morphik 1.2.2__tar.gz → 1.2.3__tar.gz

{morphik-1.2.2 → morphik-1.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: morphik
-Version: 1.2.2
+Version: 1.2.3
 Summary: Morphik Python Client
 Author-email: Morphik <founders@morphik.ai>
 Requires-Python: >=3.8
@@ -106,6 +106,11 @@ renamed = moved.rename("specs-v2")
 # Scope queries to a path and include descendants with folder_depth=-1
 chunks = folder.retrieve_chunks(query="design notes", folder_depth=-1)
 docs = db.list_documents(folder_name="/projects/alpha", folder_depth=-1)
+
+# List only the fields you need. The server reads and returns just those columns, so
+# the full document text is never downloaded — fast for large corpora.
+for doc in db.list_documents(fields=["metadata"]).documents:
+    print(doc.external_id, doc.metadata)
 ```
 
 `Folder.full_path` is exposed on folder objects, and `Document.folder_path` mirrors server responses for tracing scope.

{morphik-1.2.2 → morphik-1.2.3}/README.md

@@ -93,6 +93,11 @@ renamed = moved.rename("specs-v2")
 # Scope queries to a path and include descendants with folder_depth=-1
 chunks = folder.retrieve_chunks(query="design notes", folder_depth=-1)
 docs = db.list_documents(folder_name="/projects/alpha", folder_depth=-1)
+
+# List only the fields you need. The server reads and returns just those columns, so
+# the full document text is never downloaded — fast for large corpora.
+for doc in db.list_documents(fields=["metadata"]).documents:
+    print(doc.external_id, doc.metadata)
 ```
 
 `Folder.full_path` is exposed on folder objects, and `Document.folder_path` mirrors server responses for tracing scope.

{morphik-1.2.2 → morphik-1.2.3}/morphik/__init__.py

@@ -14,4 +14,4 @@ __all__ = [
     "DocumentQueryResponse",
 ]
 
-__version__ = "1.2.2"
+__version__ = "1.2.3"

{morphik-1.2.2 → morphik-1.2.3}/morphik/_internal.py

@@ -428,6 +428,7 @@ class _MorphikClientLogic:
         completed_only: bool,
         sort_by: Optional[str],
         sort_direction: str,
+        fields: Optional[List[str]] = None,
     ) -> Tuple[Dict[str, Any], Dict[str, Any]]:
         """Prepare request for list_docs endpoint"""
         params = {}
@@ -450,6 +451,15 @@ class _MorphikClientLogic:
             "sort_by": sort_by,
             "sort_direction": sort_direction,
         }
+        if fields:
+            # Always include the fields required to reconstruct a Document client-side, so
+            # projected responses still parse into Document objects. When any metadata field
+            # is requested, also pull metadata_types so typed values (datetime/date/decimal)
+            # are reconstructed instead of returned as raw strings.
+            projected = ["external_id", "content_type", *fields]
+            if any(field.split(".", 1)[0] == "metadata" for field in fields):
+                projected.append("metadata_types")
+            data["fields"] = list(dict.fromkeys(projected))
         return params, data
 
     def _prepare_batch_get_documents_request(

{morphik-1.2.2 → morphik-1.2.3}/morphik/_scoped_ops.py

@@ -277,6 +277,7 @@ class _ScopedOperationsMixin:
         completed_only: bool,
         sort_by: Optional[str],
         sort_direction: str,
+        fields: Optional[List[str]] = None,
     ):
         params, data = self._logic._prepare_list_documents_request(
             skip,
@@ -291,6 +292,7 @@ class _ScopedOperationsMixin:
             completed_only,
             sort_by,
             sort_direction,
+            fields,
         )
 
         return self._execute_scoped_operation(

{morphik-1.2.2 → morphik-1.2.3}/morphik/async_.py

@@ -12,11 +12,11 @@ from pydantic import BaseModel
 from ._internal import FinalChunkResult, _MorphikClientLogic
 from ._scoped_ops import _ScopedOperationsMixin
 from ._shared import (
+    build_create_app_payload,
+    build_document_by_filename_params,
     build_folder_endpoint_identifier,
     build_folder_move_payload,
     build_folder_rename_path,
-    build_create_app_payload,
-    build_document_by_filename_params,
     build_list_apps_params,
     build_logs_params,
     build_rename_app_params,
@@ -267,8 +267,15 @@ class _AsyncScopedClientOps:
         completed_only: bool = False,
         sort_by: Optional[str] = "updated_at",
         sort_direction: str = "desc",
+        fields: Optional[List[str]] = None,
     ) -> ListDocsResponse:
-        """List documents within this scope (async)."""
+        """List documents within this scope (async).
+
+        Args:
+            fields: Optional list of fields to return for each document (e.g. ["metadata"]).
+                Only those fields are read and returned, so the full document text is never
+                downloaded. external_id and content_type are always included.
+        """
         effective_folder = self._merge_folders(additional_folders)
         return await self._client._scoped_list_documents(
             skip=skip,
@@ -283,6 +290,7 @@ class _AsyncScopedClientOps:
             completed_only=completed_only,
             sort_by=sort_by,
             sort_direction=sort_direction,
+            fields=fields,
         )
 
     async def batch_get_documents(
@@ -1231,6 +1239,7 @@ class AsyncMorphik(_ScopedOperationsMixin):
         completed_only: bool = False,
         sort_by: Optional[str] = "updated_at",
         sort_direction: str = "desc",
+        fields: Optional[List[str]] = None,
     ) -> ListDocsResponse:
         """
         List accessible documents.
@@ -1247,6 +1256,9 @@ class AsyncMorphik(_ScopedOperationsMixin):
             completed_only: Only return completed documents
             sort_by: Field to sort by (created_at, updated_at, filename, external_id)
             sort_direction: Sort direction (asc, desc)
+            fields: Optional list of fields to return for each document (e.g. ["metadata"]).
+                Only those fields are read and returned, so the full document text is never
+                downloaded. external_id and content_type are always included.
         Returns:
             ListDocsResponse: Response with documents and metadata
 
