szkp-map-service/app/Support/StudyTourDeclarationParser.php

<?php

namespace App\Support;

use App\Models\DictItem;
use App\Models\Venue;
use Illuminate\Support\Collection;

class StudyTourDeclarationParser
{
    private const SECTION_MARKERS = [
        'basic' => '一、线路基本情况',
        'intro' => '二、线路简介',
        'route' => '三、线路规划',
        'courses' => '四、研学课程',
        'fee' => '五、线路收费标准',
        'impl' => '六、线路计划实施情况',
    ];

    /**
     * @return array{parsed: array<string, mixed>, warnings: array<int, string>}
     */
    public static function parseFile(string $path, string $extension): array
    {
        $text = DocTextExtractor::extract($path, $extension);

        return self::parseText($text);
    }

    /**
     * @return array{parsed: array<string, mixed>, warnings: array<int, string>}
     */
    public static function parseText(string $text): array
    {
        $warnings = [];
        $lines = self::splitLines($text);
        $sections = self::splitSections($lines);

        $basicLines = $sections['basic'] ?? [];
        $basic = self::parseBasicSection($basicLines);

        $introText = self::joinSectionLines($sections['intro'] ?? []);
        $feeText = self::joinSectionLines($sections['fee'] ?? []);
        $implText = self::joinSectionLines($sections['impl'] ?? []);

        $routePlans = self::parseRouteSection($sections['route'] ?? []);
        $courses = self::parseCoursesSection($sections['courses'] ?? []);

        $venueResult = self::matchVenueItems((string) ($basic['venue_raw'] ?? ''), $warnings);
        unset($basic['venue_raw']);

        if ($basic['name'] === '') {
            $warnings[] = '未识别到线路名称，请手动填写';
        }
        if ($venueResult['items'] === []) {
            $warnings[] = '未识别到线路点位/场馆，请手动添加';
        }

        $parsed = StudyTourPayload::normalizeIncoming([
            'name' => $basic['name'],
            'org_name' => $basic['org_name'],
            'seasons' => $basic['seasons'],
            'suitable_count' => $basic['suitable_count'],
            'grade_levels' => $basic['grade_levels'],
            'duration' => $basic['duration'],
            'contact_person' => $basic['contact_person'],
            'contact_phones' => $basic['contact_phones'],
            'venue_items' => $venueResult['items'],
            'intro_html' => self::plainTextToHtml($introText),
            'route_plans' => $routePlans,
            'courses' => $courses,
            'fee_html' => self::plainTextToHtml($feeText),
            'implementation_html' => self::plainTextToHtml($implText),
            'tags' => [],
            'cover_image' => '',
            'sort' => 0,
            'is_on_shelf' => true,
        ]);

        return [
            'parsed' => $parsed,
            'warnings' => array_values(array_unique($warnings)),
        ];
    }

    /**
     * @return array<int, string>
     */
    private static function splitLines(string $text): array
    {
        $text = str_replace(["\r\n", "\r", "\f"], "\n", $text);
        $parts = preg_split('/\n/u', $text) ?: [];

        return array_map(fn ($line) => trim((string) $line), $parts);
    }

    /**
     * @param  array<int, string>  $lines
     * @return array<string, array<int, string>>
     */
    private static function splitSections(array $lines): array
    {
        $keys = array_keys(self::SECTION_MARKERS);
        $sections = array_fill_keys($keys, []);
        $current = null;

        foreach ($lines as $line) {
            $matched = null;
            foreach (self::SECTION_MARKERS as $key => $marker) {
                if ($line === $marker || str_starts_with($line, $marker)) {
                    $matched = $key;
                    break;
                }
            }
            if ($matched !== null) {
                $current = $matched;
                continue;
            }
            if ($current !== null) {
                $sections[$current][] = $line;
            }
        }

        return $sections;
    }

    /**
     * @param  array<int, string>  $lines
     * @return array<string, mixed>
     */
    private static function parseBasicSection(array $lines): array
    {
        $fields = [
            'org_name' => '',
            'name' => '',
            'seasons' => [],
            'venue_raw' => '',
            'suitable_count' => '',
            'grade_levels' => [],
            'duration' => '',
            'contact_person' => '',
            'contact_phones' => '',
        ];

