aip-agents-binary 0.5.15__py3-none-macosx_13_0_arm64.whl → 0.5.17__py3-none-macosx_13_0_arm64.whl

aip_agents/tools/document_loader/__init__.py

@@ -12,6 +12,8 @@ References:
 
 try:
     from aip_agents.tools.document_loader.base_reader import (  # noqa: F401
+        DOCPROC_AVAILABLE,
+        DOCPROC_MISSING_MESSAGE,
         BaseDocumentReaderTool,
         DocumentReaderInput,
     )
@@ -28,10 +30,15 @@ try:
         "ExcelReaderTool",
         "PDFSplitter",
     ]
+    if not DOCPROC_AVAILABLE:
+        import warnings
+
+        warnings.warn(DOCPROC_MISSING_MESSAGE, ImportWarning)
 except ImportError:
     import warnings
 
     warnings.warn(
-        "Document loader tools not available. Install with: pip install aip-agents[document-loader]", ImportWarning
+        "Document loader tools not available. Install gllm-docproc to enable them.",
+        ImportWarning,
     )
     __all__ = []

aip_agents/tools/document_loader/base_reader.py

@@ -10,8 +10,8 @@ References:
 import gc
 import os
 from abc import ABC, abstractmethod
+from typing import TYPE_CHECKING, Any
 
-from gllm_docproc.loader.pipeline_loader import PipelineLoader
 from langchain_core.runnables import RunnableConfig
 from langchain_core.tools import BaseTool
 from pydantic import BaseModel, Field
@@ -21,6 +21,46 @@ from aip_agents.utils.logger import get_logger
 
 logger = get_logger(__name__)
 
+DOCPROC_MISSING_MESSAGE = (
+    "gllm-docproc is required for document loader tools but is not installed. "
+    "Install it from your internal registry to enable document processing."
+)
+
+try:
+    from gllm_docproc.loader.pipeline_loader import PipelineLoader
+except ImportError:
+    PipelineLoader = None  # type: ignore[assignment]
+
+if TYPE_CHECKING:
+    from gllm_docproc.loader.pipeline_loader import PipelineLoader as PipelineLoaderType
+else:
+    PipelineLoaderType = Any
+
+
+class _MissingDocprocLoader:
+    """Fallback loader that errors when document processing is attempted."""
+
+    def __init__(self) -> None:
+        self.loaders: list[object] = []
+
+    def add_loader(self, loader: object) -> None:
+        self.loaders.append(loader)
+
+    def load(self, *_args: object, **_kwargs: object) -> list[dict[str, str]]:
+        raise ImportError(DOCPROC_MISSING_MESSAGE)
+
+    def clear_cache(self) -> None:
+        self.loaders.clear()
+
+
+def _build_pipeline_loader() -> "PipelineLoader":
+    if PipelineLoader is None:
+        return _MissingDocprocLoader()
+    return PipelineLoader()  # type: ignore[misc]
+
+
+DOCPROC_AVAILABLE = PipelineLoader is not None
+
 
 class BaseDocumentConfig(BaseModel):
     """Base tool configuration schema for document processing with batching functionality.
@@ -78,7 +118,7 @@ class BaseDocumentReaderTool(BaseTool, ABC):
     description: str = "Read a document file and extract its text content."
     args_schema: type[BaseModel] = DocumentReaderInput
     tool_config_schema: type[BaseModel] = BaseDocumentConfig
-    loader: PipelineLoader = Field(default_factory=PipelineLoader)
+    loader: PipelineLoaderType = Field(default_factory=_build_pipeline_loader)
 
     def __init__(self):
         """Initialize the base document reader tool."""

aip_agents/tools/document_loader/base_reader.pyi

@@ -1,11 +1,24 @@
 from _typeshed import Incomplete
 from abc import ABC
 from aip_agents.utils.logger import get_logger as get_logger
-from gllm_docproc.loader.pipeline_loader import PipelineLoader
+from gllm_docproc.loader.pipeline_loader import PipelineLoader as PipelineLoaderType
 from langchain_core.tools import BaseTool
 from pydantic import BaseModel
+from typing import Any
 
