mdify-cli 3.0.7__tar.gz → 3.1.0__tar.gz

{mdify_cli-3.0.7 → mdify_cli-3.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdify-cli
-Version: 3.0.7
+Version: 3.1.0
 Summary: Convert PDFs and document images into structured Markdown for LLM workflows
 Author: tiroq
 License-Expression: MIT

{mdify_cli-3.0.7 → mdify_cli-3.1.0}/mdify/__init__.py

@@ -1,3 +1,3 @@
 """mdify - Convert documents to Markdown via Docling container."""
 
-__version__ = "3.0.7"
+__version__ = "3.1.0"

{mdify_cli-3.0.7 → mdify_cli-3.1.0}/mdify/cli.py

@@ -8,12 +8,14 @@ is lightweight and has no ML dependencies.
 """
 
 import argparse
+import asyncio
 import json
 import os
 import platform
 import shutil
 import subprocess
 import sys
+import tempfile
 import threading
 import time
 from pathlib import Path
@@ -883,7 +885,7 @@ Examples:
         "--timeout",
         type=int,
         default=None,
-        help="Conversion timeout in seconds (default: 1200, can be set via MDIFY_TIMEOUT env var)",
+        help="Conversion timeout in seconds (default: 1200s for local, 3600s for remote with large PDFs, can be set via MDIFY_TIMEOUT env var)",
     )
 
     parser.add_argument(
@@ -1057,6 +1059,10 @@ def main_async_remote(args) -> int:
         # Resolve timeout value: CLI > env > default 1200
         timeout = args.timeout or int(os.environ.get("MDIFY_TIMEOUT", 1200))
         
+        # For remote operations, extend timeout significantly for large PDF processing
+        # Remote conversions include network latency, file upload/download, and OCR processing
+        remote_conversion_timeout = max(timeout, 3600)  # At least 1 hour for remote conversion
+        
         # Build SSH config from CLI arguments and SSH config files
         try:
             # Build config with proper precedence (lowest to highest):
@@ -1178,7 +1184,8 @@ def main_async_remote(args) -> int:
                 return 1
             
             if not args.quiet:
-                print(color.cyan(f"\nFound {len(files_to_convert)} file(s) to convert"), file=sys.stderr)
+                print(color.cyan(f"Found {len(files_to_convert)} file(s) to convert"), file=sys.stderr)
+                print(color.cyan(f"Conversion timeout: {remote_conversion_timeout}s (for large PDFs with OCR)"), file=sys.stderr)
             
             # Import remote container and transfer manager
             from mdify.ssh.transfer import FileTransferManager
@@ -1242,7 +1249,8 @@ def main_async_remote(args) -> int:
                 if isinstance(exc, SSHConnectionError):
                     return True
                 msg = str(exc).lower()
-                return "broken pipe" in msg or "connection closed" in msg
+                # Errno 32 = Broken pipe, Errno 54 = Connection reset by peer
+                return any(x in msg for x in ["broken pipe", "connection closed", "connection reset", "errno 32", "errno 54", "ssh connection"])
             
             try:
                 for idx, input_file in enumerate(files_to_convert, 1):
@@ -1309,8 +1317,13 @@ def main_async_remote(args) -> int:
                             remote_output_path = f"{work_dir}/{input_file.stem}.md"
                             
                             # Build conversion command on remote - use -F for multipart form data
+                            # Important: use generous timeouts since large PDFs with OCR take time
+                            # --connect-timeout: max time to establish connection (60s)
+                            # --max-time: max total operation time (extended timeout)
                             convert_cmd = (
                                 f"curl -X POST "
+                                f"--connect-timeout 60 "
+                                f"--max-time {remote_conversion_timeout} "
                                 f"-F 'files=@{remote_file_path}' "
                                 f"-F 'to_formats=md' "
                                 f"-F 'do_ocr=true' "
@@ -1326,27 +1339,50 @@ def main_async_remote(args) -> int:
                             while conversion_attempt < 3 and not conversion_success:
                                 try:
                                     if conversion_attempt > 0 and not args.quiet:
-                                        print(f"  ↻ Conversion retry {conversion_attempt}...", file=sys.stderr)
+                                        # Exponential backoff: 2s, 4s, 8s
+                                        backoff_delay = 2 ** conversion_attempt
+                                        print(f"  ↻ Conversion retry {conversion_attempt} (waiting {backoff_delay}s for server recovery)...", file=sys.stderr)
+                                        await asyncio.sleep(backoff_delay)
                                     