        $labels = [
            '组织单位名称' => 'org_name',
            '线路名称' => 'name',
            '线路点位' => 'venue_raw',
            '适宜人数' => 'suitable_count',
            '研学时长' => 'duration',
            '线路联络人' => 'contact_person',
            '咨询电话' => 'contact_phones',
        ];

        $seasonBuffer = [];
        $gradeBuffer = [];

        for ($i = 0; $i < count($lines); $i++) {
            $line = $lines[$i];
            if ($line === '') {
                continue;
            }

            if (str_starts_with($line, '对应季节')) {
                $seasonBuffer[] = $line;
                if (($lines[$i + 1] ?? '') === '（可多选）') {
                    $i++;
                }
                while (($lines[$i + 1] ?? '') !== '' && ! self::isBasicLabelLine($lines[$i + 1])) {
                    $seasonBuffer[] = $lines[++$i];
                }
                $fields['seasons'] = self::parseSeasons(implode(' ', $seasonBuffer));
                continue;
            }

            if (str_starts_with($line, '适配学段')) {
                $gradeBuffer[] = $line;
                if (($lines[$i + 1] ?? '') === '（可多选）') {
                    $i++;
                }
                while (($lines[$i + 1] ?? '') !== '' && ! self::isBasicLabelLine($lines[$i + 1])) {
                    $gradeBuffer[] = $lines[++$i];
                }
                $fields['grade_levels'] = self::parseGrades(implode(' ', $gradeBuffer));
                continue;
            }

            foreach ($labels as $label => $key) {
                if ($line !== $label) {
                    continue;
                }
                $valueLines = [];
                while (($lines[$i + 1] ?? '') !== '' && ! self::isBasicLabelLine($lines[$i + 1]) && ! str_starts_with($lines[$i + 1], '对应季节') && ! str_starts_with($lines[$i + 1], '适配学段')) {
                    $valueLines[] = $lines[++$i];
                }
                $fields[$key] = trim(implode("\n", $valueLines));
                continue 2;
            }
        }

        $fields['suitable_count'] = self::normalizeBlankPlaceholder($fields['suitable_count']);
        $fields['duration'] = self::normalizeDuration($fields['duration']);
        $fields['contact_phones'] = StudyTourPayload::normalizeContactPhones($fields['contact_phones']);
        foreach (['org_name', 'name', 'contact_person', 'venue_raw'] as $key) {
            $fields[$key] = StudyTourPayload::compactText((string) $fields[$key]);
        }

        return $fields;
    }

    private static function isBasicLabelLine(string $line): bool
    {
        if ($line === '（可多选）') {
            return true;
        }

        return array_key_exists($line, [
            '组织单位名称' => true,
            '线路名称' => true,
            '线路点位' => true,
            '适宜人数' => true,
            '研学时长' => true,
            '线路联络人' => true,
            '咨询电话' => true,
        ]);
    }

    /**
     * @return array<int, string>
     */
    private static function parseSeasons(string $raw): array
    {
        $map = [
            '春季' => 'spring',
            '夏季' => 'summer',
            '秋季' => 'autumn',
            '冬季' => 'winter',
        ];
        $selected = [];
        foreach ($map as $label => $value) {
            if (preg_match('/(?:[☑✅✔]|■)\s*'.preg_quote($label, '/').'/u', $raw)) {
                $selected[] = $value;
            }
        }

        return self::filterDictValues('study_tour_season', $selected);
    }

    /**
     * @return array<int, string>
     */
    private static function parseGrades(string $raw): array
    {
        $map = [
            '幼儿园' => 'kindergarten',
            '小学' => 'primary',
            '初中' => 'junior',
            '高中' => 'high',
            '全学段' => 'all',
        ];
        $selected = [];
        foreach ($map as $label => $value) {
            if (preg_match('/(?:[☑✅✔]|■)\s*'.preg_quote($label, '/').'/u', $raw)) {
                $selected[] = $value;
            }
        }

        return self::filterDictValues('study_tour_grade_level', $selected);
    }

    /**
     * @param  array<int, string>  $values
     * @return array<int, string>
     */
    private static function filterDictValues(string $dictType, array $values): array
    {
        $allowed = DictItem::query()
            ->where('dict_type', $dictType)
            ->where('is_active', true)
            ->pluck('item_value')
            ->all();

