Version 1.10.0

What's Changed

• Replace editdistance with rapidfuzz for Levenshtein distance by @Belval in #452
• Fix undefined name 't2' in get_layout_csv_from_trp2 by @Belval in #453
• Add Table.to_list() for dependency-free row export by @Belval in #454
• Add optional boto3 config parameter to Textractor by @Belval in #455
• Fix get_cells_by_type returning only COLUMN_HEADER cells by @Belval in #457
• Declare and test support for Python 3.13 and 3.14 by @Belval in #456
• Fix failing Documentation CI deploy by @Belval in #458
• Version 1.10.0 by @Belval in #459

Full Changelog: v1.9.2...v1.10.0

Version 1.9.2

What's Changed

• Ignore KV elements in LAYOUT_LIST by @Belval in #425

Full Changelog: v1.9.1...v1.9.2

Version 1.9.1

What's Changed

• Fix s3 client instantiation in _get_document_images_from_path by @Belval in #423

Full Changelog: v1.9.0...v1.9.1

Version 1.9.0

What's Changed

• Add function to split larger element at table insertion by @Belval in #417
• Replace logging calls with module logger by @Belval in #419
• Fix short ids using long ids and vice-versa by @Belval in #418
• Add separator parameter to kv_to_csv by @DGarbs51 in #407
• export_kv_to_pandas by @Chuukwudi in #405

Full Changelog: v1.8.5...v1.9.0

Version 1.8.5

What's Changed

• Fix bug in convert that caused an exception on empty pages.

Full Changelog: v1.8.4...v1.8.5

Version 1.8.4

What's Changed

• Add check for None bounding boxes for AnalyzeExpense by @Belval
• Allow Custom Separator in Document.export_kv_to_csv() by @Chuukwudi
• Update analyze_document type hint by @ryangamble
• Fix invalid escape in BoundingBox docstring by @simonschmidt in #395

Full Changelog: v1.8.3...v1.8.4

Version 1.8.3

What's Changed

• Id in html output by @Belval in #386
• Escape html output by @Belval in #387
• Fix table indexing returning too many cells

⚠️ Breaking changes

• To support ids in HTML, layout Table created for TABLE predictions will no longer share the same ID as the table.

Full Changelog: v1.8.2...v1.8.3

Version 1.8.2

What's Changed

• Fix pypdfium2 failing to parse PDFs in bytearray format by @Belval

Full Changelog: v1.8.1...v1.8.2

Version 1.8.1

What's Changed

• Fix .to_markdown() raising an exception on missing local config by @Belval in #381

Full Changelog: v1.8.0...v1.8.1

Version 1.8.0

What's Changed

• Improve HTML linearization
  • Add HTML table linearization format that uses merged cells information for colspan and rowspan
  • Add prefix and suffix for LAYOUT_FOOTER and LAYOUT_ENTITY
  • Add <html><body>...</body></html> to the output when calling Document.to_html()
• Use pypdfium2 for PDF rasterization when available instead of pdf2image. This allows for better portability as the former does not have a dependency on OS libraries and should work out of the box with Lambda and SageMaker.
• Fix expenses with no summary fields
• Replace region mismatch with invalid S3 object exception

Backward-incompatible changes

• This update removes s3_output_path from the synchronous functions as s3_output_path is not a supported parameter for the Textract Synchronous API
• This update changes the exception raised by the textractor.py functions which will no longer raise RegionMismatchError (which is however kept in textractor.exceptions for backward compatibility.
• This update removes confidence_score from KeyValue entities in favour of _confidence which is used for all other entities.

Full Changelog: v1.7.12...v1.8.0