cohere-compass-sdk 2.2.0__tar.gz → 2.2.2__tar.gz

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cohere-compass-sdk
-Version: 2.2.0
+Version: 2.2.2
 Summary: Cohere Compass SDK
 Requires-Python: >=3.11,<4.0
 Classifier: Programming Language :: Python :: 3

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/clients/compass.py

@@ -18,6 +18,7 @@ from collections.abc import Iterable
 from dataclasses import dataclass
 from datetime import timedelta
 from statistics import mean
+from types import TracebackType
 from typing import Any, Literal
 
 # 3rd party imports
@@ -267,9 +268,9 @@ class CompassClient:
             if httpx_client.timeout.read
             else DEFAULT_COMPASS_CLIENT_TIMEOUT
         )
-        self.httpx_client = httpx_client or httpx.Client(
-            timeout=self.timeout.total_seconds()
-        )
+        self.httpx = httpx_client or httpx.Client(timeout=self.timeout.total_seconds())
+        self._own_httpx_client = httpx_client is None
+        self._closed = False
 
         self.bearer_token = bearer_token
 
@@ -281,8 +282,23 @@ class CompassClient:
         self.retry_wait = retry_wait
 
     def close(self):
-        """Close the HTTP client connection."""
-        self.httpx_client.close()
+        """Close the httpx client if it was created by this instance."""
+        if self._own_httpx_client and not self._closed:
+            self.httpx.close()
+            self._closed = True
+
+    def __enter__(self):
+        """For use by "with" statements."""
+        return self
+
+    def __exit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_value: BaseException | None,
+        traceback: TracebackType | None,
+    ) -> None:
+        """For use by "with" statements."""
+        self.close()
 
     def get_models(
         self,
@@ -615,10 +631,10 @@ class CompassClient:
         index_name: str,
         filename: str,
         filebytes: bytes,
-        content_type: ContentTypeEnum,
         document_id: str,
         attributes: DocumentAttributes = DocumentAttributes(),
         config: ParseableDocumentConfig = ParseableDocumentConfig(),
+        content_type: ContentTypeEnum | None = None,
         max_retries: int | None = None,
         retry_wait: timedelta | None = None,
         timeout: timedelta | None = None,
@@ -629,7 +645,8 @@ class CompassClient:
         :param index_name: The name of the index.
         :param filename: The filename of the document.
         :param filebytes: The raw bytes of the document.
-        :param content_type: The content type of the document.
+        :param content_type: optional content type of the document.
+            Recommended to pass it otherwise auto-detected.
         :param document_id: The ID to assign to the document.
         :param attributes: Additional attributes to add to the document.
         :param config: Configuration for the document parsing.
@@ -1392,27 +1409,27 @@ class CompassClient:
             headers = {"Authorization": f"Bearer {self.bearer_token}"}
 
         if http_method == "GET":
-            response = self.httpx_client.get(
+            response = self.httpx.get(
                 target_path,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "POST":
-            response = self.httpx_client.post(
+            response = self.httpx.post(
                 target_path,
                 json=data_dict,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "PUT":
-            response = self.httpx_client.put(
+            response = self.httpx.put(
                 target_path,
                 json=data_dict,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "DELETE":
-            response = self.httpx_client.delete(
+            response = self.httpx.delete(
                 target_path,
                 headers=headers,
                 timeout=timeout.total_seconds(),

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/clients/compass_async.py

@@ -14,6 +14,7 @@ from collections import deque
 from collections.abc import AsyncIterable, Iterable
 from datetime import timedelta
 from statistics import mean
+from types import TracebackType
 from typing import Any, Literal
 
 # 3rd party imports
@@ -133,9 +134,11 @@ class CompassAsyncClient:
             if httpx_client.timeout.read
             else DEFAULT_COMPASS_CLIENT_TIMEOUT
         )
-        self.httpx_client = httpx_client or httpx.AsyncClient(
+        self.httpx = httpx_client or httpx.AsyncClient(
             timeout=self.timeout.total_seconds()
         )
+        self._own_httpx_client = httpx_client is None
+        self._closed = False
 
         self.bearer_token = bearer_token
 
@@ -149,8 +152,25 @@ class CompassAsyncClient:
         self.retry_wait = retry_wait
 
