xref: /plugin/aichat/cli.php (revision 553920162c56b922c6ae5be71ff4442e666a63d3) 
1<?php
2
3use dokuwiki\plugin\aichat\backend\Chunk;
4use splitbrain\phpcli\Colors;
5use splitbrain\phpcli\Options;
6
7
8/**
9 * DokuWiki Plugin aichat (CLI Component)
10 *
11 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
12 * @author  Andreas Gohr <gohr@cosmocode.de>
13 */
14class cli_plugin_aichat extends \dokuwiki\Extension\CLIPlugin
15{
16    /** @var helper_plugin_aichat */
17    protected $helper;
18
19    public function __construct($autocatch = true)
20    {
21        parent::__construct($autocatch);
22        $this->helper = plugin_load('helper', 'aichat');
23        $this->helper->getEmbeddings()->setLogger($this);
24    }
25
26    /** @inheritDoc */
27    protected function setup(Options $options)
28    {
29        $options->useCompactHelp();
30
31        $options->setHelp(
32            'Manage and query the AI chatbot data. Please note that calls to your LLM provider will be made. ' .
33            'This may incur costs.'
34        );
35
36        $options->registerCommand(
37            'embed',
38            'Create embeddings for all pages. This skips pages that already have embeddings'
39        );
40        $options->registerOption(
41            'clear',
42            'Clear all existing embeddings before creating new ones',
43            'c', false, 'embed'
44        );
45
46        $options->registerCommand('similar', 'Search for similar pages');
47        $options->registerArgument('query', 'Look up chunks similar to this query', true, 'similar');
48
49        $options->registerCommand('ask', 'Ask a question');
50        $options->registerArgument('question', 'The question to ask', true, 'ask');
51
52        $options->registerCommand('chat', 'Start an interactive chat session');
53
54        $options->registerCommand('split', 'Split a page into chunks (for debugging)');
55        $options->registerArgument('page', 'The page to split', true, 'split');
56
57        $options->registerCommand('info', 'Get Info about the vector storage');
58    }
59
60    /** @inheritDoc */
61    protected function main(Options $options)
62    {
63        switch ($options->getCmd()) {
64
65            case 'embed':
66                $this->createEmbeddings($options->getOpt('clear'));
67                break;
68            case 'similar':
69                $this->similar($options->getArgs()[0]);
70                break;
71            case 'ask':
72                $this->ask($options->getArgs()[0]);
73                break;
74            case 'chat':
75                $this->chat();
76                break;
77            case 'split':
78                $this->split($options->getArgs()[0]);
79                break;
80            case 'info':
81                $this->treeinfo();
82                break;
83            default:
84                echo $options->help();
85        }
86    }
87
88    /**
89     * @return void
90     */
91    protected function treeinfo()
92    {
93        $stats = $this->helper->getEmbeddings()->getStorage()->statistics();
94        foreach ($stats as $key => $value) {
95            echo $key . ': ' . $value . "\n";
96        }
97    }
98
99    /**
100     * Split the given page into chunks and print them
101     *
102     * @param string $page
103     * @return void
104     * @throws Exception
105     */
106    protected function split($page)
107    {
108        $text = rawWiki($page);
109        $chunks = $this->helper->getEmbeddings()->splitIntoChunks($text);
110        foreach ($chunks as $chunk) {
111            echo $chunk;
112            echo "\n";
113            $this->colors->ptln('--------------------------------', Colors::C_LIGHTPURPLE);
114        }
115        $this->success('Split into ' . count($chunks) . ' chunks');
116    }
117
118    /**
119     * Interactive Chat Session
120     *
121     * @return void
122     * @throws Exception
123     */
124    protected function chat()
125    {
126        $history = [];
127        while ($q = $this->readLine('Your Question')) {
128            $this->helper->getOpenAI()->resetUsageStats();
129            if ($history) {
130                $question = $this->helper->rephraseChatQuestion($q, $history);
131                $this->colors->ptln("Interpretation: $question", Colors::C_LIGHTPURPLE);
132            } else {
133                $question = $q;
134            }
135            $result = $this->helper->askQuestion($question);
136            $history[] = [$q, $result['answer']];
137            $this->printAnswer($result);
138        }
139    }
140
141    /**
142     * Handle a single, standalone question
143     *
144     * @param string $query
145     * @return void
146     * @throws Exception
147     */
148    protected function ask($query)
149    {
150        $result = $this->helper->askQuestion($query);
151        $this->printAnswer($result);
152    }
153
154    /**
155     * Get the pages that are similar to the query
156     *
157     * @param string $query
158     * @return void
159     */
160    protected function similar($query)
161    {
162        $sources = $this->helper->getEmbeddings()->getSimilarChunks($query);
163        foreach ($sources as $source) {
164            $this->colors->ptln($source->getPage(), Colors::C_LIGHTBLUE);
165        }
166    }
167
168    /**
169     * Recreate chunks and embeddings for all pages
170     *
171     * @return void
172     * @todo make skip regex configurable
173     */
174    protected function createEmbeddings($clear)
175    {
176        ini_set('memory_limit', -1); // we may need a lot of memory here
177        $this->helper->getEmbeddings()->createNewIndex('/(^|:)(playground|sandbox)(:|$)/', $clear);
178        $this->notice('Peak memory used: {memory}', ['memory' => filesize_h(memory_get_peak_usage(true))]);
179    }
180
181    /**
182     * Print the given detailed answer in a nice way
183     *
184     * @param array $answer
185     * @return void
186     */
187    protected function printAnswer($answer)
188    {
189        $this->colors->ptln($answer['answer'], Colors::C_LIGHTCYAN);
190        echo "\n";
191        foreach ($answer['sources'] as $source) {
192            /** @var Chunk $source */
193            $this->colors->ptln("\t" . $source->getPage(), Colors::C_LIGHTBLUE);
194        }
195        echo "\n";
196        $this->printUsage();
197    }
198
199    /**
200     * Print the usage statistics for OpenAI
201     *
202     * @return void
203     */
204    protected function printUsage() {
205        $this->info(
206            'Made {requests} requests in {time}s to OpenAI. Used {tokens} tokens for about ${cost}.',
207            $this->helper->getOpenAI()->getUsageStats()
208        );
209    }
210
211    /**
212     * Interactively ask for a value from the user
213     *
214     * @param string $prompt
215     * @return string
216     */
217    protected function readLine($prompt)
218    {
219        $value = '';
220
221        while ($value === '') {
222            echo $prompt;
223            echo ': ';
224
225            $fh = fopen('php://stdin', 'r');
226            $value = trim(fgets($fh));
227            fclose($fh);
228        }
229
230        return $value;
231    }
232}
233
234