 logger: Incomplete
+DOCPROC_MISSING_MESSAGE: str
+PipelineLoaderType = Any
+
+class _MissingDocprocLoader:
+    """Fallback loader that errors when document processing is attempted."""
+    loaders: list[object]
+    def __init__(self) -> None: ...
+    def add_loader(self, loader: object) -> None: ...
+    def load(self, *_args: object, **_kwargs: object) -> list[dict[str, str]]: ...
+    def clear_cache(self) -> None: ...
+
+DOCPROC_AVAILABLE: Incomplete
 
 class BaseDocumentConfig(BaseModel):
     """Base tool configuration schema for document processing with batching functionality.
@@ -50,7 +63,7 @@ class BaseDocumentReaderTool(BaseTool, ABC):
     description: str
     args_schema: type[BaseModel]
     tool_config_schema: type[BaseModel]
-    loader: PipelineLoader
+    loader: PipelineLoaderType
     def __init__(self) -> None:
         """Initialize the base document reader tool."""
     def cleanup_memory(self) -> None:

aip_agents/tools/document_loader/docx_reader_tool.py

@@ -8,9 +8,19 @@ References:
       reader/docx_reader_tool.py
 """
 
-from gllm_docproc.loader.docx import DOCX2PythonLoader, PythonDOCXTableLoader
+try:
+    from gllm_docproc.loader.docx import DOCX2PythonLoader, PythonDOCXTableLoader
+except ImportError as exc:
+    DOCX2PythonLoader = None  # type: ignore[assignment]
+    PythonDOCXTableLoader = None  # type: ignore[assignment]
+    _DOCPROC_IMPORT_ERROR: Exception | None = exc
+else:
+    _DOCPROC_IMPORT_ERROR = None
 
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool
+from aip_agents.tools.document_loader.base_reader import DOCPROC_MISSING_MESSAGE, BaseDocumentReaderTool
+from aip_agents.utils.logger import get_logger
+
+logger = get_logger(__name__)
 
 
 class DocxReaderTool(BaseDocumentReaderTool):
@@ -20,6 +30,11 @@ class DocxReaderTool(BaseDocumentReaderTool):
     description: str = "Read a Word document and extract its text content. Input should be the path to the Word file."
 
     def _setup_loader(self):
+        if DOCX2PythonLoader is None or PythonDOCXTableLoader is None:
+            logger.warning(DOCPROC_MISSING_MESSAGE)
+            if _DOCPROC_IMPORT_ERROR is not None:
+                logger.debug("gllm_docproc import failed: %s", _DOCPROC_IMPORT_ERROR)
+            return
         self.loader.add_loader(DOCX2PythonLoader())
         self.loader.add_loader(PythonDOCXTableLoader())
 

aip_agents/tools/document_loader/docx_reader_tool.pyi

@@ -1,4 +1,8 @@
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool
+from _typeshed import Incomplete
+from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool, DOCPROC_MISSING_MESSAGE as DOCPROC_MISSING_MESSAGE
+from aip_agents.utils.logger import get_logger as get_logger
+
+logger: Incomplete
 
 class DocxReaderTool(BaseDocumentReaderTool):
     """Tool to read and extract text from Word documents."""

aip_agents/tools/document_loader/excel_reader_tool.py

@@ -10,9 +10,15 @@ Authors:
 import zipfile
 from pathlib import Path
 
-from gllm_docproc.loader.xlsx import OpenpyxlLoader
-
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool
+try:
+    from gllm_docproc.loader.xlsx import OpenpyxlLoader
+except ImportError as exc:
+    OpenpyxlLoader = None  # type: ignore[assignment]
+    _DOCPROC_IMPORT_ERROR: Exception | None = exc
+else:
+    _DOCPROC_IMPORT_ERROR = None
+
+from aip_agents.tools.document_loader.base_reader import DOCPROC_MISSING_MESSAGE, BaseDocumentReaderTool
 from aip_agents.utils.logger import get_logger
 
 logger = get_logger(__name__)
@@ -50,6 +56,11 @@ class ExcelReaderTool(BaseDocumentReaderTool):
         This method initializes the OpenpyxlLoader which handles extraction
         of content from Excel files and formatting as Markdown tables.
         """
+        if OpenpyxlLoader is None:
+            logger.warning(DOCPROC_MISSING_MESSAGE)
+            if _DOCPROC_IMPORT_ERROR is not None:
+                logger.debug("gllm_docproc import failed: %s", _DOCPROC_IMPORT_ERROR)
+            return
         self.loader.add_loader(OpenpyxlLoader())
 
