pdftopdf: Added pdftopdf-form-flattening option to select form flatte…

…ning method.

README

@@ -28,7 +28,7 @@ INTRODUCTION
     by both CUPS and this package. Then the filters of this package
     should be used.
 
-    For compiling and using this package CUPS, libqpdf (8.1.0 or
+    For compiling and using this package CUPS, libqpdf (8.3.0 or
     newer), libjpeg, libpng, libtiff, freetype, fontconfig, liblcms
     (liblcms2 recommended), libavahi-common, libavahi-client, libdbus,
     and glib are needed. It is highly recommended, especially if
@@ -845,9 +845,54 @@ non-landscape pages are rotated instead.
 Note: Some pages might end up 180 degree rotated (instead of 0 degree).
 Those should probably be rotated manually before binding the pages together.
 
+3) Method of flattening interactive PDF forms and annotations.
+
+Some PDF files (like application forms) contain interactive forms
+which the user can fill in inside a PDF viewer like evince. The filled
+in data is not integrated in each page of the PDF file but stored in
+an extra layer. Due to this the data gets lost when applying
+manipulations like scaling or N-up to the pages. To prevent the loss
+of the data pdftopdf flattens the form before doing the
+manipulations. This means the PDF will be converted into a static PDF
+file with the data being integral part of the pages.
+
+The same flattening is needed for annotations in PDF files.
+
+By default the actual flattening work is done by QPDF, as QPDF is also
+doing everything else in pdftopdf. This way no external utilities need
+to be called and so extra piping between processes and extra PDF
+interpreter runs are avoided which makes the filtering process faster.
+
+As we did not test the new QPDF-based form-flattening with thousands
+of PDF files yet and it has not been available to actual users yet it
+is possible that there are still some bugs. To give users a
+possibility to work around possible bugs in QPDF's form flattening, we
+have introduced an option to get back to the old flattening by the
+external tools pdftocairo or Ghostscript.
+
+The selection of the method is done by the "pdftopdf-form-flattening"
+option, setting it to "auto", "qpdf", "pdftocairo", "ghostscript",
+"gs", "internal" or "external":
+
+Per-job:           lpr -o pdftopdf-form-flattening=pdftocairo ...
+Per-queue default: lpadmin -p printer -o pdftopdf-form-flattening-default=gs
+Remove default:    lpadmin -p printer -R pdftopdf-form-flattening-default
+
+By default, pdftopdf uses QPDF if the option is not supplied, also the
+settings "auto" and "internal" make QPDF being used. "external"
+auto-selects from the two external utilities, trying pdftocairo at
+first and on failure Ghostscript. If the selected utility fails, the
+form stays unflattened and so the filled in data will possibly not get
+printed.
+
 Native PDF Printer / JCL Support
 --------------------------------
 
+Note that for most modern native PDF printers JCL is not needed any
+more as they are controlled via IPP. For these the PPD files get
+auto-generated by the support of CUPS and cups-filters for driverless
+IPP printing.
+
 pdftopdf will emit JCL when provided with a PPD file that includes the
 "*JCLToPDFInterpreter:" keyword.
 
@@ -915,7 +960,7 @@ License
 
 pdftopdf is released under the MIT license.
 
-The required libqpdf is available under version 2 of the Artistic License,
+The required libqpdf is available under version 2.0 of the Apache License,
 e.g. here: https://github.com/qpdf/qpdf
 
 

filter/pdftopdf/pdftopdf.cc

@@ -1026,7 +1026,7 @@ int main(int argc,char **argv)
     //param.mirror=true;
     //param.reverse=true;
     //param.numCopies=3;
-    if (!proc1->loadFilename("in.pdf")) return 2;
+    if (!proc1->loadFilename("in.pdf",1)) return 2;
     param.dump();
     if (!processPDFTOPDF(*proc1,param)) return 3;
     emitComment(*proc1,param);
@@ -1063,30 +1063,61 @@ int main(int argc,char **argv)
     param.dump();
 #endif
 
