| 1 |
/** |
| 2 |
* Content Analysis Utilities |
| 3 |
* |
| 4 |
* Utility functions for SEO content analysis, migrated from jQuery implementation |
| 5 |
* to maintain consistency with original functionality |
| 6 |
* |
| 7 |
* @package ThinkRank |
| 8 |
* @since 1.0.0 |
| 9 |
*/ |
| 10 |
|
| 11 |
/** |
| 12 |
* Calculate real word count from text |
| 13 |
* Industry standard word counting for SEO analysis |
| 14 |
* |
| 15 |
* @param {string} text - Text content to analyze |
| 16 |
* @returns {number} Word count |
| 17 |
*/ |
| 18 |
export const calculateWordCount = (text) => { |
| 19 |
if (!text || typeof text !== 'string') return 0; |
| 20 |
|
| 21 |
// Remove extra whitespace and split by spaces |
| 22 |
const words = text.trim().split(/\s+/).filter(word => word.length > 0); |
| 23 |
return words.length; |
| 24 |
}; |
| 25 |
|
| 26 |
/** |
| 27 |
* Calculate keyword density using industry standard methodology |
| 28 |
* |
| 29 |
* @param {string} content - Content to analyze |
| 30 |
* @param {string} keyword - Target keyword |
| 31 |
* @returns {number} Keyword density percentage (rounded to 1 decimal) |
| 32 |
*/ |
| 33 |
export const calculateKeywordDensity = (content, keyword) => { |
| 34 |
if (!content || !keyword) return 0; |
| 35 |
|
| 36 |
// Convert to lowercase for case-insensitive matching |
| 37 |
const lowerContent = content.toLowerCase(); |
| 38 |
const lowerKeyword = keyword.toLowerCase(); |
| 39 |
|
| 40 |
// Count total words |
| 41 |
const totalWords = calculateWordCount(content); |
| 42 |
if (totalWords === 0) return 0; |
| 43 |
|
| 44 |
// Count keyword occurrences (including partial matches) |
| 45 |
const keywordOccurrences = (lowerContent.match(new RegExp(lowerKeyword, 'g')) || []).length; |
| 46 |
|
| 47 |
// Calculate density as percentage |
| 48 |
const density = (keywordOccurrences / totalWords) * 100; |
| 49 |
return Math.round(density * 10) / 10; // Round to 1 decimal place |
| 50 |
}; |
| 51 |
|
| 52 |
/** |
| 53 |
* Estimate syllable count for readability calculations |
| 54 |
* Simple approximation used in Flesch Reading Ease formula |
| 55 |
* |
| 56 |
* @param {string} text - Text to analyze |
| 57 |
* @returns {number} Estimated syllable count |
| 58 |
*/ |
| 59 |
export const estimateSyllables = (text) => { |
| 60 |
const words = text.toLowerCase().match(/\b[a-z]+\b/g) || []; |
| 61 |
let syllableCount = 0; |
| 62 |
|
| 63 |
words.forEach(word => { |
| 64 |
// Simple syllable counting heuristic |
| 65 |
const vowels = word.match(/[aeiouy]+/g) || []; |
| 66 |
let syllables = vowels.length; |
| 67 |
|
| 68 |
// Adjust for silent 'e' |
| 69 |
if (word.endsWith('e') && syllables > 1) { |
| 70 |
syllables--; |
| 71 |
} |
| 72 |
|
| 73 |
// Minimum 1 syllable per word |
| 74 |
syllableCount += Math.max(1, syllables); |
| 75 |
}); |
| 76 |
|
| 77 |
return syllableCount; |
| 78 |
}; |
| 79 |
|
| 80 |
/** |
| 81 |
* Calculate readability score using industry standard Flesch Reading Ease |
| 82 |
* Official formula used by academic institutions and professional tools |
| 83 |
* |
| 84 |
* @param {string} text - Text content to analyze |
| 85 |
* @returns {string} Readability assessment with score |
| 86 |
*/ |
| 87 |
export const calculateReadabilityScore = (text) => { |
| 88 |
if (!text) return 'No content'; |
| 89 |
|
| 90 |
const words = calculateWordCount(text); |
| 91 |
const sentences = (text.match(/[.!?]+/g) || []).length; |
| 92 |
const syllables = estimateSyllables(text); |
| 93 |
|
| 94 |
if (sentences === 0 || words === 0) return 'Too short'; |
| 95 |
|
| 96 |
const avgWordsPerSentence = words / sentences; |
| 97 |
const avgSyllablesPerWord = syllables / words; |
| 98 |
|
| 99 |
// Official Flesch Reading Ease formula |
| 100 |
const fleschScore = 206.835 - (1.015 * avgWordsPerSentence) - (84.6 * avgSyllablesPerWord); |
| 101 |
const roundedScore = Math.round(fleschScore); |
| 102 |
|
| 103 |
// Industry standard Flesch Reading Ease scale |
| 104 |
let level = ''; |
| 105 |
