Spaces:
Running
Running
Update translator.py
Browse files- translator.py +98 -110
translator.py
CHANGED
|
@@ -1,48 +1,22 @@
|
|
| 1 |
"""
|
| 2 |
translator.py
|
| 3 |
-
腾讯云 + 百度翻译 API 轮询封装
|
| 4 |
-
⚠️ 需在 HF 空间的 “Variables” 页设置以下环境变量
|
| 5 |
-
------------------------------------------------------------------
|
| 6 |
-
TENCENT_SECRET_ID 腾讯云 SecretId
|
| 7 |
-
TENCENT_SECRET_KEY 腾讯云 SecretKey
|
| 8 |
-
TENCENT_TRANSLATE_URL (可选) 默认 https://tmt.tencentcloudapi.com
|
| 9 |
-
BAIDU_TRANSLATE_URL (可选) 默认 https://fanyi-api.baidu.com/api/trans/vip/translate
|
| 10 |
-
BAIDU_CREDENTIALS_JSON 形如:
|
| 11 |
-
[
|
| 12 |
-
{"app_id": "xxxx", "secret_key": "yyyy"},
|
| 13 |
-
{"app_id": "aaaa", "secret_key": "bbbb"}
|
| 14 |
-
]
|
| 15 |
-
------------------------------------------------------------------
|
| 16 |
"""
|
| 17 |
import hashlib, hmac, json, os, random, time
|
| 18 |
from datetime import datetime
|
| 19 |
-
from typing import List, Sequence, Optional
|
| 20 |
|
| 21 |
import requests
|
| 22 |
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
# 内部轮询索引
|
| 34 |
-
_baidu_idx: int = 0
|
| 35 |
-
def _next_baidu_cred():
|
| 36 |
-
global _baidu_idx
|
| 37 |
-
if not BAIDU_CREDENTIALS:
|
| 38 |
-
return None
|
| 39 |
-
cred = BAIDU_CREDENTIALS[_baidu_idx]
|
| 40 |
-
_baidu_idx = (_baidu_idx + 1) % len(BAIDU_CREDENTIALS)
|
| 41 |
-
return cred
|
| 42 |
|
| 43 |
-
# ------------------------------------------------------------------
|
| 44 |
-
# 腾讯翻译
|
| 45 |
-
# ------------------------------------------------------------------
|
| 46 |
def _sign(key: bytes, msg: str) -> bytes:
|
| 47 |
return hmac.new(key, msg.encode("utf-8"), hashlib.sha256).digest()
|
| 48 |
|
|
@@ -52,109 +26,123 @@ def _tc3_signature(secret_key: str, date: str, service: str, string_to_sign: str
|
|
| 52 |
secret_signing = _sign(secret_service, "tc3_request")
|
| 53 |
return hmac.new(secret_signing, string_to_sign.encode("utf-8"), hashlib.sha256).hexdigest()
|
| 54 |
|
| 55 |
-
def
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
|
|
|
|
|
|
| 66 |
algorithm = "TC3-HMAC-SHA256"
|
| 67 |
|
| 68 |
-
payload = {
|
| 69 |
-
"SourceText": "\n".join(texts),
|
| 70 |
-
"Source": src,
|
| 71 |
-
"Target": tgt,
|
| 72 |
-
"ProjectId": 0,
|
| 73 |
-
}
|
| 74 |
payload_str = json.dumps(payload, ensure_ascii=False)
|
| 75 |
|
| 76 |
-
# ---------- step‑1 canonical request ----------
|
| 77 |
canonical_request = "\n".join([
|
| 78 |
-
"POST",
|
| 79 |
-
"
|
| 80 |
-
"",
|
| 81 |
-
f"content-type:application/json; charset=utf-8\nhost:{host}\nx-tc-action:{action.lower()}\n",
|
| 82 |
-
"content-type;host;x-tc-action",
|
| 83 |
-
hashlib.sha256(payload_str.encode()).hexdigest(),
|
| 84 |
])
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
string_to_sign = "\n".join([
|
| 89 |
-
algorithm, str(ts), credential_scope,
|
| 90 |
-
hashlib.sha256(canonical_request.encode()).hexdigest(),
|
| 91 |
])
|
| 92 |
-
|
| 93 |
-
