xref: /plugin/aichat/cli.php (revision 99b713bf5ca50ca020487a25994f0098a1240134) 
1<?php
2
3use dokuwiki\Extension\CLIPlugin;
4use dokuwiki\plugin\aichat\Chunk;
5use dokuwiki\Search\Indexer;
6use splitbrain\phpcli\Colors;
7use splitbrain\phpcli\Options;
8use splitbrain\phpcli\TableFormatter;
9
10/**
11 * DokuWiki Plugin aichat (CLI Component)
12 *
13 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
14 * @author  Andreas Gohr <gohr@cosmocode.de>
15 */
16class cli_plugin_aichat extends CLIPlugin
17{
18    /** @var helper_plugin_aichat */
19    protected $helper;
20
21    public function __construct($autocatch = true)
22    {
23        parent::__construct($autocatch);
24        $this->helper = plugin_load('helper', 'aichat');
25        $this->helper->setLogger($this);
26    }
27
28    /** @inheritDoc */
29    protected function setup(Options $options)
30    {
31        $options->useCompactHelp();
32
33        $options->setHelp(
34            'Manage and query the AI chatbot data. Please note that calls to your LLM provider will be made. ' .
35            'This may incur costs.'
36        );
37
38        $options->registerCommand(
39            'embed',
40            'Create embeddings for all pages. This skips pages that already have embeddings'
41        );
42        $options->registerOption(
43            'clear',
44            'Clear all existing embeddings before creating new ones',
45            'c',
46            false,
47            'embed'
48        );
49
50        $options->registerCommand('maintenance', 'Run storage maintenance. Refer to the documentation for details.');
51
52        $options->registerCommand('similar', 'Search for similar pages');
53        $options->registerArgument('query', 'Look up chunks similar to this query', true, 'similar');
54
55        $options->registerCommand('ask', 'Ask a question');
56        $options->registerArgument('question', 'The question to ask', true, 'ask');
57
58        $options->registerCommand('chat', 'Start an interactive chat session');
59
60        $options->registerCommand('models', 'List available models');
61
62        $options->registerCommand('info', 'Get Info about the vector storage and other stats');
63
64        $options->registerCommand('split', 'Split a page into chunks (for debugging)');
65        $options->registerArgument('page', 'The page to split', true, 'split');
66
67        $options->registerCommand('page', 'Check if chunks for a given page are available (for debugging)');
68        $options->registerArgument('page', 'The page to check', true, 'page');
69        $options->registerOption('dump', 'Dump the chunks', 'd', false, 'page');
70
71        $options->registerCommand('tsv', 'Create TSV files for visualizing at http://projector.tensorflow.org/' .
72            ' Not supported on all storages.');
73        $options->registerArgument('vector.tsv', 'The vector file', false, 'tsv');
74        $options->registerArgument('meta.tsv', 'The meta file', false, 'tsv');
75    }
76
77    /** @inheritDoc */
78    protected function main(Options $options)
79    {
80        $this->loadConfig();
81        ini_set('memory_limit', -1);
82        switch ($options->getCmd()) {
83            case 'embed':
84                $this->createEmbeddings($options->getOpt('clear'));
85                break;
86            case 'maintenance':
87                $this->runMaintenance();
88                break;
89            case 'similar':
90                $this->similar($options->getArgs()[0]);
91                break;
92            case 'ask':
93                $this->ask($options->getArgs()[0]);
94                break;
95            case 'chat':
96                $this->chat();
97                break;
98            case 'models':
99                $this->models();
100                break;
101            case 'split':
102                $this->split($options->getArgs()[0]);
103                break;
104            case 'page':
105                $this->page($options->getArgs()[0], $options->getOpt('dump'));
106                break;
107            case 'info':
108                $this->showinfo();
109                break;
110            case 'tsv':
111                $args = $options->getArgs();
112                $vector = $args[0] ?? 'vector.tsv';
113                $meta = $args[1] ?? 'meta.tsv';
114                $this->tsv($vector, $meta);
115                break;
116            default:
117                echo $options->help();
118        }
119    }
120
121    /**
122     * @return void
123     */
124    protected function showinfo()
125    {
126        $stats = [
127            'chat model' => $this->getConf('chatmodel'),
128            'embed model' => $this->getConf('embedmodel'),
129        ];
130        $stats = array_merge(
131            $stats,
132            array_map('dformat', $this->helper->getRunData()),
133            $this->helper->getStorage()->statistics()
134        );
135        $this->printTable($stats);
136    }
137
138    /**
139     * Print key value data as tabular data
140     *
141     * @param array $data
142     * @param int $level
143     * @return void
144     */
145    protected function printTable($data, $level = 0)
146    {
147        $tf = new TableFormatter($this->colors);
148        foreach ($data as $key => $value) {
149            if (is_array($value)) {
150                echo $tf->format(
151                    [$level * 2, 20, '*'],
152                    ['', $key, ''],
153                    [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
154                );
155                $this->printTable($value, $level + 1);
