Python xlrd: suppress warning messages

I am using xlrd to process Ex开发者_运维百科cel files. I am running a script on a folder that contains many files, and I am printing messages related to the files. However, for each file I run, I get the following xlrd-generated error message as well:

WARNING *** OLE2 inconsistency: SSCS size is 0 but SSAT size is non-zero

Is there a way to suppress this error message, so the CLI will only print the message I want it to?

The answer by John works, but has a small problem:

xlrd writes that warning message and the following newline character separately to the logfile. Therefore you will get an empty line in your stdout instead of the message, if you use the filter class proposed by John. You shouldn't simply filter out all newlines from the log output either though, because there could be "real" warnings that would then be missing the newlines.

If you want to simply ignore all log output by xlrd, this is probably the simplest solution:

book = xlrd.open_workbook("foo.xls", logfile=open(os.devnull, 'w'))

Check out the relevant part of the xlrd docs. The 2nd arg of the open_workbook function is logfile which should be an open file object or act-alike. All it needs to support is a write method. It defaults to sys.stdout.

So, something like this (untested) should do the job:

class MyFilter(object):
    def __init__(self, mylogfile=sys.stdout):
        self.f = mylogfile
    def write(self, data):
        if "WARNING *** OLE2 inconsistency" not in data:
            self.f.write(data)

#start up
log = open("the_log_file.txt", "w")
log_filter = MyFilter(log)
book = xlrd.open_workbook("foo.xls", logfile=log_filter)

# shut down
log.close()
# or use a "with" statement

Update in response to answer by @DaniloBargen:

It's not xlrd that's writing the newline separately, it's the Python print statement/function. This script:

class FakeFile(object):
    def write(self, data):
        print repr(data)

ff = FakeFile()
for x in "foo bar baz".split():
    print >> ff, x

produces this output for all Pythons 2.2 to 2.7 both inclusive:

'foo'
'\n'
'bar'
'\n'
'baz'
'\n'

A suitably modernised script (print as a function instead of a statement) produces identical output for 2.6, 2.7, 3.1, 3.2, and 3.3. You can work around this with a more complicated filter class. The following example additionally allows a sequence of phrases to be checked for:

import sys, glob, xlrd

class MyFilter(object):
    def __init__(self, mylogfile=sys.stdout, skip_list=()):
        self.f = mylogfile
        self.state = 0
        self.skip_list = skip_list
    def write(self, data):
        if self.state == 0:
            found = any(x in data for x in self.skip_list)
            if not found:
                self.f.write(data)
                return
            if data[-1] != '\n':
                self.state = 1
        else:
            if data != '\n':
                self.f.write(data)
            self.state = 0

logf = open("the_log_file.txt", "w")
skip_these = (
    "WARNING *** OLE2 inconsistency",
    )
try:        
    log_filter = MyFilter(logf, skip_these)
    for fname in glob.glob(sys.argv[1]):
        logf.write("=== %s ===\n" % fname)
        book = xlrd.open_workbook(fname, logfile=log_filter)
finally:
    logf.close()

If you're getting this warning with pandas' read_excel function, then you can do this:

import os
import xlrd
import pandas as pd

workbook = xlrd.open_workbook(your_path, logfile=open(os.devnull, "w"))
df = pd.read_excel(workbook) # read_excel accepts workbooks too

import warnings

def fxn():
    warnings.warn("deprecated", DeprecationWarning)

with warnings.catch_warnings():
    warnings.simplefilter("ignore")
    fxn()

-> http://docs.python.org/library/warnings.html#temporarily-suppressing-warnings

For what it's worth I had the same warning message; when I deleted the first row (which was empty) the warning disappeared.

Based on John Machin's answer, here is some working code (tested with Python 3.6) that works with sys.stdout:

import io
import sys
import xlrd

path = "somefile.xls" # path to an XLS file to load or a file-like object for one
encoding_override = None

# Mute xlrd warnings for OLE inconsistencies
class LogFilter(io.TextIOWrapper):
    def __init__(self, buffer=sys.stdout, *args, **kwargs):
        self.buffer = buffer
        super(LogFilter, self).__init__(buffer, *args, **kwargs)
    def write(self, data):
        if isinstance(data, str):
            if not data.startswith("WARNING *** OLE2 inconsistency: "):
                super(LogFilter, self).write(data)
        elif isinstance(data, bytes):
            super(LogFilter, self).write(data.decode(self.buffer.encoding))
        else:
            super(LogFilter, self).write(data)

def open_workbook(file_contents, encoding_override):
    logfilter = LogFilter()
    return xlrd.open_workbook(file_contents=file_contents, logfile=logfilter,
                              encoding_override=encoding_override)

if hasattr(path, 'read'):
    book = open_workbook(file_contents=path.read(), encoding_override=encoding_override)
else:
    with open(path, 'rb') as f:
        book = open_workbook(file_contents=f.read(), encoding_override=encoding_override)