Error checking and getting meta-data right

src/calibre/ebooks/pdf/pdftrim.py

@@ -29,7 +29,7 @@ def config(defaults=None):
     c.add_opt('top_right_y', [ '-w', '--righty'], default=default_crop,
           help=_('Number of pixels to crop from the right most y (default is %d)')%default_crop )
     c.add_opt('bounding', ['-b', '--bounding'],
-          help=_('A file generated by ghostscript which allows each page to be individually cropped'))
+          help=_('A file generated by ghostscript which allows each page to be individually cropped [gs -dSAFER -dNOPAUSE -dBATCH -sDEVICE=bbox > bounding] '))
     return c
 
 
@@ -38,14 +38,24 @@ def option_parser():
     return c.option_parser(usage=_('''\
 	%prog [options] file.pdf
 
-	Crop a pdf. 
+	Crops a pdf. 
 	'''))
 
 def main(args=sys.argv):
     parser = option_parser()
     opts, args = parser.parse_args(args)
+    try:
         source = os.path.abspath(args[1])
         input_pdf = PdfFileReader(file(source, "rb"))
+    except:
+        print "Unable to read input"
+	return 2
+    info = input_pdf.getDocumentInfo()
+    title   = 'Unknown'
+    author  = 'Unknown'
+    if info.title:
+       title   = info.title
+       author  = info.author
     if opts.bounding != None:
         try:
             bounding = open( opts.bounding , 'r' )
@@ -53,7 +63,7 @@ def main(args=sys.argv):
         except:
             print 'Error opening %s' % opts.bounding 
             return 1
-    output_pdf = PdfFileWriter()
+    output_pdf = PdfFileWriter(title=title,author=author)
     for page_number in range (0, input_pdf.getNumPages() ):
         page = input_pdf.getPage(page_number)
         if opts.bounding != None:

src/pyPdf/pdf.py

@@ -55,7 +55,7 @@ from utils import readNonWhitespace, readUntilWhitespace, ConvertFunctionsToVirt
 # This class supports writing PDF files out, given pages produced by another
 # class (typically {@link #PdfFileReader PdfFileReader}).
 class PdfFileWriter(object):
-    def __init__(self):
+    def __init__(self,title=u"Unknown",author=u"Unknown"):
         self._header = "%PDF-1.3"
         self._objects = []  # array of indirect objects
 
@@ -71,7 +71,9 @@ class PdfFileWriter(object):
         # info object
         info = DictionaryObject()
         info.update({
-                NameObject("/Producer"): createStringObject(u"Python PDF Library - http://pybrary.net/pyPdf/")
+                NameObject("/Producer"): createStringObject(u"Python PDF Library - http://pybrary.net/pyPdf/"),
+                NameObject("/Author"): createStringObject(author),
+                NameObject("/Title"): createStringObject(title),
                 })
         self._info = self._addObject(info)