mdify-cli 3.0.7__py3-none-any.whl → 3.0.8__py3-none-any.whl

mdify/__init__.py

@@ -1,3 +1,3 @@
 """mdify - Convert documents to Markdown via Docling container."""
 
-__version__ = "3.0.7"
+__version__ = "3.0.8"

mdify/cli.py

@@ -883,7 +883,7 @@ 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 +1057,10 @@ 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 +1182,8 @@ 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
@@ -1309,8 +1314,13 @@ 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' "
@@ -1328,7 +1338,7 @@ def main_async_remote(args) -> int:
                                     if conversion_attempt > 0 and not args.quiet:
                                         print(f"  ↻ Conversion retry {conversion_attempt}...", file=sys.stderr)
                                     
-                                    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

{mdify_cli-3.0.7.dist-info → mdify_cli-3.0.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mdify-cli
-Version: 3.0.7
+Version: 3.0.8
 Summary: Convert PDFs and document images into structured Markdown for LLM workflows
 Author: tiroq
 License-Expression: MIT

{mdify_cli-3.0.7.dist-info → mdify_cli-3.0.8.dist-info}/RECORD

@@ -1,7 +1,7 @@
 assets/mdify.png,sha256=qUj7WXWqNwpI2KNXOW79XJwqFqa-UI0JEkmt1mmy4Rg,1820418
-mdify/__init__.py,sha256=NGfDaPTBZHjbvFgFVRcyOxA_eJgrlosmtH8KKYmakx4,90
+mdify/__init__.py,sha256=iUfoOWZ_3PErQD0XOXLNy8UyG2lK1eTPJFlW9l5NDdc,90
 mdify/__main__.py,sha256=bhpJ00co6MfaVOdH4XLoW04NtLYDa_oJK7ODzfLrn9M,143
-mdify/cli.py,sha256=5du0Eq5t8zW_rm0HY6etpWPgvSA0xuYe72k2UbW4vqA,77849
+mdify/cli.py,sha256=-HYbPCEjRU4iRcurfRpmMCdbHtQS6oNah-zajkzjojA,78732
 mdify/container.py,sha256=BjL5ZR__n1i_WHifXKllTPoqO7IuOUdPDo5esuNg0Iw,8213
 mdify/docling_client.py,sha256=xuQR6sC1v3EPloOSwExoHCqT4uUxE8myYq-Yeby3C2I,7975
 mdify/formatting.py,sha256=lJKhMbDPcaWCdyEa7aKwAm_desaWvkfDc8C3EP7LWp4,790
@@ -10,9 +10,9 @@ mdify/ssh/client.py,sha256=nO7gAQ6eWxIXFIOplW2F2ya0-1ZEFlLmz3ovi1TEFTg,14997
 mdify/ssh/models.py,sha256=jpbDS1yGhd7Xwq2tW7bZv14mTBlR8DCfhT4x-Xf2Wq4,17676
 mdify/ssh/remote_container.py,sha256=KB8rrsp3h5s_2BKPBXhzkOdQFafyPGQhSnFbK20b4yQ,10230
 mdify/ssh/transfer.py,sha256=aQuWa_B81McrgZBBWo_CxnjwoiGHmoxAoqObm19JAk8,10940
-mdify_cli-3.0.7.dist-info/licenses/LICENSE,sha256=NWM66Uv-XuSMKaU-gaPmvfyk4WgE6zcIPr78wyg6GAo,1065
-mdify_cli-3.0.7.dist-info/METADATA,sha256=7tq8pSgMPVs-CbhYvjO_pDYVPFFZKabiDLZG14UTbm8,14766
-mdify_cli-3.0.7.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-mdify_cli-3.0.7.dist-info/entry_points.txt,sha256=0Xki8f5lADQUtwdt6Eq_FEaieI6Byhk8UE7BuDhChMg,41
-mdify_cli-3.0.7.dist-info/top_level.txt,sha256=qltzf7h8owHq7dxCdfCkSHY8gT21hn1_E8P-VWS_OKM,6
-mdify_cli-3.0.7.dist-info/RECORD,,
+mdify_cli-3.0.8.dist-info/licenses/LICENSE,sha256=NWM66Uv-XuSMKaU-gaPmvfyk4WgE6zcIPr78wyg6GAo,1065
+mdify_cli-3.0.8.dist-info/METADATA,sha256=ap0J1JAN1eHngW32_OFd_5jQJrZrm4gcvCEcOMHNGNI,14766
+mdify_cli-3.0.8.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+mdify_cli-3.0.8.dist-info/entry_points.txt,sha256=0Xki8f5lADQUtwdt6Eq_FEaieI6Byhk8UE7BuDhChMg,41
+mdify_cli-3.0.8.dist-info/top_level.txt,sha256=qltzf7h8owHq7dxCdfCkSHY8gT21hn1_E8P-VWS_OKM,6
+mdify_cli-3.0.8.dist-info/RECORD,,