# ---------- step‑3 signature ----------
|
| 94 |
-
signature = _tc3_signature(TENCENT_SECRET_KEY, date, service, string_to_sign)
|
| 95 |
-
|
| 96 |
-
# ---------- step‑4 headers ----------
|
| 97 |
authorization = (
|
| 98 |
-
f"{algorithm} Credential={
|
| 99 |
f"SignedHeaders=content-type;host;x-tc-action, Signature={signature}"
|
| 100 |
)
|
| 101 |
headers = {
|
| 102 |
-
"Authorization":
|
| 103 |
-
"
|
| 104 |
-
"Host": host,
|
| 105 |
-
"X-TC-Action": action,
|
| 106 |
-
"X-TC-Timestamp": str(ts),
|
| 107 |
-
"X-TC-Version": version,
|
| 108 |
-
"X-TC-Region": region,
|
| 109 |
}
|
| 110 |
|
| 111 |
-
# ---------- request ----------
|
| 112 |
try:
|
| 113 |
-
resp = requests.post(
|
| 114 |
resp.raise_for_status()
|
| 115 |
data = resp.json()
|
| 116 |
return data["Response"]["TargetText"].split("\n")
|
| 117 |
except Exception as e:
|
| 118 |
-
print(f"[translator] Tencent API error → {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 119 |
return None
|
| 120 |
|
| 121 |
-
# ------------------------------------------------------------------
|
| 122 |
-
# 百度翻译
|
| 123 |
-
# ------------------------------------------------------------------
|
| 124 |
-
def _translate_with_baidu(texts: Sequence[str], src="auto", tgt="zh") -> Optional[List[str]]:
|
| 125 |
-
creds = _next_baidu_cred()
|
| 126 |
-
if creds is None:
|
| 127 |
-
return None # 未配置凭证
|
| 128 |
-
app_id, secret_key = creds["app_id"], creds["secret_key"]
|
| 129 |
salt = random.randint(32768, 65536)
|
| 130 |
query = "\n".join(texts)
|
| 131 |
sign = hashlib.md5((app_id + query + str(salt) + secret_key).encode()).hexdigest()
|
| 132 |
-
params = {
|
| 133 |
-
"q": query, "from": src, "to": tgt,
|
| 134 |
-
"appid": app_id, "salt": salt, "sign": sign,
|
| 135 |
-
}
|
| 136 |
try:
|
| 137 |
-
resp = requests.get(
|
| 138 |
resp.raise_for_status()
|
| 139 |
data = resp.json()
|
| 140 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 141 |
except Exception as e:
|
| 142 |
-
print(f"[translator] Baidu API error → {e}")
|
| 143 |
return None
|
| 144 |
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
""
|
| 152 |
-
|
| 153 |
-
"""
|
| 154 |
if not texts:
|
| 155 |
return []
|
| 156 |
|
| 157 |
-
|
| 158 |
-
|
| 159 |
-
out
|
| 160 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
"""
|
| 2 |
translator.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
"""
|
| 4 |
import hashlib, hmac, json, os, random, time
|
| 5 |
from datetime import datetime
|
| 6 |
+
from typing import List, Sequence, Optional, Dict, Any
|
| 7 |
|
| 8 |
import requests
|
| 9 |
|
| 10 |
+
DEFAULT_TENCENT_SECRET_ID = os.environ.get("TENCENT_SECRET_ID")
|
| 11 |
+
DEFAULT_TENCENT_SECRET_KEY = os.environ.get("TENCENT_SECRET_KEY")
|
| 12 |
+
DEFAULT_TENCENT_URL = os.environ.get("TENCENT_TRANSLATE_URL", "https://tmt.tencentcloudapi.com")
|
| 13 |
+
|
| 14 |
+
DEFAULT_BAIDU_URL = os.environ.get("BAIDU_TRANSLATE_URL", "https://fanyi-api.baidu.com/api/trans/vip/translate")
