airbyte-agent-mcp 0.1.64__py3-none-any.whl → 0.1.68__py3-none-any.whl

airbyte_agent_mcp/_vendored/connector_sdk/executor/local_executor.py

@@ -495,6 +495,14 @@ class LocalExecutor:
                 print(result.data)
         """
         try:
+            # Check for hosted-only actions before converting to Action enum
+            if config.action == "search":
+                raise NotImplementedError(
+                    "search is only available in hosted execution mode. "
+                    "Initialize the connector with external_user_id, airbyte_client_id, "
+                    "and airbyte_client_secret to use this feature."
+                )
+
             # Convert config to internal format
             action = Action(config.action) if isinstance(config.action, str) else config.action
             params = config.params or {}
@@ -1214,15 +1222,22 @@ class LocalExecutor:
     def _extract_metadata(
         self,
         response_data: dict[str, Any],
+        response_headers: dict[str, str],
         endpoint: EndpointDefinition,
     ) -> dict[str, Any] | None:
         """Extract metadata from response using meta extractor.
 
-        Each field in meta_extractor dict is independently extracted using JSONPath.
+        Each field in meta_extractor dict is independently extracted using JSONPath
+        for body extraction, or special prefixes for header extraction:
+        - @link.{rel}: Extract URL from RFC 5988 Link header by rel type
+        - @header.{name}: Extract raw header value by header name
+        - Otherwise: JSONPath expression for body extraction
+
         Missing or invalid paths result in None for that field (no crash).
 
         Args:
             response_data: Full API response (before record extraction)
+            response_headers: HTTP response headers
             endpoint: Endpoint with optional meta extractor configuration
 
         Returns:
@@ -1233,11 +1248,15 @@ class LocalExecutor:
         Example:
             meta_extractor = {
                 "pagination": "$.records",
-                "request_id": "$.requestId"
+                "request_id": "$.requestId",
+                "next_page_url": "@link.next",
+                "rate_limit": "@header.X-RateLimit-Remaining"
             }
             Returns: {
                 "pagination": {"cursor": "abc", "total": 100},
-                "request_id": "xyz123"
+                "request_id": "xyz123",
+                "next_page_url": "https://api.example.com/data?cursor=abc",
+                "rate_limit": "99"
             }
         """
         # Check if endpoint has meta extractor
@@ -1247,26 +1266,96 @@ class LocalExecutor:
         extracted_meta: dict[str, Any] = {}
 
         # Extract each field independently
-        for field_name, jsonpath_expr_str in endpoint.meta_extractor.items():
+        for field_name, extractor_expr in endpoint.meta_extractor.items():
             try:
-                # Parse and apply JSONPath expression
-                jsonpath_expr = parse_jsonpath(jsonpath_expr_str)
-                matches = [match.value for match in jsonpath_expr.find(response_data)]
-
-                if matches:
-                    # Return first match (most common case)
-                    extracted_meta[field_name] = matches[0]
+                if extractor_expr.startswith("@link."):
+                    # RFC 5988 Link header extraction
+                    rel = extractor_expr[6:]
+                    extracted_meta[field_name] = self._extract_link_url(response_headers, rel)
+                elif extractor_expr.startswith("@header."):
+                    # Raw header value extraction (case-insensitive lookup)
+                    header_name = extractor_expr[8:]
+                    extracted_meta[field_name] = self._get_header_value(response_headers, header_name)
                 else:
-                    # Path not found - set to None
-                    extracted_meta[field_name] = None
+                    # JSONPath body extraction
+                    jsonpath_expr = parse_jsonpath(extractor_expr)
+                    matches = [match.value for match in jsonpath_expr.find(response_data)]
+
+                    if matches:
+                        # Return first match (most common case)
+                        extracted_meta[field_name] = matches[0]
+                    else:
+                        # Path not found - set to None
+                        extracted_meta[field_name] = None
 
