tougao_web/src/utils/referenceJournalAbbrev.js

import { htmlToPlainText } from '@/utils/exportTableWord';

const STOP_WORDS = new Set([
    'a', 'an', 'the', 'of', 'and', 'for', 'in', 'on', 'at', 'to', 'from', 'with', 'by',
    'de', 'du', 'des', 'la', 'le', 'les', 'et', 'und', 'der', 'die', 'das'
]);

const KNOWN_WORD_ABBREVIATIONS = {
    journal: 'J',
    international: 'Int',
    american: 'Am',
    european: 'Eur',
    british: 'Br',
    medicine: 'Med',
    medical: 'Med',
    nursing: 'Nurs',
    inquiry: 'Inq',
    science: 'Sci',
    sciences: 'Sci',
    research: 'Res',
    review: 'Rev',
    reports: 'Rep',
    health: 'Health',
    healthcare: 'Health',
    clinical: 'Clin',
    surgery: 'Surg',
    pediatric: 'Pediatr',
    paediatric: 'Paediatr',
    oncology: 'Oncol',
    psychiatry: 'Psychiatry',
    psychology: 'Psychol',
    education: 'Educ',
    technology: 'Technol',
    engineering: 'Eng',
    management: 'Manage',
    studies: 'Stud',
    practice: 'Pract',
    traditional: 'Tradit',
    tradit: 'Tradit',
    clinical: 'Clin',
    complementary: 'Complement',
    alternative: 'Altern',
    integrative: 'Integr',
    pharmacology: 'Pharmacol',
    biochemistry: 'Biochem',
    physiology: 'Physiol',
    pathology: 'Pathol',
    radiology: 'Radiol',
    cardiology: 'Cardiol',
    neurology: 'Neurol',
    dermatology: 'Dermatol',
    obstetrics: 'Obstet',
    gynecology: 'Gynecol',
    gynaecology: 'Gynaecol',
    anesthesiology: 'Anesthesiol',
    anaesthesiology: 'Anaesthesiol',
    epidemiology: 'Epidemiol',
    immunology: 'Immunol',
    microbiology: 'Microbiol',
    nutrition: 'Nutr',
    public: 'Public',
    environmental: 'Environ',
    occupational: 'Occup',
    behavioral: 'Behav',
    behavioural: 'Behav',
    biological: 'Biol',
    molecular: 'Mol',
    cellular: 'Cell',
    applied: 'Appl',
    advanced: 'Adv',
    general: 'Gen',
    internal: 'Intern',
    family: 'Fam',
    community: 'Community',
    global: 'Glob',
    world: 'World',
    asian: 'Asian',
    chinese: 'Chin',
    japanese: 'Jpn',
    korean: 'Korean',
    indian: 'Indian',
    african: 'Afr',
    australian: 'Aust',
    canadian: 'Can',
    quarterly: 'Q',
    monthly: 'Mon',
    weekly: 'Wkly',
    annual: 'Annu',
    proceedings: 'Proc',
    transactions: 'Trans',
    bulletin: 'Bull',
    letters: 'Lett',
    communications: 'Commun',
    perspectives: 'Perspect',
    insights: 'Insights',
    trends: 'Trends',
    systems: 'Syst',
    informatics: 'Inform',
    bioinformatics: 'Bioinform',
    therapeutics: 'Ther',
    diagnostics: 'Diagn',
    imaging: 'Imaging',
    rehabilitation: 'Rehabil',
    emergency: 'Emerg',
    critical: 'Crit',
    care: 'Care',
    patient: 'Patient',
    patients: 'Patient',
    hospital: 'Hosp',
    hospitals: 'Hosp',
    university: 'Univ',
    academy: 'Acad',
    society: 'Soc',
    association: 'Assoc',
    federation: 'Fed',
    institute: 'Inst',
    college: 'Coll',
    school: 'Sch',
    department: 'Dep',
    division: 'Div',
    section: 'Sect',
    series: 'Ser',
    supplement: 'Suppl',
    edition: 'Ed',
    online: 'Online',
    open: 'Open',
    access: 'Access'
};

