nur-sery/src/ext/spreadsheet/wsutils.php

<?php
namespace nur\sery\ext\spreadsheet;

use nur\sery\ValueException;
use PhpOffice\PhpSpreadsheet\Cell\Coordinate;
use PhpOffice\PhpSpreadsheet\Spreadsheet;
use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet;

class wsutils {
  static function get_ws(?string $wsname, Spreadsheet $ss, bool $create=false): ?Worksheet {
    if ($wsname == null) {
      $ws = $ss->getActiveSheet();
    } elseif (is_numeric($wsname)) {
      $sheetCount = $ss->getSheetCount();
      if ($wsname < 1 || $wsname > $sheetCount) {
        throw ValueException::invalid_value($wsname, "sheet index");
      }
      $ws = $ss->getSheet($wsname - 1);
    } else {
      $ws = $ss->getSheetByName($wsname);
      if ($ws === null) {
        if ($create) $ws = $ss->createSheet()->setTitle($wsname);
        else throw ValueException::invalid_value($wsname, "sheet name");
      }
    }
    return $ws;
  }

  static function get_highest_coords(Worksheet $ws): array {
    $highestColumnA = $ws->getHighestColumn();
    $highestCol = Coordinate::columnIndexFromString($highestColumnA);
    $highestRow = $ws->getHighestRow();
    return [$highestCol, $highestRow];
  }

  /**
   * @var int nombre de colonnes/lignes au bout desquels on arrête de chercher
   * si on n'a trouvé que des cellules vides.
   *
   * c'est nécessaire à cause de certains fichiers provenant d'Excel que j'ai
   * reçus qui ont jusqu'à 10000 colonne vides et/ou 1048576 lignes vides. un
   * algorithme "bête" perd énormément de temps à chercher dans le vide, donnant
   * l'impression que le processus a planté.
   */
  const MAX_EMPTY_THRESHOLD = 150;

  static function compute_max_coords(Worksheet $ws): array {
    [$highestCol, $highestRow] = self::get_highest_coords($ws);

    $maxCol = 1;
    $maxRow = 1;
    $maxEmptyRows = self::MAX_EMPTY_THRESHOLD;
    for ($row = 1; $row <= $highestRow; $row++) {
      $emptyRow = true;
      $maxEmptyCols = self::MAX_EMPTY_THRESHOLD;
      for ($col = 1; $col <= $highestCol; $col++) {
        $value = null;
        if ($ws->cellExistsByColumnAndRow($col, $row)) {
          $value = $ws->getCellByColumnAndRow($col, $row)->getValue();
        }
        if ($value === null) {
          $maxEmptyCols--;
          if ($maxEmptyCols == 0) break;
        } else {
          $maxEmptyCols = self::MAX_EMPTY_THRESHOLD;
          if ($row > $maxRow) $maxRow = $row;
          if ($col > $maxCol) $maxCol = $col;
          $emptyRow = false;
        }
      }
      if ($emptyRow) {
        $maxEmptyRows--;
        if ($maxEmptyRows == 0) break;
      } else {
        $maxEmptyRows = self::MAX_EMPTY_THRESHOLD;
      }
    }
    return [$maxCol, $maxRow];
  }
}