<?php

namespace app\common;

use think\Db;

/**
 * 参考文献引用堆叠统计：同刊、同作者、自引（实时计算，不入库）
 */
class ReferenceStackingStatsService
{
    const DETAIL_JOURNAL = 'journal';
    const DETAIL_AUTHOR = 'author';

    const THRESHOLD_SAME_AUTHOR = 0.15;
    const THRESHOLD_SAME_JOURNAL = 0.20;
    const THRESHOLD_SELF_CITATION = 0.10;

    /** @var ReferenceCheckService */
    private $refUtil;

    /** @var CrossrefService */
    private $crossref;

    /** @var ReferenceAuthorIdentityService */
    private $identity;

    public function __construct()
    {
        $this->refUtil = new ReferenceCheckService();
        $this->crossref = new CrossrefService([
            'mailto' => trim((string)\think\Env::get('crossref_mailto', '')),
        ]);
        $this->identity = new ReferenceAuthorIdentityService();
    }

    /**
     * 按阈值规则实时统计：同作者(>15%)、同刊(>20%)、自引(>10%)
     *
     * @param int $pArticleId
     * @return array
     */
    public function getThresholdStackingByPArticleId($pArticleId)
    {
        $pArticleId = intval($pArticleId);
        if ($pArticleId <= 0) {
            throw new \InvalidArgumentException('p_article_id is required');
        }

        return $this->formatThresholdStackingReport($this->compute($pArticleId));
    }

    /**
     * @param array $full compute/analyze 或 getStored 的完整结果
     */
    public function formatThresholdStackingReport(array $full)
    {
        $total = max(1, intval($full['total_references'] ?? 0));
        $referMap = (array)($full['refer_map'] ?? []);
        if (empty($referMap) && intval($full['p_article_id'] ?? 0) > 0) {
            $referMap = $this->loadReferMapByPArticleId(intval($full['p_article_id']));
        }

        $authorItems = [];
        foreach ((array)($full['author_details'] ?? []) as $item) {
            $count = intval($item['ref_count'] ?? 0);
            $ratio = $count / $total;
            if ($ratio <= self::THRESHOLD_SAME_AUTHOR) {
                continue;
            }
            $pReferIds = array_values((array)($item['p_refer_ids'] ?? []));
            $authorItems[] = [
                'author_name'   => (string)($item['group_name'] ?? ''),
                'orcid'         => (string)($item['orcid'] ?? ''),
                'cite_count'    => $count,
                'cite_ratio'    => round($ratio, 4),
                'threshold'     => self::THRESHOLD_SAME_AUTHOR,
                'exceeded'      => true,
                'p_refer_ids'   => $pReferIds,
                'reference_nos' => array_values((array)($item['reference_nos'] ?? [])),
                'references'    => $this->buildReferBriefs($pReferIds, $referMap),
            ];
        }

        $journalItems = [];
        foreach ((array)($full['journal_details'] ?? []) as $item) {
            $count = intval($item['ref_count'] ?? 0);
            $ratio = $count / $total;
            if ($ratio <= self::THRESHOLD_SAME_JOURNAL) {
                continue;
            }
            $pReferIds = array_values((array)($item['p_refer_ids'] ?? []));
            $journalItems[] = [
                'journal_name'  => (string)($item['group_name'] ?? ''),
                'cite_count'    => $count,
                'cite_ratio'    => round($ratio, 4),
                'threshold'     => self::THRESHOLD_SAME_JOURNAL,
                'exceeded'      => true,
                'p_refer_ids'   => $pReferIds,
                'reference_nos' => array_values((array)($item['reference_nos'] ?? [])),
                'references'    => $this->buildReferBriefs($pReferIds, $referMap),
            ];
        }

