filters: use rb instead of r

--HG--
branch : WINDOWSPORT
This commit is contained in:
Jean-Francois Dockes 2015-09-14 11:36:36 +02:00
parent 002eb67185
commit 7337e5a9ff
5 changed files with 5 additions and 5 deletions

View File

@ -99,7 +99,7 @@ class WordFilter:
rtfprolog ="{\\rtf1"
docprolog = b"\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1"
try:
f = open(fn, "r")
f = open(fn, "rb")
except:
return ""
data = f.read(100)

View File

@ -109,7 +109,7 @@ class European8859TextClassifier:
if __name__ == "__main__":
f = open(sys.argv[1])
f = open(sys.argv[1], "rb")
rawtext = f.read()
f.close()

View File

@ -110,7 +110,7 @@ class SVGExtractor:
fn = params["filename:"]
try:
data = open(fn, 'r').read()
data = open(fn, 'rb').read()
docdata = rclxslt.apply_sheet_data(stylesheet_all, data)
except Exception as err:
self.em.rclog("%s: bad data: " % (fn, err))

View File

@ -20,7 +20,7 @@ class TxtDump:
# No charset, so recoll will have to use its config to guess it
txt = '<html><head><title></title></head><body><pre>'
try:
f = open(fn, "r")
f = open(fn, "rb")
txt += self.em.htmlescape(f.read())
except Exception as err:
self.em.rclog("TxtDump: %s : %s" % (fn, err))

View File

@ -68,7 +68,7 @@ class XMLExtractor:
fn = params["filename:"]
try:
data = open(fn, 'r').read()
data = open(fn, 'rb').read()
docdata = rclxslt.apply_sheet_data(stylesheet_all, data)
except Exception as err:
self.em.rclog("%s: bad data: " % (fn, err))