-                                    conversion_output, _, conv_code = await ssh_client.run_command(convert_cmd, timeout=timeout)
+                                    conversion_output, _, conv_code = await ssh_client.run_command(convert_cmd, timeout=remote_conversion_timeout)
                                     
                                     if conv_code == 0:
                                         conversion_success = True
                                         break
                                     else:
-                                        conversion_attempt += 1
+                                        # Non-zero exit code - fail without retry for non-connection errors
+                                        break
                                 except Exception as conv_exc:
-                                    if is_connection_error(conv_exc) and conversion_attempt < 2:
+                                    is_conn_err = is_connection_error(conv_exc)
+                                    if is_conn_err and conversion_attempt < 2:
                                         conversion_attempt += 1
                                         if not args.quiet:
-                                            print(f"  ↻ Connection lost during conversion. Reconnecting (attempt {conversion_attempt})...", file=sys.stderr)
+                                            # Exponential backoff: 5s, 10s
+                                            backoff_delay = 5 * conversion_attempt
+                                            print(f"  ↻ Connection reset during conversion. Reconnecting in {backoff_delay}s...", file=sys.stderr)
+                                        
+                                        await asyncio.sleep(backoff_delay)
+                                        
                                         try:
                                             await ssh_client.disconnect()
                                         except Exception:
                                             pass
-                                        await ssh_client.connect()
+                                        
+                                        # Reconnect with retry
+                                        try:
+                                            await ssh_client.connect()
+                                        except Exception:
+                                            if not args.quiet:
+                                                print(f"  ⚠ Reconnection failed: retrying...", file=sys.stderr)
+                                            continue
                                     else:
-                                        conversion_attempt += 1
+                                        # Either not a connection error, or we've exhausted retries
+                                        if not args.quiet:
+                                            print(f"  [DEBUG] Breaking loop: not conn_err or exhausted retries", file=sys.stderr)
+                                        if conversion_attempt >= 2 and is_conn_err:
+                                            if not args.quiet:
+                                                print(f"  ↻ Connection error on final retry attempt", file=sys.stderr)
+                                        break
                             
                             if not conversion_success:
                                 print(f"  ✗ Failed: Conversion failed after {conversion_attempt} attempt(s)", file=sys.stderr)
@@ -1386,12 +1422,37 @@ def main_async_remote(args) -> int:
                                         # Ultimate fallback
                                         markdown_content = conversion_output
                                 
-                                # Write markdown content to remote file
-                                write_cmd = f"cat > {remote_output_path} << 'MDIFY_EOF'\n{markdown_content}\nMDIFY_EOF"
-                                _, _, write_code = await ssh_client.run_command(write_cmd, timeout=30)
+                                # Write markdown content to local temp file first, then upload via SFTP
+                                # (Piping large content through SSH here-documents can crash the connection)
+                                content_size_kb = len(markdown_content) / 1024
+                                if not args.quiet:
+                                    print(f"  {color.cyan('Writing')} {content_size_kb:.1f}KB markdown via SFTP...", file=sys.stderr)
                                 
-                                if write_code != 0:
-                                    print(f"  ✗ Failed to write markdown output", file=sys.stderr)
+                                try:
+                                    # Write to temporary local file
+                                    with tempfile.NamedTemporaryFile(mode='w', suffix='.md', delete=False) as temp_file:
+                                        temp_file.write(markdown_content)
+                                        temp_path = temp_file.name
+                                    
+                                    # Upload via SFTP (more reliable for large files)
+                                    await transfer_manager.upload_file(
+                                        local_path=temp_path,
+                                        remote_path=remote_output_path,
+                                        overwrite=True,
+                                        compress=False,
+                                    )
+                                    
+                                    # Cleanup temp file
+                                    try:
+                                        os.unlink(temp_path)
+                                    except Exception:
+                                        pass
+                                    
+                                    if not args.quiet:
+                                        print(f"  {color.green('✓')} Markdown written", file=sys.stderr)
+                                except Exception as write_exc:
+                                    if not args.quiet:
+                                        print(f"  ✗ Failed to write markdown: {write_exc}", file=sys.stderr)
                                     failed += 1
                                     break
                                 

{mdify_cli-3.0.7 → mdify_cli-3.1.0}/mdify_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdify-cli
-Version: 3.0.7
+Version: 3.1.0
 Summary: Convert PDFs and document images into structured Markdown for LLM workflows
 Author: tiroq
 License-Expression: MIT

{mdify_cli-3.0.7 → mdify_cli-3.1.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "mdify-cli"
-version = "3.0.7"
+version = "3.1.0"
 description = "Convert PDFs and document images into structured Markdown for LLM workflows"
 readme = "README.md"
 requires-python = ">=3.10"