        $selfDetails = (array)($full['self_citation_details'] ?? []);
        $selfCount = count($selfDetails);
        $selfRatio = $selfCount / $total;
        $selfItems = [];
        foreach ($selfDetails as $item) {
            $pReferId = intval($item['p_refer_id'] ?? 0);
            $selfItems[] = [
                'manuscript_author' => (string)($item['matched_manuscript_author'] ?? ''),
                'manuscript_orcid'  => (string)($item['matched_orcid'] ?? ''),
                'matched_refer_author' => (string)($item['matched_refer_author'] ?? ''),
                'reference_no'      => intval($item['reference_no'] ?? 0),
                'p_refer_id'        => $pReferId,
                'reference'         => $this->buildReferBriefs([$pReferId], $referMap)[0] ?? null,
            ];
        }

        return [
            'p_article_id'     => intval($full['p_article_id'] ?? 0),
            'article_id'       => intval($full['article_id'] ?? 0),
            'total_references' => intval($full['total_references'] ?? 0),
            'same_author_stacking' => [
                'threshold' => self::THRESHOLD_SAME_AUTHOR,
                'exceeded'  => !empty($authorItems),
                'items'     => $authorItems,
            ],
            'same_journal_stacking' => [
                'threshold' => self::THRESHOLD_SAME_JOURNAL,
                'exceeded'  => !empty($journalItems),
                'items'     => $journalItems,
            ],
            'self_citation' => [
                'threshold'     => self::THRESHOLD_SELF_CITATION,
                'exceeded'      => $selfRatio > self::THRESHOLD_SELF_CITATION,
                'cite_count'    => $selfCount,
                'cite_ratio'    => round($selfRatio, 4),
                'reference_nos' => array_values((array)($full['self_citation_reference_nos'] ?? [])),
                'items'         => $selfItems,
                'note'          => (string)($full['author_identity_note'] ?? ''),
            ],
            'computed_at' => (string)($full['computed_at'] ?? ''),
        ];
    }

    /**
     * @param int $pArticleId
     * @return array
     */
    public function compute($pArticleId)
    {
        $pArticleId = intval($pArticleId);
        if ($pArticleId <= 0) {
            throw new \InvalidArgumentException('p_article_id is required');
        }

        DbReconnectHelper::release();

        $articleId = $this->resolveArticleId($pArticleId);

        $refers = Db::name('production_article_refer')
            ->field('p_refer_id,index,author,joura,refer_type,refer_doi,doilink,refer_content,refer_frag')
            ->where('p_article_id', $pArticleId)
            ->where('state', 0)
            ->order('index asc')
            ->select();

        $manuscriptAuthors = $this->identity->resolveManuscriptAuthors($pArticleId);
        $ambiguousManuscriptNameKeys = $this->buildAmbiguousManuscriptNameKeys($manuscriptAuthors);
        $doiCache = [];
        $referMap = [];

        $journalBuckets = [];
        $authorBuckets = [];
        $selfCitationDetails = [];

        foreach ($refers as $refer) {
            $refNo = intval($refer['index']) + 1;
            $pReferId = intval($refer['p_refer_id']);
            $referMap[$pReferId] = $refer;

            $meta = $this->resolveReferMeta($refer, $doiCache);
            $joura = (string)$meta['joura'];
            $journalKey = $this->normalizeJournalKey($joura);
            if ($journalKey !== '') {
                if (!isset($journalBuckets[$journalKey])) {
                    $journalBuckets[$journalKey] = [
                        'group_key'      => $journalKey,
                        'name'           => trim(trim($joura), '.'),
                        'count'          => 0,
                        'reference_nos'  => [],
                        'p_refer_ids'    => [],
                    ];
                }
                $journalBuckets[$journalKey]['count']++;
                $journalBuckets[$journalKey]['reference_nos'][] = $refNo;
                $journalBuckets[$journalKey]['p_refer_ids'][] = $pReferId;
            }

            $referAuthors = $this->resolveReferAuthorsWithMeta($pReferId, $meta);
            $this->accumulateAuthorBuckets($authorBuckets, $referAuthors, $refNo, $pReferId);