+    /* Check with which method we will flatten interactive PDF forms
+       and annotations so that they get printed also after page
+       manipulations (scaling, N-up, ...). Flattening means to
+       integrate the filled in data and the printable annotations into
+       the pages themselves instead of holding them in an extra
+       layer. Default method is using QPDF, alternatives are the
+       external utilities pdftocairo or Ghostscript, but these make
+       the processing slower, especially due to extra piping of the
+       data between processes. */
+    int qpdf_flatten = 1;
+    int pdftocairo_flatten = 0;
+    int gs_flatten = 0;
+    int external_auto_flatten = 0;
+    const char	*val;
+    if ((val = cupsGetOption("pdftopdf-form-flattening", num_options, options)) != NULL) {
+      if (strcasecmp(val, "qpdf") == 0 || strcasecmp(val, "internal") == 0 ||
+	  strcasecmp(val, "auto") == 0) {
+	qpdf_flatten = 1;
+      } else if (strcasecmp(val, "external") == 0) {
+	qpdf_flatten = 0;
+	external_auto_flatten = 1;
+      } else if (strcasecmp(val, "pdftocairo") == 0) {
+	qpdf_flatten = 0;
+	pdftocairo_flatten = 1;
+      } else if (strcasecmp(val, "ghostscript") == 0 || strcasecmp(val, "gs") == 0) {
+	qpdf_flatten = 0;
+	gs_flatten = 1;
+      } else
+	fprintf(stderr,
+		"WARNING: Invalid value for \"pdftopdf-form-flattening\": \"%s\"\n", val);
+    }
+
     cupsFreeOptions(num_options,options);
 
     std::unique_ptr<PDFTOPDF_Processor> proc(PDFTOPDF_Factory::processor());
 
     FILE *tmpfile = NULL;
     if (argc==7) {
-      if (!proc->loadFilename(argv[6])) {
+      if (!proc->loadFilename(argv[6],qpdf_flatten)) {
         ppdClose(ppd);
         return 1;
       }
     } else {
       tmpfile = copy_stdin_to_temp();
       if ((!tmpfile)||
-	  (!proc->loadFile(tmpfile,WillStayAlive))) {
+	  (!proc->loadFile(tmpfile,WillStayAlive,qpdf_flatten))) {
         ppdClose(ppd);
         return 1;
       }
     }
 
-    /* The input file contains a PDF form. To not loose the data filled
-       into the form during our further manipulations we need to flatten
-       the form, meaning that we integrate the filled in data into the
-       pages themselves instead of holding them in an extra layer */
-    if (proc->hasAcroForm()) {
+    /* If the input file contains a PDF form and we opted for not
+       using QPDF for flattening the form, we pipe the PDF through
+       pdftocairo or Ghostscript here */
+    if (!qpdf_flatten && proc->hasAcroForm()) {
       /* Prepare the input file for being read by the form flattening
 	 process */
       FILE *infile = NULL;
@@ -1119,22 +1150,26 @@ int main(int argc,char **argv)
       const char *command;
       cups_array_t *args;
       /* Choose the utility to be used and create its command line */
-      /* Try pdftocairo first, the preferred utility for form-flattening */
-      command = CUPS_POPPLER_PDFTOCAIRO;
-      args = cupsArrayNew(NULL, NULL);
-      cupsArrayAdd(args, strdup(command));
-      cupsArrayAdd(args, strdup("-pdf"));
-      cupsArrayAdd(args, strdup("-"));
-      cupsArrayAdd(args, strdup(buf));
-      /* Run the pdftocairo form flattening process */
-      rewind(infile);
-      int status = sub_process_spawn (command, args, infile);
-      cupsArrayDelete(args);
-      if (status == 0)
-	flattening_done = 1;
-      else {
-	error("Unable to execute pdftocairo for form flattening!");
-	/* Try Ghostscript, currently the only working alternative */
+      if (pdftocairo_flatten || external_auto_flatten) {
+	/* Try pdftocairo first, the preferred utility for form-flattening */
+	command = CUPS_POPPLER_PDFTOCAIRO;
+	args = cupsArrayNew(NULL, NULL);
+	cupsArrayAdd(args, strdup(command));
+	cupsArrayAdd(args, strdup("-pdf"));
+	cupsArrayAdd(args, strdup("-"));
+	cupsArrayAdd(args, strdup(buf));
+	/* Run the pdftocairo form flattening process */
+	rewind(infile);
+	int status = sub_process_spawn (command, args, infile);
+	cupsArrayDelete(args);
+	if (status == 0)
+	  flattening_done = 1;
+	else
+	  error("Unable to execute pdftocairo for form flattening!");
+      }
+      if (flattening_done == 0 &&
+	  (gs_flatten || external_auto_flatten)) {
+	/* Try Ghostscript */
 	command = CUPS_GHOSTSCRIPT;
 	args = cupsArrayNew(NULL, NULL);
 	cupsArrayAdd(args, strdup(command));
@@ -1157,11 +1192,12 @@ int main(int argc,char **argv)
 	cupsArrayDelete(args);
 	if (status == 0)
 	  flattening_done = 1;
-	else {
+	else
 	  error("Unable to execute Ghostscript for form flattening!");
-	  error("No suitable utility for flattening filled PDF forms available, no flattening performed. Filled in content will not be printed.");
-	  rewind(infile);
-	}
+      }
+      if (flattening_done == 0) {
+	error("No suitable utility for flattening filled PDF forms available, no flattening performed. Filled in content will possibly not be printed.");
+	rewind(infile);
       }
       /* Clean up */
       if (infile != tmpfile)
@@ -1170,12 +1206,13 @@ int main(int argc,char **argv)
       if (flattening_done) {
 	rewind(outfile);
 	unlink(buf);
-	if (!proc->loadFile(outfile,TakeOwnership)) {
+	if (!proc->loadFile(outfile,TakeOwnership,0)) {
 	  error("Unable to create a PDF processor on the flattened form!"); 
 	  return 1;
 	}
       }
-    }
+    } else if (qpdf_flatten)
+      fprintf(stderr, "DEBUG: PDF interactive form and annotation flattening done via QPDF\n");
 
