huahuibao/biz-ghs/lakala/classes/AliyunOcrClass.php

<?php

namespace bizGhs\lakala\classes;

use AlibabaCloud\Client\AlibabaCloud;
use AlibabaCloud\Client\Exception\ClientException;
use AlibabaCloud\Client\Exception\ServerException;
use common\components\util;
use Yii;

class AliyunOcrClass
{
    public static function recognize($id, $mainId, $shopId, $imgType, $imageUrl)
    {
        $application = LakalaApplicationClass::getByCondition([
            'id' => $id,
            'mainId' => $mainId,
            'shopId' => $shopId,
        ]);
        if (empty($application)) {
            util::fail('没有找到申请记录');
        }

        try {
            $response = self::request($imgType, self::formatImageUrl($imageUrl));
        } catch (ClientException $e) {
            Yii::warning('阿里云OCR客户端异常：' . $e->getErrorMessage(), __METHOD__);
            util::fail('OCR识别失败，请稍后重试');
        } catch (ServerException $e) {
            Yii::warning('阿里云OCR服务端异常：' . $e->getErrorMessage(), __METHOD__);
            util::fail('OCR识别失败，请稍后重试');
        } catch (\Exception $e) {
            Yii::warning('阿里云OCR异常：' . $e->getMessage(), __METHOD__);
            util::fail('OCR识别失败，请稍后重试');
        }

        return [
            'status' => '00',
            'result' => self::formatResult($imgType, $response),
        ];
    }

    private static function request($imgType, $imageUrl)
    {
        $accessKeyId = getenv('OCR_ACCESS_KEY_ID');
        $accessKeySecret = getenv('OCR_ACCESS_KEY_SECRET');
        AlibabaCloud::accessKeyClient(
            $accessKeyId,
            $accessKeySecret
        )->regionId('cn-shanghai')->asDefaultClient();

        $action = 'RecognizeCharacter';
        $query = [
            'RegionId' => 'cn-shanghai',
            'ImageURL' => $imageUrl,
        ];
        if ($imgType === 'ID_CARD_FRONT') {
            $action = 'RecognizeIdentityCard';
            $query['Side'] = 'face';
        } elseif ($imgType === 'ID_CARD_BEHIND') {
            $action = 'RecognizeIdentityCard';
            $query['Side'] = 'back';
        } elseif ($imgType === 'BANK_CARD') {
            $action = 'RecognizeBankCard';
        }

        return AlibabaCloud::rpc()
            ->product('ocr')
            ->version('2019-12-30')
            ->action($action)
            ->method('POST')
            ->options([
                'query' => $query,
            ])
            ->request()
            ->toArray();
    }

    private static function formatImageUrl($imageUrl)
    {
        $imageUrl = trim((string)$imageUrl);
        $host = parse_url($imageUrl, PHP_URL_HOST);
        if ($host && strpos($host, '.aliyuncs.com') !== false) {
            return $imageUrl;
        }

        $objectPath = $imageUrl;
        if (strpos($imageUrl, 'http://') === 0 || strpos($imageUrl, 'https://') === 0) {
            $objectPath = parse_url($imageUrl, PHP_URL_PATH);
        }
        $objectPath = ltrim((string)$objectPath, '/');
        if ($objectPath === '') {
            util::fail('图片地址无效');
        }

        return sprintf(
            'https://%s.%s/%s',
            Yii::$app->params['ossBucket'],
            Yii::$app->params['endpoint'],
            $objectPath
        );
    }

    private static function formatResult($imgType, array $response)
    {
        $data = self::decodeData($response);
        $lines = self::extractLines($data);
        $text = implode("\n", $lines);
        $result = [
            'text' => $text,
            'lines' => $lines,
        ];

        if ($imgType === 'ID_CARD_FRONT') {
            $result = array_merge($result, self::parseIdCardFront($text, $data));
        } elseif ($imgType === 'ID_CARD_BEHIND') {
            $result = array_merge($result, self::parseIdCardBehind($text, $data));
        } elseif ($imgType === 'BANK_CARD') {
            $result = array_merge($result, self::parseBankCard($text, $lines, $data));
        }

        return $result;
    }

    private static function decodeData(array $response)
    {
        $data = isset($response['Data']) ? $response['Data'] : $response;
        if (is_string($data)) {
            $decoded = json_decode($data, true);
            return is_array($decoded) ? $decoded : ['text' => $data];
        }
        return $data;
    }