            $matchedManuscript = $this->matchManuscriptAuthorForSelfCitation(
                $referAuthors,
                $manuscriptAuthors,
                $ambiguousManuscriptNameKeys
            );
            if ($matchedManuscript !== null) {
                $referAuthorNames = array_map(function ($ra) {
                    return (string)($ra['name'] ?? '');
                }, $referAuthors);
                $selfCitationDetails[] = [
                    'reference_no'              => $refNo,
                    'p_refer_id'                => $pReferId,
                    'refer_author'              => implode(', ', array_filter($referAuthorNames)),
                    'matched_refer_author'      => (string)($matchedManuscript['matched_refer_author'] ?? ''),
                    'matched_manuscript_author' => (string)($matchedManuscript['display_name'] ?? ''),
                    'matched_orcid'             => (string)($matchedManuscript['orcid'] ?? ''),
                    'match_confidence'          => ReferenceAuthorIdentityService::MATCH_FUZZY,
                    'meta_source'               => (string)$meta['meta_source'],
                ];
            }
        }

        $journalDetails = $this->formatBucketDetails($journalBuckets, self::DETAIL_JOURNAL);
        $authorDetails = $this->formatAuthorBucketDetails($authorBuckets, self::DETAIL_AUTHOR);
        $selfDetails = $this->formatSelfCitationDetails($selfCitationDetails);

        $selfCitationRefNos = array_column($selfCitationDetails, 'reference_no');

        return [
            'article_id'                  => $articleId,
            'p_article_id'                => $pArticleId,
            'total_references'            => count($refers),
            'self_citation_reference_nos' => array_values(array_unique($selfCitationRefNos)),
            'journal_details'             => $journalDetails,
            'author_details'              => $authorDetails,
            'self_citation_details'       => $selfDetails,
            'refer_map'                   => $referMap,
            'author_identity_note'        => '同作者堆叠按 citation_name（空则 display_name）姓名精确匹配，同名即同人；本文出现重名作者（如两位 Lin）时跳过该姓名的自引判定。优先读 t_production_article_refer_author，否则解析 refer.author。',
            'computed_at'                 => date('Y-m-d H:i:s'),
        ];
    }

    private function formatAuthorBucketDetails(array $buckets, $detailType)
    {
        $list = [];
        foreach ($buckets as $bucket) {
            $item = [
                'detail_type'      => $detailType,
                'group_key'        => (string)($bucket['group_key'] ?? ''),
                'group_name'       => (string)($bucket['name'] ?? ''),
                'ref_count'        => intval($bucket['count'] ?? 0),
                'reference_nos'    => array_values((array)($bucket['reference_nos'] ?? [])),
                'p_refer_ids'      => array_values((array)($bucket['p_refer_ids'] ?? [])),
                'match_confidence' => (string)($bucket['match_confidence'] ?? ''),
            ];
            if (!empty($bucket['openalex_id'])) {
                $item['openalex_id'] = (string)$bucket['openalex_id'];
            }
            if (!empty($bucket['orcid'])) {
                $item['orcid'] = (string)$bucket['orcid'];
            }
            $list[] = $item;
        }

        usort($list, function ($a, $b) {
            $cmp = intval($b['ref_count']) <=> intval($a['ref_count']);
            if ($cmp !== 0) {
                return $cmp;
            }
            return strcmp((string)$a['group_name'], (string)$b['group_name']);
        });

        return $list;
    }

    /**
     * @return array<int, array{name:string,orcid:string}>
     */
    private function resolveReferAuthorsWithMeta($pReferId, array $meta)
    {
        $pReferId = intval($pReferId);
        $list = [];

        if ($pReferId > 0) {
            $rows = Db::name('production_article_refer_author')
                ->where('p_refer_id', $pReferId)
                ->order('author_seq asc, id asc')
                ->field('display_name,citation_name,orcid')
                ->select();
            foreach ($rows as $row) {
                $name = trim((string)($row['citation_name'] ?? ''));
                if ($name === '') {
                    $name = trim((string)($row['display_name'] ?? ''));
                }
                if ($name === '') {
                    continue;
                }
                $list[] = [
                    'name'  => $name,
                    'orcid' => $this->cleanOrcid($row['orcid'] ?? ''),
                ];
            }
        }

        if (!empty($list)) {
            return $list;
        }

        foreach ($this->parseAuthorStringParts((string)($meta['author'] ?? '')) as $name) {
            $list[] = [
                'name'  => $name,
                'orcid' => '',
            ];
        }

        return $list;
    }

    /**
     * @return string[]
     */
    private function parseAuthorStringParts($authorString)
    {
        $authorString = trim(trim((string)$authorString), '.');
        if ($authorString === '') {
            return [];
        }