             except Exception as e:
                 # Log error but continue with other fields
-                logging.warning(f"Failed to apply meta extractor for field '{field_name}' with path '{jsonpath_expr_str}': {e}. Setting to None.")
+                logging.warning(f"Failed to apply meta extractor for field '{field_name}' with expression '{extractor_expr}': {e}. Setting to None.")
                 extracted_meta[field_name] = None
 
         return extracted_meta
 
+    @staticmethod
+    def _extract_link_url(headers: dict[str, str], rel: str) -> str | None:
+        """Extract URL from RFC 5988 Link header by rel type.
+
+        Parses Link header format: <url>; param1="value1"; rel="next"; param2="value2"
+
+        Supports:
+        - Multiple parameters per link in any order
+        - Both quoted and unquoted rel values
+        - Multiple links separated by commas
+
+        Args:
+            headers: Response headers dict
+            rel: The rel type to extract (e.g., "next", "prev", "first", "last")
+
+        Returns:
+            The URL for the specified rel type, or None if not found
+        """
+        link_header = headers.get("Link") or headers.get("link", "")
+        if not link_header:
+            return None
+
+        for link_segment in re.split(r",(?=\s*<)", link_header):
+            link_segment = link_segment.strip()
+
+            url_match = re.match(r"<([^>]+)>", link_segment)
+            if not url_match:
+                continue
+
+            url = url_match.group(1)
+            params_str = link_segment[url_match.end() :]
+
+            rel_match = re.search(r';\s*rel="?([^";,]+)"?', params_str, re.IGNORECASE)
+            if rel_match and rel_match.group(1).strip() == rel:
+                return url
+
+        return None
+
+    @staticmethod
+    def _get_header_value(headers: dict[str, str], header_name: str) -> str | None:
+        """Get header value with case-insensitive lookup.
+
+        Args:
+            headers: Response headers dict
+            header_name: Header name to look up
+
+        Returns:
+            Header value or None if not found
+        """
+        # Try exact match first
+        if header_name in headers:
+            return headers[header_name]
+
+        # Case-insensitive lookup
+        header_name_lower = header_name.lower()
+        for key, value in headers.items():
+            if key.lower() == header_name_lower:
+                return value
+
+        return None
+
     def _validate_required_body_fields(self, endpoint: Any, params: dict[str, Any], action: Action, entity: str) -> None:
         """Validate that required body fields are present for CREATE/UPDATE operations.
 
@@ -1394,7 +1483,7 @@ class _StandardOperationHandler:
                 request_kwargs = self.ctx.determine_request_format(endpoint, body)
 
                 # Execute async HTTP request
-                response = await self.ctx.http_client.request(
+                response_data, response_headers = await self.ctx.http_client.request(
                     method=endpoint.method,
                     path=path,
                     params=query_params if query_params else None,
@@ -1403,10 +1492,10 @@ class _StandardOperationHandler:
                 )
 
                 # Extract metadata from original response (before record extraction)
-                metadata = self.ctx.executor._extract_metadata(response, endpoint)
+                metadata = self.ctx.executor._extract_metadata(response_data, response_headers, endpoint)
 
                 # Extract records if extractor configured
-                response = self.ctx.extract_records(response, endpoint)
+                response = self.ctx.extract_records(response_data, endpoint)
 
                 # Assume success with 200 status code if no exception raised
                 status_code = 200
@@ -1532,7 +1621,7 @@ class _DownloadOperationHandler:
                     request_format = self.ctx.determine_request_format(operation, request_body)
                     self.ctx.validate_required_body_fields(operation, params, action, entity)
 
-                    metadata_response = await self.ctx.http_client.request(
+                    metadata_response, _ = await self.ctx.http_client.request(
                         method=operation.method,
                         path=path,
                         params=query_params,
@@ -1547,7 +1636,7 @@ class _DownloadOperationHandler:
                     )
 
