mirror of
https://github.com/denizsafak/abogen.git
synced 2026-07-18 13:40:27 +02:00
JB
be37f03109
- Implemented number conversion to words for grouped numbers in text normalization. - Added configuration options for number conversion in ApostropheConfig. - Updated the normalize_apostrophes function to include number normalization. - Enhanced the dashboard UI with new sections for manuscript and narrator defaults. - Improved voice preview functionality with better handling of audio playback and status updates. - Refactored HTML templates for cleaner structure and added new fields for speaker settings. - Updated CSS styles for improved layout and responsiveness. - Added tests to ensure correct spelling of grouped numbers in the normalization process. - Included num2words as a new dependency in pyproject.toml.
628 lines
20 KiB
Python
628 lines
20 KiB
Python
from __future__ import annotations
|
||
import re
|
||
import unicodedata
|
||
from dataclasses import dataclass
|
||
from typing import Callable, Iterable, List, Optional, Sequence, Tuple
|
||
try: # pragma: no cover - optional dependency guard
|
||
from num2words import num2words
|
||
except Exception: # pragma: no cover - graceful degradation
|
||
num2words = None # type: ignore
|
||
|
||
# ---------- Configuration Dataclass ----------
|
||
|
||
@dataclass
|
||
class ApostropheConfig:
|
||
contraction_mode: str = "expand" # expand|collapse|keep
|
||
possessive_mode: str = "keep" # keep|collapse
|
||
plural_possessive_mode: str = "collapse" # keep|collapse
|
||
irregular_possessive_mode: str = "keep" # keep|expand (expand just means keep or add hints; modify if needed)
|
||
sibilant_possessive_mode: str = "mark" # keep|mark|approx
|
||
fantasy_mode: str = "keep" # keep|mark|collapse_internal
|
||
acronym_possessive_mode: str = "keep" # keep|collapse_add_s
|
||
decades_mode: str = "expand" # keep|expand
|
||
leading_elision_mode: str = "expand" # keep|expand
|
||
ambiguous_past_modal_mode: str = "keep" # keep|expand_prefer_would|expand_prefer_had
|
||
add_phoneme_hints: bool = True # Whether to emit markers like ‹IZ›
|
||
fantasy_marker: str = "‹FAP›" # Marker inserted if fantasy_mode == mark
|
||
sibilant_iz_marker: str = "‹IZ›" # Marker for /ɪz/ insertion
|
||
joiner: str = "" # Replacement used when collapsing internal apostrophes
|
||
lowercase_for_matching: bool = True # Normalize to lower for rule matching (not output)
|
||
protect_cultural_names: bool = True # Always keep O'Brien, D'Angelo, etc.
|
||
convert_numbers: bool = True # Convert grouped numbers such as 12,500 to words
|
||
number_lang: str = "en" # num2words language code
|
||
|
||
# ---------- Dictionaries / Patterns ----------
|
||
|
||
