Added draft implementation of built-in preprocessing for RAG.

llama_stack/apis/preprocessing/preprocessing.py

@@ -26,8 +26,8 @@ class PreprocessingDataType(Enum):
 @json_schema_type
 class PreprocessingInput(BaseModel):
     preprocessor_input_id: str
-    preprocessor_input_type: Optional[PreprocessingDataType]
+    preprocessor_input_type: Optional[PreprocessingDataType] = None
-    path_or_content: str | URL
+    path_or_content: str | InterleavedContent | URL
 
 
 PreprocessorOptions = Dict[str, Any]
@@ -36,7 +36,7 @@ PreprocessorOptions = Dict[str, Any]
 @json_schema_type
 class PreprocessingResponse(BaseModel):
     status: bool
-    results: Optional[List[str | InterleavedContent | Chunk]]
+    results: Optional[List[str | InterleavedContent | Chunk]] = None
 
 
 class PreprocessorStore(Protocol):

llama_stack/providers/inline/preprocessing/basic/__init__.py

@@ -0,0 +1,18 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from .config import InlineBasicPreprocessorConfig
+
+
+async def get_provider_impl(
+    config: InlineBasicPreprocessorConfig,
+    _deps,
+):
+    from .basic import InclineBasicPreprocessorImpl
+
+    impl = InclineBasicPreprocessorImpl(config)
+    await impl.initialize()
+    return impl

llama_stack/providers/inline/preprocessing/basic/basic.py

@@ -0,0 +1,121 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+import logging
+import re
+from typing import List
+
+import httpx
+
+from llama_stack.apis.common.content_types import URL
+from llama_stack.apis.preprocessing import (
+    Preprocessing,
+    PreprocessingDataType,
+    PreprocessingInput,
+    PreprocessingResponse,
+    Preprocessor,
+    PreprocessorOptions,
+)
+from llama_stack.providers.datatypes import PreprocessorsProtocolPrivate
+from llama_stack.providers.inline.preprocessing.basic.config import InlineBasicPreprocessorConfig
+from llama_stack.providers.utils.inference.prompt_adapter import interleaved_content_as_str
+from llama_stack.providers.utils.memory.vector_store import content_from_data, parse_pdf
+
+log = logging.getLogger(__name__)
+
+
+class InclineBasicPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate):
+    # this preprocessor can either receive documents (text or binary) or document URIs
+    INPUT_TYPES = [
+        PreprocessingDataType.binary_document,
+        PreprocessingDataType.raw_text_document,
+        PreprocessingDataType.document_uri,
+    ]
+
+    # this preprocessor optionally retrieves the documents and converts them into plain text
+    OUTPUT_TYPES = [PreprocessingDataType.raw_text_document]
+
+    URL_VALIDATION_PATTERN = re.compile("^(https?://|file://|data:)")
+
+    def __init__(self, config: InlineBasicPreprocessorConfig) -> None:
+        self.config = config
+
+    async def initialize(self) -> None: ...
+
+    async def shutdown(self) -> None: ...
+
+    async def register_preprocessor(self, preprocessor: Preprocessor) -> None: ...
+
+    async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
+
+    async def preprocess(
+        self,
+        preprocessor_id: str,
+        preprocessor_inputs: List[PreprocessingInput],
+        options: PreprocessorOptions,
+    ) -> PreprocessingResponse:
+        results = []
+
+        for inp in preprocessor_inputs:
+            is_pdf = options["binary_document_type"] == "pdf"
+            input_type = self._resolve_input_type(inp, is_pdf)
+
+            if input_type == PreprocessingDataType.document_uri:
+                document = await self._fetch_document(inp, is_pdf)
+                if document is None:
+                    continue
+            elif input_type == PreprocessingDataType.binary_document:
+                document = inp.path_or_content
+                if not is_pdf:
+                    log.error(f"Unsupported binary document type: {options['binary_document_type']}")
+                    continue
+            elif input_type == PreprocessingDataType.raw_text_document:
+                document = interleaved_content_as_str(inp.path_or_content)
+            else:
+                log.error(f"Unexpected preprocessor input type: {inp.preprocessor_input_type}")
+                continue
+
+            if is_pdf:
+                document = parse_pdf(document)
+
+            results.append(document)
+
+        return PreprocessingResponse(status=True, results=results)
+
+    @staticmethod
+    async def _resolve_input_type(preprocessor_input: PreprocessingInput, is_pdf: bool) -> PreprocessingDataType:
+        if preprocessor_input.preprocessor_input_type is not None:
+            return preprocessor_input.preprocessor_input_type
+
+        if isinstance(preprocessor_input.path_or_content, URL):
+            return PreprocessingDataType.document_uri
+        if InclineBasicPreprocessorImpl.URL_VALIDATION_PATTERN.match(preprocessor_input.path_or_content):
+            return PreprocessingDataType.document_uri
+        if is_pdf:
+            return PreprocessingDataType.binary_document
+
+        return PreprocessingDataType.raw_text_document
+
+    @staticmethod
+    async def _fetch_document(preprocessor_input: PreprocessingInput, is_pdf: bool) -> str | None:
+        if isinstance(preprocessor_input.path_or_content, str):
+            url = preprocessor_input.path_or_content
+            if not InclineBasicPreprocessorImpl.URL_VALIDATION_PATTERN.match(url):
+                log.error(f"Unexpected URL: {url}")
+                return None
+        elif isinstance(preprocessor_input.path_or_content, URL):
+            url = preprocessor_input.path_or_content.uri
+        else:
+            log.error(
+                f"Unexpected type {type(preprocessor_input.path_or_content)} for input {preprocessor_input.path_or_content}, skipping this input."
+            )
+            return None
+
+        if url.startswith("data:"):
+            return content_from_data(url)
+
+        async with httpx.AsyncClient() as client:
+            r = await client.get(url)
+        return r.content if is_pdf else r.text

