math_cms/app/Services/OCRProcessingService.php

<?php

namespace App\Services;

use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Facades\Storage;

class OCRProcessingService
{
    protected string $driver; // aliyun 或 baidu
    protected array $aliyunConfig;
    protected array $baiduConfig;

    public function __construct()
    {
        $this->driver = env('OCR_DRIVER', 'aliyun');
        $this->aliyunConfig = [
            'access_key_id' => env('ALIYUN_ACCESS_KEY_ID'),
            'access_key_secret' => env('ALIYUN_ACCESS_KEY_SECRET'),
            'endpoint' => env('ALIYUN_OCR_ENDPOINT', 'ocr-api.cn-hangzhou.aliyuncs.com'),
        ];
        $this->baiduConfig = [
            'app_id' => env('BAIDU_MATH_APP_ID'),
            'app_key' => env('BAIDU_MATH_APP_KEY'),
            'secret_key' => env('BAIDU_MATH_SECRET_KEY'),
        ];
    }

    /**
     * 处理图片OCR识别
     */
    public function processImage(string $imagePath, ?string $paperId = null): array
    {
        $imageUrl = $this->getImageUrl($imagePath);
        Log::info('开始OCR识别', [
            'image_url' => $imageUrl,
            'driver' => $this->driver,
            'paper_id' => $paperId
        ]);

        // 获取标准题目（如果选择了试卷）
        $standardQuestions = [];
        if ($paperId) {
            $questionBankService = app(\App\Services\QuestionBankService::class);
            $standardQuestions = $questionBankService->getPaperQuestions($paperId);
            Log::info('获取标准题目', [
                'paper_id' => $paperId,
                'question_count' => count($standardQuestions)
            ]);
        }

        // 根据配置选择OCR服务
        if ($this->driver === 'aliyun') {
            $results = $this->processWithAliyun($imageUrl, $standardQuestions);
        } else {
            $results = $this->processWithBaidu($imageUrl, $standardQuestions);
        }

        // 如果有多个提供商，可以合并结果
        return $this->formatResults($results);
    }

    /**
     * 阿里云OCR识别
     */
    protected function processWithAliyun(string $imageUrl, array $standardQuestions = []): array
    {
        try {
            Log::info('调用阿里云OCR API', ['standard_questions_count' => count($standardQuestions)]);

            // TODO: 实现阿里云OCR具体调用逻辑
            // 这里需要使用阿里云OCR SDK进行调用
            // 参考：https://help.aliyun.com/document_detail/306402.html

            // 如果有标准题目，可以根据题目数量来验证OCR结果
            $expectedQuestions = count($standardQuestions);
            Log::info('预期题目数量', ['expected' => $expectedQuestions]);

            // 模拟返回数据（实际需要实现）
            $answers = [];
            $questionCount = $expectedQuestions > 0 ? $expectedQuestions : 4; // 默认4题

            for ($i = 1; $i <= $questionCount; $i++) {
                $questionType = 'choice';
                if ($standardQuestions) {
                    // 使用标准题目的类型
                    $stdQuestion = $standardQuestions[$i - 1] ?? null;
                    $questionType = $stdQuestion['question_type'] ?? $stdQuestion['type'] ?? 'choice';
                }

                $answers[] = [
                    'q' => $i,
                    'type' => $questionType,
                    'value' => $this->generateMockAnswer($questionType),
                    'confidence' => 0.90 + ($i * 0.02), // 模拟不同题目的置信度
                ];
            }

            return ['answers' => $answers];
        } catch (\Exception $e) {
            Log::error('阿里云OCR调用失败', ['error' => $e->getMessage()]);
            throw new \Exception('OCR识别失败：' . $e->getMessage());
        }
    }

    /**
     * 百度OCR识别
     */
    protected function processWithBaidu(string $imageUrl, array $standardQuestions = []): array
    {
        try {
            Log::info('调用百度OCR API', ['standard_questions_count' => count($standardQuestions)]);

            // TODO: 实现百度OCR具体调用逻辑
            // 这里需要使用百度OCR API进行调用
            // 参考：https://ai.baidu.com/ai-doc/REFERENCE/Ck3dwjhhu

