xref: /plugin/aichat/cli.php (revision ab1f8dde36106432cc0a6f320220da5fae6971fe) 
1<?php
2
3use dokuwiki\Extension\CLIPlugin;
4use dokuwiki\plugin\aichat\AbstractCLI;
5use dokuwiki\plugin\aichat\Chunk;
6use dokuwiki\plugin\aichat\ModelFactory;
7use dokuwiki\Search\Indexer;
8use splitbrain\phpcli\Colors;
9use splitbrain\phpcli\Options;
10use splitbrain\phpcli\TableFormatter;
11
12/**
13 * DokuWiki Plugin aichat (CLI Component)
14 *
15 * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
16 * @author  Andreas Gohr <gohr@cosmocode.de>
17 */
18class cli_plugin_aichat extends AbstractCLI
19{
20    /** @var helper_plugin_aichat */
21    protected $helper;
22
23    /** @inheritDoc */
24    protected function setup(Options $options)
25    {
26        parent::setup($options);
27
28        $options->setHelp(
29            'Manage and query the AI chatbot data. Please note that calls to your LLM provider will be made. ' .
30            'This may incur costs.'
31        );
32
33        $options->registerOption(
34            'model',
35            'Overrides the chat and rephrasing model settings and uses this model instead',
36            '',
37            'model'
38        );
39
40        $options->registerCommand(
41            'embed',
42            'Create embeddings for all pages. This skips pages that already have embeddings'
43        );
44        $options->registerOption(
45            'clear',
46            'Clear all existing embeddings before creating new ones',
47            'c',
48            false,
49            'embed'
50        );
51
52        $options->registerCommand('maintenance', 'Run storage maintenance. Refer to the documentation for details.');
53
54        $options->registerCommand('similar', 'Search for similar pages');
55        $options->registerArgument('query', 'Look up chunks similar to this query', true, 'similar');
56
57        $options->registerCommand('ask', 'Ask a question');
58        $options->registerArgument('question', 'The question to ask', true, 'ask');
59
60        $options->registerCommand('chat', 'Start an interactive chat session');
61
62        $options->registerCommand('models', 'List available models');
63
64        $options->registerCommand('info', 'Get Info about the vector storage and other stats');
65
66        $options->registerCommand('split', 'Split a page into chunks (for debugging)');
67        $options->registerArgument('page', 'The page to split', true, 'split');
68
69        $options->registerCommand('page', 'Check if chunks for a given page are available (for debugging)');
70        $options->registerArgument('page', 'The page to check', true, 'page');
71        $options->registerOption('dump', 'Dump the chunks', 'd', false, 'page');
72
73        $options->registerCommand('tsv', 'Create TSV files for visualizing at http://projector.tensorflow.org/' .
74            ' Not supported on all storages.');
75        $options->registerArgument('vector.tsv', 'The vector file', false, 'tsv');
76        $options->registerArgument('meta.tsv', 'The meta file', false, 'tsv');
77    }
78
79    /** @inheritDoc */
80    protected function main(Options $options)
81    {
82        parent::main($options);
83
84        $model = $options->getOpt('model');
85        if($model) {
86            $this->helper->updateConfig(
87                ['chatmodel' => $model, 'rephasemodel' => $model]
88            );
89        }
90
91        switch ($options->getCmd()) {
92            case 'embed':
93                $this->createEmbeddings($options->getOpt('clear'));
94                break;
95            case 'maintenance':
96                $this->runMaintenance();
97                break;
98            case 'similar':
99                $this->similar($options->getArgs()[0]);
100                break;
101            case 'ask':
102                $this->ask($options->getArgs()[0]);
103                break;
104            case 'chat':
105                $this->chat();
106                break;
107            case 'models':
108                $this->models();
109                break;
110            case 'split':
111                $this->split($options->getArgs()[0]);
112                break;
113            case 'page':
114                $this->page($options->getArgs()[0], $options->getOpt('dump'));
115                break;
116            case 'info':
117                $this->showinfo();
118                break;
119            case 'tsv':
120                $args = $options->getArgs();
121                $vector = $args[0] ?? 'vector.tsv';
122                $meta = $args[1] ?? 'meta.tsv';
123                $this->tsv($vector, $meta);
124                break;
125            default:
126                echo $options->help();
127        }
128    }
129
130    /**
131     * @return void
132     */
133    protected function showinfo()
134    {
135        $stats = [
136            'chat model' => $this->getConf('chatmodel'),
137            'embed model' => $this->getConf('embedmodel'),
138        ];
139        $stats = array_merge(
140            $stats,
141            array_map('dformat', $this->helper->getRunData()),
142            $this->helper->getStorage()->statistics()
143        );
144        $this->printTable($stats);
145    }
146
147    /**
148     * Print key value data as tabular data
149     *
150     * @param array $data
151     * @param int $level
152     * @return void
153     */
154    protected function printTable($data, $level = 0)
155    {
156        $tf = new TableFormatter($this->colors);
157        foreach ($data as $key => $value) {
