Smalot\PdfParser\PDFObject::getTextArray()

Complexity

Conditions	68
Paths	> 20000

Size

Total Lines	366
Code Lines	207

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	110
CRAP Score	68.3806

Importance

Changes	3
Bugs	0	Features	0

<?php
    public function getTextArray(Page $page = null): array
    {
        $result = [];
        $text = [];

        $marked_stack = [];
        $last_written_position = false;

        $sections = $this->getSectionsText($this->content);
        $current_font = $this->getDefaultFont($page);
        $current_font_size = 1;
        $current_text_leading = 0;

        $current_position = ['x' => false, 'y' => false];
        $current_position_tm = [
            'a' => 1, 'b' => 0, 'c' => 0,
            'i' => 0, 'j' => 1, 'k' => 0,
            'x' => 0, 'y' => 0, 'z' => 1,
        ];
        $current_position_td = ['x' => 0, 'y' => 0];
        $current_position_cm = [
            'a' => 1, 'b' => 0, 'c' => 0,
            'i' => 0, 'j' => 1, 'k' => 0,
            'x' => 0, 'y' => 0, 'z' => 1,
        ];

        $clipped_font = [];
        $clipped_position_cm = [];

        self::$recursionStack[] = $this->getUniqueId();

        foreach ($sections as $section) {
            $commands = $this->getCommandsText($section);
            foreach ($commands as $command) {
                switch ($command[self::OPERATOR]) {
                    // Begin text object
                    case 'BT':
                        // Reset text positioning matrices
                        $current_position_tm = [
                            'a' => 1, 'b' => 0, 'c' => 0,
                            'i' => 0, 'j' => 1, 'k' => 0,
                            'x' => 0, 'y' => 0, 'z' => 1,
                        ];
                        $current_position_td = ['x' => 0, 'y' => 0];
                        $current_text_leading = 0;
                        break;

                        // Begin marked content sequence with property list
                    case 'BDC':
                        if (preg_match('/(<<.*>>)$/', $command[self::COMMAND], $match)) {
                            $dict = $this->parseDictionary($match[1]);

                            // Check for ActualText block
                            if (isset($dict['ActualText']) && \is_string($dict['ActualText']) && '' !== $dict['ActualText']) {
                                if ('[' == $dict['ActualText'][0]) {
                                    // Simulate a 'TJ' command on the stack
                                    $marked_stack[] = [
                                        'ActualText' => $this->getCommandsText($dict['ActualText'].'TJ')[0],
                                    ];
                                } elseif ('<' == $dict['ActualText'][0] || '(' == $dict['ActualText'][0]) {
                                    // Simulate a 'Tj' command on the stack
                                    $marked_stack[] = [
                                        'ActualText' => $this->getCommandsText($dict['ActualText'].'Tj')[0],
                                    ];
                                }
                            }
                        }
                        break;

                        // Begin marked content sequence
                    case 'BMC':
                        if ('ReversedChars' == $command[self::COMMAND]) {
                            // Upon encountering a ReversedChars command,
                            // add the characters we've built up so far to
                            // the result array
                            $result = array_merge($result, $text);

                            // Start a fresh $text array that will contain
                            // reversed characters
                            $text = [];

                            // Add the reversed text flag to the stack
                            $marked_stack[] = ['ReversedChars' => true];
                        }
                        break;

                        // set graphics position matrix
                    case 'cm':
                        $args = preg_split('/\s+/s', $command[self::COMMAND]);
                        $current_position_cm = [
                            'a' => (float) $args[0], 'b' => (float) $args[1], 'c' => 0,
                            'i' => (float) $args[2], 'j' => (float) $args[3], 'k' => 0,
                            'x' => (float) $args[4], 'y' => (float) $args[5], 'z' => 1,
                        ];
                        break;

                    case 'Do':
                        if (null !== $page) {
                            $args = preg_split('/\s/s', $command[self::COMMAND]);
                            $id = trim(array_pop($args), '/ ');
                            $xobject = $page->getXObject($id);

                            // @todo $xobject could be a ElementXRef object, which would then throw an error
                            if (\is_object($xobject) && $xobject instanceof self && !\in_array($xobject->getUniqueId(), self::$recursionStack)) {
                                // Not a circular reference.
                                $text[] = $xobject->getText($page);
                            }
                        }
                        break;

