super/crawler-plugin
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

同步 Gitee 应用代码更新

Browse Source
This commit is contained in:
super
2026-05-29 19:47:23 +08:00
parent 266c0f17c1
commit 080625567b
7 changed files with 103 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
app/amazon/detail_spider.py
View File

@@ -13,7 +13,7 @@ from amazon.amazon_base import TaskBase
from amazon.chrome_base import ChromeAmzoneBase
from config import runing_task, runing_shop,base_dir,DELETE_BRAND_API_BASE
from config import runing_task,runing_shop,base_dir,DELETE_BRAND_API_BASE
class ChromeAmzone(ChromeAmzoneBase):
@@ -92,7 +92,8 @@ class ChromeAmzone(ChromeAmzoneBase):
"""
data = {
'image_url': "",
'title': ""
'title': "",
"sku" : ""
}
try:
@@ -103,26 +104,39 @@ class ChromeAmzone(ChromeAmzoneBase):
data["title"] = title
sku_ele_ls = self.tab.eles('xpath://ul[@class="a-unordered-list a-vertical a-spacing-mini"]', timeout=20)
if len(sku_ele_ls) > 0:
data["title"] += sku_ele_ls[0].text
data["sku"] = sku_ele_ls[0].text
imge_ele = self.tab.ele('xpath://div[@id="imgTagWrapperId"]//img',timeout=20)
image_url = ""
min_image_url = ""
data_a_dynamic_image = imge_ele.attr("data-a-dynamic-image")
if data_a_dynamic_image:
dynamic_image_json = json.loads(data_a_dynamic_image)
self.log(f"图片信息:{dynamic_image_json}")
max_area = 0
min_area = 0
for url, (width, height) in dynamic_image_json.items():
area = width * height
if area > max_area:
max_area = area
image_url = url
if min_area == 0:
min_area = area
min_image_url = url
if area < min_area:
min_area = area
min_image_url = url
if not image_url:
image_url = imge_ele.attr("src")
if not min_image_url:
min_image_url = imge_ele.attr("src")
data["image_url"] = image_url
data["min_image_url"] = min_image_url
return data
except Exception as e:
@@ -266,6 +280,7 @@ class SpiderTask(TaskBase):
"country": i.get("country"),
"url": return_data.get("image_url"),
"title": return_data.get("title"),
"sku": return_data.get("sku")
}
for i in items
]