summaryrefslogtreecommitdiffstats
path: root/gallery_dl/extractor/shimmie2.py
diff options
context:
space:
mode:
Diffstat (limited to 'gallery_dl/extractor/shimmie2.py')
-rw-r--r--gallery_dl/extractor/shimmie2.py220
1 files changed, 65 insertions, 155 deletions
diff --git a/gallery_dl/extractor/shimmie2.py b/gallery_dl/extractor/shimmie2.py
index 285cd8f..912e601 100644
--- a/gallery_dl/extractor/shimmie2.py
+++ b/gallery_dl/extractor/shimmie2.py
@@ -18,21 +18,23 @@ class Shimmie2Extractor(BaseExtractor):
filename_fmt = "{category}_{id}{md5:?_//}.{extension}"
archive_fmt = "{id}"
- def __init__(self, match):
- BaseExtractor.__init__(self, match)
-
+ def _init(self):
try:
instance = INSTANCES[self.category]
except KeyError:
- pass
- else:
- cookies = instance.get("cookies")
- if cookies:
- domain = self.root.rpartition("/")[2]
- self._update_cookies_dict(cookies, domain=domain)
- file_url = instance.get("file_url")
- if file_url:
- self.file_url_fmt = file_url
+ return
+
+ cookies = instance.get("cookies")
+ if cookies:
+ domain = self.root.rpartition("/")[2]
+ self.cookies_update_dict(cookies, domain=domain)
+
+ file_url = instance.get("file_url")
+ if file_url:
+ self.file_url_fmt = file_url
+
+ if self.category == "giantessbooru":
+ self.posts = self._posts_giantessbooru
def items(self):
data = self.metadata()
@@ -64,10 +66,6 @@ class Shimmie2Extractor(BaseExtractor):
INSTANCES = {
- "mememuseum": {
- "root": "https://meme.museum",
- "pattern": r"meme\.museum",
- },
"loudbooru": {
"root": "https://loudbooru.com",
"pattern": r"loudbooru\.com",
@@ -85,11 +83,11 @@ INSTANCES = {
"cavemanon": {
"root": "https://booru.cavemanon.xyz",
"pattern": r"booru\.cavemanon\.xyz",
- "file_url": "{0}/index.php?q=image/{2}.{4}"
+ "file_url": "{0}/index.php?q=image/{2}.{4}",
},
}
-BASE_PATTERN = Shimmie2Extractor.update(INSTANCES) + r"/(?:index\.php\?q=)?"
+BASE_PATTERN = Shimmie2Extractor.update(INSTANCES) + r"/(?:index\.php\?q=/?)?"
class Shimmie2TagExtractor(Shimmie2Extractor):
@@ -98,33 +96,7 @@ class Shimmie2TagExtractor(Shimmie2Extractor):
directory_fmt = ("{category}", "{search_tags}")
file_url_fmt = "{}/_images/{}/{}%20-%20{}.{}"
pattern = BASE_PATTERN + r"post/list/([^/?#]+)(?:/(\d+))?()"
- test = (
- ("https://meme.museum/post/list/animated/1", {
- "pattern": r"https://meme\.museum/_images/\w+/\d+%20-%20",
- "count": ">= 30"
- }),
- ("https://loudbooru.com/post/list/original_character/1", {
- "pattern": r"https://loudbooru\.com/_images/[0-9a-f]{32}/\d+",
- "range": "1-100",
- "count": 100,
- }),
- ("https://giantessbooru.com/post/list/smiling/1", {
- "pattern": r"https://giantessbooru\.com/_images/[0-9a-f]{32}/\d+",
- "range": "1-100",
- "count": 100,
- }),
- ("https://tentaclerape.net/post/list/comic/1", {
- "pattern": r"https://tentaclerape\.net/_images/[0-9a-f]{32}/\d+",
- "range": "1-100",
- "count": 100,
- }),
- ("https://booru.cavemanon.xyz/index.php?q=post/list/Amber/1", {
