Skip to content

Commit

Permalink
Some preparation for future support of m17n input methods offering mu…
Browse files Browse the repository at this point in the history 
…ltiple candidates

Related: #523

This just extends m17n_translit.py to return the data Typing Booster
needs to supports m17n input methods offering multiple candidates.
Typing Booster does not use this yet.
  • Loading branch information
@mike-fabian
mike-fabian committed Jan 8, 2025
1 parent 042c1b6 commit 346c863
Show file tree
Hide file tree
Showing 3 changed files with 251 additions and 59 deletions.
3 changes: 2 additions & 1 deletion engine/hunspell_table.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -7721,7 +7721,8 @@ def _process_key_event(self, key: itb_util.KeyEvent) -> bool:
if (not self.is_empty()
and not self._typed_compose_sequence
and self._is_restricted_engine()):
(committed, committed_index, preedit) = self._transliterators[
(committed, committed_index, preedit,
_cursor_pos, _status, _candidates) = self._transliterators[
self.get_current_imes()[0]].transliterate_parts(
self._typed_string, ascii_digits=self._ascii_digits)
if self._debug_level > 1:
Expand Down
143 changes: 117 additions & 26 deletions engine/m17n_translit.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -917,6 +917,7 @@ def convert_digits_to_ascii(text: str) -> str:
return text.translate(DIGIT_TRANS_TABLE)