            // 如果有标准题目，可以根据题目数量来验证OCR结果
            $expectedQuestions = count($standardQuestions);
            Log::info('预期题目数量', ['expected' => $expectedQuestions]);

            // 模拟返回数据（实际需要实现）
            $answers = [];
            $questionCount = $expectedQuestions > 0 ? $expectedQuestions : 4; // 默认4题

            for ($i = 1; $i <= $questionCount; $i++) {
                $questionType = 'choice';
                if ($standardQuestions) {
                    // 使用标准题目的类型
                    $stdQuestion = $standardQuestions[$i - 1] ?? null;
                    $questionType = $stdQuestion['question_type'] ?? $stdQuestion['type'] ?? 'choice';
                }

                $answers[] = [
                    'q' => $i,
                    'type' => $questionType,
                    'value' => $this->generateMockAnswer($questionType),
                    'confidence' => 0.88 + ($i * 0.02), // 模拟不同题目的置信度
                ];
            }

            return ['answers' => $answers];
        } catch (\Exception $e) {
            Log::error('百度OCR调用失败', ['error' => $e->getMessage()]);
            throw new \Exception('OCR识别失败：' . $e->getMessage());
        }
    }

    /**
     * 生成模拟答案
     */
    protected function generateMockAnswer(string $type): string
    {
        return match ($type) {
            'choice' => ['A', 'B', 'C', 'D'][array_rand(['A', 'B', 'C', 'D'])],
            'fill' => (string) rand(1, 100),
            'solve' => '解答步骤和答案',
            default => '未知答案',
        };
    }

    /**
     * 获取图片URL
     */
    protected function getImageUrl(string $imagePath): string
    {
        // 如果是完整URL，直接返回
        if (filter_var($imagePath, FILTER_VALIDATE_URL)) {
            return $imagePath;
        }

        // 如果是相对路径，构建完整URL
        if (strpos($imagePath, 'http') !== 0) {
            return asset('storage/' . $imagePath);
        }

        return $imagePath;
    }

    /**
     * 格式化结果
     */
    protected function formatResults(array $rawResults): array
    {
        $answers = [];

        foreach ($rawResults['answers'] as $result) {
            $answers[] = [
                'q' => $result['q'],
                'type' => $result['type'],
                'value' => is_array($result['value'] ?? $result['steps'] ?? '')
                    ? implode('|', $result['value'] ?? $result['steps'])
                    : ($result['value'] ?? ''),
                'confidence' => $result['confidence'] ?? 0.9,
                'provider' => $this->driver,
            ];
        }

        return ['answers' => $answers];
    }

    /**
     * 提取选择题答案
     */
    protected function extractChoiceAnswers(string $text): array
    {
        $answers = [];
        preg_match_all('/(\d+)[\.\、]\s*[A-D]/', $text, $matches);

        foreach ($matches[1] as $index => $questionNum) {
            $questionNum = intval($questionNum);
            preg_match('/' . $questionNum . '[\.\、]\s*([A-D])/', $text, $answerMatch);
            $answers[$questionNum] = $answerMatch[1] ?? null;
        }

        return $answers;
    }

    /**
     * 提取填空题答案
     */
    protected function extractFillAnswers(string $text): array
    {
        $answers = [];
        preg_match_all('/(\d+)[\.\、]\s*[：:]\s*([^\s\n]+)/', $text, $matches);

        foreach ($matches[1] as $index => $questionNum) {
            $questionNum = intval($questionNum);
            $answers[$questionNum] = $matches[2][$index] ?? '';
        }

        return $answers;
    }

    /**
     * 提取解答题答案
     */
    protected function extractSolveAnswers(string $text): array
    {
        $answers = [];
        // 简化处理，实际需要更复杂的解析逻辑
        preg_match_all('/(\d+)[\.\、]([\s\S]+?)(?=\d+\.|$)/', $text, $matches);

        foreach ($matches[1] as $index => $questionNum) {
            $questionNum = intval($questionNum);
            $answers[$questionNum] = trim($matches[2][$index] ?? '');
        }

        return $answers;
    }
}