Spaces:
Running
Running
Update main.py
Browse files
main.py
CHANGED
|
@@ -5,6 +5,12 @@ from fastapi.responses import StreamingResponse
|
|
| 5 |
from pydantic import BaseModel
|
| 6 |
import re
|
| 7 |
import replicate
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
|
| 10 |
class Item(BaseModel):
|
|
@@ -86,8 +92,6 @@ async def root(item: Item):
|
|
| 86 |
|
| 87 |
|
| 88 |
|
| 89 |
-
|
| 90 |
-
|
| 91 |
@app.post("/extract-content")
|
| 92 |
async def root(item: Item):
|
| 93 |
|
|
@@ -102,5 +106,41 @@ async def root(item: Item):
|
|
| 102 |
|
| 103 |
except requests.RequestException as e:
|
| 104 |
return {"error": str(e), "status_code": 500}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 105 |
|
| 106 |
|
|
|
|
| 5 |
from pydantic import BaseModel
|
| 6 |
import re
|
| 7 |
import replicate
|
| 8 |
+
import OSError
|
| 9 |
+
from supabase import create_client, Client
|
| 10 |
+
|
| 11 |
+
url: str = os.environ.get("DB_URL")
|
| 12 |
+
key: str = os.environ.get("DB_KEY")
|
| 13 |
+
supabase: Client = create_client(url, key)
|
| 14 |
|
| 15 |
|
| 16 |
class Item(BaseModel):
|
|
|
|
| 92 |
|
| 93 |
|
| 94 |
|
|
|
|
|
|
|
| 95 |
@app.post("/extract-content")
|
| 96 |
async def root(item: Item):
|
| 97 |
|
|
|
|
| 106 |
|
| 107 |
except requests.RequestException as e:
|
| 108 |
return {"error": str(e), "status_code": 500}
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
async def insert_image(url , id):
|
| 112 |
+
data, count = await supabase.table('news')
|
| 113 |
+
.update({'image_url': url})
|
| 114 |
+
.eq('id', id)
|
| 115 |
+
.execute()
|
| 116 |
+
|
| 117 |
+
|
| 118 |
+
@app.get("/extract-images")
|
| 119 |
+
async def root(site: str = 'abcnews.go'):
|
| 120 |
+
|
| 121 |
+
try:
|
| 122 |
+
|
| 123 |
+
response = supabase.table('news').select("*")
|
| 124 |
+
.eq('source', 'www.'+ site + '.com').is_('image_url', 'null')
|
| 125 |
+
.order('published_date', desc=True).limit(15).execute()
|
| 126 |
+
|
| 127 |
+
|
| 128 |
+
for i in range(len(response.data)):
|
| 129 |
+
|
| 130 |
+
url = response.data[i]['article_url']
|
| 131 |
+
try:
|
| 132 |
+
res = requests.get(url)
|
| 133 |
+
soup = BeautifulSoup(res.text, 'html.parser')
|
| 134 |
+
results = soup.find_all(['img'])
|
| 135 |
+
url = results[1]['src']
|
| 136 |
+
|
| 137 |
+
await insert_image(url, response.data[i]['id'] )
|
| 138 |
+
|
| 139 |
+
except Exception as e:
|
| 140 |
+
results = None
|
| 141 |
+
|
| 142 |
+
|
| 143 |
+
except requests.RequestException as e:
|
| 144 |
+
return {"error": str(e), "status_code": 500}
|
| 145 |
|
| 146 |
|