glchat-plugin 0.2.11__py3-none-any.whl → 0.3.0__py3-none-any.whl

glchat_plugin/config/constant.py

@@ -62,3 +62,21 @@ class SearchType(StrEnum):
             stacklevel=2,
         )
         return cls._DEEP_RESEARCH
+
+
+class ReferenceFormatterType(StrEnum):
+    """The type of reference formatter to use.
+
+    Attributes:
+        SIMILARITY: Use similarity based reference formatter.
+        LM: Use LM based reference formatter.
+        NONE: No reference formatter is used.
+    """
+
+    SIMILARITY = "similarity"
+    LM = "lm"
+    NONE = "none"
+
+
+WEB_SEARCH_BLACKLIST_DEFAULT = "[]"
+WEB_SEARCH_WHITELIST_DEFAULT = "[]"

glchat_plugin/pipeline/base_pipeline_preset_config.py

@@ -9,9 +9,14 @@ References:
 
 from typing import Any
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
-from glchat_plugin.config.constant import SearchType
+from glchat_plugin.config.constant import (
+    WEB_SEARCH_BLACKLIST_DEFAULT,
+    WEB_SEARCH_WHITELIST_DEFAULT,
+    ReferenceFormatterType,
+    SearchType,
+)
 
 
 class BasePipelinePresetConfig(BaseModel):
@@ -25,6 +30,27 @@ class BasePipelinePresetConfig(BaseModel):
         support_multimodal (bool): Whether the pipeline supports multimodal.
         use_docproc (bool): Whether to use the document processor.
         search_types (list[SearchType]): The supported search types.
+        anonymize_em (bool): Whether to anonymize before using the embedding model.
+        anonymize_lm (bool): Whether to anonymize before using the language model.
+        augment_context (bool): Whether context augmentation from the knowledge base is allowed.
+        chat_history_limit (int): The chat history limit. If the value is negative, no limit will be applied.
+        enable_guardrails (bool): Whether to enable guardrails.
+        enable_smart_search_integration (bool): Whether to enable smart search integration.
+        normal_search_top_k (int): The top k for normal search. Must be greater than or equal to 1.
+        prompt_context_char_threshold (int): The character limit above which the prompt is assumed
+            to have contained the context.
+        reference_formatter_batch_size (int): The reference formatter batch size.
+        reference_formatter_threshold (float): The reference formatter threshold.
+        reference_formatter_type (ReferenceFormatterType): The reference formatter type.
+        rerank_kwargs (str): The rerank kwargs.
+        rerank_type (str): The rerank type.
+        smart_search_top_k (int): The top k for smart search. Must be greater than or equal to 1.
+        use_cache (bool): Whether to use cache.
+        use_model_knowledge (bool): Whether to use model knowledge.
+        vector_weight (float): The vector weight. Must be between 0 and 1 (inclusive).
+        web_search_blacklist (str): The web search blacklist.
+        web_search_top_k (int): The top k for web search. Must be greater than or equal to 1.
+        web_search_whitelist (str): The web search whitelist.
     """
 
     pipeline_preset_id: str
@@ -34,3 +60,23 @@ class BasePipelinePresetConfig(BaseModel):
     support_multimodal: bool
     use_docproc: bool
     search_types: list[SearchType]
+    anonymize_em: bool
+    anonymize_lm: bool
+    augment_context: bool
+    chat_history_limit: int
+    enable_guardrails: bool = False
+    enable_smart_search_integration: bool = False
+    normal_search_top_k: int = Field(ge=1)
+    prompt_context_char_threshold: int = 32000
+    reference_formatter_batch_size: int = Field(ge=1)
+    reference_formatter_threshold: float = Field(ge=0, le=1)
+    reference_formatter_type: ReferenceFormatterType
+    rerank_kwargs: str = "{}"
+    rerank_type: str = ""
+    smart_search_top_k: int = Field(ge=1)
+    use_cache: bool
+    use_model_knowledge: bool
+    vector_weight: float = Field(ge=0, le=1)
+    web_search_blacklist: str = Field(default=WEB_SEARCH_BLACKLIST_DEFAULT)
+    web_search_top_k: int = Field(ge=1)
+    web_search_whitelist: str = Field(default=WEB_SEARCH_WHITELIST_DEFAULT)

glchat_plugin/pipeline/pipeline_handler.py

@@ -66,6 +66,9 @@ class ChatbotPresetMapping(BaseModel):
 logger = LoggerManager().get_logger(__name__)
 
 
+INTERNAL_PIPELINES = ["preprocessing", "postprocessing"]
+
+
 class PipelineHandler(PluginHandler):
     """Handler for pipeline builder plugins.
 
