更新代码

main.py

@@ -2,10 +2,11 @@ from source import XHS
 
 
 def example():
-    test_cookie = "abRequestId=25c57ab7-8cbf-5383-b020-08852c1704e1; webBuild=3.4.1; xsecappid=xhs-pc-web; a1=18a033d274338lwsfacj9x5bpf4fznhhc8xrakemj50000250510; webId=93c0636350d85103d93bca88da2959cd; websectiga=2a3d3ea002e7d92b5c9743590ebd24010cf3710ff3af8029153751e41a6af4a3; sec_poison_id=ae1f0190-4d0c-45f5-a75f-68c2d87c5573; web_session=030037a3ed570060b3a43845a6234a2f100ef4; gid=yY08qqfYy0SSyY08qqfJWYTd4qqY1EMi0SVjC6VC2DUi4F28iuIxx0888J282y880JfWY0Di; cache_feeds=[]"
+    image_demo = "https://www.xiaohongshu.com/explore/64d1b406000000000103ee8d"
-    demo = "https://www.xiaohongshu.com/explore/64a3a5170000000031008914"
+    video_demo = "https://www.xiaohongshu.com/explore/64c05652000000000c0378e7"
     xhs = XHS()
-    print(xhs.get_image(demo, cookie=test_cookie))
+    print(xhs.get_image(image_demo))
+    print(xhs.get_video(video_demo))
 
 
 if __name__ == '__main__':

source/Image.py

@@ -1,14 +1,23 @@
-from .Params import ID
+from re import compile
-from .Params import IMAGE_API
 
 
-def get_id(html: str) -> list:
+class Image:
-    return ID.findall(html)
+    IMAGE_API = "https://sns-img-qc.xhscdn.com/"
+    IMAGE_ID = compile(r'"traceId":"(.*?)"')
 
+    def __init__(self, html, params):
+        self.html = html
+        self.params = params
 
-def generate_url(ids: list) -> list:
+    def get_image_link(self, url: str, download: bool):
-    return [IMAGE_API + i for i in ids]
+        html = self.html.get_html(url)
+        return self.__get_image_links(html)
 
+    def __get_id(self, html: str) -> list:
+        return self.IMAGE_ID.findall(html)
 
-def get_url(html: str) -> list:
+    def __generate_url(self, ids: list) -> list:
-    return generate_url(get_id(html))
+        return [self.IMAGE_API + i for i in ids]
+
+    def __get_image_links(self, html: str) -> list:
+        return self.__generate_url(self.__get_id(html))

source/Video.py

@@ -0,0 +1,20 @@
+from re import compile
+
+
+class Video:
+    VIDEO_ID = compile(r'"masterUrl":"(.*?)"')
+
+    def __init__(self, html, params):
+        self.html = html
+        self.params = params
+
+    def get_video_link(self, url: str, download: bool):
+        html = self.html.get_html(url)
+        return self.__get_video_link(html)
+
+    def __get_video_link(self, html: str) -> list:
+        return [self.clean_url(u) for u in self.VIDEO_ID.findall(html)]
+
+    @staticmethod
+    def clean_url(url: str) -> str:
+        return bytes(url, "utf-8").decode("unicode_escape")

source/__init__.py

@@ -1,22 +1,24 @@
-from .Html import get_html
+from .Html import Html
-from .Image import get_url
+from .Image import Image
 from .Params import Params
+from .Video import Video
+
+HEADERS = {
+    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/115.0.0.0 Safari/537.36 Edg/115.0.1901.203",
+    "Cookie": "abRequestId=fd245483-beed-57b0-abfc-440b6a6be2aa; webBuild=3.4.1; xsecappid=xhs-pc-web; a1=189fe37918ezx1jqcbe9fin95cnxqj2ewcbc250yp50000234538; webId=9fff21309cfd3e4f380a6c75ed463803; websectiga=f47eda31ec99545da40c2f731f0630efd2b0959e1dd10d5fedac3dce0bd1e04d; sec_poison_id=003395d3-6520-4a02-851a-17d093203251; web_session=030037a3efee2e602d5d16fca4234a8a44466c; gid=yYjidqWi2KE4yYjidqWjyS28YduCyVASDdjiDvU3Ij2SIS28CAVJdJ888Jq42qY88J44DyjS",
+}
 
 
 class XHS:
-    def __init__(self, path="./"):
+    def __init__(self, path="./", headers=None):
         self.params = Params(path)
-        self.image = Image(self.params)
+        self.html = Html(headers or HEADERS)
+        self.image = Image(self.html, self.params)
+        self.video = Video(self.html, self.params)
+        self._cookie = ""
 
-    def get_image(self, url: str, cookie=None, ):
+    def get_image(self, url: str, download=False):
-        return self.image.get_image_link(url, cookie)
+        return self.image.get_image_link(url, download)
 
-
+    def get_video(self, url: str, download=False):
-class Image:
+        return self.video.get_video_link(url, download)
-    def __init__(self, params):
-        self.params = params
-
-    @staticmethod
-    def get_image_link(url: str, cookie=None, ):
-        html = get_html(url, cookie=cookie, )
-        return get_url(html)