        $authorString = preg_replace('/\s+et\s+al\.?\s*$/iu', '', $authorString);
        $names = [];
        foreach (preg_split('/,\s*/u', $authorString) as $part) {
            $part = trim($part);
            if ($part === '' || preg_match('/^et\s+al\.?$/iu', $part)) {
                continue;
            }
            $names[] = $part;
        }

        return $names;
    }

    /**
     * 本文多位作者姓名归一化后相同（如两位 Lin）则视为歧义，不参与自引匹配
     *
     * @return array<string,bool>
     */
    private function buildAmbiguousManuscriptNameKeys(array $manuscriptAuthors)
    {
        $counts = [];
        foreach ($manuscriptAuthors as $author) {
            $key = $this->normalizeAuthorNameKey((string)($author['display_name'] ?? ''));
            if ($key === '') {
                continue;
            }
            if (!isset($counts[$key])) {
                $counts[$key] = 0;
            }
            $counts[$key]++;
        }

        $ambiguous = [];
        foreach ($counts as $key => $count) {
            if ($count > 1) {
                $ambiguous[$key] = true;
            }
        }

        return $ambiguous;
    }

    /**
     * @param array<int, array{name:string,orcid:string}> $referAuthors
     * @param array<int, array{display_name:string,orcid:string}> $manuscriptAuthors
     * @param array<string,bool> $ambiguousNameKeys
     * @return array{display_name:string,orcid:string,matched_refer_author:string}|null
     */
    private function matchManuscriptAuthorForSelfCitation(array $referAuthors, array $manuscriptAuthors, array $ambiguousNameKeys)
    {
        foreach ($referAuthors as $referAuthor) {
            $referName = trim((string)($referAuthor['name'] ?? ''));
            $referKey = $this->normalizeAuthorNameKey($referName);
            if ($referKey === '' || !empty($ambiguousNameKeys[$referKey])) {
                continue;
            }

            foreach ($manuscriptAuthors as $manuscriptAuthor) {
                $manuscriptName = trim((string)($manuscriptAuthor['display_name'] ?? ''));
                $manuscriptKey = $this->normalizeAuthorNameKey($manuscriptName);
                if ($manuscriptKey !== '' && $manuscriptKey === $referKey) {
                    return [
                        'display_name'         => $manuscriptName,
                        'orcid'                => trim((string)($manuscriptAuthor['orcid'] ?? '')),
                        'matched_refer_author' => $referName,
                    ];
                }
            }
        }

        return null;
    }

    /**
     * @param array<int, array{name:string,orcid:string}> $referAuthors
     */
    private function accumulateAuthorBuckets(array &$buckets, array $referAuthors, $refNo, $pReferId)
    {
        if (empty($referAuthors)) {
            return;
        }

        $seenKeys = [];
        foreach ($referAuthors as $author) {
            $name = trim((string)($author['name'] ?? ''));
            if ($name === '' || preg_match('/^et\s+al\.?$/iu', $name)) {
                continue;
            }

            $key = $this->normalizeAuthorNameKey($name);
            if ($key === '' || isset($seenKeys[$key])) {
                continue;
            }
            $seenKeys[$key] = true;

            if (!isset($buckets[$key])) {
                $buckets[$key] = [
                    'group_key'        => $key,
                    'name'             => $name,
                    'orcid'            => '',
                    'match_confidence' => ReferenceAuthorIdentityService::MATCH_FUZZY,
                    'count'            => 0,
                    'reference_nos'    => [],
                    'p_refer_ids'      => [],
                ];
            }

