process_sheet/fastadmin/application/common/library/ps/LlmClient.php

<?php

namespace app\common\library\ps;

use think\Config;
use think\Exception;

/**
 * 大模型客户端。本地 Ollama 做简单任务，线上 OpenAI 兼容接口做拆版。
 */
class LlmClient
{
    protected $profile;
    protected $driver;
    protected $baseUrl;
    protected $apiKey;
    protected $timeout;
    protected $textModel;
    protected $visionModel;
    protected $visionModelComplex;
    protected $temperature;
    protected $onlineEnable;
    protected $forceEnable;

    /**
     * @param array $config profile=local|online，可覆盖 base_url / api_key / 模型名
     */
    public function __construct(array $config = [])
    {
        $file = Config::get('llm') ?: [];
        $site = Config::get('site') ?: [];
        $this->profile = ($config['profile'] ?? 'local') === 'online' ? 'online' : 'local';
        $this->forceEnable = !empty($config['force']);
        $this->temperature = (float)($config['temperature'] ?? $file['temperature'] ?? 0.1);

        if ($this->profile === 'online') {
            $this->driver = 'openai';
            $this->onlineEnable = self::isOn($config['online_enable'] ?? $site['llm_online_enable'] ?? $file['online_enable'] ?? '0');
            $this->baseUrl = rtrim((string)($config['base_url'] ?? self::siteOr($site, 'llm_online_base_url', $file['online_base_url'] ?? '')), '/');
            $this->apiKey = (string)($config['api_key'] ?? self::siteOr($site, 'llm_online_api_key', $file['online_api_key'] ?? ''));
            $this->timeout = (int)($config['timeout'] ?? self::siteOr($site, 'llm_online_timeout', $file['online_timeout'] ?? 180));
            $this->textModel = (string)($config['text_model'] ?? self::siteOr($site, 'llm_online_model', $file['online_model'] ?? 'qwen-plus'));
            $this->visionModel = (string)($config['vision_model'] ?? self::siteOr($site, 'llm_online_vision_model', $file['online_vision_model'] ?? 'qwen-vl-max'));
            $this->visionModelComplex = $this->visionModel;
            if ($this->timeout < 30) {
                $this->timeout = 30;
            }
            return;
        }

        $this->driver = 'ollama';
        $this->onlineEnable = false;
        $this->baseUrl = rtrim((string)($config['base_url'] ?? self::siteOr($site, 'llm_base_url', $file['base_url'] ?? 'http://127.0.0.1:11434')), '/');
        $this->apiKey = (string)($config['api_key'] ?? self::siteOr($site, 'llm_api_key', $file['api_key'] ?? ''));
        $this->timeout = (int)($config['timeout'] ?? self::siteOr($site, 'llm_timeout', $file['timeout'] ?? 300));
        $this->textModel = (string)($config['text_model'] ?? self::siteOr($site, 'llm_text_model', $file['text_model'] ?? 'qwen2.5:7b-instruct'));
        $this->visionModel = (string)($config['vision_model'] ?? self::siteOr($site, 'llm_vision_model', $file['vision_model'] ?? 'minicpm-v4.6'));
        $this->visionModelComplex = (string)($config['vision_model_complex'] ?? self::siteOr($site, 'llm_vision_complex', $file['vision_model_complex'] ?? 'llama3.2-vision:11b'));
    }

    /**
     * 本地 Ollama，简单文本 / 备用识版。
     * @return self
     */
    public static function local()
    {
        return new self(['profile' => 'local']);
    }

    /**
     * 线上 OpenAI 兼容接口。
     * @return self
     */
    public static function online()
    {
        return new self(['profile' => 'online']);
    }

    /**
     * 拆版：线上已启用则走线上，否则本地。
     * @return self
     */
    public static function forRecognize()
    {
        $online = self::online();
        return $online->enabled() ? $online : self::local();
    }

    /**
     * @return bool
     */
    public function enabled()
    {
        if ($this->profile === 'online') {
            if (!$this->forceEnable && !$this->onlineEnable) {
                return false;
            }
            return $this->baseUrl !== '' && $this->apiKey !== '' && $this->visionModel !== '';
        }
        return $this->baseUrl !== '';
    }

    /**
     * @return string local|online
     */
    public function profile()
    {
        return $this->profile;
    }

