API

api/common/Service/PdfHandler.php

@@ -1,15 +1,15 @@
 <?php
 namespace Contabilidad\Common\Service;
 
+use Contabilidad\Common\Concept\DocumentHandler;
 use GuzzleHttp\Client;
 
-class PdfHandler {
+class PdfHandler extends DocumentHandler {
   protected Client $client;
-  protected string $folder;
   protected string $url;
   public function __construct(Client $client, string $pdf_folder, string $url) {
+    parent::__construct($pdf_folder);
     $this->client = $client;
-    $this->folder = $pdf_folder;
     $this->url = $url;
   }
 
@@ -25,6 +25,51 @@ class PdfHandler {
     }
     $response = $this->client->post($this->url, ['json' => ['files' => $output]]);
     $output = json_decode($response->getBody());
-    return $output;
+    return $this->build($output);
+  }
+  protected function build(array $data): ?array {
+    foreach ($data as &$file) {
+      $i = $this->findStartRow($file->text);
+      if ($i === -1) {
+        continue;
+      }
+      $e = $this->findEndRow($file->text, $i);
+      if ($e == $i) {
+        continue;
+      }
+      $file->data = array_filter($file->text, function($key) use ($i, $e) {
+        return ($key >= $i) and ($key <= $e);
+      }, ARRAY_FILTER_USE_KEY);
+    }
+    return $data;
+  }
+  protected function findStartRow(array $data): int {
+    foreach ($data as $i => $row) {
+      if (!is_array($row)) {
+        continue;
+      }
+      $maybe = false;
+      foreach ($row as $cell) {
+        if (str_contains($cell, '/')) {
+          $maybe = true;
+        }
+        if ($maybe and str_contains($cell, '$')) {
+          return $i - 1;
+        }
+      }
+    }
+    return -1;
+  }
+  protected function findEndRow(array $data, int $start): int {
+    $l = count($data[$start]);
+    for ($i = $start; $i < count($data); $i ++) {
+      if (!is_array($data[$i])) {
+        return $i - 1;
+      }
+      if (count($data[$i]) != $l) {
+        return $i - 1;
+      }
+    }
+    return $start;
   }
 }