Merge pull request #40 from LlmLaraHub/source_prompting

Source prompting
LlmLaraHub · Jul 24, 2024 · 54aea8e · 54aea8e
2 parents ce275fe + e75649f
commit 54aea8e
Show file tree

Hide file tree

Showing 74 changed files with 8,043 additions and 792 deletions.
diff --git a/app/Domains/Chat/MetaDataDto.php b/app/Domains/Chat/MetaDataDto.php
@@ -15,6 +15,7 @@ public function __construct(
         public mixed $date_range = '',
         public mixed $input = '',
         public mixed $driver = '',
+        public mixed $source = '',
         public mixed $reference_collection_id = '',
     ) {
 

diff --git a/app/Domains/Documents/Transformers/CSVTransformer.php b/app/Domains/Documents/Transformers/CSVTransformer.php
@@ -28,7 +28,6 @@ public function handle(Document $document): array
 
         $filePath = $this->document->pathToFile();
 
-        //$filePath = null, string $disk = null, string $readerType = null
         $collection = (new DocumentsImport())
             ->toCollection($filePath, null, $this->readerType);
 

diff --git a/app/Domains/EmailParser/Client.php b/app/Domains/EmailParser/Client.php
@@ -18,7 +18,7 @@ class Client
         'Drafts',
     ];
 
-    public function handle(): void
+    public function handle(int $limit = 10): void
     {
         $mail = [];
 
@@ -31,7 +31,7 @@ public function handle(): void
 
             $full_name = data_get($folder, 'full_name');
             if (! in_array($full_name, $this->ignore)) {
-                $messages = $folder->messages()->all()->limit(10, 0)->get();
+                $messages = $folder->messages()->all()->limit($limit, 0)->get();
 
                 logger('[LaraChain] - Email Count', [
                     'count' => $messages->count(),
@@ -40,6 +40,12 @@ public function handle(): void
 
                 /** @var Message $message */
                 foreach ($messages as $message) {
+                    //@NOTE the Seen flag made it too hard to
+                    // then have different sources
+                    // check the same email box.
+                    // the Source will track repeats
+                    //$flags = $message->getFlags();
+
                     $messageDto = MailDto::from([
                         'to' => $message->getTo()->toString(),
                         'from' => $message->getFrom()->toString(),
@@ -63,7 +69,7 @@ public function handle(): void
                             'slug' => $slug,
                         ]);
                         $mail[] = new MailBoxParserJob($messageDto);
-                        $message->delete(expunge: true);
+                        $message->addFlag('Seen');
                     } else {
                         \Illuminate\Support\Facades\Log::info('Did not find Source with Slug To', [
                             'to' => $message->getTo()->toString(),

diff --git a/app/Domains/EmailParser/EmailClient.php b/app/Domains/EmailParser/EmailClient.php
@@ -35,7 +35,9 @@ public function setConfig(array $config): self
      * @throws \Webklex\PHPIMAP\Exceptions\MessageNotFoundException
      * @throws \Webklex\PHPIMAP\Exceptions\RuntimeException
      */
-    public function handle(CredentialsDto $credentials, bool $delete = true): array
+    public function handle(CredentialsDto $credentials,
+        bool $delete = false,
+        int $limit = 10): array
     {
         $mail = [];
 
@@ -63,6 +65,7 @@ public function handle(CredentialsDto $credentials, bool $delete = true): array
         $client = EmailClientFacade::setConfig($config);
 
         try {
+
             $client->connect();
 
             Log::info('Connected to email box', [
@@ -88,7 +91,7 @@ public function handle(CredentialsDto $credentials, bool $delete = true): array
                         'folders_to_check' => $foldersToCheck,
                     ]);
 
-                    $messages = $folder->messages()->all()->get();
+                    $messages = $folder->messages()->all()->limit($limit, 0)->get();
 
                     Log::info('[LaraChain] - Email Box Count', [
                         'count' => $messages->count(),
@@ -97,30 +100,25 @@ public function handle(CredentialsDto $credentials, bool $delete = true): array
 