    /**
     * @param array  $site
     * @param string $key
     * @param mixed  $default
     * @return mixed
     */
    protected static function siteOr(array $site, $key, $default)
    {
        if (isset($site[$key]) && $site[$key] !== '' && $site[$key] !== null) {
            return $site[$key];
        }
        return $default;
    }

    /**
     * @param mixed $v
     * @return bool
     */
    protected static function isOn($v)
    {
        return $v === 1 || $v === '1' || $v === true || $v === 'true' || $v === 'on';
    }

    /**
     * 列出本地已安装模型（Ollama /api/tags）
     * @return array
     */
    public function listModels()
    {
        if ($this->driver !== 'ollama') {
            return $this->listOnlineModels();
        }
        $result = $this->httpJson('GET', $this->baseUrl . '/api/tags', null, (int)Config::get('llm.list_timeout') ?: 8);
        $names = [];
        foreach ($result['models'] ?? [] as $item) {
            if (!empty($item['name'])) {
                $names[] = $item['name'];
            }
        }
        return $names;
    }

    /**
     * @return array
     */
    protected function listOnlineModels()
    {
        if ($this->baseUrl === '' || $this->apiKey === '') {
            return [];
        }
        try {
            $result = $this->httpJson('GET', $this->openaiRoot() . '/models', null, 15, true);
        } catch (\Exception $e) {
            return [];
        }
        $names = [];
        foreach ($result['data'] ?? [] as $item) {
            if (!empty($item['id'])) {
                $names[] = $item['id'];
            }
        }
        return $names;
    }

    /**
     * @return string
     */
    public function textModel()
    {
        return $this->textModel;
    }

    /**
     * @return string
     */
    public function visionModel()
    {
        return $this->visionModel;
    }

    /**
     * @return string
     */
    public function visionModelComplex()
    {
        return $this->visionModelComplex;
    }

    /**
     * 指定单个视觉模型。CSV 识版不要开 JSON 模式，弱模型会去抄示例 JSON。
     * @param string $prompt
     * @param string $imagePath
     * @param string $model
     * @param bool   $jsonMode
     * @return string
     */
    public function visionRaw($prompt, $imagePath, $model, $jsonMode = false)
    {
        $content = $this->requestChat($model, $prompt, [$this->encodeImage($imagePath)], $jsonMode);
        if (trim((string)$content) === '' && $jsonMode) {
            $content = $this->requestChat($model, $prompt, [$this->encodeImage($imagePath)], false);
        }
        if (is_array($content)) {
            $content = json_encode($content, JSON_UNESCAPED_UNICODE);
        }
        return trim((string)$content);
    }

    /**
     * 按文件类型给出默认模型：图片用视觉，PDF 用文本。
     * @param string $fileType image|pdf
     * @return string
     */
    public function defaultModel($fileType = 'image')
    {
        return $fileType === 'pdf' ? $this->textModel : $this->visionModel;
    }

    /**
     * 后台下拉：线上识版优先，本地模型保留作简单任务。
     * @return array name => label
     */
    public function formOptions()
    {
        $options = [];
        $online = $this->profile === 'online' ? $this : self::online();
        $local = $this->profile === 'local' ? $this : self::local();
        if ($online->enabled()) {
            if ($online->visionModel) {
                $options[$online->visionModel] = $online->visionModel . '（线上识版，推荐）';
            }
            if ($online->textModel && !isset($options[$online->textModel])) {
                $options[$online->textModel] = $online->textModel . '（线上文本）';
            }
        }
        if ($local->visionModel) {
            $key = $local->visionModel;
            if (!isset($options[$key])) {
                $options[$key] = $key . '（本地视觉，简单/备用）';
            }
        }
        if ($local->visionModelComplex && !isset($options[$local->visionModelComplex])) {
            $options[$local->visionModelComplex] = $local->visionModelComplex . '（本地视觉备用）';
        }
        if ($local->textModel && !isset($options[$local->textModel])) {
            $options[$local->textModel] = $local->textModel . '（本地文本，简单任务）';
        }
        try {
            foreach ($local->listModels() as $name) {
                if ($name && !isset($options[$name])) {
                    $options[$name] = $name . '（本地）';
                }
            }
        } catch (\Exception $e) {
            // Ollama 未启动时仍展示已配置项
        }
        return $options;
    }