if (roundedScore >= 90) level = 'Very Easy'; |
| 106 |
else if (roundedScore >= 80) level = 'Easy'; |
| 107 |
else if (roundedScore >= 70) level = 'Fairly Easy'; |
| 108 |
else if (roundedScore >= 60) level = 'Standard'; |
| 109 |
else if (roundedScore >= 50) level = 'Fairly Difficult'; |
| 110 |
else if (roundedScore >= 30) level = 'Difficult'; |
| 111 |
else level = 'Very Difficult'; |
| 112 |
|
| 113 |
// Return professional format |
| 114 |
return `${level} (${roundedScore})`; |
| 115 |
}; |
| 116 |
|
| 117 |
/** |
| 118 |
* Analyze heading structure from HTML content |
| 119 |
* |
| 120 |
* @param {string} htmlContent - HTML content to analyze |
| 121 |
* @returns {Object} Heading analysis with count and structure description |
| 122 |
*/ |
| 123 |
export const analyzeHeadingStructure = (htmlContent) => { |
| 124 |
if (!htmlContent) return { count: 0, structure: 'No content' }; |
| 125 |
|
| 126 |
// Create a temporary div to parse HTML |
| 127 |
const tempDiv = document.createElement('div'); |
| 128 |
tempDiv.innerHTML = htmlContent; |
| 129 |
|
| 130 |
// Count different heading levels |
| 131 |
const h1Count = tempDiv.querySelectorAll('h1').length; |
| 132 |
const h2Count = tempDiv.querySelectorAll('h2').length; |
| 133 |
const h3Count = tempDiv.querySelectorAll('h3').length; |
| 134 |
const h4Count = tempDiv.querySelectorAll('h4').length; |
| 135 |
const h5Count = tempDiv.querySelectorAll('h5').length; |
| 136 |
const h6Count = tempDiv.querySelectorAll('h6').length; |
| 137 |
|
| 138 |
const totalHeadings = h1Count + h2Count + h3Count + h4Count + h5Count + h6Count; |
| 139 |
|
| 140 |
let structure = ''; |
| 141 |
if (totalHeadings === 0) { |
| 142 |
structure = 'No headings'; |
| 143 |
} else if (h1Count === 0) { |
| 144 |
structure = `${totalHeadings} headings (No H1)`; |
| 145 |
} else if (h1Count > 1) { |
| 146 |
structure = `${totalHeadings} headings (${h1Count} H1s)`; |
| 147 |
} else { |
| 148 |
structure = `${totalHeadings} headings (Good H1)`; |
| 149 |
} |
| 150 |
|
| 151 |
return { count: totalHeadings, structure: structure }; |
| 152 |
}; |
| 153 |
|
| 154 |
/** |
| 155 |
* Calculate content quality using industry standard SEO methodology |
| 156 |
* Based on established content analysis best practices |
| 157 |
* |
| 158 |
* @param {string} plainText - Plain text content |
| 159 |
* @param {string} htmlContent - HTML content for structure analysis |
| 160 |
* @returns {string} Content quality assessment |
| 161 |
*/ |
| 162 |
export const calculateContentQuality = (plainText, htmlContent) => { |
| 163 |
if (!plainText) return 'No content'; |
| 164 |
|
| 165 |
const wordCount = calculateWordCount(plainText); |
| 166 |
let issues = []; |
| 167 |
let score = 0; |
| 168 |
|
| 169 |
// 1. Word count analysis (Industry standard: 300+ words) |
| 170 |
if (wordCount < 300) { |
| 171 |
issues.push('Too short'); |
| 172 |
score += 0; |
| 173 |
} else if (wordCount >= 300 && wordCount < 600) { |
| 174 |
score += 50; |
| 175 |
} else { |
| 176 |
score += 100; // 600+ words is excellent |
| 177 |
} |
| 178 |
|
| 179 |
// 2. Paragraph length analysis (SEO best practice: avoid long paragraphs) |
| 180 |
const tempDiv = document.createElement('div'); |
| 181 |
tempDiv.innerHTML = htmlContent; |
| 182 |
const paragraphs = tempDiv.querySelectorAll('p'); |
| 183 |
let longParagraphs = 0; |
| 184 |
|
| 185 |
paragraphs.forEach(p => { |
| 186 |
const pText = p.textContent || p.innerText || ''; |
| 187 |
const pWordCount = calculateWordCount(pText); |
| 188 |
if (pWordCount > 150) { // Industry standard: flag paragraphs over 150 words |
| 189 |
longParagraphs++; |
| 190 |
} |
| 191 |
}); |
| 192 |
|
| 193 |
if (longParagraphs === 0) { |
| 194 |
score += 100; |
| 195 |
} else if (longParagraphs <= 2) { |
| 196 |
score += 50; |
| 197 |
issues.push('Some long paragraphs'); |
| 198 |
} else { |
| 199 |
score += 0; |
| 200 |
issues.push('Too many long paragraphs'); |
| 201 |
} |
| 202 |
|
| 203 |
// 3. Subheading distribution (SEO best practice: subheadings every 300 words) |
