mdify-cli 2.11.5__tar.gz → 2.11.7__tar.gz

{mdify_cli-2.11.5/mdify_cli.egg-info → mdify_cli-2.11.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdify-cli
-Version: 2.11.5
+Version: 2.11.7
 Summary: Convert PDFs and document images into structured Markdown for LLM workflows
 Author: tiroq
 License-Expression: MIT

{mdify_cli-2.11.5 → mdify_cli-2.11.7}/mdify/__init__.py

@@ -1,3 +1,3 @@
 """mdify - Convert documents to Markdown via Docling container."""
 
-__version__ = "2.11.5"
+__version__ = "2.11.7"

{mdify_cli-2.11.5 → mdify_cli-2.11.7}/mdify/cli.py

@@ -752,6 +752,13 @@ Examples:
         help="Conversion timeout in seconds (default: 1200, can be set via MDIFY_TIMEOUT env var)",
     )
 
+    parser.add_argument(
+        "--memory",
+        type=str,
+        default=None,
+        help="Container memory limit (e.g., 2g, 512m, 4096m). Default: no limit",
+    )
+
     # Utility options
     parser.add_argument(
         "--check-update",
@@ -961,6 +968,7 @@ def main() -> int:
             args.port,
             timeout=timeout,
             keep_container=DEBUG,
+            memory=args.memory,
         ) as container:
             # Convert files
             conversion_start = time.time()
@@ -1020,6 +1028,60 @@ def main() -> int:
                                 f"{progress} {input_file.name} ✗ ({format_duration(elapsed)})"
                             )
                             print(f"    Error: {error_msg}", file=sys.stderr)
+                            
+                            # Check if it's a connection error and retrieve logs
+                            is_connection_error = "Connection refused" in error_msg or "Connection aborted" in error_msg or "RemoteDisconnected" in error_msg
+                            if is_connection_error:
+                                container_alive = container.is_ready()
+                                if container_alive:
+                                    print(
+                                        "    Connection lost (server may have crashed and restarted)",
+                                        file=sys.stderr,
+                                    )
+                                else:
+                                    print(
+                                        "    Container crashed while processing file",
+                                        file=sys.stderr,
+                                    )
+                                    print(
+                                        "    File may be too complex, large, or malformed",
+                                        file=sys.stderr,
+                                    )
+                                
+                                # Always show logs for connection errors
+                                print("    Retrieving container logs...", file=sys.stderr)
+                                logs, log_error = container.get_logs(tail=50)
+                                if logs:
+                                    print("    Container logs (last 50 lines):", file=sys.stderr)
+                                    for line in logs.strip().split("\n"):
+                                        if line.strip():
+                                            print(f"      {line}", file=sys.stderr)
+                                elif log_error:
+                                    print(f"    Error retrieving logs: {log_error}", file=sys.stderr)
+                                else:
+                                    print("    No logs available (container may have been removed)", file=sys.stderr)
+                                
+                                # Restart container if it crashed
+                                if not container_alive:
+                                    print("    Container crashed - attempting to restart...", file=sys.stderr)
+                                    try:
+                                        # Stop and remove the dead container
+                                        container.stop()
+                                        container.remove()
+                                        # Generate new container name to avoid conflicts
+                                        import uuid
+                                        container.container_name = f"mdify-serve-{uuid.uuid4().hex[:8]}"
+                                        # Start a new one
+                                        container.start(timeout=120)
+                                        print("    Container restarted successfully", file=sys.stderr)
+                                        print("    Continuing with next file...", file=sys.stderr)
+                                    except Exception as restart_error:
+                                        print(f"    Failed to restart container: {restart_error}", file=sys.stderr)
+                                        if DEBUG:
+                                            import traceback
+                                            traceback.print_exc()
+                                        print("    Stopping remaining conversions", file=sys.stderr)
+                                        break
                 except Exception as e:
                     elapsed = time.time() - start_time
                     failed_count += 1
@@ -1031,6 +1093,10 @@ def main() -> int:
                     error_msg = str(e)
                     is_connection_error = "Connection refused" in error_msg or "Connection aborted" in error_msg or "RemoteDisconnected" in error_msg
                     
+                    if DEBUG:
+                        print(f"    DEBUG: Exception caught: {type(e).__name__}", file=sys.stderr)
+                        print(f"    DEBUG: is_connection_error={is_connection_error}", file=sys.stderr)
+                    
                     if is_connection_error:
                         container_alive = container.is_ready()
                         if not args.quiet:
@@ -1054,13 +1120,21 @@ def main() -> int:
 
                             # Always show logs for connection errors to surface root cause
                             print("    Retrieving container logs...", file=sys.stderr)
-                            logs = container.get_logs(tail=30)
+                            logs, log_error = container.get_logs(tail=50)
                             if logs:
-                                print("    Container logs (last 30 lines):", file=sys.stderr)
+                                print("    Container logs (last 50 lines):", file=sys.stderr)
                                 for line in logs.strip().split("\n"):
-                                    print(f"      {line}", file=sys.stderr)
+                                    if line.strip():  # Skip empty lines
+                                        print(f"      {line}", file=sys.stderr)
+                            elif log_error:
+                                print(f"    Error retrieving logs: {log_error}", file=sys.stderr)
+                                if not DEBUG:
+                                    print(
+                                        "    Tip: re-run with MDIFY_DEBUG=1 to preserve container for inspection",
+                                        file=sys.stderr,
+                                    )
                             else:
-                                print("    No logs available", file=sys.stderr)
+                                print("    No logs available (container may have been removed)", file=sys.stderr)
                                 if not DEBUG:
                                     print(
                                         "    Tip: re-run with MDIFY_DEBUG=1 to preserve container logs",
@@ -1070,9 +1144,27 @@ def main() -> int:
                             if not container_alive:
                                 print("    Stopping remaining conversions", file=sys.stderr)
 
