morphik 0.2.4__py3-none-any.whl → 0.2.5__py3-none-any.whl

morphik/__init__.py

@@ -12,4 +12,4 @@ __all__ = [
     "Document",
 ]
 
-__version__ = "0.2.4"
+__version__ = "0.2.5"

morphik/_internal.py

@@ -303,6 +303,7 @@ class _MorphikClientLogic:
         use_colpali: bool,
         folder_name: Optional[Union[str, List[str]]],
         end_user_id: Optional[str],
+        padding: int = 0,
     ) -> Dict[str, Any]:
         """Prepare request for retrieve_chunks endpoint"""
         request = {
@@ -316,6 +317,8 @@ class _MorphikClientLogic:
             request["folder_name"] = folder_name
         if end_user_id:
             request["end_user_id"] = end_user_id
+        if padding > 0:
+            request["padding"] = padding
         return request
 
     def _prepare_retrieve_docs_request(

morphik/async_.py

@@ -288,6 +288,7 @@ class AsyncFolder:
         min_score: float = 0.0,
         use_colpali: bool = True,
         additional_folders: Optional[List[str]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Retrieve relevant chunks within this folder.
@@ -299,13 +300,14 @@ class AsyncFolder:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model
             additional_folders: Optional list of additional folder names to further scope operations
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
 
         Returns:
             List[FinalChunkResult]: List of relevant chunks
         """
         effective_folder = self._merge_folders(additional_folders)
         payload = self._client._logic._prepare_retrieve_chunks_request(
-            query, filters, k, min_score, use_colpali, effective_folder, None
+            query, filters, k, min_score, use_colpali, effective_folder, None, padding
         )
         response = await self._client._request("POST", "retrieve/chunks", data=payload)
         return self._client._logic._parse_chunk_result_list_response(response)
@@ -826,6 +828,7 @@ class AsyncUserScope:
         min_score: float = 0.0,
         use_colpali: bool = True,
         additional_folders: Optional[List[str]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Retrieve relevant chunks as this end user.
@@ -837,13 +840,14 @@ class AsyncUserScope:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model
             additional_folders: Optional list of additional folder names to further scope operations
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
 
         Returns:
             List[FinalChunkResult]: List of relevant chunks
         """
         effective_folder = self._merge_folders(additional_folders)
         payload = self._client._logic._prepare_retrieve_chunks_request(
-            query, filters, k, min_score, use_colpali, effective_folder, self._end_user_id
+            query, filters, k, min_score, use_colpali, effective_folder, self._end_user_id, padding
         )
         response = await self._client._request("POST", "retrieve/chunks", data=payload)
         return self._client._logic._parse_chunk_result_list_response(response)
@@ -1478,6 +1482,7 @@ class AsyncMorphik:
         min_score: float = 0.0,
         use_colpali: bool = True,
         folder_name: Optional[Union[str, List[str]]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Search for relevant chunks.
@@ -1489,6 +1494,7 @@ class AsyncMorphik:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model to retrieve chunks
                 (only works for documents ingested with `use_colpali=True`)
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
         Returns:
             List[FinalChunkResult]
 
@@ -1496,13 +1502,14 @@ class AsyncMorphik:
             ```python
             chunks = await db.retrieve_chunks(
                 "What are the key findings?",
-                filters={"department": "research"}
+                filters={"department": "research"},
+                padding=2  # Get 2 pages before and after each matched page
             )
             ```
         """
         effective_folder = folder_name if folder_name is not None else None
         payload = self._logic._prepare_retrieve_chunks_request(
-            query, filters, k, min_score, use_colpali, effective_folder, None
+            query, filters, k, min_score, use_colpali, effective_folder, None, padding
         )
         response = await self._request("POST", "retrieve/chunks", data=payload)
         return self._logic._parse_chunk_result_list_response(response)
@@ -2575,7 +2582,7 @@ class AsyncMorphik:
         self,
         graph_name: str,
         timeout_seconds: int = 300,
-        check_interval_seconds: int = 5,
+        check_interval_seconds: int = 2,
     ) -> Graph:
         """Block until the specified graph finishes processing (async).
 
@@ -2662,6 +2669,30 @@ class AsyncMorphik:
         params = {"run_id": run_id} if run_id else None
         return await self._request("GET", f"graph/workflow/{workflow_id}/status", params=params)
 
+    async def get_graph_status(
+        self, graph_name: str, folder_name: Optional[str] = None, end_user_id: Optional[str] = None
+    ) -> Dict[str, Any]:
+        """Get the current status of a graph with pipeline stage information.
+
+        This is a lightweight endpoint that checks local database status and
+        optionally syncs with external workflow status if the graph is processing.
+
+        Args:
+            graph_name: Name of the graph to check
+            folder_name: Optional folder name for scoping
+            end_user_id: Optional end user ID for scoping
+
+        Returns:
+            Dict containing status, pipeline_stage (if processing), and other metadata
+        """
+        params = {}
+        if folder_name:
+            params["folder_name"] = folder_name
+        if end_user_id:
+            params["end_user_id"] = end_user_id
+
+        return await self._request("GET", f"graph/{graph_name}/status", params=params if params else None)
+
     # ------------------------------------------------------------------
     # Document download helpers ----------------------------------------
     # ------------------------------------------------------------------

morphik/models.py

@@ -317,7 +317,7 @@ class Graph(BaseModel):
     def error(self) -> str | None:
         return self.system_metadata.get("error") if self.system_metadata else None
 