                        // Marked content point with (DP) & without (MP) property list
                    case 'DP':
                    case 'MP':
                        break;

                        // End text object
                    case 'ET':
                        break;

                        // Store current selected font and graphics matrix
                    case 'q':
                        $clipped_font[] = [$current_font, $current_font_size];
                        $clipped_position_cm[] = $current_position_cm;
                        break;

                        // Restore previous selected font and graphics matrix
                    case 'Q':
                        list($current_font, $current_font_size) = array_pop($clipped_font);
                        $current_position_cm = array_pop($clipped_position_cm);
                        break;

                        // End marked content sequence
                    case 'EMC':
                        $data = false;
                        if (\count($marked_stack)) {
                            $marked = array_pop($marked_stack);
                            $action = key($marked);
                            $data = $marked[$action];

                            switch ($action) {
                                // If we are in ReversedChars mode...
                                case 'ReversedChars':
                                    // Reverse the characters we've built up so far
                                    foreach ($text as $key => $t) {
                                        $text[$key] = implode('', array_reverse(
                                            mb_str_split($t, 1, mb_internal_encoding())

It seems like mb_internal_encoding() can also be of type true; however, parameter $encoding of mb_str_split() does only seem to accept null|string, maybe add an additional type check?

                                        ));
                                    }

                                    // Add these characters to the result array
                                    $result = array_merge($result, $text);

                                    // Start a fresh $text array that will contain
                                    // non-reversed characters
                                    $text = [];
                                    break;

                                case 'ActualText':
                                    // Use the content of the ActualText as a command
                                    $command = $data;
                                    break;
                            }
                        }

                        // If this EMC command has been transformed into a 'Tj'
                        // or 'TJ' command because of being ActualText, then bypass
                        // the break to proceed to the writing section below.
                        if ('Tj' != $command[self::OPERATOR] && 'TJ' != $command[self::OPERATOR]) {
                            break;
                        }

                        // no break
                    case "'":
                    case '"':
                        if ("'" == $command[self::OPERATOR] || '"' == $command[self::OPERATOR]) {
                            // Move to next line and write text
                            $current_position['x'] = 0;
                            $current_position_td['x'] = 0;
                            $current_position_td['y'] += $current_text_leading;
                        }
                        // no break
                    case 'Tj':
                        $command[self::COMMAND] = [$command];
                        // no break
                    case 'TJ':
                        // Check the marked content stack for flags
                        $actual_text = false;
                        $reverse_text = false;
                        foreach ($marked_stack as $marked) {
                            if (isset($marked['ActualText'])) {
                                $actual_text = true;
                            }
                            if (isset($marked['ReversedChars'])) {
                                $reverse_text = true;
                            }
                        }

                        // Account for text position ONLY just before we write text
                        if (false === $actual_text && \is_array($last_written_position)) {
                            // If $last_written_position is an array, that
                            // means we have stored text position coordinates
                            // for placing an ActualText
                            $currentX = $last_written_position[0];
                            $currentY = $last_written_position[1];
                            $last_written_position = false;
                        } else {
                            $currentX = $current_position_cm['x'] + $current_position_tm['x'] + $current_position_td['x'];
                            $currentY = $current_position_cm['y'] + $current_position_tm['y'] + $current_position_td['y'];
                        }
                        $whiteSpace = '';

                        $factorX = -$current_font_size * $current_position_tm['a'] - $current_font_size * $current_position_tm['i'];
                        $factorY = $current_font_size * $current_position_tm['b'] + $current_font_size * $current_position_tm['j'];

                        if (true === $this->addPositionWhitespace && false !== $current_position['x']) {
                            $curY = $currentY - $current_position['y'];
                            if (abs($curY) >= abs($factorY) / 4) {
                                $whiteSpace = "\n";
                            } else {
                                if (true === $reverse_text) {
                                    $curX = $current_position['x'] - $currentX;
                                } else {
                                    $curX = $currentX - $current_position['x'];
                                }

                                // In abs($factorX * 7) below, the 7 is chosen arbitrarily
                                // as the number of apparent "spaces" in a document we
                                // would need before considering them a "tab". In the
                                // future, we might offer this value to users as a config
                                // option.
                                if ($curX >= abs($factorX * 7)) {
                                    $whiteSpace = "\t";
                                } elseif ($curX >= abs($factorX * 2)) {
                                    $whiteSpace = ' ';
                                }
                            }
                        }