     async def aclose(self):
-        """Close the HTTP client."""
-        await self.httpx_client.aclose()
+        """Close the httpx client if it was created by the CompassAsyncClient."""
+        if self._own_httpx_client and not self._closed:
+            await self.httpx.aclose()
+            self._closed = True
+
+    close = aclose  # Alias for consistency with sync client
+
+    async def __aenter__(self):
+        """For use by "async with" statements."""
+        return self
+
+    async def __aexit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_value: BaseException | None,
+        traceback: TracebackType | None,
+    ) -> None:
+        """For use by "async with" statements."""
+        await self.aclose()
 
     async def get_models(
         self,
@@ -470,10 +490,10 @@ class CompassAsyncClient:
         index_name: str,
         filename: str,
         filebytes: bytes,
-        content_type: ContentTypeEnum,
         document_id: str,
         attributes: DocumentAttributes = DocumentAttributes(),
         config: ParseableDocumentConfig = ParseableDocumentConfig(),
+        content_type: ContentTypeEnum | None = None,
         max_retries: int | None = None,
         retry_wait: timedelta | None = None,
         timeout: timedelta | None = None,
@@ -484,7 +504,8 @@ class CompassAsyncClient:
         :param index_name: the name of the index
         :param filename: the filename of the document
         :param filebytes: the bytes of the document
-        :param content_type: the content type of the document
+        :param content_type: optional content type of the document.
+            Recommended to pass it otherwise auto-detected.
         :param document_id: the id of the document (optional)
         :param attributes: Additional attributes to add to the document.
         :param config: Configuration for the document parsing.
@@ -1265,27 +1286,27 @@ class CompassAsyncClient:
             headers = {"Authorization": f"Bearer {self.bearer_token}"}
 
         if http_method == "GET":
-            response = await self.httpx_client.get(
+            response = await self.httpx.get(
                 target_path,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "POST":
-            response = await self.httpx_client.post(
+            response = await self.httpx.post(
                 target_path,
                 json=data_dict,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "PUT":
-            response = await self.httpx_client.put(
+            response = await self.httpx.put(
                 target_path,
                 json=data_dict,
                 headers=headers,
                 timeout=timeout.total_seconds(),
             )
         elif http_method == "DELETE":
-            response = await self.httpx_client.delete(
+            response = await self.httpx.delete(
                 target_path,
                 headers=headers,
                 timeout=timeout.total_seconds(),

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/clients/parser.py

@@ -12,6 +12,7 @@ import logging
 from collections.abc import Callable, Iterable
 from concurrent.futures import ThreadPoolExecutor
 from datetime import timedelta
+from types import TracebackType
 from typing import Any
 
 # 3rd party imports
@@ -42,9 +43,7 @@ from cohere_compass.models import (
 )
 from cohere_compass.utils.fs import open_document, scan_folder
 from cohere_compass.utils.iter import imap_parallel
-from cohere_compass.utils.retry import (
-    is_retryable_compass_exception,
-)
+from cohere_compass.utils.retry import is_retryable_compass_exception
 
 Fn_or_Dict = dict[str, Any] | Callable[[CompassDocument], dict[str, Any]]
 
@@ -123,15 +122,34 @@ class CompassParserClient:
             if httpx_client.timeout.read
             else DEFAULT_COMPASS_PARSER_CLIENT_TIMEOUT
         )
-        self.httpx_client = httpx_client or httpx.Client(
-            timeout=self.timeout.total_seconds()
-        )
+        self.httpx = httpx_client or httpx.Client(timeout=self.timeout.total_seconds())
+        self._own_httpx_client = httpx_client is None
+        self._closed = False
 
         self.metadata_config = metadata_config
         logger.info(
             f"CompassParserClient initialized with parser_url: {self.parser_url}"
         )
 
+    def close(self):
+        """Close the httpx client if it was created by this instance."""
+        if self._own_httpx_client and not self._closed:
+            self.httpx.close()
+            self._closed = True
+
+    def __enter__(self):
+        """For use by "with" statements."""
+        return self
+
+    def __exit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_value: BaseException | None,
+        traceback: TracebackType | None,
+    ) -> None:
+        """For use by "with" statements."""
+        self.close()
+
     def process_folder(
         self,
         *,
@@ -390,7 +408,7 @@ class CompassParserClient:
             headers = {"Authorization": f"Bearer {self.bearer_token}"}
 
         with handle_httpx_exceptions():
-            res = self.httpx_client.post(
+            res = self.httpx.post(
                 url=f"{self.parser_url}/v1/process_file",
                 data={"data": json.dumps(params.model_dump())},
                 files={"file": (filename, file_bytes)},