158            if (is_array($value)) {
159                echo $tf->format(
160                    [$level * 2, 20, '*'],
161                    ['', $key, ''],
162                    [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
163                );
164                $this->printTable($value, $level + 1);
165            } else {
166                echo $tf->format(
167                    [$level * 2, 20, '*'],
168                    ['', $key, $value],
169                    [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTGRAY]
170                );
171            }
172        }
173    }
174
175    /**
176     * Check chunk availability for a given page
177     *
178     * @param string $page
179     * @return void
180     */
181    protected function page($page, $dump = false)
182    {
183        $indexer = new Indexer();
184        $pages = $indexer->getPages();
185        $pos = array_search(cleanID($page), $pages);
186
187        if ($pos === false) {
188            $this->error('Page not found');
189            return;
190        }
191
192        $storage = $this->helper->getStorage();
193        $chunks = $storage->getPageChunks($page, $pos * 100);
194        if ($chunks) {
195            $this->success('Found ' . count($chunks) . ' chunks');
196            if ($dump) {
197                echo json_encode($chunks, JSON_PRETTY_PRINT);
198            }
199        } else {
200            $this->error('No chunks found');
201        }
202    }
203
204    /**
205     * Split the given page into chunks and print them
206     *
207     * @param string $page
208     * @return void
209     * @throws Exception
210     */
211    protected function split($page)
212    {
213        $chunks = $this->helper->getEmbeddings()->createPageChunks($page, 0);
214        foreach ($chunks as $chunk) {
215            echo $chunk->getText();
216            echo "\n";
217            $this->colors->ptln('--------------------------------', Colors::C_LIGHTPURPLE);
218        }
219        $this->success('Split into ' . count($chunks) . ' chunks');
220    }
221
222    /**
223     * Interactive Chat Session
224     *
225     * @return void
226     * @throws Exception
227     */
228    protected function chat()
229    {
230        $history = [];
231        while ($q = $this->readLine('Your Question')) {
232            $this->helper->getChatModel()->resetUsageStats();
233            $this->helper->getRephraseModel()->resetUsageStats();
234            $this->helper->getEmbeddingModel()->resetUsageStats();
235            $result = $this->helper->askChatQuestion($q, $history);
236            $this->colors->ptln("Interpretation: {$result['question']}", Colors::C_LIGHTPURPLE);
237            $history[] = [$result['question'], $result['answer']];
238            $this->printAnswer($result);
239        }
240    }
241
242    /**
243     * Print information about the available models
244     *
245     * @return void
246     */
247    protected function models()
248    {
249        $result = (new ModelFactory($this->conf))->getModels();
250
251        $td = new TableFormatter($this->colors);
252        $cols = [30, 20, 20, '*'];
253        echo "==== Chat Models ====\n\n";
254        echo $td->format(
255            $cols,
256            ['Model', 'Token Limits', 'Price USD/M', 'Description'],
257            [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
258        );
259        foreach ($result['chat'] as $name => $info) {
260            echo $td->format(
261                $cols,
262                [
263                    $name,
264                    sprintf(" In: %7d\nOut: %7d", $info['inputTokens'], $info['outputTokens']),
265                    sprintf(" In: %.2f\nOut: %.2f", $info['inputTokenPrice'], $info['outputTokenPrice']),
266                    $info['description'] . "\n"
267                ],
268                [
269                    $info['instance'] ? Colors::C_LIGHTGREEN : Colors::C_LIGHTRED,
270                ]
271            );
272        }
273
274        $cols = [30, 10, 10, 10, '*'];
275        echo "==== Embedding Models ====\n\n";
276        echo $td->format(
277            $cols,
278            ['Model', 'Token Limits', 'Price USD/M', 'Dimensions', 'Description'],
279            [Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE, Colors::C_LIGHTBLUE]
280        );
281        foreach ($result['embedding'] as $name => $info) {
282            echo $td->format(
283                $cols,
284                [
285                    $name,
286                    sprintf("%7d", $info['inputTokens']),
287                    sprintf("%.2f", $info['inputTokenPrice']),
288                    $info['dimensions'],
289                    $info['description'] . "\n"
290                ],
291                [
292                    $info['instance'] ? Colors::C_LIGHTGREEN : Colors::C_LIGHTRED,
293                ]
294            );
295        }
296
297        $this->colors->ptln('Current prices may differ', Colors::C_RED);
298    }
299
300    /**
301     * Handle a single, standalone question
302     *
303     * @param string $query
304     * @return void
305     * @throws Exception
306     */
307    protected function ask($query)
308    {
309        $result = $this->helper->askQuestion($query);
310        $this->printAnswer($result);
311    }
312
313    /**
314     * Get the pages that are similar to the query
315     *
316     * @param string $query
317     * @return void
318     */
319    protected function similar($query)
320    {
321        $langlimit = $this->helper->getLanguageLimit();
