Lots of updates when using ollama to batch the jobs so they can wait …

…else the jobs stack up and timeout

Signed-off-by: alnutile <[email protected]>

Modules/LlmDriver/app/BaseClient.php

@@ -107,6 +107,11 @@ public function isAsync(): bool
         return true;
     }
 
+    public function hasFunctions(): bool
+    {
+        return count($this->getFunctions()) > 0;
+    }
+
     public function getFunctions(): array
     {
         $functions = LlmDriverFacade::getFunctions();
@@ -157,4 +162,9 @@ protected function remapMessages(array $messages): array
     {
         return $messages;
     }
+
+    public function onQueue(): string
+    {
+        return 'default';
+    }
 }

Modules/LlmDriver/app/OllamaClient.php

@@ -172,6 +172,10 @@ public function getFunctions(): array
     {
         $functions = LlmDriverFacade::getFunctions();
 
+        if (! Feature::activate('ollama-functions')) {
+            return [];
+        }
+
         return collect($functions)->map(function ($function) {
             $function = $function->toArray();
             $properties = [];
@@ -205,4 +209,9 @@ public function isAsync(): bool
     {
         return false;
     }
+
+    public function onQueue(): string
+    {
+        return 'ollama';
+    }
 }

Modules/LlmDriver/app/Orchestrate.php

@@ -125,6 +125,7 @@ public function handle(array $messagesArray, Chat $chat): ?string
 
             return $this->response;
         } else {
+            Log::info('[LaraChain] Orchestration No Fucntions SearchAnd Summarize');
             /**
              * @NOTE
              * this assumes way too much
@@ -135,6 +136,10 @@ function ($message) {
                 }
             )->content;
 
+            Log::info('[LaraChain] Orchestration No Fucntions SearchAnd Summarize', [
+                'message' => $message,
+            ]);
+
             return SearchOrSummarizeChatRepo::search($chat, $message);
         }
     }

Modules/LlmDriver/app/SimpleSearchAndSummarizeOrchestrate.php

@@ -0,0 +1,35 @@
+<?php
+
+namespace LlmLaraHub\LlmDriver;
+
+use App\Events\ChatUiUpdateEvent;
+use App\Models\Chat;
+use Facades\App\Domains\Messages\SearchOrSummarizeChatRepo;
+use Illuminate\Support\Facades\Log;
+
+class SimpleSearchAndSummarizeOrchestrate
+{
+    protected string $response = '';
+
+    protected bool $requiresFollowup = false;
+
+    public function handle(string $message, Chat $chat): ?string
+    {
+        Log::info('[LaraChain] Skipping over functions doing search and summarize');
+
+        ChatUiUpdateEvent::dispatch(
+            $chat->chatable,
+            $chat,
+            'Searching data now to summarize content'
+        );
+
+        $response = SearchOrSummarizeChatRepo::search($chat, $message);
+
+        return $response;
+    }
+
+    protected function hasFunctions(array $functions): bool
+    {
+        return is_array($functions) && count($functions) > 0;
+    }
+}

app/Domains/Documents/Transformers/PdfTransformer.php

@@ -12,6 +12,9 @@
 use Illuminate\Bus\Batch;
 use Illuminate\Support\Facades\Bus;
 use Illuminate\Support\Facades\Log;
+use LlmLaraHub\LlmDriver\LlmDriverFacade;
+use LlmLaraHub\TagFunction\Database\Seeders\TagFunctionDatabaseSeeder;
+use LlmLaraHub\TagFunction\Functions\TaggingFunction;
 use Smalot\PdfParser\Parser;
 
 class PdfTransformer
@@ -50,7 +53,7 @@ public function handle(Document $document): Document
                 $chunks[] = [
                     new VectorlizeDataJob($DocumentChunk),
                     new SummarizeDataJob($DocumentChunk),
-                    //new TagDataJob($this->document),
+                    //new TaggingFunction($this->document),
                     //then mark it all as done and notify the ui
                 ];
 
