xref: /plugin/aichat/helper.php (revision 01f06932bbd74c60ea6c93ab68b0d6cf32d05aea) 
1<?php
2
3use dokuwiki\plugin\aichat\Model\AbstractModel;
4use dokuwiki\plugin\aichat\Chunk;
5use dokuwiki\plugin\aichat\Embeddings;
6use dokuwiki\plugin\aichat\Model\OpenAI\GPT35Turbo;
7use dokuwiki\plugin\aichat\Storage\AbstractStorage;
8use dokuwiki\plugin\aichat\Storage\SQLiteStorage;
9
10require_once __DIR__ . '/vendor/autoload.php';
11
12/**
13 * DokuWiki Plugin aichat (Helper Component)
14 *
15 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
16 * @author  Andreas Gohr <gohr@cosmocode.de>
17 */
18class helper_plugin_aichat extends \dokuwiki\Extension\Plugin
19{
20    /** @var AbstractModel */
21    protected $model;
22    /** @var Embeddings */
23    protected $embeddings;
24    /** @var AbstractStorage */
25    protected $storage;
26
27    /**
28     * Check if the current user is allowed to use the plugin (if it has been restricted)
29     *
30     * @return bool
31     */
32    public function userMayAccess()
33    {
34        global $auth;
35        global $USERINFO;
36        global $INPUT;
37
38        if (!$auth) return true;
39        if (!$this->getConf('restrict')) return true;
40        if (!isset($USERINFO)) return false;
41
42        return auth_isMember($this->getConf('restrict'), $INPUT->server->str('REMOTE_USER'), $USERINFO['grps']);
43    }
44
45    /**
46     * Access the OpenAI client
47     *
48     * @return GPT35Turbo
49     */
50    public function getModel()
51    {
52        if ($this->model === null) {
53            $class = '\\dokuwiki\\plugin\\aichat\\Model\\' . $this->getConf('model');
54
55            if (!class_exists($class)) {
56                throw new \RuntimeException('Configured model not found: ' . $class);
57            }
58            // FIXME for now we only have OpenAI models, so we can hardcode the auth setup
59            $this->model = new $class([
60                'key' => $this->getConf('openaikey'),
61                'org' => $this->getConf('openaiorg')
62            ]);
63        }
64
65        return $this->model;
66    }
67
68    /**
69     * Access the Embeddings interface
70     *
71     * @return Embeddings
72     */
73    public function getEmbeddings()
74    {
75        if ($this->embeddings === null) {
76            // FIXME we currently have only one storage backend, so we can hardcode it
77            $this->embeddings = new Embeddings($this->getModel(), $this->getStorage());
78        }
79
80        return $this->embeddings;
81    }
82
83    /**
84     * Access the Storage interface
85     *
86     * @return AbstractStorage
87     */
88    public function getStorage()
89    {
90        if ($this->storage === null) {
91            $this->storage = new SQLiteStorage();
92        }
93
94        return $this->storage;
95    }
96
97    /**
98     * Ask a question with a chat history
99     *
100     * @param string $question
101     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
102     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
103     * @throws Exception
104     */
105    public function askChatQuestion($question, $history = [])
106    {
107        if ($history) {
108            $standaloneQuestion = $this->rephraseChatQuestion($question, $history);
109        } else {
110            $standaloneQuestion = $question;
111        }
112        return $this->askQuestion($standaloneQuestion);
113    }
114
115    /**
116     * Ask a single standalone question
117     *
118     * @param string $question
119     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
120     * @throws Exception
121     */
122    public function askQuestion($question)
123    {
124        $similar = $this->getEmbeddings()->getSimilarChunks($question);
125        if ($similar) {
126            $context = implode("\n", array_map(function (Chunk $chunk) {
127                return "\n```\n" . $chunk->getText() . "\n```\n";
128            }, $similar));
129            $prompt = $this->getPrompt('question', ['context' => $context]);
130        } else {
131            $prompt = $this->getPrompt('noanswer');
132        }
133
134        $messages = [
135            [
136                'role' => 'system',
137                'content' => $prompt
138            ],
139            [
140                'role' => 'user',
141                'content' => $question
142            ]
143        ];
144
145        $answer = $this->getModel()->getAnswer($messages);
146
147        return [
148            'question' => $question,
149            'answer' => $answer,
150            'sources' => $similar,
151        ];
152    }
153
154    /**
155     * Rephrase a question into a standalone question based on the chat history
156     *
157     * @param string $question The original user question
158     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
159     * @return string The rephrased question
160     * @throws Exception
161     */
162    public function rephraseChatQuestion($question, $history)
163    {
164        // go back in history as far as possible without hitting the token limit
165        $chatHistory = '';
166        $history = array_reverse($history);
167        foreach ($history as $row) {
168            if (
169                count($this->getEmbeddings()->getTokenEncoder()->encode($chatHistory)) >
170                $this->getModel()->getMaxRephrasingTokenLength()
171            ) {
172                break;
173            }
174
175            $chatHistory =
176                "Human: " . $row[0] . "\n" .
177                "Assistant: " . $row[1] . "\n" .
178                $chatHistory;
179        }
180
181        // ask openAI to rephrase the question
182        $prompt = $this->getPrompt('rephrase', ['history' => $chatHistory, 'question' => $question]);
183        $messages = [['role' => 'user', 'content' => $prompt]];
184        return $this->getModel()->getRephrasedQuestion($messages);
185    }
186
187    /**
188     * Load the given prompt template and fill in the variables
189     *
190     * @param string $type
191     * @param string[] $vars
192     * @return string
193     */
194    protected function getPrompt($type, $vars = [])
195    {
196        $template = file_get_contents($this->localFN('prompt_' . $type));
197
198        $replace = array();
199        foreach ($vars as $key => $val) {
200            $replace['{{' . strtoupper($key) . '}}'] = $val;
201        }
202
203        return strtr($template, $replace);
204    }
205}
206
207