| 204 |
const headings = tempDiv.querySelectorAll('h2, h3, h4, h5, h6').length; |
| 205 |
const expectedHeadings = Math.floor(wordCount / 300); |
| 206 |
|
| 207 |
if (headings >= expectedHeadings && headings > 0) { |
| 208 |
score += 100; |
| 209 |
} else if (headings > 0) { |
| 210 |
score += 50; |
| 211 |
issues.push('Could use more subheadings'); |
| 212 |
} else { |
| 213 |
score += 0; |
| 214 |
issues.push('No subheadings'); |
| 215 |
} |
| 216 |
|
| 217 |
// Calculate final score (0-100) |
| 218 |
const finalScore = Math.round(score / 3); // Average of 3 criteria |
| 219 |
|
| 220 |
// Return professional assessment |
| 221 |
if (finalScore >= 80) return 'Good'; |
| 222 |
else if (finalScore >= 50) return 'OK'; |
| 223 |
else return 'Needs improvement'; |
| 224 |
}; |
| 225 |
|
| 226 |
/** |
| 227 |
* Calculate keyword optimization using industry standard SEO methodology |
| 228 |
* Based on established SEO best practices |
| 229 |
* |
| 230 |
* @param {string} plainText - Plain text content |
| 231 |
* @param {string} htmlContent - HTML content for structure analysis |
| 232 |
* @param {string} title - Post title |
| 233 |
* @param {string} targetKeyword - Focus keyword |
| 234 |
* @param {string} metaDescription - Meta description |
| 235 |
* @returns {string} Keyword optimization assessment |
| 236 |
*/ |
| 237 |
export const calculateKeywordOptimization = (plainText, htmlContent, title, targetKeyword, metaDescription = '') => { |
| 238 |
if (!targetKeyword || !targetKeyword.trim()) { |
| 239 |
return 'No focus keyword'; |
| 240 |
} |
| 241 |
|
| 242 |
if (!plainText) return 'No content'; |
| 243 |
|
| 244 |
const keywordLower = targetKeyword.toLowerCase(); |
| 245 |
const contentLower = plainText.toLowerCase(); |
| 246 |
const titleLower = title.toLowerCase(); |
| 247 |
|
| 248 |
let checks = 0; |
| 249 |
let totalChecks = 0; |
| 250 |
|
| 251 |
// 1. Keyword in title (Essential SEO check) |
| 252 |
totalChecks++; |
| 253 |
if (titleLower.includes(keywordLower)) { |
| 254 |
checks++; |
| 255 |
} |
| 256 |
|
| 257 |
// 2. Keyword density (Industry standard: 0.5% - 2.5%) |
| 258 |
totalChecks++; |
| 259 |
const density = calculateKeywordDensity(plainText, targetKeyword); |
| 260 |
if (density >= 0.5 && density <= 2.5) { |
| 261 |
checks++; |
| 262 |
} |
| 263 |
|
| 264 |
// 3. Keyword in first paragraph (SEO best practice) |
| 265 |
totalChecks++; |
| 266 |
const tempDiv = document.createElement('div'); |
| 267 |
tempDiv.innerHTML = htmlContent; |
| 268 |
const firstParagraph = tempDiv.querySelector('p'); |
| 269 |
if (firstParagraph) { |
| 270 |
const firstParagraphText = (firstParagraph.textContent || firstParagraph.innerText || '').toLowerCase(); |
| 271 |
if (firstParagraphText.includes(keywordLower)) { |
| 272 |
checks++; |
| 273 |
} |
| 274 |
} |
| 275 |
|
| 276 |
// 4. Keyword in subheadings (Content structure optimization) |
| 277 |
totalChecks++; |
| 278 |
const subheadings = tempDiv.querySelectorAll('h2, h3, h4, h5, h6'); |
| 279 |
let keywordInSubheading = false; |
| 280 |
subheadings.forEach(heading => { |
| 281 |
const headingText = (heading.textContent || heading.innerText || '').toLowerCase(); |
| 282 |
if (headingText.includes(keywordLower)) { |
| 283 |
keywordInSubheading = true; |
| 284 |
} |
| 285 |
}); |
| 286 |
if (keywordInSubheading) { |
| 287 |
checks++; |
| 288 |
} |
| 289 |
|
| 290 |
// 5. Keyword in meta description (Meta optimization) |
| 291 |
if (metaDescription) { |
| 292 |
totalChecks++; |
| 293 |
if (metaDescription.toLowerCase().includes(keywordLower)) { |
| 294 |
checks++; |
| 295 |
} |
| 296 |
} |
| 297 |
|
| 298 |
// Calculate score based on checks passed |
| 299 |
const percentage = Math.round((checks / totalChecks) * 100); |
| 300 |
|
| 301 |
// Return professional assessment |
| 302 |
if (percentage >= 80) return 'Good'; |
| 303 |
else if (percentage >= 60) return 'OK'; |
| 304 |
else return 'Needs improvement'; |
| 305 |
}; |
| 306 |
|
