xref: /plugin/aichat/helper.php (revision e33a1d7adcbf36c57f516e2f829ec8ad59cdb47b) 
1<?php
2
3use dokuwiki\Extension\CLIPlugin;
4use dokuwiki\plugin\aichat\AIChat;
5use dokuwiki\plugin\aichat\Chunk;
6use dokuwiki\plugin\aichat\Embeddings;
7use dokuwiki\plugin\aichat\Model\AbstractModel;
8use dokuwiki\plugin\aichat\Model\OpenAI\GPT35Turbo;
9use dokuwiki\plugin\aichat\Storage\AbstractStorage;
10use dokuwiki\plugin\aichat\Storage\PineconeStorage;
11use dokuwiki\plugin\aichat\Storage\SQLiteStorage;
12
13require_once __DIR__ . '/vendor/autoload.php';
14
15/**
16 * DokuWiki Plugin aichat (Helper Component)
17 *
18 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
19 * @author  Andreas Gohr <gohr@cosmocode.de>
20 */
21class helper_plugin_aichat extends \dokuwiki\Extension\Plugin
22{
23    /** @var CLIPlugin $logger */
24    protected $logger;
25    /** @var AbstractModel */
26    protected $model;
27    /** @var Embeddings */
28    protected $embeddings;
29    /** @var AbstractStorage */
30    protected $storage;
31
32    /**
33     * Use the given CLI plugin for logging
34     *
35     * @param CLIPlugin $logger
36     * @return void
37     */
38    public function setLogger($logger)
39    {
40        $this->logger = $logger;
41    }
42
43    /**
44     * Check if the current user is allowed to use the plugin (if it has been restricted)
45     *
46     * @return bool
47     */
48    public function userMayAccess()
49    {
50        global $auth;
51        global $USERINFO;
52        global $INPUT;
53
54        if (!$auth) return true;
55        if (!$this->getConf('restrict')) return true;
56        if (!isset($USERINFO)) return false;
57
58        return auth_isMember($this->getConf('restrict'), $INPUT->server->str('REMOTE_USER'), $USERINFO['grps']);
59    }
60
61    /**
62     * Access the OpenAI client
63     *
64     * @return GPT35Turbo
65     */
66    public function getModel()
67    {
68        if ($this->model === null) {
69            $class = '\\dokuwiki\\plugin\\aichat\\Model\\' . $this->getConf('model');
70
71            if (!class_exists($class)) {
72                throw new \RuntimeException('Configured model not found: ' . $class);
73            }
74            // FIXME for now we only have OpenAI models, so we can hardcode the auth setup
75            $this->model = new $class([
76                'key' => $this->getConf('openaikey'),
77                'org' => $this->getConf('openaiorg')
78            ]);
79        }
80
81        return $this->model;
82    }
83
84    /**
85     * Access the Embeddings interface
86     *
87     * @return Embeddings
88     */
89    public function getEmbeddings()
90    {
91        if ($this->embeddings === null) {
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            $prev = end($history);
136        } else {
137            $standaloneQuestion = $question;
138            $prev = [];
139        }
140        return $this->askQuestion($standaloneQuestion, $prev);
141    }
142
143    /**
144     * Ask a single standalone question
145     *
146     * @param string $question
147     * @param array $previous [user, ai] of the previous question
148     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
149     * @throws Exception
150     */
151    public function askQuestion($question, $previous = [])
152    {
153        $similar = $this->getEmbeddings()->getSimilarChunks($question, $this->getLanguageLimit());
154        if ($similar) {
155            $context = implode("\n", array_map(function (Chunk $chunk) {
156                return "\n```\n" . $chunk->getText() . "\n```\n";
157            }, $similar));
158            $prompt = $this->getPrompt('question', [
159                'context' => $context,
160                'language' => $this->getLanguagePrompt()
161            ]);
162        } else {
163            $prompt = $this->getPrompt('noanswer');
164        }
165
166        $messages = [
167            [
168                'role' => 'system',
169                'content' => $prompt
170            ],
171            [
172                'role' => 'user',
173                'content' => $question
174            ]
175        ];
176
177        if ($previous) {
178            array_unshift($messages, [
179                'role' => 'assistant',
180                'content' => $previous[1]
181            ]);
182            array_unshift($messages, [
183                'role' => 'user',
184                'content' => $previous[0]
185            ]);
186        }
187
188        $answer = $this->getModel()->getAnswer($messages);
189
190        return [
191            'question' => $question,
192            'answer' => $answer,
193            'sources' => $similar,
194        ];
195    }
196
197    /**
198     * Rephrase a question into a standalone question based on the chat history
199     *
200     * @param string $question The original user question
201     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
202     * @return string The rephrased question
203     * @throws Exception
204     */
205    public function rephraseChatQuestion($question, $history)
206    {
207        // go back in history as far as possible without hitting the token limit
208        $chatHistory = '';
209        $history = array_reverse($history);
210        foreach ($history as $row) {
211            if (
212                count($this->getEmbeddings()->getTokenEncoder()->encode($chatHistory)) >
213                $this->getModel()->getMaxRephrasingTokenLength()
214            ) {
215                break;
216            }
217
218            $chatHistory =
219                "Human: " . $row[0] . "\n" .
220                "Assistant: " . $row[1] . "\n" .
221                $chatHistory;
222        }
223
224        // ask openAI to rephrase the question
225        $prompt = $this->getPrompt('rephrase', ['history' => $chatHistory, 'question' => $question]);
226        $messages = [['role' => 'user', 'content' => $prompt]];
227        return $this->getModel()->getRephrasedQuestion($messages);
228    }
229
230    /**
231     * Load the given prompt template and fill in the variables
232     *
233     * @param string $type
234     * @param string[] $vars
235     * @return string
236     */
237    protected function getPrompt($type, $vars = [])
238    {
239        $template = file_get_contents($this->localFN('prompt_' . $type));
240
241        $replace = array();
242        foreach ($vars as $key => $val) {
243            $replace['{{' . strtoupper($key) . '}}'] = $val;
244        }
245
246        return strtr($template, $replace);
247    }
248
249    /**
250     * Construct the prompt to define the answer language
251     *
252     * @return string
253     */
254    protected function getLanguagePrompt()
255    {
256        global $conf;
257
258        if ($this->getConf('preferUIlanguage') > AIChat::LANG_AUTO_ALL) {
259            $isoLangnames = include(__DIR__ . '/lang/languages.php');
260            if (isset($isoLangnames[$conf['lang']])) {
261                $languagePrompt = 'Always answer in ' . $isoLangnames[$conf['lang']] . '.';
262                return $languagePrompt;
263            }
264        }
265
266        $languagePrompt = 'Always answer in the user\'s language.';
267        return $languagePrompt;
268    }
269
270    /**
271     * Should sources be limited to current language?
272     *
273     * @return string The current language code or empty string
274     */
275    public function getLanguageLimit()
276    {
277        if ($this->getConf('preferUIlanguage') >= AIChat::LANG_UI_LIMITED) {
278            global $conf;
279            return $conf['lang'];
280        } else {
281            return '';
282        }
283    }
284}
285
286