xref: /plugin/aichat/helper.php (revision 2071dced6f96936ea7b9bf5dbe8a117eef598448) 
1<?php
2
3use dokuwiki\Extension\CLIPlugin;
4use dokuwiki\Extension\Plugin;
5use dokuwiki\plugin\aichat\AIChat;
6use dokuwiki\plugin\aichat\Chunk;
7use dokuwiki\plugin\aichat\Embeddings;
8use dokuwiki\plugin\aichat\Model\ChatInterface;
9use dokuwiki\plugin\aichat\Model\EmbeddingInterface;
10use dokuwiki\plugin\aichat\ModelFactory;
11use dokuwiki\plugin\aichat\Storage\AbstractStorage;
12
13/**
14 * DokuWiki Plugin aichat (Helper Component)
15 *
16 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
17 * @author  Andreas Gohr <gohr@cosmocode.de>
18 */
19class helper_plugin_aichat extends Plugin
20{
21    /** @var ModelFactory */
22    public $factory;
23
24    /** @var CLIPlugin $logger */
25    protected $logger;
26
27    /** @var Embeddings */
28    protected $embeddings;
29    /** @var AbstractStorage */
30    protected $storage;
31
32    /** @var array where to store meta data on the last run */
33    protected $runDataFile;
34
35
36    /**
37     * Constructor. Initializes vendor autoloader
38     */
39    public function __construct()
40    {
41        require_once __DIR__ . '/vendor/autoload.php'; // FIXME obsolete from Kaos onwards
42        global $conf;
43        $this->runDataFile = $conf['metadir'] . '/aichat__run.json';
44        $this->loadConfig();
45        $this->factory = new ModelFactory($this->conf);
46    }
47
48    /**
49     * Use the given CLI plugin for logging
50     *
51     * @param CLIPlugin $logger
52     * @return void
53     */
54    public function setLogger($logger)
55    {
56        $this->logger = $logger;
57    }
58
59    /**
60     * Check if the current user is allowed to use the plugin (if it has been restricted)
61     *
62     * @return bool
63     */
64    public function userMayAccess()
65    {
66        global $auth;
67        global $USERINFO;
68        global $INPUT;
69
70        if (!$auth) return true;
71        if (!$this->getConf('restrict')) return true;
72        if (!isset($USERINFO)) return false;
73
74        return auth_isMember($this->getConf('restrict'), $INPUT->server->str('REMOTE_USER'), $USERINFO['grps']);
75    }
76
77    /**
78     * Access the Chat Model
79     *
80     * @return ChatInterface
81     */
82    public function getChatModel()
83    {
84        return $this->factory->getChatModel();
85    }
86
87    /**
88     * @return ChatInterface
89     */
90    public function getRephraseModel()
91    {
92        return $this->factory->getRephraseModel();
93    }
94
95    /**
96     * Access the Embedding Model
97     *
98     * @return EmbeddingInterface
99     */
100    public function getEmbeddingModel()
101    {
102        return $this->factory->getEmbeddingModel();
103    }
104
105    /**
106     * Access the Embeddings interface
107     *
108     * @return Embeddings
109     */
110    public function getEmbeddings()
111    {
112        if ($this->embeddings instanceof Embeddings) {
113            return $this->embeddings;
114        }
115
116        $this->embeddings = new Embeddings(
117            $this->getChatModel(),
118            $this->getEmbeddingModel(),
119            $this->getStorage(),
120            $this->conf
121        );
122        if ($this->logger) {
123            $this->embeddings->setLogger($this->logger);
124        }
125
126        return $this->embeddings;
127    }
128
129    /**
130     * Access the Storage interface
131     *
132     * @return AbstractStorage
133     */
134    public function getStorage()
135    {
136        if ($this->storage instanceof AbstractStorage) {
137            return $this->storage;
138        }
139
140        $class = '\\dokuwiki\\plugin\\aichat\\Storage\\' . $this->getConf('storage') . 'Storage';
141        $this->storage = new $class($this->conf);
142
143        if ($this->logger) {
144            $this->storage->setLogger($this->logger);
145        }
146
147        return $this->storage;
148    }
149
150    /**
151     * Ask a question with a chat history
152     *
153     * @param string $question
154     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
155     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
156     * @throws Exception
157     */
158    public function askChatQuestion($question, $history = [])
159    {
160        if ($history && $this->getConf('rephraseHistory') > 0) {
161            $standaloneQuestion = $this->rephraseChatQuestion($question, $history);
162        } else {
163            $standaloneQuestion = $question;
164        }
165        return $this->askQuestion($standaloneQuestion, $history);
166    }
167
168    /**
169     * Ask a single standalone question
170     *
171     * @param string $question
172     * @param array $history [user, ai] of the previous question
173     * @return array ['question' => $question, 'answer' => $answer, 'sources' => $sources]
174     * @throws Exception
175     */
176    public function askQuestion($question, $history = [])
177    {
178        $similar = $this->getEmbeddings()->getSimilarChunks($question, $this->getLanguageLimit());
179        if ($similar) {
180            $context = implode(
181                "\n",
182                array_map(static fn(Chunk $chunk) => "\n```\n" . $chunk->getText() . "\n```\n", $similar)
183            );
184            $prompt = $this->getPrompt('question', [
185                'context' => $context,
186                'question' => $question,
187            ]);
188        } else {
189            $prompt = $this->getPrompt('noanswer', [
190                'question' => $question,
191            ]);
192            $history = [];
193        }
194
195        $messages = $this->prepareMessages(
196            $this->getChatModel(),
197            $prompt,
198            $history,
199            $this->getConf('chatHistory')
200        );
201        $answer = $this->getChatModel()->getAnswer($messages);
202
203        return [
204            'question' => $question,
205            'answer' => $answer,
206            'sources' => $similar,
207        ];
208    }
209
210    /**