        return array_values(array_intersect($values, $allowed));
    }

    private static function normalizeBlankPlaceholder(string $raw): string
    {
        $text = StudyTourPayload::compactText($raw);
        $text = preg_replace('/_+/u', '', $text) ?? $text;

        return StudyTourPayload::compactText($text);
    }

    private static function normalizeDuration(string $raw): string
    {
        $text = StudyTourPayload::compactMultilineText($raw);
        if ($text === '') {
            return '';
        }
        $text = preg_replace('/_+/u', '', $text) ?? $text;

        return StudyTourPayload::compactText(str_replace("\n", ' ', $text));
    }

    /**
     * @param  array<int, string>  $lines
     * @return array<int, array{date_label: string, items: array<int, array{time: string, activity: string, location: string}>}>
     */
    private static function parseRouteSection(array $lines): array
    {
        $start = 0;
        foreach ($lines as $idx => $line) {
            if (in_array($line, ['日期', '时间', '行程安排', '地点'], true)) {
                $start = $idx + 1;
            }
        }

        $groups = [];
        $currentIndex = null;
        for ($i = $start; $i < count($lines); $i++) {
            $line = $lines[$i];
            if ($line === '') {
                continue;
            }
            if (self::isRouteDateLabel($line)) {
                $groups[] = [
                    'date_label' => StudyTourPayload::compactText($line),
                    'items' => [],
                ];
                $currentIndex = count($groups) - 1;
                continue;
            }
            if ($currentIndex === null) {
                continue;
            }
            if (! self::isTimeLine($line)) {
                continue;
            }

            $time = $line;
            $activity = '';
            $location = '';

            if ($i + 1 < count($lines) && ($lines[$i + 1] ?? '') !== '' && ! self::isTimeLine($lines[$i + 1]) && ! self::isRouteDateLabel($lines[$i + 1])) {
                $activity = $lines[++$i];
            }

            while ($i + 1 < count($lines) && ($lines[$i + 1] ?? '') === '') {
                $i++;
            }

            if ($i + 1 < count($lines) && ($lines[$i + 1] ?? '') !== '' && ! self::isTimeLine($lines[$i + 1]) && ! self::isRouteDateLabel($lines[$i + 1])) {
                $location = $lines[++$i];
            }

            $groups[$currentIndex]['items'][] = [
                'time' => StudyTourPayload::compactText($time),
                'activity' => StudyTourPayload::compactText($activity),
                'location' => StudyTourPayload::compactText($location),
            ];
        }

        return StudyTourPayload::normalizeRoutePlans($groups);
    }

    /**
     * @param  array<int, string>  $lines
     * @return array<int, array{sort: int, name: string, content: string}>
     */
    private static function parseCoursesSection(array $lines): array
    {
        $start = 0;
        foreach ($lines as $idx => $line) {
            if (in_array($line, ['序号', '课程名称', '课程内容'], true)) {
                $start = $idx + 1;
            }
        }

        $courses = [];
        $sort = 1;
        for ($i = $start; $i < count($lines); $i++) {
            $line = $lines[$i];
            if ($line === '' || ! preg_match('/^\d+$/', $line)) {
                continue;
            }
            $name = StudyTourPayload::compactText((string) ($lines[$i + 1] ?? ''));
            $content = StudyTourPayload::compactText((string) ($lines[$i + 2] ?? ''));
            if ($name === '' && $content === '') {
                $i += 2;
                continue;
            }
            $courses[] = [
                'sort' => $sort++,
                'name' => $name,
                'content' => $content,
            ];
            $i += 2;
        }

        return StudyTourPayload::normalizeCourses($courses);
    }

    /**
     * @return array{items: array<int, array<string, mixed>>, warnings: array<int, string>}
     */
    private static function matchVenueItems(string $raw, array &$warnings): array
    {
        $raw = trim($raw);
        if ($raw === '') {
            return ['items' => [], 'warnings' => []];
        }

        $parts = preg_split('#[+＋、,，/|；;\n]+#u', $raw) ?: [];
        $parts = array_values(array_filter(array_map('trim', $parts), fn ($p) => $p !== ''));
        if ($parts === []) {
            $parts = [$raw];
        }

