User:PharyngealImplosive7/IPA config.yaml

Description

Actual code

# IPA Processing Configuration
max_ipa_length:1
# Bracket mappings for IPA notation
brackets:
"(":")"
"[":"]"
"{":"}"
"/":"/"
"⟨":"⟩"
brackets_to_remove:
-"("
-")"
-"["
-"]"
-"{"
-"}"
-"/"
-"⟨"
-"⟩"
# Multi-character IPA exceptions that should be treated as single units
multi_char_exceptions:
-"kp"
-"gb"
-"ŋm"
-"tp"
-"db"
-"nm"
-"qp"
-"qʡ"
-"ʈp"
-"ɖb"
-"ɳm"
-"gɓ"
-"ɠɓ"
-"ɠ̊ɓ̥"
-"ƙƥ"
-"ŋ͡m"
-"ŋm"
-"ŋ͡mg͡b"
-"ŋmgb"
-"ŋ͡mgb"
-"ŋmg͡b"
-"pɸ"
-"bβ"
-"pf"
-"bv"
-"p̪f"
-"b̪v"
-"t̪θ"
-"tθ"
-"d̪ð"
-"dð"
-"tɹ̝̊"
-"dɹ̝"
-"t̠ɹ̠̊˔"
-"d̠ɹ̠˔"
-"cç"
-"ɟʝ"
-"kx"
-"ɡɣ"
-"qχ"
-"ɢʁ"
-"ʡʜ"
-"ʡʢ"
-"ʔh"
-"ts"
-"dz"
-"t̠ʃ"
-"d̠ʒ"
-"tʂ"
-"dʐ"
-"tɕ"
-"dʑ"
-"tʃ"
-"dʒ"
-"tɬ"
-"dɮ"
-"tꞎ"
-"d𝼅"
-"c𝼆"
-"ɟʎ̝"
-"k𝼄"
-"ɡʟ̝"
-"ʈʂ"
-"ɖʐ"
# Vowel symbols for diphthong detection
vowel_symbols:
-"i"
-"y"
-"ɨ"
-"ʉ"
-"ɯ"
-"u"
-"ɪ"
-"ʏ"
-"ʊ"
-"e"
-"ø"
-"ɘ"
-"ɵ"
-"ɤ"
-"o"
-"ə"
-"ɛ"
-"œ"
-"ɜ"
-"ɞ"
-"ʌ"
-"ɔ"
-"æ"
-"ɐ"
-"a"
-"ɶ"
-"ɑ"
-"ɒ"
# Tone symbols for tonal languages
tone_symbols:
-"˥"
-"˦"
-"˧"
-"˨"
-"˩"
-"1"
-"2"
-"3"
-"4"
-"5"
-"0"
-"1"
-"2"
-"3"
-"4"
-"5"
-"˦˥"
-"˧˦"
-"˨˧"
-"˩˨"
-"˩˥"
-"˥˩"
-"˦˩"
-"˧˩"
-"˨˩"
-"˩˥"
-"˥˧"
-"˦˧"
-"˧˥"
-"˨˥"
-"˩˧"
# Click consonant symbols
click_symbols:
-"ʘ"
-"ǀ"
-"ǃ"
-"ǂ"
-"ǁ"
-"𝼊"
-"!!"
-"ʗ"
-"¡"
-"ʇ"
# IPA diacritics (combining marks)
diacritics:
-"̃"
-"ː"
-"ˑ"
-"̥"
-"̊"
-"̥"
-"̬"
-"̤"
-"̩"
-"̯"
-"̰"
-"̱"
-"ʕ"
-"ˁ"
-"ʼ"
-"ʻ"
-"ʽ"
-"̚"
-"̆"
-"̘"
-"̙"
-"̝"
-"̞"
-"̟"
-"̠"
-"̼"
-"̪"
-"̺"
-"̻"
-"̹"
-"̜"
-"̽"
-"̴"
-"̩̚"
-"͋"
-"͌"
-"͇"
-"͉"
# Non-IPA diacritics to avoid
non_ipa_diacritics:
-"̀"
-"́"
-"̂"
-"̄"
-"̆"
-"̇"
-"̉"
-"̋"
-"̌"
-"̍"
-"̎"
-"̏"
-"̐"
-"̒"
-"̓"
-"̔"
-"̕"
-"̖"
-"̗"
-"̡"
-"̢"
-"̣"
-"̦"
-"̨"
-"̬"
-"̭"
-"̮"
-"̯"
-"̱"
-"̲"
-"̳"
-"̴"
-"̵"
-"̶"
-"̷"
-"̸"
# Allophone separators
allophone_separators:
-"~"
-","
-"/"
-"•"
-" "
-"<br/>"
-"<br/>"
# Phonetic terminology for table detection
phonetic_terms:
-"labial"
-"bilabial"
-"alveolar"
-"dental"
-"coronal"
-"velar"
-"palatal"
-"uvular"
-"dorsal"
-"glottal"
-"close"
-"high"
-"low"
-"back"
-"mid"
-"central"
-"voiced"
-"unvoiced"
-"aspirated"
-"ejective"
-"plosive"
-"stop"
-"affricate"
-"fricative"
-"nasal"
-"vowel"
-"consonant"
-"epiglottal"
-"pharyngeal"
-"labiodental"
-"approximant"
-"lateral"
-"oral"
-"tap"
-"flap"
-"labialized"
-"prenasalized"
-"linguolabial"
-"trill"
-"rhotic"
-"liquid"
-"rounded"
-"unrounded"
-"laryngeal"
-"voiceless"
-"implosive"
-"click"
-"near"
-"far"
# Special characters
tiebar:"͡"
ejective_marker:"ʼ"
# Regex patterns
superscript_regex_pattern:"[\u02B0-\u02FF\u1D00-\u1DBF\u2070-\u209F\u1D2C-\u1D6B]"
# Separator sets for tokenization
separator_symbols:
-"~"
-","
-"•"
-"&nbsp;"
-"<br/>"
-"<br/>"
-"<br>"
# Bracket matchings for template conversion
ipa_brackets:
"[":
-"]"
-"IPAblink"
"/":
-"/"
-"IPAslink"