@@ -186,6 +189,25 @@ class PipelineHandler(PluginHandler):
         plugin.lmrp_catalogs = instance._chatbot_configs[chatbot_id].lmrp_catalogs
         instance._builders[chatbot_id] = plugin
 
+        for pipeline_type in INTERNAL_PIPELINES:
+            internal_plugin = instance._plugins.get(pipeline_type)
+            if internal_plugin:
+                try:
+                    internal_plugin.prompt_builder_catalogs = instance._chatbot_configs[
+                        chatbot_id
+                    ].prompt_builder_catalogs
+                    internal_plugin.lmrp_catalogs = instance._chatbot_configs[chatbot_id].lmrp_catalogs
+                    pipeline_config = instance._chatbot_configs[chatbot_id].pipeline_config.copy()
+                    pipeline = await internal_plugin.build(pipeline_config)
+                    pipeline_key = (chatbot_id, f"__{pipeline_type}__")
+                    instance._chatbot_pipeline_keys.setdefault(chatbot_id, set()).add(pipeline_key)
+                    instance._pipeline_cache[pipeline_key] = pipeline
+                except Exception as e:
+                    logger.warning(f"Failed when ainit plugin {traceback.format_exc()}")
+                    logger.warning(
+                        f"Error building internal pipeline `{pipeline_type}` for chatbot `{chatbot_id}`: {e}"
+                    )
+
         for model in supported_models:
             try:
                 model_id = model.get("model_id", model.get("name"))
@@ -206,7 +228,7 @@ class PipelineHandler(PluginHandler):
                 instance._chatbot_pipeline_keys.setdefault(chatbot_id, set()).add(pipeline_key)
                 instance._pipeline_cache[pipeline_key] = pipeline
             except Exception as e:
-                logger.warning(f"Failed when ainit pliugin {traceback.format_exc()}")
+                logger.warning(f"Failed when ainit plugin {traceback.format_exc()}")
                 logger.warning(f"Error building pipeline for chatbot `{chatbot_id}` model `{model_id}`: {e}")
 
     def get_pipeline_builder(self, chatbot_id: str) -> Plugin:

glchat_plugin/pipeline/pipeline_plugin.py

@@ -31,7 +31,8 @@ class PipelineBuilderPlugin(Plugin, Generic[PipelineState, PipelinePresetConfig]
         name (str): The name of the plugin.
         description (str): The description of the plugin.
         version (str): The version of the plugin.
-        catalog (BaseCatalog): The catalog instance.
+        lmrp_catalogs (dict[str, BaseCatalog[Any]] | None): The LM request processor catalogs.
+        prompt_builder_catalogs (dict[str, BaseCatalog[Any]] | None): The prompt builder catalogs.
         additional_config_class (Type[PipelineRuntimeConfig] | None): The additional runtime configuration class.
         preset_config_class (Type[PipelinePresetConfig] | None): The preset configuration class.
     """
@@ -40,7 +41,8 @@ class PipelineBuilderPlugin(Plugin, Generic[PipelineState, PipelinePresetConfig]
     description: str = "Pipeline builder plugin"
     version: str = "0.0.0"
 
-    catalog: BaseCatalog[Any]
+    lmrp_catalogs: dict[str, BaseCatalog[Any]] | None = None
+    prompt_builder_catalogs: dict[str, BaseCatalog[Any]] | None = None
 
     additional_config_class: Type[PipelineRuntimeConfig] | None = None
     preset_config_class: Type[PipelinePresetConfig] | None = None
@@ -64,6 +66,7 @@ class PipelineBuilderPlugin(Plugin, Generic[PipelineState, PipelinePresetConfig]
         self,
         request_config: dict[str, Any],
         pipeline_config: dict[str, Any],
+        previous_state: dict[str, Any] | None = None,
         **kwargs: Any,
     ) -> PipelineState:
         """Build the initial pipeline state.
@@ -71,6 +74,7 @@ class PipelineBuilderPlugin(Plugin, Generic[PipelineState, PipelinePresetConfig]
         Args:
             request_config (dict[str, Any]): Request configuration.
             pipeline_config (dict[str, Any]): Pipeline configuration.
+            previous_state (dict[str, Any] | None): Previous state.
             kwargs (Any): Additional state arguments.
 