@@ -1264,6 +1276,7 @@ class AsyncMorphik(_ScopedOperationsMixin):
             completed_only=completed_only,
             sort_by=sort_by,
             sort_direction=sort_direction,
+            fields=fields,
         )
 
     async def get_document(self, document_id: str) -> Document:

{morphik-1.2.2 → morphik-1.2.3}/morphik/sync.py

@@ -12,11 +12,11 @@ from pydantic import BaseModel
 from ._internal import FinalChunkResult, _MorphikClientLogic
 from ._scoped_ops import _ScopedOperationsMixin
 from ._shared import (
+    build_create_app_payload,
+    build_document_by_filename_params,
     build_folder_endpoint_identifier,
     build_folder_move_payload,
     build_folder_rename_path,
-    build_create_app_payload,
-    build_document_by_filename_params,
     build_list_apps_params,
     build_logs_params,
     build_rename_app_params,
@@ -283,9 +283,16 @@ class _ScopedClientOps:
         completed_only: bool = False,
         sort_by: Optional[str] = "updated_at",
         sort_direction: str = "desc",
+        fields: Optional[List[str]] = None,
     ) -> ListDocsResponse:
         """
         List documents within this scope.
+
+        Args:
+            fields: Optional list of fields to return for each document (e.g.
+                ["metadata"]). Only those fields are read and returned, so the full
+                document text is never downloaded. external_id and content_type are
+                always included.
         """
         effective_folder = self._merge_folders(additional_folders)
         return self._client._scoped_list_documents(
@@ -301,6 +308,7 @@ class _ScopedClientOps:
             completed_only=completed_only,
             sort_by=sort_by,
             sort_direction=sort_direction,
+            fields=fields,
         )
 
     def batch_get_documents(
@@ -1270,6 +1278,7 @@ class Morphik(_ScopedOperationsMixin):
         completed_only: bool = False,
         sort_by: Optional[str] = "updated_at",
         sort_direction: str = "desc",
+        fields: Optional[List[str]] = None,
     ) -> ListDocsResponse:
         """
         List accessible documents.
@@ -1286,6 +1295,9 @@ class Morphik(_ScopedOperationsMixin):
             completed_only: Only return completed documents
             sort_by: Field to sort by (created_at, updated_at, filename, external_id)
             sort_direction: Sort direction (asc, desc)
+            fields: Optional list of fields to return for each document (e.g. ["metadata"]).
+                Only those fields are read and returned, so the full document text is never
+                downloaded. external_id and content_type are always included.
         Returns:
             ListDocsResponse: Response with documents and metadata
 
@@ -1303,6 +1315,7 @@ class Morphik(_ScopedOperationsMixin):
             completed_only=completed_only,
             sort_by=sort_by,
             sort_direction=sort_direction,
+            fields=fields,
         )
 
     def get_document(self, document_id: str) -> Document:

{morphik-1.2.2 → morphik-1.2.3}/morphik/tests/test_scoped_ops_unit.py

@@ -117,6 +117,39 @@ def test_sync_list_documents_payloads_across_scopes():
         client.close()
 
 
+def test_sync_list_documents_fields_projection():
+    client, calls = _make_sync_client()
+    try:
+        # external_id + content_type are always added so the response parses into a Document;
+        # metadata_types is added so typed metadata values are reconstructed, not left as strings.
+        client.list_documents(fields=["metadata"])
+        assert calls.pop()["data"]["fields"] == ["external_id", "content_type", "metadata", "metadata_types"]
+
+        # Already-included required fields are not duplicated; order is preserved.
+        client.list_documents(fields=["external_id", "filename", "metadata"])
+        assert calls.pop()["data"]["fields"] == [
+            "external_id",
+            "content_type",
+            "filename",
+            "metadata",
+            "metadata_types",
+        ]
+
+        # Nested metadata paths also trigger metadata_types.
+        client.list_documents(fields=["metadata.client"])
+        assert calls.pop()["data"]["fields"] == ["external_id", "content_type", "metadata.client", "metadata_types"]
+
+        # Non-metadata projection does not pull metadata_types.
+        client.list_documents(fields=["filename"])
+        assert calls.pop()["data"]["fields"] == ["external_id", "content_type", "filename"]
+
+        # No fields -> no projection requested (full documents).
+        client.list_documents()
+        assert "fields" not in calls.pop()["data"]
+    finally:
+        client.close()
+
+
 def test_async_client_http2_toggle(monkeypatch):
     captured = []
 

{morphik-1.2.2 → morphik-1.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "morphik"
-version = "1.2.2"
+version = "1.2.3"
 authors = [
     { name = "Morphik", email = "founders@morphik.ai" },
 ]