CHANGELOG.md

Change Log

0.12.2 - Thursday February 13th 2014 [Feb 2014]

• html: fix intermitent whitespace normalization issues
• pdf: fix a crash when aMetadata is null (e.g. metadata without the --all argument)

0.12.1 - Wednesday February 12th 2014 [Feb 2014]

• fix running make check on clean machines
• fix dictionary (output order independent) tests [Debian Wheezy]
• fix mimetype alias resolution [shared-mime-info >= 1.1]
• clean up the Vim syntax support files
• update the man documentation for the cainteoir and dictionary programs
• add a man file for the phoneme-converter program

0.12 - Wednesday February 5th 2014 [Feb 2014]

• support using Pico TTS
• espeak: support pronouncing text via the Cainteoir TTS API
• audio: support ALSA output
• support private use language tags
• improve mapping eSpeak's language codes to RFC5656 language codes
• various API changes and enhancements
• improved and extended the API documentation
• improved CSS Counter Styles Level 3 support
• make use of the ucd-tools project for handling Unicode codepoint classification

new applications

• dictionary: manage pronunciation dictionaries
• phoneme-converter: convert between different phonetic transcriptions

text-to-speech support

• support tokenizing text into groups (words, numbers, punctuation, etc.)
• support basic context analysis (cardinal/ordinal number detection)
• support converting cardinal and ordinal numbers to words
• support a feature-based phonetic model and different phonetic transcription schemes
• support using a pronunciation dictionary to look up word pronunciations

document support

• epub: support ePub3 Media Overlays [1]
• epub: support ePub3 navigation documents [1]
• html: improved the whitespace normalization logic [1]
• html: improved parsing HTML documents [1]
• html: fixed handling of unquoted percentage characters in attributes
• opf: correctly handle @id attributes on dublin-core elements (dc:*).
• zip/epub: support ZIP archives containing data descriptors
• email: support archived email files
• email: support Debian bugmail mbox files

1. This feature was funded by the CSIR (Council for Scientific and Industrial Research).

0.11 - Saturday December 1st 2012 [Nov 2012]

• preserve the event information from the readers into the document model
• rework the document event structure to use a CSS-like subset
• model the CSS3 Counter Styles (previously Lists) spec for defining the disc and decimal list types

document support

• epub3: support for onix and marc metadata
• mht: fix reading MHT documents

0.10 - Monday October 1st 2012 [Sep 2012]

• fix compilation on clang 3.2
• add support for the mb-cz2 voice
• add a configure option to disable MBROLA support
• use C++11 range-based for loops instead of foreach_iter
• complete API documentation with doxygen (except the deprecated cainteoir::document API)
• rework and simplify the public buffer API
• improved space normalization logic with tests

document support

• html: initial support for MS Word generated HTML (http://www.w3.org/TR/REC-html40 XML namespace)
• html: fix anchor generation for the root item (fixes navigation in ZIP files containing HTML documents)
• mime: support story-based Title, Story, Author and Keywords mime-like markup
• mime: support 'Article ... of ...' and 'From ...' first lines
• mime: improve email and username identification in From sections
• opf: dc:subject statements with comma-separated keywords are now supported

RDF

• add support for RDF Query Language expression templates
• optimize rdf.select(s, _, _) queries
• remove rql::either support

languages

• update the IANA Language Sub-Tag Repository to 2012/09/04
• use RDF/RQL in the language tests
• fetch the extlang data from the IANA Language Sub-Tag Repository
• improve running the language tests -- set the LANGUAGE and LANG environment variables when setting the language

phoneme transcription schemes

• improve the chart generation and feature support
• fixes for ascii-ipa; make each entry unique
• create a vim syntax file for the phoneme transcription scheme format (*.phon)
• add support for arpabet, Unicode IPA, x-sampa and sampa-de transcription schemes
• add support for en1, us1, us2, us3, de1, de2, de4 and de5 MBROLA voice phoneme transcription schemes

0.9 - Tuesday July 31st 2012 [Jul 2012]

• doc2doc: improve text formatting
• improve argument and error handling for the command-line tools
• switch NEWS and README to markdown format

character encodings:

• xml: detault to windows-1252 encoding
• xml: switch encoding via <?xml encoding="..."?>
• mime/email: switch encoding via Content-Type: ...; charset="..."
• html: switch encoding via <meta charset="...">
• html: switch encoding via <meta http-equiv="Content-Type" content="...; charset=...">

document formats:

• mime: support mimetype/alternative
• email: support email embedded within html <pre> tags
• html: fix parsing of html containing implicit (missing) tags
• html: support table markup
• ePub: be more relaxed about checking the epub mimetype file

0.8 - Friday June 1st 2012 [Jun 2012]

• define an archive interface with a zip file implementation
• report the filename as a toc entry if no toc is present
• move the UTF-8 helpers to a common location
• optimize decoding of utf-8 and us-ascii encodings
• do not load a new decoder when switching to the same encoding
• switch document parsing from an event model to a reader model

applications:

• doc2doc: a simple command-line tool for converting documents from one format to another

document formats:

• pdf: support PDF documents using the Poppler library
• (x)html: support extracting metadata from several meta tag names
• zip: support reading generic zipped document collections
• ePub: make the ePub parser use the archive interface

0.7 - Friday March 30th 2012 [Mar 2012]

• use BCP 47 for interpreting and comparing language tags
• translate the language and region names using the iso-codes package
• support for espeak using installed mbrola voices

document formats:

• (x)html: report h1..h6 sections as table of content items
• email: improve email content detection and support newsgroup information
• rdf/xml: support property attributes on an empty property element
• xml: support entities in DTD blocks

bug fixes:

• fix detection of html that is valid xml, but is not xhtml
• improve detection of control characters when identifying binary data

cleanup/refactoring:

• simplified and cleaned up the RDF model and query APIs
• use std::shared_ptr consistently over std::tr1::shared_ptr
• use std::make_shared instead of allocating the object explicitly
• remove the libxml2 dependency and make all xml-based parsers use xmlreader

0.6: Sunday January 29th 2012 [Jan 2012]

• languages.rdf -- represent ISO-639-2/B, ISO-639-2/T, ISO-639-3, ISO-639-5, espeak and wikimedia language codes;
• add document format implementation status documentation.

xmlreader xml parsing support:

• parse entities in attribues;
• support single-quote and html-style attributes;
• parse xml namespaces;
• unify xml and rdf namespace/uri support;
• support element/attribute mapping to context entry objects.

document parsers:

• epub -- support epub 3 @refines and @datatype metadata;
• html -- support all known implicit close tags and recognise more elements;
• ocf -- use the xmlreader api and expose a reader api;
• opf -- use the xmlreader api.

bug fixes:

• release the file handle when finished recording ogg files.

0.5 - Monday November 21st 2011 [Nov 2011]

• use the shared-mime-info database for MIME type detection.
• use iconv for character encoding conversions.
• support all Content-Transfer-Encoding types for MIME headers, including base64.
• improve audio error handling during reading/recording.
• support more language code mappings for new espeak voices.
• support UND as a language identifier (for Calibri eBooks).
• support XML encoded HTML without an associated xmlns.

0.4 - Tuesday July 26th 2011 [Jul 2011]

file formats:

• support ncx-based table of content in epub documents;
• support epub 3.0 metadata;
• basic support for SSML.

cainteoir application and engine:

• support listing a document's table of contents and read/record a specified content range;
• support setting the voice reading speed, pitch and volume.

0.3 - Friday July 1st 2011 [Jun 2011]

file formats:

• support for single-file HTML pages (MHTML, MHT);
• support for RTF documents;
• enhanced HTML support.

bug fixes:

• don't segfault if the OPF document is missing from an epub file.

cainteoir:

• improved total reading time estimating heuristics.

0.2 - Sunday May 29th 2011 [May 2011]

file formats:

• support email (RFC822) mbox format;
• basic support for RTF documents.

metadata:

• add rdfs:domain, rdfs:range and OWL markup to the text-to-speech schema;
• install the text-to-speech RDF schema used by the Cainteoir Engine;
• extract language and title metadata from HTML documents;
• extract metadata from email From and Subject headers.

cainteoir:

• list supported document and audio formats in the --help output;
• don't output progress when writing audio data to stdout;
• don't crash when using the default voice (no dc:language metadata).

0.1 - Wednesday April 27th 2011 [Apr 2011]

cainteoir:

• command-line text-to-speech program;
• support for using espeak text-to-speech voices;
• listen to text, html and epub documents on the command line;
• record text, html and epub documents to ogg/vorbis and wave audio;
• see the reading/recording progress and (estimated) total time;
• preserve metadata from document to audio (currently epub to ogg/vorbis).

metadata:

• extract metadata from epub, smil, rdf/xml and other documents;
• generate RDF N-Triples
• generate RDF Turtle statements
• generate VorbisComment metadata