function splitJournalWords(journalText) {
    return String(journalText || '')
        .replace(/\./g, ' ')
        .split(/[\s/&]+/)
        .map(function (part) {
            return part.trim();
        })
        .filter(Boolean);
}

function abbreviateJournalWord(word) {
    const cleaned = String(word || '').replace(/[^\p{L}\-]/gu, '');
    if (!cleaned) {
        return '';
    }

    const lower = cleaned.toLowerCase();
    if (KNOWN_WORD_ABBREVIATIONS[lower]) {
        return KNOWN_WORD_ABBREVIATIONS[lower];
    }

    if (cleaned.length <= 4) {
        return cleaned.charAt(0).toUpperCase() + cleaned.slice(1);
    }

    const vowels = 'aeiouy';
    let result = cleaned.charAt(0).toUpperCase();
    const rest = cleaned.slice(1).toLowerCase();
    let index = 0;

    while (result.length < 4 && index < rest.length) {
        const ch = rest.charAt(index);
        index += 1;
        if (!vowels.includes(ch)) {
            result += ch;
        } else if (result.length === 1) {
            result += ch;
        }
    }

    if (result.length < 3) {
        const fallback = cleaned.slice(0, Math.min(4, cleaned.length));
        result = fallback.charAt(0).toUpperCase() + fallback.slice(1).toLowerCase();
    }

    return result;
}

function normalizeJournalText(text) {
    return String(text || '')
        .replace(/\./g, ' ')
        .replace(/\s+/g, ' ')
        .trim();
}

/** 是否已是期刊缩写形式（如 Nurs Inq、Tradit Med Res） */
export function isAbbreviatedJournalName(journalText) {
    const plain = normalizeJournalText(htmlToPlainText(journalText));
    if (!plain) {
        return true;
    }

    const words = splitJournalWords(plain);
    if (!words.length) {
        return true;
    }

    return !words.some(function (word) {
        if (STOP_WORDS.has(word.toLowerCase())) {
            return false;
        }
        return word.length >= 7;
    });
}

/** 将期刊全称转为无句点缩写（如 Nursing Inquiry -> Nurs Inq） */
export function buildJournalAbbreviation(journalText) {
    const plain = normalizeJournalText(htmlToPlainText(journalText));
    if (!plain) {
        return '';
    }

    const abbreviated = splitJournalWords(plain)
        .map(function (word) {
            if (STOP_WORDS.has(word.toLowerCase())) {
                return '';
            }
            return abbreviateJournalWord(word);
        })
        .filter(Boolean);

    return abbreviated.join(' ');
}

/** journal 参考文献期刊名是否需要缩写建议 */
export function needsJournalAbbreviation(journalText) {
    const plain = normalizeJournalText(htmlToPlainText(journalText));
    if (!plain || isAbbreviatedJournalName(plain)) {
        return false;
    }

    const abbreviated = buildJournalAbbreviation(plain);
    if (!abbreviated) {
        return false;
    }

    return plain.toLowerCase() !== abbreviated.toLowerCase();
}

function isJournalReference(ref) {
    return String(ref && ref.refer_type || '').toLowerCase() === 'journal';
}

export function extractJournalSource(ref) {
    if (!isJournalReference(ref)) {
        return '';
    }

    const jouraPlain = htmlToPlainText(ref.joura).trim();
    if (jouraPlain) {
        return jouraPlain;
    }

    return '';
}

export function buildJournalAbbrevComment(ref) {
    const revision = buildJournalAbbrevRevision(ref);
    if (!revision) {
        return '';
    }

    return '期刊名称缩写：' + revision.suggested;
}

export function buildJournalAbbrevRevision(ref) {
    const journalSource = extractJournalSource(ref);
    if (!journalSource || !needsJournalAbbreviation(journalSource)) {
        return null;
    }

    const suggested = buildJournalAbbreviation(journalSource);
    if (!suggested || journalSource.toLowerCase() === suggested.toLowerCase()) {
        return null;
    }

    return {
        original: journalSource,
        suggested: suggested,
        suffix: '. ',
        italics: true
    };
}