diff options
Diffstat (limited to 'gallery_dl/extractor/exhentai.py')
| -rw-r--r-- | gallery_dl/extractor/exhentai.py | 157 |
1 files changed, 52 insertions, 105 deletions
diff --git a/gallery_dl/extractor/exhentai.py b/gallery_dl/extractor/exhentai.py index 9cd7ae4..44bfe7d 100644 --- a/gallery_dl/extractor/exhentai.py +++ b/gallery_dl/extractor/exhentai.py @@ -23,25 +23,28 @@ class ExhentaiExtractor(Extractor): directory_fmt = ("{category}", "{gid} {title[:247]}") filename_fmt = "{gid}_{num:>04}_{image_token}_{filename}.{extension}" archive_fmt = "{gid}_{num}" - cookienames = ("ipb_member_id", "ipb_pass_hash") - cookiedomain = ".exhentai.org" + cookies_domain = ".exhentai.org" + cookies_names = ("ipb_member_id", "ipb_pass_hash") root = "https://exhentai.org" request_interval = 5.0 LIMIT = False def __init__(self, match): - # allow calling 'self.config()' before 'Extractor.__init__()' - self._cfgpath = ("extractor", self.category, self.subcategory) + Extractor.__init__(self, match) + self.version = match.group(1) - version = match.group(1) + def initialize(self): domain = self.config("domain", "auto") if domain == "auto": - domain = ("ex" if version == "ex" else "e-") + "hentai.org" + domain = ("ex" if self.version == "ex" else "e-") + "hentai.org" self.root = "https://" + domain - self.cookiedomain = "." + domain + self.cookies_domain = "." + domain - Extractor.__init__(self, match) + Extractor.initialize(self) + + if self.version != "ex": + self.cookies.set("nw", "1", domain=self.cookies_domain) self.original = self.config("original", True) limits = self.config("limits", False) @@ -51,10 +54,6 @@ class ExhentaiExtractor(Extractor): else: self.limits = False - self.session.headers["Referer"] = self.root + "/" - if version != "ex": - self.session.cookies.set("nw", "1", domain=self.cookiedomain) - def request(self, url, **kwargs): response = Extractor.request(self, url, **kwargs) if response.history and response.headers.get("Content-Length") == "0": @@ -66,17 +65,20 @@ class ExhentaiExtractor(Extractor): """Login and set necessary cookies""" if self.LIMIT: raise exception.StopExtraction("Image limit reached!") - if self._check_cookies(self.cookienames): + + if self.cookies_check(self.cookies_names): return + username, password = self._get_auth_info() if username: - self._update_cookies(self._login_impl(username, password)) - else: - self.log.info("no username given; using e-hentai.org") - self.root = "https://e-hentai.org" - self.original = False - self.limits = False - self.session.cookies["nw"] = "1" + return self.cookies_update(self._login_impl(username, password)) + + self.log.info("no username given; using e-hentai.org") + self.root = "https://e-hentai.org" + self.cookies_domain = ".e-hentai.org" + self.cookies.set("nw", "1", domain=self.cookies_domain) + self.original = False + self.limits = False @cache(maxage=90*24*3600, keyarg=1) def _login_impl(self, username, password): @@ -97,7 +99,7 @@ class ExhentaiExtractor(Extractor): response = self.request(url, method="POST", headers=headers, data=data) if b"You are now logged in as:" not in response.content: raise exception.AuthenticationError() - return {c: response.cookies[c] for c in self.cookienames} + return {c: response.cookies[c] for c in self.cookies_names} class ExhentaiGalleryExtractor(ExhentaiExtractor): @@ -106,61 +108,7 @@ class ExhentaiGalleryExtractor(ExhentaiExtractor): pattern = (BASE_PATTERN + r"(?:/g/(\d+)/([\da-f]{10})" r"|/s/([\da-f]{10})/(\d+)-(\d+))") - test = ( - ("https://exhentai.org/g/1200119/d55c44d3d0/", { - "options": (("original", False),), - "keyword": { - "cost": int, - "date": "dt:2018-03-18 20:14:00", - "eh_category": "Non-H", - "expunged": False, - "favorites": r"re:^[12]\d$", - "filecount": "4", - "filesize": 1488978, - "gid": 1200119, - "height": int, - "image_token": "re:[0-9a-f]{10}", - "lang": "ja", - "language": "Japanese", - "parent": "", - "rating": r"re:\d\.\d+", - "size": int, - "tags": [ - "parody:komi-san wa komyushou desu.", - "character:shouko komi", - "group:seventh lowlife", - "other:sample", - ], - "thumb": "https://exhentai.org/t/ce/0a/ce0a5bcb583229a9b07c0f8" - "3bcb1630ab1350640-624622-736-1036-jpg_250.jpg", - "title": "C93 [Seventh_Lowlife] Komi-san ha Tokidoki Daitan de" - "su (Komi-san wa Komyushou desu) [Sample]", - "title_jpn": "(C93) [Comiketjack (わ！)] 