math_cms/app/Services/MarkdownQuestionParser.php

<?php

namespace App\Services;

use App\Support\LogContext;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;

class MarkdownQuestionParser
{
    private string $aiDriver;
    private string $deepseekBaseUrl;
    private string $deepseekModel;
    private int $deepseekTimeout;
    private string $openAiBaseUrl;
    private string $openAiModel;
    private int $openAiTimeout;

    public function __construct()
    {
        $this->aiDriver = config('ai.driver', env('AI_DRIVER', 'deepseek'));

        $this->deepseekBaseUrl = rtrim((string) config('ai.deepseek.base_url', 'https://api.deepseek.com/v1'), '/');
        $this->deepseekModel = (string) config('ai.deepseek.model', 'deepseek-chat');
        $this->deepseekTimeout = (int) config('ai.deepseek.timeout', 30);

        $this->openAiBaseUrl = rtrim((string) config('ai.openai.base_url', 'https://api.openai.com/v1'), '/');
        $this->openAiModel = (string) config('ai.openai.model', 'gpt-3.5-turbo');
        $this->openAiTimeout = (int) config('ai.openai.timeout', 30);
    }

    /**
     * 解析 Markdown 文本，返回候选题数组
     */
    public function parse(string $markdown): array
    {
        $splitter = app(AsyncMarkdownSplitter::class);
        $blocks = $splitter->split($markdown);
        if (!$splitter->validate($blocks)) {
            Log::warning('Markdown split validation failed; continue with best-effort parsing', [
                'blocks_count' => count($blocks),
            ]);
        }

        $candidates = [];

        foreach ($blocks as $block) {
            $candidates[] = $this->parseRawMarkdown(
                (string) ($block['raw_markdown'] ?? ''),
                (int) ($block['index'] ?? 0),
            );
        }

        return $candidates;
    }

    /**
     * 解析单题 raw_markdown，返回候选题结构
     */
    public function parseRawMarkdown(string $rawMarkdown, int $index): array
    {
        Log::debug('Parse raw_markdown start', [
            'index' => $index,
            'raw_len' => strlen($rawMarkdown),
            'raw_sha1' => LogContext::sha1($rawMarkdown),
            'raw_excerpt' => LogContext::excerpt($rawMarkdown),
        ]);

        $candidate = $this->parseBlock($rawMarkdown, $index);

        // AI 结构化解析（失败则回退为启发式提取 + AI 判题）
        $aiStructured = $this->parseWithAi($candidate['raw_markdown'], $candidate['index']);
        if ($aiStructured !== null) {
            Log::debug('Parse raw_markdown done (ai_structured)', [
                'index' => $index,
                'keys' => array_keys($aiStructured),
                'is_question_candidate' => $aiStructured['is_question_candidate'] ?? null,
                'ai_confidence' => $aiStructured['ai_confidence'] ?? null,
                'options_count' => is_array($aiStructured['options'] ?? null) ? count($aiStructured['options']) : 0,
                'images_count' => is_array($aiStructured['images'] ?? null) ? count($aiStructured['images']) : 0,
                'tables_count' => is_array($aiStructured['tables'] ?? null) ? count($aiStructured['tables']) : 0,
            ]);
            return array_merge($candidate, $aiStructured);
        }

        $this->enhanceWithAi($candidate);

        Log::debug('Parse raw_markdown done (heuristic+detect)', [
            'index' => $index,
            'is_question_candidate' => $candidate['is_question_candidate'] ?? null,
            'ai_confidence' => $candidate['ai_confidence'] ?? null,
        ]);

        return $candidate;
    }

    /**
     * 解析单个题目块
     */
    private function parseBlock(string $block, int $index): array
    {
        $candidate = [
            'index' => $index,
            'raw_markdown' => $block,
            'stem' => null,
            'options' => null,
            'images' => [],
            'tables' => [],
            'is_question_candidate' => false,
            'ai_confidence' => null,
        ];

        // ② Stem 提取
        $candidate['stem'] = $this->extractStem($block);

        // ③ 选项识别
        $candidate['options'] = $this->extractOptions($block);

        // ④ 图片识别
        $candidate['images'] = $this->extractImages($block);

