Switch to using pdftk to set PDF metadata on windows and linux as it is more robust

installer/windows/freeze.py

@@ -11,6 +11,7 @@ LIBUSB_DIR       = 'C:\\libusb'
 LIBUNRAR         = 'C:\\Program Files\\UnrarDLL\\unrar.dll'
 PDFTOHTML        = 'C:\\cygwin\\home\\kovid\\poppler-0.10.6\\rel\\pdftohtml.exe'
 IMAGEMAGICK_DIR  = 'C:\\ImageMagick'
+PDFTK            = 'C:\\pdftk.exe'
 FONTCONFIG_DIR   = 'C:\\fontconfig'
 VC90             = r'C:\VC90.CRT'
 
@@ -100,6 +101,9 @@ class BuildEXE(py2exe.build_exe.py2exe):
         shutil.copyfile(PDFTOHTML, os.path.join(PY2EXE_DIR, os.path.basename(PDFTOHTML)))
         shutil.copyfile(PDFTOHTML+'.manifest', os.path.join(PY2EXE_DIR,
             os.path.basename(PDFTOHTML)+'.manifest'))
+        print '\tAdding pdftk'
+        shutil.copyfile(PDFTK, os.path.join(PY2EXE_DIR, os.path.basename(PDFTK)))
+
         print '\tAdding ImageMagick'
         for f in os.listdir(IMAGEMAGICK_DIR):
             shutil.copyfile(os.path.join(IMAGEMAGICK_DIR, f), os.path.join(PY2EXE_DIR, f))

src/calibre/ebooks/metadata/pdf.py

@@ -9,6 +9,7 @@ from threading import Thread
 from calibre import FileWrapper
 from calibre.ebooks.metadata import MetaInformation, authors_to_string, get_parser
 from pyPdf import PdfFileReader, PdfFileWriter
+from calibre.utils.pdftk import set_metadata as pdftk_set_metadata
 
 def get_metadata(stream):
     """ Return metadata as a L{MetaInfo} object """
@@ -49,6 +50,12 @@ class MetadataWriter(Thread):
 
 def set_metadata(stream, mi):
     stream.seek(0)
+    try:
+        pdftk_set_metadata(stream, mi)
+    except:
+        pass
+    else:
+        return
 
     # Use a StringIO object for the pdf because we will want to over
     # write it later and if we are working on the stream directly it

src/calibre/ptempfile.py

@@ -1,8 +1,8 @@
 from __future__ import with_statement
 __license__   = 'GPL v3'
 __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
-""" 
+"""
-Provides platform independent temporary files that persist even after 
+Provides platform independent temporary files that persist even after
 being closed.
 """
 import tempfile, os, atexit, shutil
@@ -15,38 +15,38 @@ def cleanup(path):
         if os.path.exists(path):
             os.remove(path)
     except:
-        pass   
+        pass
-    
+
 class PersistentTemporaryFile(object):
-    """ 
+    """
     A file-like object that is a temporary file that is available even after being closed on
     all platforms. It is automatically deleted on normal program termination.
     """
     _file = None
-    
+
     def __init__(self, suffix="", prefix="", dir=None, mode='w+b'):
-        if prefix == None: 
+        if prefix == None:
             prefix = ""
         fd, name = tempfile.mkstemp(suffix, __appname__+"_"+ __version__+"_" + prefix,
                                     dir=dir)
-        self._file = os.fdopen(fd, 'w+b')
+        self._file = os.fdopen(fd, mode)
         self._name = name
         atexit.register(cleanup, name)
-        
+
     def __getattr__(self, name):
         if name == 'name':
             return self.__dict__['_name']
         return getattr(self.__dict__['_file'], name)
-    
+
     def __enter__(self):
         return self
-    
+
     def __exit__(self, *args):
         self.close()
-        
+
     def __del__(self):
         self.close()
-          
+
 