                     # Step 3: Stream file from extracted URL
-                    file_response = await self.ctx.http_client.request(
+                    file_response, _ = await self.ctx.http_client.request(
                         method="GET",
                         path=file_url,
                         headers=headers,
@@ -1555,7 +1644,7 @@ class _DownloadOperationHandler:
                     )
                 else:
                     # One-step direct download: stream file directly from endpoint
-                    file_response = await self.ctx.http_client.request(
+                    file_response, _ = await self.ctx.http_client.request(
                         method=operation.method,
                         path=path,
                         params=query_params,

airbyte_agent_mcp/_vendored/connector_sdk/http_client.py

@@ -421,10 +421,14 @@ class HTTPClient:
         headers: dict[str, str] | None = None,
         *,
         stream: bool = False,
-    ):
+    ) -> tuple[dict[str, Any], dict[str, str]]:
         """Execute a single HTTP request attempt (no retries).
 
         This is the core request logic, separated from retry handling.
+
+        Returns:
+            Tuple of (response_data, response_headers) for non-streaming requests.
+            For streaming requests, returns (response_object, response_headers).
         """
         # Ensure auth credentials are initialized (proactive refresh if needed)
         await self._ensure_auth_initialized()
@@ -475,7 +479,7 @@ class HTTPClient:
                     status_code=status_code,
                     response_body=f"<binary content, {response.headers.get('content-length', 'unknown')} bytes>",
                 )
-                return response
+                return response, dict(response.headers)
 
             # Parse response - handle non-JSON responses gracefully
             content_type = response.headers.get("content-type", "")
@@ -501,7 +505,7 @@ class HTTPClient:
                 status_code=status_code,
                 response_body=response_data,
             )
-            return response_data
+            return response_data, dict(response.headers)
 
         except AuthenticationError as e:
             # Auth error (401, 403) - handle token refresh
@@ -631,7 +635,7 @@ class HTTPClient:
         *,
         stream: bool = False,
         _auth_retry_attempted: bool = False,
-    ):
+    ) -> tuple[dict[str, Any], dict[str, str]]:
         """Make an async HTTP request with optional streaming and automatic retries.
 
         Args:
@@ -644,8 +648,9 @@ class HTTPClient:
             stream: If True, do not eagerly read the body (useful for downloads)
 
         Returns:
-            - If stream=False: Parsed JSON (dict) or empty dict
-            - If stream=True: Response object suitable for streaming
+            Tuple of (response_data, response_headers):
+            - If stream=False: (parsed JSON dict or empty dict, response headers dict)
+            - If stream=True: (response object suitable for streaming, response headers dict)
 
         Raises:
             HTTPStatusError: If request fails with 4xx/5xx status after all retries

{airbyte_agent_mcp-0.1.64.dist-info → airbyte_agent_mcp-0.1.68.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: airbyte-agent-mcp
-Version: 0.1.64
+Version: 0.1.68
 Summary: MCP server that exposes Airbyte Connector SDK as MCP tools
 Author-email: Airbyte Support <support@airbyte.io>
 Requires-Python: >=3.13

