Update app.py
Browse files
app.py
CHANGED
|
@@ -17,9 +17,15 @@ async def fetch_url(url, session, max_retries=3, timeout=180):
|
|
| 17 |
try:
|
| 18 |
async with session.get(url, timeout=ClientTimeout(total=timeout)) as response:
|
| 19 |
response.raise_for_status()
|
| 20 |
-
|
|
|
|
| 21 |
except asyncio.TimeoutError:
|
| 22 |
print(f"Attempt {attempt + 1} timed out after {timeout} seconds", flush=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
except aiohttp.ClientError as e:
|
| 24 |
print(f"Attempt {attempt + 1} failed: {str(e)}", flush=True)
|
| 25 |
|
|
@@ -29,8 +35,8 @@ async def fetch_url(url, session, max_retries=3, timeout=180):
|
|
| 29 |
raise Exception(f"Failed to fetch URL after {max_retries} attempts")
|
| 30 |
|
| 31 |
async def extract_and_transform_proxies(input_text):
|
| 32 |
-
print("Original input data:")
|
| 33 |
-
print(input_text)
|
| 34 |
print("------------------------")
|
| 35 |
|
| 36 |
try:
|
|
@@ -47,15 +53,19 @@ async def extract_and_transform_proxies(input_text):
|
|
| 47 |
proxies_text = proxies_match.group(1)
|
| 48 |
proxies_list = yaml.safe_load(proxies_text)
|
| 49 |
else:
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
| 51 |
except yaml.YAMLError as e:
|
| 52 |
return f"YAML解析错误: {str(e)}"
|
| 53 |
|
| 54 |
if not proxies_list:
|
| 55 |
return "未找到有效的代理配置"
|
| 56 |
|
| 57 |
-
print("
|
| 58 |
-
print(
|
|
|
|
| 59 |
print("------------------------")
|
| 60 |
|
| 61 |
transformed_proxies = []
|
|
|
|
| 17 |
try:
|
| 18 |
async with session.get(url, timeout=ClientTimeout(total=timeout)) as response:
|
| 19 |
response.raise_for_status()
|
| 20 |
+
content = await response.read()
|
| 21 |
+
return content.decode('utf-8', errors='ignore')
|
| 22 |
except asyncio.TimeoutError:
|
| 23 |
print(f"Attempt {attempt + 1} timed out after {timeout} seconds", flush=True)
|
| 24 |
+
except aiohttp.ClientPayloadError as e:
|
| 25 |
+
print(f"Payload error on attempt {attempt + 1}: {str(e)}", flush=True)
|
| 26 |
+
if response.content_length and len(content) >= response.content_length:
|
| 27 |
+
print(f"Received data length: {len(content)}, expected: {response.content_length}", flush=True)
|
| 28 |
+
return content.decode('utf-8', errors='ignore')
|
| 29 |
except aiohttp.ClientError as e:
|
| 30 |
print(f"Attempt {attempt + 1} failed: {str(e)}", flush=True)
|
| 31 |
|
|
|
|
| 35 |
raise Exception(f"Failed to fetch URL after {max_retries} attempts")
|
| 36 |
|
| 37 |
async def extract_and_transform_proxies(input_text):
|
| 38 |
+
print("Original input data (first 1000 characters):")
|
| 39 |
+
print(input_text[:1000])
|
| 40 |
print("------------------------")
|
| 41 |
|
| 42 |
try:
|
|
|
|
| 53 |
proxies_text = proxies_match.group(1)
|
| 54 |
proxies_list = yaml.safe_load(proxies_text)
|
| 55 |
else:
|
| 56 |
+
# 如果仍然找不到,尝试提取所有看起来像代理配置的部分
|
| 57 |
+
proxy_pattern = r'{[^}]+}'
|
| 58 |
+
possible_proxies = re.findall(proxy_pattern, input_text)
|
| 59 |
+
proxies_list = [yaml.safe_load(proxy) for proxy in possible_proxies]
|
| 60 |
except yaml.YAMLError as e:
|
| 61 |
return f"YAML解析错误: {str(e)}"
|
| 62 |
|
| 63 |
if not proxies_list:
|
| 64 |
return "未找到有效的代理配置"
|
| 65 |
|
| 66 |
+
print(f"Found {len(proxies_list)} possible proxy configurations")
|
| 67 |
+
print("Sample of parsed proxies list:")
|
| 68 |
+
print(proxies_list[:5]) # 只打印前5个代理配置
|
| 69 |
print("------------------------")
|
| 70 |
|
| 71 |
transformed_proxies = []
|