llama_stack/providers/inline/preprocessing/basic/config.py

@@ -0,0 +1,9 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+from pydantic import BaseModel
+
+
+class InlineBasicPreprocessorConfig(BaseModel): ...

llama_stack/providers/inline/preprocessing/docling/docling.py

@@ -3,6 +3,7 @@
 #
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
+import logging
 from typing import List
 
 from docling.document_converter import DocumentConverter
@@ -21,6 +22,8 @@ from llama_stack.apis.vector_io import Chunk
 from llama_stack.providers.datatypes import PreprocessorsProtocolPrivate
 from llama_stack.providers.inline.preprocessing.docling import InlineDoclingConfig
 
+log = logging.getLogger(__name__)
+
 
 class InclineDoclingPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate):
     # this preprocessor receives URLs / paths to documents as input
@@ -58,7 +61,10 @@ class InclineDoclingPreprocessorImpl(Preprocessing, PreprocessorsProtocolPrivate
             elif isinstance(inp.path_or_content, URL):
                 url = inp.path_or_content.uri
             else:
-                raise ValueError(f"Unexpected type {type(inp.path_or_content)} for input {inp.path_or_content}")
+                log.error(
+                    f"Unexpected type {type(inp.path_or_content)} for input {inp.path_or_content}, skipping this input."
+                )
+                continue
 
             converted_document = self.converter.convert(url).document
             if self.config.chunk:

llama_stack/providers/inline/preprocessing/simple_chunking/__init__.py

@@ -0,0 +1,18 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+from .config import InclineSimpleChunkingConfig
+
+
+async def get_provider_impl(
+    config: InclineSimpleChunkingConfig,
+    _deps,
+):
+    from .simple_chunking import InclineSimpleChunkingImpl
+
+    impl = InclineSimpleChunkingImpl(config)
+    await impl.initialize()
+    return impl

llama_stack/providers/inline/preprocessing/simple_chunking/config.py

@@ -0,0 +1,11 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+from pydantic import BaseModel
+
+
+class InclineSimpleChunkingConfig(BaseModel):
+    chunk_size_in_tokens: int = 512
+    chunk_overlap_ratio: int = 4

