GroupDocs.Conversion for Python via .NET 26.9 Release Notes

This release upgrades GroupDocs.Conversion engine from 26.3 to 26.8. It adds the ConversionEvents API for observing a conversion as it runs, support for XAR archives and draw.io / Mermaid diagrams, and several new load and convert options. It also fixes a long-standing defect that made every delegate-typed property impossible to assign from Python.

The API surface is almost entirely additive, with two breaking changes to check before upgrading: the duplicate groupdocs.conversion.file_types module is gone (use groupdocs.conversion.filetypes), and PdfOptions.remove_pdf_acompliance is now remove_pdf_a_compliance. Both are one-line fixes — see API Changes.

Full List of Changes in This Release

Major Features

Conversion events

ConversionEvents lets you observe a conversion while it runs — progress, per-page and per-document completion, failures, and font substitution. Attach ordinary Python functions:

from groupdocs.conversion import ConversionEvents, Converter, ConverterSettings
from groupdocs.conversion.options.convert import PdfConvertOptions

events = ConversionEvents()
events.on_conversion_started = lambda: print("started")
events.on_conversion_progress = lambda percent: print(f"{percent}%")
events.on_document_converted = lambda ctx: print("document done")
events.on_font_substituted = lambda ctx: print(
    f"{ctx.original_font_name} -> {ctx.substitute_font_name} ({ctx.reason})")
events.on_conversion_completed = lambda: print("completed")

with Converter("business-plan.docx", ConverterSettings(), events) as converter:
    converter.convert("business-plan.pdf", PdfConvertOptions())

Note the constructor takes (path, settings, events) — there is no (path, events) overload, so pass a ConverterSettings() even when you do not need to configure anything.

The available hooks are on_conversion_started, on_conversion_progress, on_conversion_completed, on_document_converted, on_document_failed, on_page_converted, on_page_failed, on_compression_completed and on_font_substituted.

New formats

CAD layout scope

from groupdocs.conversion.options.load import CadLayoutScope, CadLoadOptions

load_options = CadLoadOptions()
load_options.layout_scope = CadLayoutScope.MODEL     # MODEL | LAYOUTS | BOTH

BOTH is the default and does not restrict the conversion.

API Changes

New classes

New members on existing classes

• FluentConverter.with_events(...), and the same on IConversionFrom and IConversionSettings
• Converter(...) — four new constructor overloads accepting a ConversionEvents
• CadLoadOptions.layout_scope
• EmailLoadOptions.page_layout_options
• WebLoadOptions.margin_settings, .orientation_settings, .size_settings
• MarkdownOptions.image_saving_callback
• ImageConvertOptions.min_resolution, .cap_resolution_to_page_content
• WordProcessingLoadOptions.auto_detect_rtl_direction
• CompressionFileType.XAR, DiagramFileType.DRAWIO, DiagramFileType.MMD

Removed — groupdocs.conversion.file_types

The module groupdocs.conversion.file_types and its 23 classes are gone. The classes themselves are not — every one of them still exists, unchanged, in groupdocs.conversion.filetypes.

file_types was a duplicate that shipped by accident. Both modules defined the same 23 classes and both mapped to the same underlying GroupDocs.Conversion.FileTypes.* types, but only filetypes was ever exported from the package root, and only filetypes was kept up to date. 26.9.0 drops the stale copy.

If you imported from it, change the import — nothing else:

# before
from groupdocs.conversion.file_types import WordProcessingFileType
# after
from groupdocs.conversion.filetypes import WordProcessingFileType

The API reference pages under /conversion/python-net/groupdocs.conversion.file_types/ are removed for the same reason; use groupdocs.conversion.filetypes.

Renamed — PdfOptions.remove_pdf_acompliance

# before
options.remove_pdf_acompliance = True
# after
options.remove_pdf_a_compliance = True

The underlying .NET property (RemovePdfACompliance) did not change; the Python spelling did, so that PDF/A reads as pdf_a rather than pdf + acompliance. This is the only renamed member in the release.

Fixed

Python 3.5 – 3.10 now actually work

The wheel has always declared python_requires = ">=3.5,<3.15". Until this release it did not deliver it: 26.5.0 works only on Python 3.11 and newer.

Two separate defects, both fixed:

• On 3.5 and 3.6 the generated sources used syntax those versions do not have (from __future__ import annotations, f-strings), so the package could not even be imported.
• On 3.7 – 3.10 an enum argument was serialised by name rather than by value, producing [FileType.DOCX, …] instead of [3, …] — not valid JSON, so the call was rejected before it reached the engine. Passing a FileType or any options enum is routine, so in practice most calls failed. From 3.11 Python returns the bare number from str() on an IntEnum, which is why newer versions were unaffected.

If you are on Python 3.10 or earlier, 26.9.0 is the first release you can actually use.

Indexers are callable again

Classes with a .NET indexer — every *DocumentInfo type and PossibleConversions — exposed it as a property named item. A property cannot take an index, so there was no way to actually use it. It is now a method plus the normal Python subscript:

info = converter.get_document_info()
value = info.get("Author")     # or: info["Author"]

An API-surface diff against 26.5.0 reports this as 60 removed properties. Nothing was lost — item was never callable.

Delegate-typed properties could not be assigned

Assigning a Python function to any property whose .NET type is a delegate raised:

GroupDocsConversionException: Object must implement IConvertible.

This affected 17 properties, including every ConversionEvents hook and, in earlier releases, ConverterSettings.on_conversion_failed, ConverterSettings.on_conversion_by_page_failed, ConverterSettings.on_compression_completed, WebLoadOptions.configure_headers, WebLoadOptions.credentials_provider, XmlLoadOptions.xsl_fo_factory, XmlLoadOptions.xslt_factory and EmailConvertOptions.attachment_content_handler. Those eight were affected in 26.5 and 26.6 as well; all of them now work.

Misleading type annotations on those properties

The same properties were annotated with the wrong type — on_page_converted was typed as taking a ConvertedPageContext when it takes a callback receiving one, and on_conversion_progress was typed int. They are now annotated Callable. This changes eight type annotations; it does not change any runtime behaviour that previously worked, because assignment previously raised.

Documentation

• The API reference now carries a Developer Guide section with task-oriented, runnable Python examples.
• Reference pages were added for groupdocs.conversion.integration.audio, .heic, .ocr and .video, which had no documentation previously.

Requirements

• Python 3.5 – 3.14 (python_requires = ">=3.5,<3.15") — every version in that range is exercised in CI; see Fixed if you are upgrading from 26.5 or earlier on Python 3.10 or below
• Windows, Linux or macOS (Intel and Apple Silicon) — no Microsoft Office or Adobe Acrobat required
• The .NET runtime is embedded in the wheel; no separate install is needed

Wheel Distribution

pip install groupdocs-conversion-net==26.9.0

Additional Resources

• Product page
• Documentation
• API reference
• Code examples

Feedback

Questions and issues are welcome on the GroupDocs.Conversion forum.