added Unstructured fallback to auto provider. Wire remote::unstructured-api into the auto file processor as an optional fallback. When an unstructured api key is provided by the user in config, the auto provider will route supported file formats to Unstructured before checking for/returning a 422 error. This provides 65+ additional format support (including EML, DOC, MSG) when users opt-in with an API key, while maintaining priority routing to pypdf for pdfs and markitdown for office/image/audio. Tested with .eml file, successfully routes to Unstructured. pypdf and markitdown routing unchanged.

Author: sahana-sreeram

docs/docs/providers/file_processors/inline_auto.mdx

@@ -18,9 +18,10 @@ Composite file processor that automatically dispatches to the appropriate backen
 | `default_chunk_overlap_tokens` | `int` | No | 400 | Default chunk overlap in tokens when chunking_strategy type is 'auto' |
 | `extract_metadata` | `bool` | No | True | Whether to extract PDF metadata (title, author, etc.) |
 | `clean_text` | `bool` | No | True | Whether to clean extracted text (remove extra whitespace, normalize line breaks) |
+| `unstructured_api_key` | `SecretStr \| None` | No |  | Optional Unstructured.io API key. If provided, uses Unstructured as fallback for unsupported file types. |
 
 ## Sample Configuration
 
 ```yaml
-{}
+unstructured_api_key: ${env.UNSTRUCTURED_API_KEY}
 ```

src/ogx/distributions/ci-tests/config.yaml

@@ -193,6 +193,8 @@ providers:
   file_processors:
   - provider_id: auto
     provider_type: inline::auto
+    config:
+      unstructured_api_key: ${env.UNSTRUCTURED_API_KEY}
   interactions:
   - provider_id: builtin
     provider_type: inline::builtin

src/ogx/distributions/ci-tests/run-with-postgres-store.yaml

@@ -193,6 +193,8 @@ providers:
   file_processors:
   - provider_id: auto
     provider_type: inline::auto
+    config:
+      unstructured_api_key: ${env.UNSTRUCTURED_API_KEY}
   interactions:
   - provider_id: builtin
     provider_type: inline::builtin

src/ogx/distributions/starter/config.yaml

@@ -187,6 +187,8 @@ providers:
   file_processors:
   - provider_id: auto
     provider_type: inline::auto
+    config:
+      unstructured_api_key: ${env.UNSTRUCTURED_API_KEY}
   interactions:
   - provider_id: builtin
     provider_type: inline::builtin

src/ogx/distributions/starter/run-with-postgres-store.yaml

@@ -187,6 +187,8 @@ providers:
   file_processors:
   - provider_id: auto
     provider_type: inline::auto
+    config:
+      unstructured_api_key: ${env.UNSTRUCTURED_API_KEY}
   interactions:
   - provider_id: builtin
     provider_type: inline::builtin

src/ogx/providers/inline/file_processor/auto/auto.py

@@ -12,6 +12,8 @@
 from ogx.providers.inline.file_processor.markitdown.markitdown_processor import MarkItDownFileProcessor
 from ogx.providers.inline.file_processor.pypdf.config import PyPDFFileProcessorConfig
 from ogx.providers.inline.file_processor.pypdf.pypdf import PyPDFFileProcessor
+from ogx.providers.remote.file_processor.unstructured_api.config import UnstructuredApiFileProcessorConfig
+from ogx.providers.remote.file_processor.unstructured_api.unstructured_api import UnstructuredApiFileProcessor
 from ogx_api.file_processors import ProcessFileRequest, ProcessFileResponse
 from ogx_api.files import RetrieveFileRequest
 
@@ -50,7 +52,8 @@
 SUPPORTED_DESCRIPTION = (
     "PDF, text (txt, csv, md, json, xml, html, code), "
     "office (DOCX, PPTX, XLSX, XLS, DOC, PPT, RTF), "
-    "EPUB, RSS, ZIP, images, and audio"
+    "EPUB, RSS, ZIP, images, audio, "
+    "and additional formats via Unstructured.io (if user API key provided)"
 )
 
 
@@ -80,6 +83,15 @@ def __init__(self, config: AutoFileProcessorConfig, files_api) -> None:
         )
         self.markitdown = MarkItDownFileProcessor(markitdown_config, files_api)
 
+        # Initialize Unstructured if API key is provided
+        self.unstructured = None
+        if config.unstructured_api_key:
+            unstructured_config = UnstructuredApiFileProcessorConfig(
+                api_key=config.unstructured_api_key,
+                default_chunk_size_tokens=config.default_chunk_size_tokens,
+            )
+            self.unstructured = UnstructuredApiFileProcessor(unstructured_config, files_api)
+
     async def process_file(
         self,
         request: ProcessFileRequest,
@@ -100,6 +112,10 @@ async def process_file(
         if mime_type in MARKITDOWN_MIME_TYPES:
             return await self.markitdown.process_file(request=request, file=file)
 
+        # Try Unstructured as fallback for unsupported types
+        if self.unstructured:
+            return await self.unstructured.process_file(request=request, file=file)
+
         raise HTTPException(
             status_code=422,
             detail=f"File type '{mime_type or 'unknown'}' is not supported. Supported types: {SUPPORTED_DESCRIPTION}.",

src/ogx/providers/inline/file_processor/auto/config.py

@@ -6,7 +6,7 @@
 
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, SecretStr
 
 from ogx_api.vector_io import VectorStoreChunkingStrategyStaticConfig
 
@@ -39,6 +39,13 @@ class AutoFileProcessorConfig(BaseModel):
         default=True, description="Whether to clean extracted text (remove extra whitespace, normalize line breaks)"
     )
 
+    unstructured_api_key: SecretStr | None = Field(
+        default=None,
+        description="Optional Unstructured.io API key. If provided, uses Unstructured as fallback for unsupported file types.",
+    )
+
     @classmethod
     def sample_run_config(cls, **kwargs: Any) -> dict[str, Any]:
-        return {}
+        return {
+            "unstructured_api_key": "${env.UNSTRUCTURED_API_KEY}",
+        }