Mercurial > repos > adam-novak > hexagram
comparison hexagram-6ae12361157c/hexagram/tsv.py @ 0:1407e3634bcf draft default tip
Uploaded r11 from test tool shed.
| author | adam-novak |
|---|---|
| date | Tue, 22 Oct 2013 14:17:59 -0400 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:1407e3634bcf |
|---|---|
| 1 #!/usr/bin/env python2.7 | |
| 2 # tsv.py: a module for writing TSV (tab-separated value) files | |
| 3 """ | |
| 4 This module defines two classes: a TsvWriter, which can be constructed on a | |
| 5 stream to allow writing TSV data lines and #-delimited comments to that stream, | |
| 6 and a TsvReader, which can be constructed on a stream and iterated over to | |
| 7 obtain lists of the values from each non-comment line in the stream. | |
| 8 | |
| 9 TSV is most useful as the basis for other, more tightly specified file formats. | |
| 10 | |
| 11 """ | |
| 12 | |
| 13 class TsvWriter(object): | |
| 14 """ | |
| 15 Represents a writer for tab-separated value files containing #-delimited | |
| 16 comments. | |
| 17 | |
| 18 """ | |
| 19 def __init__(self, stream): | |
| 20 """ | |
| 21 Make a new TsvWriter for writing TSV data to the given stream. | |
| 22 """ | |
| 23 | |
| 24 # This holds the stream | |
| 25 self.stream = stream | |
| 26 | |
| 27 | |
| 28 def line(self, *args): | |
| 29 """ | |
| 30 Write the given values to the file, as a TSV line. Args holds a list of | |
| 31 all arguments passed. Any argument that stringifies to a string legal as | |
| 32 a TSV data item can be written. | |
| 33 | |
| 34 """ | |
| 35 | |
| 36 self.list_line(args) | |
| 37 | |
| 38 | |
| 39 def list_line(self, line): | |
| 40 """ | |
| 41 Write the given iterable of values (line) to the file as items on the | |
| 42 same line. Any argument that stringifies to a string legal as a TSV data | |
| 43 item can be written. | |
| 44 | |
| 45 Does not copy the line or build a big string in memory. | |
| 46 """ | |
| 47 | |
| 48 if len(line) == 0: | |
| 49 return | |
| 50 | |
| 51 self.stream.write(str(line[0])) | |
| 52 | |
| 53 for item in line[1:]: | |
| 54 self.stream.write("\t") | |
| 55 self.stream.write(str(item)) | |
| 56 | |
| 57 self.stream.write("\n") | |
| 58 | |
| 59 def comment(self, text): | |
| 60 """ | |
| 61 Write the given text as a TSV comment. text must be a string containing | |
| 62 no newlines. | |
| 63 | |
| 64 """ | |
| 65 | |
| 66 self.stream.write("# {}\n".format(text)) | |
| 67 | |
| 68 def close(self): | |
| 69 """ | |
| 70 Close the underlying stream. | |
| 71 """ | |
| 72 | |
| 73 self.stream.close() | |
| 74 | |
| 75 class TsvReader(object): | |
| 76 """ | |
| 77 Represents a reader for tab-separated value files. Skips over comments | |
| 78 starting with #. Can be iterated over. | |
| 79 | |
| 80 Field values consisting of only whitespace are not allowed. | |
| 81 """ | |
| 82 | |
| 83 def __init__(self, stream): | |
| 84 """ | |
| 85 Make a new TsvReader to read from the given stream. | |
| 86 """ | |
| 87 | |
| 88 self.stream = stream | |
| 89 | |
| 90 def __iter__(self): | |
| 91 """ | |
| 92 Yields lists of all fields on each line, as strings, until all lines are | |
| 93 exhausted. Strips whitespace around field contents. | |
| 94 """ | |
| 95 | |
| 96 for line in self.stream: | |
| 97 line = line.strip() | |
| 98 if line == "" or line[0] == "#": | |
| 99 # Skip comments and empty lines | |
| 100 continue | |
| 101 | |
| 102 yield map(str.strip, line.split("\t")) | |
| 103 | |
| 104 def close(self): | |
| 105 """ | |
| 106 Close the underlying stream. | |
| 107 """ | |
| 108 | |
| 109 self.stream.close() |