-    def wait_for_completion(self, timeout_seconds: int = 300, check_interval_seconds: int = 5) -> "Graph":
+    def wait_for_completion(self, timeout_seconds: int = 300, check_interval_seconds: int = 2) -> "Graph":
         """Poll the server until the graph processing is finished."""
         import time
 

morphik/sync.py

@@ -289,6 +289,7 @@ class Folder:
         min_score: float = 0.0,
         use_colpali: bool = True,
         additional_folders: Optional[List[str]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Retrieve relevant chunks within this folder.
@@ -300,21 +301,16 @@ class Folder:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model
             additional_folders: Optional list of extra folders to include in the scope
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
 
         Returns:
             List[FinalChunkResult]: List of relevant chunks
         """
         effective_folder = self._merge_folders(additional_folders)
-        request = {
-            "query": query,
-            "filters": filters,
-            "k": k,
-            "min_score": min_score,
-            "use_colpali": use_colpali,
-            "folder_name": effective_folder,
-        }
-
-        response = self._client._request("POST", "retrieve/chunks", request)
+        payload = self._client._logic._prepare_retrieve_chunks_request(
+            query, filters, k, min_score, use_colpali, effective_folder, None, padding
+        )
+        response = self._client._request("POST", "retrieve/chunks", payload)
         return self._client._logic._parse_chunk_result_list_response(response)
 
     def retrieve_docs(
@@ -864,6 +860,7 @@ class UserScope:
         min_score: float = 0.0,
         use_colpali: bool = True,
         additional_folders: Optional[List[str]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Retrieve relevant chunks as this end user.
@@ -875,26 +872,16 @@ class UserScope:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model
             additional_folders: Optional list of extra folders to include in the scope
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
 
         Returns:
             List[FinalChunkResult]: List of relevant chunks
         """
         effective_folder = self._merge_folders(additional_folders)
-        request = {
-            "query": query,
-            "filters": filters,
-            "k": k,
-            "min_score": min_score,
-            "use_colpali": use_colpali,
-            "end_user_id": self._end_user_id,  # Add end user ID here
-            "folder_name": effective_folder,  # Add folder name if provided
-        }
-
-        # Add folder name if scoped to a folder
-        if self._folder_name:
-            request["folder_name"] = self._folder_name
-
-        response = self._client._request("POST", "retrieve/chunks", request)
+        payload = self._client._logic._prepare_retrieve_chunks_request(
+            query, filters, k, min_score, use_colpali, effective_folder, self._end_user_id, padding
+        )
+        response = self._client._request("POST", "retrieve/chunks", payload)
         return self._client._logic._parse_chunk_result_list_response(response)
 
     def retrieve_docs(
@@ -1623,6 +1610,7 @@ class Morphik:
         min_score: float = 0.0,
         use_colpali: bool = True,
         folder_name: Optional[Union[str, List[str]]] = None,
+        padding: int = 0,
     ) -> List[FinalChunkResult]:
         """
         Retrieve relevant chunks.
@@ -1634,6 +1622,7 @@ class Morphik:
             min_score: Minimum similarity threshold (default: 0.0)
             use_colpali: Whether to use ColPali-style embedding model to retrieve the chunks
                 (only works for documents ingested with `use_colpali=True`)
+            padding: Number of additional chunks/pages to retrieve before and after matched chunks (ColPali only, default: 0)
         Returns:
             List[ChunkResult]
 
@@ -1646,7 +1635,7 @@ class Morphik:
             ```
         """
         payload = self._logic._prepare_retrieve_chunks_request(
-            query, filters, k, min_score, use_colpali, folder_name, None
+            query, filters, k, min_score, use_colpali, folder_name, None, padding
         )
         response = self._request("POST", "retrieve/chunks", data=payload)
         return self._logic._parse_chunk_result_list_response(response)
@@ -2748,7 +2737,7 @@ class Morphik:
         self,
         graph_name: str,
         timeout_seconds: int = 300,
-        check_interval_seconds: int = 5,
+        check_interval_seconds: int = 2,
     ) -> Graph:
         """Block until the specified graph finishes processing.
 
