diff options
author | Harel Ben-Attia <harelba@gmail.com> | 2014-10-30 03:19:48 -0400 |
---|---|---|
committer | Harel Ben-Attia <harelba@gmail.com> | 2014-10-30 03:19:48 -0400 |
commit | 5043992fc46ebe1555a37fc6cc88c44cb90e688c (patch) | |
tree | 1d0e7b4d98a59a33c911e1d7558722c31c24e2ae /bin | |
parent | 8d45a4aeb8cc47768368e6cfbed01ea0130dd277 (diff) |
Fixed proper encoding when opening file, limiting zipping from stdin
Diffstat (limited to 'bin')
-rwxr-xr-x | bin/q | 22 |
1 files changed, 16 insertions, 6 deletions
@@ -290,6 +290,11 @@ class BadHeaderException(Exception): return repr(self.msg) +class CannotUnzipStdInException(Exception): + + def __init__(self): + pass + class EmptyDataException(Exception): def __init__(self): @@ -714,13 +719,15 @@ class TableCreator(object): # Check if it's standard input or a file if filename == '-': - f = sys.stdin + codec_info = codecs.lookup(self.encoding) + f = codecs.StreamReaderWriter(sys.stdin,codec_info.streamreader,codec_info.streamwriter,None) + if self.gzipped: + raise CannotUnzipStdInException() else: - f = file(filename, 'rb') - - # Wrap it with gzip decompression if needed - if self.gzipped or filename.endswith('.gz'): - f = gzip.GzipFile(fileobj=f) + if self.gzipped or filename.endswith('.gz'): + f = gzip.GzipFile(fileobj=file(filename,'rb')) + else: + f = codecs.open(filename, 'rb',encoding=self.encoding) self.read_file_using_csv(f, analyze_only) if not self.table_created: @@ -1054,6 +1061,9 @@ except (UnicodeDecodeError, UnicodeError), e: except BadHeaderException, e: print >>sys.stderr, "Bad header row: %s" % e.msg sys.exit(35) +except CannotUnzipStdInException,e: + print >>sys.stderr,"Cannot decompress standard input. Pipe the input through zcat in order to decompress." + sys.exit(36) except KeyboardInterrupt: print >>sys.stderr, "Interrupted" sys.exit(0) |