322        if ($langlimit) {
323            $this->info('Limiting results to {lang}', ['lang' => $langlimit]);
324        }
325
326        $sources = $this->helper->getEmbeddings()->getSimilarChunks($query, $langlimit);
327        $this->printSources($sources);
328    }
329
330    /**
331     * Run the maintenance tasks
332     *
333     * @return void
334     */
335    protected function runMaintenance()
336    {
337        $start = time();
338        $this->helper->getStorage()->runMaintenance();
339        $this->notice('Peak memory used: {memory}', ['memory' => filesize_h(memory_get_peak_usage(true))]);
340        $this->notice('Spent time: {time}min', ['time' => round((time() - $start) / 60, 2)]);
341
342        $data = $this->helper->getRunData();
343        $data['maintenance ran at'] = time();
344        $this->helper->setRunData($data);
345    }
346
347    /**
348     * Recreate chunks and embeddings for all pages
349     *
350     * @return void
351     */
352    protected function createEmbeddings($clear)
353    {
354        [$skipRE, $matchRE] = $this->getRegexps();
355
356        $start = time();
357        $this->helper->getEmbeddings()->createNewIndex($skipRE, $matchRE, $clear);
358        $this->notice('Peak memory used: {memory}', ['memory' => filesize_h(memory_get_peak_usage(true))]);
359        $this->notice('Spent time: {time}min', ['time' => round((time() - $start) / 60, 2)]);
360
361        $data = $this->helper->getRunData();
362        $data['embed ran at'] = time();
363        $this->helper->setRunData($data);
364    }
365
366    /**
367     * Dump TSV files for debugging
368     *
369     * @return void
370     */
371    protected function tsv($vector, $meta)
372    {
373
374        $storage = $this->helper->getStorage();
375        $storage->dumpTSV($vector, $meta);
376        $this->success('written to ' . $vector . ' and ' . $meta);
377    }
378
379    /**
380     * Print the given detailed answer in a nice way
381     *
382     * @param array $answer
383     * @return void
384     */
385    protected function printAnswer($answer)
386    {
387        $this->colors->ptln($answer['answer'], Colors::C_LIGHTCYAN);
388        echo "\n";
389        $this->printSources($answer['sources']);
390        echo "\n";
391        $this->printUsage();
392    }
393
394    /**
395     * Print the given sources
396     *
397     * @param Chunk[] $sources
398     * @return void
399     */
400    protected function printSources($sources)
401    {
402        foreach ($sources as $source) {
403            /** @var Chunk $source */
404            $this->colors->ptln(
405                "\t" . $source->getPage() . ' ' . $source->getId() . ' (' . $source->getScore() . ')',
406                Colors::C_LIGHTBLUE
407            );
408        }
409    }
410
411    /**
412     * Print the usage statistics for OpenAI
413     *
414     * @return void
415     */
416    protected function printUsage()
417    {
418        $chat = $this->helper->getChatModel()->getUsageStats();
419        $rephrase = $this->helper->getRephraseModel()->getUsageStats();
420        $embed = $this->helper->getEmbeddingModel()->getUsageStats();
421
422        $this->info(
423            'Made {requests} requests in {time}s to models. Used {tokens} tokens for about ${cost}.',
424            [
425                'requests' => $chat['requests'] + $rephrase['requests'] + $embed['requests'],
426                'time' => $chat['time'] + $rephrase['time'] + $embed['time'],
427                'tokens' => $chat['tokens'] + $chat['tokens'] + $embed['tokens'],
428                'cost' => $chat['cost'] + $chat['cost'] + $embed['cost'],
429            ]
430        );
431    }
432
433    /**
434     * Interactively ask for a value from the user
435     *
436     * @param string $prompt
437     * @return string
438     */
439    protected function readLine($prompt)
440    {
441        $value = '';
442
443        while ($value === '') {
444            echo $prompt;
445            echo ': ';
446
447            $fh = fopen('php://stdin', 'r');
448            $value = trim(fgets($fh));
449            fclose($fh);
450        }
451
452        return $value;
453    }
454
455    /**
456     * Read the skip and match regex from the config
457     *
458     * Ensures the regular expressions are valid
459     *
460     * @return string[] [$skipRE, $matchRE]
461     */
462    protected function getRegexps()
463    {
464        $skip = $this->getConf('skipRegex');
465        $skipRE = '';
466        $match = $this->getConf('matchRegex');
467        $matchRE = '';
468
469        if ($skip) {
470            $skipRE = '/' . $skip . '/';
471            if (@preg_match($skipRE, '') === false) {
472                $this->error(preg_last_error_msg());
473                $this->error('Invalid regular expression in $conf[\'skipRegex\']. Ignored.');
474                $skipRE = '';
475            } else {
476                $this->success('Skipping pages matching ' . $skipRE);
477            }
478        }
479
480        if ($match) {
481            $matchRE = '/' . $match . '/';
482            if (@preg_match($matchRE, '') === false) {
483                $this->error(preg_last_error_msg());
484                $this->error('Invalid regular expression in $conf[\'matchRegex\']. Ignored.');
485                $matchRE = '';
486            } else {
487                $this->success('Only indexing pages matching ' . $matchRE);
488            }
489        }
490        return [$skipRE, $matchRE];
491    }
492}
493