from smolagents.tools import Tool
import string
import pronouncing
import json

class ParodyWordSuggestionTool(Tool):
    name = "parody_word_suggester"
    description = """Suggests rhyming funny words using CMU dictionary and custom 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)', 'default': '0.5', 'nullable': True}, 'custom_phones': {'type': 'object', 'description': 'Optional dictionary of custom word pronunciations', 'default': None, 'nullable': True}}
    output_type = "string"
    VOWEL_REF = "AH,AX|UH|AE,EH|IY,IH|AO,AA|UW|AY,EY|OW,AO|AW,AO|OY,OW|ER,AXR"
    CONSONANT_REF = "M,N,NG|P,B|T,D|K,G|F,V|TH,DH|S,Z|SH,ZH|L,R|W,Y"

    def _get_vowel_groups(self):
        groups = []
        group_strs = self.VOWEL_REF.split("|")
        for group_str in group_strs:
            groups.append(group_str.split(","))
        return groups


    def _get_word_phones(self, word, custom_phones=None):
        """Get phones for a word, checking custom dictionary first."""
        if custom_phones and word in custom_phones:
            return custom_phones[word]["primary_phones"]
        
        import pronouncing
        phones = pronouncing.phones_for_word(word)
        return phones[0] if phones else None


    def _get_last_syllable(self, phones: list) -> tuple:
        """Extract the last syllable (vowel + remaining consonants)."""
        last_vowel_idx = -1
        last_vowel = None
        vowel_groups = self._get_vowel_groups()
    
        for i, phone in enumerate(phones):
            base_phone = phone.rstrip('012')
            for group in vowel_groups:
                if base_phone in group:
                    last_vowel_idx = i
                    last_vowel = base_phone
                    break
    
        if last_vowel_idx == -1:
            return None, []
        
        remaining = phones[last_vowel_idx + 1:]
        return last_vowel, remaining


    def _strip_stress(self, phones: list) -> list:
        result = []
        for phone in phones:
            result.append(phone.rstrip('012'))
        return result


    def _vowels_match(self, v1: str, v2: str) -> bool:
        v1 = v1.rstrip('012')
        v2 = v2.rstrip('012')
    
        if v1 == v2:
            return True
        
        vowel_groups = self._get_vowel_groups()
        for group in vowel_groups:
            if v1 in group and v2 in group:
                return True
        return False


    def _strip_common_suffix(self, phones: list) -> tuple:
        """Strip common suffixes and return base and suffix phones."""
        # Initialize variables
        suffix_name = ""
        suffix_phones = []
        phone1 = ""
        phone2 = ""
    
        # Common suffix patterns in CMU phonetic representation
        SUFFIXES = {
            'ING': ['IH0', 'NG'],  # -ing
            'ED': ['EH0', 'D'],    # -ed
            'ER': ['ER0'],         # -er
            'EST': ['EH0', 'S', 'T'],  # -est
            'LY': ['L', 'IY0'],    # -ly
            'NESS': ['N', 'EH0', 'S'],  # -ness
        }
    
        for suffix_name, suffix_phones in SUFFIXES.items():
            if len(phones) > len(suffix_phones):
                if all(phone1.rstrip('012') == phone2.rstrip('012') 
                      for phone1, phone2 in zip(phones[-len(suffix_phones):], suffix_phones)):
                    return phones[:-len(suffix_phones)], suffix_phones
    
        return phones, []


    def _get_consonant_groups(self):
        """Get consonant groups from reference string."""
        groups = []
        group_strs = self.CONSONANT_REF.split("|")
        for group_str in group_strs:
            groups.append(group_str.split(","))
        return groups


    def _consonants_similarity(self, c1: str, c2: str) -> float:
        """Calculate similarity score between two consonants."""
        if c1 == c2:
            return 1.0
        
        # Check if they're in the same group
        consonant_groups = self._get_consonant_groups()
        for group in consonant_groups:
            if c1 in group and c2 in group:
                # Nasals (first group) are more similar to each other
                if group == consonant_groups[0]:  # M,N,NG group
                    return 0.8
                return 0.5
            
        return 0.0


    def _calculate_similarity(self, word1, phones1, word2, phones2):
        """Calculate similarity score with enhanced consonant matching."""
        # Initialize variables as before
        phone_list1 = phones1.split()
        phone_list2 = phones2.split()
    
        # Get stressed vowels and their positions
        vowel_idx1 = -1
        vowel_idx2 = -1
        primary_vowel1 = None
        primary_vowel2 = None
    
        for i, phone in enumerate(phone_list1):
            if '1' in phone:  # Primary stress
                vowel_idx1 = i
                primary_vowel1 = phone.rstrip('012')
                break
            
        for i, phone in enumerate(phone_list2):
            if '1' in phone:
                vowel_idx2 = i
                primary_vowel2 = phone.rstrip('012')
                break