@@ -2852,10 +2841,34 @@ class Morphik:
         return self._request("GET", f"graph/{name}/visualization", params=params)
 
     def check_workflow_status(self, workflow_id: str, run_id: Optional[str] = None) -> Dict[str, Any]:
-        """Poll the status of an asynchronous graph build/update workflow."""
+        """Poll the status of an async graph build/update workflow."""
         params = {"run_id": run_id} if run_id else None
         return self._request("GET", f"graph/workflow/{workflow_id}/status", params=params)
 
+    def get_graph_status(
+        self, graph_name: str, folder_name: Optional[str] = None, end_user_id: Optional[str] = None
+    ) -> Dict[str, Any]:
+        """Get the current status of a graph with pipeline stage information.
+
+        This is a lightweight endpoint that checks local database status and
+        optionally syncs with external workflow status if the graph is processing.
+
+        Args:
+            graph_name: Name of the graph to check
+            folder_name: Optional folder name for scoping
+            end_user_id: Optional end user ID for scoping
+
+        Returns:
+            Dict containing status, pipeline_stage (if processing), and other metadata
+        """
+        params = {}
+        if folder_name:
+            params["folder_name"] = folder_name
+        if end_user_id:
+            params["end_user_id"] = end_user_id
+
+        return self._request("GET", f"graph/{graph_name}/status", params=params if params else None)
+
     # ------------------------------------------------------------------
     # Document download helpers ----------------------------------------
     # ------------------------------------------------------------------

{morphik-0.2.4.dist-info → morphik-0.2.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: morphik
-Version: 0.2.4
+Version: 0.2.5
 Summary: Morphik Python Client
 Author-email: Morphik <founders@morphik.ai>
 Requires-Python: >=3.8

{morphik-0.2.4.dist-info → morphik-0.2.5.dist-info}/RECORD

@@ -1,10 +1,10 @@
-morphik/__init__.py,sha256=cVuKREdIqRjF_k9B2QmVuIP3muiBa3o6JZJhrCD3gBg,242
-morphik/_internal.py,sha256=2BL2JDkzd2AegrHSMeCD6BS1bwXABQOTkECHJydeJ9c,19773
-morphik/async_.py,sha256=CXXaOPpM83W0GfeeGTNlTZEE54ssgwhnsOUmWHY3K-4,101709
+morphik/__init__.py,sha256=k5NWdg7h0isdg4FCBQjso3OBcDpMV6Nhu1lv5ZtN_kU,242
+morphik/_internal.py,sha256=kme9o2XH6R887sqNZzOmi-Q-pe2MY6CLSlVT35ZfUzY,19864
+morphik/async_.py,sha256=4xME77Ib7thEMqPJEEvBHktptGItoSXhr5LGPLSgTCs,103237
 morphik/exceptions.py,sha256=v4XGmfq5B0KrZEF6M1ID8A50-45-SRAQZTrXGXM6n0Q,260
-morphik/models.py,sha256=JFMeGLbEke-Bls08JXCBv3y_z4eI3PBW_mbX-989i5I,20942
+morphik/models.py,sha256=kbQtPgMZmc8IwF_-S8DyjIeijntTyQsUl1PmUku6SVM,20942
 morphik/rules.py,sha256=z3YUx0f_b9O7BLbsevAkyMiQg03iUjwfx6OCh7xGKF0,3344
-morphik/sync.py,sha256=klOsUYc6eoqZ4InGiipVFP9Y7FGGAW5zuGSjBprhuM0,106500
+morphik/sync.py,sha256=RwJhVZa1QKpXrDIrGrZreJGSNUk-ToCjVjwcw9ikakk,107566
 morphik/tests/README.md,sha256=jtJDDK8cS5E4SbygFQDy7t6Y-kQwNYtZajRwVJDR62U,1069
 morphik/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 morphik/tests/example_usage.py,sha256=ls8n7355q-8gY43pZLKd4SzI-01MdFeXbT8bZ4U8MCg,11561
@@ -13,6 +13,6 @@ morphik/tests/test_sync.py,sha256=Reqa25Q259mCr-tzWzc1RDcs5KZuDfBkJRKOyyxhDtE,13
 morphik/tests/test_docs/sample1.txt,sha256=Fx6TElSiKdxyFeBp1iHthzHctFVZm38DrqcbdZMoidY,507
 morphik/tests/test_docs/sample2.txt,sha256=PE97gPv59J27A7CSNvi_0tRBIN3Mj6pyTFElCLfs3TE,686
 morphik/tests/test_docs/sample3.txt,sha256=OzrnJ_XsDUntEV0jk-ansa3_KIa6GnpvS5EVmlh6BHo,732
-morphik-0.2.4.dist-info/METADATA,sha256=RjNufGEb_wpEgOLbu9RGFq6YoqrW1aYKlnu09n0WPxg,3377
-morphik-0.2.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-morphik-0.2.4.dist-info/RECORD,,
+morphik-0.2.5.dist-info/METADATA,sha256=_L2GQtwTLb1wC5W0tb0AajHXhurlLOPq_Hoodk-4nu0,3377
+morphik-0.2.5.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+morphik-0.2.5.dist-info/RECORD,,