feat: add tongyi model bury point header (langgenius#2241)

* feat: add tongyi model bury point header

* feat: add tongyi model bury point header

* feat: add tongyi model bury point header, update version

* feat: add tongyi model bury point header, update version

* feat: add tongyi model bury point header, update version

* feat: add tongyi model bury point header, update version

---------

Co-authored-by: zhaojiangang-it <zjg02419448@alibaba-inc.com>

Author: zhaojiangang-it

models/tongyi/PRIVACY.md

@@ -0,0 +1,3 @@
+## Privacy
+
+For full terms and privacy policy of tongyi, please visit https://help.aliyun.com/zh/model-studio/privacy-notice

models/tongyi/manifest.yaml

@@ -25,5 +25,5 @@ resource:
     model:
       enabled: false
 type: plugin
-version: 0.1.8
+version: 0.1.9
 created_at: "2024-12-10T16:13:50.29298939+08:00"

models/tongyi/models/constant.py

@@ -0,0 +1,4 @@
+# https://help.aliyun.com/zh/marketplace/partner-driven-large-model-tokens-consumption-buried-point-proposal?spm=a2c4g.11186623.help-menu-30488.d_2_2_0.53a66065QheeNw&scm=20140722.H_2927986._.OR_help-T_cn~zh-V_1
+# bury point header for aliyun
+BURY_POINT_HEADER = {
+    'x-dashscope-euid': '{"bizType":"B2B", "moduleType":"Third-partyproducts", "moduleCode":"market_91320506MACC8K9EXF", "accountType":"Aliyun", "accountId":""}'}

models/tongyi/models/llm/llm.py

@@ -60,6 +60,7 @@
 )
 from dify_plugin.interfaces.model.large_language_model import LargeLanguageModel
 from openai import OpenAI
+from ..constant import BURY_POINT_HEADER
 
 logger = logging.getLogger(__name__)
 
@@ -247,13 +248,18 @@ def _generate(
             params["messages"] = self._convert_prompt_messages_to_tongyi_messages(
                 credentials, prompt_messages, rich_content=True
             )
-            response = MultiModalConversation.call(**params, stream=stream, incremental_output=incremental_output)
+            response = MultiModalConversation.call(
+                **params,
+                stream=stream,
+                headers=BURY_POINT_HEADER,
+                incremental_output=incremental_output)
         else:
             params["messages"] = self._convert_prompt_messages_to_tongyi_messages(
                 credentials, prompt_messages
             )
             response = Generation.call(
                 **params,
+                headers=BURY_POINT_HEADER,
                 result_format="message",
                 stream=stream,
                 incremental_output=incremental_output,

models/tongyi/models/rerank/rerank.py

@@ -19,7 +19,7 @@
     InvokeServerUnavailableError,
 )
 from dify_plugin.interfaces.model.rerank_model import RerankModel
-
+from ..constant import BURY_POINT_HEADER
 
 class GTERerankModel(RerankModel):
     """
@@ -54,7 +54,7 @@ def _invoke(
             dashscope.base_http_api_url = "https://dashscope-intl.aliyuncs.com/api/v1"
         dashscope.api_key = credentials["dashscope_api_key"]
         response = dashscope.TextReRank.call(
-            query=query, documents=docs, model=model, top_n=top_n, return_documents=True
+            query=query, headers=BURY_POINT_HEADER, documents=docs, model=model, top_n=top_n, return_documents=True
         )
         rerank_documents = []
         if not response.output:

models/tongyi/models/speech2text/speech2text.py

@@ -6,6 +6,7 @@
 import dashscope
 from dify_plugin import OAICompatSpeech2TextModel
 from dashscope.audio.asr import *
+from ..constant import BURY_POINT_HEADER
 
 class TongyiSpeech2TextModel(OAICompatSpeech2TextModel):
     """
@@ -41,7 +42,7 @@ def _invoke(self, model: str, credentials: dict, file: IO[bytes], user: Optional
                 sample_rate=int(sample_rate),
                 callback=None,
             )
-            result = recognition.call(file_path)
+            result = recognition.call(file=file_path, headers=BURY_POINT_HEADER)
             sentence_list = result.get_sentence()
             if sentence_list is None:
                 return ''

models/tongyi/models/text_embedding/text_embedding.py

@@ -7,7 +7,7 @@
 from dify_plugin.errors.model import CredentialsValidateFailedError
 from dify_plugin.interfaces.model.text_embedding_model import TextEmbeddingModel
 from models._common import _CommonTongyi
-
+from ..constant import BURY_POINT_HEADER
 
 class TongyiTextEmbeddingModel(_CommonTongyi, TextEmbeddingModel):
     """
@@ -109,7 +109,8 @@ def call_embedding_api(text):
                 return dashscope.TextEmbedding.call(
                     api_key=credentials_kwargs["dashscope_api_key"], 
                     model=model, 
-                    input=text, 
+                    input=text,
+                    headers=BURY_POINT_HEADER,
                     text_type="document"
                 )
             except Exception as e:
@@ -119,7 +120,6 @@ def call_embedding_api(text):
         for text in texts:
             # First attempt
             response = call_embedding_api(text)
-            
             # Handle rate limit error (429)
             # Check if response is an exception with rate limit info
             if hasattr(response, 'status_code') and response.status_code == 429:

models/tongyi/models/tts/tts.py

@@ -8,7 +8,7 @@
 from dify_plugin.errors.model import CredentialsValidateFailedError, InvokeBadRequestError
 from dify_plugin.interfaces.model.tts_model import TTSModel
 from models._common import _CommonTongyi
-
+from ..constant import BURY_POINT_HEADER
 
 class TongyiText2SpeechModel(_CommonTongyi, TTSModel):
     """
@@ -117,6 +117,7 @@ def _process_sentence(sentence: str, credentials: dict, voice: str, audio_type:
             sample_rate=48000,
             api_key=credentials.get("dashscope_api_key"),
             text=sentence.strip(),
+            headers=BURY_POINT_HEADER,
             format=audio_type,
         )
         if isinstance(response.get_audio_data(), bytes):