Merge 8b2b013b36ff05b3f98afed416fe3e87023b121a into 5b5de3aff6901b4553015eb40028504e64a25524

When exploring a PDF, it may be convenient to have the typical
informations delivered by some "Document Properties"---and some more
about the MediaBox(es).

So just add the function to get the ModDate and extend the
pdfioinfo example as an example of what the library do have
and pdfioinfo as a debugging tool also.

Signed-off-by: Thierry LARONDE <tlaronde@kergis.com>

CHANGES.md

@@ -1,6 +1,12 @@
 Changes in PDFio
 ================
 
+v1.?.? - YYYY-MM-DD
+-------------------
+
+- Updated the pdf2txt example to support font encodings.
+
+
 v1.4.1 - 2025-01-24
 -------------------
 

examples/pdfioinfo.c

@@ -25,11 +25,18 @@ main(int  argc,				// I - Number of command-line arguments
 {
   const char	*filename;		// PDF filename
   pdfio_file_t	*pdf;			// PDF file
-  const char	*author;		// Author name
-  time_t	creation_date;		// Creation date
-  struct tm	*creation_tm;		// Creation date/time information
-  char		creation_text[256];	// Creation date/time as a string
+  const char	*author,		// Author name
+  		*creator,		// Creator name
+  		*producer;		// Producer name
+  time_t	creation_date,		// Creation date
+		mod_date;		// Modification date
+  struct tm	*creation_tm,		// Creation date/time information
+		*mod_tm;		// Mod. date/time information
+  char		creation_text[256],	// Creation date/time as a string
+		mod_text[256];		// Mod. date/time human fmt string
   const char	*title;			// Title
+  size_t	num_pages;		// PDF number of pages
+  bool		has_acroform;		// AcroForm or not
 
 
   // Get the filename from the command-line...
@@ -48,9 +55,12 @@ main(int  argc,				// I - Number of command-line arguments
   if (pdf == NULL)
     return (1);
 
-  // Get the title and author...
+  // Get the title, author...
   author = pdfioFileGetAuthor(pdf);
   title  = pdfioFileGetTitle(pdf);
+  creator = pdfioFileGetCreator(pdf);
+  producer = pdfioFileGetProducer(pdf);
+  num_pages = pdfioFileGetNumPages(pdf);
 
   // Get the creation date and convert to a string...
   if ((creation_date = pdfioFileGetCreationDate(pdf)) > 0)
@@ -63,12 +73,82 @@ main(int  argc,				// I - Number of command-line arguments
     snprintf(creation_text, sizeof(creation_text), "-- not set --");
   }
 
+  // Get the modification date and convert to a string...
+  if ((mod_date = pdfioFileGetModDate(pdf)) > 0)
+  {
+    mod_tm   = localtime(&mod_date);
+    strftime(mod_text, sizeof(mod_text), "%c", mod_tm);
+  }
+  else
+  {
+    snprintf(mod_text, sizeof(mod_text), "-- not set --");
+  }
+
+  // Detect simply if AcroFrom is a dict in catalog
+  {
+    pdfio_dict_t	*dict;			// some Object dictionary
+   
+    dict = pdfioFileGetCatalog(pdf);
+    has_acroform = (dict != NULL && pdfioDictGetObj(dict, "AcroForm") != NULL)?
+                   true : false;
+  }
+
   // Print file information to stdout...
   printf("%s:\n", filename);
   printf("         Title: %s\n", title ? title : "-- not set --");
   printf("        Author: %s\n", author ? author : "-- not set --");
+  printf("       Creator: %s\n", creator ? creator : "-- not set --");
+  printf("      Producer: %s\n", producer ? producer : "-- not set --");
   printf("    Created On: %s\n", creation_text);
-  printf("  Number Pages: %u\n", (unsigned)pdfioFileGetNumPages(pdf));
+  printf("   Modified On: %s\n", mod_text);
+  printf("       Version: %s\n", pdfioFileGetVersion(pdf));
+  printf("      AcroForm: %s\n", has_acroform ? "Yes" : "No");
+  printf("  Number Pages: %u\n", (unsigned)num_pages);
+  printf("    MediaBoxes:");
+  // There can be a different MediaBox per page
+  // Loop and report MediaBox and number of consecutive pages of this size
+  {
+    pdfio_obj_t		*obj;			// Object
+    pdfio_dict_t	*dict;			// Object dictionary
+    pdfio_rect_t 	prev,			// MediaBox previous
+     			now;			// MediaBox now
+    size_t 		n,			// Page index
+          		nprev;			// Number previous prev size
+
+    // MediaBox should be set at least on the root
+    for (n = nprev = 0; n < num_pages; n++)
+    {
+      obj = pdfioFileGetPage(pdf, n);
+      while (obj != NULL)
+      {
+        dict = pdfioObjGetDict(obj);
+        if (pdfioDictGetRect(dict, "MediaBox", &now))
+        {
+          if ( 
+                nprev == 0
+                || (
+                     now.x1 != prev.x1 || now.y1 != prev.y1
+                     || now.x2 != prev.x2 || now.y2 != prev.y2
+                   )
+             )
+          {
+            if (nprev) printf("(%zd) ", nprev);
+            prev = now;
+            printf("[%.7g %.7g %.7g %.7g]", now.x1, now.y1, now.x2, now.y2);
+            nprev = 1;
+          }
+          else
+            ++nprev;
+	  obj = NULL;
+        }
+        else
+	  obj = pdfioDictGetObj(dict, "Parent");
+      }
+    }
+    printf("(%zd)", nprev);
+  }
+  printf("\n");
+
 
   // Close the PDF file...
   pdfioFileClose(pdf);

pdfio-file.c

@@ -801,6 +801,18 @@ pdfioFileGetKeywords(pdfio_file_t *pdf)	// I - PDF file
 }
 
 
+//
+// 'pdfioFileGetModDate()' - Get the most recent modification date for a PDF file.
+//
+
+time_t					// O - Modification date or `0` for none
+pdfioFileGetModDate(
+    pdfio_file_t *pdf)			// I - PDF file
+{
+  return (pdf && pdf->info_obj ? pdfioDictGetDate(pdfioObjGetDict(pdf->info_obj), "ModDate") : 0);
+}
+
+
 //
 // 'pdfioFileGetName()' - Get a PDF's filename.
 //

pdfio.h

@@ -201,6 +201,7 @@ extern time_t		pdfioFileGetCreationDate(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern const char	*pdfioFileGetCreator(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern pdfio_array_t	*pdfioFileGetID(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern const char	*pdfioFileGetKeywords(pdfio_file_t *pdf) _PDFIO_PUBLIC;
+extern time_t		pdfioFileGetModDate(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern const char	*pdfioFileGetName(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern size_t		pdfioFileGetNumObjs(pdfio_file_t *pdf) _PDFIO_PUBLIC;
 extern size_t		pdfioFileGetNumPages(pdfio_file_t *pdf) _PDFIO_PUBLIC;

pdfio1.def

@@ -204,6 +204,7 @@ pdfioFileGetCreationDate
 pdfioFileGetCreator
 pdfioFileGetID
 pdfioFileGetKeywords
+pdfioFileGetModDate
 pdfioFileGetName
 pdfioFileGetNumObjs
 pdfioFileGetNumPages