 /* TODO
     // color management

filter/pdftopdf/pdftopdf_processor.h

@@ -119,8 +119,8 @@ class PDFTOPDF_Processor { // abstract interface
   virtual ~PDFTOPDF_Processor() {}
 
   // TODO: ... qpdf wants password at load time
-  virtual bool loadFile(FILE *f,ArgOwnership take=WillStayAlive) =0;
-  virtual bool loadFilename(const char *name) =0;
+  virtual bool loadFile(FILE *f,ArgOwnership take=WillStayAlive,int flatten_forms=1) =0;
+  virtual bool loadFilename(const char *name,int flatten_forms=1) =0;
 
   // TODO? virtual bool may_modify/may_print/?
   virtual bool check_print_permissions() =0;

filter/pdftopdf/qpdf_pdftopdf_processor.cc

@@ -381,7 +381,7 @@ void QPDF_PDFTOPDF_Processor::error(const char *fmt,...) // {{{
 
 // TODO?  try/catch for PDF parsing errors?
 
-bool QPDF_PDFTOPDF_Processor::loadFile(FILE *f,ArgOwnership take) // {{{
+bool QPDF_PDFTOPDF_Processor::loadFile(FILE *f,ArgOwnership take,int flatten_forms) // {{{
 {
   closeFile();
   if (!f) {
@@ -416,12 +416,12 @@ bool QPDF_PDFTOPDF_Processor::loadFile(FILE *f,ArgOwnership take) // {{{
     error("loadFile with MustDuplicate is not supported");
     return false;
   }
-  start();
+  start(flatten_forms);
   return true;
 }
 // }}}
 
-bool QPDF_PDFTOPDF_Processor::loadFilename(const char *name) // {{{
+bool QPDF_PDFTOPDF_Processor::loadFilename(const char *name,int flatten_forms) // {{{
 {
   closeFile();
   try {
@@ -431,20 +431,22 @@ bool QPDF_PDFTOPDF_Processor::loadFilename(const char *name) // {{{
     error("loadFilename failed: %s",e.what());
     return false;
   }
-  start();
+  start(flatten_forms);
   return true;
 }
 // }}}
 
-void QPDF_PDFTOPDF_Processor::start() // {{{
+void QPDF_PDFTOPDF_Processor::start(int flatten_forms) // {{{
 {
   assert(pdf);
 
-  QPDFAcroFormDocumentHelper afdh(*pdf);
-  afdh.generateAppearancesIfNeeded();
+  if (flatten_forms) {
+    QPDFAcroFormDocumentHelper afdh(*pdf);
+    afdh.generateAppearancesIfNeeded();
 
-  QPDFPageDocumentHelper dh(*pdf);
-  dh.flattenAnnotations(an_print);
+    QPDFPageDocumentHelper dh(*pdf);
+    dh.flattenAnnotations(an_print);
+  }
 
   pdf->pushInheritedAttributesToPage();
   orig_pages=pdf->getAllPages();

filter/pdftopdf/qpdf_pdftopdf_processor.h

@@ -34,8 +34,8 @@ class QPDF_PDFTOPDF_PageHandle : public PDFTOPDF_PageHandle {
 
 class QPDF_PDFTOPDF_Processor : public PDFTOPDF_Processor {
  public:
-  virtual bool loadFile(FILE *f,ArgOwnership take=WillStayAlive);
-  virtual bool loadFilename(const char *name);
+  virtual bool loadFile(FILE *f,ArgOwnership take=WillStayAlive,int flatten_forms=1);
+  virtual bool loadFilename(const char *name,int flatten_forms=1);
 
   // TODO: virtual bool may_modify/may_print/?
   virtual bool check_print_permissions();
@@ -61,7 +61,7 @@ class QPDF_PDFTOPDF_Processor : public PDFTOPDF_Processor {
  private:
   void closeFile();
   void error(const char *fmt,...);
-  void start();
+  void start(int flatten_forms);
  private:
   std::unique_ptr<QPDF> pdf;
   std::vector<QPDFObjectHandle> orig_pages;