     def _run_with_batching(self, file_path: str, batch_size: int) -> str:

aip_agents/tools/document_loader/excel_reader_tool.pyi

@@ -1,5 +1,5 @@
 from _typeshed import Incomplete
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool
+from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool, DOCPROC_MISSING_MESSAGE as DOCPROC_MISSING_MESSAGE
 from aip_agents.utils.logger import get_logger as get_logger
 
 logger: Incomplete

aip_agents/tools/document_loader/pdf_reader_tool.py

@@ -8,9 +8,16 @@ References:
       reader/pdf_reader_tool.py
 """
 
-from gllm_docproc.loader.pdf import PDFPlumberLoader, PyMuPDFLoader
-
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool
+try:
+    from gllm_docproc.loader.pdf import PDFPlumberLoader, PyMuPDFLoader
+except ImportError as exc:
+    PDFPlumberLoader = None  # type: ignore[assignment]
+    PyMuPDFLoader = None  # type: ignore[assignment]
+    _DOCPROC_IMPORT_ERROR: Exception | None = exc
+else:
+    _DOCPROC_IMPORT_ERROR = None
+
+from aip_agents.tools.document_loader.base_reader import DOCPROC_MISSING_MESSAGE, BaseDocumentReaderTool
 from aip_agents.tools.document_loader.pdf_splitter import PDFSplitter
 from aip_agents.utils.logger import get_logger
 
@@ -24,6 +31,11 @@ class PDFReaderTool(BaseDocumentReaderTool):
     description: str = "Read a PDF file and extract its text content. Input should be the path to the PDF file."
 
     def _setup_loader(self):
+        if PDFPlumberLoader is None or PyMuPDFLoader is None:
+            logger.warning(DOCPROC_MISSING_MESSAGE)
+            if _DOCPROC_IMPORT_ERROR is not None:
+                logger.debug("gllm_docproc import failed: %s", _DOCPROC_IMPORT_ERROR)
+            return
         self.loader.add_loader(PyMuPDFLoader())
         self.loader.add_loader(PDFPlumberLoader())
 
@@ -43,7 +55,7 @@ class PDFReaderTool(BaseDocumentReaderTool):
         Raises:
             FileNotFoundError: If the input PDF file doesn't exist
             ValueError: If batch_size is invalid or PDF processing fails
-            Exception: For other unexpected errors during PDF splitting
+            RuntimeError: For other unexpected errors during PDF splitting
         """
         logger.info(f"Splitting PDF file '{file_path}' with batch_size={batch_size}")
 
@@ -64,4 +76,4 @@ class PDFReaderTool(BaseDocumentReaderTool):
             raise
         except Exception as e:
             logger.error(f"Unexpected error splitting PDF '{file_path}': {str(e)}")
-            raise Exception(f"Failed to split PDF file: {str(e)}") from e
+            raise RuntimeError(f"Failed to split PDF file: {str(e)}") from e

aip_agents/tools/document_loader/pdf_reader_tool.pyi

@@ -1,5 +1,5 @@
 from _typeshed import Incomplete
-from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool
+from aip_agents.tools.document_loader.base_reader import BaseDocumentReaderTool as BaseDocumentReaderTool, DOCPROC_MISSING_MESSAGE as DOCPROC_MISSING_MESSAGE
 from aip_agents.tools.document_loader.pdf_splitter import PDFSplitter as PDFSplitter
 from aip_agents.utils.logger import get_logger as get_logger
 