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/clients/parser_async.py

@@ -11,6 +11,7 @@ import logging
 from collections.abc import Callable
 from concurrent.futures import ThreadPoolExecutor
 from datetime import timedelta
+from types import TracebackType
 from typing import Any
 
 # 3rd party imports
@@ -39,9 +40,7 @@ from cohere_compass.models import (
 )
 from cohere_compass.utils.asyn import async_map
 from cohere_compass.utils.fs import open_document, scan_folder
-from cohere_compass.utils.retry import (
-    is_retryable_compass_exception,
-)
+from cohere_compass.utils.retry import is_retryable_compass_exception
 
 Fn_or_Dict = dict[str, Any] | Callable[[CompassDocument], dict[str, Any]]
 
@@ -123,12 +122,35 @@ class CompassParserAsyncClient:
         self.httpx = httpx_client or httpx.AsyncClient(
             timeout=self.timeout.total_seconds()
         )
+        self._own_httpx_client = httpx_client is None
+        self._closed = False
 
         self.metadata_config = metadata_config
         logger.info(
             f"CompassParserClient initialized with parser_url: {self.parser_url}"
         )
 
+    async def aclose(self):
+        """Close the httpx client if it was created by the CompassParserAsyncClient."""
+        if self._own_httpx_client and not self._closed:
+            await self.httpx.aclose()
+            self._closed = True
+
+    close = aclose  # Alias for consistency with sync client
+
+    async def __aenter__(self):
+        """For use by "async with" statements."""
+        return self
+
+    async def __aexit__(
+        self,
+        exc_type: type[BaseException] | None,
+        exc_value: BaseException | None,
+        traceback: TracebackType | None,
+    ) -> None:
+        """For use by "async with" statements."""
+        await self.aclose()
+
     def process_folder(
         self,
         *,

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/models/config.py

@@ -113,6 +113,19 @@ class PresentationParsingStrategy(str, Enum):
         return cls.Unstructured
 
 
+class DocxParsingStrategy(str, Enum):
+    """Enum for specifying the parsing strategy for DOCX files."""
+
+    # Uses https://github.com/microsoft/markitdown
+    MarkItDown = "MarkItDown"
+    # Converts the DOCX to PDF and uses the PDF parsing strategy
+    ConvertToPDF = "ConvertToPDF"
+
+    @classmethod
+    def _missing_(cls, value: Any):
+        return cls.MarkItDown
+
+
 class ParsingStrategy(str, Enum):
     """Enum for specifying the parsing strategy to use."""
 
@@ -193,12 +206,12 @@ class ParserConfig(BaseModel):
     vertical_table_crop_margin: int = 100
     horizontal_table_crop_margin: int = 100
 
+    pdf_parsing_config: PDFParsingConfig = PDFParsingConfig()
     pdf_parsing_strategy: PDFParsingStrategy = PDFParsingStrategy.QuickText
     tabular_parsing_strategy: TabularParsingStrategy = TabularParsingStrategy.Granular
-
-    pdf_parsing_config: PDFParsingConfig = PDFParsingConfig()
-
     presentation_parsing_strategy: PresentationParsingStrategy | None = None
+    docx_parsing_strategy: DocxParsingStrategy | None = None
+
     enable_assets_returned_as_base64: bool = True
 
 

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/cohere_compass/models/documents.py

@@ -300,7 +300,7 @@ class ParseableDocument(BaseModel):
     filename: Annotated[
         str, StringConstraints(min_length=1)
     ]  # Ensures the filename is a non-empty string
-    content_type: str
+    content_type: str | None = None
     content_length_bytes: PositiveInt  # File size must be a non-negative integer
     content_encoded_bytes: str  # Base64-encoded file contents
     attributes: DocumentAttributes

{cohere_compass_sdk-2.2.0 → cohere_compass_sdk-2.2.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "cohere-compass-sdk"
-version = "2.2.0"
+version = "2.2.2"
 authors = []
 description = "Cohere Compass SDK"
 readme = "README.md"
@@ -93,4 +93,4 @@ omit = [
 ]
 
 [tool.coverage.html]
-directory = "coverage_html"
+directory = "coverage_html"