                     /** @var Message $message */
                     foreach ($messages as $message) {
-                        $flags = $message->getFlags();
-
-                        if (! $flags->contains('Seen')) {
-                            $messageDto = MailDto::from([
-                                'to' => $message->getTo()->toString(),
-                                'from' => $message->getFrom()->toString(),
-                                'body' => $message->getTextBody(),
-                                'subject' => $message->getSubject(),
-                                'date' => $message->getDate()->toString(),
-                                'header' => $message->getHeader()->raw,
-                            ]);
-
-                            $mail[] = $messageDto;
-
-                            if ($delete) {
-                                $message->delete(expunge: true);
-                            } else {
-                                $message->addFlag('Seen');
-                            }
-                        } else {
-                            Log::info('[LaraChain] - Flag Seen', [
-                                'flags' => $flags->toArray(),
-                            ]);
-                        }
+                        //@NOTE the Seen flag made it too hard to
+                        // then have different sources
+                        // check the same email box.
+                        // the Source will track repeats
+                        //$flags = $message->getFlags();
+
+                        $messageDto = MailDto::from([
+                            'to' => $message->getTo()->toString(),
+                            'from' => $message->getFrom()->toString(),
+                            'body' => $message->getTextBody(),
+                            'subject' => $message->getSubject(),
+                            'date' => $message->getDate()->toString(),
+                            'header' => $message->getHeader()->raw,
+                            'email_message' => $message,
+                        ]);
+
+                        $mail[] = $messageDto;
+
+                        $message->addFlag('Seen');
                     }
 
                 }

diff --git a/app/Domains/EmailParser/MailDto.php b/app/Domains/EmailParser/MailDto.php
@@ -3,6 +3,7 @@
 namespace App\Domains\EmailParser;
 
 use Spatie\LaravelData\Data;
+use Webklex\PHPIMAP\Message;
 
 class MailDto extends Data
 {
@@ -12,7 +13,8 @@ public function __construct(
         public ?string $to,
         public ?string $body,
         public ?string $header,
-        public ?string $date
+        public ?string $date,
+        public ?Message $email_message = null
     ) {
     }
 

diff --git a/app/Domains/Prompts/EventPagePrompt.php b/app/Domains/Prompts/EventPagePrompt.php
@@ -0,0 +1,62 @@
+<?php
+
+namespace App\Domains\Prompts;
+
+use Illuminate\Support\Facades\Log;
+
+class EventPagePrompt
+{
+    public static function prompt(string $context): string
+    {
+        Log::info('[LaraChain] - EventPagePrompt');
+
+        return <<<PROMPT
+<ROLE>
+You are an AI assistant tasked with extracting event data from website content.
+
+<INSTRUCTIONS>
+1. Analyze the provided website HTML content below the <CONTENT> tag.
+2. Look for information about events within the content.
+3. If no event data is found, respond with a single word: false
+4. If event data is found, extract the following information for each event:
+   - Event Title
+   - Start Date
+   - End Date
+   - Location
+   - Description
+   - Any other relevant data
+5. Format the extracted data as a JSON array according to the specifications below.
+
+<OUTPUT_FORMAT>
+If events are found, return a JSON array with the following structure:
+
+[
+  {
+    "title": "Event Title",
+    "startDate": "Start Date",
+    "endDate": "End Date",
+    "location": "Location",
+    "description": "Description",
+    "additionalInfo": "Any other relevant data"
+  },
+  {
+    "title": "Event Title",
+    "startDate": "Start Date",
+    "endDate": "End Date",
+    "location": "Location",
+    "description": "Description",
+    "additionalInfo": "Any other relevant data"
+  }
+]
+
+If no events are found, return an empty JSON array: []
+
+<CONTENT>
+$context
+</CONTENT>
+
+Respond only with the JSON array or 'false' if no events are found. Do not include any explanations or additional text in your response.
+
+PROMPT;
+    }
+}
diff --git a/app/Domains/Prompts/SpecificTopic.php b/app/Domains/Prompts/SpecificTopic.php
@@ -0,0 +1,37 @@
+<?php
+
+namespace App\Domains\Prompts;
+
+use Illuminate\Support\Facades\Log;
+
+class SpecificTopic
+{
+    public static function prompt(string $context): string
+    {
+
+        Log::info('[LaraChain] - SpecificTopic');
+
+        return <<<'PROMPT'
+<ROLE>
+You are an email reading assistant who will follow the prompts to help parse my email box. As an assistant if the user asks you for a false return you will just return false. NOTHING MORE
+
+<TASKS>
+If the email content passed in is about Web Application work the frame work then keep and and summarize it. Else if it is about anything else just return the word false and only the word false. Please IGNORE Spam emails or Subjects that are about web applications but then the body is SPAM
+
+<Format>
+On a non false response, Summary and original message as Markdown.
+On a false response just the word false,
+
+<EXAMPLE>
+I would like to hire you to build an awesome application for me with DailyAi
+"You have an email from Teddy asking you to use DailAi to automate his business.
+
+I would like to sell you property in Alaska
+False
+<END EXAMPLES>
+
+<CONTENT>
+[CONTEXT]
+PROMPT;
+    }
+}
diff --git a/app/Domains/Sources/BaseSource.php b/app/Domains/Sources/BaseSource.php
@@ -4,6 +4,7 @@
 