llama_stack/providers/inline/preprocessing/simple_chunking/simple_chunking.py

@@ -0,0 +1,96 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+import logging
+from enum import Enum
+from typing import List, Tuple
+
+from llama_models.llama3.api import Tokenizer
+
+from llama_stack.apis.preprocessing import (
+    Preprocessing,
+    PreprocessingDataType,
+    PreprocessingInput,
+    PreprocessingResponse,
+    Preprocessor,
+    PreprocessorOptions,
+)
+from llama_stack.apis.vector_io import Chunk
+from llama_stack.providers.datatypes import PreprocessorsProtocolPrivate
+from llama_stack.providers.inline.preprocessing.simple_chunking import InclineSimpleChunkingConfig
+
+log = logging.getLogger(__name__)
+
+
+class SimpleChunkingOptions(Enum):
+    chunk_size_in_tokens = "chunk_size_in_tokens"
+    chunk_overlap_ratio = "chunk_overlap_ratio"
+
+
+class InclineSimpleChunkingImpl(Preprocessing, PreprocessorsProtocolPrivate):
+    # this preprocessor receives plain text and returns chunks
+    INPUT_TYPES = [PreprocessingDataType.raw_text_document]
+    OUTPUT_TYPES = [PreprocessingDataType.chunks]
+
+    def __init__(self, config: InclineSimpleChunkingConfig) -> None:
+        self.config = config
+
+    async def initialize(self) -> None: ...
+
+    async def shutdown(self) -> None: ...
+
+    async def register_preprocessor(self, preprocessor: Preprocessor) -> None: ...
+
+    async def unregister_preprocessor(self, preprocessor_id: str) -> None: ...
+
+    async def preprocess(
+        self,
+        preprocessor_id: str,
+        preprocessor_inputs: List[PreprocessingInput],
+        options: PreprocessorOptions,
+    ) -> PreprocessingResponse:
+        chunks = []
+
+        window_len, overlap_len = self._resolve_chunk_size_params(options)
+
+        for inp in preprocessor_inputs:
+            new_chunks = self.make_overlapped_chunks(
+                inp.preprocessor_input_id, inp.path_or_content, window_len, overlap_len
+            )
+            chunks.extend(new_chunks)
+
+        return PreprocessingResponse(status=True, results=chunks)
+
+    def _resolve_chunk_size_params(self, options: PreprocessorOptions) -> Tuple[int, int]:
+        window_len = options.get(str(SimpleChunkingOptions.chunk_size_in_tokens), self.config.chunk_size_in_tokens)
+
+        chunk_overlap_ratio = options.get(
+            str(SimpleChunkingOptions.chunk_overlap_ratio), self.config.chunk_overlap_ratio
+        )
+        overlap_len = window_len // chunk_overlap_ratio
+
+        return window_len, overlap_len
+
+    @staticmethod
+    def make_overlapped_chunks(document_id: str, text: str, window_len: int, overlap_len: int) -> List[Chunk]:
+        tokenizer = Tokenizer.get_instance()
+        tokens = tokenizer.encode(text, bos=False, eos=False)
+
+        chunks = []
+        for i in range(0, len(tokens), window_len - overlap_len):
+            toks = tokens[i : i + window_len]
+            chunk = tokenizer.decode(toks)
+            # chunk is a string
+            chunks.append(
+                Chunk(
+                    content=chunk,
+                    metadata={
+                        "token_count": len(toks),
+                        "document_id": document_id,
+                    },
+                )
+            )
+
+        return chunks

llama_stack/providers/registry/preprocessing.py

@@ -23,4 +23,20 @@ def available_providers() -> List[ProviderSpec]:
             config_class="llama_stack.providers.inline.preprocessing.docling.InlineDoclingConfig",
             api_dependencies=[],
         ),
+        InlineProviderSpec(
+            api=Api.preprocessing,
+            provider_type="inline::basic",
+            pip_packages=["httpx", "pypdf"],
+            module="llama_stack.providers.inline.preprocessing.basic",
+            config_class="llama_stack.providers.inline.preprocessing.basic.InlineBasicPreprocessorConfig",
+            api_dependencies=[],
+        ),
+        InlineProviderSpec(
+            api=Api.preprocessing,
+            provider_type="inline::simple_chunking",
+            pip_packages=[],
+            module="llama_stack.providers.inline.preprocessing.simple_chunking",
+            config_class="llama_stack.providers.inline.preprocessing.simple_chunking.InclineSimpleChunkingConfig",
+            api_dependencies=[],
+        ),
     ]