335 lines
12 KiB
Python
335 lines
12 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
Thai Readability Analyzer
|
|
|
|
Analyze Thai text readability with PyThaiNLP integration.
|
|
Detects formality level, grade level, and sentence structure.
|
|
"""
|
|
|
|
import argparse
|
|
import json
|
|
import re
|
|
from typing import Dict, List
|
|
|
|
try:
|
|
from pythainlp import word_tokenize, sent_tokenize
|
|
THAI_SUPPORT = True
|
|
except ImportError:
|
|
THAI_SUPPORT = False
|
|
print("Warning: PyThaiNLP not installed. Install with: pip install pythainlp")
|
|
|
|
|
|
class ThaiReadabilityAnalyzer:
|
|
"""Analyze Thai text readability"""
|
|
|
|
def __init__(self):
|
|
self.formal_particles = [
|
|
'ครับ', 'ค่ะ', 'ข้าพเจ้า', 'กระผม', 'ดิฉัน', 'ท่าน', 'ซึ่ง', 'อัน',
|
|
'ย่อม', 'ย่อมเป็น', 'ประการ', 'ดังกล่าว', 'ดังกล่าวแล้ว', 'ดังนี้'
|
|
]
|
|
|
|
self.informal_particles = [
|
|
'นะ', 'จ้ะ', 'อ่ะ', 'มั้ย', 'เปล่าว่ะ', 'gue', 'mang', 'เว้ย',
|
|
'วะ', 'เหอะ', 'ซิ', 'นู่น', 'นี่', 'นั่น', 'โครต', 'มาก'
|
|
]
|
|
|
|
def count_sentences(self, text: str) -> int:
|
|
"""Count Thai sentences"""
|
|
if not THAI_SUPPORT:
|
|
# Fallback: count Thai sentence endings
|
|
thai_endings = ['.', '!', '?', '।', '๏']
|
|
count = sum(text.count(e) for e in thai_endings)
|
|
return max(count, 1)
|
|
|
|
sentences = sent_tokenize(text, engine="whitespace")
|
|
return len([s for s in sentences if s.strip()])
|
|
|
|
def count_words(self, text: str) -> int:
|
|
"""Count Thai words"""
|
|
if not THAI_SUPPORT:
|
|
return len(text.split())
|
|
|
|
tokens = word_tokenize(text, engine="newmm")
|
|
return len([t for t in tokens if t.strip()])
|
|
|
|
def calculate_avg_sentence_length(self, text: str) -> float:
|
|
"""Calculate average sentence length"""
|
|
if not THAI_SUPPORT:
|
|
sentences = re.split(r'[.!?]', text)
|
|
sentences = [s for s in sentences if s.strip()]
|
|
if not sentences:
|
|
return 0
|
|
|
|
words = text.split()
|
|
return len(words) / len(sentences)
|
|
|
|
sentences = sent_tokenize(text, engine="whitespace")
|
|
sentences = [s for s in sentences if s.strip()]
|
|
|
|
if not sentences:
|
|
return 0
|
|
|
|
total_words = sum(
|
|
len(word_tokenize(s, engine="newmm"))
|
|
for s in sentences
|
|
)
|
|
|
|
return total_words / len(sentences)
|
|
|
|
def detect_formality(self, text: str) -> Dict:
|
|
"""Detect Thai formality level"""
|
|
formal_count = sum(text.count(p) for p in self.formal_particles)
|
|
informal_count = sum(text.count(p) for p in self.informal_particles)
|
|
|
|
total = formal_count + informal_count
|
|
|
|
if total == 0:
|
|
ratio = 0.5 # Neutral
|
|
else:
|
|
ratio = formal_count / total
|
|
|
|
if ratio > 0.6:
|
|
level = "เป็นทางการ (Formal)"
|
|
score = 80
|
|
elif ratio < 0.4:
|
|
level = "กันเอง (Casual)"
|
|
score = 20
|
|
else:
|
|
level = "ปกติ (Normal)"
|
|
score = 50
|
|
|
|
return {
|
|
'level': level,
|
|
'score': score,
|
|
'formal_particle_count': formal_count,
|
|
'informal_particle_count': informal_count,
|
|
'ratio': round(ratio, 2)
|
|
}
|
|
|
|
def estimate_grade_level(self, avg_sentence_length: float, formality_score: int) -> Dict:
|
|
"""Estimate Thai grade level"""
|
|
# Thai grade level estimation based on sentence complexity
|
|
if avg_sentence_length < 15:
|
|
grade_th = "ง่าย (ม.6-ม.9)"
|
|
grade_num = 6-9
|
|
elif avg_sentence_length < 25:
|
|
grade_th = "ปานกลาง (ม.10-ม.12)"
|
|