    private static function extractLines($data)
    {
        $lines = [];
        self::collectLines($data, $lines);
        $lines = array_values(array_unique(array_filter(array_map('trim', $lines))));
        return $lines;
    }

    private static function collectLines($data, array &$lines)
    {
        if (is_string($data)) {
            $lines[] = $data;
            return;
        }
        if (!is_array($data)) {
            return;
        }

        foreach (['word', 'words', 'content', 'text', 'value', 'textContent'] as $key) {
            if (isset($data[$key]) && is_string($data[$key])) {
                $lines[] = $data[$key];
            }
        }
        foreach ($data as $value) {
            self::collectLines($value, $lines);
        }
    }

    private static function parseIdCardFront($text, array $data)
    {
        $result = [];
        $name = self::findValue($data, ['Name', 'name', '姓名']);
        if ($name !== '') {
            $result['name'] = $name;
        }
        $idNumber = self::findValue($data, ['IDNumber', 'IdNumber', 'idNumber', 'CardNumber', '公民身份号码', '身份证号']);
        if ($idNumber !== '') {
            $result['id_number'] = strtoupper($idNumber);
        }
        if (preg_match('/姓名\s*([\x{4e00}-\x{9fa5}·]{2,20})/u', $text, $match)) {
            $result['name'] = $result['name'] ?? $match[1];
        }
        if (preg_match('/\d{17}[\dXx]/', $text, $match)) {
            $result['id_number'] = $result['id_number'] ?? strtoupper($match[0]);
        }
        return $result;
    }

    private static function parseIdCardBehind($text, array $data)
    {
        $result = [];
        $startDate = self::findValue($data, ['StartDate', 'startDate', 'ValidStartDate', '有效期限起始日期']);
        $endDate = self::findValue($data, ['EndDate', 'endDate', 'ValidEndDate', '有效期限结束日期']);
        if ($startDate !== '' && $endDate !== '') {
            $result['validity'] = self::normalizeDate($startDate) . '-' . self::normalizeDate($endDate);
        }
        if (preg_match_all('/\d{4}[.年-]\d{1,2}[.月-]\d{1,2}日?/', $text, $matches) && count($matches[0]) >= 2) {
            $result['validity'] = $result['validity'] ?? (self::normalizeDate($matches[0][0]) . '-' . self::normalizeDate($matches[0][1]));
        } elseif (strpos($text, '长期') !== false && preg_match('/\d{4}[.年-]\d{1,2}[.月-]\d{1,2}日?/', $text, $match)) {
            $result['validity'] = $result['validity'] ?? (self::normalizeDate($match[0]) . '-9999-12-31');
        }
        return $result;
    }

    private static function parseBankCard($text, array $lines, array $data)
    {
        $result = [];
        $cardNumber = self::findValue($data, ['CardNumber', 'cardNumber', 'BankCardNumber', 'bankCardNumber', '银行卡号']);
        if ($cardNumber !== '') {
            $result['card_number'] = preg_replace('/\s+/', '', $cardNumber);
            $result['bank_card_number'] = $result['card_number'];
        }
        $bankName = self::findValue($data, ['BankName', 'bankName', '银行名称', '银行']);
        if ($bankName !== '') {
            $result['bank_name'] = $bankName;
        }
        $numberText = preg_replace('/\s+/', '', $text);
        if (preg_match('/\d{12,30}/', $numberText, $match)) {
            $result['card_number'] = $result['card_number'] ?? $match[0];
            $result['bank_card_number'] = $result['bank_card_number'] ?? $match[0];
        }
        foreach ($lines as $line) {
            if (strpos($line, '银行') !== false) {
                $result['bank_name'] = $result['bank_name'] ?? $line;
                break;
            }
        }
        return $result;
    }

    private static function normalizeDate($date)
    {
        if (preg_match('/(\d{4})[.年-](\d{1,2})[.月-](\d{1,2})日?/', $date, $match)) {
            return sprintf('%04d-%02d-%02d', $match[1], $match[2], $match[3]);
        }
        return $date;
    }

    private static function findValue(array $data, array $keys)
    {
        foreach ($data as $key => $value) {
            if (in_array($key, $keys, true) && (is_string($value) || is_numeric($value))) {
                return trim((string)$value);
            }
            if (is_array($value)) {
                $found = self::findValue($value, $keys);
                if ($found !== '') {
                    return $found;
                }
            }
        }
        return '';
    }
}