# Common contraction expansions (straightforward unambiguous)
|
||
CONTRACTIONS_EXACT = {
|
||
"it's": "it is",
|
||
"that's": "that is",
|
||
"what's": "what is",
|
||
"where's": "where is",
|
||
"who's": "who is",
|
||
"when's": "when is",
|
||
"how's": "how is",
|
||
"there's": "there is",
|
||
"here's": "here is",
|
||
"let's": "let us",
|
||
"i'm": "i am",
|
||
"you're": "you are",
|
||
"we're": "we are",
|
||
"they're": "they are",
|
||
"i've": "i have",
|
||
"you've": "you have",
|
||
"we've": "we have",
|
||
"they've": "they have",
|
||
"i'll": "i will",
|
||
"you'll": "you will",
|
||
"he'll": "he will",
|
||
"she'll": "she will",
|
||
"we'll": "we will",
|
||
"they'll": "they will",
|
||
"i'd": "i would", # ambiguous (had/would), treat default
|
||
"you'd": "you would",
|
||
"he'd": "he would",
|
||
"she'd": "she would",
|
||
"we'd": "we would",
|
||
"they'd": "they would",
|
||
"can't": "can not", # or "cannot"
|
||
"won't": "will not",
|
||
"don't": "do not",
|
||
"doesn't": "does not",
|
||
"didn't": "did not",
|
||
"isn't": "is not",
|
||
"aren't": "are not",
|
||
"wasn't": "was not",
|
||
"weren't": "were not",
|
||
"haven't": "have not",
|
||
"hasn't": "has not",
|
||
"hadn't": "had not",
|
||
"couldn't": "could not",
|
||
"shouldn't": "should not",
|
||
"wouldn't": "would not",
|
||
"mustn't": "must not",
|
||
"mightn't": "might not",
|
||
"shan't": "shall not",
|
||
}
|
||
|
||
# For ambiguous 'd and 's we handle separately
|
||
_NUMBER_WITH_GROUP_RE = re.compile(r"(?<![\w\d])(-?\d{1,3}(?:,\d{3})+)(?![\w\d])")
|
||
AMBIGUOUS_D_BASES = {"i","you","he","she","we","they"}
|
||
AMBIGUOUS_S_BASES = {"it","that","what","where","who","when","how","there","here"}
|
||
|
||
# Irregular possessives that are not formed by simple + 's logic
|
||
IRREGULAR_POSSESSIVES = {
|
||
"children's": "children's",
|
||
"men's": "men's",
|
||
"women's": "women's",
|
||
"people's": "people's",
|
||
"geese's": "geese's",
|
||
"mouse's": "mouse's", # singular irregular
|
||
}
|
||
|
||
SIBILANT_END_RE = re.compile(r"(?:[sxz]|(?:ch|sh))$", re.IGNORECASE)
|
||
|
||
DECADE_RE = re.compile(r"^'\d0s$", re.IGNORECASE) # '90s, '80s
|
||
LEADING_ELISION = {
|
||
"'tis": "it is",
|
||
"'twas": "it was",
|
||
"'cause": "because",
|
||
"'em": "them",
|
||
"'round": "around",
|
||
"'til": "until",
|
||
}
|
||
|
||
CULTURAL_NAME_PATTERNS = [
|
||
re.compile(r"^O'[A-Z][a-z]+$"),
|
||
re.compile(r"^D'[A-Z][a-z]+$"),
|
||
re.compile(r"^L'[A-Za-z].*$"),
|
||
re.compile(r"^Mc[A-Z].*$"), # not apostrophe, but often relevant (kept anyway)
|
||
]
|
||
|
||
ACRONYM_POSSESSIVE_RE = re.compile(r"^[A-Z]{2,}'s$")
|
||
|
||
INTERNAL_APOSTROPHE_RE = re.compile(r"[A-Za-z]'.+[A-Za-z]") # apostrophe not at edge