grade_num = 10-12
|
|
else:
|
|
grade_th = "ยาก (ม.13+)"
|
|
grade_num = 13
|
|
|
|
# Adjust for formality
|
|
if formality_score > 70:
|
|
grade_th += " (ทางการ)"
|
|
elif formality_score < 30:
|
|
grade_th += " (กันเอง)"
|
|
|
|
return {
|
|
'thai': grade_th,
|
|
'numeric_range': grade_num,
|
|
'us_equivalent': self._thai_to_us_grade(grade_num)
|
|
}
|
|
|
|
def _thai_to_us_grade(self, thai_grade_range) -> str:
|
|
"""Convert Thai grade to US equivalent"""
|
|
if isinstance(thai_grade_range, range):
|
|
avg = sum(thai_grade_range) / len(thai_grade_range)
|
|
elif isinstance(thai_grade_range, int):
|
|
avg = thai_grade_range
|
|
else:
|
|
avg = 10
|
|
|
|
# Very rough conversion
|
|
if avg <= 9:
|
|
return "6th-8th grade"
|
|
elif avg <= 12:
|
|
return "9th-12th grade"
|
|
else:
|
|
return "College+"
|
|
|
|
def analyze_paragraph_structure(self, text: str) -> Dict:
|
|
"""Analyze paragraph structure"""
|
|
paragraphs = [p for p in text.split('\n\n') if p.strip()]
|
|
|
|
if not paragraphs:
|
|
return {
|
|
'paragraph_count': 0,
|
|
'avg_length_words': 0,
|
|
'avg_length_sentences': 0
|
|
}
|
|
|
|
paragraph_lengths = [
|
|
self.count_words(p)
|
|
for p in paragraphs
|
|
]
|
|
|
|
paragraph_sentences = [
|
|
self.count_sentences(p)
|
|
for p in paragraphs
|
|
]
|
|
|
|
return {
|
|
'paragraph_count': len(paragraphs),
|
|
'avg_length_words': round(sum(paragraph_lengths) / len(paragraphs), 1),
|
|
'avg_length_sentences': round(sum(paragraph_sentences) / len(paragraphs), 1),
|
|
'shortest_paragraph': min(paragraph_lengths),
|
|
'longest_paragraph': max(paragraph_lengths)
|
|
}
|
|
|
|
def calculate_readability_score(self, avg_sentence_length: float, formality_score: int,
|
|
paragraph_score: float) -> float:
|
|
"""
|
|
Calculate overall readability score (0-100)
|
|
|
|
Factors:
|
|
- Sentence length (optimal: 15-25 words)
|
|
- Formality (optimal: 40-60 for general content)
|
|
- Paragraph structure (optimal: varied lengths)
|
|
"""
|
|
# Sentence length score (0-40)
|
|
if 15 <= avg_sentence_length <= 25:
|
|
sentence_score = 40
|
|
elif 10 <= avg_sentence_length < 15 or 25 < avg_sentence_length <= 30:
|
|
sentence_score = 30
|
|
elif avg_sentence_length < 10:
|
|
sentence_score = 20
|
|
else:
|
|
sentence_score = 15
|
|
|
|
# Formality score (0-30)
|
|
# Optimal: 40-60 (normal/formal mix)
|
|
if 40 <= formality_score <= 60:
|
|
formality_points = 30
|
|
elif 30 <= formality_score < 40 or 60 < formality_score <= 70:
|
|
formality_points = 25
|
|
else:
|
|
formality_points = 15
|
|
|
|
# Paragraph score (0-30)
|
|
paragraph_points = min(30, paragraph_score * 30)
|
|
|
|
total = sentence_score + formality_points + paragraph_points
|
|
|
|
return round(total, 1)
|
|
|
|
def get_recommendations(self, analysis: Dict) -> List[str]:
|
|
"""Generate recommendations"""
|
|
recs = []
|
|
|
|
avg_len = analysis['avg_sentence_length']
|
|
if avg_len < 15:
|
|
recs.append("ประโยคสั้นเกินไป พิจารณาเพิ่มรายละเอียดบ้าง")
|
|
elif avg_len > 25:
|
|
recs.append("ประโยคยาวเกินไป แบ่งออกเป็น 2-3 ประโยคจะอ่านง่ายขึ้น")
|
|
|
|
formality = analysis['formality']['level']
|
|
if "เป็นทางการ" in formality:
|
|
recs.append("ภาษาเป็นทางการเกินไปสำหรับเนื้อหาทั่วไป พิจารณาใช้ภาษาที่เป็นกันเองมากขึ้น")
|
|
elif "กันเอง" in formality:
|
|
recs.append("ภาษาเป็นกันเองมาก ตรวจสอบว่าเหมาะกับกลุ่มเป้าหมายหรือไม่")
|
|
|
|
para = analysis['paragraph_structure']
