mcp-stata 1.7.6__py3-none-any.whl → 1.13.0__py3-none-any.whl

mcp_stata/streaming_io.py

@@ -2,11 +2,13 @@ import queue
 import threading
 import time
 from typing import Any, Awaitable, Callable, Optional
+import logging
 
 import anyio
 
 
 _SENTINEL = object()
+logger = logging.getLogger("mcp_stata")
 
 
 class StreamBuffer:
@@ -92,7 +94,7 @@ class StreamingTeeIO:
                 self._on_chunk_callback(text)
             except Exception:
                 # Don't let callback errors break streaming
-                pass
+                logger.debug("Streaming chunk callback failed", exc_info=True)
 
         with self._lock:
             if self._closed:

mcp_stata/test_stata.py

@@ -0,0 +1,54 @@
+import stata_setup
+stata_setup.config("/Applications/StataNow/", "mp")
+from pystata import stata
+import tempfile
+import os
+
+print("=== Testing multiple concurrent logs ===\n")
+
+# Create temp files for logs
+log1_path = tempfile.mktemp(suffix='.smcl')
+log2_path = tempfile.mktemp(suffix='.smcl')
+
+stata.run("sysuse auto, clear")
+
+try:
+    # Start first (unnamed) log - simulating user's log
+    print("1. Starting unnamed user log...")
+    stata.run(f'log using "{log1_path}", replace smcl')
+    
+    # Start second (named) log - our capture log
+    print("2. Starting named capture log...")
+    stata.run(f'log using "{log2_path}", replace smcl name(_capture)')
+    
+    # Run a command - should go to both logs
+    print("3. Running command...")
+    stata.run("summarize price mpg")
+    
+    # Close named log first
+    print("4. Closing named log...")
+    stata.run("log close _capture")
+    
+    # Close unnamed log
+    print("5. Closing unnamed log...")
+    stata.run("log close")
+    
+    print("\n=== SUCCESS: Multiple concurrent logs work! ===\n")
+    
+    # Show contents
+    print("--- User log contents (first 500 chars) ---")
+    with open(log1_path, 'r') as f:
+        print(f.read()[:500])
+    
+    print("\n--- Capture log contents (first 500 chars) ---")
+    with open(log2_path, 'r') as f:
+        print(f.read()[:500])
+
+except Exception as e:
+    print(f"\n=== FAILED: {e} ===\n")
+
+finally:
+    # Cleanup
+    for p in [log1_path, log2_path]:
+        if os.path.exists(p):
+            os.unlink(p)

mcp_stata/ui_http.py

@@ -4,11 +4,26 @@ import secrets
 import threading
 import time
 import uuid
+import logging
 from dataclasses import dataclass
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 from typing import Any, Callable, Optional
 
 from .stata_client import StataClient
+from .config import (
+    DEFAULT_HOST,
+    DEFAULT_PORT,
+    MAX_ARROW_LIMIT,
+    MAX_CHARS,
+    MAX_LIMIT,
+    MAX_REQUEST_BYTES,
+    MAX_VARS,
+    TOKEN_TTL_S,
+    VIEW_TTL_S,
+)
+
+
+logger = logging.getLogger("mcp_stata")
 
 
 def _stable_hash(payload: dict[str, Any]) -> str:
@@ -39,14 +54,15 @@ class UIChannelManager:
         self,
         client: StataClient,
         *,
-        host: str = "127.0.0.1",
-        port: int = 0,
-        token_ttl_s: int = 20 * 60,
-        view_ttl_s: int = 30 * 60,
-        max_limit: int = 500,
-        max_vars: int = 200,
-        max_chars: int = 500,
-        max_request_bytes: int = 1_000_000,
+        host: str = DEFAULT_HOST,
+        port: int = DEFAULT_PORT,
+        token_ttl_s: int = TOKEN_TTL_S,
+        view_ttl_s: int = VIEW_TTL_S,
+        max_limit: int = MAX_LIMIT,
+        max_vars: int = MAX_VARS,
+        max_chars: int = MAX_CHARS,
+        max_request_bytes: int = MAX_REQUEST_BYTES,
+        max_arrow_limit: int = MAX_ARROW_LIMIT,
     ):
         self._client = client
         self._host = host
@@ -57,6 +73,7 @@ class UIChannelManager:
         self._max_vars = max_vars
         self._max_chars = max_chars
         self._max_request_bytes = max_request_bytes
+        self._max_arrow_limit = max_arrow_limit
 
         self._lock = threading.Lock()
         self._httpd: ThreadingHTTPServer | None = None
@@ -87,7 +104,7 @@ class UIChannelManager:
             return UIChannelInfo(base_url=base_url, token=self._token or "", expires_at=self._expires_at)
 