         Returns:

{glchat_plugin-0.2.11.dist-info → glchat_plugin-0.3.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: glchat-plugin
-Version: 0.2.11
+Version: 0.3.0
 Author-email: GenAI SDK Team <gat-sdk@gdplabs.id>
 Requires-Python: <3.13,>=3.11
 Description-Content-Type: text/markdown

{glchat_plugin-0.2.11.dist-info → glchat_plugin-0.3.0.dist-info}/RECORD

@@ -1,15 +1,15 @@
 glchat_plugin/__init__.py,sha256=SHSBMz7JDU6MecyIrhHu5-3NVs89JkXhyvD3ZGOkWOE,37
 glchat_plugin/config/__init__.py,sha256=DNnX8B_TvAN89oyMgq32zG1DeaezODrihiAXTwOPT5o,39
 glchat_plugin/config/app_config.py,sha256=9_ShYtaQ7Rp14sSkrIFLoOAMlbwVlm13EuCxzOn4NCI,426
-glchat_plugin/config/constant.py,sha256=D8BCSkwL7AbXUT_zdsXEdLmmuSlLThSJn7-02niZeXI,2340
+glchat_plugin/config/constant.py,sha256=iRiY-wN7dhIHmxl8zzmYtOtE9B-oBFadThqALcnSoEE,2744
 glchat_plugin/context/__init__.py,sha256=3Wx_apMIS6z-m6eRs6hoyOsJFLJfKmMFOkrPDkPQfJI,40
 glchat_plugin/context/context_manager.py,sha256=0lhO0w_hd5dUdIEJQ2LOJFZsgpzitQU_aPZfTfQK3vw,1302
 glchat_plugin/handler/__init__.py,sha256=H5DJaAfwwtRsvMcOaEzHfGMQk25H7la0E7uPfksWtoQ,40
 glchat_plugin/handler/base_post_login_handler.py,sha256=48xSbe_LwTCjRY-lCuzWXqbnEr1ql8bAhQih1Xeh8f8,2835
 glchat_plugin/pipeline/__init__.py,sha256=Sk-NfIGyA9VKIg0Bt5OHatNUYyWVPh9i5xhE5DFAfbo,41
-glchat_plugin/pipeline/base_pipeline_preset_config.py,sha256=2DxJlroZ_tvVO6DYo1R07cjhWgKQJpysw1GsJ3BNRnw,1100
-glchat_plugin/pipeline/pipeline_handler.py,sha256=vDvIPI7Y39K_DI3jvQZsXC9YzQLUI4TuC-boyasdey8,24405
-glchat_plugin/pipeline/pipeline_plugin.py,sha256=1i7w7WepM5gON4Kx10SjL7NV7-SYVAmRE1EW2m_dOYg,4028
+glchat_plugin/pipeline/base_pipeline_preset_config.py,sha256=F-u4WzMuNlzEmKE-xgeAG-mI8TB6d4JLTsjnh_89-Y0,3677
+glchat_plugin/pipeline/pipeline_handler.py,sha256=aCRvhS6Dkhmqsx_Ya-2t2PbMseacw1VI6PUEOQq0RsM,25620
+glchat_plugin/pipeline/pipeline_plugin.py,sha256=fozvxVrOphgwLIF7uPrEkF8ZQcu8xgifYAQyuxj9628,4393
 glchat_plugin/service/__init__.py,sha256=9T4qzyYL052qLqva5el1F575OTRNaaf9tb9UvW-leTc,47
 glchat_plugin/service/base_rate_limiter_service.py,sha256=tgKwdr4EqnGo5iDRVJPnlg8W9q0hiUzfeewAtdW4IjU,1232
 glchat_plugin/service/base_tenant_service.py,sha256=CB-jJWXi87nTcjO1XhPoSgNMf2YA_LfXoHr30ye0VtI,734
@@ -19,7 +19,7 @@ glchat_plugin/storage/base_anonymizer_storage.py,sha256=oFwovWrsjM7v1YjeN-4p-M3O
 glchat_plugin/storage/base_chat_history_storage.py,sha256=JvUUFMu_9jRBQ9yug_x7S4rQjZEA1vM5ombDvz-7zCE,11095
 glchat_plugin/tools/__init__.py,sha256=OFotHbgQ8mZEbdlvlv5aVMdxfubPvkVWAcTwhIPdIqQ,542
 glchat_plugin/tools/decorators.py,sha256=AvQBV18wzXWdC483RSSmpfh92zsqTyp8SzDLIkreIGU,3925
-glchat_plugin-0.2.11.dist-info/METADATA,sha256=2rPdxeHViAjDibsyvbElHc-Lli63qsPIzUotDFFdRB4,2064
-glchat_plugin-0.2.11.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-glchat_plugin-0.2.11.dist-info/top_level.txt,sha256=fzKSXmct5dY4CAKku4-mkdHX-QPAyQVvo8vpQj8qizY,14
-glchat_plugin-0.2.11.dist-info/RECORD,,
+glchat_plugin-0.3.0.dist-info/METADATA,sha256=hl5d1CqDpK27hGQCDKYX8Uw2JlWWCshBr9rphZRvDsE,2063
+glchat_plugin-0.3.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+glchat_plugin-0.3.0.dist-info/top_level.txt,sha256=fzKSXmct5dY4CAKku4-mkdHX-QPAyQVvo8vpQj8qizY,14
+glchat_plugin-0.3.0.dist-info/RECORD,,