@@ -66,6 +69,7 @@ public function handle(Document $document): Document
                 SummarizeDocumentJob::dispatch($document);
             })
             ->allowFailures()
+            ->onQueue(LlmDriverFacade::driver($document->getDriver())->onQueue())
             ->dispatch();
 
         return $this->document;

app/Domains/Documents/Transformers/PowerPointTransformer.php

@@ -0,0 +1,71 @@
+<?php
+
+namespace App\Domains\Documents\Transformers;
+
+use App\Domains\Collections\CollectionStatusEnum;
+use App\Events\CollectionStatusEvent;
+use App\Jobs\SummarizeDataJob;
+use App\Jobs\VectorlizeDataJob;
+use App\Models\Document;
+use App\Models\DocumentChunk;
+use Illuminate\Support\Facades\Log;
+use PhpOffice\PhpPresentation\IOFactory;
+use PhpOffice\PhpPresentation\Shape\RichText;
+
+class PowerPointTransformer
+{
+    protected Document $document;
+
+    public function handle(Document $document): array
+    {
+        $this->document = $document;
+
+        $filePath = $this->document->pathToFile();
+
+        $parser = IOFactory::createReader('PowerPoint2007');
+        if (! $parser->canRead($filePath)) {
+            throw new \Exception('Can not read the document '.$filePath);
+        }
+
+        $oPHPPresentation = $parser->load($filePath);
+
+        $chunks = [];
+        foreach ($oPHPPresentation->getAllSlides() as $page_number => $page) {
+            try {
+                foreach ($page->getShapeCollection() as $shape) {
+                    // Check if shape contains text
+                    if ($shape instanceof RichText) {
+                        // Get the text from the shape
+                        $page_number = $page_number + 1;
+                        $pageContent = $shape->getPlainText();
+                        $guid = $filePath.'_'.$page_number;
+                        $DocumentChunk = DocumentChunk::updateOrCreate(
+                            [
+                                'guid' => $guid,
+                                'document_id' => $this->document->id,
+                            ],
+                            [
+                                'content' => $pageContent,
+                                'sort_order' => $page_number,
+                            ]
+                        );
+
+                        $chunks[] = [
+                            new VectorlizeDataJob($DocumentChunk),
+                            new SummarizeDataJob($DocumentChunk),
+                            //new TagDataJob($this->document),
+                            //then mark it all as done and notify the ui
+                        ];
+                    }
+                }
+                CollectionStatusEvent::dispatch($document->collection, CollectionStatusEnum::PROCESSING);
+            } catch (\Exception $e) {
+                Log::error('Error parsing PDF', ['error' => $e->getMessage()]);
+            }
+        }
+
+        Log::info('PowerPointTransformer:handle', ['chunks' => count($chunks)]);
+
+        return $chunks;
+    }
+}

app/Domains/Messages/SearchOrSummarizeChatRepo.php

@@ -5,6 +5,7 @@
 use App\Models\Chat;
 use App\Models\DocumentChunk;
 use Illuminate\Support\Facades\Log;
+use Laravel\Pennant\Feature;
 use LlmLaraHub\LlmDriver\LlmDriverFacade;
 use LlmLaraHub\LlmDriver\Responses\CompletionResponse;
 use LlmLaraHub\LlmDriver\Responses\EmbeddingsResponseDto;
@@ -19,7 +20,7 @@ public function search(Chat $chat, string $input): string
          * by searching the data or a summary of the data.
          * For now we will search.
          */
-        Log::info('ChatController:chat getting embedding', ['input' => $input]);
+        Log::info('[LaraChain] Embedding and Searching');
 
         /** @var EmbeddingsResponseDto $embedding */
         $embedding = LlmDriverFacade::driver(
@@ -42,12 +43,16 @@ public function search(Chat $chat, string $input): string
         $content = [];
 
         foreach ($results as $result) {
-            $content[] = remove_ascii(reduce_text_size($result->content)); //reduce_text_size seem to mess up Claude?
+            $contentString = remove_ascii($result->content);
+            if (Feature::active('reduce_text')) {
+                $result = reduce_text_size($contentString);
+            }
+            $content[] = $contentString; //reduce_text_size seem to mess up Claude?
         }
 
         $content = implode(' ', $content);
 
-        $content = "This is data from the search results when entering the users prompt which is ### START PROMPT ### {$input} ### END PROMPT ### please use this with the following context and only this and return as markdown so I can render it: ".$content;
+        $content = "This is data from the search results when entering the users prompt which is ### START PROMPT ### {$input} ### END PROMPT ###  please use this with the following context and only this, summarize it for the user and return as markdown so I can render it and strip out and formatting like extra spaces, tabs, periods etc: ".$content;
 
         $chat->addInput(
             message: $content,
@@ -57,7 +62,7 @@ public function search(Chat $chat, string $input): string
         );
 