     def capabilities(self) -> dict[str, bool]:
-        return {"dataBrowser": True, "filtering": True, "sorting": True}
+        return {"dataBrowser": True, "filtering": True, "sorting": True, "arrowStream": True}
 
     def current_dataset_id(self) -> str:
         with self._lock:
@@ -193,6 +210,7 @@ class UIChannelManager:
             manager = self
 
             class Handler(BaseHTTPRequestHandler):
+
                 def _send_json(self, status: int, payload: dict[str, Any]) -> None:
                     data = json.dumps(payload).encode("utf-8")
                     self.send_response(status)
@@ -201,7 +219,17 @@ class UIChannelManager:
                     self.end_headers()
                     self.wfile.write(data)
 
+                def _send_binary(self, status: int, data: bytes, content_type: str) -> None:
+                    self.send_response(status)
+                    self.send_header("Content-Type", content_type)
+                    self.send_header("Content-Length", str(len(data)))
+                    self.end_headers()
+                    self.wfile.write(data)
+
                 def _error(self, status: int, code: str, message: str, *, stata_rc: int | None = None) -> None:
+                    if status >= 500 or code == "internal_error":
+                        logger.error("UI HTTP error %s: %s", code, message)
+                        message = "Internal server error"
                     body: dict[str, Any] = {"error": {"code": code, "message": message}}
                     if stata_rc is not None:
                         body["error"]["stataRc"] = stata_rc
@@ -288,20 +316,31 @@ class UIChannelManager:
                     if not self._require_auth():
                         return
 