|
| 15 |
+
try:
|
| 16 |
+
DEFAULT_BAIDU_CREDENTIALS = json.loads(os.environ.get("BAIDU_CREDENTIALS_JSON", "[]"))
|
| 17 |
+
except json.JSONDecodeError:
|
| 18 |
+
DEFAULT_BAIDU_CREDENTIALS = []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
|
|
|
|
|
|
|
|
|
| 20 |
def _sign(key: bytes, msg: str) -> bytes:
|
| 21 |
return hmac.new(key, msg.encode("utf-8"), hashlib.sha256).digest()
|
| 22 |
|
|
|
|
| 26 |
secret_signing = _sign(secret_service, "tc3_request")
|
| 27 |
return hmac.new(secret_signing, string_to_sign.encode("utf-8"), hashlib.sha256).hexdigest()
|
| 28 |
|
| 29 |
+
def _translate_with_tencent_dynamic(
|
| 30 |
+
texts: Sequence[str],
|
| 31 |
+
secret_id: str,
|
| 32 |
+
secret_key: str,
|
| 33 |
+
url: str,
|
| 34 |
+
src="auto",
|
| 35 |
+
tgt="zh"
|
| 36 |
+
) -> Optional[List[str]]:
|
| 37 |
+
if not (secret_id and secret_key):
|
| 38 |
+
return None
|
| 39 |
+
service, host, action, version, region = "tmt", "tmt.tencentcloudapi.com", "TextTranslate", "2018-03-21", "ap-beijing"
|
| 40 |
+
ts = int(time.time())
|
| 41 |
+
date = datetime.utcfromtimestamp(ts).strftime("%Y-%m-%d")
|
| 42 |
algorithm = "TC3-HMAC-SHA256"
|
| 43 |
|
| 44 |
+
payload = {"SourceText": "\n".join(texts), "Source": src, "Target": tgt, "ProjectId": 0}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
payload_str = json.dumps(payload, ensure_ascii=False)
|
| 46 |
|
|
|
|
| 47 |
canonical_request = "\n".join([
|
| 48 |
+
"POST", "/", "", f"content-type:application/json; charset=utf-8\nhost:{host}\nx-tc-action:{action.lower()}\n",
|
| 49 |
+
"content-type;host;x-tc-action", hashlib.sha256(payload_str.encode()).hexdigest(),
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
])
|
| 51 |
+
credential_scope = f"{date}/{service}/tc3_request"
|
| 52 |
+
string_to_sign = "\n".join([
|
| 53 |
+
algorithm, str(ts), credential_scope, hashlib.sha256(canonical_request.encode()).hexdigest(),
|
|
|
|
|
|
|
|
|
|
| 54 |
])
|
| 55 |
+
signature = _tc3_signature(secret_key, date, service, string_to_sign)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
authorization = (
|
| 57 |
+
f"{algorithm} Credential={secret_id}/{credential_scope}, "
|
| 58 |
f"SignedHeaders=content-type;host;x-tc-action, Signature={signature}"
|
| 59 |
)
|
| 60 |
headers = {
|
| 61 |
+
"Authorization": authorization, "Content-Type": "application/json; charset=utf-8", "Host": host,
|
| 62 |
+
"X-TC-Action": action, "X-TC-Timestamp": str(ts), "X-TC-Version": version, "X-TC-Region": region,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
}
|
| 64 |
|
|
|
|
| 65 |
try:
|
| 66 |
+
resp = requests.post(url, headers=headers, data=payload_str, timeout=8)
|
| 67 |
resp.raise_for_status()
|
| 68 |
data = resp.json()
|
| 69 |
return data["Response"]["TargetText"].split("\n")
|
| 70 |
except Exception as e:
|
| 71 |
+
print(f"[translator] Dynamic Tencent API error → {e}")
|
| 72 |
+
return None
|
