Hyland.DocumentFilters.Linux-Musl-x64 25.2.0

Learn more about Target Frameworks and .NET Standard.

• • Hyland.DocumentFilters (>= 25.2.0)

# Enhancements

- Improved conversion of Microsoft Outlook files to Markdown with better mapping of lists and paragraphs. (DF-2365)
- Introduced a flexible OCR callback mechanism that allows AI and ML systems to integrate custom OCR engines—such as Python-based solutions like EasyOCR—directly into the Document Filters workflow. This enhancement enables users to invoke OCR functions in the language of their host application, providing greater control and adaptability for image-based document processing, with extracted text flowing seamlessly into standard Document Filters outputs like Markdown, PDF, HTML, and text. See the Document Filters documentation for implementation guidance and sample integrations. (DF-2265)
- Introduced a new `IGR_Open_Ex` callback, `IGR_OPEN_CALLBACK_ACTION_APPROVE_EXTERNAL_RESOURCE`, enabling applications to approve or deny the download of external images in HTML-based documents. (DF-1722)
- Introduced a new `IGR_Open_Ex` callback, `IGR_OPEN_CALLBACK_ACTION_GET_RESOURCE`, allowing applications to provide external resource bytes, such as images, when processing HTML-based documents. This enables greater flexibility by allowing developers to supply resource data directly via an `IGR_Stream`, eliminating dependencies on strict directory structures or URL-based downloads. (DF-2372)
- Introduced support for adding positional metadata to Markdown and JSON MDAST outputs, enabling AI and ML systems to map extracted content back to its original position in the source document. When enabled via "MARKDOWN_INCLUDE_LOCATIONS" or "JSON_INCLUDE_BOUNDS", positional data is embedded as comments in Markdown or as structured properties in JSON MDAST, enhancing contextual understanding for use cases like search, content analysis, and document review. See the Document Filters documentation for more details. (DF-2341)
- Introduced support for converting complex tables—including those with lists and other unsupported Markdown content—by embedding sanitized HTML tables within the Markdown output. This ensures complete and accurate content preservation without data loss, improving reliability for AI, data curation, and documentation workflows. Standard tables without complex content continue to render as native Markdown for optimal readability. (DF-2345)
- Introduced support for normalizing dashes into the content cleaning feature in Markdown and JSON outputs using the `normalize_dashes` cleaning function. This option normalizes dashes and characters that commonly serve as dashes into the ASCII minus sign (45/0x2D) to improve readability and machine-friendliness of generated content, making it ideal for downstream processing in AI/ML systems. (DF-2220)
- Introduced support to identify and extract tables from XFA PDF files, preserving the logical structure of tables, rows, and cells for enhanced AI data analysis.  It can be enabled using the `PDF_TABLE_DETECTION` option, defaulting to OFF. (DF-2196)
- Introduced the `MARKDOWN_TABLE_DETECT_TITLES` option to enable detection of table titles when the first row spans all columns. (DF-2206)
- Upgraded FreeType to 2.13.3. (DF-2446)
- Upgraded OpenJPEG to 2.5.3. (DF-2360)

