Use PoDoFo in pdfmanipulate info

src/calibre/ebooks/pdf/manipulate/info.py

@@ -9,16 +9,14 @@ __docformat__ = 'restructuredtext en'
 Merge PDF files into a single PDF document.
 '''
 
-import os, re, sys, time
+import os, sys
-from optparse import OptionGroup, Option
 
 from calibre.utils.config import OptionParser
 from calibre.utils.logging import Log
 from calibre.constants import preferred_encoding
-from calibre.customize.conversion import OptionRecommendation
+from calibre.ebooks.pdf.verify import is_valid_pdfs, is_encrypted
-from calibre.ebooks.pdf.verify import is_valid_pdfs, is_encrypted, is_encrypted
+from calibre import prints
-
+from calibre.utils.podofo import podofo, podofo_err
-from pyPdf import PdfFileWriter, PdfFileReader
 
 USAGE = '\n%prog %%name ' + _('''\
 file.pdf ...
@@ -35,40 +33,36 @@ def option_parser(name):
     return OptionParser(usage=usage)
 
 def print_info(pdf_path):
-    with open(os.path.abspath(pdf_path), 'rb') as pdf_file:
+    if not podofo:
-        pdf = PdfFileReader(pdf_file)
+        raise RuntimeError('Failed to load PoDoFo with error:'+podofo_err)
-        print _('Title:                 %s' % pdf.documentInfo.title)
+    p = podofo.PDFDoc()
-        print _('Author:                %s' % pdf.documentInfo.author)
+    p.open(pdf_path)
-        print _('Subject:               %s' % pdf.documentInfo.subject)
+
-        print _('Creator:               %s' % pdf.documentInfo.creator)
+    fmt = lambda x, y: '%-20s: %s'%(x, y)
-        print _('Producer:              %s' % pdf.documentInfo.producer)
+
-        #print _('Creation Date:         %s' % time.strftime('%a %b %d %H:%M:%S %Y', time.gmtime(os.path.getctime(pdf_path))))
+    print
-        #print _('Modification Date:     %s' % time.strftime('%a %b %d %H:%M:%S %Y', time.gmtime(os.path.getmtime(pdf_path))))
+
-        print _('Pages:                 %s' % pdf.numPages)
+    prints(fmt(_('Title'), p.title))
-        #print _('Encrypted:             %s' % pdf.isEncrypted)
+    prints(fmt(_('Author'), p.author))
-        try:
+    prints(fmt(_('Subject'), p.subject))
-            print _('File Size:             %s bytes' % os.path.getsize(pdf_path))
+    prints(fmt(_('Creator'), p.creator))
-        except: pass
+    prints(fmt(_('Producer'), p.producer))
-        try:
+    prints(fmt(_('Pages'), p.pages))
-            pdf_file.seek(0)
+    prints(fmt(_('File Size'), os.stat(pdf_path).st_size))
-            vline = pdf_file.readline()
+    prints(fmt(_('PDF Version'), p.version if p.version else _('Unknown')))
-            mo = re.search('(?iu)^%...-(?P<version>\d+\.\d+)', vline)
-            if mo != None:
-                print _('PDF Version:           %s' % mo.group('version'))
-        except: pass
 
 def main(args=sys.argv, name=''):
     log = Log()
     parser = option_parser(name)
-    
+
     opts, args = parser.parse_args(args)
     args = args[1:]
-    
+
     if len(args) < 1:
         print 'Error: No PDF sepecified.\n'
         print_help(parser, log)
         return 1
-    
+
     bad_pdfs = is_valid_pdfs(args)
     if bad_pdfs != []:
         for pdf in bad_pdfs:
@@ -85,7 +79,7 @@ def main(args=sys.argv, name=''):
 
     for pdf in args:
         print_info(pdf)
-    
+
     return 0
 
 if __name__ == '__main__':

src/calibre/utils/podofo/podofo.cpp

@@ -108,6 +108,40 @@ podofo_PDFDoc_pages_getter(podofo_PDFDoc *self, void *closure) {
     return ans;
 }
 
+static PyObject *
+podofo_PDFDoc_version_getter(podofo_PDFDoc *self, void *closure) {
+    int version;
+    try {
+        version = self->doc->GetPdfVersion();
+    } catch(const PdfError & err) {
+        podofo_set_exception(err);
+        return NULL;
+    }
+    switch(version) {
+        case ePdfVersion_1_0:
+            return Py_BuildValue("s", "1.0");
+        case ePdfVersion_1_1:
+            return Py_BuildValue("s", "1.1");
+        case ePdfVersion_1_2:
+            return Py_BuildValue("s", "1.2");
+        case ePdfVersion_1_3:
+            return Py_BuildValue("s", "1.3");
+        case ePdfVersion_1_4:
+            return Py_BuildValue("s", "1.4");
+        case ePdfVersion_1_5:
+            return Py_BuildValue("s", "1.5");
+        case ePdfVersion_1_6:
+            return Py_BuildValue("s", "1.6");
+        case ePdfVersion_1_7:
+            return Py_BuildValue("s", "1.7");
+        default:
+            return Py_BuildValue("");
+    }
+    return Py_BuildValue("");
+}
+
+
+ 
 static PyObject *
 podofo_PDFDoc_delete_pages(podofo_PDFDoc *self, PyObject *args, PyObject *kwargs) {
     int first_page, num_pages;
@@ -315,6 +349,10 @@ static PyGetSetDef podofo_PDFDoc_getsetters[] = {
      (getter)podofo_PDFDoc_pages_getter, NULL,
      (char *)"Number of pages in document (read only)",
      NULL},
+    {(char *)"version", 
+     (getter)podofo_PDFDoc_version_getter, NULL,
+     (char *)"The PDF version (read only)",
+     NULL},
 
     {NULL}  /* Sentinel */
 };