    /**
     * 是否为本地 Ollama 模型名（含冒号标签）。
     * @param string $name
     * @return bool
     */
    public static function isLocalModelName($name)
    {
        $name = trim((string)$name);
        if ($name === '') {
            return false;
        }
        $local = self::local();
        $known = [$local->textModel(), $local->visionModel(), $local->visionModelComplex()];
        if (in_array($name, $known, true)) {
            return true;
        }
        return strpos($name, ':') !== false;
    }

    /**
     * 文本对话
     * @param string      $prompt
     * @param string|null $model
     * @return string
     */
    public function chat($prompt, $model = null, $jsonMode = false)
    {
        return $this->requestChat($model ?: $this->textModel, $prompt, [], $jsonMode);
    }

    /**
     * 视觉识版。优先用指定模型，失败时回退复杂视觉模型。
     * @param string      $prompt
     * @param string      $imagePath 本地文件路径
     * @param string|null $preferredModel
     * @return array {content, model}
     */
    public function vision($prompt, $imagePath, $preferredModel = null)
    {
        $models = [];
        if ($preferredModel) {
            $models[] = $preferredModel;
        }
        if ($this->visionModel && !in_array($this->visionModel, $models, true)) {
            $models[] = $this->visionModel;
        }
        if ($this->visionModelComplex && !in_array($this->visionModelComplex, $models, true)) {
            $models[] = $this->visionModelComplex;
        }
        if (!$models) {
            $models[] = $this->visionModel;
        }
        $lastError = null;
        foreach ($models as $model) {
            try {
                $content = $this->requestChat($model, $prompt, [$this->encodeImage($imagePath)], true);
                if (trim((string)$content) === '') {
                    $content = $this->requestChat($model, $prompt, [$this->encodeImage($imagePath)], false);
                }
                if (trim((string)$content) !== '') {
                    return ['content' => $content, 'model' => $model];
                }
            } catch (\Exception $e) {
                $lastError = $e;
                if (!$this->isModelLoadError($e)) {
                    throw $e;
                }
            }
        }
        throw $lastError ?: new Exception('视觉模型没有返回内容，请再试一次');
    }

    /**
     * 从模型输出中抽取 JSON
     * @param string $text
     * @return array
     */
    public static function extractJson($text)
    {
        $text = trim((string)$text);
        $text = preg_replace('/<think>[\s\S]*?<\/think>/i', '', $text);
        $text = preg_replace('/^```(?:json)?\s*/i', '', $text);
        $text = preg_replace('/\s*```$/', '', $text);
        $text = trim($text);
        if ($text === '') {
            throw new Exception('识版模型没有返回内容，请再点一次「重新 AI 拆版」');
        }
        $candidates = [$text];
        if (preg_match('/\{[\s\S]*\}/', $text, $m)) {
            $candidates[] = $m[0];
        }
        foreach ($candidates as $candidate) {
            $decoded = self::decodeLooseJson($candidate);
            if (is_array($decoded)) {
                return $decoded;
            }
        }
        $salvaged = self::salvageJson($text);
        if (is_array($salvaged)) {
            return $salvaged;
        }
        throw new Exception('识版结果不完整，请再点一次「重新 AI 拆版」');
    }

    /**
     * @param string $text
     * @return array|null
     */
    protected static function decodeLooseJson($text)
    {
        $text = trim((string)$text);
        if ($text === '') {
            return null;
        }
        $decoded = json_decode($text, true);
        if (is_array($decoded)) {
            return $decoded;
        }
        $text = preg_replace('/,\s*([}\]])/', '$1', $text);
        $text = preg_replace('/"([A-Za-z_][A-Za-z0-9_]*)"\s*([}\]])/', '"$1":0$2', $text);
        $text = preg_replace('/":\s*([}\],])/', ':0$1', $text);
        $decoded = json_decode($text, true);
        return is_array($decoded) ? $decoded : null;
    }