- "pattern": r"https://booru\.cavemanon\.xyz"
- r"/index\.php\?q=image/\d+\.\w+",
- "range": "1-100",
- "count": 100,
- }),
- )
+ example = "https://loudbooru.com/post/list/TAG/1"
def __init__(self, match):
Shimmie2Extractor.__init__(self, match)
@@ -184,120 +156,43 @@ class Shimmie2TagExtractor(Shimmie2Extractor):
if not extr("/{}'>{}<".format(pnum, pnum), ">"):
return
+ def _posts_giantessbooru(self):
+ pnum = text.parse_int(self.page, 1)
+ file_url_fmt = (self.root + "/index.php?q=/image/{}.jpg").format
+
+ while True:
+ url = "{}/index.php?q=/post/list/{}/{}".format(
+ self.root, self.tags, pnum)
+ extr = text.extract_from(self.request(url).text)
+
+ while True:
+ pid = extr('href="./index.php?q=/post/view/', '&')
+ if not pid:
+ break
+
+ tags, dimensions, size = extr('title="', '"').split(" // ")
+ width, _, height = dimensions.partition("x")
+
+ yield {
+ "file_url": file_url_fmt(pid),
+ "id": pid,
+ "md5": "",
+ "tags": tags,
+ "width": width,
+ "height": height,
+ "size": text.parse_bytes(size[:-1]),
+ }
+
+ pnum += 1
+ if not extr('/{}">{}<'.format(pnum, pnum), ">"):
+ return
+
class Shimmie2PostExtractor(Shimmie2Extractor):
"""Extractor for single shimmie2 posts"""
subcategory = "post"
pattern = BASE_PATTERN + r"post/view/(\d+)"
- test = (
- ("https://meme.museum/post/view/10243", {
- "pattern": r"https://meme\.museum/_images/105febebcd5ca791ee332adc"
- r"49971f78/10243%20-%20g%20beard%20open_source%20richar"
- r"d_stallman%20stallman%20tagme%20text\.jpg",
- "content": "45565f3f141fc960a8ae1168b80e718a494c52d2",
- "keyword": {
- "extension": "jpg",
- "file_url": "https://meme.museum/_images/105febebcd5ca791ee332"
- "adc49971f78/10243%20-%20g%20beard%20open_source%2"
- "0richard_stallman%20stallman%20tagme%20text.jpg",
- "filename": "10243 - g beard open_source richard_stallman "
- "stallman tagme text",
- "height": 451,
- "id": 10243,
- "md5": "105febebcd5ca791ee332adc49971f78",
- "size": 0,
- "subcategory": "post",
- "tags": "/g/ beard open_source "
- "richard_stallman stallman tagme text",
- "width": 480,
- },
- }),
- ("https://loudbooru.com/post/view/33828", {
- "pattern": r"https://loudbooru\.com/_images/.+\.png",
- "content": "a4755f787ba23ae2aa297a46810f802ca9032739",
- "keyword": {
- "extension": "png",
- "file_url": "https://loudbooru.com/_images/ca2638d903c86e8337f"
- "e9aeb4974be88/33828%20-%202020%20artist%3Astikyfi"
- "nkaz%20character%3Alisa_loud%20cover%20fanfiction"
- "%3Aplatz_eins%20frowning%20half-closed_eyes%20sol"
- "o%20text%20title_card.png",
- "filename": "33828 - 2020 artist:stikyfinkaz character:lisa_"
- "loud cover fanfiction:platz_eins frowning "
- "half-closed_eyes solo text title_card",
- "height": 1920,
- "id": 33828,
- "md5": "ca2638d903c86e8337fe9aeb4974be88",
- "tags": "2020 artist:stikyfinkaz character:lisa_loud cover "
- "fanfiction:platz_eins frowning half-closed_eyes "
- "solo text title_card",
- "width": 1078,
- },
- }),
