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

[patreon] include full metadata with each URL (#4286 #8498)

Browse Source 
* include full metadata object with each url
    Fixes #4286
* rename 'url_metadata' to 'file'
This commit is contained in:
Octavia Togami
2025-11-20 10:30:13 -08:00
committed by GitHub
parent d74809c538
commit 685777e43e
2 changed files with 10 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
gallery_dl/extractor/patreon.py
View File

@@ -53,13 +53,14 @@ class PatreonExtractor(Extractor):
post["num"] = 0 post["num"] = 0
hashes = set() hashes = set()
for kind, url, name in itertools.chain.from_iterable( for kind, file, url, name in itertools.chain.from_iterable(
g(post) for g in generators): g(post) for g in generators):
fhash = self._filehash(url) fhash = self._filehash(url)
if fhash not in hashes or not fhash: if fhash not in hashes or not fhash:
hashes.add(fhash) hashes.add(fhash)
post["hash"] = fhash post["hash"] = fhash
post["type"] = kind post["type"] = kind
post["file"] = file
post["num"] += 1 post["num"] += 1
text.nameext_from_url(name, post) text.nameext_from_url(name, post)
if text.ext_from_url(url) == "m3u8": if text.ext_from_url(url) == "m3u8":
@@ -86,7 +87,7 @@ class PatreonExtractor(Extractor):
name = url name = url
else: else:
name = self._filename(url) or url name = self._filename(url) or url
return (("postfile", url, name),) return (("postfile", postfile, url, name),)
return () return ()
def _images(self, post): def _images(self, post):
@@ -94,7 +95,7 @@ class PatreonExtractor(Extractor):
for image in images: for image in images:
if url := self._images_url(image): if url := self._images_url(image):
name = image.get("file_name") or self._filename(url) or url name = image.get("file_name") or self._filename(url) or url
yield "image", url, name yield "image", image, url, name
def _images_url(self, image): def _images_url(self, image):
return image.get("download_url") return image.get("download_url")
@@ -109,24 +110,24 @@ class PatreonExtractor(Extractor):
if image := post.get("image"): if image := post.get("image"):
if url := image.get("large_url"): if url := image.get("large_url"):
name = image.get("file_name") or self._filename(url) or url name = image.get("file_name") or self._filename(url) or url
return (("image_large", url, name),) return (("image_large", image, url, name),)
return () return ()
def _attachments(self, post): def _attachments(self, post):
for attachment in post.get("attachments") or (): for attachment in post.get("attachments") or ():
if url := self.request_location(attachment["url"], fatal=False): if url := self.request_location(attachment["url"], fatal=False):
yield "attachment", url, attachment["name"] yield "attachment", attachment, url, attachment["name"]
for attachment in post.get("attachments_media") or (): for attachment in post.get("attachments_media") or ():
if url := attachment.get("download_url"): if url := attachment.get("download_url"):
yield "attachment", url, attachment["file_name"] yield "attachment", attachment, url, attachment["file_name"]
def _content(self, post): def _content(self, post):
if content := post.get("content"): if content := post.get("content"):
for img in text.extract_iter( for img in text.extract_iter(
content, '<img data-media-id="', '>'): content, '<img data-media-id="', '>'):
if url := text.extr(img, 'src="', '"'): if url := text.extr(img, 'src="', '"'):
yield "content", url, self._filename(url) or url yield "content", None, url, self._filename(url) or url
def posts(self): def posts(self):
"""Return all relevant post objects""" """Return all relevant post objects"""

2
test/results/patreon.py
View File

@@ -118,6 +118,8 @@ __tests__ = (
"#comment" : "postfile + attachments", "#comment" : "postfile + attachments",
"#class" : patreon.PatreonPostExtractor, "#class" : patreon.PatreonPostExtractor,
"#count" : 4, "#count" : 4,
"file": {dict, None},
}, },
{ {