211     * Rephrase a question into a standalone question based on the chat history
212     *
213     * @param string $question The original user question
214     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
215     * @return string The rephrased question
216     * @throws Exception
217     */
218    public function rephraseChatQuestion($question, $history)
219    {
220        $prompt = $this->getPrompt('rephrase', [
221            'question' => $question,
222        ]);
223        $messages = $this->prepareMessages(
224            $this->getRephraseModel(),
225            $prompt,
226            $history,
227            $this->getConf('rephraseHistory')
228        );
229        return $this->getRephraseModel()->getAnswer($messages);
230    }
231
232    /**
233     * Prepare the messages for the AI
234     *
235     * @param ChatInterface $model The used model
236     * @param string $promptedQuestion The user question embedded in a prompt
237     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
238     * @param int $historySize The maximum number of messages to use from the history
239     * @return array An OpenAI compatible array of messages
240     */
241    protected function prepareMessages(
242        ChatInterface $model,
243        string $promptedQuestion,
244        array $history,
245        int $historySize
246    ): array {
247        // calculate the space for context
248        $remainingContext = $model->getMaxInputTokenLength();
249        $remainingContext -= $this->countTokens($promptedQuestion);
250        $safetyMargin = $remainingContext * 0.05; // 5% safety margin
251        $remainingContext -= $safetyMargin;
252        // FIXME we may want to also have an upper limit for the history and not always use the full context
253
254        $messages = $this->historyMessages($history, $remainingContext, $historySize);
255        $messages[] = [
256            'role' => 'user',
257            'content' => $promptedQuestion
258        ];
259        return $messages;
260    }
261
262    /**
263     * Create an array of OpenAI compatible messages from the given history
264     *
265     * Only as many messages are used as fit into the token limit
266     *
267     * @param array[] $history The chat history [[user, ai], [user, ai], ...]
268     * @param int $tokenLimit The maximum number of tokens to use
269     * @param int $sizeLimit The maximum number of messages to use
270     * @return array
271     */
272    protected function historyMessages(array $history, int $tokenLimit, int $sizeLimit): array
273    {
274        $remainingContext = $tokenLimit;
275
276        $messages = [];
277        $history = array_reverse($history);
278        $history = array_slice($history, 0, $sizeLimit);
279        foreach ($history as $row) {
280            $length = $this->countTokens($row[0] . $row[1]);
281            if ($length > $remainingContext) {
282                break;
283            }
284            $remainingContext -= $length;
285
286            $messages[] = [
287                'role' => 'assistant',
288                'content' => $row[1]
289            ];
290            $messages[] = [
291                'role' => 'user',
292                'content' => $row[0]
293            ];
294        }
295        return array_reverse($messages);
296    }
297
298    /**
299     * Get an aproximation of the token count for the given text
300     *
301     * @param $text
302     * @return int
303     */
304    protected function countTokens($text)
305    {
306        return count($this->getEmbeddings()->getTokenEncoder()->encode($text));
307    }
308
309    /**
310     * Load the given prompt template and fill in the variables
311     *
312     * @param string $type
313     * @param string[] $vars
314     * @return string
315     */
316    protected function getPrompt($type, $vars = [])
317    {
318        $template = file_get_contents($this->localFN($type, 'prompt'));
319        $vars['language'] = $this->getLanguagePrompt();
320
321        $replace = [];
322        foreach ($vars as $key => $val) {
323            $replace['{{' . strtoupper($key) . '}}'] = $val;
324        }
325
326        return strtr($template, $replace);
327    }
328
329    /**
330     * Construct the prompt to define the answer language
331     *
332     * @return string
333     */
334    protected function getLanguagePrompt()
335    {
336        global $conf;
337        $isoLangnames = include(__DIR__ . '/lang/languages.php');
338
339        $currentLang = $isoLangnames[$conf['lang']] ?? 'English';
340
341        if ($this->getConf('preferUIlanguage') > AIChat::LANG_AUTO_ALL) {
342            if (isset($isoLangnames[$conf['lang']])) {
343                $languagePrompt = 'Always answer in ' . $isoLangnames[$conf['lang']] . '.';
344                return $languagePrompt;
345            }
346        }
347
348        $languagePrompt = 'Always answer in the user\'s language. ' .
349            "If you are unsure about the language, speak $currentLang.";
350        return $languagePrompt;
351    }
352
353    /**
354     * Should sources be limited to current language?
355     *
356     * @return string The current language code or empty string
357     */
358    public function getLanguageLimit()
359    {
360        if ($this->getConf('preferUIlanguage') >= AIChat::LANG_UI_LIMITED) {
361            global $conf;
362            return $conf['lang'];
363        } else {
364            return '';
365        }
366    }
367
368    /**
369     * Store info about the last run
370     *
371     * @param array $data
372     * @return void
373     */
374    public function setRunData(array $data)
375    {
376        file_put_contents($this->runDataFile, json_encode($data, JSON_PRETTY_PRINT));
377    }
378
379    /**
380     * Get info about the last run
381     *
382     * @return array
383     */
384    public function getRunData()
385    {
386        if (!file_exists($this->runDataFile)) {
387            return [];
388        }
389        return json_decode(file_get_contents($this->runDataFile), true);
390    }
391}
392