+
+                    if self.path == "/v1/arrow":
+                        body = self._read_json()
+                        if body is None:
+                            return
+                        try:
+                            resp_bytes = handle_arrow_request(manager, body, view_id=None)
+                            self._send_binary(200, resp_bytes, "application/vnd.apache.arrow.stream")
+                            return
+                        except HTTPError as e:
+                            self._error(e.status, e.code, e.message, stata_rc=e.stata_rc)
+                            return
+                        except Exception as e:
+                            self._error(500, "internal_error", str(e))
+                            return
+
                     if self.path == "/v1/page":
                         body = self._read_json()
                         if body is None:
                             return
-                        # Debug logging to diagnose limit parameter issues
-                        import sys
-                        print(f"[DEBUG] /v1/page request body: {body}", file=sys.stderr, flush=True)
-                        print(f"[DEBUG] limit value: {body.get('limit')!r} (type: {type(body.get('limit')).__name__})", file=sys.stderr, flush=True)
                         try:
                             resp = handle_page_request(manager, body, view_id=None)
                             self._send_json(200, resp)
                             return
                         except HTTPError as e:
-                            print(f"[DEBUG] HTTPError: {e.code} - {e.message}", file=sys.stderr, flush=True)
                             self._error(e.status, e.code, e.message, stata_rc=e.stata_rc)
                             return
                         except Exception as e:
@@ -354,16 +393,31 @@ class UIChannelManager:
                         body = self._read_json()
                         if body is None:
                             return
-                        # Debug logging to diagnose limit parameter issues
-                        import sys
-                        print(f"[DEBUG] /v1/views/{view_id}/page request body: {body}", file=sys.stderr, flush=True)
-                        print(f"[DEBUG] limit value: {body.get('limit')!r} (type: {type(body.get('limit')).__name__})", file=sys.stderr, flush=True)
                         try:
                             resp = handle_page_request(manager, body, view_id=view_id)
                             self._send_json(200, resp)
                             return
                         except HTTPError as e:
-                            print(f"[DEBUG] HTTPError: {e.code} - {e.message}", file=sys.stderr, flush=True)
+                            self._error(e.status, e.code, e.message, stata_rc=e.stata_rc)
+                            return
+                        except Exception as e:
+                            self._error(500, "internal_error", str(e))
+                            return
+
+                    if self.path.startswith("/v1/views/") and self.path.endswith("/arrow"):
+                        parts = self.path.split("/")
+                        if len(parts) != 5:
+                            self._error(404, "not_found", "Not found")
+                            return
+                        view_id = parts[3]
+                        body = self._read_json()
+                        if body is None:
+                            return
+                        try:
+                            resp_bytes = handle_arrow_request(manager, body, view_id=view_id)
+                            self._send_binary(200, resp_bytes, "application/vnd.apache.arrow.stream")
+                            return
+                        except HTTPError as e:
                             self._error(e.status, e.code, e.message, stata_rc=e.stata_rc)
                             return
                         except Exception as e:
@@ -593,3 +647,108 @@ def handle_page_request(manager: UIChannelManager, body: dict[str, Any], *, view
             "missing": ".",
         },
     }
+
+
+def handle_arrow_request(manager: UIChannelManager, body: dict[str, Any], *, view_id: str | None) -> bytes:
+    max_limit, max_vars, max_chars, _ = manager.limits()
+    # Use the specific Arrow limit instead of the general UI page limit
+    chunk_limit = getattr(manager, "_max_arrow_limit", 1_000_000)
+
+    if view_id is None:
+        dataset_id = str(body.get("datasetId", ""))
+        frame = str(body.get("frame", "default"))
+    else:
+        view = manager.get_view(view_id)
+        if view is None:
+            raise HTTPError(404, "not_found", "View not found")
+        dataset_id = view.dataset_id
+        frame = view.frame
+
+    # Parse offset (default 0)
+    try:
+        offset = int(body.get("offset") or 0)
+    except (ValueError, TypeError):
+        raise HTTPError(400, "invalid_request", "offset must be a valid integer")
+
+    # Parse limit (required)
+    limit_raw = body.get("limit")
+    if limit_raw is None:
+        # Default to the max arrow limit if not specified? 
+        # The previous code required it. Let's keep it required but allow large values.
+        raise HTTPError(400, "invalid_request", "limit is required")
+    try:
+        limit = int(limit_raw)
+    except (ValueError, TypeError):
+        raise HTTPError(400, "invalid_request", "limit must be a valid integer")
+
+    vars_req = body.get("vars", [])
+    include_obs_no = bool(body.get("includeObsNo", False))
+    sort_by = body.get("sortBy", [])
+
+    if offset < 0:
+        raise HTTPError(400, "invalid_request", "offset must be >= 0")
+    if limit <= 0:
+        raise HTTPError(400, "invalid_request", "limit must be > 0")
+    # Arrow streams are efficient, but we still respect a (much larger) max limit
+    if limit > chunk_limit:
+        raise HTTPError(400, "request_too_large", f"limit must be <= {chunk_limit}")
+
+    if not isinstance(vars_req, list) or not all(isinstance(v, str) for v in vars_req):
+        raise HTTPError(400, "invalid_request", "vars must be a list of strings")
+    if len(vars_req) > max_vars:
+        raise HTTPError(400, "request_too_large", f"vars length must be <= {max_vars}")
+
+    current_id = manager.current_dataset_id()
+    if dataset_id != current_id:
+        raise HTTPError(409, "dataset_changed", "Dataset changed")
+
+    if view_id is None:
+        obs_indices = None
+    else:
+        assert view is not None
+        obs_indices = view.obs_indices
+
+    try:
+        # Apply sorting if requested
+        if sort_by:
+            if not isinstance(sort_by, list) or not all(isinstance(s, str) for s in sort_by):
+                raise HTTPError(400, "invalid_request", "sortBy must be a list of strings")
+            try:
+                manager._client.apply_sort(sort_by)
+                if view_id is not None:
+                    # encapsulated re-computation if view is active
+                    # Note: original code only does this for view_id is not None
+                    # But if we sort global dataset, existing views might become invalid unless
+                    # they rely on stable indices. Stata indices change on sort.
+                    # The current implementation of create_view computes indices once.
+                    # If we sort, those indices point to different rows! 
+                    # The original code handles this by re-computing view indices on sort.
+                    assert view is not None
+                    obs_indices = manager._client.compute_view_indices(view.filter_expr)
+            except ValueError as e:
+                raise HTTPError(400, "invalid_request", f"Invalid sort: {e}")
+            except RuntimeError as e:
+                raise HTTPError(500, "internal_error", f"Sort failed: {e}")
+
+        arrow_bytes = manager._client.get_arrow_stream(
+            offset=offset,
+            limit=limit,
+            vars=vars_req,
+            include_obs_no=include_obs_no,
+            obs_indices=obs_indices,
+        )
+        return arrow_bytes
+
+    except RuntimeError as e:
+        msg = str(e) or "No data in memory"
+        if "no data" in msg.lower():
+            raise HTTPError(400, "no_data_in_memory", msg)
+        raise HTTPError(500, "internal_error", msg)
+    except ValueError as e:
+        msg = str(e)
+        if "invalid variable" in msg.lower():
+            raise HTTPError(400, "invalid_variable", msg)
+        raise HTTPError(400, "invalid_request", msg)
+    except Exception as e:
+        raise HTTPError(500, "internal_error", str(e))
+

