
("https://hentaihand.com/en/tag/full-color"), ("https://hentaihand.com/en/tag/full-color"),
("https://hentaihand.com/fr/language/japanese"), ("https://hentaihand.com/fr/language/japanese"),

View File

@ -1,6 +1,6 @@
# -*- coding: utf-8 -*- # -*- coding: utf-8 -*-
# Copyright 2015-2020 Mike Fährmann # Copyright 2015-2021 Mike Fährmann
# #
# This program is free software; you can redistribute it and/or modify # 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 # it under the terms of the GNU General Public License version 2 as
@ -56,7 +56,7 @@ class MangaparkChapterExtractor(MangaparkBase, ChapterExtractor):
test = ( test = (
("https://mangapark.net/manga/gosu/i811653/c055/1", { ("https://mangapark.net/manga/gosu/i811653/c055/1", {
"count": 50, "count": 50,
"keyword": "8344bdda8cd8414e7729a4e148379f147e3437da", "keyword": "db1ed9af4f972756a25dbfa5af69a8f155b043ff",
}), }),
(("https://mangapark.net/manga" (("https://mangapark.net/manga"
"/ad-astra-per-aspera-hata-kenjirou/i662051/c001.2/1"), { "/ad-astra-per-aspera-hata-kenjirou/i662051/c001.2/1"), {
@ -121,7 +121,7 @@ class MangaparkMangaExtractor(MangaparkBase, MangaExtractor):
r"(/manga/[^/?#]+)/?$") r"(/manga/[^/?#]+)/?$")
test = ( test = (
("https://mangapark.net/manga/aria", { ("https://mangapark.net/manga/aria", {
"url": "f07caf0bc5097c9b32c8c0d6f446bce1bf4bd329", "url": "b8f7db2f581404753c4af37af66c049a41273b94",
"keyword": "2c0d28efaf84fcfe62932b6931ef3c3987cd48c0", "keyword": "2c0d28efaf84fcfe62932b6931ef3c3987cd48c0",
}), }),
("https://mangapark.me/manga/aria"), ("https://mangapark.me/manga/aria"),

