FPDI extension to preserve external hyperlinks

fpdi_with_annots.php

<?php

// FPDI extension that preserves hyperlinks when copying PDF pages.
//
// (c) 2012, Andrey Tarantsov <andrey@tarantsov.com>, provided under the MIT license.
//
// Published at: https://gist.github.com/2020422
//
// Note: the free version of FPDI requires unprotected PDFs conforming to spec version 1.4.
// I use qpdf (http://qpdf.sourceforge.net/) to preprocess PDFs before running through this
// code, invoking it like this:
//
//     qpdf --decrypt --stream-data=uncompress --force-version=1.4 src.pdf temp.pdf
//
// then, after processing temp.pdf into out.pdf with FPDI, I run the following to re-establish
// protection:
//
//     qpdf --encrypt "" "" 40 --extract=n -- out.pdf final.pdf
//
class FPDI_with_annots extends FPDI {

    // default maxdepth prevents an infinite recursion on malformed PDFs (not theoretical, actually found in the wild)
    function resolve(&$parser, $smt, $maxdepth=10) {
        if ($maxdepth == 0)
            return $smt;

        if ($smt[0] == PDF_TYPE_OBJREF) {
            $result = $parser->pdf_resolve_object($parser->c, $smt);
            return $this->resolve($parser, $result, $maxdepth-1);

        } else if ($smt[0] == PDF_TYPE_OBJECT) {
            return $this->resolve($parser, $smt[1], $maxdepth-1);

        } else if ($smt[0] == PDF_TYPE_ARRAY) {
            $result = array();
            foreach ($smt[1] as $item) {
                $result[] = $this->resolve($parser, $item, $maxdepth-1);
            }
            $smt[1] = $result;
            return $smt;

        } else if ($smt[0] == PDF_TYPE_DICTIONARY) {
            $result = array();
            foreach ($smt[1] as $key => $item) {
                $result[$key] = $this->resolve($parser, $item, $maxdepth-1);
            }
            $smt[1] = $result;
            return $smt;

        } else {
            return $smt;
        }
    }

    function findPageNoForRef(&$parser, $pageRef) {
        $ref_obj = $pageRef[1]; $ref_gen = $pageRef[2];

        foreach ($parser->pages as $index => $page) {
          $page_obj = $page['obj']; $page_gen = $page['gen'];
          if ($page_obj == $ref_obj && $page_gen == $ref_gen) {
              return $index + 1;
          }
        }

        return -1;
    }

    function importPage($pageno, $boxName = '/CropBox') {
        $tplidx = parent::importPage($pageno, $boxName);

        $tpl =& $this->tpls[$tplidx];
        $parser =& $tpl['parser'];

        // look for hyperlink annotations and store them in the template
        if (isset($parser->pages[$pageno - 1][1][1]['/Annots'])) {
            $annots = $parser->pages[$pageno - 1][1][1]['/Annots'];
            $annots = $this->resolve($parser, $annots);

            $links = array();
            foreach ($annots[1] as $annot) if ($annot[0] == PDF_TYPE_DICTIONARY) {
                // all links look like:  << /Type /Annot /Subtype /Link /Rect [...] ... >>
                if ($annot[1]['/Type'][1] == '/Annot' && $annot[1]['/Subtype'][1] == '/Link') {
                    $rect = $annot[1]['/Rect'];
                    if ($rect[0] == PDF_TYPE_ARRAY && count($rect[1]) == 4) {
                        $x = $rect[1][0][1]; $y = $rect[1][1][1];
                        $x2 = $rect[1][2][1]; $y2 = $rect[1][3][1];
                        $w = $x2 - $x; $h = $y2 - $y;
                        $h = -$h;
                    }

                    if (isset($annot[1]['/A'])) {
                        $A = $annot[1]['/A'];

                        if ($A[0] == PDF_TYPE_DICTIONARY && isset($A[1]['/S'])) {
                            $S = $A[1]['/S'];

                            //  << /Type /Annot ... /A << /S /URI /URI ... >> >>
                            if ($S[1] == '/URI' && isset($A[1]['/URI'])) {
                                $URI = $A[1]['/URI'];

                                if (is_string($URI[1])) {
                                    $uri = str_replace("\\000", '', trim($URI[1]));
                                    if (!empty($uri)) {
                                        $links[] = array($x, $y, $w, $h, $uri);
                                    }
                                }

                            //  << /Type /Annot ... /A << /S /GoTo /D [%d 0 R /Fit] >> >>
                            } else if ($S[1] == '/GoTo' && isset($A[1]['/D'])) {
                                $D = $A[1]['/D'];
                                if ($D[0] == PDF_TYPE_ARRAY && count($D[1]) > 0 && $D[1][0][0] == PDF_TYPE_OBJREF) {
                                    $target_pageno = $this->findPageNoForRef($parser, $D[1][0]);
                                    if ($target_pageno >= 0) {
                                        $links[] = array($x, $y, $w, $h, $target_pageno);
                                    }
                                }
                            }
                        }

                    } else if (isset($annot[1]['/Dest'])) {
                        $Dest = $annot[1]['/Dest'];

