Spaces:
Configuration error
Configuration error
| from smolagents.tools import Tool | |
| import pronouncing | |
| import difflib | |
| import json | |
| import string | |
| class ParodyWordSuggestionTool(Tool): | |
| name = "parody_word_suggester" | |
| description = """Suggests rhyming funny words using CMU dictionary pronunciations. | |
| Returns similar-sounding words that rhyme, especially focusing on common vowel sounds.""" | |
| inputs = {'target': {'type': 'string', 'description': 'The word you want to find rhyming alternatives for'}, 'word_list_str': {'type': 'string', 'description': 'JSON string of word list (e.g. \'["word1", "word2"]\')'}, 'min_similarity': {'type': 'string', 'description': 'Minimum similarity threshold (0.0-1.0)', 'nullable': True}} | |
| output_type = "string" | |
| def _are_vowels_similar(self, v1: str, v2: str) -> bool: | |
| """Check if two vowel sounds are similar enough to rhyme.""" | |
| # Strip stress markers | |
| v1 = v1.rstrip('012') | |
| v2 = v2.rstrip('012') | |
| # Define groups of similar vowel sounds | |
| similar_vowels = [ | |
| {'AH', 'UH'}, # Short u sounds | |
| {'AE', 'EH'}, # Short e/a sounds | |
| {'IY', 'IH'}, # Long/short i sounds | |
| {'AO', 'AA'}, # Open o/a sounds | |
| {'UW', 'UH'}, # Long/short oo sounds | |
| ] | |
| # Direct match | |
| if v1 == v2: | |
| return True | |
| # Check if they're in the same similarity group | |
| for group in similar_vowels: | |
| if v1 in group and v2 in group: | |
| return True | |
| return False | |
| def _contains_vowel(self, phone: str, vowels: list) -> bool: | |
| """Helper function to check if a phone contains any vowel from the list.""" | |
| for v in vowels: | |
| if v in phone: | |
| return True | |
| return False | |
| def forward(self, target: str, word_list_str: str, min_similarity: str = "0.5") -> str: | |
| """Get rhyming word suggestions.""" | |
| import pronouncing | |
| import string | |
| import json | |
| from difflib import SequenceMatcher | |
| VOWEL_LETTERS = ['A', 'E', 'I', 'O', 'U'] | |
| target = target.lower().strip(string.punctuation) | |
| min_similarity = float(min_similarity) | |
| suggestions = [] | |
| # Parse JSON string to list | |
| try: | |
| words = json.loads(word_list_str) | |
| except json.JSONDecodeError: | |
| return json.dumps({ | |
| "error": "Invalid JSON string for word_list_str", | |
| "suggestions": [] | |
| }, indent=2) | |
| # Get target pronunciation | |
| target_phones = pronouncing.phones_for_word(target) | |
| if not target_phones: | |
| return json.dumps({ | |
| "error": f"'{target}' not found in CMU dictionary", | |
| "suggestions": [] | |
| }, indent=2) | |
| target_phones = target_phones[0] | |
| target_phone_list = target_phones.split() | |
| # Find the last vowel in target | |
| target_last_vowel = None | |
| target_last_vowel_idx = -1 | |
| for i, phone in enumerate(target_phone_list): | |
| if self._contains_vowel(phone, VOWEL_LETTERS): | |
| target_last_vowel = phone | |
| target_last_vowel_idx = i | |
| # Check each word | |
| for word in words: | |
| word = word.lower().strip(string.punctuation) | |
| phones = pronouncing.phones_for_word(word) | |
| if phones: | |
| word_phones = phones[0] | |
| word_phone_list = word_phones.split() | |
| # Find last vowel in word | |
| word_last_vowel = None | |
| word_last_vowel_idx = -1 | |
| for i, phone in enumerate(word_phone_list): | |
| if self._contains_vowel(phone, VOWEL_LETTERS): | |
| word_last_vowel = phone | |
| word_last_vowel_idx = i | |
| # 1. Rhyme score (most important - 60%) | |
| rhyme_score = 0 | |
| if word_last_vowel and target_last_vowel: | |
| # Check if the vowels are similar | |
| if self._are_vowels_similar(word_last_vowel, target_last_vowel): | |
| # Check if the endings after the vowel match | |
| if (word_phone_list[word_last_vowel_idx:] == | |
| target_phone_list[target_last_vowel_idx:]): | |
| rhyme_score = 1.0 | |
| # Partial match for similar endings | |
| elif (len(word_phone_list) > word_last_vowel_idx + 1 and | |
| len(target_phone_list) > target_last_vowel_idx + 1 and | |
| word_phone_list[-1] == target_phone_list[-1]): | |
| rhyme_score = 0.8 | |
| # 2. Syllable match (25%) | |
| target_syl = pronouncing.syllable_count(target_phones) | |
| word_syl = pronouncing.syllable_count(word_phones) | |
| syllable_score = 1.0 if target_syl == word_syl else 0.0 | |
| # 3. Overall similarity (15%) | |
| string_similarity = SequenceMatcher(None, target, word).ratio() | |
| # Combined score | |
| similarity = (rhyme_score * 0.6) + (syllable_score * 0.25) + (string_similarity * 0.15) | |
| if similarity >= min_similarity: | |
| suggestions.append({ | |
| "word": word, | |
| "similarity": round(similarity, 3), | |
| "rhyme_match": rhyme_score > 0, | |
| "rhyme_score": round(rhyme_score, 3), | |
| "syllable_match": syllable_score == 1.0, | |
| "string_similarity": round(string_similarity, 3), | |
| "syllables": word_syl, | |
| "phones": word_phones | |
| }) | |
| # Sort by similarity score descending | |
| suggestions.sort(key=lambda x: x["similarity"], reverse=True) | |
| result = { | |
| "target": target, | |
| "target_syllables": pronouncing.syllable_count(target_phones), | |
| "target_phones": target_phones, | |
| "suggestions": suggestions | |
| } | |
| return json.dumps(result, indent=2) | |
| def __init__(self, *args, **kwargs): | |
| self.is_initialized = False | |