gkxk
1f7a6a2054
- Fix unicode surrogate pairs and emoji slicing via Array.from in ORP calculator - Add slot smoothing constraint to pivot element to prevent horizontal jitter - Implement adaptive timestamp compensation and visibilitychange auto-pause - Slim storage layer by updating only STORE_META on progress changes with memory fallback - Align EPUB chapter indices to token array indices instead of raw character offsets - Support standalone emoji tokens via Extended_Pictographic regex - Code-split JSZip via dynamic import to optimize initial bundle size Co-Authored-By: Claude Code <noreply@anthropic.com>
81 lines
2.1 KiB
TypeScript
81 lines
2.1 KiB
TypeScript
/**
|
|
* Optimal Recognition Point (ORP) Calculation for Spritz RSVP Reading.
|
|
* Fixates the eye at ~30%-35% into the word where human ocular fixation naturally rests.
|
|
* Uses unicode-safe code point segmentation (Array.from) to eliminate surrogate pair
|
|
* and emoji splitting defects.
|
|
*/
|
|
|
|
const CJK_REGEX = /[一-龥-ヿ가-]/;
|
|
|
|
export function isCJKChar(char: string): boolean {
|
|
return CJK_REGEX.test(char);
|
|
}
|
|
|
|
export function isCJKWord(word: string): boolean {
|
|
return CJK_REGEX.test(word);
|
|
}
|
|
|
|
/**
|
|
* Calculates the 0-based index of the ORP character in cleanWord.
|
|
* Based on unicode code-point length (Array.from(cleanWord).length).
|
|
*/
|
|
export function calculateOrpIndex(cleanWord: string): number {
|
|
const chars = Array.from(cleanWord);
|
|
const len = chars.length;
|
|
if (len <= 1) return 0;
|
|
|
|
if (isCJKWord(cleanWord)) {
|
|
// CJK characters have high semantic density per glyph
|
|
if (len === 2) return 0;
|
|
if (len <= 4) return 1;
|
|
if (len <= 6) return 2;
|
|
return Math.min(len - 1, Math.floor((len - 1) * 0.35));
|
|
}
|
|
|
|
// Standard Spritz ORP Table for alphabetic words
|
|
if (len <= 5) return 1;
|
|
if (len <= 9) return 2;
|
|
if (len <= 13) return 3;
|
|
return 4;
|
|
}
|
|
|
|
export interface SplitWord {
|
|
prefix: string;
|
|
pivot: string;
|
|
suffix: string;
|
|
orpIndex: number;
|
|
}
|
|
|
|
/**
|
|
* Splits a token into prefix, pivot (ORP character), and suffix.
|
|
* Handles leading punctuation (e.g. "“", "《", "(") and trailing punctuation.
|
|
* Operates purely on unicode code points to prevent corrupting surrogate pairs.
|
|
*/
|
|
export function splitOrp(
|
|
coreWord: string,
|
|
leadingPunct: string = '',
|
|
trailingPunct: string = ''
|
|
): SplitWord {
|
|
if (!coreWord) {
|
|
return {
|
|
prefix: leadingPunct,
|
|
pivot: '',
|
|
suffix: trailingPunct,
|
|
orpIndex: 0,
|
|
};
|
|
}
|
|
|
|
const chars = Array.from(coreWord);
|
|
const orpIndex = calculateOrpIndex(coreWord);
|
|
const prefix = leadingPunct + chars.slice(0, orpIndex).join('');
|
|
const pivot = chars[orpIndex] || '';
|
|
const suffix = chars.slice(orpIndex + 1).join('') + trailingPunct;
|
|
|
|
return {
|
|
prefix,
|
|
pivot,
|
|
suffix,
|
|
orpIndex,
|
|
};
|
|
}
|