156            } else {
157                echo $tf->format(
158                    [$level * 2, 20, '*'],
159                    ['', $key, $value],
160                    [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTGRAY]
161                );
162            }
163        }
164    }
165
166    /**
167     * Check chunk availability for a given page
168     *
169     * @param string $page
170     * @return void
171     */
172    protected function page($page, $dump = false)
173    {
174        $indexer = new Indexer();
175        $pages = $indexer->getPages();
176        $pos = array_search(cleanID($page), $pages);
177
178        if ($pos === false) {
179            $this->error('Page not found');
180            return;
181        }
182
183        $storage = $this->helper->getStorage();
184        $chunks = $storage->getPageChunks($page, $pos * 100);
185        if ($chunks) {
186            $this->success('Found ' . count($chunks) . ' chunks');
187            if ($dump) {
188                echo json_encode($chunks, JSON_PRETTY_PRINT);
189            }
190        } else {
191            $this->error('No chunks found');
192        }
193    }
194
195    /**
196     * Split the given page into chunks and print them
197     *
198     * @param string $page
199     * @return void
200     * @throws Exception
201     */
202    protected function split($page)
203    {
204        $text = rawWiki($page);
205        $chunks = $this->helper->getEmbeddings()->splitIntoChunks($text);
206        foreach ($chunks as $chunk) {
207            echo $chunk;
208            echo "\n";
209            $this->colors->ptln('--------------------------------', Colors::C_LIGHTPURPLE);
210        }
211        $this->success('Split into ' . count($chunks) . ' chunks');
212    }
213
214    /**
215     * Interactive Chat Session
216     *
217     * @return void
218     * @throws Exception
219     */
220    protected function chat()
221    {
222        if ($this->loglevel['debug']['enabled']) {
223            $this->helper->getChatModel()->setDebug(true);
224        }
225
226        $history = [];
227        while ($q = $this->readLine('Your Question')) {
228            $this->helper->getChatModel()->resetUsageStats();
229            $result = $this->helper->askChatQuestion($q, $history);
230            $this->colors->ptln("Interpretation: {$result['question']}", Colors::C_LIGHTPURPLE);
231            $history[] = [$result['question'], $result['answer']];
232            $this->printAnswer($result);
233        }
234    }
235
236    protected function models()
237    {
238        $result = [
239            'chat' => [],
240            'embedding' => [],
241        ];
242
243
244        $jsons = glob(__DIR__ . '/Model/*/models.json');
245        foreach ($jsons as $json) {
246            $models = json_decode(file_get_contents($json), true);
247            foreach ($models as $type => $model) {
248                $namespace = basename(dirname($json));
249                foreach ($model as $name => $info) {
250
251
252                    $class = '\\dokuwiki\\plugin\\aichat\\Model\\' . $namespace . '\\' . ucfirst($type) . 'Model';
253                    try {
254                        new $class($name, $this->conf);
255                        $info['confok'] = true;
256                    } catch (Exception $e) {
257                        $info['confok'] = false;
258                    }
259
260                    $result[$type]["$namespace $name"] = $info;
261                }
262            }
263        }
264
265        $td = new TableFormatter($this->colors);
266        $cols = [30, 20, 20, '*'];
267        echo "==== Chat Models ====\n\n";
268        echo $td->format(
269            $cols,
270            ['Model', 'Token Limits', 'Price USD/M', 'Description'],
271            [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
272        );
273        foreach ($result['chat'] as $name => $info) {
274            echo $td->format(
275                $cols,
276                [
277                    $name,
278                    sprintf(" In: %7d\nOut: %7d", $info['inputTokens'], $info['outputTokens']),
279                    sprintf(" In: %.2f\nOut: %.2f", $info['inputTokenPrice'], $info['outputTokenPrice']),
280                    $info['description']."\n"
281                ],
282                [
283                    $info['confok'] ? Colors::C_LIGHTGREEN : Colors::C_LIGHTRED,
284                ]
285            );
286        }
287
288        $cols = [30, 10, 10, 10, '*'];
289        echo "==== Embedding Models ====\n\n";
290        echo $td->format(
291            $cols,
292            ['Model', 'Token Limits', 'Price USD/M', 'Dimensions', 'Description'],
293            [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
294        );
295        foreach ($result['embedding'] as $name => $info) {
296            echo $td->format(
297                $cols,
298                [
299                    $name,
300                    sprintf("%7d", $info['inputTokens']),
301                    sprintf("%.2f", $info['inputTokenPrice']),
302                    $info['dimensions'],
303                    $info['description']."\n"
304                ],
305                [
306                    $info['confok'] ? Colors::C_LIGHTGREEN : Colors::C_LIGHTRED,
307                ]
308            );
309        }
310
311        $this->colors->ptln('Current prices may differ', Colors::C_RED);
312    }
313
314    /**
315     * Handle a single, standalone question
316     *
317     * @param string $query