            $orcid = trim((string)($author['orcid'] ?? ''));
            if ($orcid !== '' && trim((string)($buckets[$key]['orcid'] ?? '')) === '') {
                $buckets[$key]['orcid'] = $orcid;
            }

            $buckets[$key]['count']++;
            $buckets[$key]['reference_nos'][] = $refNo;
            $buckets[$key]['p_refer_ids'][] = $pReferId;
        }
    }

    private function normalizeAuthorNameKey($name)
    {
        $name = trim(preg_replace('/\.+$/u', '', trim((string)$name)));
        $name = preg_replace('/\s+/u', ' ', $name);
        if ($name === '') {
            return '';
        }

        return mb_strtolower($name);
    }

    /**
     * @param int[] $pReferIds
     * @param array<int,array> $referMap
     * @return array<int,array{p_refer_id:int,reference_no:int,refer_text:string}>
     */
    private function buildReferBriefs(array $pReferIds, array $referMap)
    {
        $list = [];
        foreach ($pReferIds as $pReferId) {
            $pReferId = intval($pReferId);
            if ($pReferId <= 0 || empty($referMap[$pReferId])) {
                continue;
            }
            $refer = $referMap[$pReferId];
            $list[] = [
                'p_refer_id'   => $pReferId,
                'reference_no' => intval($refer['index'] ?? 0) + 1,
                'refer_text'   => $this->referSnippet($refer),
            ];
        }

        return $list;
    }

    private function loadReferMapByPArticleId($pArticleId)
    {
        $pArticleId = intval($pArticleId);
        if ($pArticleId <= 0) {
            return [];
        }

        DbReconnectHelper::release();
        $refers = Db::name('production_article_refer')
            ->field('p_refer_id,index,author,joura,refer_type,refer_doi,doilink,refer_content,refer_frag')
            ->where('p_article_id', $pArticleId)
            ->where('state', 0)
            ->order('index asc')
            ->select();

        $map = [];
        foreach ($refers as $refer) {
            $map[intval($refer['p_refer_id'])] = $refer;
        }

        return $map;
    }

    private function formatBucketDetails(array $buckets, $detailType)
    {
        $list = [];
        foreach ($buckets as $bucket) {
            $list[] = [
                'detail_type'   => $detailType,
                'group_key'     => (string)($bucket['group_key'] ?? ''),
                'group_name'    => (string)($bucket['name'] ?? ''),
                'ref_count'     => intval($bucket['count'] ?? 0),
                'reference_nos' => array_values((array)($bucket['reference_nos'] ?? [])),
                'p_refer_ids'   => array_values((array)($bucket['p_refer_ids'] ?? [])),
            ];
        }

        usort($list, function ($a, $b) {
            $cmp = intval($b['ref_count']) <=> intval($a['ref_count']);
            if ($cmp !== 0) {
                return $cmp;
            }
            return strcmp((string)$a['group_name'], (string)$b['group_name']);
        });

        return $list;
    }

    private function formatSelfCitationDetails(array $items)
    {
        usort($items, function ($a, $b) {
            return intval($a['reference_no']) <=> intval($b['reference_no']);
        });
        return $items;
    }

    /**
     * 合并 refer 行已有字段、Crossref、refer_frag/refer_content 解析结果
     *
     * @return array{
     *   author:string,
     *   joura:string,
     *   author_keys:string[],
     *   first_author_display:string,
     *   meta_source:string,
     *   resolved:bool
     * }
     */
    private function resolveReferMeta(array $refer, array &$doiCache)
    {
        $author = trim(trim((string)($refer['author'] ?? '')), '.');
        $joura = trim(trim((string)($refer['joura'] ?? '')), '.');
        $sources = [];
        $authorKeys = [];

        if ($author !== '') {
            $sources[] = 'local';
            $authorKeys = $this->extractAuthorKeysFromAuthorString($author);
        }