# Updates

- HD Mode: Resolved a condition for DOCX files where numbered lists were sometimes skipping sequence numbers on nested levels. (DF-2290)
- HD Mode: Resolved a condition when converting an HTML file where some special characters would not have their size recognized correctly. (DF-2338)
- HD Mode: Resolved a condition where converting a PDF with an invalid page size results in an empty file. Introduced options `PDF_PAGE_SIZE_WIDTH` and `PDF_PAGE_SIZE_HEIGHT` to set the maximum allowed page width and height for PDF input. Both default to 14400 points (200 inches). When the limit is exceeded, Document Filters will return a descriptive error. (DF-2233)
- HD Mode: Resolved a condition where converting an XLSX file could fail with a "bad allocation" error. (DF-2335)
- HD Mode: Resolved an issue in Markdown conversion of Office documents where certain bullet lists could trigger the error: "pop() called with mismatched type". (DF-2435)
- HD Mode: Resolved an issue in RTF formatting where images were mislocated and some paragraphs could disappear. (DF-2373)
- HD Mode: Resolved an issue where paragraph borders were rendered improperly when placed inside a table. (DF-2120)
- HD Mode: Resolved an issue where tables disappeared when processing Open Office Impress files. (DF-2120)
- Resolved an issue in PDF table detection where header rows were not recognized if they were outside a bordered area. (DF-2368)
- Resolved an issue where Microsoft Office documents encrypted using SHA-1 could fail to decrypt. (DF-2430)
- Resolved an issue where references with special characters in XLS could cause "invalid string position" error. (DF-2351)
- Resolved an issue with HTML files that could cause truncation when a file contained improperly encoded multi-byte sequences (e.g., invalid UTF-8). The document may still display "broken" characters in place of the invalid sequences; correcting these requires changes to the source file. (DF-2427)
- Security: Chrome: CVE-2025-1918: not-affected: The vulnerable code is not compiled into Document Filters. (DF-2418)
- Security: FreeType: CVE-2025-27363: not-affected: The vulnerable code no longer exists in version 2.13.3 of the library. (DF-2400)
- Security: Giflib: CVE-2024-45993: not-exploitable: Issue exists in CLI, not in the library; does not impact Document Filters. (DF-2401)
- Security: Giflib: CVE-2025-31344: not-exploitable: Issue exists in CLI, not in the library; does not impact Document Filters.  (DF-2443)
- Security: OpenJPEG: CVE-2023-39327: patched: Applied security patch to address this issue. (DF-2417)
- Security: OpenJPEG: CVE-2023-39328: patched: Applied security patch to address this issue. (DF-2415)
- Security: OpenJPEG: CVE-2023-39329: patched: Applied security patch to address this issue. (DF-2414)
- Security: OpenJPEG: CVE-2024-56826: not-exploitable: Issue does not exist in version. (DF-2413)
- Security: OpenJPEG: CVE-2024-56827: not-exploitable: Issue does not exist in version. (DF-2412)
- Security: Protobuf: CVE-2024-24786: not-exploitable: Vulnerability requires access to Unmarshal, which is not reachable in Document Filters. (DF-2404)
- Security: Protobuf: CVE-2024-7254: not-exploitable: Vulnerability only affects Java Protobuf code; these parts are not incorporated into Document Filters. (DF-2403)
- Security: Skia: CVE-2024-43768: not-exploitable: Document Filters use zlib library directly to process PDF files, not Skia. (DF-2407)
- Security: Skia: CVE-2025-0436: not-exploitable: Document Filters does not use Skia library for HTML conversions. (DF-2405)
- Security: Skia: CVE-2025-0444: not-exploitable: Issue exists in Blink rendering engine; does not impact Document Filters. (DF-2406)
- Security: Snappy: CVE-2024-36114: not-exploitable: Issue exists for Aircompressor library; does not impact Document Filters. (DF-2408)
- Security: Tesseract: CVE-2024-29511: not-exploitable: Document Filters does not use Artifex Ghostscript software. (DF-2409)
- Security: Xpdf: CVE-2025-2574: patched: Applied security patch to address this issue. (DF-2419)
- Security: Xpdf: CVE-2025-3154: patched: Applied security patch to address this issue. (DF-2425)
- Security: bzip2: CVE-2024-3508: not-exploitable: Issue exists for Bombastic sBOM editor; does not impact Document Filters. (DF-2402)
- Security: zlib: CVE-2024-57923: not-exploitable: Document Filters does not use zlib for Btrfs or target s390 platforms. (DF-2429)
- Security: zlib: CVE-2025-0725: not-exploitable: Document Filters does not use zlib for processing of HTTP responses. (DF-2411)
- Security: zlib: CVE-2025-25293: not-exploitable: Document Filters does not use zlib for SAML processing. (DF-2410)