{mcp_stata-1.7.6.dist-info → mcp_stata-1.13.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mcp-stata
-Version: 1.7.6
+Version: 1.13.0
 Summary:  A lightweight Model Context Protocol (MCP) server for Stata. Execute commands, inspect data, retrieve stored results (`r()`/`e()`), and view graphs in your chat interface. Built for economists who want to integrate LLM assistance into their Stata workflow. 
 Project-URL: Homepage, https://github.com/tmonk/mcp-stata
 Project-URL: Repository, https://github.com/tmonk/mcp-stata
@@ -21,6 +21,8 @@ Requires-Python: >=3.12
 Requires-Dist: httpx<0.28.0,>=0.27.0
 Requires-Dist: mcp[cli]>=1.0.0
 Requires-Dist: pandas>=2.0.0
+Requires-Dist: polars>=1.36.1
+Requires-Dist: pyarrow>=14.0.0
 Requires-Dist: pydantic>=2.0.0
 Requires-Dist: pystata>=0.0.1
 Requires-Dist: pytest-asyncio>=1.3.0
@@ -30,6 +32,7 @@ Requires-Dist: build>=1.3.0; extra == 'dev'
 Requires-Dist: hatch>=1.16.2; extra == 'dev'
 Requires-Dist: pytest-cov>=4.0.0; extra == 'dev'
 Requires-Dist: pytest>=7.0.0; extra == 'dev'
+Requires-Dist: python-semantic-release>=9.8.0; extra == 'dev'
 Requires-Dist: ruff>=0.4.0; extra == 'dev'
 Requires-Dist: twine>=6.2.0; extra == 'dev'
 Description-Content-Type: text/markdown
@@ -244,6 +247,9 @@ VS Code documents `.vscode/mcp.json` and the `servers` schema, including `type`
   - Always writes output to a temporary log file and emits a single `notifications/logMessage` containing `{"event":"log_path","path":"..."}` so the client can tail it locally.
   - May emit `notifications/progress` when the client provides a progress token/callback.
 * `read_log(path, offset=0, max_bytes=65536)`: Read a slice of a previously-provided log file (JSON: `path`, `offset`, `next_offset`, `data`).
+* `find_in_log(path, query, start_offset=0, max_bytes=5_000_000, before=2, after=2, case_sensitive=False, regex=False, max_matches=50)`: Search a log file for text and return context windows.
+  - Returns JSON with `matches` (context lines, line indices), `next_offset`, and `truncated` if `max_matches` is hit.
+  - Supports literal or regex search with bounded read window for large logs.
 * `load_data(source, clear=True, as_json=True, raw=False, max_output_lines=None)`: Heuristic loader (sysuse/webuse/use/path/URL) with JSON envelope unless `raw=True`. Supports output truncation.
 * `get_data(start=0, count=50)`: View dataset rows (JSON response, capped to 500 rows).
 * `get_ui_channel()`: Return a short-lived localhost HTTP endpoint + bearer token for the UI-only data browser.
@@ -301,7 +307,8 @@ Call the MCP tool `get_ui_channel()` and parse the JSON:
   "capabilities": {
     "dataBrowser": true,
     "filtering": true,
-    "sorting": true
+    "sorting": true,
+    "arrowStream": true
   }
 }
 ```
@@ -309,9 +316,10 @@ Call the MCP tool `get_ui_channel()` and parse the JSON:
 Server-enforced limits (current defaults):
 
 - **maxLimit**: 500
-- **maxVars**: 200
+- **maxVars**: 32,767
 - **maxChars**: 500
 - **maxRequestBytes**: 1,000,000
+- **maxArrowLimit**: 1,000,000 (specific to `/v1/arrow`)
 
 ### Endpoints
 
@@ -323,10 +331,14 @@ All endpoints are under `baseUrl` and require the bearer token.