        /** @var Collection<int, Venue> $venues */
        $venues = Venue::query()->orderBy('sort')->orderBy('id')->get(['id', 'name']);
        $items = [];
        $usedVenueIds = [];

        foreach ($parts as $part) {
            $part = self::cleanVenueToken($part);
            if ($part === '') {
                continue;
            }

            $match = self::findVenueMatch($part, $venues, $usedVenueIds);
            if ($match !== null) {
                $items[] = ['type' => 'system', 'venue_id' => $match->id];
                $usedVenueIds[] = $match->id;
                continue;
            }

            $items[] = ['type' => 'custom', 'name' => $part];
            $warnings[] = "场馆「{$part}」未在系统中匹配，已作为自定义场馆添加";
        }

        return ['items' => $items, 'warnings' => []];
    }

    /**
     * @param  Collection<int, Venue>  $venues
     * @param  array<int, int>  $usedVenueIds
     */
    private static function findVenueMatch(string $token, Collection $venues, array $usedVenueIds): ?Venue
    {
        $tokenNorm = self::normalizeVenueName($token);

        $exact = $venues->first(function (Venue $v) use ($tokenNorm, $usedVenueIds) {
            if (in_array($v->id, $usedVenueIds, true)) {
                return false;
            }

            return self::normalizeVenueName((string) $v->name) === $tokenNorm;
        });
        if ($exact !== null) {
            return $exact;
        }

        $contains = $venues->filter(function (Venue $v) use ($tokenNorm, $usedVenueIds) {
            if (in_array($v->id, $usedVenueIds, true)) {
                return false;
            }
            $nameNorm = self::normalizeVenueName((string) $v->name);

            return $nameNorm !== '' && (str_contains($nameNorm, $tokenNorm) || str_contains($tokenNorm, $nameNorm));
        })->sortByDesc(fn (Venue $v) => mb_strlen((string) $v->name))->first();

        return $contains;
    }

    private static function cleanVenueToken(string $token): string
    {
        $token = StudyTourPayload::compactText($token);
        $token = preg_replace('/等$/u', '', $token) ?? $token;
        $token = preg_replace('/[（(].*[)）]/u', '', $token) ?? $token;

        return StudyTourPayload::compactText($token);
    }

    private static function normalizeVenueName(string $name): string
    {
        $name = mb_strtolower(trim($name));
        $name = str_replace([' ', '　', '·', '•'], '', $name);

        return $name;
    }

    private static function isRouteDateLabel(string $line): bool
    {
        if (self::isTimeLine($line)) {
            return false;
        }

        return (bool) preg_match('/^(线路[一二三四五六七八九十百零\d]+|第[一二三四五六七八九十百零\d]+天|上午|中午|下午|晚上)/u', $line);
    }

    private static function isTimeLine(string $line): bool
    {
        return (bool) preg_match('/^\d{1,2}:\d{2}/', $line);
    }

    /**
     * @param  array<int, string>  $lines
     */
    private static function joinSectionLines(array $lines): string
    {
        $chunks = [];
        $buf = [];
        foreach ($lines as $line) {
            if ($line === '') {
                if ($buf !== []) {
                    $chunks[] = trim(implode("\n", $buf));
                    $buf = [];
                }
                continue;
            }
            $buf[] = $line;
        }
        if ($buf !== []) {
            $chunks[] = trim(implode("\n", $buf));
        }

        return trim(implode("\n\n", array_filter($chunks, fn ($c) => $c !== '')));
    }

    private static function plainTextToHtml(string $text): string
    {
        $text = StudyTourPayload::compactMultilineText($text);
        if ($text === '') {
            return '';
        }

        $paragraphs = preg_split("/\n{2,}/u", $text) ?: [$text];
        $html = [];
        foreach ($paragraphs as $paragraph) {
            $paragraph = StudyTourPayload::compactMultilineText($paragraph);
            if ($paragraph === '') {
                continue;
            }
            $escaped = htmlspecialchars($paragraph, ENT_QUOTES | ENT_SUBSTITUTE, 'UTF-8');
            $escaped = nl2br($escaped, false);
            $html[] = '<p>'.$escaped.'</p>';
        }

        return implode('', $html);
    }
}