|
||
|
||
# Capture contiguous runs of Unicode letters/digits/apostrophes/hyphens, otherwise fall back to
|
||
# single-character tokens (punctuation, symbols, etc.).
|
||
WORD_TOKEN_RE = re.compile(
|
||
r"[0-9A-Za-z'’\u00C0-\u1FFF\u2C00-\uD7FF\-]+|[^0-9A-Za-z\s]",
|
||
re.UNICODE,
|
||
)
|
||
|
||
APOSTROPHE_CHARS = "’`´ꞌʼ"
|
||
|
||
TERMINAL_PUNCTUATION = {".", "?", "!", "…", ";", ":"}
|
||
CLOSING_PUNCTUATION = '"\'”’)]}»›'
|
||
ELLIPSIS_SUFFIXES = ("...", "…")
|
||
_LINE_SPLIT_RE = re.compile(r"(\n+)")
|
||
|
||
TITLE_ABBREVIATIONS = {
|
||
"mr": "mister",
|
||
"mrs": "missus",
|
||
"ms": "miz",
|
||
"dr": "doctor",
|
||
"prof": "professor",
|
||
"rev": "reverend",
|
||
}
|
||
|
||
SUFFIX_ABBREVIATIONS = {
|
||
"jr": "junior",
|
||
"sr": "senior",
|
||
}
|
||
|
||
_TITLE_PATTERN = re.compile(
|
||
r"\b(?P<abbr>" + "|".join(sorted(TITLE_ABBREVIATIONS.keys(), key=len, reverse=True)) + r")\.",
|
||
re.IGNORECASE,
|
||
)
|
||
_SUFFIX_PATTERN = re.compile(
|
||
r"\b(?P<abbr>" + "|".join(sorted(SUFFIX_ABBREVIATIONS.keys(), key=len, reverse=True)) + r")\.",
|
||
re.IGNORECASE,
|
||
)
|
||
|
||
# ---------- Utility Functions ----------
|
||
|
||
def normalize_unicode_apostrophes(text: str) -> str:
|
||
text = unicodedata.normalize("NFKC", text)
|
||
for ch in APOSTROPHE_CHARS:
|
||
text = text.replace(ch, "'")
|
||
return text
|
||
|
||
def tokenize(text: str) -> List[str]:
|
||
# Simple tokenization preserving punctuation tokens
|
||
return WORD_TOKEN_RE.findall(text)
|
||
|
||
|
||
def _cleanup_spacing(text: str) -> str:
|
||
if not text:
|
||
return text
|
||
|
||
for marker in ("\ufeff", "\u200b", "\u200c", "\u200d", "\u2060"):
|
||
text = text.replace(marker, "")
|
||
|
||
# Collapse spaces before closing punctuation.
|
||
text = re.sub(r"\s+([,.;:!?%])", r"\1", text)
|
||
text = re.sub(r"\s+([’\"”»›)\]\}])", r"\1", text)
|
||
|
||
# Remove spaces directly after opening punctuation/quotes.
|
||
text = re.sub(r"([«‹“‘\"'(\[\{])\s+", r"\1", text)
|
||
|
||
# Ensure spaces exist after sentence punctuation when followed by a word/quote.
|
||
text = re.sub(r"([,.;:!?%])(?![\s”'\"’»›)])", r"\1 ", text)
|
||
text = re.sub(r"([”\"’])(?![\s.,;:!?\"”’»›)])", r"\1 ", text)
|
||
|
||
# Tighten hyphen/em dash spacing between word characters.
|
||
text = re.sub(r"(?<=\w)\s*([-–—])\s*(?=\w)", r"\1", text)
|
||
|
||
# Normalize multiple spaces.
|
||
text = re.sub(r"\s{2,}", " ", text)
|
||
return text.strip()
|
||
|
||
|
||
_ROMAN_VALUE_MAP = {
|
||
"I": 1,
|
||
"V": 5,
|
||
"X": 10,
|
||
"L": 50,
|
||
"C": 100,
|
||
"D": 500,
|
||
"M": 1000,
|
||
}
|
||
|
||
_ROMAN_COMPOSE_ORDER = [
|
||
(1000, "M"),
|
||
(900, "CM"),
|
||
(500, "D"),
|
||
(400, "CD"),
|
||
(100, "C"),
|
||
(90, "XC"),
|
||
(50, "L"),
|
||
(40, "XL"),
|
||
(10, "X"),
|
||
(9, "IX"),
|
||
(5, "V"),
|
||
(4, "IV"),
|
||
(1, "I"),
|
||
]
|
||
|
||
_ROMAN_PREFIX_RE = re.compile(r"^(?P<roman>[IVXLCDM]+)(?P<sep>[\s\.:,;\-–—]*)", re.IGNORECASE)
|
||
|
||
|
||
def _roman_to_int(token: str) -> Optional[int]:
|
||
if not token:
|
||
return None
|
||
total = 0
|
||
prev = 0
|
||
token_upper = token.upper()
|
||
for char in reversed(token_upper):
|
||
value = _ROMAN_VALUE_MAP.get(char)
|
||
if value is None:
|
||
return None
|
||
if value < prev:
|
||
total -= value
|
||
else:
|
||
total += value
|
||
prev = value
|
||
if total <= 0:
|
||
return None
|
||
if _int_to_roman(total) != token_upper:
|
||
return None
|
||
return total
|
||
|
||
|
||
def _int_to_roman(value: int) -> str:
|
||
parts: List[str] = []
|
||
remaining = value
|
||
for amount, symbol in _ROMAN_COMPOSE_ORDER:
|
||
while remaining >= amount:
|
||
parts.append(symbol)
|
||
remaining -= amount
|
||
return "".join(parts)
|
||
|
||
|
||
def normalize_roman_numeral_titles(
|
||
titles: Sequence[str],
|
||
*,
|
||
threshold: float = 0.5,
|
||
) -> List[str]:
|
||
if not titles:
|
||
return []
|
||
|
||
normalized: List[str] = []
|
||
matches: List[Tuple[int, str, int, str, str]] = []
|
||
non_empty = 0
|
||
|
||
for index, raw in enumerate(titles):
|
||
title = "" if raw is None else str(raw)
|
||
stripped = title.lstrip()
|
||
leading_ws = title[: len(title) - len(stripped)]
|
||
if not stripped:
|
||
normalized.append(title)
|
||
continue
|
||
|
||
non_empty += 1
|
||
match = _ROMAN_PREFIX_RE.match(stripped)
|
||
if not match:
|
||
normalized.append(title)
|
||
continue
|
||
|
||
roman_token = match.group("roman")
|
||
separator = match.group("sep") or ""
|
||
rest = stripped[match.end():]
|
||
|
||
if not separator and rest and rest[:1].isalnum():
|
||
normalized.append(title)
|
||
continue
|
||
|
||
numeric_value = _roman_to_int(roman_token)
|
||
if numeric_value is None:
|
||
normalized.append(title)
|
||
continue
|
||
|
||
matches.append((index, leading_ws, numeric_value, separator, rest))
|
||
normalized.append(title)
|
||
|
||
if not matches or non_empty == 0:
|
||
return list(normalized)
|
||
|
||
if len(matches) <= non_empty * threshold:
|
||
return list(normalized)
|
||
|
||
output = list(normalized)
|
||
for idx, leading_ws, value, separator, rest in matches:
|
||
new_title = f"{leading_ws}{value}"
|
||
if separator:
|
||
new_title += separator
|
||
elif rest and not rest[0].isspace() and rest[0] not in ".-–—:;,":
|
||
new_title += " "
|
||
new_title += rest
|
||
output[idx] = new_title
|
||
|
||
return output
|
||
|
||
|
||
def _match_casing(template: str, replacement: str) -> str:
|
||
if template.isupper():
|
||
return replacement.upper()
|
||
if template[:1].isupper() and template[1:].islower():
|
||
return replacement.capitalize()
|
||
if template[:1].isupper():
|
||
# Mixed case (e.g., Mc), fall back to title case
|
||
return replacement.capitalize()
|
||
return replacement
|
||
|
||
|
||
def expand_titles_and_suffixes(text: str) -> str:
|
||
def _replace(match: re.Match[str], mapping: dict[str, str]) -> str:
|
||
abbr = match.group("abbr")
|
||
lookup = mapping.get(abbr.lower())
|
||
if not lookup:
|
||
return match.group(0)
|
||
return _match_casing(abbr, lookup)
|
||
|
||
text = _TITLE_PATTERN.sub(lambda m: _replace(m, TITLE_ABBREVIATIONS), text)
|
||
text = _SUFFIX_PATTERN.sub(lambda m: _replace(m, SUFFIX_ABBREVIATIONS), text)
|
||
return text
|
||
|
||
|
||
def ensure_terminal_punctuation(text: str) -> str:
|
||
def _amend(segment: str) -> str:
|
||
if not segment or not segment.strip():
|
||
return segment
|
||
|
||
stripped = segment.rstrip()
|
||
trailing_ws = segment[len(stripped) :]