{aip_agents_binary-0.5.15.dist-info → aip_agents_binary-0.5.17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aip-agents-binary
-Version: 0.5.15
+Version: 0.5.17
 Summary: A library for managing agents in Gen AI applications.
 Author-email: Raymond Christopher <raymond.christopher@gdplabs.id>
 Requires-Python: <3.13,>=3.11
@@ -38,7 +38,6 @@ Requires-Dist: browser-use==0.5.9; extra == "browser-use-tool"
 Requires-Dist: steel-sdk>=0.7.0; extra == "browser-use-tool"
 Requires-Dist: json-repair>=0.52.3; extra == "browser-use-tool"
 Provides-Extra: document-loader
-Requires-Dist: gllm-docproc-binary[docx,pdf,xlsx]<0.8.0,>=0.7.20; extra == "document-loader"
 Requires-Dist: PyPDF2<4.0.0,>=3.0.0; extra == "document-loader"
 Requires-Dist: unidecode<2.0.0,>=1.3.0; extra == "document-loader"
 Provides-Extra: gl-connector
@@ -50,9 +49,8 @@ Requires-Dist: steel-sdk>=0.7.0; extra == "local"
 Requires-Dist: json-repair>=0.52.3; extra == "local"
 Requires-Dist: PyPDF2<4.0.0,>=3.0.0; extra == "local"
 Requires-Dist: unidecode<2.0.0,>=1.3.0; extra == "local"
+Requires-Dist: gllm-docproc-binary<0.8.0,>=0.7.20; extra == "local"
 Requires-Dist: bosa-connectors-binary<0.4.0,>=0.3.1; extra == "local"
-Requires-Dist: gllm-docproc-binary[docx,pdf,xlsx]<0.8.0,>=0.7.20; extra == "local"
-Requires-Dist: gllm-datastore-binary[chroma,elasticsearch]==0.5.40; extra == "local"
 Provides-Extra: dev
 Requires-Dist: coverage<8.0.0,>=7.4.4; extra == "dev"
 Requires-Dist: mypy<2.0.0,>=1.15.0; extra == "dev"