        # Calculate vowel similarity (50% of total score)
        vowel_score = 0.0
        if primary_vowel1 and primary_vowel2:
            if primary_vowel1 == primary_vowel2:
                vowel_score = 1.0
            elif self._vowels_match(primary_vowel1, primary_vowel2):
                vowel_score = 0.8

        # Calculate consonant similarity (30% of total score)
        consonant_score = 0.0
        if vowel_idx1 >= 0 and vowel_idx2 >= 0:
            # Compare consonants around the stressed vowel
            pre_c1 = phone_list1[vowel_idx1-1] if vowel_idx1 > 0 else None
            pre_c2 = phone_list2[vowel_idx2-1] if vowel_idx2 > 0 else None
            post_c1 = phone_list1[vowel_idx1+1] if vowel_idx1 < len(phone_list1)-1 else None
            post_c2 = phone_list2[vowel_idx2+1] if vowel_idx2 < len(phone_list2)-1 else None
        
            if pre_c1 and pre_c2:
                consonant_score += self._consonants_similarity(pre_c1, pre_c2)
            if post_c1 and post_c2:
                consonant_score += self._consonants_similarity(post_c1, post_c2)
            
            consonant_score = consonant_score / 2  # Normalize to 0-1

        # Pattern/length similarity (20% of total score)
        pattern_score = 0.0
        if len(phone_list1) == len(phone_list2):
            pattern_score = 1.0
        else:
            pattern_score = 1.0 - (abs(len(phone_list1) - len(phone_list2)) / max(len(phone_list1), len(phone_list2)))

        # Combined weighted score
        similarity = (
            (vowel_score * 0.5) +      # Vowel similarity most important
            (consonant_score * 0.3) +   # Consonant similarity next
            (pattern_score * 0.2)       # Pattern/length least important
        )

        return {
            "similarity": round(similarity, 3),
            "vowel_score": round(vowel_score, 3),
            "consonant_score": round(consonant_score, 3),
            "pattern_score": round(pattern_score, 3),
            "primary_vowels": f"{primary_vowel1}-{primary_vowel2}",
            "consonants": "similar" if consonant_score > 0.5 else "different"
        }


    def forward(self, target: str, word_list_str: str, min_similarity: str = "0.5", custom_phones: dict = None) -> str:
        import pronouncing
        import string
        import json
    
        # Initialize all variables
        target = target.lower().strip(string.punctuation)
        min_similarity = float(min_similarity)
        suggestions = []
        word_vowel = None
        word_end = []
        target_vowel = None
        target_end = []
        valid_words = []
        invalid_words = []
        target_phone_list = []
        words = []
        target_phones = ""
        word_phones = ""
        word = ""
        word_phone_list = []
        similarity_result = {}
    
        # 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 = self._get_word_phones(target, custom_phones)
        if not target_phones:
            return json.dumps({
                "error": f"Target word '{target}' not found in dictionary or custom phones",
                "suggestions": []
            }, indent=2)
    
        # Filter word list
        valid_words = []
        invalid_words = []
        for word in words:
            word = word.lower().strip(string.punctuation)
            if self._get_word_phones(word, custom_phones):
                valid_words.append(word)
            else:
                invalid_words.append(word)
    
        if not valid_words:
            return json.dumps({
                "error": "No valid words found in dictionary or custom phones",
                "invalid_words": invalid_words,
                "suggestions": []
            }, indent=2)
    
        target_phone_list = target_phones.split()
        target_vowel, target_end = self._get_last_syllable(target_phone_list)
    
        # Check each word
        # Check each word
        for word in valid_words:
            word_phones = self._get_word_phones(word, custom_phones)
            if word_phones:
                similarity_result = self._calculate_similarity(word, word_phones, target, target_phones)
            
                if similarity_result["similarity"] >= min_similarity:
                    word_phone_list = word_phones.split()
                    word_vowel, word_end = self._get_last_syllable(word_phone_list)
                
                    suggestions.append({
                        "word": word,
                        "similarity": similarity_result["similarity"],
                        "vowel_score": similarity_result["vowel_score"],
                        "consonant_score": similarity_result["consonant_score"],
                        "pattern_score": similarity_result["pattern_score"],
                        "primary_vowels": similarity_result["primary_vowels"],
                        "consonants": similarity_result["consonants"],
                        "phones": word_phones,
                        "last_vowel": word_vowel,
                        "ending": " ".join(word_end) if word_end else "",
                        "is_custom": word in custom_phones if custom_phones else False
                    })
    
        # Sort by similarity score descending
        suggestions.sort(key=lambda x: x["similarity"], reverse=True)
    
        result = {
            "target": target,
            "target_phones": target_phones,
            "target_last_vowel": target_vowel,
            "target_ending": " ".join(target_end) if target_end else "",
            "invalid_words": invalid_words,
            "suggestions": suggestions
        }
    
        return json.dumps(result, indent=2)


    def __init__(self, *args, **kwargs):
        self.is_initialized = False