                        $newtext = $this->getTJUsingFontFallback(
                            $current_font,
                            $command[self::COMMAND],
                            $page,
                            $factorX
                        );

                        // If there is no ActualText pending then write
                        if (false === $actual_text) {
                            $newtext = str_replace(["\r", "\n"], '', $newtext);
                            if (false !== $reverse_text) {
                                // If we are in ReversedChars mode, add the whitespace last
                                $text[] = preg_replace('/  $/', ' ', $newtext.$whiteSpace);
                            } else {
                                // Otherwise add the whitespace first
                                if (' ' === $whiteSpace && isset($text[\count($text) - 1])) {
                                    $text[\count($text) - 1] = preg_replace('/ $/', '', $text[\count($text) - 1]);
                                }
                                $text[] = preg_replace('/^[ \t]{2}/', ' ', $whiteSpace.$newtext);
                            }

                            // Record the position of this inserted text for comparison
                            // with the next text block.
                            // Provide a 'fudge' factor guess on how wide this text block
                            // is based on the number of characters. This helps limit the
                            // number of tabs inserted, but isn't perfect.
                            $factor = $factorX / 2;
                            $current_position = [
                                'x' => $currentX - mb_strlen($newtext) * $factor,
                                'y' => $currentY,
                            ];
                        } elseif (false === $last_written_position) {
                            // If there is an ActualText in the pipeline
                            // store the position this undisplayed text
                            // *would* have been written to, so the
                            // ActualText is displayed in the right spot
                            $last_written_position = [$currentX, $currentY];
                            $current_position['x'] = $currentX;
                        }
                        break;

                        // move to start of next line
                    case 'T*':
                        $current_position['x'] = 0;
                        $current_position_td['x'] = 0;
                        $current_position_td['y'] += $current_text_leading;
                        break;

                        // set character spacing
                    case 'Tc':
                        break;

                        // move text current point and set leading
                    case 'Td':
                    case 'TD':
                        // move text current point
                        $args = preg_split('/\s+/s', $command[self::COMMAND]);
                        $y = (float) array_pop($args);
                        $x = (float) array_pop($args);

                        if ('TD' == $command[self::OPERATOR]) {
                            $current_text_leading = -$y * $current_position_tm['b'] - $y * $current_position_tm['j'];
                        }

                        $current_position_td = [
                            'x' => $current_position_td['x'] + $x * $current_position_tm['a'] + $x * $current_position_tm['i'],
                            'y' => $current_position_td['y'] + $y * $current_position_tm['b'] + $y * $current_position_tm['j'],
                        ];
                        break;

                    case 'Tf':
                        $args = preg_split('/\s/s', $command[self::COMMAND]);
                        $size = (float) array_pop($args);
                        $id = trim(array_pop($args), '/');
                        if (null !== $page) {
                            $new_font = $page->getFont($id);
                            // If an invalid font ID is given, do not update the font.
                            // This should theoretically never happen, as the PDF spec states for the Tf operator:
                            // "The specified font value shall match a resource name in the Font entry of the default resource dictionary"
                            // (https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/PDF32000_2008.pdf, page 435)
                            // But we want to make sure that malformed PDFs do not simply crash.
                            if (null !== $new_font) {
                                $current_font = $new_font;
                                $current_font_size = $size;
                            }
                        }
                        break;

                        // set leading
                    case 'TL':
                        $y = (float) $command[self::COMMAND];
                        $current_text_leading = -$y * $current_position_tm['b'] + -$y * $current_position_tm['j'];
                        break;

                        // set text position matrix
                    case 'Tm':
                        $args = preg_split('/\s+/s', $command[self::COMMAND]);
                        $current_position_tm = [
                            'a' => (float) $args[0], 'b' => (float) $args[1], 'c' => 0,
                            'i' => (float) $args[2], 'j' => (float) $args[3], 'k' => 0,
                            'x' => (float) $args[4], 'y' => (float) $args[5], 'z' => 1,
                        ];
                        break;

                        // set text rendering mode
                    case 'Ts':
                        break;

                        // set super/subscripting text rise
                    case 'Ts':
                        break;

                        // set word spacing
                    case 'Tw':
                        break;

                        // set horizontal scaling
                    case 'Tz':
                        break;

                    default:
                }
            }
        }

        $result = array_merge($result, $text);

        return $result;
    }