 def PersistentTemporaryDirectory(suffix='', prefix='', dir=None):
     '''
@@ -66,12 +66,37 @@ class TemporaryDirectory(object):
         self.prefix = prefix
         self.dir = dir
         self.keep = keep
-    
+
     def __enter__(self):
         self.tdir = tempfile.mkdtemp(self.suffix, __appname__+"_"+ __version__+"_" +self.prefix, self.dir)
         return self.tdir
-    
+
     def __exit__(self, *args):
         if not self.keep and os.path.exists(self.tdir):
             shutil.rmtree(self.tdir, ignore_errors=True)
-            
+
+class TemporaryFile(object):
+
+    def __init__(self, suffix="", prefix="", dir=None, mode='w+b'):
+        if prefix == None:
+            prefix = ''
+        if suffix is None:
+            suffix = ''
+        self.prefix, self.suffix, self.dir, self.mode = prefix, suffix, dir, mode
+        self._file = None
+
+    def __enter__(self):
+        fd, name = tempfile.mkstemp(self.suffix,
+                __appname__+"_"+ __version__+"_" + self.prefix,
+                                    dir=self.dir)
+        self._file = os.fdopen(fd, self.mode)
+        self._name = name
+        self._file.close()
+        return name
+
+    def __exit__(self, *args):
+        cleanup(self._name)
+
+
+
+

src/calibre/trac/plugins/download.py

@@ -20,6 +20,7 @@ DEPENDENCIES = [
             ('BeautifulSoup', '3.0.5', 'beautifulsoup', 'python-beautifulsoup', 'python-BeautifulSoup'),
             ('dnspython', '1.6.0', 'dnspython', 'dnspython', 'dnspython', 'dnspython'),
             ('poppler', '0.10.5', 'poppler', 'poppler', 'poppler', 'poppler'),
+            ('pdftk', '1.12', 'pdftk', 'pdftk', 'pdftk', 'pdftk'),
             ]
 
 

src/calibre/utils/pdftk.py

@@ -0,0 +1,80 @@
+#!/usr/bin/env python
+# vim:fileencoding=UTF-8:ts=4:sw=4:sta:et:sts=4:ai
+from __future__ import with_statement
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Kovid Goyal <kovid@kovidgoyal.net>'
+__docformat__ = 'restructuredtext en'
+
+
+import sys, subprocess, os, errno
+from functools import partial
+from contextlib import nested
+
+from calibre.ptempfile import TemporaryFile
+from calibre.constants import iswindows
+
+PDFTK = 'pdftk'
+popen = subprocess.Popen
+#if isosx and hasattr(sys, 'frameworks_dir'):
+#    PDFTK = os.path.join(getattr(sys, 'frameworks_dir'), 'pdftk')
+if iswindows and hasattr(sys, 'frozen'):
+    PDFTK = os.path.join(os.path.dirname(sys.executable), 'pdftk.exe')
+    popen = partial(subprocess.Popen, creationflags=0x08) # CREATE_NO_WINDOW=0x08 so that no ugly console is popped up
+
+class PdftkError(Exception): pass
+
+def mi_to_info(mi):
+    ans = []
+    if mi.title:
+        ans.extend(('InfoKey: Title', 'InfoValue: '+mi.title))
+    if mi.authors:
+        from calibre.ebooks.metadata import authors_to_string
+        ans.extend(('InfoKey: Author', 'InfoValue: ' +
+            authors_to_string(mi.authors)))
+    return u'\n'.join(ans)
+
+def set_metadata(stream, mi):
+    raw = mi_to_info(mi)
+    if not raw: return
+    raw = raw.encode('utf-8')
+    with nested(TemporaryFile('.pdf'), TemporaryFile('.pdf'),
+            TemporaryFile('.info')) as (input, output, meta):
+        oi = getattr(stream, 'name', None)
+        if not oi or not os.access(oi, os.R_OK):
+            stream.seek(0)
+            with open(input, 'wb') as f: f.write(stream.read())
+        else:
+            input = oi
+        with open(meta, 'wb') as f: f.write(raw)
+        if os.path.exists(output):
+            os.remove(output)
+        cmd = (PDFTK, input, 'update_info', meta, 'output', output)
+        p = popen(cmd)
+
+        while True:
+            try:
+                ret = p.wait()
+                break
+            except OSError, e:
+                if e.errno == errno.EINTR:
+                    continue
+                else:
+                    raise
+
+        if os.stat(output).st_size < 2048:
+            raise PdftkError('Output file too small')
+
+        with open(output, 'rb') as f: raw = f.read()
+        if raw:
+            stream.seek(0)
+            stream.truncate()
+            stream.write(raw)
+            stream.flush()
+
+if __name__ == '__main__':
+    args = sys.argv
+    from calibre.ebooks.metadata import MetaInformation
+    mi = MetaInformation(args[2], [args[3]])
+    x = open(args[1], 'r+b')
+    set_metadata(x, mi)