|
|
if para['avg_length_words'] > 200:
|
|
recs.append("บางย่อหน้ายาวเกินไป แบ่งย่อหน้าเพื่อให้อ่านง่ายขึ้น")
|
|
|
|
if para['paragraph_count'] < 5:
|
|
recs.append("เพิ่มจำนวนย่อหน้าเพื่อให้อ่านง่ายขึ้น")
|
|
|
|
return recs
|
|
|
|
def analyze(self, text: str) -> Dict:
|
|
"""Full readability analysis"""
|
|
avg_sentence_length = self.calculate_avg_sentence_length(text)
|
|
formality = self.detect_formality(text)
|
|
grade_level = self.estimate_grade_level(avg_sentence_length, formality['score'])
|
|
paragraph_structure = self.analyze_paragraph_structure(text)
|
|
|
|
# Calculate paragraph score (0-1)
|
|
para_score = 0.5 # Default
|
|
if paragraph_structure['paragraph_count'] > 0:
|
|
# Score based on variety
|
|
lengths = [paragraph_structure['avg_length_words']]
|
|
if paragraph_structure['shortest_paragraph'] != paragraph_structure['longest_paragraph']:
|
|
para_score = 0.8 # Good variety
|
|
else:
|
|
para_score = 0.6 # Same length
|
|
|
|
readability_score = self.calculate_readability_score(
|
|
avg_sentence_length,
|
|
formality['score'],
|
|
para_score
|
|
)
|
|
|
|
recommendations = self.get_recommendations({
|
|
'avg_sentence_length': avg_sentence_length,
|
|
'formality': formality,
|
|
'paragraph_structure': paragraph_structure
|
|
})
|
|
|
|
return {
|
|
'avg_sentence_length': round(avg_sentence_length, 1),
|
|
'sentence_count': self.count_sentences(text),
|
|
'word_count': self.count_words(text),
|
|
'grade_level': grade_level,
|
|
'formality': formality,
|
|
'paragraph_structure': paragraph_structure,
|
|
'readability_score': readability_score,
|
|
'recommendations': recommendations
|
|
}
|
|
|
|
|
|
def main():
|
|
"""Main entry point"""
|
|
parser = argparse.ArgumentParser(
|
|
description='Analyze Thai text readability'
|
|
)
|
|
|
|
parser.add_argument(
|
|
'--text', '-t',
|
|
required=True,
|
|
help='Text content to analyze'
|
|
)
|
|
|
|
parser.add_argument(
|
|
'--output', '-o',
|
|
choices=['json', 'text'],
|
|
default='text',
|
|
help='Output format (default: text)'
|
|
)
|
|
|
|
args = parser.parse_args()
|
|
|
|
# Analyze
|
|
analyzer = ThaiReadabilityAnalyzer()
|
|
result = analyzer.analyze(args.text)
|
|
|
|
# Output
|
|
if args.output == 'json':
|
|
print(json.dumps(result, indent=2, ensure_ascii=False))
|
|
else:
|
|
print("\n📖 Thai Readability Analysis\n")
|
|
print(f"Sentence Count: {result['sentence_count']}")
|
|
print(f"Word Count: {result['word_count']}")
|
|
print(f"Avg Sentence Length: {result['avg_sentence_length']} words")
|
|
print(f"\nGrade Level: {result['grade_level']['thai']}")
|
|
print(f"US Equivalent: {result['grade_level']['us_equivalent']}")
|
|
print(f"\nFormality: {result['formality']['level']} (score: {result['formality']['score']})")
|
|
print(f" - Formal particles: {result['formality']['formal_particle_count']}")
|
|
print(f" - Informal particles: {result['formality']['informal_particle_count']}")
|
|
print(f"\nParagraph Structure:")
|
|
print(f" - Count: {result['paragraph_structure']['paragraph_count']}")
|
|
print(f" - Avg length: {result['paragraph_structure']['avg_length_words']} words")
|
|
print(f"\nReadability Score: {result['readability_score']}/100")
|
|
|
|
if result['recommendations']:
|
|
print(f"\n💡 Recommendations:")
|
|
for rec in result['recommendations']:
|
|
print(f" • {rec}")
|
|
|
|
print()
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|