|
||
|
||
match = re.match(rf"^(.*?)([{re.escape(CLOSING_PUNCTUATION)}]*)$", stripped)
|
||
if not match:
|
||
return segment
|
||
|
||
body, closers = match.groups()
|
||
if not body:
|
||
return segment
|
||
|
||
normalized_body = body.rstrip()
|
||
trailing_body_ws = body[len(normalized_body) :]
|
||
|
||
if normalized_body.endswith(ELLIPSIS_SUFFIXES):
|
||
return normalized_body + trailing_body_ws + closers + trailing_ws
|
||
|
||
last_char = normalized_body[-1]
|
||
if last_char in TERMINAL_PUNCTUATION:
|
||
return normalized_body + trailing_body_ws + closers + trailing_ws
|
||
|
||
return normalized_body + "." + trailing_body_ws + closers + trailing_ws
|
||
|
||
parts = _LINE_SPLIT_RE.split(text)
|
||
amended: List[str] = []
|
||
for part in parts:
|
||
if not part:
|
||
continue
|
||
if part.startswith("\n"):
|
||
amended.append(part)
|
||
else:
|
||
amended.append(_amend(part))
|
||
if not parts:
|
||
return _amend(text)
|
||
return "".join(amended)
|
||
|
||
|
||
def is_cultural_name(token: str, cfg: ApostropheConfig) -> bool:
|
||
if not cfg.protect_cultural_names:
|
||
return False
|
||
for pat in CULTURAL_NAME_PATTERNS:
|
||
if pat.match(token):
|
||
return True
|
||
return False
|
||
|
||
def classify_token(token: str, cfg: ApostropheConfig) -> Tuple[str, str]:
|
||
"""
|
||
Classify apostrophe usage and propose normalized form.
|
||
Returns (category, normalized_token_or_same).
|
||
Categories: contraction, ambiguous_contraction_s, ambiguous_contraction_d,
|
||
plural_possessive, irregular_possessive, sibilant_possessive,
|
||
singular_possessive, acronym_possessive, decade, leading_elision,
|
||
fantasy_internal, other
|
||
"""
|
||
if "'" not in token:
|
||
return "other", token
|
||
|
||
raw = token
|
||
low = token.lower()
|
||
|
||
# 1. Decades
|
||
if DECADE_RE.match(token):
|
||
if cfg.decades_mode == "expand":
|
||
# '90s -> 1990s (you could also choose 90s)
|
||
return "decade", f"19{token[2:4]}s"
|
||
return "decade", token
|
||
|
||
# 2. Leading elision
|
||
if low in LEADING_ELISION:
|
||
if cfg.leading_elision_mode == "expand":
|
||
return "leading_elision", LEADING_ELISION[low]
|
||
return "leading_elision", token
|
||
|
||
# 3. Exact contraction
|
||
if low in CONTRACTIONS_EXACT:
|
||
if cfg.contraction_mode == "expand":
|
||
return "contraction", CONTRACTIONS_EXACT[low]
|
||
elif cfg.contraction_mode == "collapse":
|
||
# collapse: remove apostrophe only (it's -> its)
|
||
return "contraction", low.replace("'", "")
|
||
else:
|
||
return "contraction", token
|
||
|
||
# 4. Ambiguous 'd
|
||
if low.endswith("'d"):
|
||
base = low[:-2]
|
||
if base in AMBIGUOUS_D_BASES:
|
||
if cfg.ambiguous_past_modal_mode == "expand_prefer_would":
|
||
return "ambiguous_contraction_d", base + " would"
|
||
elif cfg.ambiguous_past_modal_mode == "expand_prefer_had":
|
||
return "ambiguous_contraction_d", base + " had"
|
||
elif cfg.contraction_mode == "collapse":
|
||
return "ambiguous_contraction_d", base + "d"
|
||
return "ambiguous_contraction_d", token
|
||
|
||
# 5. Ambiguous 's
|
||
if low.endswith("'s"):
|
||
base = low[:-2]
|
||
if base in AMBIGUOUS_S_BASES:
|
||
# treat as contraction 'is' under chosen mode
|
||
