Spaces:
Configuration error
Configuration error
Upload tool
Browse files- requirements.txt +1 -1
- tool.py +21 -64
requirements.txt
CHANGED
@@ -1,2 +1,2 @@
|
|
1 |
-
pronouncing
|
2 |
smolagents
|
|
|
|
|
|
1 |
smolagents
|
2 |
+
pronouncing
|
tool.py
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
from smolagents.tools import Tool
|
2 |
-
import pronouncing
|
3 |
-
import difflib
|
4 |
import json
|
5 |
import string
|
|
|
|
|
6 |
|
7 |
class ParodyWordSuggestionTool(Tool):
|
8 |
name = "parody_word_suggester"
|
@@ -11,39 +11,21 @@ class ParodyWordSuggestionTool(Tool):
|
|
11 |
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}}
|
12 |
output_type = "string"
|
13 |
|
14 |
-
def
|
15 |
-
"""Check if
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
# Define groups of similar vowel sounds
|
21 |
-
similar_vowels = [
|
22 |
-
{'AH', 'UH'}, # Short u sounds
|
23 |
-
{'AE', 'EH'}, # Short e/a sounds
|
24 |
-
{'IY', 'IH'}, # Long/short i sounds
|
25 |
-
{'AO', 'AA'}, # Open o/a sounds
|
26 |
-
{'UW', 'UH'}, # Long/short oo sounds
|
27 |
-
]
|
28 |
-
|
29 |
-
# Direct match
|
30 |
-
if v1 == v2:
|
31 |
-
return True
|
32 |
-
|
33 |
-
# Check if they're in the same similarity group
|
34 |
-
for group in similar_vowels:
|
35 |
-
if v1 in group and v2 in group:
|
36 |
return True
|
37 |
-
|
38 |
return False
|
39 |
|
40 |
|
41 |
-
def
|
42 |
-
"""
|
43 |
-
for
|
44 |
-
if
|
45 |
-
return
|
46 |
-
return
|
47 |
|
48 |
|
49 |
def forward(self, target: str, word_list_str: str, min_similarity: str = "0.5") -> str:
|
@@ -53,8 +35,6 @@ class ParodyWordSuggestionTool(Tool):
|
|
53 |
import json
|
54 |
from difflib import SequenceMatcher
|
55 |
|
56 |
-
VOWEL_LETTERS = ['A', 'E', 'I', 'O', 'U']
|
57 |
-
|
58 |
target = target.lower().strip(string.punctuation)
|
59 |
min_similarity = float(min_similarity)
|
60 |
suggestions = []
|
@@ -75,17 +55,10 @@ class ParodyWordSuggestionTool(Tool):
|
|
75 |
"error": f"'{target}' not found in CMU dictionary",
|
76 |
"suggestions": []
|
77 |
}, indent=2)
|
78 |
-
|
79 |
target_phones = target_phones[0]
|
80 |
target_phone_list = target_phones.split()
|
81 |
-
|
82 |
-
# Find the last vowel in target
|
83 |
-
target_last_vowel = None
|
84 |
-
target_last_vowel_idx = -1
|
85 |
-
for i, phone in enumerate(target_phone_list):
|
86 |
-
if self._contains_vowel(phone, VOWEL_LETTERS):
|
87 |
-
target_last_vowel = phone
|
88 |
-
target_last_vowel_idx = i
|
89 |
|
90 |
# Check each word
|
91 |
for word in words:
|
@@ -94,29 +67,11 @@ class ParodyWordSuggestionTool(Tool):
|
|
94 |
if phones:
|
95 |
word_phones = phones[0]
|
96 |
word_phone_list = word_phones.split()
|
97 |
-
|
98 |
-
# Find last vowel in word
|
99 |
-
word_last_vowel = None
|
100 |
-
word_last_vowel_idx = -1
|
101 |
-
for i, phone in enumerate(word_phone_list):
|
102 |
-
if self._contains_vowel(phone, VOWEL_LETTERS):
|
103 |
-
word_last_vowel = phone
|
104 |
-
word_last_vowel_idx = i
|
105 |
|
106 |
# 1. Rhyme score (most important - 60%)
|
107 |
-
|
108 |
-
if
|
109 |
-
# Check if the vowels are similar
|
110 |
-
if self._are_vowels_similar(word_last_vowel, target_last_vowel):
|
111 |
-
# Check if the endings after the vowel match
|
112 |
-
if (word_phone_list[word_last_vowel_idx:] ==
|
113 |
-
target_phone_list[target_last_vowel_idx:]):
|
114 |
-
rhyme_score = 1.0
|
115 |
-
# Partial match for similar endings
|
116 |
-
elif (len(word_phone_list) > word_last_vowel_idx + 1 and
|
117 |
-