 use App\Domains\Prompts\PromptMerge;
 use App\Domains\UnStructured\StructuredTypeEnum;
+use App\Helpers\ChatHelperTrait;
 use App\Jobs\DocumentProcessingCompleteJob;
 use App\Jobs\SummarizeDocumentJob;
 use App\Jobs\VectorlizeDataJob;
@@ -17,12 +18,17 @@
 use Illuminate\Support\Facades\Bus;
 use Illuminate\Support\Facades\Log;
 use LlmLaraHub\LlmDriver\LlmDriverFacade;
+use LlmLaraHub\LlmDriver\ToolsHelper;
 use LlmLaraHub\TagFunction\Jobs\TagDocumentJob;
 
 abstract class BaseSource
 {
+    use ChatHelperTrait, ToolsHelper;
+
     public string $batchTitle = 'Chunking Source';
 
+    public bool $promptPower = true;
+
     public static string $description = 'Sources are ways we get data into the system. They are the core of the system.';
 
     public ?Document $document = null;
@@ -215,4 +221,17 @@ protected function getEmailSummary(Document $document): string
 
         return $content;
     }
+
+    public function getSourceFromSlug(string $slug): ?Source
+    {
+        $source = Source::where('type', $this->sourceTypeEnum)
+            ->slug($slug)
+            ->first();
+
+        if ($source) {
+            return $source;
+        }
+
+        return null;
+    }
 }
diff --git a/app/Domains/Sources/EmailBoxSource.php b/app/Domains/Sources/EmailBoxSource.php
@@ -5,8 +5,7 @@
 use App\Domains\EmailParser\CredentialsDto;
 use App\Models\Source;
 use Facades\App\Domains\EmailParser\EmailClient;
-use Facades\App\Domains\Transformers\EmailTransformer;
-use Illuminate\Support\Facades\Log;
+use Facades\App\Domains\Sources\EmailSource as EmailSourceFacade;
 
 class EmailBoxSource extends EmailSource
 {
@@ -22,43 +21,7 @@ public function handle(Source $source): void
         $this->source = $source;
 
         foreach ($mails as $mailDto) {
-            $this->mailDto = $mailDto;
-
-            $this->content = $this->mailDto->getContent();
-
-            $this->documentSubject = $this->mailDto->subject;
-
-            $this->meta_data = $this->mailDto->toArray();
-
-            $this->transformers = $source->transformers;
-
-            Log::info('[LaraChain] - Running Email Source');
-
-            try {
-                Log::info('Do something!');
-                $baseSource = EmailTransformer::transform(baseSource: $this);
-                foreach ($source->transformers as $transformerChainLink) {
-                    Log::info("[LaraChain] - Source has Transformers let's figure out which one to run", [
-                        'type' => $transformerChainLink->type->name,
-                    ]);
-
-                    $class = '\\App\\Domains\\Transformers\\'.$transformerChainLink->type->name;
-                    if (class_exists($class)) {
-                        $facade = '\\Facades\\App\\Domains\\Transformers\\'.$transformerChainLink->type->name;
-                        $baseSource = $facade::transform($this);
-                    } else {
-                        Log::info('[LaraChain] - No Class found ', [
-                            'class' => $class,
-                        ]);
-                    }
-                }
-
-                $this->batchTransformedSource($baseSource, $source);
-            } catch (\Exception $e) {
-                Log::error('[LaraChain] - Error running Email Source', [
-                    'error' => $e->getMessage(),
-                ]);
-            }
+            EmailSourceFacade::setMailDto($mailDto)->handle($source);
         }
 
     }