Smalot\PdfParser\Page::getText()

Complexity

Conditions	10
Paths	7

Size

Total Lines	48
Code Lines	26

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	19
CRAP Score	14.0992

Importance

Changes	6
Bugs	1	Features	0

<?php
    public function getText(self $page = null): string
    {
        if ($contents = $this->get('Contents')) {
            if ($contents instanceof ElementMissing) {
                return '';
            } elseif ($contents instanceof ElementNull) {
                return '';
            } elseif ($contents instanceof PDFObject) {
                $elements = $contents->getHeader()->getElements();

                if (is_numeric(key($elements))) {
                    $new_content = '';

                    foreach ($elements as $element) {
                        if ($element instanceof ElementXRef) {
                            $new_content .= $element->getObject()->getContent();
                        } else {
                            $new_content .= $element->getContent();
                        }
                    }

                    $header = new Header([], $this->document);
                    $contents = new PDFObject($this->document, $header, $new_content, $this->config);
                }
            } elseif ($contents instanceof ElementArray) {
                // Create a virtual global content.
                $new_content = '';

                foreach ($contents->getContent() as $content) {
                    $new_content .= $content->getContent()."\n";
                }

                $header = new Header([], $this->document);
                $contents = new PDFObject($this->document, $header, $new_content, $this->config);
            }

            /*
             * Elements referencing each other on the same page can cause endless loops during text parsing.
             * To combat this we keep a recursionStack containing already parsed elements on the page.
             * The stack is only emptied here after getting text from a page.
             */
            $contentsText = $contents->getText($this);
            PDFObject::$recursionStack = [];

            return $contentsText;
        }

        return '';
    }