Browse the data

1import requests
2import re
3import hashlib
4import secrets
5from fastapi import FastAPI, HTTPException, Header
6from pydantic import BaseModel
7from urllib.parse import urlparse
8 
9app = FastAPI()
10 
11users = {}
12tokens = {}
13thumbnails = {}
14thumbnail_id_counter = 1
15 
16class SignupRequest(BaseModel):
17    username: str
18    password: str
19 
20class LoginRequest(BaseModel):
21    username: str
22    password: str
23 
24class ThumbnailGetRequest(BaseModel):
25    video_page_url: str
26 
27def hash_password(password: str):
28    return hashlib.sha256(password.encode()).hexdigest()
29 
30def generate_token():
31    return secrets.token_hex(32)
32 
33def get_user_from_token(authorization: str = Header(None)):
34    if not authorization:
35        raise HTTPException(status_code=401, detail="Missing Authorization header")
36    token = authorization.replace("Bearer ", "")
37    if token not in tokens:
38        raise HTTPException(status_code=401, detail="Invalid token")
39    return tokens[token]
40 
41@app.post("/signup")
42def signup(req: SignupRequest):
43    if req.username in users:
44        raise HTTPException(status_code=400, detail="Username already exists")
45    users[req.username] = hash_password(req.password)
46    return {"message": "User created"}
47 
48@app.post("/login")
49def login(req: LoginRequest):
50    if req.username not in users or users[req.username] != hash_password(req.password):
51        raise HTTPException(status_code=401, detail="Invalid credentials")
52    token = generate_token()
53    tokens[token] = req.username
54    return {"token": token}
55 
56@app.post("/thumbnail/get")
57def get_thumbnail(req: ThumbnailGetRequest, authorization: str = Header(None)):
58    user = get_user_from_token(authorization)
59    global thumbnail_id_counter
60 
61    video_url = req.video_page_url
62 
63    try:
64        response = requests.get(video_url, headers={"User-Agent": "Mozilla/5.0"})
65        response.raise_for_status()
66        html = response.text
67    except Exception as e:
68        raise HTTPException(status_code=400, detail=f"Failed to fetch video page: {str(e)}")
69 
70    # Try multiple patterns to extract thumbnail URLs
71    thumbnail_url = None
72 
73    # YouTube-style: yt-thumbnail or thumbnail_url in JSON
74    patterns = [
75        r'<link\s+rel="image_src"\s+href="([^"]+)"',
76        r'"thumbnail_url"\s*:\s*"([^"]+)"',
77        r'"thumbnail"\s*:\s*{\s*"url"\s*:\s*"([^"]+)"',
78        r'property="og:image"\s+content="([^"]+)"',
79        r'content="([^"]+)"\s+property="og:image"',
80        r'<img[^>]+class="[^"]*thumbnail[^"]*"[^>]+src="([^"]+)"',
81        r'data-thumbnail="([^"]+)"',
82        r'poster="([^"]+)"',
83    ]
84 
85    for pattern in patterns:
86        match = re.search(pattern, html, re.IGNORECASE)
87        if match:
88            candidate = match.group(1)
89            if candidate.startswith("//"):
90                candidate = "https:" + candidate
91            if candidate.startswith("http"):
92                thumbnail_url = candidate
93                break
94 
95    if not thumbnail_url:
96        raise HTTPException(status_code=404, detail="No thumbnail found on page")
97 
98    # Try to get highest resolution by replacing quality parameters
99    # YouTube specific: replace hqdefault with maxresdefault
100    if "hqdefault" in thumbnail_url:
101        high_res_url = thumbnail_url.replace("hqdefault", "maxresdefault")
102    elif "sddefault" in thumbnail_url:
103        high_res_url = thumbnail_url.replace("sddefault", "maxresdefault")
104    else:
105        high_res_url = thumbnail_url
106 
107    # For YouTube, also try different resolutions
108    if "ytimg.com" in high_res_url:
109        # Try maxresdefault first
110        test_url = high_res_url
111        try:
112            test_resp = requests.head(test_url, timeout=5)
113            if test_resp.status_code != 200:
114                # Fall back to hqdefault
115                test_url = thumbnail_url
116        except:
117            test_url = thumbnail_url
118        high_res_url = test_url
119 
120    thumbnail_data = {
121        "id": thumbnail_id_counter,
122        "video_page_url": video_url,
123        "thumbnail_url": high_res_url,
124        "user": user
125    }
126    thumbnails[thumbnail_id_counter] = thumbnail_data
127    thumbnail_id_counter += 1
128 
129    return thumbnail_data
130 
131@app.get("/thumbnail/{id}")
132def get_thumbnail_by_id(id: int, authorization: str = Header(None)):
133    user = get_user_from_token(authorization)
134    if id not in thumbnails:
135        raise HTTPException(status_code=404, detail="Thumbnail not found")
136    return thumbnails[id]

1fastapi
2uvicorn
3requests