[tenor] add 'user' extractor (#6075)

This commit is contained in:
Mike Fährmann
2025-03-04 21:47:16 +01:00
parent b7bb40c06b
commit f5073605f6
3 changed files with 80 additions and 40 deletions

View File

@@ -940,7 +940,7 @@ Consider all listed sites to potentially be NSFW.
<tr>
<td>Tenor</td>
<td>https://tenor.com/</td>
<td>individual Images, Search Results</td>
<td>individual Images, Search Results, User Profiles</td>
<td></td>
</tr>
<tr>

View File

@@ -57,6 +57,47 @@ class TenorExtractor(Extractor):
if fmt in media_formats:
return media_formats[fmt]
def _search_results(self, query):
url = "https://tenor.googleapis.com/v2/search"
params = {
"appversion": "browser-r20250225-1",
"prettyPrint": "false",
"key": "AIzaSyC-P6_qz3FzCoXGLk6tgitZo4jEJ5mLzD8",
"client_key": "tenor_web",
"locale": "en",
"anon_id": "",
"q": query,
"limit": "50",
"contentfilter": "low",
"media_filter": "gif,gif_transparent,mediumgif,tinygif,"
"tinygif_transparent,webp,webp_transparent,"
"tinywebp,tinywebp_transparent,tinymp4,mp4,webm,"
"originalgif,gifpreview",
"fields": "next,results.id,results.media_formats,results.title,"
"results.h1_title,results.long_title,results.itemurl,"
"results.url,results.created,results.user,"
"results.shares,results.embed,results.hasaudio,"
"results.policy_status,results.source_id,results.flags,"
"results.tags,results.content_rating,results.bg_color,"
"results.legacy_info,results.geographic_restriction,"
"results.content_description",
"pos": None,
"component": "web_desktop",
}
headers = {
"Referer": self.root + "/",
"Origin" : self.root,
}
while True:
data = self.request(url, params=params, headers=headers).json()
yield from data["results"]
params["pos"] = data.get("next")
if not params["pos"]:
return
def metadata(self):
return False
@@ -93,43 +134,14 @@ class TenorSearchExtractor(TenorExtractor):
return {"search_tags": self.search_tags}
def gifs(self):
url = "https://tenor.googleapis.com/v2/search"
params = {
"appversion": "browser-r20250225-1",
"prettyPrint": "false",
"key": "AIzaSyC-P6_qz3FzCoXGLk6tgitZo4jEJ5mLzD8",
"client_key": "tenor_web",
"locale": "en",
"anon_id": "",
"q": self.search_tags,
"limit": "50",
"contentfilter": "low",
"media_filter": "gif,gif_transparent,mediumgif,tinygif,"
"tinygif_transparent,webp,webp_transparent,"
"tinywebp,tinywebp_transparent,tinymp4,mp4,webm,"
"originalgif,gifpreview",
"fields": "next,results.id,results.media_formats,results.title,"
"results.h1_title,results.long_title,results.itemurl,"
"results.url,results.created,results.user,"
"results.shares,results.embed,results.hasaudio,"
"results.policy_status,results.source_id,results.flags,"
"results.tags,results.content_rating,results.bg_color,"
"results.legacy_info,results.geographic_restriction,"
"results.content_description",
"searchfilter": "none",
"pos": None,
"component": "web_desktop",
}
headers = {
"Referer": self.root + "/",
"Origin" : self.root,
}
return self._search_results(self.search_tags)
while True:
data = self.request(url, params=params, headers=headers).json()
yield from data["results"]
class TenorUserExtractor(TenorExtractor):
subcategory = "user"
directory_fmt = ("{category}", "@{user[username]}")
pattern = BASE_PATTERN + r"/users/([^/?#]+)"
example = "https://tenor.com/users/USER"
params["pos"] = data.get("next")
if not params["pos"]:
return
def gifs(self):
return self._search_results("@" + self.groups[0])

View File

@@ -17,7 +17,7 @@ __tests__ = (
"content_description": "an illustration of a tree with green leaves",
"created" : 1687512768.687436,
"date" : "dt:2023-06-23 09:32:48",
"embed" : "<div class=\"tenor-gif-embed\" data-postid=\"8525772382434057283\" data-share-method=\"host\" data-aspect-ratio=\"0.937751\" data-width=\"100%\"><a href=\"https://tenor.com/view/moving-gif-8525772382434057283\">Moving Sticker</a>from <a href=\"https://tenor.com/search/moving-stickers\">Moving Stickers</a></div> <script type=\"text/javascript\" async src=\"https://tenor.com/embed.js\"></script>",
"embed" : r"re:<div class=.+",
"extension": "gif",
"filename" : "moving",
"h1_title" : "Moving Sticker",
@@ -30,7 +30,7 @@ __tests__ = (
"long_title": "Moving Sticker - Moving Stickers",
"media_formats": dict,
"policy_status": "POLICY_STATUS_UNSPECIFIED",
"shares" : 42528,
"shares" : range(40000, 60000),
"source_id": "",
"title" : "Moving Sti",
"url" : "https://tenor.com/kjYh53rdMGt.gif",
@@ -88,4 +88,32 @@ __tests__ = (
"search_tags": "trees water wind sun <&>",
},
{
"#url" : "https://tenor.com/users/robloxfan123",
"#class" : tenor.TenorUserExtractor,
"#urls" : "https://media1.tenor.com/m/1auSjzCikuoAAAAC/2016-roblox.gif",
"user": {
"profile_id": "8180139772821505417",
"url" : "https://tenor.com/users/ROBLOXfan123",
"userid" : "11206759",
"username" : "ROBLOXfan123",
"usertype" : "user",
},
},
{
"#url" : "https://tenor.com/users/annetv",
"#class" : tenor.TenorUserExtractor,
"#pattern": r"https://media\d+\.tenor\.com/m/[\w-]+/[\w%-]+\.gif",
"#count" : range(70, 100),
"user": {
"profile_id": "14727075564983373376",
"url" : "https://tenor.com/users/annetv",
"userid" : "8529134",
"username" : "annetv",
},
},
)