class Transliterator:
# pylint: disable=line-too-long
'''A class for transliterators using libm17n
If initializing the transliterator fails, for example because a
Expand Down Expand Up @@ -950,23 +951,23 @@ class Transliterator:
'नमस्ते'
>>> trans.transliterate_parts(list('n'))
('', 0, 'न्')
('', 0, 'न्', 2, 'क', [])
>>> trans.transliterate_parts(list('n '))
('न ', 2, '')
('न ', 2, '', 0, 'क', [])
>>> trans.transliterate_parts(list('na'))
('', 0, 'न')
('', 0, 'न', 1, 'क', [])
>>> trans.transliterate_parts(list('nam'))
('न', 2, 'म्')
('न', 2, 'म्', 2, 'क', [])
>>> trans.transliterate_parts(list('nama'))
('न', 2, 'म')
('न', 2, 'म', 1, 'क', [])
>>> trans.transliterate_parts(list('namas'))
('नम', 4, 'स्')
('नम', 4, 'स्', 2, 'क', [])
>>> trans.transliterate_parts(list('namast'))
('नम', 4, 'स्त्')
('नम', 4, 'स्त्', 4, 'क', [])
>>> trans.transliterate_parts(list('namaste'))
('नम', 4, 'स्ते')
('नम', 4, 'स्ते', 4, 'क', [])
>>> trans.transliterate_parts(list('namaste '))
('नमस्ते ', 8, '')
('नमस्ते ', 8, '', 0, 'क', [])
>>> trans.transliterate(list('. '))
'। '
Expand Down Expand Up @@ -1015,30 +1016,47 @@ class Transliterator:
>>> trans = Transliterator('t-latn-post')
>>> trans.transliterate_parts(list('u'))
('', 0, 'u')
('', 0, 'u', 1, 'Latin-post', [])
>>> trans.transliterate_parts(list('u"'))
('', 0, 'ü')
('', 0, 'ü', 1, 'Latin-post', [])
>>> trans.transliterate_parts(list('u""'))
('u"', 3, '')
('u"', 3, '', 0, 'Latin-post', [])
>>> trans.transliterate_parts(list('u"u'))
('ü', 2, 'u')
('ü', 2, 'u', 1, 'Latin-post', [])
>>> trans.transliterate_parts(list('üu"u'))
('üü', 3, 'u')
('üü', 3, 'u', 1, 'Latin-post', [])
>>> trans = Transliterator('t-rfc1345')
>>> trans.transliterate_parts(list('&'))
('', 0, '&')
('', 0, '&', 1, 'RFC1345', [])
>>> trans.transliterate_parts(list('&C'))
('', 0, '&C')
('', 0, '&C', 2, 'RFC1345', [])
>>> trans.transliterate_parts(list('&Co'))
('©', 3, '')
('©', 3, '', 0, 'RFC1345', [])
>>> trans.transliterate_parts(list('&f'))
('', 0, '&f')
('', 0, '&f', 2, 'RFC1345', [])
>>> trans.transliterate_parts(list('&ff'))
('', 0, 'ff')
('', 0, 'ff', 1, 'RFC1345', [])
>>> trans.transliterate_parts(list('&ffi'))
('ffi', 4, '')
('ffi', 4, '', 0, 'RFC1345', [])
>>> trans = Transliterator('t-lsymbol')
>>> trans.transliterate_parts(list('/:)'))
('', 0, '☺️', 2, 'lsymbol', ['☺️', '😃', '😅', '😆', '😉', '😇', '😂', '😏', '😛', '😜', '😝', '😋', '😉', '💏', '💋', '😍', '😘', '😚', '😽', '😻'])
>>> trans.transliterate_parts(list('a'))
('a', 1, '', 0, 'lsymbol', [])
>>> trans.transliterate_parts(list('a/'))
('a', 1, '/', 1, 'lsymbol', [])
>>> trans.transliterate_parts(list('a/:'))
('a', 1, '/:', 2, 'lsymbol', [])
>>> trans.transliterate_parts(list('a/:('))
('a', 1, '😢', 1, 'lsymbol', ['😢', '😩', '😡', '😭', '😪', '🙈', '🙊', '🙉'])
>>> trans.transliterate_parts(list('a/:(b'))
('a😢b', 5, '', 0, 'lsymbol', [])
For a test transliterating parts using 'ja-anthy' see 'tests/test_m17n_translit.py'.
'''
# pylint: enable=line-too-long
def __init__(self, ime: str) -> None:
'''Initialize the input method to use for the transliteration
Expand Down Expand Up @@ -1074,7 +1092,7 @@ def __init__(self, ime: str) -> None:
def transliterate_parts(
self,
msymbol_list: Iterable[str],
ascii_digits: bool = False) -> Tuple[str, int, str]:
ascii_digits: bool = False) -> Tuple[str, int, str, int, str, List[str]]:
'''Transliterate a list of Msymbol names
:param msymbol_list: A list of strings which are interpreted
Expand All @@ -1084,16 +1102,40 @@ def transliterate_parts(
Msymbols is just joined to a single string.
:param ascii_digits: If true, convert language specific digits
to ASCII digits
:return: The transliteration in two parts: (committed, preedit)
:return: The transliteration in several parts:
(committed, committed_index, preedit, cursor_pos, status, candidates)
committed: str The part of the transliteration which cannot be
changed anymore by adding more input, could be
committed already if desired.
committed_index: int The index up to which the msymbol_list input
was “used up” to create the “committed” text.
preedit: The transliteration of the remaining input,
may still change by adding more input.
cursor_pos: int The cursor position in the preedit.
Counted in codepoints, not glyphs.
Usually this is at the end of the preedit
but an input method may move the cursor
within the preedit!
(I think only ja-anthy.mim actually uses this)
status: str May change for some input methods to
indicate a state.
For example in case of ja-anthy.mim,
this is 'aあ' before Henkan and changes
to '漢' in Henkan mode.
candidates: List[str] May contain a list of candidates if the
input method can produce multiple candidates.
'''
if not isinstance(msymbol_list, list):
raise ValueError('Argument of transliterate() must be a list.')
if self._dummy:
return (''.join(msymbol_list), 0, '')
return (''.join(msymbol_list), 0, '', 0, '', [])
libm17n__minput_reset_ic(self._ic) # type: ignore
committed = ''
committed_index = 0
preedit = ''
candidates: List[str] = []
for index, symbol in enumerate(msymbol_list):
if len(symbol) == 1 and not itb_util.is_ascii(symbol):
symbol = IBus.keyval_name(IBus.unicode_to_keyval(symbol))
Expand Down Expand Up @@ -1124,6 +1166,42 @@ def transliterate_parts(
except Exception as error: # pylint: disable=broad-except
# This should never happen:
raise ValueError('Problem accessing preedit') from error
plist = self._ic.contents.candidate_list
while bool(plist): # NULL pointers have a False boolean value
key = libm17n__mplist_key(plist) # type: ignore
if not bool(key):
break
key_name = libm17n__msymbol_name(key.contents) # type: ignore
if key_name == b'mtext':
characters = mtext_to_string(
ctypes.cast(libm17n__mplist_value(plist), # type: ignore
ctypes.POINTER(libm17n__MText)))
candidates += list(characters)
elif key_name == b'plist':
candidate_plist = ctypes.cast(
libm17n__mplist_value(plist), # type: ignore
ctypes.POINTER(libm17n__MPlist))
while True:
candidate_plist_key = libm17n__mplist_key( # type: ignore
candidate_plist)
if not bool(candidate_plist_key):
break
candidate_plist_key_name = libm17n__msymbol_name( # type: ignore
candidate_plist_key.contents)
if candidate_plist_key_name != b'mtext':
break
candidate = mtext_to_string(
ctypes.cast(
libm17n__mplist_value(candidate_plist), # type: ignore
ctypes.POINTER(libm17n__MText)))
candidates.append(candidate)
candidate_plist = libm17n__mplist_next( # type: ignore
candidate_plist)
else:
break
plist = libm17n__mplist_next(plist) # type: ignore
cursor_pos = self._ic.contents.cursor_pos
status = mtext_to_string(self._ic.contents.status)
# From the m17n-lib documentation:
#
# The minput_reset_ic () function resets input context $IC by
Expand Down Expand Up @@ -1151,10 +1229,18 @@ def transliterate_parts(
if committed and not preedit:
committed_index = len(msymbol_list)
if not ascii_digits:
return (committed, committed_index, preedit)
return (committed,
committed_index,
preedit,
cursor_pos,
status,
candidates)
return (convert_digits_to_ascii(committed),
committed_index,
convert_digits_to_ascii(preedit))
convert_digits_to_ascii(preedit),
cursor_pos,
status,
candidates)

def transliterate(self, msymbol_list: Iterable[str], ascii_digits: bool = False) -> str:
'''Transliterate a list of Msymbol names
Expand All @@ -1168,7 +1254,12 @@ def transliterate(self, msymbol_list: Iterable[str], ascii_digits: bool = False)
to ASCII digits
:return: The transliteration in one string
'''
(committed, _committed_index, preedit) = self.transliterate_parts(
(committed,
_committed_index,
preedit,
_cursor_pos,
_status,
_candidates) = self.transliterate_parts(
msymbol_list, ascii_digits)
return committed + preedit

Expand Down
Loading

0 comments on commit 346c863

Please sign in to comment.