- Pull translations from Zanata (de, pl, uk updated)
- Make the list of characters to auto commit configurable
(Empty list by default)
- Fix duplicates in the candidate list caused by overwriting
input_phrase with the NFC version
- Don’t show the special candidates for missing dictionaries for
Japanese and Chinese
- Implement do_cursor_up() and do_cursor_down() to make scrolling
the lookup table with the mouse wheel work (Needs also a patch in ibus)
- Add an “Off the record mode” (also gets a property menu)
- Tooltips don’t seem to work on sub-properties, remove the tooltips there
- Add a property menu for the emoji prediction mode
- Make triggering a commit with “Left” or “Control+Left” work
correctly in “Tab enable mode ” again
- Down, Up, Page_Down, and Page_Up should trigger a commit and
be passed to the application if possible
- If “☑ Enable suggestions by Tab key” is on make it possible
to close the lookup table with Escape but keep the preëdit
- If “☑ Enable suggestions by Tab key” is on, don’t autocommit digits
- Make autocommitting much more rare (for characters which are not
the first typed character)
- Don’t autocommit the first typed character unless absolutely necessary
- Even when “☑ Enable suggestions by Tab key” is used,
don’t complete empty strings
- Reduce the number of characters which cause immediate commits a lot
- Load CLDR data for *all* languages in the _expand_languages() list
- Currency symbols should neither be stripped from tokens nor
trigger an immediate commit
- Fix bidi reordering problem in the candidate list for
right-to-left candidates followed by comments
- Update emoji annotations from CLDR (de_CH and sr_Latn new,
the others updated)
- Remove category 'Pc' from categories to commit immediately
(allow _ to be typed into the preëdit always)
- Remove button to install pyhunspell from the setup tool
(python3-enchant is preferred and even required by the Fedora rpm)
- Include more currency symbols and fullwidth symbols
- Add category from UnicodeData.txt to emoji dictionary
(For better results when looking up related characters)
- Add 'Sc', # Symbol, Currency to VALID_CATEGORIES
(to make the currency symbols work)
- Add list of valid characters (to include special characters
manually)
- Add mouse binding Alt+Mouse3 anywhere in the candidate list
to start the setup tool
- Pull translations form Zanata (de, pl, and uk updated because of
the new “About” tab)
- If “☑ Enable suggestions by Tab key” option is on, any preëdit
change should hide the lookup table
- Make showing of similar emoji work even if emoji preditions are off
- Display whether emoji predictions are turned on in the auxiliary string
- Add key and mouse bindings to toggle the emoji predictions
(AltGr+F6 and Control+Mouse3 anywhere in the candidate list)
- Add AltGr+F10 key binding to open the setup tool
- Allow any amount of white space and '_' characters to seperate words
in an emoji query string
- Add an “About” tab to the setup tool and put links to home page and
online documentation there.
- Update README with latest key binding and mouse binding documentation
- Accent insensitive matching
- Update pl.po from zanata
- Add cache for the suggestions from the hunspell dictionaries
- Make Control+MouseButton1 remove the clicked candidate from
the user database (was MouseButton2)
- Change key binding for looking up related candidates
from Alt+F12 to AltGr+F12
- Change label of the emoji option to
“☑ Unicode symbols and emoji predictions”
- get_supported_imes(self) and def get_current_imes(self) should
return copies not the lists directly
- Resolves: rhbz#1372660
- Update emojione.json, version from 2016-07-16
- Pull translations from Zanata: Fixes for fr and pl. New: uk
- Changes in itb_emoji.py necessary because of the update of
the CLDR emoji annotations.
- Update emoji annotations from CLDR (be, bs, cy, eu, gl, zu
are new, the others updated).
- Shortcut keys which look up related candidates should enable
the candidate list
- Show ⏳ HOURGLASS WITH FLOWING SAND in the auxiliary text when
the lookup table is being updated
- Fix bug when committing the preëdit with Space when no
candidates are available
- Improve the behaviour of the “Tab” key
- Improve the behaviour of the “Escape” key.
- Make mouse clicks in the candidate list behave differently
depending on the mouse button
- Add hu-rovas-post.mim to hu_HU.conf
- If the query string in EmojiMatcher.candidates() is an emoji
itself, match similar ones (useful when backspacing to an emoji
to correct it)
- Data files should not be stored gzipped in the repository
- Change displayed input method name from “Hunspell” to “Typing Booster”
- Use Zanata to get more translations
- French translations added (100% translated)
- Polish translations added (100% translated)
- Add Recommends: gdouros-symbola-fonts
- If the lookup table shows related words, “Escape” shows the
original lookup table
- Use itb_nltk.py to find related words (synonyms, hypernyms, and hyponyms)
- Add a module to find related words using NLTK
- Add a feature to find similar emoji
- Add predictions for emoji (optional, on by default)
- Add a module to match emoji using Unicode, CLDR, and emojione data
- Make typing-booster.appdata.xml translatable
- When ignoring key release events, “False” should be returned, not “True”
- Resolves: rhbz#1365497
- Make typing smoother by updating the candidates using GLib.idle_add()
- Make it possible to enter a space into the preëdit by
typing “G- ” (AltGr+Space)
- Do not colourize the preëdit dark blue, that is unreadable on
dark backgrounds
- Resolves: rhbz#1335201
- Set the size of the libm17n mconv conversion buffer correctly
- Resolves: rhbz#1335021
- self._current_imes needs to be updated before self.init_transliterators()
- Resolves: rhbz#1334579
- typing-booster.appdata.xml is now included and installed from the tarball
- libtranslit is not required anymore (neither for building nor at runtime)
but m17n-lib is required at runtime.