| 73 |
+
|
| 74 |
+
def _translate_with_baidu_dynamic(
|
| 75 |
+
texts: Sequence[str],
|
| 76 |
+
baidu_creds_list: List[Dict[str, str]],
|
| 77 |
+
url: str,
|
| 78 |
+
src="auto",
|
| 79 |
+
tgt="zh"
|
| 80 |
+
) -> Optional[List[str]]:
|
| 81 |
+
if not baidu_creds_list:
|
| 82 |
+
return None
|
| 83 |
+
|
| 84 |
+
cred = random.choice(baidu_creds_list)
|
| 85 |
+
app_id, secret_key = cred.get("app_id"), cred.get("secret_key")
|
| 86 |
+
if not (app_id and secret_key):
|
| 87 |
return None
|
| 88 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
salt = random.randint(32768, 65536)
|
| 90 |
query = "\n".join(texts)
|
| 91 |
sign = hashlib.md5((app_id + query + str(salt) + secret_key).encode()).hexdigest()
|
| 92 |
+
params = {"q": query, "from": src, "to": tgt, "appid": app_id, "salt": salt, "sign": sign}
|
|
|
|
|
|
|
|
|
|
| 93 |
try:
|
| 94 |
+
resp = requests.get(url, params=params, timeout=8)
|
| 95 |
resp.raise_for_status()
|
| 96 |
data = resp.json()
|
| 97 |
+
if "trans_result" in data:
|
| 98 |
+
return [item["dst"] for item in data["trans_result"]]
|
| 99 |
+
else:
|
| 100 |
+
print(f"[translator] Baidu API returned error: {data.get('error_msg', 'Unknown error')}")
|
| 101 |
+
return None
|
| 102 |
except Exception as e:
|
| 103 |
+
print(f"[translator] Dynamic Baidu API error → {e}")
|
| 104 |
return None
|
| 105 |
|
| 106 |
+
def translate_texts_with_dynamic_keys(
|
| 107 |
+
texts: Sequence[str],
|
| 108 |
+
tencent_id: str,
|
| 109 |
+
tencent_key: str,
|
| 110 |
+
baidu_json_str: str,
|
| 111 |
+
src_lang: str = "auto",
|
| 112 |
+
tgt_lang: str = "zh"
|
| 113 |
+
) -> List[str]:
|
|
|
|
| 114 |
if not texts:
|
| 115 |
return []
|
| 116 |
|
| 117 |
+
if tencent_id and tencent_key:
|
| 118 |
+
out = _translate_with_tencent_dynamic(texts, tencent_id, tencent_key, DEFAULT_TENCENT_URL, src_lang, tgt_lang)
|
| 119 |
+
if out is not None:
|
| 120 |
+
return out
|
| 121 |
+
|
| 122 |
+
baidu_creds = []
|
| 123 |
+
if baidu_json_str:
|
| 124 |
+
try:
|
| 125 |
+
baidu_creds = json.loads(baidu_json_str)
|
| 126 |
+
if not isinstance(baidu_creds, list): baidu_creds = []
|
| 127 |
+
except json.JSONDecodeError:
|
| 128 |
+
print("[translator] Invalid Baidu JSON provided by user.")
|
| 129 |
+
baidu_creds = []
|
| 130 |
+
|
| 131 |
+
if baidu_creds:
|
| 132 |
+
out = _translate_with_baidu_dynamic(texts, baidu_creds, DEFAULT_BAIDU_URL, src_lang, tgt_lang)
|
| 133 |
+
if out is not None:
|
| 134 |
+
return out
|
| 135 |
+
|
| 136 |
+
return list(texts)
|
| 137 |
+
|
| 138 |
+
def translate_texts(texts: Sequence[str],
|
| 139 |
+
src_lang: str = "auto",
|
| 140 |
+
tgt_lang: str = "zh") -> List[str]:
|
| 141 |
+
return translate_texts_with_dynamic_keys(
|
| 142 |
+
texts,
|
| 143 |
+
DEFAULT_TENCENT_SECRET_ID,
|
| 144 |
+
DEFAULT_TENCENT_SECRET_KEY,
|
| 145 |
+
json.dumps(DEFAULT_BAIDU_CREDENTIALS), # 转换回 JSON 字符串以匹配接口
|
| 146 |
+
src_lang,
|
| 147 |
+
tgt_lang
|
| 148 |
+
)
|