View File

@ -215,7 +215,7 @@ class JoyreactorTagExtractor(ReactorTagExtractor):
"count": ">= 15", "count": ">= 15",
}), }),
("http://joyreactor.com/tag/Cirno", { ("http://joyreactor.com/tag/Cirno", {
"url": "de1e60c15bfb07a0e9603b00dc3d05f60edc7914", "url": "aa59090590b26f4654881301fe8fe748a51625a8",
}), }),
) )
@ -243,7 +243,7 @@ class JoyreactorUserExtractor(ReactorUserExtractor):
test = ( test = (
("http://joyreactor.cc/user/hemantic"), ("http://joyreactor.cc/user/hemantic"),
("http://joyreactor.com/user/Tacoman123", { ("http://joyreactor.com/user/Tacoman123", {
"url": "452cd0fa23e2ad0e122c296ba75aa7f0b29329f6", "url": "60ce9a3e3db791a0899f7fb7643b5b87d09ae3b5",
}), }),
) )
@ -254,23 +254,27 @@ class JoyreactorPostExtractor(ReactorPostExtractor):
pattern = JR_BASE_PATTERN + r"/post/(\d+)" pattern = JR_BASE_PATTERN + r"/post/(\d+)"
test = ( test = (
("http://joyreactor.com/post/3721876", { # single image ("http://joyreactor.com/post/3721876", { # single image
"url": "6ce09f239d8b7fdf6dd1664c2afc39618cc87663", "pattern": r"http://img\d\.joyreactor\.com/pics/post/full"
"keyword": "147ed5b9799ba43cbd16168450afcfae5ddedbf3", r"/cartoon-painting-monster-lake-4841316.jpeg",
"count": 1,
"keyword": "2207a7dfed55def2042b6c2554894c8d7fda386e",
}), }),
("http://joyreactor.com/post/3713804", { # 4 images ("http://joyreactor.com/post/3713804", { # 4 images
"url": "f08ac8493ca0619a3e3c6bedb8d8374af3eec304", "pattern": r"http://img\d\.joyreactor\.com/pics/post/full"
"keyword": "f12c6f3c2f298fed9b12bd3e70fb823870aa9b93", r"/movie-tv-godzilla-monsters-\d+\.jpeg",
"count": 4,
"keyword": "d7da9ba7809004c809eedcf6f1c06ad0fbb3df21",
}), }),
("http://joyreactor.com/post/3726210", { # gif / video ("http://joyreactor.com/post/3726210", { # gif / video
"url": "33a48e1eca6cb2d298fbbb6536b3283799d6515b", "url": "60f3b9a0a3918b269bea9b4f8f1a5ab3c2c550f8",
"keyword": "d173cc6e88f02a63904e475eacd7050304eb1967", "keyword": "8949d9d5fc469dab264752432efbaa499561664a",
}), }),
("http://joyreactor.com/post/3668724", { # youtube embed ("http://joyreactor.com/post/3668724", { # youtube embed
"url": "bf1666eddcff10c9b58f6be63fa94e4e13074214", "url": "bf1666eddcff10c9b58f6be63fa94e4e13074214",
"keyword": "e18b1ffbd79d76f9a0e90b6d474cc2499e343f0b", "keyword": "e18b1ffbd79d76f9a0e90b6d474cc2499e343f0b",
}), }),
("http://joyreactor.cc/post/1299", { # "malformed" JSON ("http://joyreactor.cc/post/1299", { # "malformed" JSON
"url": "ac900743ed7cf1baf3db3b531c3bc414bf1ffcde", "url": "ab02c6eb7b4035ad961b29ee0770ee41be2fcc39",
}), }),
) )
@ -311,7 +315,7 @@ class PornreactorUserExtractor(ReactorUserExtractor):
test = ( test = (
("http://pornreactor.cc/user/Disillusion", { ("http://pornreactor.cc/user/Disillusion", {
"range": "1-25", "range": "1-25",
"count": ">= 25", "count": ">= 20",
}), }),
("http://fapreactor.com/user/Disillusion"), ("http://fapreactor.com/user/Disillusion"),
) )
@ -324,10 +328,10 @@ class PornreactorPostExtractor(ReactorPostExtractor):
pattern = PR_BASE_PATTERN + r"/post/(\d+)" pattern = PR_BASE_PATTERN + r"/post/(\d+)"
test = ( test = (
("http://pornreactor.cc/post/863166", { ("http://pornreactor.cc/post/863166", {
"url": "680db1e33ca92ff70b2c0e1708c471cbe2201324", "url": "a09fb0577489e1f9564c25d0ad576f81b19c2ef3",
"content": "ec6b0568bfb1803648744077da082d14de844340", "content": "ec6b0568bfb1803648744077da082d14de844340",
}), }),
("http://fapreactor.com/post/863166", { ("http://fapreactor.com/post/863166", {
"url": "864ecd5785e4898301aa8d054dd653b1165be158", "url": "2a956ce0c90e8bc47b4392db4fa25ad1342f3e54",
}), }),
) )

View File

@ -140,8 +140,8 @@ class SubscribestarUserExtractor(SubscribestarExtractor):
test = ( test = (
("https://www.subscribestar.com/subscribestar", { ("https://www.subscribestar.com/subscribestar", {
"count": ">= 20", "count": ">= 20",
"pattern": r"https://star-uploads.s\d+-us-west-\d+.amazonaws.com" "pattern": r"https://(star-uploads|ss-uploads-prod)\.s\d+-us-west-"
r"/uploads/users/11/", r"\d+\.amazonaws\.com/uploads(_v2)?/users/11/",
"keyword": { "keyword": {
"author_id": 11, "author_id": 11,
"author_name": "subscribestar", "author_name": "subscribestar",

View File

@ -71,8 +71,8 @@ class WikiartArtistExtractor(WikiartExtractor):
directory_fmt = ("{category}", "{artist[artistName]}") directory_fmt = ("{category}", "{artist[artistName]}")
pattern = BASE_PATTERN + r"/(?!\w+-by-)([\w-]+)/?$" pattern = BASE_PATTERN + r"/(?!\w+-by-)([\w-]+)/?$"
test = ("https://www.wikiart.org/en/thomas-cole", { test = ("https://www.wikiart.org/en/thomas-cole", {
"url": "5140343730331786117fa5f4c013a6153393e28e", "url": "deabec0ed7efa97e2a729ff9d08b539143106bac",
"keyword": "4d9cbc50ebddfcb186f31ff70b08833578dd0070", "keyword": "751a5457b71c8704982d3bb6485a214cd3d07bf9",
}) })
def __init__(self, match): def __init__(self, match):