        $doi = $this->refUtil->extractDoiFromRefer($refer);
        $summary = null;
        if ($doi !== '') {
            if (!array_key_exists($doi, $doiCache)) {
                try {
                    $doiCache[$doi] = $this->crossref->fetchWorkSummary($doi);
                } catch (\Throwable $e) {
                    $doiCache[$doi] = null;
                }
            }
            $summary = $doiCache[$doi];
        }

        if (is_array($summary)) {
            $sources[] = 'crossref';
            if ($joura === '' && trim((string)($summary['joura'] ?? '')) !== '') {
                $joura = trim((string)$summary['joura']);
            }
            $crossrefKeys = $this->authorKeysFromCrossrefMessage($summary['raw'] ?? []);
            if ($author === '') {
                $citationAuthor = $this->crossref->getAuthorsCitation($summary['raw'] ?? [], 3);
                if ($citationAuthor !== '') {
                    $author = $citationAuthor;
                }
                $authorKeys = !empty($crossrefKeys)
                    ? $crossrefKeys
                    : $this->extractAuthorKeysFromAuthorString($author);
            } elseif (!empty($crossrefKeys)) {
                $authorKeys = array_values(array_unique(array_merge($authorKeys, $crossrefKeys)));
            }
        }

        if ($joura === '' || $author === '') {
            $fragParsed = $this->parseStructuredReferText($refer);
            if (is_array($fragParsed)) {
                $sources[] = 'frag';
                if ($joura === '' && trim((string)($fragParsed['joura'] ?? '')) !== '') {
                    $joura = trim((string)$fragParsed['joura']);
                }
                if ($author === '' && trim((string)($fragParsed['author'] ?? '')) !== '') {
                    $author = trim((string)$fragParsed['author']);
                    $authorKeys = $this->extractAuthorKeysFromAuthorString($author);
                }
            }
        }

        $sources = array_values(array_unique($sources));
        $metaSource = empty($sources) ? 'unresolved' : implode('+', $sources);
        $resolved = ($joura !== '' || !empty($authorKeys));

        return [
            'author'               => $author,
            'joura'                => $joura,
            'author_keys'          => array_values(array_unique($authorKeys)),
            'first_author_display' => $this->firstAuthorDisplayFromAuthorString($author),
            'meta_source'          => $metaSource,
            'resolved'             => $resolved,
        ];
    }

    /**
     * 解析 refer_frag / refer_content 中「作者.标题.期刊.年卷页」四段式结构
     *
     * @return array{author:string,joura:string}|null
     */
    private function parseStructuredReferText(array $refer)
    {
        foreach (['refer_frag', 'refer_content'] as $field) {
            $text = trim((string)($refer[$field] ?? ''));
            if ($text === '') {
                continue;
            }
            $text = preg_replace('/\s+Available at:.*$/is', '', $text);
            $text = trim($text, " \t\n\r\0\x0B.");
            if ($text === '' || mb_substr_count($text, '.') !== 3) {
                continue;
            }

            $parts = explode('.', $text);
            if (count($parts) < 4) {
                continue;
            }

            $authorPart = trim((string)$parts[0]);
            $journalPart = trim((string)$parts[2]);
            if ($authorPart === '' || $journalPart === '') {
                continue;
            }

            $bj = bekjournal($journalPart);
            $joura = formateJournal(trim((string)($bj[0] ?? '')));
            $author = trim(prgeAuthor($authorPart), '.');
            if ($joura === '' && $author === '') {
                continue;
            }

            return [
                'author' => $author,
                'joura'  => $joura,
            ];
        }

        return null;
    }

    /**
     * @return string[]
     */
    private function authorKeysFromCrossrefMessage(array $message)
    {
        $keys = [];
        if (empty($message['author']) || !is_array($message['author'])) {
            return $keys;
        }

        foreach ($message['author'] as $author) {
            if (!is_array($author)) {
                continue;
            }
            $family = trim((string)($author['family'] ?? ''));
            $given = trim((string)($author['given'] ?? ''));
            if ($family === '' && $given === '') {
                $org = trim((string)($author['name'] ?? ''));
                if ($org !== '') {
                    $keys[] = $this->authorKeyFromCitationPart($org);
                }
                continue;
            }
            if ($family !== '') {
                $keys[] = mb_strtoupper($family) . '|' . $this->givenToInitials($given);
            }
        }