len(target_phone_list) > target_last_vowel_idx + 1 and
|
118 |
-
word_phone_list[-1] == target_phone_list[-1]):
|
119 |
-
rhyme_score = 0.8
|
120 |
|
121 |
# 2. Syllable match (25%)
|
122 |
target_syl = pronouncing.syllable_count(target_phones)
|
@@ -138,7 +93,8 @@ class ParodyWordSuggestionTool(Tool):
|
|
138 |
"syllable_match": syllable_score == 1.0,
|
139 |
"string_similarity": round(string_similarity, 3),
|
140 |
"syllables": word_syl,
|
141 |
-
"phones": word_phones
|
|
|
142 |
})
|
143 |
|
144 |
# Sort by similarity score descending
|
@@ -148,6 +104,7 @@ class ParodyWordSuggestionTool(Tool):
|
|
148 |
"target": target,
|
149 |
"target_syllables": pronouncing.syllable_count(target_phones),
|
150 |
"target_phones": target_phones,
|
|
|
151 |
"suggestions": suggestions
|
152 |
}
|
153 |
|
|
|
1 |
from smolagents.tools import Tool
|
|
|
|
|
2 |
import json
|
3 |
import string
|
4 |
+
import pronouncing
|
5 |
+
import difflib
|
6 |
|
7 |
class ParodyWordSuggestionTool(Tool):
|
8 |
name = "parody_word_suggester"
|
|
|
11 |
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}}
|
12 |
output_type = "string"
|
13 |
|
14 |
+
def _has_vowel(self, phone: str) -> bool:
|
15 |
+
"""Check if a phone contains a vowel."""
|
16 |
+
VOWELS = ['A', 'E', 'I', 'O', 'U']
|
17 |
+
for vowel in VOWELS:
|
18 |
+
if vowel in phone:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
return True
|
|
|
20 |
return False
|
21 |
|
22 |
|
23 |
+
def _get_rhyme_part(self, phones: list) -> list:
|
24 |
+
"""Get the rhyming part of a word (last vowel onwards)."""
|
25 |
+
for i, phone in enumerate(reversed(phones)):
|
26 |
+
if self._has_vowel(phone):
|
27 |
+
return phones[-(i+1):]
|
28 |
+
return phones
|
29 |
|
30 |
|
31 |
def forward(self, target: str, word_list_str: str, min_similarity: str = "0.5") -> str:
|
|
|
35 |
import json
|
36 |
from difflib import SequenceMatcher
|
37 |
|
|
|
|
|
38 |
target = target.lower().strip(string.punctuation)
|
39 |
min_similarity = float(min_similarity)
|
40 |
suggestions = []
|
|
|
55 |
"error": f"'{target}' not found in CMU dictionary",
|
56 |
"suggestions": []
|
57 |
}, indent=2)
|
58 |
+
|
59 |
target_phones = target_phones[0]
|
60 |
target_phone_list = target_phones.split()
|
61 |
+
target_rhyme_part = self._get_rhyme_part(target_phone_list)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
|
63 |
# Check each word
|
64 |
for word in words:
|
|
|
67 |
if phones:
|
68 |
word_phones = phones[0]
|
69 |
word_phone_list = word_phones.split()
|
70 |
+
word_rhyme_part = self._get_rhyme_part(word_phone_list)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
# 1. Rhyme score (most important - 60%)
|
73 |
+
# Perfect rhyme if the rhyming parts match exactly
|
74 |
+
rhyme_score = 1.0 if word_rhyme_part == target_rhyme_part else 0.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
|
76 |
# 2. Syllable match (25%)
|
77 |
target_syl = pronouncing.syllable_count(target_phones)
|
|
|
93 |
"syllable_match": syllable_score == 1.0,
|
94 |
"string_similarity": round(string_similarity, 3),
|
95 |
"syllables": word_syl,
|
96 |
+
"phones": word_phones,
|
97 |
+
"rhyme_part": " ".join(word_rhyme_part)
|
98 |
})
|
99 |
|
100 |
# Sort by similarity score descending
|
|
|
104 |
"target": target,
|
105 |
"target_syllables": pronouncing.syllable_count(target_phones),
|
106 |
"target_phones": target_phones,
|
107 |
+
"target_rhyme_part": " ".join(target_rhyme_part),
|
108 |
"suggestions": suggestions
|
109 |
}
|
110 |
|