Changes in version 1.2.1                        

  - Fixed bug in make_hocr()

                 Changes in version 1.2.0 (2025-11-18)                  

  - Minor refactoring

                        Changes in version 1.1.9                        

  - Complete overhaul of tests for much better coverage.
  - Small changes to checks throughout to better handle edge cases.
  - Refactoring of xml.R

                        Changes in version 1.1.1                        

  - Changed get_text() to handle empty pages gracefully.
  - Fixed bugs in get_processor_versions(), get_ids_by_type(), and
    get_versions_by_type().
  - Updated URLs in documentation.

                        Changes in version 1.1.0                        

  - Added two new functions to facilitate processor management:
    get_ids_by_type() and get_versions_by_type().
  - Added vignette on processor management.

                 Changes in version 1.0.1 (2024-11-13)                  

  - Bug fixes.

                 Changes in version 1.0.0 (2024-02-12)                  

  - Major revision.
  - Added several processor-related functions following an update of the
    Google Document AI REST API. These include list_processor_types(),
    create_processor(), enable_processor(), disable_processor(), and
    delete_processor().
  - Added get_text() and get_tables(), which replace
    text_from_dai_response(), text_from_dai_file(),
    tables_from_dai_response() and tables_from_dai_file().
  - Added get_entities() and draw_entities().
  - Removed dai_tab_sync() and dai_tab_async() following Google's
    discontinuation of the v1beta2 endpoint on 31 January 2024.
  - Modified the parameters of the draw*() functions for better
    consistency with other functions.
  - Renamed the .R files and regrouped the functions.
  - Added prettier messages using the cli package.

                 Changes in version 0.9.9 (2023-09-07)                  

  - Added the function make_hocr to convert DAI output to hOCR files,
    thereby facilitating the creation of searchable PDFs.

                        Changes in version 0.9.8                        

  - Changed the build_token_df() and build_block_df() functions so they
    can take as input response objects from dai_sync() in addition to
    json files from dai_async().
  - Changed the build_token_df() and build_block_df() functions to
    include confidence scores in the dataframe, so as to enable
    filtering on confidence.

                        Changes in version 0.9.7                        

  - Changed the "draw_*" functions so they work with response objects
    (from dai_sync() and dai_sync_tab()) as well as with json files from
    dai_async_tab().
  - Changed the "draw_*" functions to allow customizing color and
    thickness of lines around bounding boxes.
  - Modified tables_from_dai_response() so that it handles response
    objects from dai_sync() with form parser processors.

                        Changes in version 0.9.6                        

  - Added three new functions relating to processors: get_processors(),
    get_processor_info(), and get_processor_versions().
  - Added parameter proc_v to dai_sync() and dai_async, allowing for
    specification of processor version.

                        Changes in version 0.9.5                        

  - Added two new functions: dai_notify() and merge_shards()
  - Modified text_from_dai_response() and text_from_dai_file() to allow
    saving the output straight to a text file.
  - Fixed a bug in dai_status() that caused an error when processing
    responses from the v1beta2 endpoint (dai_tab_async()).

                        Changes in version 0.9.3                        

  - Changed the "draw_*" functions to allow custom output filenames.

                        Changes in version 0.9.2                        

  - Added more support documentation in connection with JOSS release.

                        Changes in version 0.9.1                        

  - Added new function (draw_blocks_new()) to inspect block bounding
    boxes after reprocessing.

                 Changes in version 0.9.0 (2021-06-11)                  

  - Initial CRAN version.

                        Changes in version 0.8.0                        

  - Simplified the auth functions to avoid modifying the global
    environment. Tokens are no longer held in an internal .auth object.
    Removed dai_has_token(), dai_deauth(), and create_folder().

                        Changes in version 0.7.0                        

  - Revised draw_blocks(), draw_paragraphs(), draw_lines(), and
    draw_tokens() functions. These functions no longer require supplying
    a pdf file, as they get the images from a base64-encoded string in
    the json file.

                        Changes in version 0.6.0                        

  - New processing functions adapted to the new stable release of
    Document AI API (v1). The dai_sync()/dai_async() functions now
    access the new v1 endpoint, which has foreign language support.
    However, the v1 endpoint currently does not support table
    extraction, so the old processing functions (which access the
    v1beta2 endpoint) are kept under the new names dai_sync_tab() and
    dai_async_tab(). I expect this to be a temporary solution until
    DAI's capabilities are consolidated in a single endpoint, at which
    stage the *_tab() functions will be phased out.

                        Changes in version 0.4.0                        

  - New table extraction functions.

                        Changes in version 0.2.0                        

  - New helper functions and more robust code.

                        Changes in version 0.1.0                        

  - Revised and expanded auth functions.

                        Changes in version 0.0.1                        

  - Initial Github release.