-                        # Stop processing if container is dead
+                        # Restart container if it crashed
                         if not container_alive:
-                            break
+                            print("    Container crashed - attempting to restart...", file=sys.stderr)
+                            try:
+                                # Stop and remove the dead container
+                                container.stop()
+                                container.remove()
+                                # Generate new container name to avoid conflicts
+                                import uuid
+                                container.container_name = f"mdify-serve-{uuid.uuid4().hex[:8]}"
+                                # Start a new one
+                                container.start(timeout=120)
+                                print("    Container restarted successfully", file=sys.stderr)
+                                print("    Continuing with next file...", file=sys.stderr)
+                            except Exception as restart_error:
+                                print(f"    Failed to restart container: {restart_error}", file=sys.stderr)
+                                if DEBUG:
+                                    import traceback
+                                    traceback.print_exc()
+                                print("    Stopping remaining conversions", file=sys.stderr)
+                                break
                     else:
                         # Non-connection error
                         if not args.quiet:

{mdify_cli-2.11.5 → mdify_cli-2.11.7}/mdify/container.py

@@ -27,6 +27,7 @@ class DoclingContainer:
         port: int = 5001,
         timeout: int = 1200,
         keep_container: bool = False,
+        memory: Optional[str] = None,
     ):
         """Initialize container manager.
 
@@ -36,12 +37,14 @@ class DoclingContainer:
             port: Host port to bind (default: 5001)
             timeout: Conversion timeout in seconds (default: 1200)
             keep_container: If True, do not auto-remove container (preserve logs)
+            memory: Memory limit (e.g., "2g", "512m"). None for no limit.
         """
         self.runtime = runtime
         self.image = image
         self.port = port
         self.timeout = timeout
         self.keep_container = keep_container
+        self.memory = memory
         self.container_name = f"mdify-serve-{uuid.uuid4().hex[:8]}"
         self.container_id: Optional[str] = None
 
@@ -110,6 +113,11 @@ class DoclingContainer:
         ]
         if not self.keep_container:
             cmd.insert(3, "--rm")  # Auto-remove on stop
+        
+        # Add memory limit if specified
+        if self.memory:
+            cmd.insert(3, self.memory)
+            cmd.insert(3, "-m")
 
         try:
             result = subprocess.run(cmd, capture_output=True, text=True, check=True)
@@ -135,28 +143,50 @@ class DoclingContainer:
                 check=False,
             )
 
-    def get_logs(self, tail: int = 50) -> str:
+    def remove(self) -> None:
+        """Remove container. Safe to call multiple times."""
+        if self.container_name:
+            subprocess.run(
+                [self.runtime, "rm", "-f", self.container_name],
+                capture_output=True,
+                check=False,
+            )
+
+    def get_logs(self, tail: int = 50) -> tuple[str, str]:
         """Get container logs for debugging.
 
         Args:
             tail: Number of lines to retrieve from end of logs
 
         Returns:
-            Container logs as string
+            Tuple of (stdout, stderr) from container logs
         """
         if not self.container_name:
-            return ""
+            return ("", "No container name set")
         
         try:
+            import os
+            runtime_name = os.path.basename(self.runtime)
+            
+            # Apple Container uses -n instead of --tail
+            if runtime_name == "container":
+                cmd = [self.runtime, "logs", "-n", str(tail), self.container_name]
+            else:
+                cmd = [self.runtime, "logs", "--tail", str(tail), self.container_name]
+            
             result = subprocess.run(
-                [self.runtime, "logs", "--tail", str(tail), self.container_name],
+                cmd,
                 capture_output=True,
                 text=True,
                 check=False,
             )
-            return result.stdout if result.returncode == 0 else ""
-        except Exception:
-            return ""
+            if result.returncode != 0:
+                return ("", f"Failed to get logs (exit {result.returncode}): {result.stderr}")
+            # Container logs come from both stdout and stderr
+            combined = result.stdout + result.stderr
+            return (combined, "")
+        except Exception as e:
+            return ("", f"Exception getting logs: {e}")
 
     def is_running(self) -> bool:
         """Check if container process is still running.

{mdify_cli-2.11.5 → mdify_cli-2.11.7/mdify_cli.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdify-cli
-Version: 2.11.5
+Version: 2.11.7
 Summary: Convert PDFs and document images into structured Markdown for LLM workflows
 Author: tiroq
 License-Expression: MIT

{mdify_cli-2.11.5 → mdify_cli-2.11.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "mdify-cli"
-version = "2.11.5"
+version = "2.11.7"
 description = "Convert PDFs and document images into structured Markdown for LLM workflows"
 readme = "README.md"
 requires-python = ">=3.8"