Regression Patch

Fixes two regression problems:

• Errors on multiple empty text elements.
• Missing code for external attributes.

Minor fixes

Just some minor fixes. Most importantly:

• Fixes the problem with missing xm-mapping package.

Other things:

• JSON output suitable for visualisation not using xm-mapping.
• JSON output is properly printed when read from CLI or pipe.
• Uses latest version of MathJax in test output.

Version 2.3.0

Implementation of ClearSpeak Rules.

• Contains the ClearSpeak implementation.

Other small additions:

• An SSML renderer with external ids for synchronised highlighting.
• Improved localisation handling.

Version 2.2.2

Summary of changes in this minor release:

• Unified locale handling in anticipation of more localisation
• Improvements to overall build and packaging system
• Overhaul of the command line interface
• Integrates some results of the AIM workshop on web accessibilty of Maths

In more details:

1. Changes to locale handling:

   • Hierarchical structuring unicode mappings directory.
   • Homogeneous separation of mappings for each locale.
   • Moves all code out of unicode mappings.
   • Removes mappings for IE.

2. Changes to npm package and build system

   • New variables module to hold info on version numbers, directories and URLs.
   • Browser goal is now included in lib directory.
   • Uses delivrjs as default file server for JSON files.
   • Minifies JSON files
   • IE mappings are constructed at build time only.
   • Unicode mappings are collated in the mathmaps subdirectory in lib directory.

3. Overhaul of CLI

   • Processes now multiple files from command line input
   • The -i, --input option is deprecated and will be removed in future
     releases.
   • Allows for multiple processors (e.g., -p, -m, ... options) to be
     specified to be run on input files. Even same processor, multiple times.
   • -o', --output` option only saves the last to file. For multiple
     file/output processing use shell piping or shell input.
   • Allows for interactive input of a single XML expression in the shell. Also enables piping of the form
     sre < input > output
     or
     sre -o output < input
     This will also same the output of all processors into the output file.
   • Fixes log file handling
   • -e, --enumerate removed. Functionality folded into --help option.

4. AIM workshop results

   • Support for external walkers via semantic tree structure similar to the collapsed attributes.
   • -r, --structure exposes this information via structure attribute in the enriched mathml.

5. A number of bug fixes.

Version 2.2.1

Main changes:

• Changes node dependencies by moving to wicked good xpath and a dedicated xmldom library.
• Major performance improvement in node.
• Implements support for all HTML5 entities.

Minor changes:

• Directory cleanup and restructuring.
• Support for as jsdoc documentation via make goal.
• Caches indexing tries.
• Some bug fixes.

Version 2.2.0

The first localisation for SRE is here! MathSpeak rules in SRE now also work in Spanish.

• Full Spanish localisation of the MathSpeak rule set.
• Spanish localisation of prefix and abstraction rule sets for use in web browsing.
• A new command line option to expose locales via ISO codes.
• In case a localisation does not exist for a particular rule set, SRE defaults to English.

Other changes in this version include:

• Improved constraint handling which speeds up rule application.
• New command line option for embedding speech into enriched MathML.

Acknowledgements: Thanks to TextHelp for funding Spanish localisation.

Patch for v2.1.0

• Minor bug fixes

Version 2.1.0

Includes

• Improved semantic recognition and treatment of equation systems and matrices,
• Enhanced walking and exploration of equation system,
• Cursor virtualisation,
• Virtual summarisation and expansion of expressions
• A new executable that can be run directly in npm repository of with npx.

Thanks to sponsorship from Benetech, the American Maths Society and others.

Version 2.0

Changes

1. New rule indexing via a generalised trie that makes speech string computation up to 10 times faster.

2. New Multi-axis constraint system for rule definition to support addition rule sets, multi-modality, localisation etc.

3. New grammar data structure for the uniform treatment of administrative information during rule dispatch, such as grammatical cases, corrections, font information, etc.

4. Prosody markup support for Audio CSS, Sable, SSML, VoiceXML.

5. Enhanced semantic recognition and handling of two dimensional structures, like equation systems, multi-line formulas, etc.

6. Emacspeak support. See also the release notes for Emacspeak v46.0 for details.

7. Uses semantic interpretations by default. This a change of behaviour from v1.x.x!

8. Improved build system with npm install using latest closure version (without NTI).

9. Removes deprecated API functions processExpression and processFile.

10. Various bug fixes.

Acknowledgements

A number of organisations provided funding and support for the continuing development of SRE. In particular, thanks to

• the Mozilla Foundation for funding work on data structure improvements under the MOSS scheme.
• Benetech for funding work on multi-line equations and two dimensional layout with a Diagram Center grant.
• AMS and the MathJax consortiumn for continuing financial and logistical support.

Also thanks to the following individuals for contributions, testing, feedback and encouragement:

Peter Krautzberger and Davide Cervone (MathJax), TV Raman (Emacspeak), Sue-Ann Ma and Anh Bui (Benetech), Jamie Teh (NVDA).