- Fix mistyped variable name
- Resolves: rhbz#1330461
- Add option to remember the preëdit input method used last
- Update German translations
- The combobox in the setup tool should show the first supported ime
from dconf
- Avoid unnessary initialization of transliterators when the set
of input methods has not changed
- Add property menu to choose the current preedit input method
- Display preëdit input method in aux_string also when number of
candidates is not shown
- Add some tooltips to the setup tool
- Update German translations
- Multilingual support, more than one language in an engine
- Simple option in the setup tool to enable bilingual support
(i.e. one language + Enlish).
- The default of the option “Add direct input” in the setup tool
should be false (bug found by Pravin Satpute).
- Changing the main input method with the setup tool should not
remove the direct input (bug found by Pravin Satpute)
- Add 0 as a digit to commit directly when using digits as select keys
- Clear dictionaries in Hunspell class before reloading
- Use libm17n directly instead of going through libtranslit
- Forward key events triggering a commit using “forward_key_event()”
instead of relying on “return False”
- Resolves: rhbz#1291238
- Add code to use F1-F9 as well as keys to select candidates
for commit or remove
- Don’t commit invisible candidates with select keys with numbers
greater than the length of a page of the candidate list
- Control-arrow-left and Control-arrow-right now commit when
the edges of the preedit string are reached
- Alt-<number> does not delete a prediction anymore,
now only Control-<number> does this
- Add an option to disable the use of the digits 1-9 as
selection keys (useful if one wants easier number input,
selection then works only with the F1-F9 keys)
- Support input methods using AltGr (e.g. mr-inscript2)
and Alt keys (e.g. ta-lk-renganathan)
- Resolves: rhbz#1051405
- Resolves: rhbz#772665
- Require Python >= 3.3
- Always write xml output in UTF-8 encoding, not in the encoding of the current locale
- Change class “KeyEvent” to store the keycode as well
- Commit when hitting the borders of the preëdit with the arrow keys (Resolves: rhbz#1140502)
- Fix behaviour of arrow right keys in preëdit (Resolves: rhbz#1049324)
- Add timestamps to entries in the user database
- Add timestamp support to user_transliteration.py
- Use a single user database for all engines
- Add *-inscript2 transliteration options to the Indian languages where these were still missing (Resolves: rhbz#1051405)
- Make it possible to use multiple hunspell dictionaries at the same time
- Make it possible to specify a list of dictionaries in the config files
- Make it possible to get a word back into preëdit by using backspace (Resolves: rhbz#1032442)
- Change wording of the option to show the total number of candidates (Resolves: rhbz#1029748)
- Commit candidate clicked on with the mouse (Resolves: rhbz#1029822)
- Use direct input also for IBus.InputPurpose.PIN
- remove unused und superfluous arguments of constructor of Hunspell class
- Add some transliteration options to .conf files which had only native keyboard enabled
- Add feature to display input method description to setup tool (Resolves: rhbz#1001581)
- Remove the options “m17n_mim_name” and “other_ime” from the .conf files
- remove tab_enable option from config files
- Use normalization form NFD internally for Korean as well
- Add check for input purpose for gnome-shell password dialog (Resolves: rhbz#10
13008 - ibus-typing-booster shows entered text in password fields)
- Commit immediately when certain punctuation characters are typed and transliteration is not used (Resolves: rhbz#981179)
- Add an option to try completion only when a minimum number of characters has been typed
- Pop up a message box when a file has been read to train the database, indicating success or failure (Resolves: rhbz#979933)
- Update German translation
- Ignore most punctuation characters and mathematical symbols when tokenizing (Resolves: rhbz#979939)
- Make TAB when used to enable/disable the lookup table work as a toogle
- Create a VIEW for “LIKE input_phrase%” in select_words() and use that
in the following SELECT statements (Makes candidate calculation more
than 10 times faster)
- Add a commit=True parameter to check_phrase_and_update_frequency()
- Fix that the page_size is shown as 0 in the setup tool if it has not been set before
- Do not use AUTOINCREMENT
- Make it possible to exit the setup tool by typing Control-C in the terminal
- Add feature to read a text file for training the user database
- Update German translations and .pot file
- Fix error when the hunspell dictionary for an engine is missing
- Don’t output page_size in “/usr/libexec/ibus-engine-typing-booster --xml” (Resolves: rhbz#975449 - ibus-daemon prints warnings because “/usr/libexec/ibus-engine-typing-booster --xml” prints the invalid element “page_size”)
