| """ | |
| Character-level text frontend for Burmese. | |
| No real G2P: we clean the text and pass characters through directly. | |
| Matches the interface of english_to_ipa2 / japanese_to_ipa2 (returns a list). | |
| """ | |
| import re | |
| # Burmese Unicode ranges | |
| _BURMESE_RANGES = ( | |
| (0x1000, 0x109F), # Myanmar | |
| (0xAA60, 0xAA7F), # Myanmar Extended-A | |
| (0xA9E0, 0xA9FF), # Myanmar Extended-B | |
| ) | |
| # Allow these through the cleaner; everything else gets dropped | |
| _ALLOWED_ASCII_PUNCT = set(".,!?'\"-:;()") | |
| _ALLOWED_BURMESE_PUNCT = set("\u104A\u104B\u104C\u104D\u104E\u104F") | |
| def _is_burmese(ch): | |
| o = ord(ch) | |
| return any(lo <= o <= hi for lo, hi in _BURMESE_RANGES) | |
| def clean_burmese(text): | |
| """Keep only Burmese chars, spaces, and whitelisted punctuation.""" | |
| out = [] | |
| for ch in text: | |
| if ch in (" ", "\t", "\n", "\r"): | |
| out.append(" ") | |
| elif _is_burmese(ch): | |
| out.append(ch) | |
| elif ch in _ALLOWED_ASCII_PUNCT or ch in _ALLOWED_BURMESE_PUNCT: | |
| out.append(ch) | |
| # else: drop silently | |
| return re.sub(r"\s+", " ", "".join(out)).strip() | |
| def burmese_to_ipa2(text): | |
| """Return a list of character tokens (matches the IPA-G2P interface).""" | |
| cleaned = clean_burmese(text) | |
| return list(cleaned) | |