        return array_values(array_unique(array_filter($keys)));
    }

    private function referSnippet(array $refer)
    {
        foreach (['refer_content', 'refer_frag'] as $field) {
            $text = trim((string)($refer[$field] ?? ''));
            if ($text !== '') {
                $text = preg_replace('/\s+/u', ' ', $text);
                return mb_substr($text, 0, 240);
            }
        }
        $doi = trim((string)($refer['refer_doi'] ?? ''));
        if ($doi !== '') {
            return 'DOI: ' . $doi;
        }
        return '';
    }

    private function resolveArticleId($pArticleId)
    {
        $row = Db::name('production_article')
            ->field('article_id')
            ->where('p_article_id', $pArticleId)
            ->whereIn('state', [0, 2])
            ->find();

        return empty($row['article_id']) ? 0 : intval($row['article_id']);
    }

    /**
     * @return string[]
     */
    private function extractAuthorKeysFromAuthorString($author)
    {
        $author = trim(trim((string)$author), '.');
        if ($author === '') {
            return [];
        }

        $keys = [];
        foreach (preg_split('/,\s*/u', $author) as $part) {
            $part = trim($part);
            if ($part === '' || preg_match('/^et\s+al\.?$/iu', $part)) {
                continue;
            }
            $key = $this->authorKeyFromCitationPart($part);
            if ($key !== '') {
                $keys[] = $key;
            }
        }

        return array_values(array_unique($keys));
    }

    private function firstAuthorDisplayFromAuthorString($author)
    {
        $author = trim(trim((string)$author), '.');
        if ($author === '') {
            return '';
        }
        $parts = preg_split('/,\s*/u', $author);
        $first = trim((string)($parts[0] ?? ''));
        if (preg_match('/^et\s+al\.?$/iu', $first)) {
            return '';
        }
        return $first;
    }

    private function authorKeyFromCitationPart($part)
    {
        $part = trim(preg_replace('/\.+$/u', '', trim((string)$part)));
        if ($part === '') {
            return '';
        }

        $tokens = preg_split('/\s+/u', $part, -1, PREG_SPLIT_NO_EMPTY);
        if (count($tokens) === 1) {
            return mb_strtoupper($tokens[0]) . '|';
        }

        $last = array_pop($tokens);
        if (preg_match('/^[A-Za-z]{1,4}$/u', $last)) {
            $family = implode(' ', $tokens);
            return mb_strtoupper(preg_replace('/\s+/u', ' ', trim($family))) . '|' . mb_strtoupper($last);
        }

        $family = $last;
        $initials = '';
        foreach ($tokens as $token) {
            $initials .= mb_strtoupper(mb_substr($token, 0, 1));
        }
        return mb_strtoupper($family) . '|' . $initials;
    }

    private function normalizeJournalKey($joura)
    {
        $joura = trim(trim((string)$joura), '.');
        if ($joura === '') {
            return '';
        }

        $mapped = formateJournal($joura);
        $key = mb_strtolower($mapped);
        $key = preg_replace('/[^\p{L}\p{N}\s]/u', '', $key);
        $key = preg_replace('/\s+/u', ' ', trim($key));
        return $key;
    }

    private function givenToInitials($given)
    {
        $given = trim((string)$given);
        if ($given === '') {
            return '';
        }
        $parts = preg_split('/[\s\-\.]+/u', $given, -1, PREG_SPLIT_NO_EMPTY);
        $initials = '';
        foreach ($parts as $part) {
            $first = mb_substr($part, 0, 1);
            if ($first !== '') {
                $initials .= mb_strtoupper($first);
            }
        }
        return $initials;
    }

    private function cleanOrcid($orcid)
    {
        $orcid = trim((string)$orcid);
        if ($orcid === '') {
            return '';
        }
        $orcid = preg_replace('#^https?://orcid\.org/#i', '', $orcid);
        return trim($orcid, " \t\n\r\0\x0B/");
    }
}