if cfg.contraction_mode == "expand":
|
||
return "ambiguous_contraction_s", base + " is"
|
||
elif cfg.contraction_mode == "collapse":
|
||
return "ambiguous_contraction_s", base + "s"
|
||
else:
|
||
return "ambiguous_contraction_s", token
|
||
|
||
# 6. Irregular possessives (keep or expand logic)
|
||
if low in IRREGULAR_POSSESSIVES:
|
||
if cfg.irregular_possessive_mode == "keep":
|
||
return "irregular_possessive", token
|
||
else:
|
||
# 'expand': we might keep same or optionally add marker
|
||
return "irregular_possessive", token
|
||
|
||
# 7. Plural possessive pattern dogs'
|
||
if re.match(r"^[A-Za-z0-9]+s'$", token):
|
||
if cfg.plural_possessive_mode == "collapse":
|
||
return "plural_possessive", token[:-1] # remove trailing apostrophe
|
||
return "plural_possessive", token
|
||
|
||
# 8. Acronym possessive NASA's
|
||
if ACRONYM_POSSESSIVE_RE.match(token):
|
||
if cfg.acronym_possessive_mode == "collapse_add_s":
|
||
return "acronym_possessive", token.replace("'", "")
|
||
return "acronym_possessive", token
|
||
|
||
# 9. Sibilant singular possessive boss's, church's
|
||
if low.endswith("'s"):
|
||
base = token[:-2]
|
||
if SIBILANT_END_RE.search(base):
|
||
if cfg.sibilant_possessive_mode == "keep":
|
||
return "sibilant_possessive", token
|
||
elif cfg.sibilant_possessive_mode == "approx":
|
||
# convert to base + "es" (boss's -> bosses)
|
||
# risk: loses possessive semantics visually
|
||
return "sibilant_possessive", base + "es"
|
||
elif cfg.sibilant_possessive_mode == "mark":
|
||
# remove apostrophe, add IZ marker
|
||
normalized = base
|
||
if cfg.add_phoneme_hints:
|
||
normalized += cfg.sibilant_iz_marker
|
||
else:
|
||
normalized += "es"
|
||
return "sibilant_possessive", normalized
|
||
|
||
# 10. Generic singular possessive (\w+'s)
|
||
if re.match(r"^[A-Za-z0-9]+'s$", token):
|
||
if cfg.possessive_mode == "collapse":
|
||
# Just remove apostrophe
|
||
return "singular_possessive", token.replace("'", "")
|
||
return "singular_possessive", token
|
||
|
||
# 11. Cultural names or fantasy internal
|
||
if is_cultural_name(token, cfg):
|
||
return "cultural_name", token
|
||
|
||
# 12. Fantasy internal apostrophes
|
||
if INTERNAL_APOSTROPHE_RE.search(token):
|
||
if cfg.fantasy_mode == "keep":
|
||
return "fantasy_internal", token
|
||
elif cfg.fantasy_mode == "mark":
|
||
out = token + (cfg.fantasy_marker if cfg.add_phoneme_hints else "")
|
||
return "fantasy_internal", out
|
||
elif cfg.fantasy_mode == "collapse_internal":
|
||
# Remove internal apostrophes only
|
||
inner = re.sub(r"(?<=\w)'+(?=\w)", cfg.joiner, token)
|
||
return "fantasy_internal", inner
|
||
|
||
# 13. Fallback: treat as other (maybe stray apostrophe)
|
||
if cfg.fantasy_mode == "collapse_internal":
|
||
# Remove any internal apostrophes
|
||
return "other", token.replace("'", cfg.joiner)
|
||
return "other", token
|
||
|
||
def normalize_apostrophes(text: str, cfg: ApostropheConfig | None = None) -> Tuple[str, List[Tuple[str,str,str]]]:
|
||
"""
|
||
Normalize apostrophes per config.
|
||
Returns normalized text AND a list of (original_token, category, normalized_token)
|
||
so you can debug or post-process (e.g., apply phoneme replacement for ‹IZ›).