古見さんは、時々大胆" - "です。 (古見さんは、コミュ症です。) [見本]", - "token": "d55c44d3d0", - "torrentcount": "0", - "uploader": "klorpa", - "width": int, - }, - "content": ("2c68cff8a7ca540a78c36fdbf5fbae0260484f87", - "e9891a4c017ed0bb734cd1efba5cd03f594d31ff"), - }), - ("https://exhentai.org/g/960461/4f0e369d82/", { - "exception": exception.NotFoundError, - }), - ("http://exhentai.org/g/962698/7f02358e00/", { - "exception": exception.AuthorizationError, - }), - ("https://exhentai.org/s/f68367b4c8/1200119-3", { - "options": (("original", False),), - "count": 2, - }), - ("https://e-hentai.org/s/f68367b4c8/1200119-3", { - "options": (("original", False),), - "count": 2, - }), - ("https://g.e-hentai.org/g/1200119/d55c44d3d0/"), - ) + example = "https://e-hentai.org/g/12345/67890abcde/" def __init__(self, match): ExhentaiExtractor.__init__(self, match) @@ -171,10 +119,25 @@ class ExhentaiGalleryExtractor(ExhentaiExtractor): self.image_token = match.group(4) self.image_num = text.parse_int(match.group(6), 1) + def _init(self): source = self.config("source") if source == "hitomi": self.items = self._items_hitomi + def favorite(self, slot="0"): + url = self.root + "/gallerypopups.php" + params = { + "gid": self.gallery_id, + "t" : self.gallery_token, + "act": "addfav", + } + data = { + "favcat" : slot, + "apply" : "Apply Changes", + "update" : "1", + } + self.request(url, method="POST", params=params, data=data) + def items(self): self.login() @@ -219,6 +182,10 @@ class ExhentaiGalleryExtractor(ExhentaiExtractor): data["_http_validate"] = None yield Message.Url, url, data + fav = self.config("fav") + if fav is not None: + self.favorite(fav) + def _items_hitomi(self): if self.config("metadata", False): data = self.metadata_from_api() @@ -390,8 +357,9 @@ class ExhentaiGalleryExtractor(ExhentaiExtractor): url = "https://e-hentai.org/home.php" cookies = { cookie.name: cookie.value - for cookie in self.session.cookies - if cookie.domain == self.cookiedomain and cookie.name != "igneous" + for cookie in self.cookies + if cookie.domain == self.cookies_domain and + cookie.name != "igneous" } page = self.request(url, cookies=cookies).text @@ -458,26 +426,10 @@ class ExhentaiSearchExtractor(ExhentaiExtractor): """Extractor for exhentai search results""" subcategory = "search" pattern = BASE_PATTERN + r"/(?:\?([^#]*)|tag/([^/?#]+))" - test = ( - ("https://e-hentai.org/?f_search=touhou"), - ("https://exhentai.org/?f_cats=767&f_search=touhou"), - ("https://exhentai.org/tag/parody:touhou+project"), - (("https://exhentai.org/?f_doujinshi=0&f_manga=0&f_artistcg=0" - "&f_gamecg=0&f_western=0&f_non-h=1&f_imageset=0&f_cosplay=0" - "&f_asianporn=0&f_misc=0&f_search=touhou&f_apply=Apply+Filter"), { - "pattern": ExhentaiGalleryExtractor.pattern, - "range": "1-30", - "count": 30, - "keyword": { - "gallery_id": int, - "gallery_token": r"re:^[0-9a-f]{10}$" - }, - }), - ) + example = "https://e-hentai.org/?f_search=QUERY" def __init__(self, match): ExhentaiExtractor.__init__(self, match) - self.search_url = self.root _, query, tag = match.groups() if tag: @@ -492,6 +444,9 @@ class ExhentaiSearchExtractor(ExhentaiExtractor): if "next" not in self.params: self.params["page"] = text.parse_int(self.params.get("page")) + def _init(self): + self.search_url = self.root + def items(self): self.login() data = {"_extractor": ExhentaiGalleryExtractor} @@ -528,15 +483,7 @@ class ExhentaiFavoriteExtractor(ExhentaiSearchExtractor): """Extractor for favorited exhentai galleries""" subcategory = "favorite" pattern = BASE_PATTERN + r"/favorites\.php(?:\?([^#]*)())?" - test = ( - ("https://e-hentai.org/favorites.php", { - "count": 1, - "pattern": r"https?://e-hentai\.org/g/1200119/d55c44d3d0" - }), - ("https://exhentai.org/favorites.php?favcat=1&f_search=touhou" - "&f_apply=Search+Favorites"), - ) + example = "https://e-hentai.org/favorites.php" - def __init__(self, match): - ExhentaiSearchExtractor.__init__(self, match) + def _init(self): self.search_url = self.root + "/favorites.php" |