        // ⑤ 表格识别
        $candidate['tables'] = $this->extractTables($block);

        return $candidate;
    }

    /**
     * AI 结构化解析：返回符合候选库字段的结构化数组，失败返回 null
     *
     * @return array{
     *   index:int,
     *   stem:?string,
     *   options:?array,
     *   images:array,
     *   tables:array,
     *   is_question_candidate:bool,
     *   ai_confidence:?float
     * }|null
     */
    private function parseWithAi(string $rawMarkdown, int $index): ?array
    {
        $template = (string) config('ai.question_parse_prompt');
        if (trim($template) === '') {
            return null;
        }

        $prompt = str_replace(['{index}', '{content}'], [(string) $index, $rawMarkdown], $template);

        try {
            Log::debug('AI structured parse request', [
                'driver' => $this->aiDriver,
                'index' => $index,
                'prompt_len' => strlen($prompt),
                'raw_sha1' => LogContext::sha1($rawMarkdown),
            ]);

            $result = $this->callAiApi($prompt);

            $normalized = [
                'index' => (int) ($result['index'] ?? $index),
                'stem' => isset($result['stem']) ? (string) $result['stem'] : null,
                'options' => isset($result['options']) && is_array($result['options']) ? $result['options'] : null,
                'images' => isset($result['images']) && is_array($result['images']) ? $result['images'] : [],
                'tables' => isset($result['tables']) && is_array($result['tables']) ? $result['tables'] : [],
                'is_question_candidate' => (bool) ($result['is_question_candidate'] ?? $result['is_question'] ?? false),
                'ai_confidence' => isset($result['ai_confidence']) ? (float) $result['ai_confidence'] : (isset($result['confidence']) ? (float) $result['confidence'] : null),
                'answer' => $result['answer'] ?? null,
                'solution' => $result['solution'] ?? null,
                'solution_steps' => $result['solution_steps'] ?? $result['steps'] ?? [],
            ];

            // 如果是简答题（没有选项）且没有分步解析，尝试使用专门的 Prompt 补全
            if (empty($normalized['options']) && empty($normalized['solution_steps']) && $normalized['is_question_candidate']) {
                $stepResult = $this->refineSolutionSteps($rawMarkdown);
                if ($stepResult) {
                    $normalized['solution'] = $stepResult['solution'] ?? $normalized['solution'];
                    $normalized['solution_steps'] = $stepResult['steps'] ?? [];
                }
            }

            Log::debug('AI structured parse response', [
                'driver' => $this->aiDriver,
                'index' => $index,
                'response_keys' => array_keys($result),
                'normalized' => [
                    'index' => $normalized['index'],
                    'is_question_candidate' => $normalized['is_question_candidate'],
                    'ai_confidence' => $normalized['ai_confidence'],
                    'options_count' => is_array($normalized['options']) ? count($normalized['options']) : 0,
                    'images_count' => is_array($normalized['images']) ? count($normalized['images']) : 0,
                    'tables_count' => is_array($normalized['tables']) ? count($normalized['tables']) : 0,
                ],
            ]);

            return $normalized;
        } catch (\Throwable $e) {
            Log::warning('AI structured parse failed, fallback to heuristic', [
                'index' => $index,
                'error' => $e->getMessage(),
                'raw_sha1' => LogContext::sha1($rawMarkdown),
            ]);
            return null;
        }
    }

    /**
     * 提取题目主干
     */
    private function extractStem(string $block): ?string
    {
        $lines = explode("\n", $block);
        $stemLines = [];

        foreach ($lines as $line) {
            $line = trim($line);

            // 跳过选项行
            if (preg_match('/^[A-D]\.\s+/', $line)) {
                break;
            }

            // 跳过空行和图片行
            if (empty($line) || preg_match('/^<img/', $line)) {
                continue;
            }

            $stemLines[] = $line;
        }

        return empty($stemLines) ? null : implode("\n", $stemLines);
    }

    /**
     * 提取选项
     */
    private function extractOptions(string $block): ?array
    {
        $options = [];

        preg_match_all('/^([A-D])\.\s+(.+)$/m', $block, $matches, PREG_SET_ORDER);

        foreach ($matches as $match) {
            $label = $match[1];
            $content = trim($match[2]);
            $options[$label] = $content;
        }

        return empty($options) ? null : $options;
    }

    /**
     * 提取图片
     */
    private function extractImages(string $block): array
    {
        $images = [];

        preg_match_all('/<img[^>]+src=["\']([^"\']+)["\'][^>]*>/i', $block, $matches);

        foreach ($matches[1] as $src) {
            $images[] = $src;
        }

        return $images;
    }

    /**
     * 提取表格
     */
    private function extractTables(string $block): array
    {
        $tables = [];

