summaryrefslogtreecommitdiff
path: root/lib/misc.py
diff options
context:
space:
mode:
authorjvoisin2011-06-21 20:41:18 +0200
committerjvoisin2011-06-21 20:41:18 +0200
commit9e69adbe1b065707f8be4f146cc3c05660cef711 (patch)
treed60509a4982d7699204059184c4343352fef52de /lib/misc.py
parentf0c9c5b56e3909ba36cc84ff82b05fab9a180911 (diff)
Add pdfrw, and many files that I have forgetten, sorry !
Diffstat (limited to '')
-rw-r--r--lib/misc.py44
1 files changed, 44 insertions, 0 deletions
diff --git a/lib/misc.py b/lib/misc.py
new file mode 100644
index 0000000..56c2274
--- /dev/null
+++ b/lib/misc.py
@@ -0,0 +1,44 @@
1import parser
2import pdfrw
3
4class PdfStripper(parser.Generic_parser):
5 '''
6 Represent a pdf file, with the help of pdfrw
7 '''
8 def __init__(self, filename):
9 self.filename = filename
10 self.trailer = pdfrw.PdfReader(self.filename)
11 self.writer = pdfrw.PdfWriter()
12
13 def remove_all(self):
14 '''
15 Remove all the files that are compromizing
16 '''
17 self.trailer.Info.Title = ''
18 self.trailer.Info.Author = ''
19 self.trailer.Info.Producer = ''
20 self.trailer.Info.Creator = ''
21 self.trailer.Info.CreationDate = ''
22 self.trailer.Info.ModDate = ''
23
24 self.writer.trailer = self.trailer
25 self.writer.write(self.filename + parser.POSTFIX)
26
27 def is_clean(self):
28 '''
29 Check if the file is clean from harmful metadatas
30 '''
31 for field in self.trailer.Info:
32 if field != '':
33 return False
34 return True
35
36 def get_meta(self):
37 '''
38 return a dict with all the meta of the file
39 '''
40 metadata = {}
41 for key, value in self.trailer.Info.iteritems():
42 metadata[key[1:]] = value[1:-1]
43 return metadata
44