{aip_agents_binary-0.5.15.dist-info → aip_agents_binary-0.5.17.dist-info}/RECORD

@@ -414,16 +414,16 @@ aip_agents/tools/code_sandbox/e2b_cloud_sandbox_extended.py,sha256=H91-1a3keS6Si
 aip_agents/tools/code_sandbox/e2b_cloud_sandbox_extended.pyi,sha256=ZnW7S-Wla5J-ZD8P92-sXuzTGZ2-z5zyv4aH97FlshI,3554
 aip_agents/tools/code_sandbox/e2b_sandbox_tool.py,sha256=sTFTQd9Li9lB2iXgsARfpJEfFz9O2_sdFYFh8gFvC6A,16954
 aip_agents/tools/code_sandbox/e2b_sandbox_tool.pyi,sha256=ylrBQaqNBa1Jppld-mHjaskSa9SOjsQD8TdcK2bnl4s,1165
-aip_agents/tools/document_loader/__init__.py,sha256=Av6CDxzGuLhlpLtwHjKaEGHVVf06nIo50Vpn30BetG4,1206
+aip_agents/tools/document_loader/__init__.py,sha256=rnQFLYJqvit5eegGIWGdjOUoEJiyOh3nW0mLSRd7xfE,1375
 aip_agents/tools/document_loader/__init__.pyi,sha256=RYZb-EdfR1btPxFBUwfmrOWs51aVgwJuw9epguxnVgQ,650
-aip_agents/tools/document_loader/base_reader.py,sha256=JxqG-VXzDHUesGoCGqmzX_fzEG2f15DwanpbhDHko9k,9192
-aip_agents/tools/document_loader/base_reader.pyi,sha256=SlXiQdIUXeNATyxSvTI9Z5v9DVVUkqkuzMn2k5MPHRU,2606
-aip_agents/tools/document_loader/docx_reader_tool.py,sha256=7JNLDk8w0boRvg-ltBt3csTo8FyPdr_KunayMrm-JJk,2032
-aip_agents/tools/document_loader/docx_reader_tool.pyi,sha256=__bJ8jrlZOJsTo33YeLuTicuhGGOje5ROg7AXxjS3fI,249
-aip_agents/tools/document_loader/excel_reader_tool.py,sha256=GQOF4DC4eUxeobRvSPmt_I0oFQ2cD2e6Px-oWO21Y1Q,6153
-aip_agents/tools/document_loader/excel_reader_tool.pyi,sha256=-I5YT2OpOO0eWQPf5Ly0vbR3n1xgVREroq9qvie0I40,875
-aip_agents/tools/document_loader/pdf_reader_tool.py,sha256=YTUunsVHAB7FfOSjZMGXhRvlU7-YLYemJw3GCMkWkZE,2575
-aip_agents/tools/document_loader/pdf_reader_tool.pyi,sha256=VWXiWU1FlO0eh_Si8siXzmPkPRB63dTaR3Z2oggLUyc,442
+aip_agents/tools/document_loader/base_reader.py,sha256=fUK5yry8VnTLgu04LiW-fZRjpWLcIcs5nIzOwBUS4AQ,10341
+aip_agents/tools/document_loader/base_reader.pyi,sha256=OLn3MKD831u97rOiPGjoL6-VcJSOS-rDhhlPiXs7eY8,3085
+aip_agents/tools/document_loader/docx_reader_tool.py,sha256=EOGgXdjC3BXKPtu-6ABzGQp0xL8Z9L9WiMM7aUDPREE,2656
+aip_agents/tools/document_loader/docx_reader_tool.pyi,sha256=LKhrTT7xSVZr-zvrJQzhcIoUe_grQTXsl9DRxPS7oNk,415
+aip_agents/tools/document_loader/excel_reader_tool.py,sha256=uu1aK81SlrRR_8hoEGApRH3_QM-CpSgMBOX0tWPwESE,6599
+aip_agents/tools/document_loader/excel_reader_tool.pyi,sha256=my50Gc51IYNyekVBLRFksbsDG6lmK85o2d-exvff9r4,927
+aip_agents/tools/document_loader/pdf_reader_tool.py,sha256=36U_r7E0n_nLl4xUXC4nDXmr3L6kASfgKHuhtJFZeqI,3109
+aip_agents/tools/document_loader/pdf_reader_tool.pyi,sha256=ZKr5E3ePTJWHdLfM3_7Ji9NGDpL29uKLFhLl-smZQqQ,494
 aip_agents/tools/document_loader/pdf_splitter.py,sha256=-QyrlnN4AXDqY_dxeUzxcgCVJJiD1vYUYR7swfhM_RQ,5752
 aip_agents/tools/document_loader/pdf_splitter.pyi,sha256=3IiRDQWDz8QR5LH_sni9O-N2qJFheFjKOQMAhiWxB7E,716
 aip_agents/tools/gl_connector/__init__.py,sha256=tpQ0vF7gMiyO32jTBndOAVor5MRDNff7yh_SC7Frv_U,136
@@ -538,7 +538,7 @@ aip_agents/utils/pii/pii_helper.py,sha256=g0yRzakfA2AA6vjUNLWHqFlcxyLql6MXQ90NN3
 aip_agents/utils/pii/pii_helper.pyi,sha256=dulZs150ikbAL3Bw2YLcz3_g4DsGmL3lciwf8mKxEjI,2939
 aip_agents/utils/pii/uuid_deanonymizer_mapping.py,sha256=Gks8l8t0cuS9pzoQnrpiK1CaLmWYksjOnTeiHh3_7EE,7348
 aip_agents/utils/pii/uuid_deanonymizer_mapping.pyi,sha256=gnWfD1rWZh_tloJjgKiZ6f6iNUuBaHpKqCSiP0d-9bs,3084
-aip_agents_binary-0.5.15.dist-info/METADATA,sha256=_rnBCf9SJp55WwHGQ18LO3dVpcX7xp98U050dNAzrQo,22786
-aip_agents_binary-0.5.15.dist-info/WHEEL,sha256=PaP4PvkDyiSc4C6Dhw6ccQmfxsWFrSv-lJQjBshu0hw,105
-aip_agents_binary-0.5.15.dist-info/top_level.txt,sha256=PEz8vcwC1bH4UrkhF0LkIYCNfXGWZUHdSklbvkBe25E,11
-aip_agents_binary-0.5.15.dist-info/RECORD,,
+aip_agents_binary-0.5.17.dist-info/METADATA,sha256=2RWWD_Yd2YUYRVCLHiCZCB5mlISRwe0rw71BU3mLHiU,22593
+aip_agents_binary-0.5.17.dist-info/WHEEL,sha256=PaP4PvkDyiSc4C6Dhw6ccQmfxsWFrSv-lJQjBshu0hw,105
+aip_agents_binary-0.5.17.dist-info/top_level.txt,sha256=PEz8vcwC1bH4UrkhF0LkIYCNfXGWZUHdSklbvkBe25E,11
+aip_agents_binary-0.5.17.dist-info/RECORD,,