- Use ~/.local/share/ibus-typing-booster/ to store user data and log files (Resolves: rhbz#949035 - don't use a hidden directory under .local/share)
- Remove mudb and use “Write-Ahead Logging”
- Introduce an environment variable IBUS_TYPING_BOOSTER_DEBUG_LEVEL for debugging
- Speed up converting an old database to the current format
- Make prediction more intelligent by using context of up to 2 previous words
- Automatically remove whitespace between the last word and a punctuation character ending a sentence
- Resolves: rhbz#969847 - Editing in the preëdit of ibus-typing-booster behaves weird, especially with transliteration
- Fix behaviour of Control+Number
- When committing by typing TAB, update frequency data in user database
- When committing by tying RETURN or ENTER, update frequency data in user database
- Do not try to match very long words in the hunspell dictionaries
- Rewrite the code for moving and editing within the preëdit (rhbz#969847)
- Fix encoding error when changing values with the setup tool
- Add ko_KR.conf and ko_KR.svg
- Use normalization forms NFD or NFKD internally and NFC externally
- Remove old way of using libtranslit via ctypes
- Get rid of “freq” column in databases
- Remove too simpleminded auto-capitalization
- Resolves: rhbz#968209 - Typing characters which are not explicitly listed as “valid_input_chars” in .conf files in ibus-typing-booster get inserted in a weird position
- Remove lots of unused and/or useless code
- Simplify some code
- Fix the problem that after “page down” the first “arrow down” does not move down in the lookup table
- Never use “-” or “=” as page up and page down keys
- Print more useful debug output when an exception happens
- Replace unencodable characters when asking pyhunspell for suggestions
- Get dictionary encoding from .aff file
- Get rid of the the variable “valid_input_chars” (https://bugzilla.redhat.com/show_bug.cgi?id=968209)
- Remove option “valid_input_chars” from .conf files and template.txt
- Replace keysym2unichr(key.code) with IBus.keyval_to_unicode(key.code)
- simplify database structure and code
- The Swedish hunspell dictionary is in UTF-8, not ISO-8859-1
- SQL LIKE should behave case sensitively
- Do not throw away the input phrase in hunspell_suggest.suggest()
- Merge candidates which have the same resulting phrase in select_words()
- Remove phrases always from the user database when typing Alt+Number
- Sync memory user database “mudb” to disk user database “user_db” on focus out
- Delete all records from mudb after syncing to user_db
- Do not prevent phrases of length < 4 to be added to the frequency database
- Resolves: #966947 - When typing a/ with the da_DK ibus-typing-booster, one gets weird matches like a/ACJSTVW
- Do not use lang_chars for matching in the hunspell dictionaries, return immediately if input contains a “/” (Resolves: #966947)
- Remove lang_chars variable
- Use re.escape() to escape the string typed by the user correctly for use in a regular expression
- When removing a phrase with Alt+Number, remove it independent of the input_phrase
- Resolves: #962609 - [abrt] ibus-typing-booster-0.0.28-1.fc19: main.py:107:__init__:AttributeError: tabsqlitedb instance has no attribute 'get_ime_property' (Fix setup tool to use the new class for parsing the config files)
- Avoid adding duplicates to the database by checking first whether phrase is already there in add_phrase()
- Resolves: #961923 - python /usr/share/ibus-typing-booster/engine/main.py --xml is extremely slow when many hunspell dictionaries are installed
- Put the input phrase into a single column in the databases instead of using one column for each character
- Get rid of tab_dict
- Resolves: #959860 - [as_IN] Wrong keymap name Assami (fix spelling error in language name for Assamese)
- Resolves: #958770 - [ibus-typing-Booster][gu-IN]- Typo error (fix spelling error in language name for Gujarati)
- Resolves: #875285 - IME names too long in gnome-shell Input Sources indicator (remove ✓ from symbol in the .conf files)
- simplify code in select_words()
- remove some unused functions
- Resolves: #910986 - The arrow icons at the bottom of the candidate lookup table of ibus-typing-booster do not work
- Use different .svg icons for all engines
- Increase number of suggestions from hunspell
- Use the auxiliary text to display the number of candidates
- Make the display of the number of candidates in the auxiliary text optional
- Display of the number of candidates needs to be updated on page-up and page-down
- Resolves: #884808 - ibus-typing-booster should also show candidates which correct spelling errors
- Use pyhunspell to add spell-checking suggestions
- Use underline for preedit
- Colourize spellchecking suggestions and system phrases already used
- Resolves: #879262 dictionary is not automatically reloaded when it is installed via the setup tool
- Make the engine reload the dictionary when the dictionary is installed via the setup tool
