Merge branch 'B11509-PDF-splitting-fails' into 'develop'

Fixes #11509. Implement a) detection of broken 'split PDF to JPEG', b) fire...

See merge request !292

Documentation/Installation.rst

@@ -46,6 +46,8 @@ The following features are only tested / supported on linux hosts:
 * PDF decrypt (used for merge with pdfunite) - command `qpdf`.
 * PDF decrypt (used for merge with pdfunite) - command `gs` - in case `qpdf` is not successful.
 * Mime type detection for uploads - command `file`.
+* Split PDF into JPG - command `convert`.
+* Repair PDF - command `pdftocairo`.
 
 
 .. _`preparation`:
@@ -401,7 +403,11 @@ Extension Manager: QFQ Configuration
 +-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+
 | cmdInkscape                       | inkscape                                              | If inkscape is not available, specify an empty string.                     |
 +-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+
-| cmdConvert                        | convert                                               | GraphicsMagics 'convert' is recommended.                                   |
+| cmdConvert                        | convert                                               | Image-/GraphicsMagics 'convert' is recommended.                            |
++-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+
+| cmdPdf2svg                        | pdf2svg                                               | Convert PDF to SVG.                                                        |
++-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+
+| cmdPdftocairo                     | pdftocairo                                            | Poppler based.                                                             |
 +-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+
 | cmdWkhtmltopdf                    | /usr/bin/wkhtmltopdf                                  | PathFilename of wkhtmltopdf. Optional variables like LD_LIBRARY_PATH=...   |
 +-----------------------------------+-------------------------------------------------------+----------------------------------------------------------------------------+

extension/Classes/Core/Constants.php

@@ -259,7 +259,7 @@ const ERROR_IO_CHMOD = 1314;
 const ERROR_IO_READ_FILE = 1315;
 const ERROR_IO_WRITE_FILE = 1316;
 
-const ERROR_PDF2SVG = 1320;
+const ERROR_PDF_SPLIT = 1320;
 const ERROR_PDF2JPEG = 1321;
 
 //Report
@@ -647,6 +647,8 @@ const SYSTEM_SHOW_ID_IN_FORM_TITLE = 'showIdInFormTitle';
 const SYSTEM_CMD_WKHTMLTOPDF = 'cmdWkhtmltopdf';
 const SYSTEM_CMD_INKSCAPE = 'cmdInkscape';
 const SYSTEM_CMD_CONVERT = 'cmdConvert';
+const SYSTEM_CMD_PDF2SVG = 'cmdPdf2svg';
+const SYSTEM_CMD_PDFTOCAIRO = 'cmdPdftocairo';
 const SYSTEM_CMD_QPDF = 'cmdQpdf';
 const SYSTEM_CMD_GS = 'cmdGs';
 const SYSTEM_CMD_PDFUNITE = 'cmdPdfunite';

extension/Classes/Core/Helper/HelperFile.php

@@ -336,7 +336,7 @@ class HelperFile {
     public static function rename($oldname, $newname) {
 
         if (false === @rename($oldname, $newname)) {
-            $msg = self::errorGetLastAsString() . " - rename($oldname ,$newname)";
+            $msg = self::errorGetLastAsString() . " - rename($oldname, $newname)";
             throw new \UserFormException(json_encode([ERROR_MESSAGE_TO_USER => 'unlink failed', ERROR_MESSAGE_TO_DEVELOPER => $msg]), ERROR_IO_RENAME);
         }
 

extension/Classes/Core/Save.php

@@ -1262,13 +1262,10 @@ class Save {
 
         switch ($fileSplitType) {
             case FE_FILE_SPLIT_SVG:
-                $cmd = 'pdf2svg "' . $newSrc . '" "' . $fileNameDest . '" all';
+                $cmd = $this->buildPdf2svg($newSrc, $fileNameDest);
                 break;
             case FE_FILE_SPLIT_JPEG:
-                if ($fileSplitTypeOptions == '') {
-                    $fileSplitTypeOptions = FE_FILE_SPLIT_OPTIONS_JPEG;
-                }
-                $cmd = "convert $fileSplitTypeOptions '$newSrc' '$fileNameDest'";
+                $cmd = $this->buildConvertSplit($newSrc, $fileNameDest, $fileSplitTypeOptions);
                 break;
             default:
                 throw new \UserFormException("Unknown 'fileSplit' type: " . $formElement[FE_FILE_SPLIT], ERROR_UNKNOWN_TOKEN);
@@ -1276,13 +1273,30 @@ class Save {
 
         // Split PDF
         HelperFile::chdir($tempDir);
-        $output = Support::qfqExec($cmd, $rc);
+        $cnt = 0;
+        do {
+            $output = Support::qfqExec($cmd, $rc);
+            $cnt++;
+
+            if ($rc != 0) {
+                // Split failed
+                $cmdRepair = $this->buildPdftocairo($newSrc, $newSrc . '.1');
+                $output .= PHP_EOL . $cmdRepair . PHP_EOL;
+                $output .= Support::qfqExec($cmdRepair, $rc1);
+                if ($rc1 != 0) {
+                    // Repair failed too: stop here.
+                    break;
+                }
+                HelperFile::rename($newSrc . '.1', $newSrc);
+            }
+        } while ($rc != 0 && $cnt < 2);
+
         HelperFile::chdir($cwd);
 
         if ($rc != 0) {
             throw new \UserFormException(
-                json_encode([ERROR_MESSAGE_TO_USER => 'pdf2svg failed', ERROR_MESSAGE_TO_DEVELOPER => "[$cwd][cmd=$cmd]$output"]),
-                ERROR_PDF2SVG);
+                json_encode([ERROR_MESSAGE_TO_USER => 'pdf split failed', ERROR_MESSAGE_TO_DEVELOPER => "[$cwd][cmd=$cmd]$output"]),
+                ERROR_PDF_SPLIT);
         }
 