    /**
     * 截断的 JSON 尽量补全到最后一个完整对象。
     * @param string $text
     * @return array|null
     */
    public static function salvageJson($text)
    {
        $start = strpos($text, '{');
        if ($start === false) {
            $start = strpos($text, '[');
        }
        if ($start === false) {
            return null;
        }
        $json = substr($text, $start);
        $stack = [];
        $inStr = false;
        $esc = false;
        $lastComplete = -1;
        $stackAtComplete = [];
        $len = strlen($json);
        for ($i = 0; $i < $len; $i++) {
            $ch = $json[$i];
            if ($inStr) {
                if ($esc) {
                    $esc = false;
                    continue;
                }
                if ($ch === '\\') {
                    $esc = true;
                    continue;
                }
                if ($ch === '"') {
                    $inStr = false;
                }
                continue;
            }
            if ($ch === '"') {
                $inStr = true;
                continue;
            }
            if ($ch === '{' || $ch === '[') {
                $stack[] = $ch;
                continue;
            }
            if ($ch === '}' || $ch === ']') {
                if ($stack) {
                    array_pop($stack);
                }
                $lastComplete = $i;
                $stackAtComplete = $stack;
                if (!$stack) {
                    $decoded = json_decode(substr($json, 0, $i + 1), true);
                    if (is_array($decoded)) {
                        return $decoded;
                    }
                }
            }
        }
        if ($lastComplete < 0) {
            return null;
        }
        $slice = rtrim(substr($json, 0, $lastComplete + 1));
        $slice = preg_replace('/,\s*$/', '', $slice);
        $close = array_reverse($stackAtComplete);
        foreach ($close as $open) {
            $slice .= $open === '{' ? '}' : ']';
        }
        $slice = preg_replace('/,\s*([}\]])/', '$1', $slice);
        $decoded = self::decodeLooseJson($slice);
        return is_array($decoded) ? $decoded : null;
    }

    /**
     * @param string $model
     * @param string $prompt
     * @param array  $images base64 列表
     * @return string
     */
    protected function requestChat($model, $prompt, array $images, $jsonMode = false)
    {
        if ($this->driver === 'openai') {
            return $this->openaiChat($model, $prompt, $images, $jsonMode);
        }
        return $this->ollamaChat($model, $prompt, $images, $jsonMode);
    }

    protected function ollamaChat($model, $prompt, array $images, $jsonMode = false)
    {
        $message = [
            'role'    => 'user',
            'content' => $prompt,
        ];
        if ($images) {
            $message['images'] = $images;
        }
        $payload = [
            'model'    => $model,
            'messages' => [$message],
            'stream'   => false,
            'options'  => [
                'temperature' => $this->temperature,
                'num_predict' => 8192,
            ],
        ];
        if ($jsonMode) {
            $payload['format'] = 'json';
        }
        try {
            $result = $this->httpJson('POST', $this->baseUrl . '/api/chat', $payload, $this->timeout);
        } catch (\Exception $e) {
            if ($jsonMode && (strpos($e->getMessage(), 'HTTP 400') !== false || strpos($e->getMessage(), 'format') !== false)) {
                unset($payload['format']);
                $result = $this->httpJson('POST', $this->baseUrl . '/api/chat', $payload, $this->timeout);
            } else {
                throw $e;
            }
        }
        $content = $result['message']['content'] ?? ($result['response'] ?? '');
        if (is_array($content)) {
            $content = json_encode($content, JSON_UNESCAPED_UNICODE);
        }
        $content = preg_replace('/<think>[\s\S]*?<\/think>/i', '', (string)$content);
        return trim($content);
    }

    /**
     * @return string
     */
    public function openaiRoot()
    {
        $base = rtrim($this->baseUrl, '/');
        if ($base === '') {
            return '';
        }
        if (preg_match('#/v\d+$#', $base) || preg_match('#/compatible-mode/v\d+$#', $base)) {
            return $base;
        }
        return $base . '/v1';
    }

