xref: /plugin/aichat/helper.php (revision 689088446f64ff8d9dfdae9ae0666b45de449da7) 
1<?php
2
3use dokuwiki\plugin\aichat\backend\Chunk;
4use dokuwiki\plugin\aichat\Embeddings;
5use dokuwiki\plugin\aichat\OpenAI;
6use TikToken\Encoder;
7
8require_once __DIR__ . '/vendor/autoload.php';
9
10/**
11 * DokuWiki Plugin aichat (Helper Component)
12 *
13 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
14 * @author  Andreas Gohr <gohr@cosmocode.de>
15 */
16class helper_plugin_aichat extends \dokuwiki\Extension\Plugin
17{
18    /** @var OpenAI */
19    protected $openAI;
20    /** @var Embeddings */
21    protected $embeddings;
22
23    public function __construct()
24    {
25        $this->openAI = new OpenAI($this->getConf('openaikey'), $this->getConf('openaiorg'));
26        $this->embeddings = new Embeddings($this->openAI);
27    }
28
29    /**
30     * Check if the current user is allowed to use the plugin (if it has been restricted)
31     *
32     * @return bool
33     */
34    public function userMayAccess()
35    {
36        global $auth;
37        global $USERINFO;
38        global $INPUT;
39
40        if (!$auth) return true;
41        if (!$this->getConf('restrict')) return true;
42        if (!isset($USERINFO)) return false;
43
44        return auth_isMember($this->getConf('restrict'), $INPUT->server->str('REMOTE_USER'), $USERINFO['grps']);
45    }
46
47    /**
48     * Access the OpenAI client
49     *
50     * @return OpenAI
51     */
52    public function getOpenAI()
53    {
54        return $this->openAI;
55    }
56
57    /**
58     * Access the Embeddings interface
59     *
60     * @return Embeddings
61     */
62    public function getEmbeddings()
63    {
64        return $this->embeddings;
65    }
66
67    /**
68     * Ask a question with a chat history
69     *
70     * @param string $question
71     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
72     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
73     * @throws Exception
74     */
75    public function askChatQuestion($question, $history = [])
76    {
77        if ($history) {
78            $standaloneQuestion = $this->rephraseChatQuestion($question, $history);
79        } else {
80            $standaloneQuestion = $question;
81        }
82        return $this->askQuestion($standaloneQuestion);
83    }
84
85    /**
86     * Ask a single standalone question
87     *
88     * @param string $question
89     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
90     * @throws Exception
91     * @todo add context until token limit is hit
92     */
93    public function askQuestion($question)
94    {
95        $similar = $this->embeddings->getSimilarChunks($question);
96        if ($similar) {
97            $context = implode("\n", array_map(function (Chunk $chunk) {
98                return "\n```\n" . $chunk->getText() . "\n```\n";
99            }, $similar));
100            $prompt = $this->getPrompt('question', ['context' => $context]);
101        } else {
102            $prompt = $this->getPrompt('noanswer');
103        }
104
105        $messages = [
106            [
107                'role' => 'system',
108                'content' => $prompt
109            ],
110            [
111                'role' => 'user',
112                'content' => $question
113            ]
114        ];
115
116        $answer = $this->openAI->getChatAnswer($messages);
117
118        return [
119            'question' => $question,
120            'answer' => $answer,
121            'sources' => $similar,
122        ];
123    }
124
125    /**
126     * Rephrase a question into a standalone question based on the chat history
127     *
128     * @param string $question The original user question
129     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
130     * @return string The rephrased question
131     * @throws Exception
132     */
133    public function rephraseChatQuestion($question, $history)
134    {
135        // go back in history as far as possible without hitting the token limit
136        $tiktok = new Encoder();
137        $chatHistory = '';
138        $history = array_reverse($history);
139        foreach ($history as $row) {
140            if (count($tiktok->encode($chatHistory)) > 3000) {
141                break;
142            }
143
144            $chatHistory =
145                "Human: " . $row[0] . "\n" .
146                "Assistant: " . $row[1] . "\n" .
147                $chatHistory;
148        }
149
150        // ask openAI to rephrase the question
151        $prompt = $this->getPrompt('rephrase', ['history' => $chatHistory, 'question' => $question]);
152        $messages = [['role' => 'user', 'content' => $prompt]];
153        return $this->openAI->getChatAnswer($messages, OpenAI::REPHRASE_MODEL);
154    }
155
156    /**
157     * Load the given prompt template and fill in the variables
158     *
159     * @param string $type
160     * @param string[] $vars
161     * @return string
162     */
163    protected function getPrompt($type, $vars = [])
164    {
165        $template = file_get_contents($this->localFN('prompt_' . $type));
166
167        $replace = array();
168        foreach ($vars as $key => $val) {
169            $replace['{{' . strtoupper($key) . '}}'] = $val;
170        }
171
172        return strtr($template, $replace);
173    }
174}
175
176