Merge pull request #285 from biblioverse/ai-context

Refactor AI context handling and update

config/services.yaml

@@ -39,6 +39,9 @@ parameters:
     AI_CONTEXT_AMAZON_ENABLED: '%env(bool:AI_CONTEXT_AMAZON_ENABLED)%'
     env(AI_CONTEXT_AMAZON_ENABLED): false
 
+    AI_CONTEXT_FULL_EPUB: '%env(bool:AI_CONTEXT_FULL_EPUB)%'
+    env(AI_CONTEXT_FULL_EPUB): false
+
 
 services:
     # default configuration for services in *this* file

doc/src/content/docs/guides/Administrator/ai-features-configuration.mdx

@@ -91,16 +91,23 @@ On a book page, you can click on the "generate summary"  or "generate tags" butt
 ## Run it through your whole library
 You can run the following command:
 ```
-docker compose exec biblioteca bin/console books:tag
+docker compose exec biblioteca bin/console books:ai <tags|summary|both>
 ```
 It will tag all your books that currently don't have tags.
 
 If you want to use a user's configured prompts:
 
 ```
-docker compose exec biblioteca bin/console books:tag <userid>
+docker compose exec biblioteca bin/console books:ai <tags|summary|both> <userid>
 ```
 
+If you want to use it on a specific book:
+
+```
+docker compose exec biblioteca bin/console books:ai <tags|summary|both> -b <book-id>
+```
+
+
 ## Add context
 There are currently 2 ways to add context for better results. Both can be enabled at the same time.
 

doc/src/content/docs/guides/Administrator/commands.md

@@ -26,15 +26,9 @@ Relocate all books to their calculated folder. This is necessary only if you wan
 ## `books:scan`                              
 Will scan the `public/books` folder and add all books to the database. If a book already exists, it will be updated.
 
-## `books:tag`
+## `books:ai`
 
-> First you will need to setup an openAi chatgpt key in a user's properties and run the command as this user
-
-Then you can run the command to generate tag for all books in the library:
-
-```bash
-bin/console books:tag <user_id>
-```
+Check the documentation for this in the AI chapter
 
 ## `cache:clear`
 Clears the cache

src/Ai/CommunicatorDefiner.php

@@ -7,20 +7,9 @@
 class CommunicatorDefiner
 {
     public const BASE_PROMPT = "
-    As a highly skilled and experienced librarian AI model, I'm here to provide you with deep insights and practical recommendations 
-    based on my vast experience in the field of literature and knowledge organization. Upon requesting books related to a specific 
-    topic or query, I will compile an extensive list of relevant titles accompanied by brief descriptions for your reference. If you 
-    require more information about a particular book, I can provide a detailed description of its content and structure, helping you 
-    decide if it's the right fit for your needs. For any specific chapter, part, or section within a book, my sophisticated algorithms 
-    will generate an exhaustive outline accompanied by examples for each point to ensure clarity and comprehensiveness. 
-    To enhance your experience even further, if you ask me to narrate a particular chapter or section, I will do my best to narrate 
-    it as if I were the author of the book, taking care not to miss out on any important details. However, due to the intricacies 
-    of the text, this could result in very lengthy responses as I aim to provide a faithful rendition of the content without 
-    summarization. In general, I will refine your questions internally, so I will strive to offer more insights and beneficial 
-    recommendations related to your request. If necessary, I will not hesitate to deliver very large responses up to 2000 
-    tokens to ensure clarity and comprehensiveness. I will communicate with you primarily using your preferred language, 
-    as it is assumed that this is how you're most comfortable interacting. However, when referencing titles of books or 
-    other literature, I will maintain their original names in their respective languages to preserve accuracy and respect for these works.";
+    As a highly skilled and experienced librarian AI model, I'm here to help you tag and summarize books as close to the
+    original as possible. I will never make up any information. I will only use the information you provide me.
+    I will communicate with you primarily using your preferred language.";
 
     /**
      * @param iterable<AiCommunicatorInterface> $handlers

src/Ai/Context/ContextBuilder.php

@@ -18,15 +18,18 @@ public function __construct(
 
     public function getContext(AbstractBookPrompt $abstractBookPrompt): AbstractBookPrompt
     {
-        $prompt = "Use the following pieces of context to answer the question at the end. If you don't know the answer, don't try to make up an answer.";
-
+        $prompt = "Use the following pieces of context to answer the query at the end. If you don't know the answer, don't try to make up an answer. Context:
+---------------------
+";
         foreach ($this->handlers as $handler) {
             if ($handler->isEnabled()) {
                 $prompt .= $handler->getContextForPrompt($abstractBookPrompt->getBook());
             }
         }
-
-        $prompt .= $abstractBookPrompt->getPrompt();
+        $prompt .= '
+---------------------
+Given the context information and not prior knowledge, answer the query.
+Query: '.$abstractBookPrompt->getPrompt();
 