    protected function openaiChat($model, $prompt, array $images, $jsonMode = false)
    {
        $content = [['type' => 'text', 'text' => $prompt]];
        foreach ($images as $b64) {
            $content[] = [
                'type'      => 'image_url',
                'image_url' => ['url' => 'data:image/jpeg;base64,' . $b64],
            ];
        }
        $payload = [
            'model'            => $model,
            'temperature'      => $this->temperature,
            'max_tokens'       => 8192,
            'enable_thinking'  => false,
            'messages'         => [
                ['role' => 'user', 'content' => $images ? $content : $prompt],
            ],
        ];
        if ($jsonMode) {
            $payload['response_format'] = ['type' => 'json_object'];
        }
        $url = $this->openaiRoot() . '/chat/completions';
        try {
            $result = $this->httpJson('POST', $url, $payload, $this->timeout, true);
        } catch (\Exception $e) {
            if ($jsonMode && (strpos($e->getMessage(), 'HTTP 400') !== false || stripos($e->getMessage(), 'response_format') !== false || stripos($e->getMessage(), 'json_object') !== false || stripos($e->getMessage(), 'enable_thinking') !== false)) {
                unset($payload['response_format'], $payload['enable_thinking']);
                $result = $this->httpJson('POST', $url, $payload, $this->timeout, true);
            } else {
                throw $e;
            }
        }
        if (!empty($result['error'])) {
            $msg = $result['error']['message'] ?? json_encode($result['error'], JSON_UNESCAPED_UNICODE);
            throw new Exception('线上模型：' . mb_substr((string)$msg, 0, 300));
        }
        $text = $result['choices'][0]['message']['content'] ?? '';
        if (is_array($text)) {
            $parts = [];
            foreach ($text as $part) {
                if (is_string($part)) {
                    $parts[] = $part;
                } elseif (is_array($part) && isset($part['text'])) {
                    $parts[] = $part['text'];
                }
            }
            $text = implode('', $parts);
        }
        return trim((string)$text);
    }

    protected function encodeImage($path)
    {
        if (!is_file($path)) {
            throw new Exception('识版文件不存在: ' . $path);
        }
        $data = file_get_contents($path);
        if ($data === false || $data === '') {
            throw new Exception('识版文件无法读取');
        }
        if (function_exists('imagecreatefromstring')) {
            $img = @imagecreatefromstring($data);
            if ($img) {
                $w = imagesx($img);
                $h = imagesy($img);
                $max = $this->driver === 'openai' ? 1280 : 1600;
                if ($w > $max || $h > $max || strlen($data) > 800 * 1024 || $this->driver === 'openai') {
                    $scale = min(1, $max / max($w, $h));
                    $nw = max(1, (int)($w * $scale));
                    $nh = max(1, (int)($h * $scale));
                    $dst = imagecreatetruecolor($nw, $nh);
                    imagecopyresampled($dst, $img, 0, 0, 0, 0, $nw, $nh, $w, $h);
                    ob_start();
                    imagejpeg($dst, null, 82);
                    $data = ob_get_clean();
                    imagedestroy($dst);
                }
                imagedestroy($img);
            }
        }
        return base64_encode($data);
    }

    protected function isModelLoadError(\Exception $e)
    {
        $msg = strtolower($e->getMessage());
        return strpos($msg, 'mllama') !== false
            || strpos($msg, 'unknown model architecture') !== false
            || strpos($msg, 'status code: 500') !== false;
    }

    /**
     * @param string $method
     * @param string $url
     * @param array|null $payload
     * @param int $timeout
     * @param bool $useBearer
     * @return array
     */
    protected function httpJson($method, $url, $payload, $timeout, $useBearer = false)
    {
        $ch = curl_init($url);
        $headers = ['Content-Type: application/json'];
        if (($useBearer || $this->driver === 'openai') && $this->apiKey) {
            $headers[] = 'Authorization: Bearer ' . $this->apiKey;
        }
        $opts = [
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_CUSTOMREQUEST  => $method,
            CURLOPT_HTTPHEADER     => $headers,
            CURLOPT_TIMEOUT        => $timeout,
            CURLOPT_CONNECTTIMEOUT => $this->driver === 'openai' ? 15 : 5,
        ];
        $ca = CONF_PATH . 'extra' . DS . 'cacert.pem';
        if (is_file($ca)) {
            $opts[CURLOPT_CAINFO] = $ca;
        }
        if ($payload !== null) {
            $opts[CURLOPT_POSTFIELDS] = json_encode($payload, JSON_UNESCAPED_UNICODE);
        }
        curl_setopt_array($ch, $opts);
        $body = curl_exec($ch);
        $errno = curl_errno($ch);
        $error = curl_error($ch);
        $code = (int)curl_getinfo($ch, CURLINFO_HTTP_CODE);
        curl_close($ch);
        if ($errno) {
            throw new Exception('大模型请求失败(' . $this->baseUrl . '): ' . $error);
        }
        if ($code >= 400) {
            throw new Exception('大模型 HTTP ' . $code . ': ' . mb_substr((string)$body, 0, 300));
        }
        $decoded = json_decode((string)$body, true);
        if (!is_array($decoded)) {
            throw new Exception('大模型返回非 JSON');
        }
        return $decoded;
    }
}