tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

rcldoc.py: port to python3. We by default exec antiword directly anyway

Browse Source
This commit is contained in:
Jean-Francois Dockes 2018-03-08 20:38:51 +01:00
parent b8fa3005dd
commit 7f49de5d97
1 changed files with 7 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
src/filters/rcldoc.py
View File

@ -85,28 +85,28 @@ class WordFilter:
self.em = em self.em = em
self.ntry = 0 self.ntry = 0
self.execdir = td self.execdir = td
self.rtfprolog = b'{\\rtf1'
self.docprolog = b'\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1'
def reset(self): def reset(self):
self.ntry = 0 self.ntry = 0
def hasControlChars(self, data): def hasControlChars(self, data):
for c in data: for c in data:
if c < chr(32) and c != '\n' and c != '\t' and \ if c < b' '[0] and c != b'\n'[0] and c != b'\t'[0] and \
c != '\f' and c != '\r': c != b'\f'[0] and c != b'\r'[0]:
return True return True
return False return False
def mimetype(self, fn): def mimetype(self, fn):
rtfprolog = b'{\\rtf1'
docprolog = b'\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1'
try: try:
f = open(fn, "rb") f = open(fn, "rb")
except: except:
return "" return ""
data = f.read(100) data = f.read(100)
if data[0:6] == rtfprolog: if data[0:6] == self.rtfprolog:
return "text/rtf" return "text/rtf"
elif data[0:8] == docprolog: elif data[0:8] == self.docprolog:
return "application/msword" return "application/msword"
elif self.hasControlChars(data): elif self.hasControlChars(data):
return "application/octet-stream" return "application/octet-stream"