{airbyte_agent_mcp-0.1.64.dist-info → airbyte_agent_mcp-0.1.68.dist-info}/RECORD

@@ -14,7 +14,7 @@ airbyte_agent_mcp/_vendored/connector_sdk/connector_model_loader.py,sha256=b88ao
 airbyte_agent_mcp/_vendored/connector_sdk/constants.py,sha256=AtzOvhDMWbRJgpsQNWl5tkogHD6mWgEY668PgRmgtOY,2737
 airbyte_agent_mcp/_vendored/connector_sdk/exceptions.py,sha256=ss5MGv9eVPmsbLcLWetuu3sDmvturwfo6Pw3M37Oq5k,481
 airbyte_agent_mcp/_vendored/connector_sdk/extensions.py,sha256=XWRRoJOOrwUHSKbuQt5DU7CCu8ePzhd_HuP7c_uD77w,21376
-airbyte_agent_mcp/_vendored/connector_sdk/http_client.py,sha256=NdccrrBHI5rW56XnXcP54arCwywIVKnMeSQPas6KlOM,27466
+airbyte_agent_mcp/_vendored/connector_sdk/http_client.py,sha256=S0eECrPVkjjro9xOhGA3QEwi2H0aWOwwoNZdfXd8OkE,27882
 airbyte_agent_mcp/_vendored/connector_sdk/introspection.py,sha256=2CyKXZHT74-1Id97uw1RLeyOi6TV24_hoNbQ6-6y7uI,10335
 airbyte_agent_mcp/_vendored/connector_sdk/secrets.py,sha256=J9ezMu4xNnLW11xY5RCre6DHP7YMKZCqwGJfk7ufHAM,6855
 airbyte_agent_mcp/_vendored/connector_sdk/types.py,sha256=CStkOsLtmZZdXylkdCsbYORDzughxygt1-Ucma0j-qE,8287
@@ -24,7 +24,7 @@ airbyte_agent_mcp/_vendored/connector_sdk/cloud_utils/__init__.py,sha256=4799Hv9
 airbyte_agent_mcp/_vendored/connector_sdk/cloud_utils/client.py,sha256=YxdRpQr9XjDzih6csSseBVGn9kfMtaqbOCXP0TPuzFY,7189
 airbyte_agent_mcp/_vendored/connector_sdk/executor/__init__.py,sha256=EmG9YQNAjSuYCVB4D5VoLm4qpD1KfeiiOf7bpALj8p8,702
 airbyte_agent_mcp/_vendored/connector_sdk/executor/hosted_executor.py,sha256=ydHcG-biRS1ITT5ELwPShdJW-KYpvK--Fos1ipNgHho,6995
-airbyte_agent_mcp/_vendored/connector_sdk/executor/local_executor.py,sha256=CMuknflYNY6_f83xrxvqewfI52MLYdPin3Rvz6HS3wU,67610
+airbyte_agent_mcp/_vendored/connector_sdk/executor/local_executor.py,sha256=_RQXpBVXMlyvI2sL-F5N82nBKq7osZIn6H27Q1fse0g,71225
 airbyte_agent_mcp/_vendored/connector_sdk/executor/models.py,sha256=lYVT_bNcw-PoIks4WHNyl2VY-lJVf2FntzINSOBIheE,5845
 airbyte_agent_mcp/_vendored/connector_sdk/http/__init__.py,sha256=y8fbzZn-3yV9OxtYz8Dy6FFGI5v6TOqADd1G3xHH3Hw,911
 airbyte_agent_mcp/_vendored/connector_sdk/http/config.py,sha256=6J7YIIwHC6sRu9i-yKa5XvArwK2KU60rlnmxzDZq3lw,3283
@@ -55,6 +55,6 @@ airbyte_agent_mcp/_vendored/connector_sdk/telemetry/__init__.py,sha256=RaLgkBU4d
 airbyte_agent_mcp/_vendored/connector_sdk/telemetry/config.py,sha256=tLmQwAFD0kP1WyBGWBS3ysaudN9H3e-3EopKZi6cGKg,885
 airbyte_agent_mcp/_vendored/connector_sdk/telemetry/events.py,sha256=8Y1NbXiwISX-V_wRofY7PqcwEXD0dLMnntKkY6XFU2s,1328
 airbyte_agent_mcp/_vendored/connector_sdk/telemetry/tracker.py,sha256=Ftrk0_ddfM7dZG8hF9xBuPwhbc9D6JZ7Q9qs5o3LEyA,5579
-airbyte_agent_mcp-0.1.64.dist-info/METADATA,sha256=4IimXK4qhjfTEjYo-Ya1aKq9Y7ABo2t7MExRZ1GOnRk,3009
-airbyte_agent_mcp-0.1.64.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-airbyte_agent_mcp-0.1.64.dist-info/RECORD,,
+airbyte_agent_mcp-0.1.68.dist-info/METADATA,sha256=lm4gWhk2YW_kH_UT1um7uUMhDZY8bhl49jygjPRW4_A,3009
+airbyte_agent_mcp-0.1.68.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+airbyte_agent_mcp-0.1.68.dist-info/RECORD,,