- ("https://giantessbooru.com/post/view/41", {
- "pattern": r"https://giantessbooru\.com/_images"
- r"/3f67e1986496806b7b14ff3e82ac5af4/41\.jpg",
- "content": "79115ed309d1f4e82e7bead6948760e889139c91",
- "keyword": {
- "extension": "jpg",
- "file_url": "https://giantessbooru.com/_images"
- "/3f67e1986496806b7b14ff3e82ac5af4/41.jpg",
- "filename": "41",
- "height": 0,
- "id": 41,
- "md5": "3f67e1986496806b7b14ff3e82ac5af4",
- "size": 0,
- "tags": "anime bare_midriff color drawing gentle giantess "
- "karbo looking_at_tinies negeyari outdoors smiling "
- "snake_girl white_hair",
- "width": 0
-
-
- },
- }),
- ("https://tentaclerape.net/post/view/10", {
- "pattern": r"https://tentaclerape\.net/\./index\.php"
- r"\?q=/image/10\.jpg",
- "content": "d0fd8f0f6517a76cb5e23ba09f3844950bf2c516",
- "keyword": {
- "extension": "jpg",
- "file_url": "https://tentaclerape.net/./index.php"
- "?q=/image/10.jpg",
- "filename": "10",
- "height": 427,
- "id": 10,
- "md5": "945db71eeccaef82ce44b77564260c0b",
- "size": 0,
- "subcategory": "post",
- "tags": "Deviant_Art Pet Tentacle artist_sche blonde_hair "
- "blouse boots green_eyes highheels leash miniskirt "
- "octopus schoolgirl white_skin willing",
- "width": 300,
- },
- }),
- # video
- ("https://tentaclerape.net/post/view/91267", {
- "pattern": r"https://tentaclerape\.net/\./index\.php"
- r"\?q=/image/91267\.mp4",
- }),
- ("https://booru.cavemanon.xyz/index.php?q=post/view/8335", {
- "pattern": r"https://booru\.cavemanon\.xyz"
- r"/index\.php\?q=image/8335\.png",
- "content": "7158f7e4abbbf143bad5835eb93dbe4d68c1d4ab",
- "keyword": {
- "extension": "png",
- "file_url": "https://booru.cavemanon.xyz"
- "/index.php?q=image/8335.png",
- "filename": "8335",
- "height": 460,
- "id": 8335,
- "md5": "",
- "size": 0,
- "tags": "Color Fang",
- "width": 459,
- },
- }),
- )
+ example = "https://loudbooru.com/post/view/12345"
def __init__(self, match):
Shimmie2Extractor.__init__(self, match)
@@ -313,7 +208,7 @@ class Shimmie2PostExtractor(Shimmie2Extractor):
"md5" : extr("/_thumbs/", "/"),
"file_url": self.root + (
extr("id='main_image' src='", "'") or
- extr("<source src='", "'")),
+ extr("<source src='", "'")).lstrip("."),
"width" : extr("data-width=", " ").strip("\"'"),
"height" : extr("data-height=", ">").partition(
" ")[0].strip("\"'"),
@@ -324,3 +219,18 @@ class Shimmie2PostExtractor(Shimmie2Extractor):
post["md5"] = text.extr(post["file_url"], "/_images/", "/")
return (post,)
+
+ def _posts_giantessbooru(self):
+ url = "{}/index.php?q=/post/view/{}".format(
+ self.root, self.post_id)
+ extr = text.extract_from(self.request(url).text)
+
+ return ({
+ "id" : self.post_id,
+ "tags" : extr(": ", "<").partition(" - ")[0].rstrip(")"),
+ "md5" : "",
+ "file_url": self.root + extr('id="main_image" src=".', '"'),
+ "width" : extr("orig_width =", ";"),
+ "height" : 0,
+ "size" : 0,
+ },)
generated by cgit v1.2.3 (git 2.46.0) at 2026-01-26 20:19:58 +0000