github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[hentaifoundry] extract 'categories' metadata (#8656)

Browse Source 
* [hentaifoundry] Add support for categories
* rename 'hf_categories' to 'categories'
* move code into '_extract_categories()' method
* update tests

---------

Co-authored-by: Mike Fährmann <mike_faehrmann@web.de>
This commit is contained in:
icelit4
2025-12-08 01:15:59 -08:00
committed by GitHub
parent 9a8746f12d
commit 2ab74dfbf1
2 changed files with 16 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
gallery_dl/extractor/hentaifoundry.py
View File

@@ -86,6 +86,7 @@ class HentaifoundryExtractor(Extractor):
.replace("\r\n", "\n")),
"ratings" : [text.unescape(r) for r in text.extract_iter(extr(
"class='ratings_box'", "</div>"), "title='", "'")],
"categories" : self._extract_categories(extr),
"date" : self.parse_datetime_iso(extr("datetime='", "'")),
"views" : text.parse_int(extr(">Views</span>", "<")),
"score" : text.parse_int(extr(">Vote Score</span>", "<")),
@@ -141,11 +142,17 @@ class HentaifoundryExtractor(Extractor):
path = extr('class="pdfLink" href="', '"')
data["src"] = self.root + path
data["index"] = text.parse_int(path.rsplit("/", 2)[1])
data["categories"] = self._extract_categories(extr)
data["ratings"] = [text.unescape(r) for r in text.extract_iter(extr(
"class='ratings_box'", "</div>"), "title='", "'")]
return text.nameext_from_url(data["src"], data)
def _extract_categories(self, extr):
return [text.unescape(text.extr(c, ">", "<"))
for c in extr('class="categoryBreadcrumbs">', "</span>")
.split("&raquo;")]
def _request_check(self, url, **kwargs):
self.request = self._request_original

13
test/results/hentaifoundry.py
View File

@@ -90,6 +90,7 @@ __tests__ = (
"#sha1_content": "91bf01497c39254b6dfb234a18e8f01629c77fd1",
"artist" : "Tenpura",
"categories" : ["Anime & Manga"],
"date" : "dt:2016-02-22 14:41:19",
"description": "Thank you!",
"height" : 700,
@@ -120,6 +121,7 @@ __tests__ = (
"#results" : "https://pictures.hentai-foundry.com/s/Soloid/186714/Soloid-186714-Osaloop.swf",
"artist" : "Soloid",
"categories" : ["Misc"],
"date" : "dt:2013-02-07 17:25:54",
"description": "It took me ages.\nI hope you'll like it.\nSorry for the bad quality, I made it on after effect because Flash works like shit when you have 44 layers to animate, and the final ae SWF file is 55mo big.",
"extension" : "swf",
@@ -204,22 +206,24 @@ Sorry for the bad quality, I made it on after effect because Flash works like sh
{
"#url" : "https://www.hentai-foundry.com/stories/user/Likelymouse",
"#class" : hentaifoundry.HentaifoundryStoriesExtractor,
"#range" : "2",
"#results" : "https://www.hentai-foundry.com/stories/user/Likelymouse/77892/The-Sweater-and-Scarf-Public-Investigators-Part-1.pdf",
"author" : "Likelymouse",
"chapters" : 4,
"categories" : ["Original", "Neko Shoujo & Kemonomimi"],
"chapters" : 11,
"comments" : 0,
"date" : "dt:2025-09-01 00:00:00",
"date" : "dt:2025-10-04 00:00:00",
"extension" : "pdf",
"filename" : "The-Sweater-and-Scarf-Public-Investigators-Part-1",
"index" : 77892,
"rating" : 0,
"src" : "https://www.hentai-foundry.com/stories/user/Likelymouse/77892/The-Sweater-and-Scarf-Public-Investigators-Part-1.pdf",
"status" : "Incomplete",
"status" : "Complete",
"title" : "The Sweater and Scarf Public Investigators, Part 1",
"user" : "Likelymouse",
"views" : range(100, 10_000),
"words" : 10661,
"words" : 47031,
"description": """\
<div style="text-align:center"><a href="https://imgur.com/a/uRDss5c"><img src="https://i.imgur.com/SCWI09e.jpeg" alt="135x240"/></a><br />
<br />
@@ -250,6 +254,7 @@ Follow 22 y/o Puffy Penelope as she descends into degeneracy, mastering her new
"#sha1_url": "5a67cfa8c3bf7634c8af8485dd07c1ea74ee0ae8",
"title": "Overwatch High Chapter Voting Location",
"categories": ["Games", "Overwatch"],
},
)