   - Returns variable metadata (`name`, `type`, `label`, `format`).
 - `POST /v1/page`
   - Returns a page of data for selected variables.
+- `POST /v1/arrow`
+  - Returns a binary Arrow IPC stream (same input as `/v1/page`).
 - `POST /v1/views`
   - Creates a server-side filtered view (handle-based filtering).
 - `POST /v1/views/:viewId/page`
   - Pages within a filtered view.
+- `POST /v1/views/:viewId/arrow`
+  - Returns a binary Arrow IPC stream from a filtered view.
 - `DELETE /v1/views/:viewId`
   - Deletes a view handle.
 - `POST /v1/filters/validate`

mcp_stata-1.13.0.dist-info/RECORD

@@ -0,0 +1,16 @@
+mcp_stata/__init__.py,sha256=kJKKRn7lGuVCuS2-GaN5VoVcvnxtNlfuswW_VOlYqwg,98
+mcp_stata/config.py,sha256=SfXltpwO_gROABca1sm0xXDhaeRmlRfQmXcnBiG4GYk,714
+mcp_stata/discovery.py,sha256=rU2udO1vZ32ZpkqH6JXUBSGcfdqltK-_w0MQUyHQKLE,20576
+mcp_stata/graph_detector.py,sha256=E-EuE9cWBBziZSq2EDjM1qQxNZy7__q3l6gekq4wujI,16492
+mcp_stata/models.py,sha256=RpxUnvgaQxnoR1Oaj23C57RWzVPFOs53HRAtHiMtm9I,1311
+mcp_stata/server.py,sha256=LrJN4kc9aPKnpzmhZnv-j-uQ9zoQlKi00btzxklTyfY,37623
+mcp_stata/stata_client.py,sha256=jPKwEMnr7w6fH8ohsz-08XjGRIqMsRjRjrgzKtSFE6M,135298
+mcp_stata/streaming_io.py,sha256=1WLStbdYEQFO1P3gCchmD7frHpmn5mHi6DuynodfiqM,7087
+mcp_stata/test_stata.py,sha256=HF_P4IjvLGlihXggbt79MZZB6PbuYHpKI3S0R2dgCeE,1512
+mcp_stata/ui_http.py,sha256=vG6oO3fokSvWkIx3fcCb-tiqlkuFKtml2ptvQrZ-o5k,30457
+mcp_stata/smcl/smcl2html.py,sha256=wi91mOMeV9MCmHtNr0toihNbaiDCNZ_NP6a6xEAzWLM,2624
+mcp_stata-1.13.0.dist-info/METADATA,sha256=WW21u_KFj1qseJWEh83eYKGVHWSnZ1YtyUqe_9Ca2Jw,16727
+mcp_stata-1.13.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+mcp_stata-1.13.0.dist-info/entry_points.txt,sha256=TcOgrtiTL4LGFEDb1pCrQWA-fUZvIujDOvQ-bWFh5Z8,52
+mcp_stata-1.13.0.dist-info/licenses/LICENSE,sha256=DZak_2itbUtvHzD3E7GNUYSRK6jdOJ-GqncQ2weavLA,34523
+mcp_stata-1.13.0.dist-info/RECORD,,

mcp_stata-1.7.6.dist-info/RECORD

@@ -1,14 +0,0 @@
-mcp_stata/__init__.py,sha256=kJKKRn7lGuVCuS2-GaN5VoVcvnxtNlfuswW_VOlYqwg,98
-mcp_stata/discovery.py,sha256=jQN9uvBNHF_hCCU9k6BDtSdDxiUVpvXcOJwpWYwo55c,17430
-mcp_stata/graph_detector.py,sha256=-dJIU1Dq_c1eQSk4eegUi0gU2N-tFqjFGM0tE1E32KM,16066
-mcp_stata/models.py,sha256=EKFawioKBhtZhRQ3pFzrKV99ui9L-qzcAuRYuk0npVg,1235
-mcp_stata/server.py,sha256=PV8ragGMeHT72zgVx5DJp3vt8CPqT8iwdvJ8GXSctds,15989
-mcp_stata/stata_client.py,sha256=Yd8SxtLf_JVxeOnqOwEM6lNGdhw4T3v2tjVxpTCf9xU,96397
-mcp_stata/streaming_io.py,sha256=GVaXgTtxx8YLY6RWqdTcO2M3QSqxLsefqkmnlNO1nTI,6974
-mcp_stata/ui_http.py,sha256=w1tYxNuwuhkjyfWHxUnpd1DcVBaakjPkEnWr-Fo1lWo,24193
-mcp_stata/smcl/smcl2html.py,sha256=wi91mOMeV9MCmHtNr0toihNbaiDCNZ_NP6a6xEAzWLM,2624
-mcp_stata-1.7.6.dist-info/METADATA,sha256=J9ki7OKvuj3Ve_kmr6sWGkZq3eQg24iDkpwqr5J-iLI,15951
-mcp_stata-1.7.6.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-mcp_stata-1.7.6.dist-info/entry_points.txt,sha256=TcOgrtiTL4LGFEDb1pCrQWA-fUZvIujDOvQ-bWFh5Z8,52
-mcp_stata-1.7.6.dist-info/licenses/LICENSE,sha256=DZak_2itbUtvHzD3E7GNUYSRK6jdOJ-GqncQ2weavLA,34523
-mcp_stata-1.7.6.dist-info/RECORD,,