         $abstractBookPrompt->setPrompt($abstractBookPrompt->replaceBookOccurrence($prompt));
 

src/Ai/Context/EpubContextBuilder.php

@@ -0,0 +1,56 @@
+<?php
+
+namespace App\Ai\Context;
+
+use App\Entity\Book;
+use App\Service\BookFileSystemManagerInterface;
+use Kiwilan\Ebook\Ebook;
+use Kiwilan\Ebook\Formats\Epub\EpubModule;
+use Symfony\Component\DependencyInjection\Attribute\Autowire;
+
+class EpubContextBuilder implements ContextBuildingInteface
+{
+    public function __construct(private readonly BookFileSystemManagerInterface $bookFileSystemManager, #[Autowire(param: 'AI_CONTEXT_FULL_EPUB')] private readonly bool $enable)
+    {
+    }
+
+    #[\Override]
+    public function isEnabled(): bool
+    {
+        return $this->enable;
+    }
+
+    #[\Override]
+    public function getContextForPrompt(Book $book): string
+    {
+        if ($book->getExtension() !== 'epub') {
+            return '';
+        }
+        $prompt = '';
+
+        $bookFile = $this->bookFileSystemManager->getBookFile($book);
+
+        $ebook = Ebook::read($bookFile->getPathname());
+
+        if (!$ebook instanceof Ebook) {
+            return '';
+        }
+
+        $epub = $ebook->getParser()?->getEpub();
+
+        if (!$epub instanceof EpubModule) {
+            return '';
+        }
+
+        $htmlArray = $epub->getHtml();
+
+        foreach ($htmlArray as $html) {
+            $text = $html->getBody();
+            $prompt .= strip_tags($text ?? '');
+        }
+
+        $prompt = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", '', $prompt);
+
+        return preg_replace("/([\n]|[\r])+/", ' ', (string) $prompt) ?? '';
+    }
+}

src/Ai/OllamaCommunicator.php

@@ -65,6 +65,7 @@ private function getOllamaUrl(string $path): string
     #[\Override]
     public function initialise(string $basePrompt): void
     {
+        $this->basePrompt = $basePrompt;
         $this->sendRequest($this->getOllamaUrl('pull'), [
             'model' => $this->model,
         ], 'POST');
@@ -73,14 +74,16 @@ public function initialise(string $basePrompt): void
     #[\Override]
     public function interrogate(BookPromptInterface $prompt): string|array
     {
-        $response = $this->sendRequest($this->getOllamaUrl('generate'), [
+        $params = [
             'model' => $this->model,
             'prompt' => $prompt->getPrompt(),
             'system' => $this->basePrompt,
             'options' => [
-                'temperature' => 0.3,
+                'temperature' => 0,
             ],
-        ], 'POST');
+        ];
+
+        $response = $this->sendRequest($this->getOllamaUrl('generate'), $params, 'POST');
 
         return $prompt->convertResult($response);
     }

src/Ai/Prompt/AbstractBookPrompt.php

@@ -33,7 +33,12 @@ public function getBook(): Book
 
     public function replaceBookOccurrence(string $prompt): string
     {
-        $bookString = '"'.$this->book->getTitle().'" by '.implode(' and ', $this->book->getAuthors());
+        $title = $this->book->getTitle();
+        if (preg_match('/T\d+/', $title) !== false) {
+            $bookString = '"'.$this->book->getTitle().'" by '.implode(' and ', $this->book->getAuthors());
+        } else {
+            $bookString = ' by '.implode(' and ', $this->book->getAuthors());
+        }
 
         if ($this->book->getSerie() !== null) {
             $bookString .= ' number '.$this->book->getSerieIndex().' in the series "'.$this->book->getSerie().'"';

src/Command/BooksTagCommand.php → src/Command/BooksAiCommand.php

@@ -5,6 +5,7 @@
 use App\Ai\AiCommunicatorInterface;
 use App\Ai\CommunicatorDefiner;
 use App\Ai\Context\ContextBuilder;
+use App\Ai\Prompt\SummaryPrompt;
 use App\Ai\Prompt\TagPrompt;
 use App\Entity\Book;
 use App\Entity\User;
@@ -18,10 +19,10 @@
 use Symfony\Component\Console\Style\SymfonyStyle;
 
 #[AsCommand(
-    name: 'books:tag',
+    name: 'books:ai',
     description: 'Add a short description for your command',
 )]
-class BooksTagCommand extends Command
+class BooksAiCommand extends Command
 {
     public function __construct(
         private readonly EntityManagerInterface $em,
@@ -35,24 +36,34 @@ public function __construct(
     protected function configure(): void
     {
         $this
+            ->addArgument('type', InputArgument::REQUIRED, '`summary`, `tags` or `both`')
             ->addArgument('userid', InputArgument::OPTIONAL, 'user for the prompts. Default prompts used if not provided')
+            ->addOption('book', 'b', InputArgument::OPTIONAL, 'book id to process (otherwise all books are processed)')
         ;
     }
 