         $files = Helperfile::getSplitFileNames($tempDir);
@@ -1326,6 +1340,54 @@ class Save {
         HelperFile::rmdir($tempDir);
     }
 
+    /**
+     * @param $src
+     * @param $dest
+     * @return string
+     * @throws \CodeException
+     * @throws \DbException
+     * @throws \UserFormException
+     * @throws \UserReportException
+     */
+    private function buildPdf2svg($src, $dest) {
+
+        $cmd = $this->store->getVar(SYSTEM_CMD_PDF2SVG, STORE_SYSTEM);
+        return "$cmd '$src' '$dest' all";
+    }
+
+    /**
+     * @param $src
+     * @param $dest
+     * @return string
+     * @throws \CodeException
+     * @throws \DbException
+     * @throws \UserFormException
+     * @throws \UserReportException
+     */
+    private function buildPdftocairo($src, $dest) {
+
+        $cmd = $this->store->getVar(SYSTEM_CMD_PDFTOCAIRO, STORE_SYSTEM);
+        return "$cmd -pdf '$src' '$dest'";
+    }
+
+    /**
+     * @param $src
+     * @param $dest
+     * @param $options
+     * @return string
+     * @throws \CodeException
+     * @throws \DbException
+     * @throws \UserFormException
+     * @throws \UserReportException
+     */
+    private function buildConvertSplit($src, $dest, $options) {
+        if ($options == '') {
+            $options = FE_FILE_SPLIT_OPTIONS_JPEG;
+        }
+        $cmd = $this->store->getVar(SYSTEM_CMD_CONVERT, STORE_SYSTEM);
+        return "$cmd $options '$src' '$dest'";
+    }
+
     /**
      * Create/update or delete the slave record.
      *

extension/Classes/Core/Store/Config.php

@@ -374,6 +374,8 @@ class Config {
             SYSTEM_CMD_WKHTMLTOPDF => '/opt/wkhtmltox/bin/wkhtmltopdf',
             SYSTEM_CMD_INKSCAPE => 'inkscape',
             SYSTEM_CMD_CONVERT => 'convert',
+            SYSTEM_CMD_PDF2SVG => 'pdf2svg',
+            SYSTEM_CMD_PDFTOCAIRO => 'pdftocairo',
             SYSTEM_CMD_QPDF => 'qpdf',
             SYSTEM_CMD_GS => 'gs',
             SYSTEM_CMD_PDFUNITE => 'pdfunite',

extension/Tests/Unit/Core/Store/StoreTest.php

@@ -345,6 +345,8 @@ class StoreTest extends TestCase {
             SYSTEM_THUMBNAIL_DIR_PUBLIC => SYSTEM_THUMBNAIL_DIR_PUBLIC_DEFAULT,
             SYSTEM_CMD_INKSCAPE => 'inkscape',
             SYSTEM_CMD_CONVERT => 'convert',
+            SYSTEM_CMD_PDF2SVG => 'pdf2svg',
+            SYSTEM_CMD_PDFTOCAIRO => 'pdftocairo',
             SYSTEM_CMD_WKHTMLTOPDF => '/opt/wkhtmltox/bin/wkhtmltopdf',
             SYSTEM_DATE_FORMAT => 'yyyy-mm-dd',
             SYSTEM_THROW_GENERAL_ERROR => 'no',

extension/ext_conf_template.txt

@@ -22,9 +22,15 @@ thumbnailDirPublic = typo3temp/qfqThumbnail
 # cat=config/config; type=string; label=Command 'inkscape':Default is 'inkscape'. Will be used to convert SVG to images (png). An empty string disables `inkscape`. If it is not available, `convert` will be used instead.
 cmdInkscape = inkscape
 
-# cat=config/config; type=string; label=Command 'convert':Default is 'convert'. Will be used to convert images to thumbnails. GraphicsMagics 'convert' is recommended.
+# cat=config/config; type=string; label=Command 'convert':Default is 'convert'. Will be used to convert images to thumbnails and to split PDF to JPEG.
 cmdConvert = convert
 
+# cat=config/config; type=string; label=Command 'pdf2svg':Default is 'pdf2svg'. Will be used to convert PDF to SVG.
+cmdPdf2svg = pdf2svg
+
+# cat=config/config; type=string; label=Command 'pdftocairo':Default is 'pdftocairo'. Will be used to repair broken PDFs to be splitted via convert.
+cmdPdftocairo = pdftocairo
+
 # cat=config/config; type=string; label=Command 'wkhtmltopdf':Default is '/opt/wkhtmltox/bin/wkhtmltopdf'. Command to convert a HTML page to a PDF.
 cmdWkhtmltopdf = /opt/wkhtmltox/bin/wkhtmltopdf