|
||
"""
|
||
if cfg is None:
|
||
cfg = ApostropheConfig()
|
||
|
||
text = normalize_unicode_apostrophes(text)
|
||
text = _normalize_grouped_numbers(text, cfg)
|
||
tokens = tokenize(text)
|
||
|
||
results = []
|
||
normalized_tokens: List[str] = []
|
||
|
||
for tok in tokens:
|
||
category, norm = classify_token(tok, cfg)
|
||
results.append((tok, category, norm))
|
||
normalized_tokens.append(norm)
|
||
|
||
filtered = [token for token in normalized_tokens if token]
|
||
normalized_text = _cleanup_spacing(" ".join(filtered))
|
||
return normalized_text, results
|
||
|
||
def _normalize_grouped_numbers(text: str, cfg: ApostropheConfig) -> str:
|
||
if not text or not cfg.convert_numbers:
|
||
return text
|
||
|
||
def _replace(match: re.Match[str]) -> str:
|
||
token = match.group(1)
|
||
cleaned = token.replace(",", "")
|
||
if not cleaned:
|
||
return token
|
||
negative = cleaned.startswith("-")
|
||
cleaned_digits = cleaned[1:] if negative else cleaned
|
||
|
||
if not cleaned_digits.isdigit():
|
||
return cleaned_digits if not negative else f"-{cleaned_digits}"
|
||
|
||
if num2words is None:
|
||
return ("-" if negative else "") + cleaned_digits
|
||
|
||
try:
|
||
value = int(cleaned)
|
||
except ValueError:
|
||
return cleaned
|
||
|
||
language = (cfg.number_lang or "en").strip() or "en"
|
||
try:
|
||
words = num2words(abs(value), lang=language)
|
||
except Exception: # pragma: no cover - unsupported locale
|
||
return str(value)
|
||
|
||
if value < 0:
|
||
words = f"minus {words}"
|
||
return words
|
||
|
||
return _NUMBER_WITH_GROUP_RE.sub(_replace, text)
|
||
|
||
# ---------- Optional phoneme hint post-processing ----------
|
||
|
||
def apply_phoneme_hints(text: str, iz_marker="‹IZ›") -> str:
|
||
"""
|
||
Replace markers with an orthographic sequence that
|
||
your phonemizer will reliably convert to /ɪz/.
|
||
"""
|
||
return text.replace(iz_marker, " iz")
|
||
|
||
|
||
DEFAULT_APOSTROPHE_CONFIG = ApostropheConfig()
|
||
|
||
|
||
def normalize_for_pipeline(text: str, *, config: Optional[ApostropheConfig] = None) -> str:
|
||
"""Normalize text for the synthesis pipeline.
|
||
|
||
This expands contractions, normalizes apostrophes, and adds phoneme hints
|
||
using the provided configuration so downstream chunking and synthesis share
|
||
the same representation.
|
||
"""
|
||
|
||
cfg = config or DEFAULT_APOSTROPHE_CONFIG
|
||
normalized, _details = normalize_apostrophes(text, cfg)
|
||
normalized = expand_titles_and_suffixes(normalized)
|
||
normalized = ensure_terminal_punctuation(normalized)
|
||
normalized = _normalize_grouped_numbers(normalized, cfg)
|
||
if cfg.add_phoneme_hints:
|
||
normalized = apply_phoneme_hints(normalized, iz_marker=cfg.sibilant_iz_marker)
|
||
return normalized
|
||
|
||
# ---------- Example Usage ----------
|
||
|
||
if __name__ == "__main__":
|
||
sample = "Bob's boss's chair. The dogs' collars. It's cold. Ta'veren and Sha'hal. O'Brien's code in the '90s. Boss's orders."
|
||
config = ApostropheConfig()
|
||
norm_text, details = normalize_apostrophes(sample, config)
|
||
norm_text = apply_phoneme_hints(norm_text)
|
||
print("Original:", sample)
|
||
print("Normalized:", norm_text)
|
||
for orig, cat, norm in details:
|
||
print(f"{orig:15} -> {norm:15} [{cat}]") |