        // 简单匹配 HTML 表格标签
        preg_match_all('/<table[^>]*>.*?<\/table>/s', $block, $matches);

        foreach ($matches[0] as $table) {
            $tables[] = $table;
        }

        return $tables;
    }

    /**
     * AI 增强：判断是否为题目
     */
    private function enhanceWithAi(array &$candidate): void
    {
        $prompt = $this->buildQuestionDetectionPrompt($candidate['raw_markdown']);

        try {
            $result = $this->callAiApi($prompt);

            if (isset($result['is_question'])) {
                $candidate['is_question_candidate'] = $result['is_question'];
                $candidate['ai_confidence'] = $result['confidence'] ?? null;
            }
        } catch (\Exception $e) {
            Log::error('AI question detection failed', [
                'error' => $e->getMessage(),
                'block' => substr($candidate['raw_markdown'], 0, 200),
            ]);

            // 默认值：不是题目
            $candidate['is_question_candidate'] = false;
            $candidate['ai_confidence'] = 0.0;
        }
    }

    /**
     * 构建题目检测 Prompt
     */
    private function buildQuestionDetectionPrompt(string $rawMarkdown): string
    {
        $template = (string) config('ai.question_detection_prompt');
        if (trim($template) === '') {
            $template = "请判断下面这段 Markdown 是否是一道数学题目。\n\n题目内容：\n{content}\n\n请输出 JSON：{\"is_question\":true|false,\"confidence\":0~1}";
        }

        return str_replace('{content}', $rawMarkdown, $template);
    }

    /**
     * 调用 AI API
     */
    private function callAiApi(string $prompt): array
    {
        if ($this->aiDriver === 'deepseek') {
            return $this->callDeepSeek($prompt);
        } elseif ($this->aiDriver === 'openai') {
            return $this->callOpenAI($prompt);
        }

        throw new \Exception("Unsupported AI driver: {$this->aiDriver}");
    }

    /**
     * DeepSeek API 调用
     */
    private function callDeepSeek(string $prompt): array
    {
        $apiKey = config('ai.deepseek.api_key', env('DEEPSEEK_API_KEY'));

        $response = Http::withHeaders([
            'Authorization' => "Bearer {$apiKey}",
            'Content-Type' => 'application/json',
        ])->timeout($this->deepseekTimeout)->post($this->deepseekBaseUrl . '/chat/completions', [
            'model' => $this->deepseekModel,
            'messages' => [
                ['role' => 'user', 'content' => $prompt]
            ],
            'temperature' => 0.1,
        ]);

        if (!$response->successful()) {
            throw new \Exception('DeepSeek API error: ' . $response->body());
        }

        $content = $response->json('choices.0.message.content');

        return $this->parseJsonResponse($content);
    }

    /**
     * OpenAI API 调用
     */
    private function callOpenAI(string $prompt): array
    {
        $apiKey = config('ai.openai.api_key', env('OPENAI_API_KEY'));

        $response = Http::withHeaders([
            'Authorization' => "Bearer {$apiKey}",
            'Content-Type' => 'application/json',
        ])->timeout($this->openAiTimeout)->post($this->openAiBaseUrl . '/chat/completions', [
            'model' => $this->openAiModel,
            'messages' => [
                ['role' => 'user', 'content' => $prompt]
            ],
            'temperature' => 0.1,
        ]);

        if (!$response->successful()) {
            throw new \Exception('OpenAI API error: ' . $response->body());
        }

        $content = $response->json('choices.0.message.content');

        return $this->parseJsonResponse($content);
    }

    /**
     * 精修简答题的解题步骤
     */
    private function refineSolutionSteps(string $rawMarkdown): ?array
    {
        $prompt = app(QuestionPromptService::class)->buildSolutionStepsPrompt($rawMarkdown);

        try {
            return $this->callAiApi($prompt);
        } catch (\Throwable $e) {
            Log::warning('Refine solution steps failed', [
                'error' => $e->getMessage(),
            ]);
            return null;
        }
    }

    /**
     * 解析 AI 返回的 JSON
     */
    private function parseJsonResponse(string $content): array
    {
        // 提取 JSON 部分
        preg_match('/\{.*\}/s', $content, $matches);

        if (empty($matches[0])) {
            throw new \Exception('No JSON found in response');
        }

        $json = json_decode($matches[0], true);

        if (json_last_error() !== JSON_ERROR_NONE) {
            throw new \Exception('Invalid JSON: ' . json_last_error_msg());
        }

        return $json;
    }
}