         $latestMessagesArray = $chat->getChatResponse();
-
+        Log::info('[LaraChain] Getting the Summary');
         /** @var CompletionResponse $response */
         $response = LlmDriverFacade::driver(
             $chat->chatable->getDriver()

app/Http/Controllers/ChatController.php

@@ -10,6 +10,9 @@
 use App\Models\Chat;
 use App\Models\Collection;
 use Facades\LlmLaraHub\LlmDriver\Orchestrate;
+use Facades\LlmLaraHub\LlmDriver\SimpleSearchAndSummarizeOrchestrate;
+use Illuminate\Support\Facades\Log;
+use LlmLaraHub\LlmDriver\LlmDriverFacade;
 use LlmLaraHub\LlmDriver\Requests\MessageInDto;
 
 class ChatController extends Controller
@@ -44,6 +47,7 @@ public function chat(Chat $chat)
     {
         $validated = request()->validate([
             'input' => 'required|string',
+            'completion' => 'boolean'
         ]);
 
         $chat->addInput(
@@ -58,7 +62,22 @@ public function chat(Chat $chat)
             'role' => 'user',
         ]);
 
-        $response = Orchestrate::handle($messagesArray, $chat);
+        if(data_get($validated, 'completion', false)) {
+            Log::info('[LaraChain] Running Simple Completion');
+            $prompt = $validated['input'];
+            $response = LlmDriverFacade::driver($chat->getDriver())->completion($prompt);
+            $response = $response->content;
+            $chat->addInput(
+                message: $response,
+                role: RoleEnum::Assistant,
+                show_in_thread: true);
+        } elseif (LlmDriverFacade::driver($chat->getDriver())->hasFunctions()) {
+            Log::info('[LaraChain] Running Orchestrate');
+            $response = Orchestrate::handle($messagesArray, $chat);
+        } else {
+            Log::info('[LaraChain] Simple Search and Summarize');
+            $response = SimpleSearchAndSummarizeOrchestrate::handle($validated['input'], $chat);
+        }
 
         ChatUpdatedEvent::dispatch($chat->chatable, $chat);
 

app/Http/Controllers/CollectionController.php

@@ -10,6 +10,7 @@
 use App\Jobs\ProcessFileJob;
 use App\Models\Collection;
 use App\Models\Document;
+use Illuminate\Support\Facades\Log;
 
 class CollectionController extends Controller
 {
@@ -92,11 +93,24 @@ public function filesUpload(Collection $collection)
         ]);
 
         foreach ($validated['files'] as $file) {
-            $document = Document::create([
-                'collection_id' => $collection->id,
-                'file_path' => $file->getClientOriginalName(),
-                'type' => TypesEnum::PDF,
-            ]);
+            $mimetype = $file->getMimeType();
+
+            //if pptx
+            Log::info($mimetype);
+
+            if ($mimetype === 'application/vnd.openxmlformats-officedocument.presentationml.presentation') {
+                $document = Document::create([
+                    'collection_id' => $collection->id,
+                    'file_path' => $file->getClientOriginalName(),
+                    'type' => TypesEnum::Pptx,
+                ]);
+            } else {
+                $document = Document::create([
+                    'collection_id' => $collection->id,
+                    'file_path' => $file->getClientOriginalName(),
+                    'type' => TypesEnum::PDF,
+                ]);
+            }
 
             $file->storeAs(
                 path: $collection->id,

app/Jobs/ParsePowerPointJob.php

@@ -0,0 +1,44 @@
+<?php
+
+namespace App\Jobs;
+
+use App\Models\Document;
+use Facades\App\Domains\Documents\Transformers\PowerPointTransformer;
+use Illuminate\Bus\Batchable;
+use Illuminate\Bus\Queueable;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Foundation\Bus\Dispatchable;
+use Illuminate\Queue\InteractsWithQueue;
+use Illuminate\Queue\SerializesModels;
+
+class ParsePowerPointJob implements ShouldQueue
+{
+    use Batchable, Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
+
+    /**
+     * Create a new job instance.
+     */
+    public function __construct(public Document $document)
+    {
+        //
+    }
+
+    /**
+     * Execute the job.
+     */
+    public function handle(): void
+    {
+        if ($this->batch()->cancelled()) {
+            // Determine if the batch has been cancelled...
+
+            return;
+        }
+
+        $chunks = PowerPointTransformer::handle($this->document);
+
+        foreach ($chunks as $chunk) {
+            $this->batch()->add($chunk);
+        }
+
+    }
+}