318     * @return void
319     * @throws Exception
320     */
321    protected function ask($query)
322    {
323        if ($this->loglevel['debug']['enabled']) {
324            $this->helper->getChatModel()->setDebug(true);
325        }
326
327        $result = $this->helper->askQuestion($query);
328        $this->printAnswer($result);
329    }
330
331    /**
332     * Get the pages that are similar to the query
333     *
334     * @param string $query
335     * @return void
336     */
337    protected function similar($query)
338    {
339        $langlimit = $this->helper->getLanguageLimit();
340        if ($langlimit) {
341            $this->info('Limiting results to {lang}', ['lang' => $langlimit]);
342        }
343
344        $sources = $this->helper->getEmbeddings()->getSimilarChunks($query, $langlimit);
345        $this->printSources($sources);
346    }
347
348    /**
349     * Run the maintenance tasks
350     *
351     * @return void
352     */
353    protected function runMaintenance()
354    {
355        $start = time();
356        $this->helper->getStorage()->runMaintenance();
357        $this->notice('Peak memory used: {memory}', ['memory' => filesize_h(memory_get_peak_usage(true))]);
358        $this->notice('Spent time: {time}min', ['time' => round((time() - $start) / 60, 2)]);
359
360        $data = $this->helper->getRunData();
361        $data['maintenance ran at'] = time();
362        $this->helper->setRunData($data);
363    }
364
365    /**
366     * Recreate chunks and embeddings for all pages
367     *
368     * @return void
369     */
370    protected function createEmbeddings($clear)
371    {
372        [$skipRE, $matchRE] = $this->getRegexps();
373
374        $start = time();
375        $this->helper->getEmbeddings()->createNewIndex($skipRE, $matchRE, $clear);
376        $this->notice('Peak memory used: {memory}', ['memory' => filesize_h(memory_get_peak_usage(true))]);
377        $this->notice('Spent time: {time}min', ['time' => round((time() - $start) / 60, 2)]);
378
379        $data = $this->helper->getRunData();
380        $data['embed ran at'] = time();
381        $this->helper->setRunData($data);
382    }
383
384    /**
385     * Dump TSV files for debugging
386     *
387     * @return void
388     */
389    protected function tsv($vector, $meta)
390    {
391
392        $storage = $this->helper->getStorage();
393        $storage->dumpTSV($vector, $meta);
394        $this->success('written to ' . $vector . ' and ' . $meta);
395    }
396
397    /**
398     * Print the given detailed answer in a nice way
399     *
400     * @param array $answer
401     * @return void
402     */
403    protected function printAnswer($answer)
404    {
405        $this->colors->ptln($answer['answer'], Colors::C_LIGHTCYAN);
406        echo "\n";
407        $this->printSources($answer['sources']);
408        echo "\n";
409        $this->printUsage();
410    }
411
412    /**
413     * Print the given sources
414     *
415     * @param Chunk[] $sources
416     * @return void
417     */
418    protected function printSources($sources)
419    {
420        foreach ($sources as $source) {
421            /** @var Chunk $source */
422            $this->colors->ptln(
423                "\t" . $source->getPage() . ' ' . $source->getId() . ' (' . $source->getScore() . ')',
424                Colors::C_LIGHTBLUE
425            );
426        }
427    }
428
429    /**
430     * Print the usage statistics for OpenAI
431     *
432     * @return void
433     */
434    protected function printUsage()
435    {
436        $this->info(
437            'Made {requests} requests in {time}s to Model. Used {tokens} tokens for about ${cost}.',
438            $this->helper->getChatModel()->getUsageStats()
439        );
440    }
441
442    /**
443     * Interactively ask for a value from the user
444     *
445     * @param string $prompt
446     * @return string
447     */
448    protected function readLine($prompt)
449    {
450        $value = '';
451
452        while ($value === '') {
453            echo $prompt;
454            echo ': ';
455
456            $fh = fopen('php://stdin', 'r');
457            $value = trim(fgets($fh));
458            fclose($fh);
459        }
460
461        return $value;
462    }
463
464    /**
465     * Read the skip and match regex from the config
466     *
467     * Ensures the regular expressions are valid
468     *
469     * @return string[] [$skipRE, $matchRE]
470     */
471    protected function getRegexps()
472    {
473        $skip = $this->getConf('skipRegex');
474        $skipRE = '';
475        $match = $this->getConf('matchRegex');
476        $matchRE = '';
477
478        if ($skip) {
479            $skipRE = '/' . $skip . '/';
480            if (@preg_match($skipRE, '') === false) {
481                $this->error(preg_last_error_msg());
482                $this->error('Invalid regular expression in $conf[\'skipRegex\']. Ignored.');
483                $skipRE = '';
484            } else {
485                $this->success('Skipping pages matching ' . $skipRE);
486            }
487        }
488
489        if ($match) {
490            $matchRE = '/' . $match . '/';
491            if (@preg_match($matchRE, '') === false) {
492                $this->error(preg_last_error_msg());
493                $this->error('Invalid regular expression in $conf[\'matchRegex\']. Ignored.');
494                $matchRE = '';
495            } else {
496                $this->success('Only indexing pages matching ' . $matchRE);
497            }
498        }
499        return [$skipRE, $matchRE];
500    }
501}
502