     #[\Override]
     protected function execute(InputInterface $input, OutputInterface $output): int
     {
         $io = new SymfonyStyle($input, $output);
-        $arg1 = $input->getArgument('userid');
+        $userId = $input->getArgument('userid');
+
+        $type = $input->getArgument('type');
+        if (!in_array($type, ['summary', 'tags', 'both'], true)) {
+            $io->error('Invalid type');
+
+            return Command::FAILURE;
+        }
+        $bookId = $input->getOption('book');
 
         $user = null;
-        if ($arg1 !== null) {
-            $user = $this->em->getRepository(User::class)->find($arg1);
+        if ($userId !== null) {
+            $user = $this->em->getRepository(User::class)->find($userId);
             if (!$user instanceof User) {
                 $io->error('User not found');
-            }
 
-            return Command::FAILURE;
+                return Command::FAILURE;
+            }
         }
 
         $communicator = $this->aiCommunicator->getCommunicator();
@@ -63,9 +74,22 @@ protected function execute(InputInterface $input, OutputInterface $output): int
             return Command::FAILURE;
         }
 
-        $qb = $this->em->getRepository(Book::class)->createQueryBuilder('book');
-        $qb->andWhere('book.tags = \'[]\'');
-        $books = $qb->getQuery()->getResult();
+        $io->title('AI data with '.$communicator::class);
+
+        if ($bookId === null) {
+            $io->note('Processing all books without tags or summary');
+            $qb = $this->em->getRepository(Book::class)->createQueryBuilder('book');
+            if ($type === 'tags' || $type === 'both') {
+                $qb->orWhere('book.tags = \'[]\'');
+            }
+            if ($type === 'summary' || $type === 'both') {
+                $qb->orWhere('book.summary is null');
+            }
+            $books = $qb->getQuery()->getResult();
+        } else {
+            $io->note('Processing book '.$bookId);
+            $books = $this->em->getRepository(Book::class)->findBy(['id' => $bookId]);
+        }
 
         if (!is_array($books)) {
             $io->error('Failed to get books');
@@ -79,16 +103,28 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         foreach ($books as $book) {
             $progress->setMessage($book->getSerie().' '.$book->getTitle().' ('.implode(' and ', $book->getAuthors()).')');
             $progress->advance();
-            $tagPrompt = new TagPrompt($book, $user);
-            $tagPrompt = $this->contextBuilder->getContext($tagPrompt);
-            $array = $communicator->interrogate($tagPrompt);
 
-            if (is_array($array)) {
-                $io->writeln('🏷️ '.implode(' 🏷️ ', $array));
-                $book->setTags($array);
+            if ($type === 'summary' || $type === 'both') {
+                $summaryPrompt = new SummaryPrompt($book, $user);
+                $summaryPrompt = $this->contextBuilder->getContext($summaryPrompt);
+                $summary = $communicator->interrogate($summaryPrompt);
+                $io->block($summary, padding: true);
+                $book->setSummary($summary);
+            }
+
+            if ($type === 'tags' || $type === 'both') {
+                $tagPrompt = new TagPrompt($book, $user);
+                $tagPrompt = $this->contextBuilder->getContext($tagPrompt);
+
+                $array = $communicator->interrogate($tagPrompt);
+
+                if (is_array($array)) {
+                    $io->block(implode(' 🏷️ ', $array), padding: true);
+                    $book->setTags($array);
+                }
             }
 
-            $this->em->flush();
+            // $this->em->flush();
         }
 
         $progress->finish();

src/Controller/ConfigurationController.php

@@ -33,6 +33,7 @@ public function index(ParameterBagInterface $parameterBagInterface): Response
             'OLLAMA_URL' => 'Url of the Ollama API. do not forget the trailing slash. Example: http://ollama:11434/api/',
             'OLLAMA_MODEL' => 'Ollama model to use',
             'AI_CONTEXT_AMAZON_ENABLED' => 'Do you want to give context from Amazon to AI completions? This will scrape the Amazon website. Use with caution.',
+            'AI_CONTEXT_FULL_EPUB' => 'Do you want to give the full epub context to AI completions? This will cause the prompt to be very long and could increase costs when using a paid model.',
             'WIKIPEDIA_API_TOKEN' => 'Wikipedia API token. Required if you want to give context from Wikipedia to AI. You will need to generate a personal api token in wikipedia.org',
         ];
         foreach ($documentedParams as $key => $value) {

src/Twig/Components/AiSuggestion.php

@@ -84,10 +84,10 @@ public function generate(): void
             default => throw new \InvalidArgumentException('Invalid field'),
         };
 
-        $promptObj = $this->contextBuilder->getContext($promptObj);
-
         $promptObj->setPrompt($this->prompt);
 
+        $promptObj = $this->contextBuilder->getContext($promptObj);
+
         $result = $communicator->interrogate($promptObj);
 
         $this->result = is_array($result) ? $result : [$result];