xref: /plugin/aichat/helper.php (revision 8285fff93bda4602771c807cbe9218d1f3b88d32) 
1<?php
2
3use dokuwiki\Extension\CLIPlugin;
4use dokuwiki\plugin\aichat\Model\AbstractModel;
5use dokuwiki\plugin\aichat\Chunk;
6use dokuwiki\plugin\aichat\Embeddings;
7use dokuwiki\plugin\aichat\Model\OpenAI\GPT35Turbo;
8use dokuwiki\plugin\aichat\Storage\AbstractStorage;
9use dokuwiki\plugin\aichat\Storage\PineconeStorage;
10use dokuwiki\plugin\aichat\Storage\SQLiteStorage;
11
12require_once __DIR__ . '/vendor/autoload.php';
13
14/**
15 * DokuWiki Plugin aichat (Helper Component)
16 *
17 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
18 * @author  Andreas Gohr <gohr@cosmocode.de>
19 */
20class helper_plugin_aichat extends \dokuwiki\Extension\Plugin
21{
22    /** @var CLIPlugin $logger */
23    protected $logger;
24    /** @var AbstractModel */
25    protected $model;
26    /** @var Embeddings */
27    protected $embeddings;
28    /** @var AbstractStorage */
29    protected $storage;
30
31    /**
32     * Use the given CLI plugin for logging
33     *
34     * @param CLIPlugin $logger
35     * @return void
36     */
37    public function setLogger($logger)
38    {
39        $this->logger = $logger;
40    }
41
42    /**
43     * Check if the current user is allowed to use the plugin (if it has been restricted)
44     *
45     * @return bool
46     */
47    public function userMayAccess()
48    {
49        global $auth;
50        global $USERINFO;
51        global $INPUT;
52
53        if (!$auth) return true;
54        if (!$this->getConf('restrict')) return true;
55        if (!isset($USERINFO)) return false;
56
57        return auth_isMember($this->getConf('restrict'), $INPUT->server->str('REMOTE_USER'), $USERINFO['grps']);
58    }
59
60    /**
61     * Access the OpenAI client
62     *
63     * @return GPT35Turbo
64     */
65    public function getModel()
66    {
67        if ($this->model === null) {
68            $class = '\\dokuwiki\\plugin\\aichat\\Model\\' . $this->getConf('model');
69
70            if (!class_exists($class)) {
71                throw new \RuntimeException('Configured model not found: ' . $class);
72            }
73            // FIXME for now we only have OpenAI models, so we can hardcode the auth setup
74            $this->model = new $class([
75                'key' => $this->getConf('openaikey'),
76                'org' => $this->getConf('openaiorg')
77            ]);
78        }
79
80        return $this->model;
81    }
82
83    /**
84     * Access the Embeddings interface
85     *
86     * @return Embeddings
87     */
88    public function getEmbeddings()
89    {
90        if ($this->embeddings === null) {
91            // FIXME we currently have only one storage backend, so we can hardcode it
92            $this->embeddings = new Embeddings($this->getModel(), $this->getStorage());
93            if ($this->logger) {
94                $this->embeddings->setLogger($this->logger);
95            }
96        }
97
98        return $this->embeddings;
99    }
100
101    /**
102     * Access the Storage interface
103     *
104     * @return AbstractStorage
105     */
106    public function getStorage()
107    {
108        if ($this->storage === null) {
109            if ($this->getConf('pinecone_apikey')) {
110                $this->storage = new PineconeStorage();
111            } else {
112                $this->storage = new SQLiteStorage();
113            }
114
115            if ($this->logger) {
116                $this->storage->setLogger($this->logger);
117            }
118        }
119
120        return $this->storage;
121    }
122
123    /**
124     * Ask a question with a chat history
125     *
126     * @param string $question
127     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
128     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
129     * @throws Exception
130     */
131    public function askChatQuestion($question, $history = [])
132    {
133        if ($history) {
134            $standaloneQuestion = $this->rephraseChatQuestion($question, $history);
135        } else {
136            $standaloneQuestion = $question;
137        }
138        return $this->askQuestion($standaloneQuestion);
139    }
140
141    /**
142     * Ask a single standalone question
143     *
144     * @param string $question
145     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
146     * @throws Exception
147     */
148    public function askQuestion($question)
149    {
150        $similar = $this->getEmbeddings()->getSimilarChunks($question);
151        if ($similar) {
152            $context = implode("\n", array_map(function (Chunk $chunk) {
153                return "\n```\n" . $chunk->getText() . "\n```\n";
154            }, $similar));
155            $prompt = $this->getPrompt('question', ['context' => $context]);
156        } else {
157            $prompt = $this->getPrompt('noanswer');
158        }
159
160        $messages = [
161            [
162                'role' => 'system',
163                'content' => $prompt
164            ],
165            [
166                'role' => 'user',
167                'content' => $question
168            ]
169        ];
170
171        $answer = $this->getModel()->getAnswer($messages);
172
173        return [
174            'question' => $question,
175            'answer' => $answer,
176            'sources' => $similar,
177        ];
178    }
179
180    /**
181     * Rephrase a question into a standalone question based on the chat history
182     *
183     * @param string $question The original user question
184     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
185     * @return string The rephrased question
186     * @throws Exception
187     */
188    public function rephraseChatQuestion($question, $history)
189    {
190        // go back in history as far as possible without hitting the token limit
191        $chatHistory = '';
192        $history = array_reverse($history);
193        foreach ($history as $row) {
194            if (
195                count($this->getEmbeddings()->getTokenEncoder()->encode($chatHistory)) >
196                $this->getModel()->getMaxRephrasingTokenLength()
197            ) {
198                break;
199            }
200
201            $chatHistory =
202                "Human: " . $row[0] . "\n" .
203                "Assistant: " . $row[1] . "\n" .
204                $chatHistory;
205        }
206
207        // ask openAI to rephrase the question
208        $prompt = $this->getPrompt('rephrase', ['history' => $chatHistory, 'question' => $question]);
209        $messages = [['role' => 'user', 'content' => $prompt]];
210        return $this->getModel()->getRephrasedQuestion($messages);
211    }
212
213    /**
214     * Load the given prompt template and fill in the variables
215     *
216     * @param string $type
217     * @param string[] $vars
218     * @return string
219     */
220    protected function getPrompt($type, $vars = [])
221    {
222        $template = file_get_contents($this->localFN('prompt_' . $type));
223
224        $replace = array();
225        foreach ($vars as $key => $val) {
226            $replace['{{' . strtoupper($key) . '}}'] = $val;
227        }
228
229        return strtr($template, $replace);
230    }
231}
232
233