                        //  << /Type /Annot ... /Dest [42 0 R ...] >>
                        if ($Dest[0] == PDF_TYPE_ARRAY && $Dest[0][1][0] == PDF_TYPE_OBJREF) {
                            $target_pageno = $this->findPageNoForRef($parser, $Dest[0][1][0]);
                            if ($target_pageno >= 0) {
                                $links[] = array($x, $y, $w, $h, $target_pageno);
                            }
                        }
                    }
                }
            }
            $tpl['links'] = $links;
        }
        // echo "Links on page $pageno:\n";
        // print_r($links);

        return $tplidx;
    }

    function useTemplate($tplidx, $_x = null, $_y = null, $_w = 0, $_h = 0, $adjustPageSize = false) {
        $result = parent::useTemplate($tplidx, $_x, $_y, $_w, $_h, $adjustPageSize);

        // apply links from the template
        $tpl =& $this->tpls[$tplidx];
        if (isset($tpl['links'])) {
            foreach ($tpl['links'] as $link) {
                // $link[4] is either a string (external URL) or an integer (page number)
                if (is_int($link[4])) {
                    $l = $this->AddLink();
                    $this->SetLink($l, 0, $link[4]);
                    $link[4] = $l;
                }
                $this->PageLinks[$this->page][] = $link;
            }
        }

        return $result;
    }

}

pdf_personalization_example.php

<?php

// uses FPDI to append another PDF file, watermarking each page with a message
class FPDI_AppendWithWatermark extends FPDI_with_annots {

    function AppendPDFWithWatermarkMessage($file, $message) {
        $pagecount = $this->setSourceFile($file);
        for ($i = 1; $i <= $pagecount; $i++) {
            $tplidx = $this->ImportPage($i);
            $s = $this->getTemplatesize($tplidx);
            $this->AddPage('P', array($s['w'], $s['h']));
            $this->useTemplate($tplidx);

            // watermark (a message printed vertically along the left margin)
            $this->SetAutoPageBreak(FALSE);
            $this->SetXY(6, -28);
            $this->Rotate(90);
            $this->SetTextColor(102, 102, 102);
            $this->SetFont('Arial', '', 8);
            $this->Cell(0, 5, utf8_decode($message),'',1,'L');
            $this->Rotate(0); // outputs Q to balance "q" added by the previous call to Rotate
        }
    }

}

// combines FPDI_AppendWithWatermark and qpdf to watermark existing PDF files
function personalize_pdf($source_file, $output_file, $temp_file, $message, $debug_mode = FALSE) {
    setlocale(LC_CTYPE, "en_US.UTF-8"); // otherwise escapeshellarg() strips non-ASCII characters

    // see the comments in FPDI_with_annots as to why we have to run this
    $cmd = sprintf('qpdf --decrypt --stream-data=uncompress --force-version=1.4 %s %s', escapeshellarg($source_file), escapeshellarg($temp_file));
    $output = shell_exec($cmd);
    if (!file_exists($temp_file) || filesize($temp_file) == 0) {
        if ($debug_mode) die("Error occurred while running:\n$cmd\n\nOutput:\n$output");
        return FALSE;
    }

    $pdf = new FPDI_AppendWithWatermark();

    // make debugging easier by leaving the output file uncompressed
    if ($debug_mode) $pdf->SetCompression(FALSE);

    $pdf->AppendPDFWithWatermarkMessage($temp_file, $message);
    $pdf->Output($temp_file, 'F');

    if ($debug_mode) {
        // make debugging easier by omitting the final processing step
        copy($temp_file, $output_file);
    } else {
        $cmd = sprintf('qpdf --encrypt "" "" 40 --extract=n -- %s %s', escapeshellarg($temp_file), escapeshellarg($output_file));
        $output = shell_exec($cmd);
        if (!file_exists($output_file) || filesize($output_file) == 0) {
            if ($debug_mode) die("Error occurred while running:\n$cmd\n\nOutput:\n$output");
            return FALSE;
        }
    }
    return TRUE;
}

Just a note for anyone that might be puzzled why this code works for some document links and not others: The code by @andreyvit works for page annotations that involve external links (these are URL links), and it works for some internal links ONLY IF those internal links are annotated in the document with an array of numeric coordinates, it DOES NOT work if the internal links are annotated in the document by name (name is a string, for example "appendix-a"), there's no code in the class file that cross references a name to a page number. Also note that the code will not preserve any bookmark "links" because those are stored in the document differently than links that are stored as "Annots".

I mention this because I spent several hours trying to figure out why sometimes links carried through and other times they didn't. After digging into document structure and experimenting I realized what was happening. And I don't know enough about PDF parsing to modify the code to work for named links or bookmarks.

Thank you publishing the snippets @andreyvit !

For Fpdi 2+ I created this
https://gist.github.com/trubit/65c7674c4d644db107f25f021acd5dfd

Curently working only with External links. Needs testing, but works for me.

@jakubnavratil Your URL seems to have changed since you posted this. Looks like this is the correct URL for people looking for a modern solution: https://gist.github.com/jakubnavratil/65c7674c4d644db107f25f021acd5dfd

If you guys are looking for a newer solution using FPDI 2.3 and TCPDF 6.4+, take a look at this answer:
https://stackoverflow.com/a/67071744/5397846

Supporting both internal and external annotation links (hyperlinks)